![Pore size distribution and sieving coeffi cient profi les for three hypothetical membranes A, B and C. a Relationship between number of pores and pore size. b Sieving coeffi cient as a function of solute molecular weight. Reprinted with permission from Ronco et al. [14].](https://www.researchgate.net/profile/Bernard-Canaud/publication/232423073/figure/fig1/AS:669989644296193@1536749387009/Pore-size-distribution-and-sieving-coeffi-cient-profi-les-for-three-hypothetical_Q320.jpg)
Content uploaded by Bernard Canaud
Author content
All content in this area was uploaded by Bernard Canaud on Jan 29, 2014
Content may be subject to copyright.
Hemodiafiltration
Contributions to Nephrology
Vol. 158
Series Editor
Claudio Ronco, Vicenza
Hemodiafiltration
Basel · Freiburg · Paris · London · New York ·
Bangalore · Bangkok · Singapore · Tokyo · Sydney
Volume Editors
Claudio Ronco, Vicenza
Bernard Canaud, Montpellier
Pedro Aljama, Cordoba
42 figures, 5 in color, and 11 tables, 2007
Claudio Ronco Bernard Canaud
Department of Nephrology Department of Nephrology
St. Bortolo Hospital Dialysis and Intensive Care
I-36100 Vicenza (Italy) University Hospital Lapeyronie
F-34295 Montpellier (France)
Pedro Aljama
Department of Nephrology
University Hospital Reina Sofía
E-14004 Cordoba (Spain)
Bibliographic Indices. This publication is listed in bibliographic services, including Current Contents® and
Index Medicus.
Disclaimer. The statements, options and data contained in this publication are solely those of the individ-
ual authors and contributors and not of the publisher and the editor(s). The appearance of advertisements in the
book is not a warranty, endorsement, or approval of the products or services advertised or of their effectiveness,
quality or safety. The publisher and the editor(s) disclaim responsibility for any injury to persons or property
resulting from any ideas, methods, instructions or products referred to in the content or advertisements.
Drug Dosage. The authors and the publisher have exerted every effort to ensure that drug selection and dos-
age set forth in this text are in accord with current recommendations and practice at the time of publication.
However, in view of ongoing research, changes in government regulations, and the constant flow of information
relating to drug therapy and drug reactions, the reader is urged to check the package insert for each drug for any
change in indications and dosage and for added warnings and precautions. This is particularly important when the
recommended agent is a new and/or infrequently employed drug.
All rights reserved. No part of this publication may be translated into other languages, reproduced or
utilized in any form or by any means electronic or mechanical, including photocopying, recording, microcopying,
or by any information storage and retrieval system, without permission in writing from the publisher.
© Copyright 2007 by S. Karger AG, P.O. Box, CH–4009 Basel (Switzerland)
www.karger.com
Printed in Switzerland on acid-free and non-aging paper (ISO 9706) by Reinhardt Druck, Basel
ISSN 0302–5144
ISBN 978–3–8055–8288–9
Library of Congress Cataloging-in-Publication Data
Contributions to Nephrology
(Founded 1975 by Geoffrey M. Berlyne)
Hemodiafi ltration / volume editors, Claudio Ronco, Bernard Canaud, Pedro
Aljama.
p. ; cm. – (Contributions to nephrology, ISSN 0302-5144 ; v. 158)
Includes bibliographical references and indexes.
ISBN 978-3-8055-8288-9 (hard cover : alk. paper)
1. Blood–Filtration. 2. Hemodialysis. I. Ronco, C. (Claudio), 1951– II.
Canaud, Bernard. III. Aljama, Pedro. IV. Series.
[DNLM: 1. Hemodiafi ltration–methods. W1 CO778UN v.158 2007 / WJ 378
H4883 2007]
RC901.7.H47H43 2007
617.4'61059–dc22
2007024039
Contents
IX Preface
Ronco, C. (Vicenza); Canaud, B. (Montpellier); Aljama, P. (Cordoba)
History and Evolution
1 The Birth of Hemodiafiltration
Henderson, L.W. (Vernon Hills, Ill.)
9 Evolution of Hemodiafiltration
Ronco, C. (Vicenza)
Basic Principles
20 Solute Removal by Hollow-Fiber Dialyzers
Clark, W.R. (Lakewood, Colo./Indianapolis, Ind.); Rocha, E. (Rio de Janeiro);
Ronco, C. (Vicenza)
34 Fluid Mechanics and Crossfiltration in Hollow-Fiber
Hemodialyzers
Ronco, C. (Vicenza)
50 Mechanisms of Solute and Fluid Removal in
Hemodiafiltration
Yamashita, A.C. (Fujisawa)
V
Contents VI
Membranes and Hardware for Hemodiafiltration
57 Membranes and Filters for Haemodiafiltration
Hoenich, N.A. (Newcastle upon Tyne)
68 Technical Aspects of Online Hemodiafiltration
Polaschegg, H.-D. (Köstenberg); Roy, T. (Bad Homburg)
Technical Aspects and Fluids in Hemodiafiltration
80 Quality of Water, Dialysate and Infusate
Cappelli, G.; Ricardi, M.; Bonucchi, D.; De Amicis, S. (Modena)
87 Fluids in Bags for Hemodiafiltration
Ledebo, I. (Lund)
94 Hemodiafiltration with Endogenous Reinfusion
Wratten, M.L. (Mirandola); Ghezzi, P.M. (Montemassi)
Hemodiafiltration Techniques
103 Low- (Classical) and High-Efficiency Haemodiafiltration
Wizemann, V. (Giessen)
110 Online Hemodiafiltration
Technical Options and Best Clinical Practices
Canaud, B. (Montpellier)
123 Mixed-Dilution Hemodiafiltration
Pedrini, L.A.; Zerbi, S. (Seriate)
131 Paired Hemodiafiltration
Pizzarelli, F. (Florence)
138 Acetate-Free Biofiltration
Santoro, A.; Guarnieri, F.; Ferramosca, E.; Grandi, F. (Bologna)
153 Mid-Dilution: An Innovative High-Quality and Safe Haemodiafiltration
Approach
Potier, J. (Cherbourg)
161 Double High-Flux Hemodiafiltration
von Albertini, B. (Lausanne)
Contents VII
169 Push/Pull Hemodiafiltration
Shinzato, T.; Maeda, K. (Nagoya)
177 Principles and Practice of Internal Hemodiafiltration
Fiore, G.B. (Milan); Ronco, C. (Vicenza)
Clinical Aspects of Hemodiafiltration
185 Clinical Aspects of Haemodiafiltration
Locatelli, F.; Di Filippo, S.; Manzoni, C. (Lecco)
194 The Biological Response to Online Hemodiafiltration
Panichi, V. (Pisa); Tetta, C. (Bad Homburg)
201 Clearance of Beta-2-Microglobulin and Middle Molecules in
Haemodiafiltration
Tattersall, J. (Leeds)
210 Inflammation and Hemodiafiltration
Ramirez, R.; Martin-Malo, A.; Aljama, P. (Cordoba)
216 Effect of Online Hemodiafiltration on Morbidity and Mortality of
Chronic Kidney Disease Patients
Canaud, B. (Montpellier)
225 Optimizing the Prescription of Hemodiafiltration
Maduell, F. (Barcelona)
232 Author Index
233 Subject Index
Preface
Exactly 40 years after the first contribution of Lee W. Henderson on the
potential use of convection as a blood-cleansing modality and exactly 30 years
after H. Leber published his first paper on a new dialysis modality called ‘hemo-
diafiltration’, this book is a tribute to the genius and creativity in the field of
artificial organs. But the book is not just a homage to the brilliant idea or to the
important investigators, it is a real updated review of the evolution, the advances
and the recent results achieved by hemodiafiltration in the clinical arena as
renal replacement modality. For this reason, this comprehensive review, made
possible by a series of outstanding scientists and physicians, represents today an
important source of information and a valuable tool for implementing hemodia-
filtration in the daily practice. For a long time, results were limited, and evi-
dence was scanty and insufficient to expand the application of hemodiafiltration;
recently however, large studies and important clinical investigations have pro-
duced enough evidence for a clinical application of hemodiafiltration on a
broader scale and even on a routine basis in some centers.
The present book is a collection of papers that include historical notes and
a journey through the evolution of different forms of hemodiafiltration, made
possible by technological developments in the field of membranes, machines
and fluids.
The subsequent group of papers describe the theoretical rationale for
hemodiafiltration with a detailed analysis of the involved mass separation pro-
cesses and the hydraulic properties of the dialyzers. In this section, fluid
mechanics and crossfiltration in hollow-fiber hemodialyzers are described in
detail.
IX
A special section has been devoted to the description of different hemodia-
filtration techniques; in each chapter, a specific technique is analyzed, and the
particular transport mechanism and related technology are reported.
Finally, the clinical effects of hemodiafiltration are described in a series of
chapters that conclude the book.
At the end of this important enterprise, we are proud of our effort to unify
the knowledge about hemodiafiltration. The book includes different technolo-
gies and therefore offers the readers a complete overview of the technical and
clinical possibilities provided by the technique in its widest concept. It is not a
case that the three editors come from three different European countries where
hemodiafiltration has found large application and interesting clinical results.
Joining our efforts has been mutually rewarding, but also a precise warranty
that a multinational view has been conveyed in the main message of the book.
We are indebted to Karger for the prompt and efficient publication of the
book that allows physicians and investigators to have a comprehensive over-
view of hemodiafiltration from its molecular basis to the most practical appli-
cation. Thus, we hope that this book will represent an important aid for beginners
and for experts, for scientists and for physicians, for students and for senior
faculty members, creating the bridges that today’s translational research is
intended to build.
Claudio Ronco
Bernard Canaud
Pedro Aljama
Preface X
History and Evolution
Ronco C, Canaud B, Aljama P (eds): Hemodiafi ltration.
Contrib Nephrol. Basel, Karger, 2007, vol 158, pp 1–8
The Birth of Hemodiafiltration
L.W. Henderson
Sorbent Therapeutics Inc., Vernon Hills, Ill., USA
It should be duly noted that there have been several relevant writings on
the origin of the fi ltration technologies to treat end-stage renal disease (ESRD)
[1–7]. In this chapter, I will provide a brief précis of this work on the early his-
tory of the use of convection, rather than diffusion, to cleanse uremic blood.
I will then focus on the ‘birth’ of ‘hemodiafi tration’ as a clinical tool for the
treatment of ESRD.
Preconception: Précis of Earlier Work
Lysaght [4] notes that ultrafi ltration of whole blood for analytical purposes
dates to 1928 with the work of Brull [8]. However, the fi rst application of ultra-
fi ltration for the purpose of removing uremic toxins may be ascribed to Malinow
and Korzon [9] who, in 1947, described the use of an 0.8-m2 cellophane ultra-
fi lter in uremic dogs that was intended to ‘duplicate glomerular function’ and
prolong their life. The blood fl ow of 100 ml/min and pressure of 500 mm Hg
yielded only 15–20 ml/min of ultrafi ltrate, which was duly replaced with inter-
mittent intravenous injections of Krebs-Ringer solution. Approximately 7 liters
of fl uid was exchanged over an 8-hour time period, and blood urea nitrogen
fell from 175 to 75 mg/dl. The report in 1944 of the rotating-drum diffusion-
based dialyzer by Kolff et al. [10] eclipsed these early fi ndings with convec-
tive transport. Others, such as Alwall [11] in Sweden, conducted ultrafi ltration
with a stationary drum hemodialysis device. The aim of these studies, however,
was to remove excess body water and was not aimed at the removal of uremic
toxins. As such, no replacement fl uid was used, and the potential for convec-
tive blood cleansing was therefore sharply limited. In the mid-1940s, Kolff and
Alwall opened the era of hemodialysis treatment for acute renal failure. The era
of hemodialysis for ESRD was introduced by Quinton, Dillard and Scribner
with their introduction of the arteriovenous shunt in 1960 [12]. This era persists
today.
Gestation: Examination of More Recent Work
The next deliberate use of ultrafi ltration to mimic glomerular blood cleans-
ing in patients with ESRD was reported by myself, Bluemle and colleagues
in 1967 [13]. My interest in convective transport came from work with the
peritoneal membrane. Here, ultrafi ltration with hypertonic glucose solutions
clearly resulted in the transport of urea and other small-molecular-weight
plasma water constituents into the peritoneal cavity [14]. I was not at this time
aware of the prior work by Malinow and Kurzon [9]. I was sharing the labora-
tory of my mentor and colleague, Lewis W. Bluemle Jr., at the University of
Pennsylvania. Bluemle held a contract with a division of the National Institutes
of Health, the National Institute of Arthritis and Metabolic Diseases (NIAMD),
to explore novel membranes for hemodialysis [5]. At that time, Alan Michaels,
the President of the then recently founded Amicon Corporation, came to
Philadelphia to discuss with Bluemle and myself a new series of membranes
that his new corporation was able to produce. The result was a 10-year col-
laboration involving funding from the NIAMD and joint research between the
University of Pennsylvania laboratories (Henderson and colleagues), Amicon
(notably Cheryl Ford and Michael Lysaght) and the Massachusetts Institute of
Technology (Clark Colton).
Noteworthy parallel investigative activities using convective transport
came from Dorson and Markovitz [15], who reported an in vitro pulsatile ultra-
fi ltration system in 1968, and of course the elegant clinical studies of Edward
Quellhorst in 1972 [16] using the Hospal AN69 membrane.
Birth: Nomenclature, Membranes and Equipment
The fi rst name applied to the convective technology aimed at cleansing
uremic blood was ‘diafi ltration’ in 1967. The name appeared to acknowl-
edge the potential roles of both diffusion (dialysis) and convection (fi ltra-
tion). However, 6 years later (1973), Ben Burton of the NIAMD suggested
that ‘hemofi ltration’ was a more apt designation, as solute kinetic studies
had shown little in the way of diffusive transport with this technique. The
name stuck. The work reported from the University of Pennsylvania was all
conducted in the predilution mode, because bench studies with postdilution
were interpreted to be less fl exible in application, owing to the limitations
Henderson 2
The Birth of Hemodiafi ltration 3
imposed by high fi ltration fractions and the lower effi ciency of removal of
toxic solutes loosely bound to protein [17]. It should be noted with appro-
priate respect that W. Leber, working in Giessen, Germany, was the fi rst
investigator to deliberately couple diffusion and convection in what is now
termed ‘hemodiafi ltration’, which technique he presented in an unpublished
report at a conference in Gstaad, Switzerland, in 1977. His subsequent
untimely death in a traffi c accident sadly ended his contributions to the con-
vective technologies.
Membranes
Unlike the existing Cuprophan® cellulosic membranes of the day that dom-
inated the fi eld of dialysis, the membranes from Amicon were made from pre-
cipitated polyelectrolyte. These membranes were neutral in charge and could be
cast in a wide range of hydraulic permeabilities. Unlike the homogeneous struc-
ture of Cuprophan, the Amicon membranes were asymmetrical, with a thin 1- to
2-μm skin supported by a thick spongy stroma. The stroma was exceedingly
open and offered no real resistance to fl uid fl ow. This structure may be found in
the polysulfone membranes in common use today. By traditional standards of
dialysis, the Amicon membrane was too thick (200 μm) and presented a signifi -
cant barrier to diffusion [18]. However, the skin was thin and suffi ciently porous
to permit inulin (5,200 Da, the traditional index solute for measuring glomerular
fi ltration rate) to pass through unrestrained with the ultrafi ltered plasma, simi-
lar to the glomerular basement membrane (i.e. a sieving coeffi cient 1.0). It
was, however, far more permeable to water than was the Cuprophan membrane.
Of note was the commercial presence of the dialysis membrane from Rhone
Poulenc (Hospal), which was synthetic in origin (polyacrylonitrile) and quite
comparable in hydraulic permeability to the Amicon polyelectrolyte membrane
[18]. It was tighter in pore structure, sieving inulin at 0.7 rather than unity. The
ready availability of this membrane in Europe boosted the investigative efforts
of Quellhorst and others in exploring convective transport. Interestingly, it car-
ried a net negative surface charge, causing the solute clearance profi le to differ
from that of the neutral membrane from Amicon [19].
Equipment
The very high water permeability of these polyelectrolyte membranes
(200–300 ml/min depending on the transmembrane pressure) made it necessary
to design and build fl uid-cycling equipment of exceptional precision. Figure 1
shows the original paired pump principle employed in the early animal work
and clinical trials. Reciprocating pistons, driven by a common shaft and cam,
presumably rendered the ultrafi ltrate into the same volume of diluting fl uid. This
volumetric approach, however, had problems with degassing the ultrafi ltrate,
Henderson 4
causing unacceptable fl uid balance errors. Subsequently, gravimetrically based
machines proved to be more accurate and are now widely available.
Postpartum: Clinical Research and Adjunctive Technologies
Clinical Research in Europe and the USA
The difference between the regulatory climates of Europe and the USA
was a signifi cant factor affecting the development of hemodiafi ltration. In the
USA in the late 1960s and, more particularly, in the early 1970s, the need for
independent review of experimental procedures involving patients was recog-
nized, and Institutional Review Boards were put in place. Even prior to this
recognition, there was a strong concern held by those of us investigating the
new convective cleansing modality for uremic blood. We feared that some criti-
cal, but as yet unknown, plasma substance of medium molecular weight, that
could be vital to the life and/or well-being of the study subject, would be swept
into the drain with the effl uent ultrafi ltrate. Therefore, when moving from ani-
mal work to the fi rst human study subject, the ultrafi ltration technique was not
Patient
QBo
QBi
QD
Ultrafiltrate
Amicon
Diluting
fluid
reservoir
Mixing
chamber
Waste
Reciprocating
paired
piston pump
Float
Metering
chamber
Variable
vacuum
source
QF
QB
Fig. 1. Diagram of the paired-pump volumetrically based system employed for
the fi rst patient trials. Errors incurred by differential pressures in the pumping chambers
resulted in degassing errors that subsequently led to abandoning this method for a gravi-
metrically based system. QB Blood fl ow rate; i inlet; o outlet; QD dialysate fl ow rate;
QF ultrafi ltration fl ow rate.
The Birth of Hemodiafi ltration 5
fully applied. Rather, a ‘creeping substitution’ protocol was instituted, in which
an initial combined therapy with a small hemofi lter and a standard Travenol
twin-coil dialyzer was applied for 30 min of the 4-hour hemodialysis. Following
a week free of adverse events, the application of the parallel ultrafi ltration
was increased in duration to 60 min, and then to 90 min, etc. As no adverse
events were encountered, we eventually applied a full-scale hemofi lter once per
week in the thrice weekly treatment schedule, but watched for clinical signs of
vital solute defi ciency in whichever way it might manifest itself. This caution,
encouraged by the regulatory climate in the USA, led us to focus on the details
of the experimental technique, resulting in the development of the fundamen-
tal mathematical relationships for hemodiafi ltration [20, 21] and the relatively
slow, deliberate movement to large clinical trials.
Europe, on the other hand, was not constrained by these formal require-
ments. There was a lively discussion at a 1977 meeting in Europe where a
number of investigators interested in the convective technologies participated.
The subject under discussion was the reprocessing of spent dialysate. Upon my
return to the USA, I began a series of experiments with uremic sheep. The fol-
lowing year, I duly reported my results with the REDY cartridge [22], only to
fi nd that a European investigator was reporting his initial clinical experience
with this technique. These patients subsequently displayed crippling osteopathy
which was felt to be due to the high aluminum concentration in the reprocessed
dialysate. This observation was critical to the formulation change of the car-
tridge and to the benefi t of all subsequent users. Furthermore, it was this free-
dom of investigators and expectation of patient study in Europe that permitted
Quellhorst to move directly into the clinical application of hemodiafi ltration
with the Hospal AN69 polyacrylonitrile membrane, reporting brilliant results in
his series of 100 study subjects and their clinical response over a 10-year time
frame [23].
In short, the regulatory climate at the time led to our investigative interest
in transport kinetics while Europe was applying the technique clinically.
Adjunctive Technologies
The birth of hemodiafi ltration spawned several useful technologies that are
worthy of comment. The continuous therapies used to treat acute renal failure
were the result of work by Silverstein et al. [24] and particularly by Kramer
et al. [25]. Online production of sterile pyrogen-free electrolyte solution was
made possible by the availability of the Amicon membrane technology. The
lack of necessity for heat sterilization reduced the cost of using the prepack-
aged replacement solution required for the fi ltration-based technologies [26].
It even provided the potential for using sterile pyrogen-free dialysate in routine
hemodialysis for those who wished to do so. Precise fl uid-monitoring machines
Henderson 6
began to be used as a requirement for performing hemofi ltration. They are now
generally available for hemodialysis as well as for the convective technologies.
Apart from the two original membranes (from Amicon and from Rhone Poulenc)
mentioned above, there are now many high-hydraulic-permeability membranes
composed of either cellulosic or synthetic materials for use in hemodiafi ltration
or other high-fl ux applications.
Infancy
It is noteworthy that, while we are still in the era of maintenance hemo-
dialysis, there are reports of the increasing use of convective technologies.
Europe leads the world in this regard (table 1) [Port F., DOPPS studies I and II,
pers. commun.] and, as reported by Canaud et al. [27], high-volume convective
therapy is associated with a marked (36%) reduction in mortality compared
with conventional hemodialysis. The USA, with its regulatory restrictions
and fi scal constraints on provision of care, now trails badly. Use of high-fl ux
membranes in which passive (internal) hemodiafi ltration occurs is widespread,
even in the USA; however, as noted in the National Institutes of Health study
of dialysis dose and membrane fl ux (HEMO-I), this degree of convective
Table 1. Prevalence of hemodiafi ltration use by coun-
try in DOPPS II and DOPPS III: percent of patients receiving
hemodiafi ltration in the initial prevalent cross-section
Country DOPPS II DOPPS III
n%n%
Australia/New Zealand 513 6.6 421 4.8
Belgium 538 5.6 482 13.1
Canada 601 0.2 523 0.0
France 528 12.9 486 17.1
Germany 571 10.0 558 5.9
Italy 576 19.1 500 16.0
Japan 1,805 0.0 1,826 4.7
Spain 613 2.8 504 6.0
Sweden 547 13.2 542 20.1
UK 565 5.0 263 8.7
USA 2,260 0.5 1,652 0.1
Hemodiafi ltration is defi ned as patients receiving 5 liters
of replacement fl uid during dialysis [28].
The Birth of Hemodiafi ltration 7
transport fails to confer the benefi t of improved survival [28]. Treatment time
as a prescription variable is now under experimental examination (National
Institutes of Health: Frequent Hemodialysis Network Trials, 2007; HEMO-
II). Longer weekly treatment duration and greater frequency of application
offer the broader solute clearance profi le achieved with hemodiafi ltration. I
am advised that high-volume hemodiafi ltration (20 l/treatment), coupled
with increased frequency and/or duration of treatment (e.g. overnight), will be
needed to match the divine prototype in terms of longevity and quality of life
[Hova J.A., pers. commun.].
References
Henderson LW: Present status of hemofi ltration; in Chang TMS (ed): Artifi cial Kidney, Artifi cial
Liver and Artifi cial Cells. New York, Plenum Publishers, 1978, pp 39–46.
Henderson LW: Historical overview and principles of hemodiafi ltration. Dial Transplant
1980;9:220–221.
Henderson LW: The beginning of hemofi ltration; in Schaefer K, Koch K, Quellhorst E, von
Herrath D (eds): Hemofi ltration. Contrib Nephrol. Basel, Karger, 1982, pp 1–19.
Lysaght M: The history of hemofi ltration; in Henderson LW, Quellhorst EA, Baldamus CA,
Lysaght MJ (eds): Hemofi ltration. Berlin, Springer, 1986, pp. 1–13.
Bluemle LW Jr: Early days of hemofi ltration. ASAIO Trans 1987;33:54–56.
Cameron S: History of the Treatment of Renal Failure by Dialysis. Oxford, Oxford University
Press, 2002, pp 239–244.
Henderson LW: The beginning of clinical hemofi ltration: a personal account. ASAIO J
2003;49:513–517.
Brull L: Réalisation de l’ultrafi ltration in vivo. Biol C R 1928;99:105–107.
Malinow MR, Korzon W: An experimental method for obtaining an ultrafi ltrate of the blood. J Lab
Clin Med 1947;32:461–471.
Kolff WJ, Berk HTHJ, Ter Welle M, van der Leg JW, van Dijk EC, van Noordwijk J: The artifi cial
kidney: a dialyser with great area. Acta Med Scand 1944;117:121–134.
Alwall N: Apparatus for dialysis of the blood in vivo. Acta Med Scand 1947;128:317–325.
Quinton WE, Dillard DH, Scribner BH: Cannulation of blood vessels for prolonged hemodialysis.
Trans Am Soc Artif Intern Organs 1960;6:104–111.
Henderson LW, Besarab A, Michaels A, Bluemle LW Jr: Blood purifi cation by ultrafi ltration and
fl uid replacement (diafi ltration). Trans Am Soc Artif Intern Organs 1967;13:216–226.
Henderson LW: Peritoneal ultrafi ltration dialysis: enhanced urea transfer using hypertonic perito-
neal dialysis fl uid. J Clin Invest 1966;45:950–955.
Dorson W, Markovitz M: A pulsating ultrafi ltration artifi cial kidney. Chem Eng Prog Symp Ser
1968;64:85–89.
Quellhorst E, Fernandez E, Scheler F: Treatment of uremia using an ultrafi ltration-fi ltration sys-
tem. Proc Eur Dial Transplant Assoc 1972;9:584–587.
Henderson LW: Pre- vs post-dilution hemofi ltration. Conference on nondialytic management of
uremia. Clin Nephrol 1979;11:120–124.
Colton CK, Lysaght MJ: Membranes for hemodialysis; in Jacobs C, Kjellstrand CM, Koch KM,
Winchester JF (eds): Replacement of Renal Function by Dialysis. New York, Kluwer Academic
Publishers, 1966, pp 103–113.
Leypoldt JK, Frigon RP, Henderson LW: Macromolecular charge effects hemofi lter solute sieving.
Trans Am Soc Artif Intern Organs 1986;32:384–387.
Colton CK, Henderson LW, Ford CA, Lysaght MJ: Kinetics of hemodiafi ltration. I. In vitro trans-
port characteristics of a hollow-fi ber blood ultrafi lter. J Lab Clin Med 1975;85:355–371.
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
Henderson 8
Henderson LW, Colton CK, Ford C: Kinetics of hemodiafi ltration. II. Clinical characteristics of a
new blood cleansing modality. J Lab Clin Med 1975;85:372–391.
Henderson LW, Parker HR, Schroeder JP, Frigon R, Sanfelippo ML: Continuous low fl ow hemofi l-
tration with sorbent regeneration of ultrafi ltrate. Trans Am Soc Artif Intern Organs 1978;24:178–
184.
Quellhorst E: Long term follow up in chronic hemofi ltration. Int J Artif Organs 1983;6:115–120.
Silverstein ME, Ford CA, Lysaght MJ, Henderson LW: Treatment of severe fl uid overload by ultra-
fi ltration. N Engl J Med 1974;291:747–751.
Kramer P, Wigger W, Rieger J, Mathaei D, Scheler F: Arteriovenous hemofi ltration: a new and
simple method for the treatment of overhydrated patients resistant to diuretics. Klin Wochenschr
1977;55:1121.
Henderson LW, Beans E: Successful production of sterile pyrogen-free electrolyte solution by
ultrafi ltration. Kidney Int 1978;14:522–525.
Canaud B, Bragg-Gresham JL, Marshall MR, Desmueles S, Gillespie BW, Depner T, Klassen
P, Port F: Mortality risk for patients receiving hemodiafi ltration versus hemodialysis: European
results from the DOPPS. Kidney Int 2006;69:2087–2093.
Eknoyen G, Beck GJ, Cheung AK, Daugirdas JT, Greene T, Kusek JW, Allon M, Bailey J, Delmez
JA, Depner TA, Dwyer JT, Levey AS, Levin NW, Milford E, Ornt DB, Rocco MV, Schulman G,
Schwab ST, Teehan BP, Toto R: Effect of dialysis dose and membrane fl ux in maintenance hemo-
dialysis. N Engl J Med 2002;347:2010–2019.
Lee W. Henderson, MD, FACP
480 Clapboard Hill Road
Guilford, CT 06437 (USA)
Tel. 1 203 458 2847, E-Mail hndrsnlee@aol.com
21
22
23
24
25
26
27
28
Ronco C, Canaud B, Aljama P (eds): Hemodiafi ltration.
Contrib Nephrol. Basel, Karger, 2007, vol 158, pp 9–19
Evolution of Hemodiafiltration
Claudio Ronco
Department of Nephrology, St. Bortolo Hospital, Vicenza, Italy
Abstract
The evolution of hemodiafi ltration (HDF) has become possible by the advances in the
construction of dialysis membranes. The availability of a high-fl ux membrane, partially
hydrophilic with high sieving coeffi cients and a reduced wall thickness has made it possible
to conveniently combine diffusion and convection for blood purifi cation. The second impor-
tant step has been the development of accurate ultrafi ltration control systems. Machines
capable of managing large amounts of fl uid turnover allowed safe and effective HDF. Some
new dialysis machines are specifi cally designed to perform HDF with adequate embedded
software. The third step regards the production of large amounts of ultrapure dialysate and
replacement fl uid. For years replacement fl uid was produced in bags, while recently, online
production has enabled high volume exchanges in HDF. All these advances, together with the
creativity of several groups, have spurred new interest in HDF and have led to the develop-
ment of various HDF techniques, which are successfully applied to uremic patients with
important clinical benefi ts.
Copyright © 2007 S. Karger AG, Basel
In 1967, Henderson et al. [1] published the rationale for a new blood-cleans-
ing modality based on ultrafi ltration and fl uid replacement: they defi ned the new
method ‘diafi ltration’. Years later, thanks to the contribution of Clark Colton and
other collaborators, the same senior author published a systematic characteriza-
tion of the new blood purifi cation modality [2]. The new term ‘hemodiafi ltration’
(HDF) was however used for the fi rst time by Leber et al. [3] in Germany. The
combination of diffusion and convection to achieve an effi cient and adequate
blood purifi cation was proposed as a new method full of potential for blood puri-
fi cation [4]. From that moment, a series of developments took place until the
modern days. These developments represent the evolution of HDF from its origi-
nal conception to the most updated and recent modalities of its application.
As in its original description, HDF is an extracorporeal renal replace-
ment technique utilizing a highly permeable membrane, in which diffusion and
Ronco 10
convection are conveniently combined to enhance solute removal in a wide
spectrum of molecular weights. Ultrafi ltration exceeds the desired fl uid loss in
the patient, and therefore replacement fl uid must be administered to achieve the
target fl uid balance. Different modalities of HDF are applied in clinical practice
today; however, the common denominator of such a therapy is the combination
of diffusion and convection to achieve solute and volume control.
Evolution of Hemodiafiltration Technology
Dialysis in its beginning was mostly a diffusive process. The possibility of
using convection in extracorporeal therapies was substantially limited by the
low hydraulic permeability of cellulosic membranes. In fact, Cuprophan®, like
all cellulosic membranes, was considerably hydrophilic with a wall thickness
in the range of 10–20 m and therefore good for diffusion, but its porosity was
insuffi cient to provide high water fl uxes and elevated sieving coeffi cients.
On the other hand, original synthetic fi bers had an internal skin layer sur-
rounded by a microporous structure with a total thickness of 75–100 m. The
polymer was hydrophobic, and its hydraulic permeability and sieving capacity
were high and thus adequate for convection, but the effi ciency in diffusion was
poor due to the marked thickness. Thus high-fl ux membranes were only used for
hemofi ltration with consequent limitations in terms of small-solute clearance. A
further and decisive evolution in the fi eld of membranes has been represented by
the new generation of synthetic polymers with a combined hydrophilic-hydro-
phobic structure and a reduced wall thickness. Such membranes are constituted
of polyethersulfone, polyamide, polymethylmethacrylate, polyacrylonitrile and
other polymers mixed in various proportions with hydrophilic compounds such
as polyvinylpyrrolidone. Thus, only after the advent of these membranes could
the use of therapies such as HDF be adequately developed with an optimization
of the combination of diffusion and convection [5].
The fi rst convective therapy made commercially available was performed
with a closed dialysate delivery system called Rhodial 75. This system was
instrumental in supporting the theory of middle molecules [6, 7]. The system
had to rely on a closed tank of dialysate to overcome the excessive ultrafi l-
tration provided by the fl at polyacrylonitrile RP6 dialyzer. Subsequently, the
technology for controlling ultrafi ltration was developed in conjunction with the
evolution of hemofi ltration and the expanded production of high-fl ux dialyzers
[8]. The real advance was obtained when ultrafi ltration control systems were
routinely applied to standard dialysis machines. In those years, closed and open
(single-pass) volumetric systems were developed together with other approaches
such as gravimetric or fl uximetric ultrafi ltration control mechanisms. The main
Evolution of HDF 11
difference between two widely used approaches was that in one case the system
was acting as fl ow equalizer and a separate pump was providing net ultrafi ltra-
tion; in the other case, the system was a differential fl ow-measuring device and
the difference between the outlet and inlet fl ow was governing transmembrane
pressure in the fi lter. Once the problem of ultrafi ltration control had been solved,
it became evident that a balancing system for fl uid replacement was needed to
perform a safe and accurate HDF. Thus, machines started to be equipped with
specifi c balancing systems to manage up to 9 liters of fl uid reinfusion and up to
15 liters of ultrafi ltration per session. Different companies implemented various
strategies to solve the technicality of HDF, and the treatment became widely
available even to those centers where partial access to technology had previ-
ously limited the implementation of HDF. A further step in development was
then made when online fi ltration of fresh dialysis fl uid through a redundant
fi ltration cascade made large quantities of ultrapure fl uid available, ready to be
reinfused in the patient during HDF.
Until few years ago, HDF was performed using replacement fl uids in bags.
Originally fl uids contained acetate or lactate while only recently could a bicar-
bonate-based solution be utilized. A signifi cant advance in the fi eld of HDF
was made when biofi ltration (a soft form of HDF) was launched on the mar-
ket. The system provided circuits, lines and a 3-liter bag in a single package
making HDF a ready-to-use treatment. Subsequently, as the confi dence of the
operators became greater, larger surface area dialyzers and larger amounts of
fl uid were made available in the package leading to a standardization around
9 liters of reinfusion. A further advance was made when acetate was completely
eliminated from the solution as in the case of acetate-free biofi ltration. Further
modifi cations were suggested in the replacement fl uid composition and even a
complete regeneration of the endogenous ultrafi ltrate by sorbents was proposed
and clinically utilized in a 2-chamber HDF technique called paired fi ltration
dialysis/HDF with endogenous reinfusion (HFR).
The advances made in the fi eld of membranes, machines and solutions was
paralleled by a signifi cant improvement of the dedicated software so that HDF
became an easy procedure, automatically managed by the machine and easily
controlled by nurses through a friendly user interface [7–13]. The typical and
most recent example for this is the modern Fresenius 5008 machine which has a
built-in design specifi cally oriented towards HDF and online HDF [14].
Evolution of Hemodiafiltration Techniques
As previously stated, HDF is a combination of hemodialysis and hemofi l-
tration. The combination of convection with a high-fl ux thin membrane allows
Ronco 12
for the simultaneous removal of small and large solutes. The relative contribu-
tion of convection to the overall solute removal increases progressively with
the increasing molecular weight. The above concepts have been implemented
in clinical practice leading to different forms of HDF. A fl ow chart of the HDF
evolution into different techniques is reported in fi gure 1 [8, 9].
The following lines intend to describe a series of techniques utilized over
the years to perform HDF; the layout of the relevant circuits is schematically
reported in fi gure 2.
Classic Hemodiafiltration
This technique is based on an average reinfusion rate of 9 l/session typi-
cally in postdilution. Fluids are contained in commercial bags. A blood fl ow
higher than 300 ml/min is required to allow the production of suffi cient rates of
ultrafi ltration at acceptable transmembrane pressure regimes. The equipment
requires an ultrafi ltration control system and a reinfusion pump operated by a
HDF
Classic
(9 liters exchange)
HFR
(Charcoal resin)
Online HDF
Classic
Biofiltration
Soft
(3–6 liters exchange)
Hard
(15–21 liters exchange)
AFB
PFD
Double high-flux HDF
Push-Pull HDF
Internal HDF
PHF
Mid-dilution HDF
Fig. 1. HDF evolution tree: the evolution of HDF from the original technique to the
different modifi cations. AFB Acetate-free biofi ltration; PFD paired fi ltration dialysis;
PHF paired HDF.
Fig. 2. The different HDF techniques. A Arterial port; V venous port; R replace-
ment fl uid; UF ultrafi ltrate; UFC UF control; Do dialysate outlet; Di dialysate inlet;
fi ltr. fi ltration; P1 pump 1; P2 pump 2. a Classic HDF. b Soft HDF or biofi ltration. c Hard
(high-volume) HDF. d Online HDF. e Internal HDF (high-fl ux dialysis). f Paired fi ltration
dialysis. g HDF with HFR. h Mid-dilution HDF. i Double high-fl ux HDF. j Push-pull HDF.
Evolution of HDF 13
DiDo UF
UFC
V
A
DiDo
UF UFC
A
VR
R9 liters
V
A
DiDo
UF UFC
R9 liters
V
A
UF DiDo
UFC
V
A
UF DiDo
UFC
R3 liters
V
A
DiDo
UF UFC
R15 liters
V
A
DiDo
UF UFC
V
DiDo
A
UF
Filtr. 1
Filtr. 2
V
A
Di
Do
UF
A1
DiDo
V
A
UF UFC
P1
P1P2
P2
a
b
c
d
e
f
g
h
i
j
Ronco 14
scale continuously weighing reinfusion bags. This technique has been used for
many years before online modalities were available. In some cases, the amount
of reinfusion was as low as 3 l/session (soft HDF) as in the case of biofi ltration
(a ready-to-use proprietary technique from Hospal, France) or up to 15 l/session
as in the case of the so-called hard HDF [8].
Acetate-Free Biofiltration
This is a special form of HDF in which even small traces of acetate were
completely eliminated both from dialysate and replacement fl uid. The average
amount of replacement was normally titrated based on the bicarbonate level of
the patient and varied between 6 and 9 l/session [10]. This technique is derived
by the soft HDF called biofi ltration in which only 3 liters of convection were
scheduled. Later, the technique became widely diffused and it became known
for the improved hemodynamic tolerance and its effects on treatment effi ciency
[11]. A special software was implemented to guide the dialysis session making
specifi c profi les of ultrafi ltration rate and sodium concentration in the dialysate,
not blindly as in the past, but driven by signals coming from the patient and
integrated in a biofeedback software technology.
High-Volume Hemodiafiltration
This form of HDF consists in a classic HDF in which however the amount
of fl uid exchange is 15 l/session or more. Because of the high ultrafi ltration rate,
high blood fl ows are required and replacement solution is sometimes infused in
predilution mode. This partially decreases the effi ciency of the therapy although
it allows for an optimal blood fl ow distribution in the hemodialyzer and a lower
protein concentration polarization and the blood membrane interface [12].
Online Hemodiafiltration
The high cost of commercially prepared fl uids in bags and the improved
technology of dialysate preparation and fl uid fi ltration has allowed it in recent
years to develop a novel technique called ‘online HDF’ [13]. In this technique,
a certain amount of freshly prepared ultrapure dialysate is taken from the dialy-
sate inlet line and processed with multiple fi ltration steps before being used as a
replacement fl uid. With such a technique, large amounts of inexpensive replace-
ment solution are made available and HDF can be carried out with a very high
fl uid turnover (up to 30–40 l/session) utilizing pre- and postdilution or even
simultaneous pre-/postdilution in different proportions. Specifi c adjustments
had to be made in the past generation of machines, whilst the latter machines
are conceived to perform online HDF with adequate embedded software and
system controls [14]. The quality of the reinfusion fl uid is excellent, and it is
guaranteed by the redundancy of the fi ltration system.
Evolution of HDF 15
Internal-Filtration Hemodiafiltration
There is a common belief that when a high-fl ux hemodialyzer is used with
a minimal net ultrafi ltration, the process is purely diffusive as in the case of
low-fl ux dialyzers. This is an incorrect statement, and the treatment features
high amounts of convection although this is masked by the internal kinetics of
crossfi ltration. As blood moves into the dialyzer, the pressure regime in local
regions along the length of the fi bers continuously changes. As a consequence,
water fl uxes in hollow-fi ber hemodialyzers are characterized by a postitive
transmembrane pressure and direct fi ltration in the regions near the arterial inlet
and a negative transmembrane pressure and reverse fi ltration (backfi ltration) in
the region near the venous outlet. In some circumstances this mechanism can
be enhanced: by applying a constriction in the middle of the fi ber bundle, by
operating an obstruction to dialysate fl ow in the dialysate compartment or by
reducing the inner diameter of the fi bers, internal fi ltration can reach values of
40–50 ml/min with a 1.8-m2 dialyzer operated at zero net fi ltration [15, 16]. As
hydraulic conditions tend to favor an increase in ultrafi ltration, the ultrafi ltra-
tion control system of the machine controls the net fl uid balance by modifying
the pressures in the system thus increasing the relative amount of backfi ltration.
Although this in general is defi ned high-fl ux dialysis, when specifi c measures
or dialyzer designs are applied to enhance the mechanism of fi ltration/backfi l-
tration, this form of therapy can be defi ned as internal HDF [17].
Paired Filtration Dialysis
This technique of HDF has been conceived in Italy for the fi rst time [18].
It is based on two fi lters placed in series: the fi rst is a hemofi lter and removes
fl uid and solutes by convection; the second is a hemodialyzer in which diffusion
is prevalently utilized. Replacement fl uid is infused in between the two units.
The meaning of this therapy is to minimize the interactions between convection
and diffusion, to prevent backfi ltration in the hemodialyzer and fi nally to make
ultrafi ltrate available for online measurements as a surrogate of plasma water
[19]. Recently further evolution of this concept has led to the HFR and paired
HDF techniques. HFR is a paired fi ltration dialysis where the ultrafi ltrate pro-
duced is purifi ed by adsorption through a resin/charcoal unit and utilized sub-
sequently as a replacement fl uid [20]. The paired HDF is another modifi cation
of the technique in which the fi st unit is used to backfi lter some fresh dialysate
acting as ultrapure online fi ltered replacement fl uid [21].
Mid-Dilution Hemodiafiltration
This technique is made possible by the use of special fi lters with two longi-
tudinal compartments in series. Blood fl ows in the fi rst compartment producing
a certain amount of ultrafi ltration; at the end of the compartment, instead of
Ronco 16
having a venous port, there is a chamber designed to receive the replacement
fl uid infusion. In this chamber, the hemoconcentrated blood – due to the ultra-
fi ltration process – is reconstituted to the original dilution and it is redirected
countercurrently in the second blood compartment. Blood then leaves the dia-
lyzer beside the arterial entry. Dialysate in this system fl ows 50% countercur-
rently to blood and 50% cocurrently [22].
Double High-Flux Hemodiafiltration
This technique utilizes two high-fl ux dialyzers in series. Filtration takes
place in the proximal unit while backfi ltration takes place in the distal unit.
High blood fl ows (500 ml/min) have been utilized for this technique, and its
high effi ciency has allowed treatments under 2 h/session [23].
Push-Pull Hemodiafiltration
This technique utilizes the mechanism of fi ltration and backfi ltration alter-
nating the rotation of a prefi lter pump (while the postfi lter pump is stopped) pro-
ducing fi ltration, and the rotation of the postfi lter pump (while the prefi lter pump
is stopped) producing a negative pressure in the blood compartment and thus
backfi ltration [24]. This technique can also be operated by a similar alternate
fl ow regime in the dialysate compartment instead of the blood compartment.
Evolution of Dialysate Filtration Techniques
The use of high-fl ux dialyzers with a high surface area and large amounts
of crossfi ltration requires a careful analysis of the quality of every fl uid involved
in the process. Thus, water treatment systems, dialysate concentrates and dialy-
sate fi ltration procedures for online replacement fl uid production must be con-
tinuously controlled and screened for quality. Online preparation of sterile and
pyrogen-free solutions for infusion during HDF is based on the use of water and
concentrates that are ultrapure and are mixed and distributed in a well-designed
and maintained fl ow path [25]. Ultrafi lters with known retention capacity are
placed in strategic positions and dimensioned to remove bacteria and endotox-
ins, which gives a satisfactory purity assurance level. Microbiological safety of
online HDF has been shown in long-term clinical studies, with a notable absence
of cytokine-inducing activity and pyrogenic reactions in patients despite the
infusion of large volumes of fl uid in every session [26, 27].
The combination of biocompatible membranes and ultrapure fl uid is benefi -
cial in reducing bioactivation of circulating leukocytes induced by blood-hemo-
dialyzer interaction. The high ultrafi ltration rates also result in protein-coating
of the membrane, rendering it even more biocompatible.
Evolution of HDF 17
Arteriovenous Hemodiafiltration and Continuous Venovenous
Hemodiafiltration for Acute Cases
HDF was fi rstly described in acute cases in 1985 [28]. In those times,
hemofi lters were designed with a single port for ultrafi ltrate, and the Vicenza
group suggested a modifi cation by adding a second port to the Amicon fi lters
so that dialysate fl uid could be circulated countercurrently to blood. This fi rst
description was characterized by an arteriovenous circuit with no pumps. At the
same time, dialysate was infused and drained from the fi lter by gravity. Only
subsequently pumped circuits started to be used in the acute settings and con-
tinuous venovenous hemofi ltration/HDF started to appear in the clinical arena.
In particular continuous venovenous HDF was made possible by the advent of
a machine specifi c for continuous renal replacement therapy (Prisma) featuring
4 pumps and thus allowing a separate management of dialysate fl ow and ultra-
fi ltration/reinfusion fl ow. Today all machines are equipped with circuitry and
software that allow a simple and safe management of continuous HDF [29].
Conclusions
The evolution of technology for HDF has made this technique simpler
and safer. Furthermore, modifi cations of the original basic HDF layout have
allowed to explore new treatment modalities that however retain the basic prin-
ciples of combining diffusion and convection. The emerging evidence that
these therapies may be superior to classic hemodialysis in terms of morbid-
ity and mortality further encourages the progress of application of HDF and
derived techniques. We think that this represents a classic example of transla-
tional research where a continuous interaction between bench experiments and
clinical testings is paralleled by an intelligent merging of clinical questions and
technological responses.
References
1 Henderson LW, Besarab A, Michaels A, Bluemle LW Jr: Blood purifi cation by UF and fl uid
replacement (diafi ltration). Trans Am Soc Artif Intern Organs 1967;17:216–221.
2 Henderson LW, Colton CK, Ford C: Kinetics of HDF. II. Clinical characterization of a new blood
cleansing modality. J Lab Clin Med 1975;85:372–375.
3 Leber H, Wizemann V, Goubeaud G, Rawer P, Schutterle G: Hemodiafi ltration: a new alternative
to hemofi ltration and conventional hemodialysis. Artif Organs 1978;2:150–153.
4 Henderson LW: Biophysics of UF and hemofi ltration; in Maher JF (ed): Replacement of Renal
Function by Dialysis: A Textbook of Dialysis, ed 3. Dordrecht, Kluwer Academic Publishers,
1989, pp 300–326.
Ronco 18
5 Hoenich N, Ronco C: Dialyzer evaluation; in Winchester J, Koch K, Jacobs C, Kjiellestrand C
(eds): Replacement of Renal Function by Dialysis. Dordrecht, Kluwer Academic Publishers, 1996,
pp 256–270.
6 Babb AL, Farrel PC, Uvelli DA, Scribner BH: Hemodialyzer evaluation by examination of solute
molecular spectra. Trans Am Soc Artif Intern Organs 1972;18:98–105.
7 Babb AL, Strand MJ, Uvelli DA, et al: Quantitative description of dialysis treatment: a dialysis
index. Kidney Int 1975;7(suppl 2):23–28.
8 Ronco C, La Greca G: The role of technology in hemodialysis; in Ronco C, La Greca G (eds):
Hemodialysis Technology. Contrib Nephrol. Basel, Karger, 2002, vol 137, pp 1–12.
9 Maduell F: Hemodiafi ltration. Hemodial Int 2005;9:47–55.
10 Pacitti A, Casino FG, Pedrini L, Santoro A, Atti M: Prescription and surveillance of the acetate-
free biofi ltration sessions: the bicarbonate cycle. Int J Artif Organs 1995;18:722–725.
11 Ronco C, Brendolan A, Milan M, Rodighiero MP, Zanella M, La Greca G: Impact of biofeedback-
induced cardiovascular stability on hemodialysis tolerance and effi ciency. Kidney Int 2000;58:
800–808.
12 Pedrini LA, Cozzi G, Faranna P, Mercieri A, Ruggiero P, Zerbi S, Feliciani A, Riva A:
Transmembrane pressure modulation in high-volume mixed hemodiafi ltration to optimize effi -
ciency and minimize protein loss. Kidney Int 2006;69:573–579.
13 Canaud B, Levesque R, Krieter D, Desmeules S, Chalabi L, Moragues H, Morena M, Cristol JP:
On-line hemodiafi ltration as routine treatment of end-stage renal failure: why pre- or mixed dilu-
tion mode is necessary in on-line hemodiafi ltration today? Blood Purif 2004;22(suppl 2):40–48.
14 Ronco C, Bowry S, Tetta C: Dialysis patients and cardiovascular problems: can technology help
solve the complex equation? Blood Purif 2006;24:39–45.
15 Ronco C, Brendolan A, Feriani M, Milan M, Conz P, Lupi A, Berto P, Bettini MC, La Greca G: A
new scintigraphic method to characterize ultrafi ltration in hollow fi ber dialyzers. Kidney Int 1992;
41:1383–1393.
16 Ronco C, Brendolan A, Lupi A, Bettini MC, La Greca G: Enhancement of convective transport by
internal fi ltration in a modifi ed experimental hemodialyzer. Kidney Int 1998;54:979–985.
17 Fiore GB, Guadagni G, Lupi A, Ricci Z, Ronco C: A new semiempirical mathematical model for
prediction of internal fi ltration in hollow fi ber hemodialyzers. Blood Purif 2006;24:555–568.
18 Botella J, Ghezzi P, Sanz-Moreno C, Milan M, Conz P, La Greca G, Ronco C: Multicentric study
on paired fi ltration dialysis as a short, highly effi cient dialysis technique. Nephrol Dial Transplant
1991;6:715–721.
19 Arrigo G, Tetta C, Santoro A, Ghezzi P, Ronco C, Colasanti G, La Greca G, Zucchelli P, D’Amico
G: Continuous urea monitoring in hemodialysis: a model approach to forecast dialytic perfor-
mance: results of a multicenter study. J Nephrol 2001;14:481–487.
20 Ursino M, Coli L, Magosso E, Capriotti P, Fiorenzi A, Baroni P, Stefoni S: A mathematical model
for the prediction of solute kinetics, osmolarity and fl uid volume changes during hemodiafi ltration
with on-line regeneration of ultrafi ltrate (HFR). Int J Artif Organs 2006;29:1031–1040.
21 Mandolfo S, Corsi A, Wratten ML, Sereni L, Imbasciati E: Evaluation of hygiene and safety con-
trols for on-line paired hemodiafi ltration (PHF). Int J Artif Organs 2006;29:160–165.
22 Santoro A, Conz PA, De Cristofaro V, Acquistapace I, Gaggi R, Ferramosca E, Renaux JL, Rizzioli
E, Wratten ML: Mid-dilution: the perfect balance between convection and diffusion; in Ronco
C, Brendolan A, Levin NW (eds): Cardiovascular Disorders in Hemodialysis. Contrib Nephrol.
Basel, Karger, 2005, vol 149, pp 107–114.
23 Miller JH, von Albertini B, Gardner PW, Shinaberger JH: Technical aspects of high fl ux hemodia-
fi ltration for adequate short (under 2 hours) treatment. Trans Am Soc Artif Intern Organs 1984;
30:377–380.
24 Miwa M, Shinzato T: Push/pull hemodiafi ltration: technical aspects and clinical effectiveness.
Artif Organs 1999;23:1123–1126.
25 Ledebo I: On-line preparation of solutions for dialysis: practical aspects versus safety and regula-
tions. J Am Soc Nephrol 2002;13(suppl 1):S78–S83.
26 Canaud B, Bosc JY, Leray-Moragues H, et al: On-line haemodiafi ltration: safety and effi cacy in
long-term clinical practice. Nephrol Dial Transplant 2000;5(suppl 1):S60–S67.
Evolution of HDF 19
27 Guth HJ, Gruska S, Kraatz G: On-line production of ultrapure substitution fl uid reduces TNF-
alpha- and IL-6 release in patients on hemodiafi ltration therapy. Int J Artif Organs 2003;26:181–
187.
28 Ronco C, Bragantini L, Brendolan A, Dell’Aquila R, Fabris A, Chiaramonte S, Feriani M,
Laquaniti L, La Greca G: Arteriovenous hemofi ltration (AVHDF) combined with continuous arte-
riovenuous hemofi ltration (CAVH). Trans Am Soc Artif Intern Organs 1985;31:349–352.
29 Ronco C, Ricci Z, Bellomo R, Baldwin I, Kellum J: Management of fl uid balance in CRRT: a
technical approach. Int J Artif Organs 2005;28:765–776.
Claudio Ronco, MD
Department of Nephrology, St. Bortolo Hospital
IT–36100 Vicenza (Italy)
Tel. 39 0444753650, Fax 39 0444753949, E-Mail cronco@goldnet.it
Ronco C, Canaud B, Aljama P (eds): Hemodiafi ltration.
Contrib Nephrol. Basel, Karger, 2007, vol 158, pp 20–33
Solute Removal by Hollow-Fiber Dialyzers
William R. Clarka,b, Eduardo Rochac, Claudio Roncod
aGambro Inc., Lakewood, Colo., and bNephrology Division, Indiana University
School of Medicine, Indianapolis, Ind., USA; cNephrology Department,
Universidade Federal do Rio de Janeiro, Rio de Janeiro, Brazil;
dNephrology Department, St. Bortolo Hospital, Vicenza, Italy
Abstract
Renal replacement therapy performed for end-stage renal disease patients now occurs
almost exclusively with hollow-fi ber dialyzers. Because renal replacement therapy utilizing
such a device requires goals to be set with regard to the rate and extent of solute removal, a
thorough understanding of the mechanisms by which solute removal occurs is necessary.
This chapter provides an overview of solute removal by hollow-fi ber dialyzers. In the fi rst
part of the chapter, the major characteristics of hollow-fi ber membranes infl uencing solute
removal are discussed. Within this section, the chemical composition and physical character-
istics of commonly used dialysis membranes and the features determining their solute per-
meability properties are reviewed. The remainder of the chapter emphasizes the major
determinants of hollow-fi ber dialyzer performance.
Copyright © 2007 S. Karger AG, Basel
Renal replacement therapy performed for end-stage renal disease patients
now occurs almost exclusively with hollow-fi ber dialyzers. Because renal
replacement therapy utilizing such a device requires goals to be set with regard
to the rate and extent of solute removal, a thorough understanding of the mecha-
nisms by which solute removal occurs is necessary. This chapter provides an
overview of solute removal by hollow-fi ber dialyzers. In the fi rst part of the
chapter, the major characteristics of hollow-fi ber membranes infl uencing sol-
ute removal are discussed. Within this section, the chemical composition and
physical characteristics of commonly used dialysis membranes and the features
determining their solute permeability properties are reviewed. The remainder
of the chapter emphasizes the major determinants of hollow-fi ber dialyzer
performance.
Basic Principles
Dialyzer Solute Transport 21
Hollow-Fiber Membranes: Classification by Material
Cellulosic Membranes
The relatively long duration of popularity of cellulosic membranes can be
explained largely by their particular suitability for a diffusion-based procedure
like hemodialysis (HD) [1]. The underlying hydrogel structure of these mem-
branes and their tensile strength allow the combination of low wall thickness
(see below) and high porosity to be attained in the fi ber spinning process [2].
These characteristics allow the attainment of high rates of diffusive membrane
transport and effi cient removal of small, water-soluble uremic solutes, such as
urea and creatinine [3, 4]. Another characteristic feature of these membranes is
symmetry with respect to composition, implying an essentially uniform resis-
tance to mass transfer over the entire wall thickness.
Synthetic Membranes
Synthetic membranes were developed essentially in response to concerns
related to the narrow scope of solute removal and the pronounced complement
activation associated with unmodifi ed cellulosic dialyzers. The AN69® mem-
brane, a copolymer of acrylonitrile and an anionic sulfonate group, was fi rst
employed in fl at sheet form in a closed-loop dialysate system in the early 1970s
[5]. Since that time, a number of other synthetic membranes have been devel-
oped, including polysulfone [6], polyamide [7], polymethylmethacrylate [8],
polyethersulfone [9] and polyarylethersulfone/polyamide [10]. Largely related
to the interest in hemofi ltration as an end-stage renal disease therapy in the late
1970s and early in the following decade, along with the inability to use low-fl ux
unmodifi ed cellulosic dialyzers for this therapy, these membranes were initially
formulated with high water permeability [11]. The large mean pore size and
thick wall structure of these membranes allowed the high ultrafi ltration rates
necessary in hemofi ltration to be achieved at relatively low transmembrane
pressures. However, with the waning of interest in hemofi ltration as a chronic
dialysis therapy in the mid-1980s, dialyzers with these highly permeable mem-
branes were used subsequently in the diffusive mode as high-fl ux dialyzers. This
latter mode continues to be the most common application of these membranes,
although they are increasingly being employed for chronic hemodiafi ltration
now [12]. Synthetic membranes have wall thicknesses of at least 20 m and
may be structurally symmetric (e.g. AN69, polymethylmethacrylate) or asym-
metric (e.g. polysulfone, polyamide, polyethersulfone, polyamide/polyaryle-
thersulfone). In the latter category, a very thin ‘skin’ (approx. 1 m) contacting
the blood compartment lumen acts primarily as the membrane’s separative ele-
ment with regard to solute removal.
Clark/Rocha/Ronco 22
Properties of Hemodialyzer Membranes Influencing Solute Removal
Membrane Diffusive Permeability
Membrane wall thickness is one important determinant of diffusive trans-
port [13]. The relatively thin-walled structure of cellulosic membranes (usu-
ally 6–15 m) is largely responsible for their particular suitability in the setting
of diffusive HD. The other major determinant of dialyzer membrane diffusive
transport is porosity, also known as pore density. Based on the cylindrical pore
model described above, membrane porosity is directly proportional to both the
number of pores and the square of the pore radius (r2). Therefore, the smaller
dependence of membrane porosity on pore size, relative to the case of water
permeability, implies a relatively greater importance of pore number in deter-
mining diffusive permeability. That the major determinants of fl ux (r4) and
diffusive permeability (number of pores, r2 and wall thickness) differ so sig-
nifi cantly implies that the two properties can be independent of one another for
a particular HD membrane. Such is the case for cellulosic high-effi ciency dia-
lyzers, which typically have very high diffusive permeability values for small
solutes but low water permeability.
Nondiffusive Membrane Considerations
A membrane represented by the cylindrical pore model described above
deviates from an actual membrane used for clinical HD in that the latter actu-
ally has a distribution of pore sizes. Ronco et al. [4] have recently discussed the
manner in which pore size distribution may differ among HD membranes and
the manner in which this distribution infl uences a membrane’s sieving prop-
erties (fi g. 1). The membrane represented by curve A has a large number of
relatively small pores while the membrane represented by curve B has a large
number of relatively large pores. Based on the relatively narrow pore size distri-
butions, the solute sieving coeffi cient versus molecular weight profi le for both
membranes has the desirable sharp cut-off, similar to that of the native kidney.
However, the molecular-weight cut-off for membrane A (approx. 10 kDa) is
consistent with a high-effi ciency membrane while that of membrane B (approx.
60 kDa) is consistent with a high-fl ux membrane. In addition, primarily due to
the large number of pores, both membranes would be expected to demonstrate
favorable diffusive transport properties. On the other hand, membrane C exhib-
its a pore size distribution that is unfavorable from both a diffusive transport
and sieving perspective. The relatively small number of pores accounts for the
poor diffusive properties. In addition, the broad distribution of pores explains
not only the ‘early’ drop-off in sieving coeffi cient at relatively low molecular
weight but also the ‘tail’ effect at high molecular weight. This latter phenom-
enon is highly undesirable as it may lead to unacceptably high albumin losses
Dialyzer Solute Transport 23
across the membrane. In actual practice, all highly permeable membranes have
measurable albumin sieving coeffi cient values such that the design of this type
of membrane involves striking a balance between optimized large-molecular-
weight toxin removal and minimal albumin losses.
Another convection-related mechanism by which large uremic toxins can
be removed relates to fl uid fl ow within the fi lter. Under normal operating con-
ditions of high-fl ux dialysis, the large axial pressure drop that occurs in such
highly permeable membranes typically results in pressures in a certain portion of
the distal (venous) end of the fi ber that are less than the corresponding dialysate
compartment pressure. This results in the routine occurrence of backfi ltration
of dialysate during high-fl ux HD [15]. Although the combination of signifi cant
backfi ltration and contaminated dialysate raises concerns related to ‘backtrans-
fer’, this internal fi ltration (‘Starling’s fl ow’) mechanism can signifi cantly aug-
ment removal of larger molecules. In fact, under normal operating conditions of
high-fl ux HD, this mechanism is typically the predominant mechanism by which
large-solute removal occurs. Attempts at accentuating this internal fi ltration
mechanism, either through a decrease in hollow-fi ber inner diameter or manipu-
lations in dialysate compartment pressure, have been described recently [16].
Adsorption (membrane binding) is another mechanism by which hydro-
phobic compounds like peptides and proteins may be removed during HD.
Although adsorption during HD is a relatively poorly understood phenom-
enon, certain membrane characteristics play an important role. First, adsorp-
tion primarily occurs within the pore structure of the membrane rather than at
the nominal surface contacting the blood only [17]. Therefore, the open pore
0
109
103104105
0 102030405060
Pores (n/cm2)
Average pore diameter (Å) Solute molecular weight (Da)
Sieving coefficient
0
A
AB
B
C
C0.2
0.4
0.6
0.8
1.0
ab
Fig. 1. Pore size distribution and sieving coeffi cient profi les for three hypothetical mem-
branes A, B and C. a Relationship between number of pores and pore size. b Sieving coeffi cient
as a function of solute molecular weight. Reprinted with permission from Ronco et al. [14].
Clark/Rocha/Ronco 24
structure of high-fl ux membranes affords more adsorptive potential than do
low-fl ux counterparts. Second, synthetic membranes, many of which are fun-
damentally hydrophobic, are generally much more adsorptive than hydrophilic
cellulosic membranes [18].
Characterization of Dialyzer Performance: The Concept of
Solute Clearance
Whole-Blood Clearance
By defi nition [19], solute clearance (K) is the ratio of mass removal rate
(N) to blood solute concentration (CB):
K N/CB. (1)
For a hemodialyzer, mass removal rate is simply the difference between the
rate of solute mass (i.e. product of fl ow rate and concentration) presented to the
dialyzer in the arterial blood line and the rate of solute mass leaving the dialyzer
in the venous blood line. This mass balance applied to the dialyzer results in the
classical (i.e. arteriovenous) whole-blood dialyzer clearance equation [20]:
KB [(QBi C Bi) (QBo C Bo)]/CBi QF (CBo/CBi) (2)
In this equation, KB is whole-blood clearance, QB is blood fl ow rate, CB is whole-
blood solute concentration, and QF is net ultrafi ltration rate. [The subscripts ‘i’
and ‘o’ refer to the inlet (arterial) and outlet (venous) blood lines.]
It is important to note that diffusive, convective and possibly adsorptive solute
removals occur simultaneously in HD. For a nonadsorbing solute like urea, diffu-
sion and convection interact in such a manner that total solute removal is signifi -
cantly less than what is expected if the individual components are simply added
together. This phenomenon is explained in the following way. Diffusive removal
results in a decrease in solute concentration in the blood compartment along the
axial length (i.e. from blood inlet to blood outlet) of the hemodialyzer. As convec-
tive solute removal is directly proportional to the blood compartment concentra-
tion, convective solute removal decreases as a function of this axial concentration
gradient. On the other hand, hemoconcentration resulting from ultrafi ltration of
plasma water causes a progressive increase in plasma protein concentration and
hematocrit (Hct) along the axial length of the dialyzer. This hemoconcentration
and resultant hyperviscosity cause an increase in diffusive mass transfer resis-
tance and a decrease in solute transport by this mechanism. The effect of this
interaction on overall solute removal has been analyzed rigorously by numerous
investigators. The most useful quantifi cation has been developed by Jaffrin [21]:
KT K D Q F Tr (3)
Dialyzer Solute Transport 25
In this equation, KT is total solute clearance, KD is diffusive clearance under
conditions of no net ultrafi ltration, and the fi nal term is the convective compo-
nent of clearance. The latter term is a function of the ultrafi ltration rate (QF) and
an experimentally derived transmittance coeffi cient (Tr), such that:
Tr S (1 K D/QB) (4)
where S is the solute sieving coeffi cient. Thus, Tr for a particular solute is depen-
dent on the effi ciency of diffusive removal. At very low values of KD/QB, diffusion
has a very small impact on blood compartment concentrations and the convective
component of clearance closely approximates the quantity S Q F
. However, with
increasing effi ciency of diffusive removal (i.e. increasing KD/QB), blood compart-
ment concentrations are signifi cantly infl uenced. The result is a decrease in Tr
and, consequently, in the convective contribution to total clearance.
Blood Water and Plasma Clearance
An implicit assumption in the determination of whole-blood clearance is
that the volume from which the solute is cleared is the actual volume of blood
transiting through the dialyzer at a certain time. This assumption is incorrect
for two reasons. First, in both the erythron and plasma components of blood,
a certain volume is comprised of solids (proteins or lipids) rather than water.
Second, for solutes like creatinine and phosphate which are distributed in both
the erythron and plasma water, slow mass transfer from the intracellular space
to the plasma space (relative to mass transfer across the dialyzer) results in
relative sequestration (compartmentalization) in the former compartment [22].
This reduces the effective volume of distribution from which these solutes can
be cleared in the dialyzer. As such, whole-blood dialyzer clearances derived
by using plasma water concentrations in conjunction with blood fl ow rates, a
common practice in dialyzer evaluations, results in a signifi cant overestimation
of actual solute removal. The more appropriate approach is to employ blood
water clearances, which account for the above Hct-dependent effects on effec-
tive intradialyzer solute distribution volume [23]:
QBW 0.93 Q B[1 Hct K(1 e–t)Hct] (5)
where QBW is blood water fl ow rate. In this equation, for a given solute, K is the
red blood cell (RBC) water/plasma water partition coeffi cient for a given sol-
ute, is the transcellular rate constant (units: time–1), and t is the characteristic
dialyzer residence time. Estimates for these parameters have been provided by
numerous prior studies and have been summarized by Shinaberger et al. [24].
(The factor 0.93 in equation 2 corrects for the volume of plasma occupied by
plasma proteins and lipids.) Finally, KBW can be calculated by substituting QBW
for QB in equation 1.
Clark/Rocha/Ronco 26
Although the distribution volume of many uremic solutes approximates
total body water, it is much more limited for other toxins, particularly those
of larger molecular weight. For example, the distribution space of 2-micro-
globulin and many other low-molecular-weight proteins is the plasma volume.
Consequently, when using equation 2 to determine 2-microglobulin clearance,
plasma fl ow rates (inlet and outlet) should replace blood fl ow rates in the fi rst
term of the right-hand side of the equation.
The distinction between whole-blood, blood water and dialysate-side clear-
ances is very important when interpreting clinical data. However, clearances
provided by dialyzer manufacturers are typically in vitro data generated from
experiments in which the blood compartment fl uid is an aqueous solution.
Although these data provide useful information to the clinician, they overes-
timate actual dialyzer performance that can be achieved clinically (under the
same conditions). This overestimation is related to the inability of aqueous-
based experiments to capture the effects of RBCs (see above) and plasma pro-
teins on solute mass transfer.
Clearance versus Mass Removal Rate
It is important to recognize that clearance is not a measure of actual dialytic
mass removal of a particular solute. As equation 1 indicates, clearance is the ratio
of mass removal rate to blood concentration for a given solute. In HD, the mass
removal rate of small solutes like urea is very high during the early stage of an
intermittent HD treatment due to a favorable transmembrane concentration gradi-
ent for diffusion at this time. As the treatment proceeds, a proportional decrease
in the blood urea nitrogen and the urea mass removal rate, which is determined by
the instantaneous blood urea nitrogen, occurs [19]. Equation 1 predicts that a pro-
portional decrease in these parameters results in a constant dialyzer clearance dur-
ing the treatment (provided that dialyzer function is preserved; fi g. 2). Despite not
being a measure of actual dialytic solute removal, clearance remains a very reason-
able parameter to assess dialyzer function. The discordance between solute clear-
ance and mass removal rate described above is a much more relevant consideration
when a whole-body (rather than dialyzer) clearance approach is used (see below).
Determinants of Diffusive Solute Clearance
Diffusion is the dominant mass transfer mechanism mediating small solute
removal in HD. Diffusive solute removal involves sequential mass transfer from
the dialyzer blood compartment, through the membrane, and into the dialysate
compartment. To quantify a dialyzer’s diffusive capabilities, the concept of mass
transfer resistance is frequently employed [25]:
Dialyzer Solute Transport 27
RO R B RM RD (7)
In the above equation, the overall resistance to diffusive mass transfer of a
particular solute (RO) by a dialyzer has three components: blood compartment
resistance (RB), resistance due to the membrane itself (RM) and dialysate com-
partment resistance (RD). In turn, RO is the inverse of the overall mass transfer
coeffi cient (KO), which is a component of the overall mass transfer-area coeffi -
cient (KOA) discussed below. After a discussion of membrane properties above,
the blood and dialysate compartments are discussed below.
Blood Compartment
A fundamental relationship exists between diffusive clearance and blood
fl ow rate for all solutes. For a given solute, a graph of clearance versus QB has
two domains [26]. In the relatively low QB regime, an effectively linear rela-
tionship exists between these two parameters. For all solutes, the line defi ned
by this relationship falls below the line of identity, thus indicating that dialyzer
clearance can never exceed the blood fl ow rate. For a given dialyzer, the slope
of the line defi ning this fl ow-limited regime is inversely related to solute size.
Beyond a certain QB, the curve defi ning the clearance versus QB relationship
for a given solute/dialyzer combination demonstrates a plateau. This plateau
defi nes the KOA-limited region. For a given solute/dialyzer combination, the
KHF urea
(ml/min)
Removal
rate
(mg/min)
Amount
removed
(mg)
1234
Time (h)
Fig. 2. Relationship between solute clearance (KHF), mass removal rate and cumula-
tive removal during a 4-hour HD treatment. Even with constant dialyzer clearance, the mass
removal rate falls during the treatment due to a reduced concentration gradient. Reprinted
with permission from Clark and Henderson [19].
Clark/Rocha/Ronco 28
KOA parameter can be regarded as the maximal clearance attainable under a
given set of fl ow conditions. Both the QB at which the transition from the blood-
fl ow-limited to the KOA-limited region occurs and the plateau clearance value
are specifi c for a given solute/dialyzer combination [26]. For a given solute,
an increase in either membrane diffusivity (KO) or area (A) has the effect of
increasing both the transition QB and the plateau clearance value.
Minimizing the mass transfer resistance in the blood compartment is
achieved primarily by the use of relatively high fl ow rates (i.e. shear rates) that
minimize effects related to boundary (unstirred) layers. A boundary layer can
be conceptualized as a stagnant fi lm of fl uid residing on the membrane surface.
However, another important factor infl uencing blood compartment resistance is
Hct. Blood is a complex fl uid in which RBCs are suspended in plasma. Plasma
is an aqueous-based solution but does have a solid component (approx. 7% by
volume) consisting of proteins and lipids. The erythron is also primarily aque-
ous, with water constituting approximately 70% of the total erythron and the
remaining solid component being comprised primarily of cellular membranes.
Although many uremic solutes are distributed in the aqueous phase of both the
RBC and plasma fractions of blood, solute removal during HD can occur only
from plasma water. Before actual dialytic removal of solutes with this type of
distribution can be achieved, mass transfer from the RBC water to the plasma
water must occur. In turn, the rate at which this latter process occurs is solute
specifi c. Prior data [27] indicate that urea movement across the RBC membrane
is relatively fast. Therefore, during HD, urea in the plasma water leaving the
dialyzer is in equilibrium with urea in the RBC water, with the ratio of these
concentrations (approx. 0.76) being determined by the ratio of the water frac-
tions of the aqueous and RBC compartments. On the other hand, the transcellu-
lar rate of movement for other uremic solutes, such as creatinine and phosphate,
is small (or negligible) relative to the rate of dialytic removal [28]. For a given
unit volume of whole blood, an increase in Hct causes a relative increase in the
distribution of solute in the RBC water, resulting in a relative sequestration of
solutes with low RBC membrane diffusivity.
The application of rheological principles to the fl ow of blood in a dialyzer
also raises concerns that blood compartment mass transfer may be impaired by
increasing Hct. For a given solute, diffusive mass transfer resistance in the blood
compartment of a dialyzer is the ratio of effective diffusive path length to effective
solute diffusivity, both of which may be infl uenced by Hct [28]. As the volume
comprised by the RBC mass per unit volume of blood increases with increas-
ing Hct, solutes diffusing to the membrane surface are relatively more likely to
encounter an RBC, causing an effective lengthening of the diffusion distance. In
addition, solute diffusivity may decrease as a function of increasing Hct due to
the latter’s effect on viscosity, itself a determinant of mass transfer resistance.
Dialyzer Solute Transport 29
Lim et al. [23] studied 5 patients in whom pre-HD Hct was raised from a
mean of 22.9 to 37.8% with the use of erythropoietin. Whole-blood (KB) and
dialysate-side (KD) clearances of urea, creatinine and phosphate were measured
under the following (prescribed) conditions: QB, 400 ml/min; dialysate fl ow
rate (QD), 500 ml/min; treatment time, 180 min. Dialysate-side clearance was
used as the truer (‘gold standard’) estimate of mass removal. The ratio KD/KB,
an estimate of the degree to which whole-blood clearance overestimates mass
removal, was observed to decrease signifi cantly for both creatinine and phos-
phate but not for urea. In urea kinetic analyses (based on the direct dialysate
quantifi cation method), both Kt/V (1.21 vs. 1.17) and percent urea reduction
(64.2 vs. 61.6%) decreased but not signifi cantly. Interestingly, in a separate
group of 7 patients in whom Hct was raised from 19.1 to 29.5% with RBC
transfusions, both Kt/V (1.32 vs. 1.19) and percent reduction (66.4 vs. 62.7%)
decreased signifi cantly.
Data from Ronco et al. [28] suggest that Hct may also infl uence fl ow
distribution within the blood compartment of a dialyzer. These investigators
employed a computerized-tomography-based technique to measure fi ber bundle
perfusion of blood with varying Hct (25–40%). A centralized distribution of
fl ow was observed. Moreover, the extent of this maldistribution was propor-
tional to Hct. In fact, at Hct 40%, the fl ow velocity and wall shear rate were
2- to 3-fold higher in the central region of the bundle than in its peripheral
region. For clinical correlation, these investigators also measured dialyzer urea
and creatinine clearances as a function of Hct. As shown in fi gure 3, this study
corroborated the differential effect of increasing Hct on urea and creatinine
clearance reported by Lim et al. [23].
0
50
100
150
200
250
300
Clearance (ml/min)
20.00 30.00 40.00 50.00
Hct (%)
Urea
y3.0723x240.65
R20.603
y1.017x246.7
R20.097
Creatinine
Fig. 3. Small-solute clearance as a function of Hct in HD. Reprinted with permission
from Ronco et al. [28].
Clark/Rocha/Ronco 30
Dialysate Compartment
In recent years, enhanced blood compartment and transmembrane small-
solute mass transfer effi ciency has been attained by the use of high blood fl ow
rates and improved membrane designs, respectively. Consequently, most recent
efforts have focused on dialysate-side mass transfer. Based on the KOA concept
introduced above, both the dialysate-side mass transfer coeffi cient and mem-
brane surface area may infl uence mass transfer. The dialysate-side mass transfer
coeffi cent is determined largely by boundary layer phenomena, as in the blood
compartment. As discussed below, effective mass transfer area is not necessar-
ily equal to the manufacturer-reported (nominal) value.
Dialyzer characteristics which infl uence dialysate-side mass transfer
include packing density, fi ber undulation (also known as crimping), and the
presence or absence of spacer yarns. Packing density is defi ned as the ratio of
the area comprised of hollow fi bers to the area of the dialyzer housing, based on
a cross-sectional cut through the dialyzer. Recent magnetic resonance imaging
and computed tomography studies [28, 29] suggest that nonoptimized pack-
ing density may be the cause of channeling of dialysate at standard fl ow rates.
These investigations demonstrate that a large proportion of the dialysate stream
may fl ow peripherally to the fi ber bundle in dialyzers that are not optimally
confi gured. From a physical perspective, the interior of a fi ber bundle packed
too tightly represents a path of relatively large resistance while the peripheral
pathway is the path of least resistance. Obviously, an inwardly situated hollow
fi ber cannot participate in diffusive mass exchange if it is not perfused with
dialysate. Packing density values beyond the optimum may account for the
recent fi nding that dialyzers with a large surface area (i.e. greater than 1.7 m2)
are generally associated with less effi cient dialysate small-solute mass transfer,
relative to dialyzers of smaller surface area [30].
Another dialyzer characteristic that infl uences hollow-fi ber perfusion with
dialysate is fi ber bundle spacing. Dialysate may not be able to perfuse the area
between adjacent fi bers that are spatially too close. As is the case for nonopti-
mized packing density, this reduces the effective membrane surface area avail-
able for mass exchange. Two recently developed approaches to address this
fi ber spacing problem are spacer yarns and a specifi c fi ber undulation pattern.
Spacer yarns are multifi lament, linear structures interspersed longitudinally in a
specifi c spatial distribution within the fi ber bundle [28]. With respect to undu-
lation, all hollow fi bers are manufactured with a relatively specifi c periodicity
(amplitude and frequency). However, as discussed below, recent evidence sug-
gests that specifi c fi ber undulation approaches improve dialysate fl ow distribu-
tion and small-solute mass transfer.
In a recent clinical evaluation, Ronco et al. [28] measured the effect of
‘microcrimping’ and spacer yarns on both small-solute removal and dialysate
Dialyzer Solute Transport 31
fl ow distribution. The microcrimped fi bers contained in the dialyzers used in
this study have a relatively low amplitude and high frequency. In comparison
to conventional dialyzers (i.e. fi bers with standard undulation and no spacer
yarns), urea clearances were found to be signifi cantly higher for dialyzers with
both microcrimped fi bers and spacer yarns. Based on a computerized-tomogra-
phy-based technique, these investigators also found that dialysate fl ow distribu-
tion was most homogeneous in dialyzers with microcrimped fi bers and least
homogeneous in conventional dialyzers, with dialyzers having spacer yarn tech-
nology in an intermediate range (fi g. 4). These data suggest that both of these
newer approaches improve dialysate fl ow distribution and, thus, increase effec-
tive membrane surface area.
In addition to this infl uence on effective surface area, microcrimping may
also reduce dialysate-side mass transfer resistance essentially by disrupting
(‘agitating’) the boundary layer. Another way in which boundary layer effects
may be attenuated is through creation of a turbulent fl ow regime with a rela-
tively high QD. At a relatively common QB and QD combination of 300 and
500 ml/min, respectively, it is possible that dialysate-side mass transfer is rate
limiting under certain conditions. For several high-effi ciency and high-fl ux dia-
lyzers, Leypoldt et al. [30] reported a mean increase of 14% in in vitro urea KOA
when QD was increased from 500 to 800 ml/min at a constant QB of 450 ml/min.
These laboratory data have been corroborated clinically [31, 32].
Standard Space yarns Moiré structure
Fig. 4. Dialysate fl ow distribution in dialyzers having different structures. (Moiré
structure corresponds to microcrimped fi bers.) Reprinted with permission from Ronco et al.
[28].
Clark/Rocha/Ronco 32
Two important points about QD-related effects on small-solute mass trans-
fer require comment. First, for QD to have a signifi cant effect on KOA, a mini-
mal QB must be achieved. Specifi cally, if the QB is much less than 50% of the
QD at baseline, an increase in the latter is not expected to derive much benefi t
[33]. Second, it is important to note that the benefi cial effect of increasing QD
on small-solute mass transfer may also be due to a reduction in channeling with
improved perfusion of the inner fi ber bundle. Thus, the mass transfer benefi t of
both microcrimping and increased dialysate fl ow mechanisms may be due to
dissipation of boundary layer effects, an increase in effective membrane surface
area or both.
References
1 Lysaght MJ: Evolution of hemodialysis membranes; in Bonimini V, Berland Y (eds): Dialysis
Membranes: Structure and Predictions. Contrib Nephrol. Basel, Karger, 1995, vol 113, pp 1–10.
2 Lysaght MJ: Hemodialysis membranes in transition; in Colasanti G, D’Amico G (eds): Nephrology
and Dialysis Updated. Contrib Nephrol. Basel, Karger, 1988, vol 61, pp 1–17.
3 Hoenich N, Woffi ndin C, Cox P, Goldfi nch M, Roberts S: Clinical characterization of Dicea, a new
cellulose membrane for haemodialysis. Clin Nephrol 1997;48:253–259.
4 Ronco C, Brendolan A, Everard P, et al: Cellulose triacetate: another membrane for continuous
renal replacement therapy. J Nephrol 1999;12:241–247.
5 Funck-Bretano J, Sausse A, Man NK, Granger A, Rondon-Nucete M, Zingraff J, Jungers P: A new
hemodialysis treatment associating a membrane highly permeable to middle molecules with a
closed circuit dialysate system. Proc EDTA 1972;9:55–66.
6 Streicher E, Schneider H: The development of a polysulfone membrane: a new perspective in
dialysis? In Streicher E, Seyffart G (eds): Highly Permeable Membranes. Contrib Nephrol. Basel,
Karger, 1985, vol 46, pp 1–13.
7 Gohl H, Buck R, Strathmann H: Basic features of polyamide membranes; in Shaldon S, Koch KM
(eds): Polyamide – The Evolution of a Synthetic Membrane for Renal Therapy. Contrib Nephrol.
Basel, Karger, 1992, vol 96, pp 1–25.
8 Bonomini M, Fiederling B, Bucciarelli T, Manfrini V, Di Ilio C, Albertazzi A: A new polymethyl-
methacrylate membrane for hemodialysis. Int J Artif Organs 1996;19:232–239.
9 Jaber BL, Gonski JA, Cendoroglo M, Balakrishnan VS, Razeghi P, Dinarello C, Pereira BJG: New
polyethersulfone dialyzers attenuate passage of cytokine-inducing substances from Pseudomonas
aeruginosa contaminated dialysate. Blood Purif 1998;16:210–219.
10 Ronco C, Crepaldi C, Brendolan A, et al: Evolution of synthetic membranes for blood purifi ca-
tion: the case of the Polyfl ux family. Nephrol Dial Transplant 2003;18(suppl 7):vii10–vii20.
11 Rockel A, Hertel J, Fiegel P, Abdelhamid S, Panitz N, Walb D: Permeability and secondary mem-
brane formation of a high fl ux polysulfone hemofi lter. Kidney Int 1986;30:429–432.
12 Ledebo I: Principles and practice of hemofi ltration and hemodiafi ltration. Artif Organs 1998;22:
20–25.
13 Clark WR, Hamburger RJ, Lysaght MJ: Effect of membrane composition and structure on perfor-
mance and biocompatibility in hemodialysis. Kidney Int 1999;56:2005–2015.
14 Ronco C, Ballestri M, Gappelli G: Dialysis membranes in convective treatments. Nephrol Dial
Transplant 2000;15(suppl 2):31–36.
15 Ronco C: Backfi ltration: a controversial issue in modern dialysis. Int J Artif Organs 1988;11:
69–74.
16 Ronco C, Brendolan A, Lupi A, Metry G, Levin N: Effects of a reduced inner diameter of hollow
fi bers in hemodialyzers. Kidney Int 2000;58:809–817.
Dialyzer Solute Transport 33
17 Clark WR, Macias WL, Molitoris BA, Wang NHL: 2-Microglobulin membrane adsorption:
equilibrium and kinetic characterization. Kidney Int 1994;46:1140–1146.
18 Clark WR, Macias WL, Molitoris A, Wang NHL: Plasma protein adsorption to highly permeable
hemodialysis membranes. Kidney Int 1995;48:481–488.
19 Clark WR, Henderson LW: Renal versus continuous versus intermittent therapies for removal of
uremic toxins. Kidney Int 2001;59(suppl 78):S298–S303.
20 Clark WR, Shinaberger JH: Clinical evaluation of a new high-effi ciency hemodialyzer: polysyn-
thane (PSNn). ASAIO J 2000;46:288–292.
21 Jaffrin MY: Convective mass transfer in hemodialysis. Artif Organs 1995;19:1162–1171.
22 Slatsky M, Schindhelm K, Farrell P: Creatinine transfer between red blood cells and plasma: a
comparison between normal and uremic subjects. Nephron 1978;22:514–521.
23 Lim V, Flanigan M, Fangman J: Effect of hematocrit on solute removal during high effi ciency
hemodialysis. Kidney Int 1990;37:1557–1562.
24 Shinaberger J, Miller J, Gardner P: Erythropoietin alert: risks of high hematocrit hemodialysis.
Trans Am Soc Artif Intern Organs 1988;34:179–184.w
25 Huang Z, Clark WR, Gao D: Determinants of small solute clearance in hemodialysis. Semin Dial
2005;18:30–35.
26 Leypoldt JK, Cheung AK: Optimal use of hemodialyzers; in Ronco C, La Greca G (eds):
Hemodialysis Technology. Contrib Nephrol. Basel, Karger, 2002, vol 137, pp 129–137.
27 Cheung A, Alford M, Wilson M, Leypoldt JK, Henderson LW: Urea movement across erythrocyte
membrane. Kidney Int 1983;23:866–869.
28 Ronco C, Brendolan A, Crepaldi C, Rodighiero M, Scabardi M, Ghezzi PM: Blood and dialysate
fl ow distributions in hollow fi ber hemodialyzers analyzed by computerized helical scanning tech-
nique. J Am Soc Nephrol 2002;13(suppl 1):S53–S61.
29 Poh CK, Hardy PA, Liao Z, Huang Z, Clark WR, Gao DY: Effect of fl ow baffl es on the dialysate
fl ow distribution in hollow-fi ber hemodialyzers: a non-intrusive experimental study using MRI. J
Biomech Eng 2003;125:481–489.
30 Leypoldt JK, Cheung AK, Agodoa LY, Daugirdas JT, Greene T, Keshaviah PR: Hemodialyzer mass
transfer-area coeffi cients for urea increase at high dialysate fl ow rates. Kidney Int 1997;51:2013–
2017.
31 Ouseph R, Ward RA: Increasing dialysate fl ow rate increases dialyzer urea mass transfer-area
coeffi cients during clinical use. Am J Kidney Dis 2001;37:316–320.
32 Hauk M, Kuhlmann M, Riegel W, Kohler H: In vivo effects of dialysate fl ow rate on Kt/V in main-
tenance hemodialysis patients. Am J Kidney Dis 2000;35:105–111.
33 Sigdell J, Tersteegen B: Clearance of a dialyzer under varying operating conditions. Artif Organs
1986;10:219–225.
William R. Clark, MD
Gambro Renal Products
4322 Wythe Lane
Indianapolis, IN 46250 (USA)
Tel. 1 317 691 1438, Fax 1 317 849 4599, E-Mail william.clark@us.gambro.com
Ronco C, Canaud B, Aljama P (eds): Hemodiafi ltration.
Contrib Nephrol. Basel, Karger, 2007, vol 158, pp 34–49
Fluid Mechanics and Crossfiltration in
Hollow-Fiber Hemodialyzers
Claudio Ronco
Department of Nephrology, St. Bortolo Hospital, Vicenza, Italy
Abstract
The effi ciency of a hemodialyzer is largely dependent on its ability to facilitate diffu-
sion between blood and dialysis solution. The diffusion process can be impaired if there is a
mismatch between blood and dialysate fl ow distribution in the dialyzer. For this reason it is
important that average and regional blood and dialysate fl ow velocities do not differ signifi -
cantly. Single-fi ber fl ow velocity should be similar in the center and the periphery of the
bundle. Similarly, dialysate fl ow in the central region of the dialyzer and in the peripheral
areas should be similar. In this way the best blood-to-dialysate fl ow countercurrent confi gura-
tion is obtained, and the diffusive process is optimized. Unfortunately this optimal situation
is hard to achieve, and frequently a signifi cant blood-to-dialysate fl ow mismatch may occur
in hollow-fi ber hemodialyzers either due to uneven blood fl ow distribution or to a dialysate
channeling phenomenon external to the fi ber bundle. Attempts to optimize fl ows has been
made in the blood compartment designing specifi c blood ports while, in the dialysate, differ-
ent options have been proposed such as space yarns (spacing fi laments preventing contact
between fi bers) or the moiré structure (waved shape of fi bers to prevent contact between
adjacent fi bers). Furthermore, the process of transmembrane crossfi ltration along the length
of the dialyzer can be very different in quantity and direction, thus interfering signifi cantly
with the diffusion process. In particular, maximal rates of direct fi ltration (blood to dialysate)
are achieved in the proximal part of the dialyzer, while in the distal part ultrafi ltration is
minimal and it can also change direction producing signifi cant amounts of backfi ltration.
Copyright © 2007 S. Karger AG, Basel
Small-solute removal is obtained primarily by diffusion [1–5]. Convection
represents an additional mechanism mostly important for larger molecules.
The effi ciency of a hemodialyzer is therefore dependent on its ability to facili-
tate the diffusion process and optimize its interaction with convection [6–9].
Diffusion is affected by blood and dialysate fl ow rates, temperature, surface
area of the dialyzer and thickness of the membrane. Assuming that all other
Flow Distribution and Crossfi ltration in Hemodialyzers 35
factors are constant, the diffusion process is basically dependent on the con-
centration gradient between blood and dialysate [10, 11]. This is strongly
affected by the blood and dialysate fl ow rates and by the distribution of the
countercurrent fl ows in their relative compartments. It is evident that any pos-
sible mismatch between blood and dialysate fl ow distributions can create a
signifi cant reduction in the effi ciency of the fi lter [10]. In some cases blood
fl ow distribution may be less than optimal due to blood viscosity properties or
to a poor distribution of the fl ow at the blood inlet port [11, 12]. In this case,
the external fi bers of the bundle may be penalized by a lower fl ow velocity
compared to the fi bers located in the central region of the bundle. On the other
hand, fi ber packing density may be higher in the central region of the bundle,
and dialysate fl ow may be limited in that region by an increased resistance.
Under these circumstances, dialysate tends to fl ow at higher speed in those
regions of the fi lter where blood fl ow velocity is minimal and vice versa [13].
This effect may be the cause of inconsistent performances of the hemodia-
lyzer, and clearance values may be lower than those expected from theoretical
calculations [14].
When convection is taken into account, the regional fl ux across the mem-
brane becomes an important factor affecting solute transport. This is due to the
solvent drag effect but also to the potential negative interference of the fi ltration
fl uxes with the diffusion process. Thus, as far as hydraulics is concerned, three
fundamental aspects should be considered to analyze the solute transport pro-
cesses inside hollow-fi ber hemodialyzers: (a) the fl uid mechanics in the blood
compartment, (b) the fl uid mechanics in the dialysate compartment, (c) the
transmembrane crossfi ltration fl ux with its direct and reverse components.
Fluid Mechanics in the Blood Compartment
The uremic syndrome is characterized by the retention of a host of solutes
that interfere with various biochemical functions. Over the past decade, much
clinical research has been carried out on the adequacy of dialysis, mainly focus-
ing on the clearance of the small-molecular-weight substances like urea, with
much less consideration for the middle-molecular-weight substances, such as
2-microglobulin [15]. Moreover, treatment of renal anemia by recombinant
human erythropoietin has increased dramatically the average hematocrit (Hct)
in the dialysis population. Although positive effects were expected on cardio-
vascular function, there has been increasing concern that increasing Hct values
beyond a certain level will adversely affect dialyzer clearance necessitating a
modifi cation in dialysis therapy. Besarab et al. [16] reported that changes in
dialyzer clearances after erythropoietin treatment were not signifi cant. Other
Ronco 36
authors [17, 18] found that in patients receiving either high-fl ux or conventional
dialysis during treatment with erythropoietin, there was only a slight decrease
in clearances of creatinine, potassium and phosphate in the presence of higher
Hcts. However, in other studies [19, 20], creatinine and phosphate clearances
signifi cantly decreased in the presence of high Hct values, while urea clearance
was minimally affected. Burr and Martin [21] noted that dialysis effi ciency for
creatinine decreased by approximately 10% in patients with high Hct.
In a specifi cally designed study we showed that creatinine and phosphate
clearances negatively correlated with the levels of Hct, whilst urea clearance
only displayed a negative trend [22]. This is a fi nding which is in agreement
with previous studies [19, 20]. Urea is a highly diffusible molecule, freely
mobile between the extracellular and intracellular compartments. As a con-
sequence, urea should not be signifi cantly affected by changes in plasma
volume/red cell volume ratio. On the other hand, creatinine and phosphate
are slowly moving across the red cell membrane, and for dialytic purposes
they can be considered confi ned to plasma water. Relative plasma volume is
reduced when Hct increases, and this may decrease the delivery of creatinine
and phosphate to the hemodialyzers due to a decrease in effective plasma
fl ow.
Although this appears the most logical explanation, our in vitro results
seem to suggest a further possible effect played by an impaired blood fl ow dis-
tribution in the dialyzer occurring when Hct increases.
Blood is a concentrated suspension of red cells in an aqueous electrolyte-
protein solution, which shows a viscoelastic property [23]. The viscoelastic-
ity of the blood is traceable to the elastic red cells. When the red cells are at
rest, they tend to aggregate and stick together (fi g. 1). In order to fl ow freely,
the aggregates must disintegrate and the cells must undergo elastic deforma-
tion and orientation to each other. This process depends upon several factors
including Hct, plasma proteins and fi brinogen [24]. The blood fl ow in a narrow
vessel with an inner diameter ⬍300 m is determined by how the cells glide
past each other and how interaction between cells and vessel walls takes place
[25]. Therefore an insight into the blood fl ow mechanics within the dialyzer is
needed to understand the complicated effects of increasing Hct levels.
One should consider the important function of the arterial blood port of the
hemodialyzer. This component is crucial to ensure a good distribution of blood
into the fi bers. The presence of turbulence, dead spaces or preferential pathways
might interfere with a good distribution, and peripheral fi bers may be penalized
in terms of fl ow delivery.
Assuming a physiological difference in fl ow delivery between the central
(higher) and the peripheral (lower) regions of the bundle, there may be an addi-
tional effect inducing a further reduction of fl ow velocity in the peripheral fi bers
Flow Distribution and Crossfi ltration in Hemodialyzers 37
(fi g. 2) [14]. Due to the hydraulic design of the hemodialyzer, a transmembrane
pressure (TMP) gradient will be equally applied to all fi bers of the bundle. If
the single-fi ber blood fl ow is slightly lower from the beginning in the peripheral
regions of the bundle, these fi bers will experience a slightly higher fi ltration
fraction. In fact, for the same permeability coeffi cient and the same TMP gra-
dient, equal amounts of ultrafi ltration will be produced in all fi bers. However,
since peripheral fi bers tend to have a lower blood fl ow per fi ber, this will result
in a higher single-fi ber fi ltration fraction and higher hemoconcentration in the
fi ber. This in turn will result in an increase in blood viscosity and a possible
increase in the resistance to fl ow in those specifi c fi bers. This phenomenon may
contribute to a further reduction in fl ow velocity and a new steady-state profi le.
The phenomenon only leads to a new steady-state profi le, and not to a pro-
gressive obstruction of the peripheral fi bers for two reasons. One is due to the
lower wall shear rates present in these fi bers and the consequent formation of a
boundary layer that limits ultrafi ltration. The second is linked to a progressive
increase in the oncotic pressure generated by plasma proteins that acts against
ultrafi ltration. Assuming that these considerations are true, one could expect a
different fl ow distribution profi le within the hemodialyzer, in the presence of
constant Hct and blood fl ow but variable levels of ultrafi ltration.
Shear rate (1/s)
Relative blood viscosity
10⫺210⫺111010
2
1
10
102
103
Deformation
Aggregation
RBC in ringer solution
Hardened RBC in ringer solution
Vmax
V
–⫽0
rV⫽1
–
2 Vmax
dv⫽shear rate
dr
Fig. 1. Characteristics of blood as a non-Newtonian viscoelastic fl uid and behavior of
red cells suspended in different media.
Ronco 38
Considering the rheological properties of blood and its typically non-
Newtonian behavior, alternative hypotheses can be formulated. Blood is a fl uid-
ized suspension of red blood cells, which has viscoelastic properties refl ecting
the cumulative effects of plasma viscosity and Hct. Since the smaller the veloc-
ity and the shear rate applied to blood, the higher its viscosity, one can speculate
that the lower fl ow velocity observed in the peripheral fi bers further aggravates
the fl ow-dynamic conditions because of a relative increase in blood viscosity in
those fi bers.
In summary, blood is distributed nonproportionally in the hemodialyzer.
This uneven distribution is strongly affected by the level of Hct. The fi nal effect
is refl ected on effi ciency and solute clearances. We assume that this effect can
be individualized to the patients, since it can be affected by several factors, e.g.
the degree of ultrafi ltration, the presence of stiff erythrocytes and the blood fl ow
rate.
ROI⫽Region of interest
DP⫽Density profile
t1 and t2⫽Subsequent times of analysis
Cr⫽Central region of the fiber bundle
Pr⫽Peripheral regions of the fiber bundle
V⫽Average flow velocity in the filter
Vmax⫽Flow velocity in the central region
Vmin⫽Flow velocity in the peripheral regions
V⫽4QB/(d2n)
where QB⫽blood flow
d⫽fiber inner diameter
n⫽number of fibers
wSh⫽4QB/r2⫽4V r2/r3⫽4V/r
where wSh⫽wall Shear rate
r⫽fiber inner radius
V⫽flow velocity in the fiber
DP 2
DP 1
ROI 2
ROI 1
Pr
1cm
Cr
1cm Pr
1cm
t1
t2
5 cm
3 cm
1 cm
ROI 3
Vmax
Vmin
Blood
Fig. 2. Flow distribution in the blood compartment and effects of local ultrafi ltration
on viscosity and concentration polarization. Two fl ow distribution profi les describing two
subsequent times (t1 and t2) are presented. Analysis of distribution profi les is carried out in
different regions of interest (ROI 1, ROI 2 and ROI 3) with a specifi c radiographic method.
Once the blood enters the blood port of the hemodialyzer, a fl ow distribution curve starts to
build up. Regional velocities are different at different moments and in different regions of
the bundle.
Flow Distribution and Crossfi ltration in Hemodialyzers 39
As far as the blood compartment is concerned, hemodialyzer design has
been quite constant over the last 20 years. The average Hct of patients has how-
ever increased signifi cantly. Therefore, a careful evaluation of dialysis adequacy
parameters is strongly advisable when dealing with patients in which Hct levels
progressively increase.
These considerations apply even more when treatments with high convec-
tive components are involved. In fact, hemodiafi ltration (HDF) may induce a
signifi cant hemoconcentration in the blood compartment, and increased vis-
cosity can be experienced along the length of the hollow fi bers. This effect is
mostly observed in postdilution modalities, while in predilution or middilution
this effect is mitigated and an improved performance of the fi ber is observed.
Nevertheless, a general recommendation should be made to increase blood
fl ows as much as possible when convective therapies are utilized.
Fluid Mechanics in the Dialysate Compartment
Some studies have demonstrated that the fl ow distribution in the dialysate
compartment might be less than optimal due to channeling phenomena [26].
Under such circumstances, the hypothesis of a blood-to-dialysate fl ow mis-
match as a major cause of dialyzer malfunction or impaired effi ciency becomes
realistic.
The fl ow distribution in the dialysate compartment can be theoretically
modeled using equations of physical chemistry and transport [27, 28] and can
be assimilated to the fl ow distribution in packed beds [29].
As in packed beds, the packing structure of the hollow fi bers is usually
quite complex, and the resulting fl ow pattern external to the fi bers is extremely
complicated. In well-packed columns, the diversity of channel diameters and of
velocities in the individual channels is small. In this case the packed bed can be
approximated to a bundle of tortuous capillary tubes. In the case of the dialysate
compartment of a hollow-fi ber hemodialyzer, some wide-diameter channels and
gaps in the packing structure may be present resulting in wide variations of local
fl ow velocity. This may cause the undesirable phenomenon of so-called chan-
neling of the fl ow. As discussed in detail in a recent publication [29], the funda-
mental principle governing the fl ow of fl uids through packed beds is Darcy’s law.
The free cross-section of the dialysate compartment bed (total internal area of
the case – total area occupied by the fi bers) is constituted by the interfi ber gaps
(interparticle porosity) and constitutes the fl uid pathway external to the fi bers.
The dimension of the specifi c permeability of the compartment is square centi-
meters but it could be given in Darcy units (1 darcy ⫽ 10–8 cm2). Unfortunately,
the bundle can be concentrated and packed in the central region of the dialyzer,
Ronco 40
and low-resistance pathways can be created in the more peripheral regions of
the compartment. This results in a greater fl ow velocity in the peripheral regions
while a signifi cant stagnation can be observed in the central region. In this case,
the specifi c permeability of the interfi ber space of the bundle in the central
region becomes much smaller than that observed in the peripheral regions and
the effi cacy of the countercurrent fl ow is impaired (fi g. 3).
The most uniform fl ow profi le in packed beds can be obtained when beds
are packed tightly with spherical particles of equal size. If the ratio of the tube
diameter to the particle diameter is less than 100, this may have a signifi cantly
positive effect on the fl ow distribution profi le.
In hollow-fi ber dialyzers, the fi bers and their external surface substitute the
particles of a porous bed. Since the fi bers are not tightly packed, preferential
Pr
1cm
Cr
1cm
5 cm
3 cm
1 cm
Dialysate
DP 2
t2
Vmax
Vmin
ROI 2
ROI 1
ROI 3
DP 1
t1
ROI⫽Region of interest
DP⫽Density profile
t1 and t2 ⫽Subsequent times of analysis
Cr⫽Central region of the fiber bundle
Pr⫽Peripheral regions of the fiber bundle
V⫽Average flow velocity in the filter
Vmax⫽Flow velocity in the central region
Vmin⫽Flow velocity in the peripheral regions
Pr
1cm
Fig. 3. Flow distribution in the dialysate compartment. Two fl ow distribution profi les
(describing two subsequent times t1 and t2) are presented. Analysis of distribution profi les is
carried out in different regions of interest (ROI 1, ROI 2 and ROI 3) with a specifi c radio-
graphic method. Once the dialysate enters the Hansen connector of the hemodialyzer, a fl ow
distribution curve starts to build up. Regional velocities are different at different moments
and in different regions of the bundle.
Flow Distribution and Crossfi ltration in Hemodialyzers 41
fl uid pathways may be generated. This explains why the packing density of hol-
low fi bers is an important parameter in the design of a hemodialyzer. Further, it
seems that special confi gurations designed to prevent a close contact of adjacent
fi bers may induce a signifi cant improvement on the fl ow distribution. Studies
have permitted the evaluation of the possible impact of new solutions oriented
to the improvement of the dialysate pathway confi guration. In particular, the
use of space yarns external to hollow fi bers may help in reducing the negative
effects due to dialysate channeling. A more homogeneous distribution of the
dialysate has been however obtained by the waved confi guration of the hollow
fi bers in the bundle, creating the so-called moiré structure [14].
The optimization of dialysate distribution in the modifi ed hemodialyzers
is also confi rmed by an improved performance in terms of urea clearance. This
suggests a defi nite improvement of the diffusion processes inside the dialyzer
due to an optimization of the countercurrent effect on blood-to-dialysate solute
gradients [11, 30].
Crossfiltration in Hollow-Fiber Hemodialyzers
Water fl ux across the dialysis membrane in each axial segment (dl) of the
dialyzer may occur in two directions: from blood to dialysate, which is termed
fi ltration, or from dialysate to blood, which is termed backfi ltration.
Backfi ltration may occur inside any kind of fi lter, and during any kind of
treatment, when the TMP gradient at a given point becomes negative – that is
when the hydraulic pressure of dialysate (PD) together with the oncotic pres-
sure exerted by plasma proteins () exceeds the hydraulic pressure of the blood
inside the fi ber (PB). This condition may happen occasionally during the treat-
ment or for the entire duration of the session affecting solute fl uxes [31–33].
TMP is generally expressed in average values with the simplifi ed equation 1:
TMP ⫽ (PBi ⫹ P Bo)/2 ⫹ (PDi ⫹ P Do)/2 ⫹ (i ⫹ )/2 (1)
where i and o represent the inlet and the outlet of the fi lter for blood and dialy-
sate, respectively. However, this representation only describes an average phe-
nomenon and it does not defi ne the actual profi le of local pressures along the
length of the fi lter. Although TMP is positive, the local pressure gradient ⌬P is
not necessarily positive at every point along the length of the fi lter. Equation 1
also assumes that the pressure drop inside the blood and dialysate compartment
is linear which according to the Hagen-Poiseuille law is only true under certain
circumstances. The pressure drop is linear only when blood viscosity remains
constant along the fi bers, and this would only occur if no crossfi ltration is pres-
ent. Crossfi ltration in fact either increases blood viscosity when it is direct or
Ronco 42
reduces viscosity when it is reverse. Thus, since crossfi ltration cannot be null,
viscosity is subject to change, and pressure drop cannot be linear.
The local water fl ux in a single surface element (ds) of the dialyzer is
described by the equation QF ⫽ KM ⫻ ⌬P (KM being the hydraulic permeability
coeffi cient of the membrane and ⌬P the algebraic sum of hydraulic and oncotic
pressures). Expanding this concept to the whole dialyzer, the overall water fl ux
in a given dialyzer will be expressed by the formula described in equation 2:
QF ⫽ 兰兰ol ⌬P ⫻ K M ⫻ ds (2)
where ds is a single surface element of the dialyzer and s is the surface of the
dialyzer, o is the initial segment of the dialyzer and l is the most distal segment
of the dialyzer after n segments of dimension dl.
Arbitrarily assuming KM to be a constant on the whole surface area s, and
⌬P to be identical in any point of a cross-sectional segment of the dialyzer,
equation 2 can be simplifi ed as follows:
QF ⫽ K M 兰ol ⌬P ⫻ dl (3)
where l is the length of the dialyzer and ⌬P is the local TMP gradient in a cross-
sectional segment of the dialyzer (dl). For simple calculation we can use the
formula:
QF ⫽ K D 兰ol ⌬P ⫻ dl/l (4)
where 兰ol ⌬P ⫻ dl/l is the average TMP gradient (avTMP) and KD is the dialyzer
ultrafi ltration coeffi cient. Thus, the overall net water fl ux will be
QF ⫽ K D ⫻ avTMP. (5)
While this is a commonly used equation to simplify the phenomena inside the
dialyzer, a more complex crossfi ltration and pressure profi les have been experi-
mentally determined by nuclear scintigraphic methods [34] (fi g. 4). Linear
models are in fact not accurate to predict fi ltration and backfi ltration fl uxes
empirically measured by the changes in concentration of a nondiffusible marker
molecule along the length of the fi bers.
The water fl ux inside the dialyzer is then the result of two opposing
fl uxes:
QF ⫽ Q F1 ⫺ Q F2 ⫽ (KM1 ⫻ 兰0x ⌬P ⫻ dl) ⫺ (KM2 ⫻ 兰xl ⌬P ⫻ dl) (6)
where QF ⫽ net ultrafi ltration; QF1 ⫽ fi ltration; QF2 ⫽ backfi ltration; KM1 ⫽
membrane direct ultrafi ltration coeffi cient; KM2 ⫽ membrane reverse ultrafi ltra-
tion coeffi cient; x ⫽ point of inversion of pressure gradient and water fl ux.
Most of the effects observed along the length of the fi lter are related to
variations in blood viscosity and plasma protein concentration. In fact, in highly
permeable hollow-fi ber hemodialyzers, although with higher fi ltration rates
Flow Distribution and Crossfi ltration in Hemodialyzers 43
backfi ltration is minimized, the fl ux of reverse fi ltration cannot be avoided since
plasma proteins operate small but signifi cant amounts of oncotic-pressure-
driven backfi ltration. Thus, while the linear model suggests to identify three
possible conditions, i.e. (a) spontaneous fi ltration, (b) critical fi ltration (the
minimal amount of fi ltration necessary to avoid backfi ltration) and (c) zero net
fi ltration (the fl ux of fi ltration equals that of backfi ltration), the experimental
fi ndings clarify that even at high fi ltration rates, small amounts of backfi ltration
are always present.
Several lines of evidence demonstrate the importance of middle-mol-
ecule removal in hemodialysis [35]. The use of high-fl ux membranes and the
increased use of convective techniques have permitted to improve the effi ciency
of hemodialysis leading to better removal of solutes in the middle-molecular-
weight range [36, 37].
The improvement achieved with synthetic membranes is mainly due to
their higher hydraulic permeability and their increased sieving capacity com-
pared to classic cellulose membranes. These properties result in higher middle-
molecule clearance. This clearance improvement is due to larger amounts of
ultrafi ltration per treatment and a more important contribution of convection to
the overall transport process [38].
C1C2aC
3
0
C2bC
2c
0
ab
2012
Qf⫽150
Qf⫽90
Qf⫽30
24 281684
V
E
R
T
I
C
A
L
Fig. 4. Examples of local crossfi ltration fl uxes along the length of the dialyzer at dif-
ferent net fi ltration rates. a The scintigraphic curves of the nondiffusible marker molecule
generated by the gamma camera are reported. b Filtration fl uxes are positive in the proximal
part of the dialyzer while backfi ltration occurs distally. Even at very high fi ltration rates,
backfi ltration cannot be avoided.
Ronco 44
Synthetic high-fl ux membranes are used both in Europe and the
USA, with different modes of application [35, 36]. In Europe, high-fl ux
dialysis and HDF are used as treatment modalities for chronic renal fail-
ure. Conventional HDF utilizes large convective transport with ultrafi ltra-
tion rates above 70 ml/min. Since the ultrafi ltration rate exceeds the rate of
desired weight loss in the patient, sterile replacement fl uid must be admin-
istered. The net ultrafi ltration rate in the patient will be equal to the differ-
ence between total ultrafi ltration rate and reinfusion rate. Total ultrafi ltration
varies between 12 and 15 l/session. The enhanced convective transport in
HDF permits an increased removal of middle- to high-molecular-weight sol-
utes such as 2-microglobulin compared to standard hemodialysis [35–39].
The major drawbacks of this treatment are the complexity of the system
and the increased costs compared to conventional hemodialysis due to large
amounts of substitution fl uid.
In the USA, high-fl ux membranes are commonly utilized in high-fl ux
dialysis in which net fi ltration rates are volumetrically controlled. This results
in a complex fl uid balance within the dialyzer where true fi ltration rates are
counterbalanced by signifi cant amounts of backfi ltration. In this setting, con-
vective transport is partially maintained and the clearance of large molecules is
improved compared to conventional hemodialysis, although not as much as in
HDF. The magnitude of net fi ltration in high-fl ux dialysis is controlled by the
dialysis machine. In contrast, the amounts of true fi ltration and backfi ltration
are determined by the hydraulic permeability and surface of the membrane, by
the geometry of the dialyzer and by the hydrostatic and oncotic forces acting
on the dialysis membrane. In previous studies, we measured the internal water
fl uxes in high-fl ux dialyzers using an experimental dialysis circuit at zero net
fi ltration [34]. In such conditions, the rate of fi ltration and concomitantly of
backfi ltration could be precisely determined.
Convective transport and the rates of fi ltration-backfi ltration can be
increased in high-fl ux dialyzers, by modifying the geometry of the fi lter but also
of the hollow fi ber. The rates of fi ltration-backfi ltration at a given blood fl ow are
directly correlated with the resistance of the fi lter, i.e. with the pressure drop
in the blood compartment and that in the dialysate compartment. Filtration-
backfi ltration rates can be increased experimentally by the application of a fi xed
O ring external to the fi ber bundle to alter the dialysate end-to-end pressure
drop [40]. In the blood compartment this effect could be achieved by modify-
ing the length of the fi lter and/or its cross-sectional area. The cross-sectional
area of a dialyzer can be modifi ed by changing the number of hollow fi bers in
the bundle or by using hollow fi bers with a different inner diameter. A reduced
inner diameter is in fact expected to increase proximal fi ltration and distal back-
fi ltration, by increasing the resistance to fl ow in the blood compartment. In all
Flow Distribution and Crossfi ltration in Hemodialyzers 45
cases, improved convective removal of large solutes is expected because of the
increased internal fi ltration [41].
Internal crossfi ltration is governed by the hydraulic and oncotic forces act-
ing along the length of the dialyzer on each side of the membrane. In each point
of the dialyzer, the local pressure differential is termed TMP. When the TMP is
positive, the water fl ux is from the blood compartment to the dialysate compart-
ment. When the TMP is negative, backfi ltration occurs. The removal of middle
molecules can be enhanced by increasing the positive pressure differential in
the proximal part of the dialyzer, thus increasing internal fi ltration. Adequate
net fi ltration is maintained by the ultrafi ltration control system by a parallel
increase in the negative pressure differential in the distal part of the dialyzer.
This would result in increased rates of proximal fi ltration and distal backfi ltra-
tion without affecting the ‘net’ fi ltration rate.
The relationship of TMP, ultrafi ltration (UF) and membrane ultrafi ltration
coeffi cient (KM) is expressed in the equation:
KM ⫽ UF/TMP. (7)
High-fl ux membranes have a KUF ⬎20 ml/h/mm Hg. Therefore, these mem-
branes can have UF rates of 4,000 ml/h with a TMP ⫽ 300 mm Hg. The clear-
ance of middle molecules at this UF rate would be ideal, but the patient cannot
tolerate these rates. HDF, as carried out in Europe, operates in conditions of
high fi ltration rates, but signifi cant amounts of replacement fl uid are required to
maintain the patient’s fl uid balance.
In high-fl ux dialysis, volumetric control regulates the net ultrafi ltration;
however as stated previously, the convective transport is limited by the rate
of internal fi ltration. Modifi cation of the dialyzer structure could increase the
peaks of positive and negative TMP along the length of the dialyzer, thereby
increasing fi ltration and backfi ltration [34].
When high rates of backfi ltration are utilized, a high quality of dialy-
sate is needed to prevent any inconvenient or side effect related to pyrogen
transfer into the patient’s circulation [42, 43]. For this treatment the use of
last-generation hemodialysis machines is strongly suggested. New machines
are equipped with a built-in pyrogen fi lter to prepare ultrapure dialysate. The
reinfusion via backfi ltration is providing an extra step of safety since the fl uid
is fi ltered again across the hemodialysis membrane prior to reaching the blood
compartment.
The modifi cation of the inner diameter of the fi ber may become an inter-
esting approach to increasing the TMP without introducing major changes in
the dialyzer design. In previous experiments, this approach resulted in a posi-
tive increase in the TMP in the proximal portion of the dialyzer and a negative
increase in the distal portion of the dialyzer [41]. The difference in pressure
Ronco 46
drop when the inner diameter is even marginally reduced becomes signifi cant as
it is predicted by the Hagen-Poiseuille formula
⌬P ⫽ Q B ⫻ (8l/r4) (8)
where ⌬P ⫽ end-to-end pressure drop, QB ⫽ blood fl ow, ⫽ blood viscosity.
Since the pressure drop in a fi ber correlates with the internal radius of
the fi ber to the fourth power, it seems logical to attempt solutions that involve
modifi cations of the fi ber geometry. One may speculate that a reduction of the
inner diameter of the hollow fi ber will also result in an increase in the average
blood fl ow velocity per fi ber and a consequent increase in wall shear rates. This
additional factor may in fact result in a ‘cleaning’ effect at the blood-membrane
interface. Higher shear rates lead in fact to a reduction of the thickness of the
protein boundary layer and improve membrane permeability counterbalancing
the concentration polarization phenomenon. This will certainly help not only
to obtain a better performance of the membrane in terms of fi ltration rates at a
given local TMP gradient, but also an optimal utilization of the sieving capaci-
ties of the membrane.
From this observation the importance of the blood fl ow rate as a major
determinant of convective clearance becomes evident. In fact, at a given blood
fl ow, the viscosity of blood and the cross-sectional area of the conduct gov-
ern the end-to-end pressure drop in the blood compartment. If blood fl ow is
increased, the end-to-end pressure drop will increase according to the Hagen-
Poiseuille law and so will the TMP gradient in the proximal and distal parts of
the hemodialyzer. For this reason, the same amount of convective transport in
specifi cally modifi ed dialyzers was obtained at lower blood fl ow rates compared
to standard dialyzers [40, 41]. This is in fact a good chance to achieve a remark-
able convective clearance during high-fl ux dialysis without reaching danger-
ously high fi ltration fractions and an increased risk of clotting in patients who
cannot be treated with traditional convective therapies such as hemofi ltration or
HDF because of insuffi cient blood fl ow.
The fi nal consideration concerns the better performance of the hollow fi bers
in the fi ltration-backfi ltration mode, as compared to the fi ltration-postdilution
mode. In the latter mode in fact, a greater impact of protein concentration polar-
ization is expected and the boundary layer at the blood-membrane interface will
be thicker resulting in a signifi cant decrease in membrane permeability.
In conclusion, two directions will probably be undertaken in the near
future: one consists in modifi cations of the design of hollow fi bers leading to
simplifi ed HDF techniques without the need for replacement solutions, but sim-
ply utilizing internal fi ltration as a main way to increase convective transport;
in contrast, the second consists in the maximization of the convective transport
utilizing replacement solutions produced online (and therefore cheaply and in
Flow Distribution and Crossfi ltration in Hemodialyzers 47
large amounts) possibly infused in predilution mode to improve performance of
the fi bers [43].
Conclusions
After the accurate analysis of the mechanics of fl uid in the blood and dialy-
sate compartments together with the profi les of crossfi ltration along the length
of the dialyzer, we can conclude that diffusion and convection are two mem-
brane separation processes that continuously interfere making it impossible to
clearly distinguish the separate contribution of each phenomenon to the fi nal
solute removal. We may say that while in hemodialysis diffusion is the prevalent
mechanism, in HDF convection may become the prevalent mechanism, and in
high-fl ux dialysis a really mixed form of transport is observed inside the fi lter
without a prevalent mechanism of removal.
References
1 Kill F: Development of a parallel fl ow artifi cial kidney in plastics. Acta Chir Scand Suppl 1960;
253:S142–146.
2 Malchesky PS, Mrava GI, Nose Y: A totally disposable presterilized dialyzer. Arch Intern Med
1971;127:278–283.
3 Kolff VJ, Watshinger B: Further development of a coil kidney. J Lab Clin Med 1956;47:969–974.
4 Haas G: Dialysieren des strömenden Blutes am Lebenden. Klin Wochenschr 1923;2:1888–1894.
5 Nakagawa S, Koshikawa S, Ishida Y, Uematsu M, Ishibashi K: Development of fl at type hollow
fi bre dialyzer (NF-01) achievement of better performance than cylinder type with same membrane
area; in Frost TH (ed): Technical Aspects of Renal Dialysis. Tunbridge Wells, Petman Medical,
1977, pp 29–37.
6 Sigdell JE: Hollow fi ber dialyzers on the market. Artif Organs 1988;12:4–11.
7 Sigdell JE: Comparison of the hollow dialyzers. Artif Organs 1981;5:4–9.
8 Sakai K: Technical determination of optimal dimension of hollow fi bre membranes for clinical
dialysis. Nephrol Dial Transplant 1989;4(suppl):S73–S77.
9 Ronco C, Ballestri M: New developments in hemodialyzers. Blood Purif 2000;18:267–275.
10 Vander Velde C, Leonard EF: Theoretical assessment of the effect of fl ow maldistributions on the
mass transfer effi ciency of artifi cial organs. Med Biol Eng Comput 1985;23:224–229.
11 Hoenich N, Ronco C: Dialyzer evaluation; in Winchester J, Koch K, Jacobs C, Kjiellestrand C
(eds): Replacement of Renal Function by Dialysis. Dordrecht, Kluwer Academic Publishers, 1996,
pp 256–270.
12 Leypoldt JK, Cheung AK, Agodoa LY, Daugirdas JT, Greene T, Keshaviah PR: Hemodialyzer
mass transfer-area coeffi cients for urea increase at high dialysate fl ow rates. The Hemodialysis
(HEMO) Study. Kidney Int 1997;51:2013–2017.
13 Ronco C, Ghezzi PM, Metry G, Spittle M, Brendolan A, Rodighiero M, Milan M, Zanella M, La
Greca G, Levin N.W: Effects of hematocrit and blood fl ow distribution on solute clearance in hol-
low-fi ber hemodialyzers. Nephron 2001;89:243–250.
14 Ronco C, Brendolan A, Crepaldi C, Rodighiero MP, Scabardi M: Blood and dialysate fl ow distri-
butions in hollow-fi ber hemodialyzers analyzed by computerized helical scanning technique. J Am
Soc Nephrol 2002;13:S53–S61.
Ronco 48
15 Skroder N, Jacobson S, Holmquist B, Kjellstrand P, Kjellstrand C: 2-Microglobulin generation
and removal in long slow and short fast hemodialysis. Am J Kidney Dis 1993,21:519–526.
16 Besarab A, Girone J, Erslev A, et al: Recent developments in the anemia of chronic renal failure.
Semin Dial 1989;2:87–97.
17 Acchiardo S, Quinn B, Burk L, et al: Are high fl ux dialysis and erythropoietin treatment in a colli-
sion course? ASAIO Trans 1989;14:19–25.
18 Paganini E, Latham D, Abdulhadi M: Practical considerations of recombinant human erythropoi-
etin therapy. Am J Kidney Dis 1989; 14(suppl 1):19–25.
19 Zehnter E, Pollock M, Longere F, Bramsiepe P, Ziegenhagen D, Baldamus C: Urea kinetics in
patients on RDT treated with recombinant erythropoietin (abstract). Kidney Int 1988;33:242.
20 Shinaberger J, Miller J, Gardner P: Erythropoietin alert: risks of high hematocrit hemodialysis.
ASAIO Trans 1988;34:179–184.
21 Burr T, Martin L: Secondary effects of erythropoietin on metabolism and dialysis effi ciency in
stable hemodialysis patients. Clin Nephrol 1990;34:230–335.
22 Chiou W, Lee H: Effect of change in blood fl ow on hemodialysis clearance studied by a simple
unifi ed organ clearance approach. Res Commun Chem Pathol Pharmacol 1989;65:393–396.
23 Thurston G: Viscoelasticity of human blood. Biophys J 1972;12:1205–1217.
24 Schweiger B, Pohl I, Driessen G, Heidtmann H, Schmid-Schönbein H: Infl uence of fi brinogen
concentration on fl ow velocity of RBCs in rat mesentery; in Gaehtgens P (ed): Recent Advances in
Microcirculatory Research. Bibl Anat. Basel, Karger, 1981, vol 20, pp 220–233.
25 Secomb T: Red blood cell mechanics and capillary blood rheology. Cell Biophys 1991;18:
231–251.
26 Ronco C, Scabardi M, Goldoni M, Brendolan A, Crepaldi C, La Greca G: Impact of spacing fi la-
ments external to hollow fi bers on dialysate fl ow distribution and dialyzer performance. Int J Artif
Organs 1997;5:261–266.
27 Jost W: Diffusion in Solids, Liquids, Gases. New York, Academic Press, 1965.
28 Scheidegger AE: The Physics of Flow through Porous Media. New York, McMillan, 1960.
29 Ronco C, Ghezzi PM, Morris A, Rosales L, Wang E, Zhu F, Metry G, De Simone L, Rahmati S,
Adhikarla R, Bashir A, Manzoni C, Spittle M, Levin NW: Blood fl ow distribution in sorbent beds:
analysis of a new sorbent device for hemoperfusion. Int J Artif Organs 2000;23:125–130.
30 Ronco C: Dialysis delivery versus dialysis prescription. Int J Artif Organs 1993;16:628–635.
31 Ronco C: Hemofi ltration and hemodiafi ltration; in Bosch JP (ed): Hemodialysis: High Effi ciency
Treatments. New York, Churchill Livingstone, 1993, pp 119–133.
32 Ronco C, Brendolan A, Bragantini L, Chiaramonte S, Feriani M, La Greca G: Technical and clini-
cal evaluation of four different highly effi cient short dialysis techniques; in Colasanti G, D’Amico
G (eds): Nephrology and Dialysis Updated. Contrib Nephrol. Basel, Karger, 1988, vol 61, pp
46–53.
33 Ronco C: Backfi ltration: a controversial issue in modern dialysis. Int J Artif Organs 1988;11:
69–74.
34 Ronco C, Brendolan A, Feriani M, Milan M, Conz P, Lupi A, Berto P, Bettini MC, La Greca G: A
new scintigraphic method to characterize ultrafi ltration in hollow fi ber dialyzers. Kidney Int 1992;
41:1383–1393.
35 Ohmi T, Muto C, Nishibori F: Investigation on plasma clearance of beta-2-microglobulin in hemo-
dialysis with PMMA membrane. Jpn J Artif Organs 1991;20:116–121.
36 Lee CjHsiong CH, Chang YL, Cheng CH, Lian JD: Statistical and parameteric analysis of beta-2
microglobulin removal from uremic patients in high fl ux hemodialysis. ASAIO J 1994;40:62–66.
37 Floege J, Granolleras C, Deschodt G, Heck M, Baudin G, Branger B, Tournier O, Reinhard B,
Eisenbach GM, Smeby LC: High fl ux synthetic versus cellulosic membranes for beta-2 microglob-
ulin removal during hemodialysis, hemodiafi ltration and hemofi ltration. Nephrol Dial Transplant
1989;4:653–657.
38 Locatelli F, Mastrangelo F, Redaelli B, Ronco C, Marcelli D, La Greca G, Orlandini G: Effects
of different membranes and dialysis technologies on patient treatment tolerance and nutritional
parameters. Kidney Int 1996;50:1293–1302.
Flow Distribution and Crossfi ltration in Hemodialyzers 49
39 Miller Jh, von Albertini B, Gardner PW, Shinaberger JH: Technical aspects of high fl ux hemodia-
fi ltration for adequate short (under 2 hours) treatment. Trans Am Soc Artif Intern Organs 1984;
30:377–380.
40 Ronco C, Brendolan A, Lupi A, Bettini MC, La Greca G: Enhancement of convective transport by
internal fi ltration in a modifi ed experimental hemodialyzer. Kidney Int 1998;54:979–985.
41 Ronco C, Brendolan A, Lupi A, Metry G: Effects of a reduced inner diameter of hollow fi bers in
hemodialyzers. Kidney Int 2000;58:809–817.
42 Ronco C, Cappelli G, Ballestri M, Lusvarghi E, Frisone P, Milan M, Dell’Aquila R, Crepaldi C,
Dissegna D, Gastaldon F, La Greca G: On-line fi ltration of dialysate: structural and functional
features of an asymmetric polysulfone hollow fi ber ultrafi lter (Diaclean). Int J Artif Organs 1994;
10:515–520.
43 Canaud B, Bosc JY, Leray-Moragues H, et al: On-line haemodiafi ltration: safety and effi cacy in
long-term clinical practice. Nephrol Dial Transplant 2000;5(suppl 1):S60–S67.
Dr. Claudio Ronco
Department of Nephrology, St. Bortolo Hospital
Viale Rodolfi 16
IT–36100 Vicenza (Italy)
Tel. ⫹39 0444753650, Fax ⫹39 0444753949, E-Mail cronco@goldnet.it
Ronco C, Canaud B, Aljama P (eds): Hemodiafi ltration.
Contrib Nephrol. Basel, Karger, 2007, vol 158, pp 50–56
Mechanisms of Solute and Fluid Removal
in Hemodiafiltration
Akihiro C. Yamashita
Department of Materials Science and Engineering, College of Engineering,
Shonan Institute of Technology, Fujisawa, Japan
Abstract
Background: Prescribing therapeutic conditions for online predilution hemodiafi ltra-
tion (HDF) with fi xed total dialysate fl ow rate QDtotal is not straightforward, since the increase
in the substitution fl ow rate QS is compensated by the decrease in the net dialysate fl ow rate
QDnet. Methods: Clearances of various solutes under online predilution HDF were clinically
evaluated with fi xed QDtotal (⫽ 520 ml/min) divided into QDnet and QS. Three polysulfone
membrane dialyzers and 5 polyester polymer alloy membrane dialyzers were chosen to mea-
sure sieving coeffi cients (SC) for albumin in vitro at 37°C to predict when the albumin loss
is greatest during clinical treatment. Results: Clearances of small solutes such as urea and
creatinine increased in vivo with the increase in blood fl ow. These values, however, slightly
but steadily decreased with the increase in QS because the increase in QS decreased QDnet.
Clearances of 2-microglobulin and ␣1-microglobuin increased with the increase in QS and
decreased with the increase in QDnet, because clearances of larger solutes were more strongly
dependent on ultrafi ltration than on diffusion. The SC for albumin in vitro showed a peak at
the beginning of the experiment in those membranes with large proportions of polyvinylpyr-
roridone (PVP), which may lead to large amounts of albumin loss at the beginning of the
treatment. Conclusions: Dialysis prescription in online predilution HDF in terms of maxi-
mizing clearance for the solute of interest may be different for each target solute. The amount
of albumin loss may be closely related to the amount of PVP included in the membrane.
Copyright © 2007 S. Karger AG, Basel
Hemodiafi ltration (HDF) is accepted as one of the intermittent blood puri-
fi cation modalities for treating end-stage renal disease (ESRD) patients, since
it removes toxins over a wide molecular-weight range [1–3]. In HDF treat-
ments, however, since molecular diffusion and considerable amount of ultra-
fi ltration occur at the same time in one dialyzer, the solute and fl uid transport
mechanisms are much more complicated than other treatment modalities in
Solute and Fluid Removal in HDF 51
which either diffusion or convection is used with a limited effect of the other
mechanism. Many factors affect the fl uid and mass transport in HDF, and the
following 2 factors need to be considered before starting discussion. The fi rst
concerns the mode of dilution. Postdilution has been the preferred form of
treatment, owing to its optimal cost-effectiveness. However, the use of online
prepared dialysate as the substitution fl uid has solved the economic problem of
predilution. Moreover, since it may be easier to control the amount of albumin
loss with predilution HDF, this mode is preferred when dialyzers with larger
pore sizes are used. Secondly, in recent years, many petroleum-based synthetic
polymers have been used as dialysis membranes; however, these materials are
too hydrophobic in nature to use for blood purifi cation treatment. A hydrophilic
agent is therefore often added to these materials to improve their compatibility
with blood, more specifi cally to prevent thrombosis. The most common is poly-
vinylpyrrolidone (PVP), which has been used with many polymers, including
polysulfone, polyether sulfone, polyamide and polyester polymer alloy (PEPA).
However, PVP affects not only biocompatibility, but also solute transport,
including the amount of albumin loss. Then we will consider online predilu-
tion HDF with fi xed total dialysate fl ow rate, which is sometimes diffi cult for
prescribing therapeutic conditions, since an increase in the substitution fl ow
rate is compensated by a decrease in the net dialysate fl ow rate. Mechanisms of
albumin loss in HDF with high-fl ux dialyzers using PVP are also discussed in
this section.
Materials and Methods
The amount of PVP used in the membrane is semiquantitatively scaled in the following
4 categories based on the membrane casting procedure: PVP⫺ ⫽ no PVP, PVP⫹ ⫽ a small
amount, PVP⫹⫹ ⫽ an increased amount, and PVP⫹⫹⫹ ⫽ a large amount.
Clearances for urea [molecular weight (MW) ⫽ 60], creatinine (MW ⫽ 113), 2-micro-
globulin (2-MG, MW ⫽ 11,800) and ␣1-microglobulin (␣1-MG, MW ⫽ 33,000) were clini-
cally measured in hemodialysis (HD) and in online predilution HDF mode with various
substitution fl ow rates (QS) from 180 to 260 ml/min. Blood fl ow rates (QB) ranged from 200
to 230 ml/min while the total dialysate fl ow rate (QDtotal) was fi xed at 520 ml/min, which was
divided into QS and net dialysate fl ow rate (QDnet) in all clinical studies.
The sieving coeffi cients (SC) for albumin were measured in vitro with 6 commercial
dialyzers (table 1), the membranes of which included various amounts of PVP. Measurements
were done at 37°C with 2 liters of phosphate buffer solution (pH ⫽ 7.4) with bovine albumin
(Wako Pure Chemical Ind. Co., Osaka, Japan) in a glass container. The fl ow rate of the test
solution (pseudo-blood) was 200 ml/min, and the ultrafi ltration rate (QF) was fi xed at 10 ml/
min. The ultrafi ltrate and the test solutions were returned to the glass container during the
course of the experiment for 630 min. Samples were taken at various times, and concentra-
tions were analyzed by either spectrophotometry or HPLC.
Yamashita 52
Theoretical Background
Dialyzer clearances (CL) were calculated for various solutes (urea, creatinine, 2-MG,
␣1-MG) using the following equation:
CCC
CQQ
L
Bi Bo
Bi
Bo F
=−+ (1)
where CBi, CBo are concentrations in the test solution at the inlet and outlet of the dialyzer,
respectively, and QBo is the blood fl ow rate at the outlet of the dialyzer.
The defi nition of the SC is the ratio between the concentration in the downstream to that
in the upstream, and the defi nitive equations may be found elsewhere. We used the following
semiquantitative defi nition SC4, proposed by the authors [4, 5]:
SC C
CC
F
Bi Bo
4= (2)
where CF is the solute concentration in the ultrafi ltrate.
Results and Discussion
Although QF was changed within a certain range (20–280 ml/min), clear-
ances for urea and creatinine in vivo increased sharply with QB, implying that
clearances for these solutes were strongly dependent on QB. The clearance for
␣1-MG, however, never showed a sharp increase with QB but showed a relatively
large deviation, caused by the change in QF. This implied that the transport of
large solutes cannot be controlled only by QB.
Table 1. Technical specifi cations of investigated ultrafi lters
Name Abbreviated
name
Surface
area, m2
Membrane
materials
Hydrophilic
agent
Pore size Manufacturer
PS-1.6UW PS 1.6 PS PVP⫹⫹⫹ NA Fresenius-
Kawasumi,
Tokyo, Japan
FLX-15GW FLX 1.5 PEPA PVP⫺standard
FDX-15GW FDX 1.5 PEPA PVP⫹standard
FDY-15GW FDY 1.5 PEPA PVP⫹larger Nikkiso Co.,
Tokyo, Japan
FDX-150GW new FDX 1.5 PEPA PVP⫹⫹ standard
FDY-150GW new FDY 1.5 PEPA PVP⫹⫹ larger
PS ⫽ Polysulfone; ⫺ ⫽ no PVP; ⫹ ⫽ small amount; ⫹⫹ ⫽ increased amount; ⫹⫹⫹ ⫽ large amount.
Solute and Fluid Removal in HDF 53
Figure 1a shows another clinical relationship between solute clearances
and QS in online predilution HDF treatments with a fi xed blood fl ow rate
QB ⫽ 200 ml/min. With the increase in QS, clearances for small solutes such as
urea and creatinine decreased slightly but steadily, while clearances for 2-MG
and ␣1-MG increased. In online treatments in general, since QDtotal is usually
fi xed at some value and was fi xed at 520 ml/min in these treatments, an increase
in QS will decrease QDnet (QDnet ⫽ QDtotal ⫺ QS). Then the abscissa of fi gure 1a
is replaced by QDnet to yield fi gure 1b, which is a mirror image of fi gure 1a.
Since urea and creatinine transport is a diffusion-limited process, a decrease
in QDnet has a more crucial negative effect on clearances of these solutes than
an increase in QS (or QF), which should have a positive effect on clearances.
However, since transport of 2-MG-and ␣1-MG is a bulk fl ow-limited (convec-
tion-limited) process, for clearances of these solutes an increase in QS (or QF)
is much more effective than a decrease in QDnet, which should have an adverse
effect. Since transport phenomena of this kind never happen in postdilution
HDF (neither online nor off-line) [6], great attention should be paid to the tar-
get solutes that should be removed by the treatment in online predilution HDF.
There is still some debate on dilution modes because in terms of solute removal
it is in general better to treat patients with postdilution HDF. Canaud et al. [7],
however, showed that solute removal performance in predilution can match that
in postdilution. Moreover, Ahrenholz et al. [8] showed that the rate of albumin
loss can exceed 7,000 mg/session in postdilution HDF; in that case, it is safer to
perform HDF in predilution mode. Since recent dialyzers have high hydraulic
200
150
Clearance (ml/min)
100
50
0
160
ab
180 200 220 240 260 280
QS (ml/min)
200
150
Clearance (ml/min)
100
50
0
240 260 280 300 320 340 360
QDnet (ml/min)
Urea
Creatinine
2-MG
␣1-MG
Fig. 1. Clearances for various solutes in online predilution HDF. QB ⫽ 200 ml/min,
QDtotal ⫽ 520 ml/min.
Yamashita 54
permeability as well as high solute permeability, ultrafi ltration from the blood
compartment to the dialysate and reverse fi ltration from the dialysate to the
blood are induced in one dialyzer at the same time, which is termed the inter-
nal fi ltration [9]. There are several commercial dialyzers that are designed to
increase the internal fi ltration rate. With an increased internal fi ltration similar
(on a limited scale) to postdilution HDF, albumin loss may also be increased,
and this should therefore be monitored. The internal fi ltration is suppressed by
increasing QF; therefore, it is minimized in predilution HDF that induces by
far the largest QF. Consequently, the reduced albumin loss in predilution HDF
is caused not only by the low albumin concentration due to diluted blood but
also by the suppressed internal fi ltration due to extremely high QF. Reducing
albumin loss without changing clearances for other solutes may in many cases
fi t clinical requirements.
The time courses of SC4 for albumin, using a PS-1.6UW dialyzer (PS,
Fresenius-Kawasumi Co., Tokyo, Japan) in aqueous solution with various albu-
min concentrations, showed strong time-dependent patterns with peak values
approximately 10 min after start of the experiments. The lower the albumin con-
centration, the higher the values and the longer the time required for achieving
steady state. Moreover, Ahrenholz et al. [8] pointed out that as much as 50% of
albumin loss occurred within the fi rst 30 min of the treatment. From these facts,
large albumin losses may be expected at the beginning of each treatment with
such membranes that take the peak SC values [10, 11].
The SC4 for albumin of 3 PEPA dialyzers, FLX, FDX and FDY gradually
increased with time and never reached peak values. The membranes used in
the FLX and FDX had the same pore size, the only difference being that the
latter contained a small amount of PVP on the inner surface of the membrane.
The fact that the FDX showed much lower SC4 values than the FLX can be
explained in the following way. The membrane used in the FDX soon reached
the adsorption saturation of albumin due to reduced hydrophobic interaction.
In contrast, since the FLX had higher adsorption characteristics to albumin, the
albumin concentration in the test solution (CBi and CBo) drastically decreased,
which decreased the value of the denominator of the SC4, while the numera-
tor did not change much because adsorbed albumin molecules may have been
slowly released from the membrane. The membrane material used in the FDX
is the same as the one used in the FDY, which had a 5% larger pore diameter
analyzed by the classic pore theory [12]. By enlarging the pore diameter in the
FDY, the SC4 increased in accordance with the enlargement.
The time courses of SC4 for albumin of the latest models of PEPA dia-
lyzers (new FDX and new FDY with PVP⫹⫹) showed peak values at 6 min
after starting the experiments. The peaks found with the new PEPA membranes
were quite similar to the one found with the PS (PVP⫹⫹⫹) dialyzer, which
Solute and Fluid Removal in HDF 55
may be because of the increased amount of PVP. We have already reported
that an increase in PVP may induce larger changes in C3a concentration in
vivo [11]. Then the peak SC4 values for albumin and the blood compatibility
of the membrane may be related to the amount of PVP or hydrophilicity of the
membrane.
Since the albumin concentrations of the test solutions were lower by a fac-
tor of 1/30–1/10 than the standard albumin concentration in human blood (3.6–
4.0 g/dl), the SC4 values for albumin shown above do not correspond directly
to the clinical results. One should, however, consider the membrane separation
characteristics that depend on the membrane materials, including both the main
material and hydrophilic agents, as well as the experimental conditions.
Conclusions
Solute removal characteristics in online predilution HDF were dem-
onstrated with a fi xed total dialysate fl ow rate. Clearances for small solutes
decreased with increasing QF, and clearances for 2-MG-and ␣1-MG increased
with increasing QF.
Since the SC for albumin reached a peak value at the beginning of the
experiment with those membranes containing relatively large amounts of PVP
(less adsorption), a large amount of albumin loss may be expected at the begin-
ning of the treatment.
The amount of PVP used in the membrane may be closely related to bio-
compatibility as well as the solute removal characteristics.
References
1 Sprenger KB: Haemodiafi ltration. Life Support Syst 1983;1:127–136.
2 Ofsthum NJ, Leypoldt JK: Ultrafi ltration and backfi ltration during hemodialysis. Artif Organs
1995;19:1143–1161.
3 Leypoldt JK: Solute fl uxes in different treatment modalities. Nephrol Dial Transplant 2000;1:3–9.
4 Yamashita AC, Sakiyama R, Hamada H, Tojo K: Two new defi nitive equations of the sieving coef-
fi cient. Kidney Dial (Jin To Toseki) 1998;45:S36–S38.
5 Yamashita AC: New dialysis membrane for removal of middle molecule uremic toxins. Am J
Kidney Dis 2001;38(suppl 1):S217–S219.
6 Masakane M: Selection of dilutional method for on-line HDF, pre- or post-dilution. Blood Purif
2004;22(suppl 2):49–54.
7 Canaud B, Levesque R, Krieter D, Desmeules S, Chalabi L, Moragues H, Morena M, Cristol JP:
On-line hemodiafi ltration as routine treatment of end-stage renal failure: why pre- or mixed dilu-
tion mode is necessary in on-line hemodiafi ltration today? Blood Purif 2004;22(suppl 2):40–48.
8 Ahrenholz PG, Winkler RE, Michelsen A, Lang DA, Bowry SK: Dialysis membrane-dependent
removal of middle molecules during hemodiafi ltration: the b2-microglobulin/albumin ratio. Clin
Nephrol 2004;62:21–28.
Yamashita 56
9 Dellana F, Baldamus CA: Does internal-fi ltration have a benefi t? Abstr Meet ‘Quality care for
haemodialysis patients’, Wiesbaden, May 11–13, 1995, p 51.
10 Yamashita AC, Tomisawa N, Takezawa A, Sato Y: Separation characteristics of newly developed
polymer alloy membrane. Proc 8th Jpn Int SAMPE Symp, Tokyo, 2003, pp 415–418.
11 Yamashita AC, Tomisawa N, Takesawa A, Sakurai K, Sakai T: Blood compatibility and fi ltration
characteristics of newly developed polyester polymer alloy (PEPA) membrane. Hemodial Int
2004;8:373–337.
12 Pappenheimer JR, Renkin EM, Borrero LM: Filtration, diffusion and molecular sieving through
peripheral capillary membranes – a contribution to the pore theory of capillary permeability. Am
J Physiol 1951;167:13–46.
Prof. Akihiro C. Yamashita, PhD
Department of Materials Science and Engineering
College of Engineering, Shonan Institute of Technology
1–1–25 Tsujido-Nishikaigan
Fujisawa, Kanagawa 251-8511 (Japan)
Tel./Fax ⫹81 466 30 0234, E-Mail yama@la.shonan-it.ac.jp
Ronco C, Canaud B, Aljama P (eds): Hemodiafi ltration.
Contrib Nephrol. Basel, Karger, 2007, vol 158, pp 57–67
Membranes and Filters for
Haemodiafiltration
Nicholas A. Hoenich
School of Clinical Medical Sciences, Faculty of Medical Sciences,
Newcastle University, Newcastle upon Tyne, UK
Abstract
Online haemodiafi ltration is an extracorporeal technique, utilizing highly permeable
and highly biocompatible membranes, which permits the combination of convective and dif-
fusive solute removal from the blood and offers increased removal of medium-weight urae-
mic solutes, compared to the more frequently used low- and high-fl ux haemodialysis. The
objective of this chapter is to review the membranes and fi lters available for haemodiafi ltra-
tion and to discuss factors that infl uence their performance during clinical use.
Copyright © 2007 S. Karger AG, Basel
The treatment of patients with chronic kidney disease in the absence of a
suitable kidney for transplantation relies upon the use of artifi cial support to sus-
tain life. The most widely used support modality is haemodialysis, a treatment
in which non-protein-bound solutes from the blood are removed by diffusion
into an electrolyte solution (dialysis fl uid) across a semi-permeable membrane.
This treatment modality is highly effi cient in removing small-molecular-weight
uraemic toxins, but offers only a limited clearance of larger-molecular-weight
substances, some of which, e.g. 2-microglobulin, are associated with the
development of long-term treatment complications. Improved removal of such
compounds can be achieved by convective therapies such as haemofi ltration,
a treatment modality introduced into clinical practice in the mid-1970s, using
hollow-fi bre devices containing highly permeable membranes.
During haemofi ltration, plasma water is fi ltered across a highly perme-
able membrane at a rate far beyond that needed to normalize intertreatment
fl uid gain. The excess fl uid removed is replaced with a sterile replacement fl uid
infused into the extracorporeal circuit either before (pre-dilution) or after (post-
Membranes and Hardware for Hemodiafiltration
Hoenich 58
dilution) the fi lter, with early treatments using prepared sterile replacement
fl uid. Owing to cost considerations, the volume of the fl uid used was typically
5–10 litres, up to a maximum of 15 litres.
As a result of water movement, solutes able to pass through the pores of the
membrane are dragged across the membrane independently of their molecular
size, provided this size is less than the size of the membrane pores. Consequently,
this treatment offered a higher rate of medium- and large-molecular removal
than conventional haemodialysis, but removal of small-molecular compounds
was inferior. Leber et al. [1], in the late 1970s, proposed combining haemodi-
alysis and haemofi ltration into a single treatment (haemodiafi ltration). As in
haemofi ltration, early applications relied on the use of prepared sterile infusate.
Today, technological advances permit the online production of fl uid derived
from bicarbonate-buffered dialysis fl uid to be used for the infusate, allowing
safe use of high-exchange volumes (online haemodiafi ltration) [2].
In this chapter, the membranes and fi lters used in haemodiafi ltration will
be reviewed, together with their functional performance, and the factors infl u-
encing performance during clinical use will be discussed.
Membrane Materials
In the early 1970s, membrane requirements focused on small-molecular
clearance and hydraulic permeability or ultrafi ltration; since the 1980s, the
emphasis has shifted to the clearance of larger molecules and fl ux. Today, major
manufacturers produce families of membranes or devices suitable for conven-
tional haemodialysis, high-fl ux dialysis, haemofi ltration or haemodiafi ltration.
Classifi cation of the membranes may be according to their hydraulic permeabil-
ity or their chemical composition (fi g. 1).
In contrast to membranes intended for use in conventional low-fl ux
haemodialysis, which have a permeability to water of 5–6 ml/h ⭈ mm Hg ⭈ m2,
membranes intended for use in haemodiafi ltration have higher permeabilities
(30–40 ml/h ⭈ mm Hg ⭈ m 2), and additionally exhibit a high removal rate for 2-
microglobulin, a compound implicated in the evolution of long-term compli-
cations associated with haemodialysis treatment, and which has become an
important surrogate parameter of dialysis effi ciency regarding medium-mol-
ecule removal. High-permeability membranes in current clinical use are not
selective, and whilst a certain degree of selectivity can be achieved by the
manipulation of the membrane structure during manufacture [3], such selec-
tivity is not optimal and is associated with some albumin loss. Short-term
clinical experience with such membranes suggests improved anaemia correc-
tion, decreased total plasma homocysteine concentrations and reduced plasma
Membranes and Filters for Haemodiafi ltration 59
concentrations of glycosylated and oxidized proteins, but it is not yet clear
whether the routine use of such membranes is warranted [4]. Furthermore, whilst
it can be speculated that a considerable albumin loss across the membrane may
lead to hypo-albuminaemia and malnutrition, at least in those patients who are
unable to supply a suffi cient protein intake, it remains impossible to quantify
an acceptable upper limit for albumin loss for extracorporeal renal replacement
therapies or to defi ne the optimum membrane permeability.
Haemodiafilter Design
Today, the most commonly used haemodialyser is the hollow-fi bre or cap-
illary design. Such a design can be subject to limitations in respect of an even
fl ow through and around the fi bres [5]. Nevertheless, it offers considerable
manufacturing fl exibility, enabling devices of a range of sizes to be manufac-
tured for use in a range of clinical applications merely by altering the number,
length or type of fi bre, and all major manufacturers of haemodialysers produce
Synthetically
modified cellulose
Cuprophan®
SCE
Cuprammonium rayon
Hemophan®
SMC®
Cellulose diacetate
Cellulose triacetate
Excebrane®
PEG-RC
Diapes®
Polyamix®
Fresenius polysulfone®
␣-Polysulfone
DIAPES®
PEPA®
EVAL®
AN69® AN69ST
PMMA
Polyamix®
Fresenius Polysulfone®
␣-Polysulfone
Arylane®
PAN
Toraysulfone®
DIAPES®
AN69® AN69ST
Polyamix®
Toraysulfone®
␣-Polysulfone
Rexeed®
PEPA®
Fresenius Polysulfone®
PAN
Arylane®
PMMA
Cellulose diacetate
Cellulose triacetate
UFC
10–20 ml/h· mm Hg
Unmodified
cellulose Synthetic
CTA
UFC
⬍10 ml/h· mm Hg
UFC
⬎20 ml/h· mm Hg
Fig. 1. Membranes suitable for haemodiafi ltration classifi ed according to their chemi-
cal composition and hydraulic permeability. UFC ⫽ Ultrafi ltration coeffi cient.
Hoenich 60
hollow-fi bre devices suitable for haemodiafi ltration in either pre- or postfi lter
dilution mode (fi g. 2a).
A deviation from this approach is the Olpur MD 190H mid-dilution fi lter
(Nephros Inc., New York, N.Y., USA). This has a circular groove incorporated into
the fi bre bundle tube at one end, such that when the header cap is added, two dis-
crete but serially connected paths for blood are formed: an outer or annular path,
and an inner or core path. The header cap splits the fi bre bundle and also functions
as a mixing chamber for the infused substitution fl uid. The other end of the device
uses a dual port header which acts as an inlet and outlet manifold for the blood. The
dialysis fl uid enters and leaves the device in the conventional manner (fi g. 2b).
Fundamentals of Performance
Solute Transport
Solute transport across the membrane can occur via diffusion or convec-
tion. Diffusive solute transport is the transport in the presence of a concentration
a
Fig. 2. a Modern hollow-fi bre devices, suitable for use as haemodiafi lters, utilizing
fi bres with a three-dimensional microwave structure, incorporated into a specifi cally designed
housing for optimized fl ow distribution in both the blood and dialysate pathways (photograph
courtesy of Fresenius Medical Care AG, Bad Homburg, Germany).
Fig. 2. b The Olpur 190 Mid-dilution fi lter for haemodiafi ltration (photographs cour-
tesy of Nephros Inc., New York, N.Y., USA).
Membranes and Filters for Haemodiafi ltration 61
Blood IN
Blood IN
Blood OUT
Blood OUT
Annular region (stage 1)
Circular groove
Core region (stage 2)
Substitution
fluid IN
Dialysate IN Dialysate OUT
End view
Internal
Wall
End view
Substitution fluid
header cap
removed
b
2-Port
Header cap
Blood
Header cap
removed
Hoenich 62
gradient. It is governed by Fick’s law, which can be expressed mathematically
as
JDA
dC
dx
D=−
Where JD is the diffusive fl ux, D is the solute diffusion coeffi cient, A the area
available for transport, and dC/dx the concentration gradient.
Convective solute transport is a consequence of fi ltration of fl uid through
the membrane. All solutes that can pass through the pores of the membrane, i.e.
which are sieved by the membrane, are carried along by the fi ltered fl uid. Thus,
the sieving properties of the membrane determine what is removed. In contrast
to diffusive transport, convective transport remains constant over a wide range
of molecular sizes but decreases as the molecular size approaches that of the
pores. Mathematically, it can be expressed as
JQCS
CF
=
Where JC is the convective fl ux, QF the rate of fl uid transfer across the mem-
brane, C the solute concentration in the fl uid, and S the sieving coeffi cient, a
parameter that represents the magnitude of restriction of the pore size relative to
the molecular size and varies between 0 for a freely permeable molecule and 1
for a completely impermeable molecule.
Solute transport in haemodiafi ltration occurs as a consequence of both dif-
fusion and convection. Early theoretical approaches assumed that the diffusive
and convective mass transfer took place sequentially, and this approach remains
applicable to the paired fi ltration dialysis system, a variant of online haemodia-
fi ltration [6].
In the presence of simultaneous convective and diffusive mass transport,
the combined solute transport is not the sum of the individual components,
because of an interaction between the convective and diffusive components.
Several models have been proposed to explain this mathematically, of which
the simplest is
KHDF ⫽ K 0 ⫹ Q FT
where K0 is the clearance at zero ultrafi ltration, QF the ultrafi ltration rate and T
the transmittance coeffi cient, a parameter which is a function of the fl ow condi-
tions and membrane properties.
An expression for the transmittance coeffi cient that is universal for all sol-
utes has been proposed by Jaffrin et al. [7], namely
KHDF ⫽ K 0 ⫹ 0.46QF
for ultrafi ltration rates below 70 ml/min, which is modifi ed to
Membranes and Filters for Haemodiafi ltration 63
KHDF ⫽ K 0 ⫹ 0.43QF ⫹ 0.00083QF2
for ultrafi ltration rates above 70 ml/min.
Hydraulic Permeability
The hydraulic permeability (Lp) of the membrane is a parameter which
refl ects the relationship between the volumetric fl ux and the transmembrane
pressure difference (⌬P). In the absence of an osmotic pressure difference the
volumetric fl ux (JV) can be expressed mathematically as
JV ⫽ L p⌬P
which in the presence of an osmotic pressure difference becomes
JV ⫽ L p⌬P ⫺ L p⌬s
where ⌬s is the osmotic pressure difference across the membrane and
the Staverman or osmotic refl ection coeffi cient, whereby a value of 0 applies
to a non-selective membrane, and a value of 1 applies to an impermeable
membrane.
Biocompatibility
Extracorporeal circulatory procedures involve repeated exposure of the
patient’s blood to foreign materials, of which the membrane represents the larg-
est contact surface. Such exposure is associated with a number of biological
sequelae, the magnitude of which is determined to a large extent by the mem-
brane’s surface characteristics; however, in the case of high-fl ux membranes,
transmembrane transport of activated components also contributes [8, 9].
Considerable uncertainty exists as to the long-term clinical impact of
membrane biocompatibility; however, what is beyond doubt is that clinical
application of haemodiafi ltration exposes patients to highly biocompatible
membranes which are used in conjunction with ultrapure dialysis fl uid and
infusion solutions, offering the highest possible treatment-associated biocom-
patibility [10].
Performance of Haemodiafilters
The laboratory performance of haemodiafi lters is generally measured by
manufacturers in accordance with an internationally recognized standard: ISO
8637:2004 – Cardiovascular implants and artifi cial organs – Haemodialysers,
haemodiafi lters, haemofi lters and haemoconcentrators. Table 1 summarizes
comparative data for currently produced devices in terms of urea clearance and
2-microglobulin and protein sieving coeffi cients.
Hoenich 64
Operational Factors Influencing Efficiency
Solute Transport
The effi ciency of haemodiafi ltration is infl uenced, as for conventional
haemodialysis, by the blood and dialysate fl ow rates and the red cell concentra-
tion. In addition, solute transport effi ciency is also infl uenced by the site of the
infusion fl uid.
In postdilution, the clearance of both small and medium molecules is
increased compared to conventional haemodialysis, and can be considered as
the most effi cient method of haemodiafi ltration. In this modality, however, as
the infusion fl uid is added after the fi lter, the blood fl owing through the fi lter is
subject to haemoconcentration. Therefore, to mimimize the risk of compromis-
ing the extracorporeal circuit through clotting, the total ultrafi ltration rate (i.e.
the sum of the substitution fl uid replacement rate and the net ultrafi ltration)
should not exceed 30% of the blood fl ow rate to ensure that the postfi lter hae-
matocrit does not exceed 50%.
Table 1. Comparative performance characteristics of hollow-fi bre devices suitable for haemodiafi l-
tration
Device Area
m2
Membrane Clearance performance
ml/mina
UFC ml/h
⭈ mm Hgb
Sieving coeffi cient
urea creatinine PO4 β2-microglobulin albumin
Polyfl uxS 1.7 Polyamix 254 229 223 71 0.63 ⬍0.01
280c262c258c
PF-170H 1.7 Purema 274 259 240 74 0.8 NA
FX80 1.8 Helixone 276 250 239 59 0.8 0.001
Rexeed 18A 1.8 Rexbrane 280 265 250 71 0.85 0.002
Olpur MD
190H
1.9 Polyethersulfone 276 264 257 90 0.8 0.005
Polyfl uxS 2.1 Polyamix 267 245 240 83 0.63 ⬍0.01
280c262c258c
PF-210H 2.1 Purema 285 272 253 80 0.8 NA
Rexeed 21A 2.1 Rexbrane 284 272 257 74 0.85 0.002
FX100 2.2 Helixone 278 261 248 73 0.8 0.001
Data shown extracted from manufacturers’ product specifi cation sheets and measured in accordance with
ISO 8637. NA ⫽ Not given in product sheet; UFC ⫽ ultrafi ltration capacity.
a At a blood fl ow rate of 300 ml/min, dialysis fl uid fl ow rate 500 ml/min, ultrafi ltration rate 0 ml/min.
b Measured using bovine blood.
c At a blood fl ow rate of 300 ml/min, dialysis fl uid fl ow rate 500 ml/min, ultrafi ltration rate 60 ml/min.
Membranes and Filters for Haemodiafi ltration 65
In predilution HDF, where the substitution fl uid is added to the blood
before the fi lter, a dilution of the blood entering the fi lter occurs. This ensures
better rheological conditions, and higher convective or large-molecular clear-
ances, but this advantage is offset by the dilution of the concentration avail-
able for diffusion, resulting in reduced small-molecular clearance, to a level
below that achieved in conventional haemodialysis [11]. In an online setting,
this approach, however, permits infusion rates of up to 10–12 l/h to be used,
which can prove useful when treating patients with high haemoglobin levels or
when high exchange volumes are required.
Simultaneous pre- and postfi lter dilution has also been proposed, but has
not been used clinically. Instead, the mid-dilution approach with its specially
designed fi lter appears to combine the high clearance of small-molecular-weight
toxins (such as urea) associated with postdilution haemodiafi ltration with the
high clearance of medium-weight molecules (such as 2-microglobulin) associ-
ated with predilution haemodiafi ltration [12].
Irrespective of the strategy used, an important contributing element to the
removal of large-molecular-weight compounds such as 2-microglobulin is
adsorption to the membrane, a characteristic which differs between membranes
[13]. Therefore, the future development of membrane materials with a greater
affi nity for adsorption of 2-microglobulin could theoretically result in higher
removal rates than achieved by the combination of convection and diffusion.
A logical extension of this is the future immobilization of proteins on the sur-
face of the membrane to act as immunoadsorptive devices for this and other
compounds.
Hydraulic Permeability
Membrane permeability is a key determinant of performance; in vivo,
the hydraulic permeability is affected by the formation of a protein layer on
the membrane surface, reducing ultrafi ltration, and convective fl uxes infl u-
encing the exchange volumes are possible as well as the convective solute
removal.
Adequacy of Treatment when Using Haemodiafiltration
In haemodiafi ltration, the additional fl ux compared to low- or high-fl ux
haemodialysis treatments increases removal of small-molecular-weight com-
pounds and facilitates the attainment of a high Kt/V. Additionally, removal
of medium-molecular-weight compounds such as 2-microglobulin is also
increased, with the increase infl uenced by the volume of the substitution
fl uid used [14]. Thus, with a typical treatment duration of 240 min, and a
Hoenich 66
blood fl ow rate of 300 ml/min, postfi lter substitution volumes of 22 litres are
potentially possible when maintaining the ultrafi ltration rate of 30% of the
blood fl ow rate, rising to 29 litres at a blood fl ow rate of 400 ml/min. Whilst
theoretically it would be possible to use higher exchange volumes with the
use of prefi lter infusion, this would require a longer treatment period to
compensate for the reduction of small-molecular clearance. Increasing the
clearance of molecules such as 2-microglobulin is subject to physiological
constraints as a recent study by Ward et al. [15] indicated. Consequently,
higher removal may require alternative strategies such as increased treatment
times or frequency of treatment, to further reduce plasma 2-microglobulin
concentrations.
References
1 Leber HW, Wizemann V, Goubeaud G, Rawer P, Schutterle G: Hemodiafi ltration: a new alternative
to hemofi ltration and conventional hemodialysis. Artif Organs 1978;2:150–153.
2 Aires I, Matias P, Gil C, Jorge C, Ferreira A: On-line haemodiafi ltration with high volume substi-
tution fl uid: long-term effi cacy and security. Nephrol Dial Transplant 2007;22:286–287.
3 Ronco C, Bowry S: Nanoscale modulation of dimensions, size distribution and structure of a new
polysulfone-based high-fl ux dialysis membrane. Int J Artif Organs 2001;24:726–735.
4 Ward RA: Protein-leaking membranes for hemodialysis: a new class of membranes in search of an
application? J Am Soc Nephrol 2005;16:2421–2430.
5 Ronco C, Levin N, Brendolan A, Nalesso F, Cruz D, Ocampo C, Kuang D, Bonello M, De Cal M,
Corradi V, Ricci Z: Flow distribution analysis by helical scanning in polysulfone hemodialyzers:
effects of fi ber structure and design on fl ow patterns and solute clearances. Hemodial Int 2006;10:
380–388.
6 Ghezzi PM, Dutto A, Gervasio R, Botella J: Hemodiafi ltration with separate convection and
diffusion: paired fi ltration dialysis; in D’Amico G, Colasanti G (eds): Clinical Nephrology:
Immunologic Considerations, Invasive Techniques and Dialytic Strategies. Contrib Nephrol.
Basel, Karger, 1989, vol 69, pp 141–161.
7 Jaffrin MY, Ding LH, Laurent JM: Simultaneous convective and diffusive mass transfers in a
hemodialyser. J Biomech Eng 1990;112:212–219.
8 Kaiser JP, Oppermann M, Gotze O, Deppisch R, Gohl H, Asmus G, Rohrich B, von Herrath D,
Schaefer K: Signifi cant reduction of factor D and immunosuppressive complement fragment Ba
by hemofi ltration. Blood Purif 1995;13:314–321.
9 Gasche Y, Pascual M, Suter PM, Favre H, Chevrolet JC, Schifferli JA: Complement depletion dur-
ing haemofi ltration with polyacrilonitrile membranes. Nephrol Dial Transplant 1996;11:117–119.
10 Macleod AM, Campbell M, Cody JD, Daly C, Donaldson C, Grant A, Khan I, Rabindranath KS,
Vale L, Wallace S: Cellulose, modifi ed cellulose and synthetic membranes in the haemodialysis of
patients with end-stage renal disease. Cochrane Database Syst Rev. 2005;3:CD003234.
11 Wizemann V, Kulz M, Techert F, Nederlof B: Effi cacy of haemodiafi ltration. Nephrol Dial
Transplant 2001;16(suppl 4):27–30.
12 Santoro A, Conz PA, De Cristofaro V, Acquistapace I, Gaggi R, Ferramosca E, Renaux JL, Rizzioli
E, Wratten ML: Mid-dilution: the perfect balance between convection and diffusion; in Ronco
C, Brendolan A, Levin NW (eds): Cardiovascular Disorders in Hemodialysis. Contrib Nephrol.
Basel, Karger, 2005, vol 149, pp107–114.
Membranes and Filters for Haemodiafi ltration 67
13 Padrini R, Canova C, Conz P, Mancini E, Rizzioli E, Santoro A: Convective and adsorptive
removal of 2-microglobulin during predilutional and postdilutional hemofi ltration. Kidney Int
2005;68:2331–2337.
14 Lornoy W, Becaus I, Billiouw JM, Sierens L, Van Malderen P, D’Haenens PR: On-line haemo-
diafi ltration. Remarkable removal of 2-microglobulin. Long-term clinical observations. Nephrol
Dial Transplant 2000;15(suppl 1):49–54.
15 Ward RA, Greene T, Hartmann B, Samtleben W: Resistance to intercompartmental mass transfer
limits 2-microglobulin removal by post-dilution hemodiafi ltration. Kidney Int 2006;69:1431–
1437.
Nicholas A. Hoenich
School of Clinical Medical Sciences
Faculty of Medical Sciences, Newcastle University
Newcastle upon Tyne NE2 4HH (UK)
Tel. ⫹44 191 222 6998, Fax ⫹44 191 222 0723, E-Mail nicholas.hoenich@ncl.ac.uk
Ronco C, Canaud B, Aljama P (eds): Hemodiafi ltration.
Contrib Nephrol. Basel, Karger, 2007, vol 158, pp 68–79
Technical Aspects of Online
Hemodiafiltration
Hans-Dietrich Polaschegga, Thomas Royb
aMedical Devices Consultant, Köstenberg, Austria; bResearch and Development,
Fresenius Medical Care Deutschland GmbH, Bad Homburg, Germany
Abstract
The chapter describes and analyzes the most commonly used systems for online substi-
tution fl uid production in hemofi ltration and hemodiafi ltration, covering in detail systems
comprising 2 multiuse fi lters, or 2 multiuse fi lters in combination with 1 disposable fi lter. A
generally applicable risk analysis for the process of online substitution fl uid preparation is
provided. It is concluded that both system variants discussed perform in a safe manner, pro-
vided the operator fully complies with the manufacturer’s instructions for use.
Copyright © 2007 S. Karger AG, Basel
Many scientists, clinicians and technologists in Japan, the USA and Europe
have contributed to the development of hemofi ltration (HF) and hemodiafi ltra-
tion (HDF) with online-produced substitution fl uid. Today’s most widely used
commercial systems were developed by the companies Gambro and Fresenius
Medical Care using various technologies. These systems formed the basis upon
which other companies designed their systems. For this reason these most com-
mon systems will be described in detail, while the later developed systems will
be discussed only briefl y, using the basic systems as a reference.
Over the years, various systems for online HDF from various manufactur-
ers have been used clinically. However, the routine application of online HDF
is technically dominated by equipment from Fresenius Medical Care (FME),
Germany, and Gambro AB, Sweden. Both companies entered the fi eld of online
fl uid preparation before others, and provided online HDF equipment fi rst on
request from enthusiastic doctors and later as a regular option for all their main-
stream hemodialysis (HD) machines (Gambro: AK series; FME: 2008 and 4008
series).
Technical Aspects of Online Hemodiafi ltration 69
The trend to use online HDF as a cost-effi cient, highly effective treatment
mode for end-stage renal disease was low before the 1990s owing to uncer-
tainties regarding the regulatory situation of these systems. However, with the
so-called Medical Devices Directive [1], coming into effect in 1993, and sub-
sequent interpretations of this directive [2], online HDF equipment including
consumables were classifi ed as medical devices and – in consequence – could
be CE marked and distributed in the European Community, leading to increased
application of this modality. For the FME 5008, the most current FME HD
machine, online HDF is part of the standard device confi guration.
Gambro and FME online HDF equipment, exemplifi ed in the following
paragraphs, underwent various modifi cations since the mid-1980s on the way
from prototype-like experimental devices to the current state of professionally
made medical equipment, but still represent the 2 most common approaches
to solve the problem of safe and effi cient online preparation of intravenously
injectable substitution fl uid.
Both systems are adapted to the specifi c properties of the underlying HD
machines, in particular to the design of the respective dialysate circuits. Other
online HDF machines that were developed later employ principles used by
either the FME or the Gambro system or both.
In general it can be said that both systems, to the authors’ knowledge, have
a clean safety record. According to FME internal reports, between the years
2000 and 2006 approximately 15 million single online HDF treatments were
performed with the current technical system (brand name Online plusTM) with-
out any report of patient injury due to failures of the specifi c online fl uid prepa-
ration components of the device.
Gambro HD machines are traditionally characterized by an ultrafi ltration
control system comprising 2 electromagnetic fl ow meters before and behind the
dialyzer (fi g. 1). Fluid balance is regulated by means of 2 fl ow restrictors and 2
pumps installed before and behind the fl ow cell unit.
Incoming process water is fi ltered through an ultrafi lter (type U8000S, sur-
face area 2.1 m2, membrane material polyamide S). The water fi lter is operated in
cross-fl ow mode, in which a small amount of fl uid is continuously drained off to
prevent accumulation of possible contaminants on the inlet side of the fi lter. In the
next step the 2 concentrate components for bicarbonate dialysate are added. In the
context of ‘intended use’, the manufacturer requires for the device that the incom-
ing water complies with at least the Association for the Advancement of Medical
Instrumentation standard for HD water quality [3], and that the bicarbonate concen-
trate is generated in situ by the machine from powder to avoid the hazards of liquid
bicarbonate concentrate, which easily develops major bacterial contamination.
After passing the fl ow cell, the now ready-to-use dialysate is fi ltered
through another U8000S ultrafi lter, this one running in dead-end mode. A valve
Polaschegg/Roy 70
in the cross-fl ow path of the fi lter is operated only during the cleaning and disin-
fection cycle of the machine. Both U8000S fi lters are multiple-use components,
are included in the cleaning and disinfection cycles of the machine, and are
routinely used for 1 month. However, depending on the microbiological test
results of the fl uid quality at the end of the routine fi lter lifetime, this period
may be extended [4].
Injectable substitution fl uid is generated from this purifi ed dialysate stream
by means of further fi ltration across a low-surface-area polyamide fi lter (0.2 m2).
This fi lter is an integral part of a sterile disposable set consisting of the fi lter, the
connection lines to the machine and to the extracorporeal circuit, and a roller
pump segment. Depending on the desired dilution mode, this fl uid is injected
into the bloodstream before or after the dialyzer.
The safe operation of online HDF systems depends on ultrafi lter integrity
during its entire lifetime. In the case of the system described here, the U8000S
fi lters are exposed to a pressure test procedure at the end of the manufacturing
Filter 1
(2.1 m2)
To patient
From patient
Filter 2
(2.1 m2)
Filter 3
(0.2 m2, disposable)
AAMI-standard
Water
Drain
Drain
A conc.
B conc.
(from powder)
Dialysate (filtered water,
unfiltered concentrates) Dialysate, 1 × filtered
Substitution fluid Dialysate, used
Flow cell
Infusate pump
Fig. 1. Gambro 3-fi lter hemodialysis system. Online HDF-related components and
simplifi ed fl ow scheme for Gambro AK 200 S Ultra. Components not required for online
HDF are not shown. AAMI ⫽ Association for the Advancement of Medical Instrumentation;
conc. ⫽ concentrate. Redrawn from HCEN9291: AK 200 Ultra S Service Manual, with kind
permission of Gambro Corporate R&D, Lund, Sweden.
Technical Aspects of Online Hemodiafi ltration 71
process. Filters showing a pressure drop exceeding a certain limit over time
(⬎0.3 bar at a test pressure of 1.6 bar) are discarded [5]. The third disposable
fi lter undergoes a similar test procedure during manufacturing and is guaran-
teed by the manufacturer.
Instead of using 3 fi lters for substitution fl uid preparation as described
above, it is possible to design systems with equal safety and performance on
the basis of 2 large-surface, multiple-use fi lters. The FME system (fi g. 2) takes
advantage of the design of the volumetric dialysate circuit, which is common
to nearly all HD machines of this manufacturer. The fl uid space behind the so-
called balancing chamber represents a precisely constant volume. In modern
HD machines, using extensive microprocessor control, it is possible to activate
all hydraulic components such as valves or pumps in an independent and highly
fl exible manner to run a variety of procedures and tests. These test capabilities
in particular are essential for the design and operation of a cost-effi cient 2-fi lter
system.
The FME system comprises 2 identical large-surface ultrafi lters (type
Diasafe® plus, surface area 2.2 m2, membrane material Fresenius Polysulfone®).
The fi lter housing and connector technology were specifi cally developed for
online fl uid preparation. In the context of ‘intended use’ the manufacturer
defi nes minimum quality requirements for process water, concentrates and
From patient
To patient
Balancing
chambers
Infusate pump
Filter 1
(2.2 m2)
Filter 2
(2.2 m2)
Dialysate, unfiltered Dialysate, 1 ⫻ filtered
Substitution fluid Dialysate, used
Fig. 2. FME 2-fi lter hemodialysis system. Online HDF-related components and sim-
plifi ed fl ow scheme for FME Online plus systems. Redrawn from Fresenius Medical Care
Online plus 7/07.03 (OP), Fresenius Medical Care, Bad Homburg, Germany.
Polaschegg/Roy 72
ready-to-use dialysate (water: ⬍100 CFU/ml and ⬍0.25 EU/ml endotoxin;
ready-to-use dialysate: ⬍1,000 CFU/ml, ⬍1 EU/ml endotoxin).
The input to the fi rst fi lter is ready-to-use dialysate. For reasons of dialysate
economy this fi lter is operated most of the time in dead-end mode. To prevent
a potentially hazardous buildup of contaminants in the fi lter, the inlet side is
frequently fl ushed to the drain via a valve. The fl uid from this fi rst fi lter passes
the inlet side of the second, identical fi lter, operated in cross-fl ow mode, and is
used to supply the dialyzer with the required fl ow.
Substitution fl uid is generated by drawing the required amount of fl uid
across the second fi lter using a standard roller pump on a simple disposable
tubing section consisting of 2 connectors, a pump segment and the required
tubing to the extracorporeal blood circuit for the desired mode of dilution (pre-,
post- or mixed dilution).
It must be emphasized here, and this is true for all systems used for online
substitution fl uid preparation, that connector technology is crucial for all com-
ponents in contact with the potentially intravenously injectable fl uid. Since it
has been known for many years that the standard dialysate connectors (Hansen
type) are prone to contamination and nearly impossible to clean and disinfect,
these connectors should not be found together with online HDF or online HF
systems. Instead, special types of connector and sealings have been developed,
mostly using the existing experience from peritoneal-dialysis connectors regard-
ing safety against touch contamination.
The nominal lifetime of the 2 fi lters is 12 weeks or 100 online HF/HDF
treatments. A fi lter must be changed when the predialysis fi lter integrity test
fails. The fi lters may be cleaned with sodium hypochlorite according to a
defi ned procedure. The HD machine monitors these procedures together with
fi lter lifetime and, for safety reasons, blocks all online specifi c functions apart
from standard dialysis in the case of deviations from the specifi ed regimen.
The design of a safe 2-fi lter system requires that the 2 fi lters are truly
redundant, which means that 1 fi lter alone has suffi cient retention capability
to protect the patient from any hazards by contamination, at least for a limited
period of time. This property must be validated during the approval procedure
for a given design.
Furthermore, it is essential for a safe 2-fi lter system that every treatment
starts with 2 functioning ultrafi lters. To verify this, the device must be able to
run a fi lter integrity test procedure before every treatment. For reasons of com-
fort and reliability this procedure should be automated.
Figure 3 demonstrates such an integrity test routine for the FME 2-fi lter
system [6]. The test makes use of the property of the fi lter membrane to let
fl uid pass but to block air. For test purposes the system is pressurized with
fi ltered air which displaces the fl uid from both fi lters. Once the respective
Technical Aspects of Online Hemodiafi ltration 73
fi lter compartments have been completely fi lled with air, the system builds
up pressure across the membrane, which is measured by a suitable pressure
transducer. After reaching the required test pressure of approximately 1 bar
the system is hermetically closed. In the case of an intact membrane, the pres-
sure drop is small or zero. A hazardous fi lter leak is shown by a pressure drop
above a certain design-specifi c limit. This test procedure is very similar to the
process used in manufacture to ensure dialyzer integrity. The test pressure is
suffi cient to detect fi lter leaks at levels which can represent a safety hazard for
the patient (see below).
After a successful test the air is removed from the system, which is then
ready for treatment. In the case of repeated negative test results the device must
not be used for treatments with online fl uid preparation, but can be operated in
standard dialysis mode. A negative test is not able to differentiate which of the
2 fi lters is faulty; this must be done later during the required servicing of the
HD machine.
Filter 1
Filter 2
Filtered air
pOnline
Dialysate or water
Filtered air
Fig. 3. Filter integrity test procedure and related components for FME Online plus
systems. p ⫽ Pressure. Redrawn from Fresenius Medical Care Online plus 07/07.03 (OP),
Fresenius Medical Care, Bad Homburg, Germany.
Polaschegg/Roy 74
Other Machines
The B. Braun system uses the principles pioneered by FME. Like the FME
machines, the B. Braun Dialog HDF Online comprises a volumetric balancing
system for ultrafi ltration control. The 2 dialysate fi lters are used in the same
confi guration: The fi rst fi lter can be periodically fl ushed while the second fi lter
is operated in fl ow-through mode. These fi lters are also periodically checked.
Unlike the FME system, which comprises special fi lter connectors excluding
the use of noncompatible fi lters by design, the B. Braun system uses standard
connectors.
The Nikkiso system uses a single dialysate fi lter that is regularly tested by
a pressure-holding test (like the FME system) in combination with a single-use
fi lter (like the Gambro system). The system was originally equipped with a dial-
ysate fi lter with conventional connectors, which makes this system compatible
to the B. Braun fi lter and vice versa. The later version comprises a special fi lter
housing and a dedicated fi lter, which improves fi lter handling and eliminates the
risk of touch contamination.
Nephros is a small company that originally embarked on the develop-
ment of an integrated HDF system comprising several novel features. This
development is well documented by patents. Eventually, however, the com-
pany decided to offer a 2-stage dialyzer for mid-dilution HDF which is mar-
keted in Europe. In a next stage the company developed an add-on device for
online HDF in combination with conventional HD machines. This device [7]
is aimed at the US market where online HDF is not yet available. It incor-
porates an integrated 2-stage dialysate fi lter [8], each with a surface area of
0.5 m2, which can be disinfected and tested by the device. This device was
recently cleared for clinical trials by the US Food and Drug Administration
(FDA).
Bellco, building upon its well-established paired-fi ltration dialysis method,
developed a device for the regeneration of fi ltrate removed from the ultrafi l-
tration stage. The regenerated fi ltrate is reinfused into the bloodstream. This
principle builds upon decades of experience with dialysate regeneration by
adsorption and ion exchange. The double fi lter can be operated with the fi ltrate
stage fi rst, followed by the dialysis stage, or vice versa.
Safety Aspects
Safety is the freedom from unacceptable risk [9]. Because risk cannot be
avoided entirely, ‘absolute safety’, a term not only used by marketing depart-
ments but also by the European Best Practice Guidelines (EBPG) [10], does
Technical Aspects of Online Hemodiafi ltration 75
not exist. Complying with safety regulations and safety standards means that
the manufacturer has weighed risks and benefi ts and has come to the conclu-
sion that the benefi ts outweigh the risks. This process is checked by notifi ed
bodies (e.g. TÜV) in the European Union, by the FDA in the USA and by other
national bodies in other countries.
The level of accepted risk depends on the application. For medical devices
it is assumed that prevention of a hazard to the patient or user after a single
fault is suffi cient (safety under single-fault condition). Because single systems
produced under normal quality conditions have failure rates of around 10–4,
redundant systems reduce the likelihood of a hazard to around 10–8, which is
equivalent to one accident in 108 treatments [11]. In Europe the manufacturer is
obliged to set up a risk management system which includes risk analysis, risk
mitigation and market surveillance. The authors have asked the major compa-
nies to lay open their risk analysis data, but received the uniform answer that
these documents contain proprietary information and will not be published.
For this reason, the risks of online HDF and HF and possible methods for
risk mitigation will be discussed using general knowledge and related to the
various methods employed by the manufacturers.
The primary risk of online HDF and HF is related to the infusion into
the bloodstream of dialysate that may be contaminated with particulate mat-
ter, bacteria and/or endotoxin. Secondary risks are the possible infusion of air
into the extracorporeal circuit and ultrafi ltration errors. These secondary risks
do not differ qualitatively from risks in regular HD with high-fl ux dialyzers. In
terms of numbers the additional risk is small and can be neglected, provided
that the alarm limits of the dialysis machine are set correctly. Unfortunately,
often this is not the case with the venous pressure monitor. Incorrect settings
of this monitor may not only cause undetected blood loss to the environment
but also undetected weight loss in the case of a leak in the infusion line for the
online-produced substitution fl uid.
The primary risk is mitigated by fi ltering contaminated dialysate through a
fi lter capable of preventing the passage of particulates, bacteria and endotoxin.
For the purpose of this discussion, the term endotoxin includes all Limulus-
amebocyte-lysate-reactive substances. The fi lters used today have membranes
similar to high-fl ux dialyzers. This means that particulate matter and bacte-
ria are held back by size exclusion. Endotoxin fractions and other potentially
hazardous debris from bacteria are not fi ltered because of the low molecular
weight. Membrane materials used for dialysate fi ltration are capable of adsorb-
ing this potentially hazardous debris on the membrane surface, including the
inner surface of the membrane pores. Tests have shown that the rejection frac-
tion of these membranes for bacteria exceeds 106–109 (detection limit) and for
endotoxin around 103 [12, 13].
Polaschegg/Roy 76
The EBPG (rule IV.4.3) [10] demand an endotoxin concentration for ultra-
pure dialysate below the detection limit. The detection limit is defi ned by these
guidelines at better than 0.03 EU/ml. With a rejection fraction of around 300
the endotoxin concentration of the incoming dialysate must not exceed 10 EU/
ml, which exceeds the limits set for regular dialysate by rule IV.4.2 EBPG of
0.25 EU/ml. Clinical improvements have been reported with lower endotoxin
concentrations (0.001 IU/ml) apparently measured with a more sensitive test
[14]. This limit can still be achieved with a typical 1-stage dialysate fi lter. This
means that a single fi lter is suffi cient if faults can be excluded.
Fault exclusion for a dialysate fi lter is not possible – any fi lter may fail. Potential
failure modes are: a gross leak either in the potting material that separates the inside
and outside of the fi lter capillaries at the fi lter ends or a broken fi ber, a leak in the
fi ber caused by wear or exposure to unsuitable disinfection fl uids (e.g. bleach for
polysulfone membranes), and saturation of the membrane by endotoxin.
The worst-case scenario for a gross leak is that large amounts of contami-
nated dialysate are infused into the bloodstream. The hazard is primarily related
to bacterial contamination. The endotoxin content of dialysate complying with
the EBPG is low compared to previous years when membranes permeable to
endotoxin (e.g. cuprophane) were used. Infusion of this endotoxin in the case of
an isolated event may be regarded as less severe.
Gross leaks can in principle be detected by measuring fi ltrate pressures or
by pressure holding tests. FME employs an automatic pressure holding fi lter
test [15, 16]. The fi lter is fi lled with air on one side while the pores and the other
side remain fi lled with fl uid. A pressure difference is generated between the air
and the fl uid side with the higher pressure on the air side. Because fi lters employ
hydrophilic membranes, air cannot penetrate the fi lters in physical form in the
absence of a gross leak. If the air side is closed, a gross leak will be detected by
a rapid pressure drop. If the air side is open, air bubbles can be detected on the
fl uid side. This test is called the bubble point test. The sensitivity of such tests
can be estimated. For the typical test conditions employed by HD machines the
diameters of leaks that can be detected is around 5 μm, which would allow the
passage of bacteria. Leaks of this kind are discussed below.
Membrane leaks caused by wear or unsuitable disinfection fl uids cannot be
detected by mechanical tests that would be applicable inside a dialysis machine.
The effect of such leaks will be the passage of bacteria and endotoxin. The upper
limit for the bypass fl ow caused by a single pore leak with a 5-μm diameter can
be estimated with the Bernoulli equation. For a dialysate fi lter with an ultrafi l-
tration coeffi cient of 5 ml/min ⭈ mm Hg, the pressure drop at 500 ml/min dialy-
sate fl ow is 100 mm Hg. This results in a fl ow of around 0.025 ml/min, which is
5 ⭈ 10–5 of the actual fl ow through a single leak not detectable by the usual pres-
sure holding test. The actual value will be less and we can use 10–6 (1 : 1 million)
Technical Aspects of Online Hemodiafi ltration 77
for the further estimate. Assuming 1,000 CFU/ml in the unfi ltered dialysate,
the fi ltered dialysate will still comply with the requirement for ultrapure dialy-
sate (0.01 CFU/ml). The same applies for endotoxin. If, however, 10 or more
such pores develop simultaneously, ultrapure dialysate is no longer guaranteed.
Similar calculations can be done for smaller pores but larger numbers.
The failure mode assumes that the diameter of existing pores is increased,
which means that the fl ow resistance will decrease, but also that the effective
surface area for diffusion of air during the pressure holding test will increase,
causing a faster drop in pressure. In principle it may be possible to detect this
fault by a sensitive pressure test, taking historical data into account. Technically
this is feasible with existing dialysis machines. It is unknown whether this pos-
sibility has been investigated by companies.
In the absence of a sensitive test, these leaks can only be avoided by strict
process control, including regular changing of the fi lter, and limiting the expo-
sure time to aggressive disinfectants according to the instructions for use.
Modern machines make use of microprocessors to remind the user about the
need to change the fi lter and may also be able to detect the use of unsuitable
disinfection fl uids.
The adsorption capacity of fi lters depends on the capillary surface area.
It also depends on the membrane material and on the physical structure of the
membrane which infl uences the total surface area of the pores. This means that
quantitative test results achieved for one type of fi lter cannot be used for other
fi lters, even if the same membrane raw material is used.
Weber et al. [13] measured the passage and adsorption capacity for syn-
thetic lipid A and lipopolysaccharide (LPS) from Pseudomonas aeruginosa.
While no passage was detected, the membranes used for dialysate fi lters typi-
cally adsorbed around 50 μg of lipid A (approx. 5,200 EU) per square meter of
nominal capillary surface area, or around 500 ng/m2 of LPS (approx. 6,000 EU).
In both cases this corresponded to around 50% of the total challenging dose.
Because only half of the material was adsorbed, it is assumed that the rest was
size excluded. With this data the ‘adsorption’ lifetime can be estimated, assum-
ing that LPS remains adsorbed on the surface. A 2-m2 fi lter would be able to
adsorb around 12,000 EU and reject approximately the same amount by size
exclusion. The sum is then estimated at around 25,000 EU. With the incom-
ing dialysate concentration limited to 0.25 EU/ml, 100,000 ml or 100 liters of
dialysate could be fi ltered, which corresponds approximately to the dialysate
consumption of 1 treatment.
Because fi lters are used for up to 100 treatments, it can be concluded that
LPS and lipid A are desorbed during the cleaning cycle or that the challenging
test mentioned above was insuffi cient to reveal the true adsorption capacity of
the membranes.
Polaschegg/Roy 78
The likelihood of a fi lter to fail because of a gross leak has not been pub-
lished or made available by the manufacturers. An estimate can be made based
on the claim that no serious adverse events related to dialysate fl uid fi lters for
HF/HDF have become known so far. Assuming 15 million treatments and with
2 fi lters in series, the likelihood of both fi lters failing during a treatment can be
estimated to be less than 10–6 to 10–8 (both fi lters having been tested before the
start of the treatment). The likelihood of a single fi lter failing is then 10–3 to 10–4
per treatment. This estimate is applicable for the FME system.
The Gambro system comprises two fi lters that are changed each month,
but not tested before treatment, and a single disposable fi lter. Estimating the
likelihood of failure is more diffi cult for this system, because the probability of
contamination between fi lters 1 and 2 must be estimated. Since manipulations
involving connectors take place when the system is being set up, contamina-
tion cannot be entirely excluded. The integer single disposable fi lter will still
be able to prevent the infusion of bacteria into the bloodstream. No published
information is known to the authors whether this single small-surface-area fi lter
is capable of adsorbing the majority of endotoxin in the case of both dialysate
fi lters failing. If not, such failures may go undetected for several treatments, and
exposure to endotoxin may go on for several weeks.
In summary it can be concluded that the application of risk analysis to the
existing systems using published data is consistent with the assumption that
these systems are safe, although ‘absolute safety’ does not exist. The limitation
of the number of treatments between fi lter changes seems to be based on clini-
cal experience rather than quantitative knowledge about the processes involved.
As long as sound research results about membrane destruction and membrane
adsorption capacity are lacking, users are strongly advised to comply fully with
the instructions given by the manufacturers.
Declaration of Conflict of Interest
Hans-Dietrich Polaschegg is an independent scientifi c consultant for medical devices.
He has no fi nancial interest in any of the companies mentioned and no consulting project for
on-line hemodiafi ltration. Thomas Roy is an employee of Fresenius Medical Care Deutschland
GmbH, Bad Homburg, Germany.
References
1 Community Directive 93/42/EEC on Medical Devices. Offi cial Journal of the European
Communities No L169, 12 July 1993.
2 Pirovano D: Regulatory issues for on-line haemodiafi ltration. Nephrol Dial Transplant
1998;13(suppl 5):21–23.
Technical Aspects of Online Hemodiafi ltration 79
3 Association for the Advancement of Medical Instrumentation: Water Treatment Equipment for
Hemodialysis Applications (ANSI/AAMI RD62:2001). American National Standard. Arlington,
AAMI, 2001.
4 AK 200 ULTRA S – Operator’s manual. HCEN9752 Revision.12.2003, Lund, Gambro AB, 2003.
5 Ledebo I: On-line preparation of solutions for dialysis: practical aspects versus safety and regula-
tions. J Am Soc Nephrol 2002;13:S78–S83.
6 Online plus™ Operating Instructions, part No 676 894 1, Revision: 7/07.03, Bad Homburg,
Fresenius Medical Care, 2003.
7 Collins GR, Summerton J, Spence E (inventors), Nephros Inc (assignee): Method and apparatus
for a hemodiafi ltration delivery module. US patent 6916424. 07/12/2005.
8 Summerton J, Collins G (inventors), Nephros Inc (assignee): Sterile fl uid fi ltration cartridge and
method for using same. US patent 6635179. 10/21/2003.
9 ISO 14971 1, 2007-02-28: Medical devices – Application of risk management to medical devices.
10 EBPG Expert Group on Haemodialysis: European best practice guidelines for haemodialysis
(part 1), section IV: dialysis fl uid purity – IV.4 haemodialysis-proportioning machine. Nephrol
Dial Transplant 2002;17:1–111.
11 Polaschegg HD, Levin N: Hemodialysis machines and monitors; in Winchester J, Koch R, Lindsay
R, Ronco C, Horl W (eds): Replacement of Renal Function by Dialysis, ed 5. New York, Kluwer
Academic Publishers, 2004, pp 323–447.
12 Frinak S, Polaschegg HD, Levin NW, Pohlod DJ, Dumler F, Saravolatz LD: Filtration of dialysate
using an on-line dialysate fi lter. Int J Artif Organs 1991;14:691–697.
13 Weber C, Linsberger I, Rafi ee-Tehrani M, Falkenhagen D: Permeability and adsorption capacity
of dialysis membranes to lipid A. Int J Artif Organs 1997;20:144–152.
14 Arizono K, Nomura K, Motoyama T, Matsushita Y, Matsuoka K, Miyazu R, Takeshita H, Fukui H:
Use of ultrapure dialysate in reduction of chronic infl ammation during hemodialysis. Blood Purif
2004;22:26–29.
15 Polaschegg HD, Mathieu B (inventors), Fresenius AG (assignee): Verfahren zum Prüfen von
Sterilfi ltern eines Hämofi ltrationsgeräts. DE patent 3448262. 06/21/1990.
16 Wamsiedler R, Matheiu B (inventors), Fresenius AG (assignee): Verfahren zur Überprüfung
von mindestens einem im Dialysierfl üssigkeitssystem einer Vorrichtung zur extrakorporalen
Blutbehandlung angeordneten Filter. DE patent 19534417. 03/20/1997.
Hans-Dietrich Polaschegg
Scientifi c Consultant
A-9231 Koestenberg (Austria)
Tel. ⫹43 4274 4045, Fax ⫹43 4274 4096, E-Mail hdp@aon.at
Ronco C, Canaud B, Aljama P (eds): Hemodiafi ltration.
Contrib Nephrol. Basel, Karger, 2007, vol 158, pp 80–86
Quality of Water, Dialysate and Infusate
Gianni Cappelli, Marco Ricardi, Decenzio Bonucchi, Sara De Amicis
Nephrology Dialysis and Renal Transplantation Unit, University Hospital of Modena,
Modena, Italy
Abstract
Great improvements in water treatment technology and the spread of ultrafi ltration for
cold sterilization have been the basic support for the development and diffusion of on-line
dialysis treatments. Some 20 years ago, nephrologists recognized that the offi cial standards
for dialysis fl uids were insuffi cient with respect to these new treatment modalities, and
ultrapure water (bacteria ⬍0.1 CFU/ml; endotoxin ⬍0.03 EU/ml) was proposed as a refer-
ence. Today, ultrapure water is included in most guidelines and recommended standards, but
there remains a need for harmonization between standards. To achieve and ensure these lev-
els of purity, technology must be supported by commitment of resources to an active quality
assurance programme with adequate maintenance, monitoring, cleaning, sanitizing and prob-
lem analysis procedures.
Copyright © 2007 S. Karger AG, Basel
Background
Several lines of evidence have accumulated in the last 10–15 years which
show that optimizing water purity in dialysis, in terms of bacterial contamina-
tion, is a fundamental issue for preventing infl ammation in most dialysis-related
pathologies in end-stage renal disease patients [1–7]. Until the mid-80s, the
concern regarding water quality was mostly concentrated on chemical con-
taminants but after the relationship between the bacterial burden in dialysate
and pyrogenic reactions had been established [8–10], the emphasis moved to
microbiological contamination. Ultrafi ltration was at that time in the phase of
preliminary experimentation, but several groups were already reporting its use
with consistent results in obtaining sterile bacteria- and endotoxin-free solu-
tions to be used as dialysate or infusate in haemo(dia)fi ltration [11–15].
Technical Aspects and Fluids in Hemodiafiltration
Quality of Water, Dialysate and Infusate 81
The Search for Harmonization of Standards
The pressure to redefi ne adequate microbiological standards became urgent
in the early 90s, due to the spreading use of highly permeable membranes, the
spread of online treatments and the feasibility offered by dialysis technology in
obtaining sterile fl uids. Now, at the beginning of the 21st century, while there
is a worldwide general agreement on maximum levels of most chemical con-
taminants, some differences between countries still remain regarding permitted
levels of microbiological contaminants.
The lack of harmonization is due to both scientifi c and economic reasons.
For the fi rst part, the clinical evidence for using ultrapure fl uids is still under
discussion, since (i) most data are from different groups with no controlled
clinical trials, (ii) endotoxin evaluation does not cover all microbially derived
products (including bacterial DNA) that have been demonstrated to cross the
dialysis membrane, and (iii) the purity of dialysis fl uids is part of a multifac-
torial aetiology for chronic infl ammatory response in end-stage renal disease
patients, which also includes vascular access, dental infection and membrane
biocompatibility.
As to the economic question, the main points are: (i) capital investments
have to be made in water systems and monitors, (ii) resources must also be
committed to quality assurance and monitoring, and (iii) in some countries
where reimbursements from government organizations are linked to adherence
to water quality standards, new limits must be implemented gradually.
A recent review [16] discusses the opportunity of harmonization to achieve
a widely applicable set of standards, taking into consideration all these aspects
and, most of all, best practice for patient safety.
Table 1 lists recommended maximum concentrations of chemical con-
taminants in water for haemodialysis, comparing the Association for the
Advancement of Medical Instrumentation (AAMI) with the European
Pharmacopoeia [17, 19]. In table 2 are the maximum levels of bacterial con-
taminants for water for dilution and standard dialysate, recently issued as
recommendations or national standards of some professional associations or
organizations for standardization.
From these tables it is clear that, in spite of a large production of standards,
the ideal purity of dialysis fl uid is still under development. In recent papers and
reviews, ultrapure fl uids, as originally defi ned (⬍0.1 CFU/ml, ⬍0.03 EU/ml)
[25], are indicated as new ‘gold standards’ for both dialysis water and dialysate,
and according to European Renal Association/European Dialysis and Transplant
Association (ERA-EDTA) guidelines they are suggested for all dialysis treat-
ment modalities [26].
Cappelli/Ricardi/Bonucchi/De Amicis 82
Technology to Obtain Ultrapure Fluids
Producing ultrapure fl uids is not a trivial task, as it requires appropriate
technology and most of all a commitment of resources and a quality control
programme to put it on a routine basis. Current water treatment techniques offer
optimal results and can be adapted to most situations. Only general indications
can be proposed for a new system, since detailed suggestions must be based on
Table 1. Recommended levels of chemical contami-
nants in water for haemodialysis
Contaminant Maximum contaminant concentration mg/l
ANSI/AAMI
RD62-2001 [17]
European
Pharmacopoeia ed. 5
[18]
Calcium 2 (0.1 mEq/l) 2
Magnesium 4 (0.3 mEq/l) 2
Potassium 8 (0.2 mEq/l) 2
Sodium 70 (3.0 mEq/l) 50
Ammonia – 0.2
Antimony 0.006 –
Arsenic 0.005 –
Barium 0.10 –
Beryllium 0.0004 –
Cadmium 0.001 –
Chromium 0.014 –
Chloride – 50
Lead 0.005 –
Mercury 0.0002 0.001
Selenium 0.09 –
Silver 0.005 –
Aluminium 0.01 0.01
Chloramine 0.10 –
Free chlorine 0.50 –
Total available
chlorine – 0.1
Copper 0.10 –
Fluoride 0.20 0.20
Nitrate (as N) 2.00 2.00
Sulphate 100 50
Thallium 0.002 –
Total heavy
metals – 0.10
Zinc 0.10 0.10
Quality of Water, Dialysate and Infusate 83
local water quality, taking into account also seasonal variability and disinfec-
tion methodology [27].
The optimal water purifi cation system is the result of a 2-stage procedure:
pretreatment and fi nal treatment (fi g. 1). Pretreatment technology is dictated by
raw water quality and is usually based on softeners and carbon tanks contain-
ing granulated activated charcoal. Pretreatment includes a depth fi lter, sometimes
added to remove particulate matter from feeding water, and a 5-m cartridge fi l-
ter used to protect the reverse osmosis (RO) membrane from leached carbon par-
ticles. Final treatment with RO provides an excellent barrier to most chemicals
and microbiological contaminants, and the use of a double RO in series improves
performance. When a deionizer is used as a supplement to RO installation, an
ultrafi lter after the deionizer is mandatory to achieve consistent microbiological
results. Distribution of the purifi ed water to individual dialysis machines is a key
point, as bacterial growth in the distribution pipes is the most common source of
contamination. New piping materials have been proposed, featuring no leaching
of contaminants, smoother surfaces and low adherence to bacterial cells in order
to prevent biofi lm formation, but frequent and periodical disinfections remain the
best strategy to assure a high microbiological level. A continuous loop design is
the recommended circuit to minimize biofi lm formation and to assure a pure water
Table 2. Recommended levels of microbiological contaminants in solutions for haemodialysis
AAMI
(USA)
[17, 19]
Eur.
Pharma.
ed. 5 [18]
ESBP
ERA-
EDTA [20]
CSA
(Canada)
[21]
Renal
Association
(UK) [22]
CARI
(Australia)
[23]
JSDT
(Japan)
[24]
Water for dialysis
Bacteria, CFU/ml 200a100 100 100 100 100 –
Endotoxin, EU/ml 2a0.25 0.25 2 0.25 0.25 0.05
Standard dialysate
Bacteria, CFU/ml 200a– 100 100 100 100 –
Endotoxin, EU/ml 2a– 0.25 2 0.25 0.25 0.05
Ultrapure dialysate
Bacteria, CFU/ml 0.1 – 0.1 – 0 0.1 –
Endotoxin, EU/ml 0.03 – 0.03 – 0.015 0.03 0.01
Infusate for
haemo(dia)fi ltration
Bacteria, CFU/ml 10-6 10-6 10-6 –010
-6 –
Endotoxin, EU/ml 0.03 0.25 0.03 – 0.015 0.03 0.01
CARI ⫽ Caring for Australasians with Renal Impairment; CSA ⫽ Canadian Standards Association;
ESBP ⫽ European Best Practice Guidelines; JSDT ⫽ Japanese Society for Dialysis Therapy.
a Defi ned an action level at 50 CFU/ml for bacteria and 1 EU/ml for endotoxins.
Cappelli/Ricardi/Bonucchi/De Amicis 84
transfer according to ultrapure dialysate prescription. As to the regulatory aspects
of components of water treatment and distribution systems, producers of water
purifi cation systems are regulated by the Food and Drug Administration (FDA) in
the USA or by Directive 93/42/EEC for medical devices in the European Union.
Water systems, dialysis machines and high-permeability dialysers are mandated
as class II medical devices by the FDA and as class II-b by the European Union
and require diligent tracking of critical components and a complaint investiga-
tion system in place. The quality level of ultrapure dialysate technology should
be user-friendly and budget-compatible, but nephrology professionals, including
nurses and technicians, should be aware that monitors and ultrafi lters must be
used according to manufacturers’ instructions, and adequate disinfection as sug-
gested by each manufacturer must be periodically performed. The choice of dis-
infectant is based on compatibility with the materials used in the water system,
on manufacturers’ suggestions and on results from monitoring the dialysis fl uid
production process. Disinfectants are also regulated as class II medical devices,
being accessories to the devices that they are intended to disinfect, and therefore
dialysis staff must strictly adhere to manufacturers’ instructions.
Depth
filter
Cartridge
filter
HD
monitor
Distribution loopFinal
treatment
Ultrapure dialysate
Ultrafilter 2
Patient
Ultrafilter 1
Online infusate
DialyserHD monitor
AB
Ultrapure water
/UF
filtration
Pretreatment
Charcoal
adsorption
Water
softener
RO1 RO2
Tap
water
a
b
Fig. 1. Water treatment system and dialysis apparatus to obtain fl uids for online treat-
ments. a Water plant with pretreatment and fi nal treatment based on a double RO in series.
b Water from distribution loop enters the dialysis monitor and, after a fi rst ultrafi ltration to
obtain ultrapure dialysate, is passed through a second ultrafi lter to obtain substitution fl uid.
HD ⫽ Haemodialysis; /UF fi ltration ⫽ micro- or ultrafi ltration.
Quality of Water, Dialysate and Infusate 85
Quality Assurance
The experience accumulated in past years with online treatments of thou-
sands of patients has overcome the problem of safety. At each stage, ultrafi ltra-
tion is able to reduce bacteria and endotoxin by factors of at least 107 and 103,
respectively, providing a ‘cold sterilization’ device [28]. To stay within opera-
tional limits the lowest possible bacterial and endotoxin mass must be presented
to the ultrafi ltration point and redundancy in terms of surface area must be pro-
vided by the device to optimize fi ltration and absorption processes. The impos-
sibility of online monitoring of results is therefore balanced by the assurance of
maintaining a high absorption capacity.
Staff should be aware of the importance of the hygiene of the entire fl uid
path, from tap water to the distribution loop, including the hydraulic dialysis-
monitoring circuit. To avoid failure of the most critical component of the water
treatment system, the human one, a quality assurance procedure has to be insti-
tuted, based on appropriate maintenance, monitoring, cleaning and sanitizing
procedures for the whole production chain. It includes acceptance or redefi nition
of limit values for each part of the system, depending on local needs, as well
as defi nition of laboratory methods and tests to check the process. Auditing for
analysis of operational problems should be part of the quality assurance system.
References
1 Baz M, Durand C, Ragon A, Jaber K, Andrieu D, Merzouk T, Purgus R, Olmer M, Reynier JP,
Berland Y: Using ultrapure water in hemodialysis delays carpal tunnel syndrome. Int J Artif
Organs 1991;14:681–685.
2 Sitter T, Bergner A, Schiffl H: Dialysate related cytokine induction and response to recombinant
human erythropoietin in haemodialysis patients. Nephrol Dial Transplant 2000;15:1207–1211.
3 Schiffl H, Lang SM, Stratakis D, Fischer R: Effects of ultrapure dialysis fl uid on nutritional status
and infl ammatory parameters. Nephrol Dial Transplant 2001;16:1863–1869.
4 Matsuhashi N, Yoshioka T: Endotoxin-free dialysate improves response to erythropoietin in hemo-
dialysis patients. Nephron 2002;92:601–604.
5 Schiffl H, Wendinger H, Lang SM: Ultrapure dialysis fl uid and responsiveness to hepatitis B vac-
cine. Nephron 2002;91:530–531.
6 Matsuhashi N, Yoshioka T: Endotoxin-free dialysate improves response to erythropoietin in hemo-
dialysis patients. Nephron 2002;92:601–604.
7 Blagg C, Twardowski Z, Bower J, Kjellstrand C: Cardiovascular instability and dialysate purity.
ASAIO J 2003;49:190.
8 Favero MS, Carson LA, Bond WW, Petersen NJ: Factors that infl uence microbial contamination of
fl uids associated with hemodialysis machines. Appl Microbiol 1974;28:822–830.
9 Gordon SM, Oettinger CW, Bland LA, Oliver JC, Arduino MJ, Aguero SM, McAllister SK, Favero
MS, Jarvis WR: Pyrogenic reactions in patients receiving conventional, high-effi ciency, or high-
fl ux hemodialysis treatments with bicarbonate dialysate containing high concentrations of bacteria
and endotoxin. J Am Soc Nephrol 1992;2:1436–1444.
10 Mion CM, Canaud B, Garred LJ, Stec F, Nguyen QV: Sterile and pyrogen-free bicarbonate dialy-
sate: a necessity for hemodialysis today. Adv Nephrol Necker Hosp 1990;19:275-314.
Cappelli/Ricardi/Bonucchi/De Amicis 86
11 Henderson LW, Beans E: Successful production of sterile pyrogen-free electrolyte solution by
ultrafi ltration. Kidney Int 1978;14:522–525.
12 Henderson LW, Sanfelippo ML, Beans E: ‘On line’ preparation of sterile pyrogen-free electrolyte
solution. Trans Am Soc Artif Intern Organs 1978;24:465–467.
13 Shaldon S, Beau MC, Deschodt G, Flavier JL, Nilsson L, Ramperez P, Mion C: Three years of
experience with on-line preparation of sterile pyrogen free infusate for hemofi ltration. Int J Artif
Organs 1983;6:25–26.
14 Canaud B, N’Guyen QV, Lagarde C, Stec F, Polaschegg HD, Mion C: Clinical evaluation of a
multipurpose dialysis system adequate for hemodialysis or for postdilution hemofi ltration/hemo-
diafi ltration with on-line preparation of substitution fl uid from dialysate; in Streicher E, Seyffart G
(eds): Highly Permeable Membranes. Contrib Nephrol. Basel, Karger, 1985, vol 46, pp 184–186.
15 Erley CM, von Herrath D, Hartenstein-Koch K, Kutschera D, Amir-Moazami B, Schaefer K: Easy
production of sterile, pyrogen-free dialysate. ASAIO Trans 1988;34:205–207.
16 Ward RA: Worldwide water standards for hemodialysis. Hemodial Int 2007;11:S18–S25.
17 Association for the Advancement of Medical Instrumentation: Water Treatment Equipment for
Hemodialysis Applications. ANSI/AAMI RD62:2001. Arlington, Association for the Advancement
of Medical Instrumentation, 2001.
18 European Pharmacopoeia Commission: Monograph 01/2005:1167 Haemodialysis Solutions,
Concentrated, Water for Diluting. European Pharmacopoiea, ed 5. Strasbourg, European
Pharmacopoeia Commission, 2006.
19 Association for the Advancement of Medical Instrumentation: Dialysate for hemodialysis. ANSI/
AAMI RD52:2004. Arlington, Association for the Advancement of Medical Instrumentation, 2004.
20 European Renal Association-European Dialysis and Transplant Association: European best prac-
tice guidelines for haemodialysis (part I), section IV: dialysis fl uid purity. Nephrol Dial Transplant
2002;17(suppl 7):45–62.
21 Canadian Standards Association: Water Treatment Equipment and Water Quality Requirements for
Hemodialysis. CSA Standard Z364.2.2–03. Mississauga, Canadian Standards Association, 2003.
22 The Renal Association and the Royal College of Physicians of London: Treatment of Adults and
Children with Renal Failure: Standards and Audit Measures, ed 3. London, Lavenham Press,
2002.
23 CARI Guidelines: Dialysis adequacy guidelines – water quality for hemodialysis. Nephrology
2005;10:S61–S80.
24 Masakane I: Review: clinical usefulness of ultrapure dialysate – recent evidence and perspectives.
Ther Apher Dial 2006;10:348–354.
25 Ledebo I, Nystrand R: Defi ning the microbiological quality of dialysis fl uid. Artif Organs 1999;23:
37–43.
26 Kjellstrand CM, Kjellstrand P: Beyond ultrapure hemodialysis: a necessary and achievable goal.
Hemodial Int 2007;11:S39-S48.
27 Cappelli G, Perrone S, Ciuffreda A: Water quality for on-line haemodiafi ltration. Nephrol Dial
Transplant 1998;13(suppl 5):12–16.
28 Canaud B, Bosc JY, Leray H, Stec F: Microbiological purity of dialysate for on-line substitution
fl uid preparation. Nephrol Dial Transpl 2000;15(suppl 2):21–30.
Prof. Gianni Cappelli
Nephrology Dialysis and Renal Transplantation Unit
Department of Medicine and Medical Specialties
University Hospital of Modena
Via Del Pozzo, 71
IT–41100 Modena (Italy)
Tel. ⫹39 059 422 5220, Fax ⫹39 059 422 2167, E-Mail cappelli@unimo.it
Ronco C, Canaud B, Aljama P (eds): Hemodiafi ltration.
Contrib Nephrol. Basel, Karger, 2007, vol 158, pp 87–93
Fluids in Bags for Hemodiafiltration
Ingrid Ledebo
Gambro Research, Lund, Sweden
Abstract
The objective of hemodiafi ltration (HDF) is to increase the convective transport of sol-
utes poorly removed by diffusion, and therefore ultrafi ltration (UF) beyond the desired weight
loss is prescribed. The excess UF is compensated for by infusion of a physiological solution
which should be sterile and nonpyrogenic. This replacement solution can be provided either
in bags containing commercially prepared infusion solution, i.e. so-called classic HDF, or by
an integrated stepwise fi ltration of the dialysis fl uid, i.e. so-called online HDF. In both cases
the composition of the replacement solution should mirror that of plasma water. When the
fl uid is provided in bags, practical handling is a limiting factor, and the amount of convection
that can be delivered is most often restricted to 10–12 l/session. Results from clinical studies
show that the degree of convective transport obtained in classic HDF corresponds to what
can be achieved in contemporary high-fl ux dialysis, where uncontrolled UF and backfi ltra-
tion take place inside the dialyzer. Classic HDF with replacement fl uid in bags offers the
possibility of delivering an HDF treatment with controlled convective dose and fl uid quality,
albeit with a limited amount of convection.
Copyright © 2007 S. Karger AG, Basel
Definitions and Terminology
The difference between hemodialysis (HD) and hemodiafi ltration (HDF)
is that in the latter therapy, ultrafi ltration (UF) exceeds the desired weight loss,
and the difference is made up with a replacement solution, according to com-
mon defi nitions [1]. The original defi nition states that the replacement solution
should come from an external source. However, with our increased understand-
ing of the transport processes that take place in blood purifi cation therapies,
it is now acknowledged that so-called internal replacement of excess UF can
produce a similar result, and therapies comprising this feature should also be
regarded as HDF [2].
Ledebo 88
The objective of excessive UF is to increase the removal of solutes that are
poorly transported by diffusion. UF is accompanied by convective transport of
all those solutes that can pass the membrane, and the amount of convection is
determined by the UF volume and the membrane permeability. Thus, in low-
fl ux HD convective transport is negligible, because the UF corresponds only
to the weight loss and the membrane has a low solute permeability. Modern
high-fl ux membranes are highly permeable to water as well as to solutes up to
a molecular weight of 20–30 kDa, and thus have the potential to provide con-
siderable convective clearance. These membranes are used in HDF as well as in
high-fl ux dialysis. When used in HD with modern equipment providing volume
control, the net UF corresponds to the desired weight loss, but signifi cant fl uid
fl ux takes place inside the dialyzer in both directions across the membrane and
generates convective solute transport [3]. Blood is ultrafi ltered at the arterial end
of the dialyzer, and the excess is replaced by dialysis fl uid backfi ltered across
the membrane at the venous end. Under conditions of countercurrent fl ow, the
backfi ltered fl uid is relatively fresh and can be regarded as replacement fl uid,
although not quite of the same composition and quality. The result is convective
solute transport and fl uid replacement, both of which occur automatically and
are uncontrolled.
With the use of externally provided replacement fl uid, the amount of con-
vection can be prescribed and controlled by settings on the machine. The total
convective volume corresponds to the volume of UF, which is a product of the
UF rate and the treatment time. The blood fl ow rate is a limiting factor, and
UF rates corresponding to 30% of the blood fl ow rate can usually be achieved
in postdilution HDF. The amount of replacement fl uid required is the total UF
volume corrected for the desired weight loss. To avoid symptoms of under- and
overhydration, it is vital that the control mechanism for fl uid removal and fl uid
replacement be accurate. The fl ow rates should be matched so that the differ-
ence corresponds to the desired weight loss rate for the specifi c patient and
treatment, just as in an HD treatment. The replacement solution is most com-
monly continuously provided through an integrated online preparation that con-
sists of stepwise UF of dialysis fl uid under controlled conditions [4]. However,
the replacement fl uid can also be provided as an intravenous solution in bags,
which is the focus of this chapter. An overview of the characteristics of the
replacement fl uids used for the various forms of HDF is given in table 1.
Composition of Replacement Fluid
The replacement fl uid as well as the dialysis fl uid should be physiologi-
cal, and the composition should mirror that of plasma water. With stable
Fluids in Bags for Hemodiafi ltration 89
HD patients who are transferred to HDF, it is recommended to start with the
same composition of the replacement fl uid as has been used for the dialysis
fl uid. This may not always be possible when using fl uid in bags, since the
choice of available compositions is limited. When alternative compositions
have to be used, it is wise to aim for more physiological solutions, since
the replacement fl uid needs no concentration gradient for transport across a
membrane.
In the early days of HDF the buffer source was a limiting factor. When
acetate was used in the dialysis fl uid, lactate was the obvious choice in the
replacement fl uid. However, when acetate was replaced with bicarbonate,
the drawbacks of using unphysiological buffer sources became obvious, and
bicarbonate-containing fl uid in bags was developed. The use of bicarbonate
in fl uids prepared, sterilized and stored in bags makes particular demands on
Table 1. Replacement fl uid characteristics for various forms of HDF
High-fl ux
dialysis
Classic HDF Online HDF
Source of fl uid internal,
dialysis fl uid
external,
bags
external,
online
Mode of preparation backfi ltration autoclaving stepwise
fi ltration
Fluid composition as dialysis fl uid choice between
available fl uids
as dialysis fl uid
Fluid quality variable, not
likely to be
sterile
sterile and
nonpyrogenic
according to
Pharmacopoeia
sterile and
nonpyrogenic
Prescription of convective
volume
not possible limited by fl uid
volume in bags
not limited by
access to fl uid
Infusion mode only
postdilution
only
postdilution
pre- or
postdilution
Extra work compared to HD simple,
automatic
handling of
bags
hygiene of
system
Cost compared to HD similar ⫹ cost of fl uid
in bags
⫹ cost of
ultrafi lters
Risk to patient use of unsterile
fl uid
low, if fl uid
quality OK
low, if system
operated
according to
manuals
Ledebo 90
the packaging. To avoid precipitation, the bicarbonate ions must be separated
from the divalent cations until the moment of mixing. Therefore, separate
bags or 2-compartment bags must be used. The benefi ts of using bicarbonate
as the sole buffer source in dialysis therapy were clearly illustrated in a trial
which tested 4 different buffer combinations, and which demonstrated that
acid-base correction as well as patient tolerance were superior when bicar-
bonate was used as the buffer source in both fl uids [5]. Today, the majority of
fl uid in bags for HDF contains bicarbonate at a concentration of 32–35 mmol/
l, which corresponds to the buffer concentration most commonly used in
dialysis fl uid.
When discussing the composition of replacement solution, it should also
be mentioned that certain forms of HDF therapy require special solutions, e.g.
acetate-free biofi ltration. This is HDF with buffer-free dialysis fl uid and an exter-
nal replacement fl uid consisting of sodium bicarbonate at a concentration of
145 mmol/l, provided in bags and used in volumes of 8–10 l/session [6]. Several
studies have documented the benefi ts of optimal acid-base correction and hemo-
dynamic stability, but comparisons have generally been made with HD rather
than HDF. For further information about acetate-free biofi ltration, readers are
referred to the chapter by Santoro et al. (pp. 138–152) in this book.
Quality of Replacement Fluid
Replacement fl uid is classifi ed as a drug, an infusion solution, which must
be sterile and which must contain less than 0.25 endotoxin units/ml (EU/ml)
according to the European Pharmacopoeia [7].
However, if fl uid of such quality were to be used in convective therapies,
the maximum recommended endotoxin exposure for healthy individuals (5 EU/
kg body weight and hour) would be exceeded already at moderate infusion rates
and body weights. It is therefore important that the replacement solution for
HDF is of considerably higher microbiological quality than just meeting the
limit prescribed by current regulations.
Example: 70 kg ⫻ 5 EU/(kg ⭈ h) ⫽ 0.25 EU/ml ⫻ 1,400 ml/h
Experience from the early days of using replacement fl uid in bags for
convective therapies, at that time mainly hemofi ltration, shows a number of
serious, even fatal, incidences of contamination from these fl uids, either from
handling of the fl uid or from the fl uid itself [8]. Quellhorst et al. [9] measured
the endotoxin concentration in various fl uids and found that the mean con-
centration in 721 samples of commercially prepared substitution solution was
17.5 ⫾ 6.8 pg/ml, well below the target recommended by the Pharmacopoeia,
Fluids in Bags for Hemodiafi ltration 91
but still 4 times higher than the average in the 1,364 samples of online pre-
pared fl uid (4.1 ⫾ 0.6 pg/ml). However, the quality of the dialysis fl uid is also
important because fl uid fl ux across the dialysis membrane may occur. Panichi
et al. [10] found that HDF using replacement fl uid in bags was associated with
production of proinfl ammatory cytokines during periods when backfi ltration
of standard-quality dialysis fl uid was known to take place, but not when back-
fi ltration was avoided.
Volume of Replacement Fluid
With the use of replacement fl uid in bags, mainly practical but also eco-
nomical considerations pose serious limitations on the volume used and thus
the dose of convection delivered. Large volumes of fl uid are heavy and diffi cult
to handle, and modern bicarbonate-containing fl uids are costly. Most users of
HDF with fl uid in bags therefore limit the amount of replacement fl uid to 9
liters, which translates into two 4.5-liter bags. Adding the UF that corresponds
to the weight loss gives a total convective volume of 10–12 liters. This limita-
tion in fl uid volume also means that administration in postdilution mode is the
only realistic alternative.
From a therapeutic point of view the volume of convective transport should
be maximized. To justify the additional cost when switching patients from high-
fl ux dialysis to HDF, the convective transport should be considerably increased.
However, the degree of convection already achieved in contemporary high-fl ux
HD appears to correspond to HDF treatments with approximately 10 liters of
convective transport, i.e. classic HDF sessions. This is shown by studies compar-
ing the clearance and plasma levels of convectively removed marker molecules,
such as 2-microglobulin. Similar levels of 2-microglobulin were achieved
with high-fl ux dialysis and classic HDF in the Italian Cooperative Dialysis
Study [11], and similar 2-microglobulin clearance and reduction ratios were
observed in high-fl ux dialysis and HDF when infusion rates of 40 ml/min were
used [12]. Thus, the major benefi t of switching from high-fl ux dialysis to HDF
with fl uid in bags appears to be the improved control of convective volume and
assurance of fl uid quality.
To be able to achieve considerably increased convective volumes there
must be no limitation in access to fl uid, and online fl uid preparation is the
only realistic alternative. An average online HDF session, performed at a
blood fl ow rate of 300–400 ml/min, would generate a UF rate of 90–120 ml/
min, assuming that 30% of the blood fl ow can be ultrafi ltered. In 4 h, 22–29
liters of ultrafi ltrate would be generated and would require some 20–26 liters
of replacement solution. The convective transport would be at least twice that
Ledebo 92
achieved with high-fl ux dialysis or classic HDF, with 9 liters of replacement
fl uid in bags.
Use of HDF with Fluid in Bags
If the purpose of a therapy prescription is to achieve as high a convective
clearance as possible in combination with a certain amount of diffusive clear-
ance, HDF with unlimited access to replacement fl uid, i.e. online HDF, is the
obvious choice. However, in situations when this is not possible due to unavail-
ability of equipment or regulatory obstacles, the use of classic HDF, i.e. with
the replacement fl uid provided in bags, is a good alternative. This will allow the
user to prescribe a certain, although limited, amount of convection and admin-
ister it under conditions of controlled fl uid quality. High-fl ux dialysis, on the
other hand, may provide similar amounts of convective clearance, but unless the
quality of the dialysis fl uid and the characteristics of the dialysis membrane are
such that endotoxin transfer is reduced to safe levels, the risk to the patient is
considerable. It may be claimed that high-fl ux dialysis is widely used and even
associated with superior survival, thus indicating that it is benefi cial rather than
risky [13]. Still, the biologic response to chronic exposure to bacterial prod-
ucts may be subtle, only evident as a microinfl ammation or even blunted by the
benefi t of enhanced convective solute removal. Finally, the use of HDF with
fl uid in bags is also justifi ed when the composition of the fl uid has a purpose
beyond mere physiological replacement, such as in acetate-free biofi ltration.
Thus, there are special cases when HDF with replacement fl uid in bags is the
better choice, but in general online fl uid preparation is an essential prerequisite
for optimal HDF therapy [14].
References
1 Gurland HJ, Davison AM, Bonomini V, et al: Defi nitions and terminology in biocompatibility.
Nephrol Dial Transplant 1994;9(suppl 2):4–10.
2 von Albertini B, Miller JH, Gardner PW, et al: High-fl ux hemodiafi ltration: under six hours/week
treatment. Trans Am Soc Artif Intern Organs 1984;30:227–231.
3 Ronco C: Backfi ltration: a controversial issue in modern dialysis. Int J Artif Organs 1988;11:
69–74.
4 Ledebo I: On-line preparation of solutions for dialysis: practical aspects versus safety and regula-
tions. J Am Soc Nephrol 2002;13:S78–S83.
5 Biasiolo S, Feriani M, Chiaramonte S, et al: Different buffers for hemodiafi ltration. a controlled
study. Int J Artif Organs 1989;12:25–30.
6 Galli G, Panzetta G: Acetate free biofi ltration (AFB): from theory to clinical results. Clin Nephrol
1998;50:28–37.
7 European Pharmacopoeia, ed 5. Strasbourg, Council of Europe, 2005.
Fluids in Bags for Hemodiafi ltration 93
8 von Herrath D, Schaefer K, Hüfl er M, Pauls A, Koch KM: Complications of hemofi ltration; in
Schaefer, K (ed): Hemofi ltration. Contrib Nephrol. Basel, Karger, 1982, vol 32, pp 146–153.
9 Quellhorst E, Hildebrand U, Solf A: Long-term morbidity: hemofi ltration versus hemodialysis; in
Berland Y, Bonomini V (eds): Dialysis Membranes: Structure and Predictions. Contrib Nephrol.
Basel, Karger, 1995, vol 113 pp 110–119.
10 Panichi V, Tetta C, Rindi P, Palla R, Lonnemann G: Plasma C-reactive protein is linked to backfi l-
tration associated interleukin-6 production. ASAIO J 1998;44:M415–M417.
11 Locatelli F, Mastrangelo F, Redaelli B, et al: Effects of different membranes and dialysis tech-
nologies on patient treatment tolerance and nutritional parameters. Kidney Int 1996;50:1293–
1302.
12 Lornoy W, Becaus I, Billiouw J-M, Sierens L, van Malderen P: Remarkable removal of beta-2-
microglobulin by on-line hemodiafi ltration. Am J Nephrol 1998;18:105–108.
13 Chaveau P, Nguyen H, Combe C, et al: Dialyzer membrane permeability and survival in hemodi-
alysis patients. Am J Kidney Dis 2005:45:565–571.
14 Ledebo I: On-line hemodiafi ltration: technique and therapy. Adv Ren Replace Ther 1999;6:195–
208.
Ingrid Ledebo, PhD
Gambro Lundia AB
Box 10101
SE–22010 Lund (Sweden)
Tel. ⫹46 46 16 91 76, Fax ⫹46 46 16 97 77, E-Mail ingrid.ledebo@gambro.com
Ronco C, Canaud B, Aljama P (eds): Hemodiafi ltration.
Contrib Nephrol. Basel, Karger, 2007, vol 158, pp 94–102
Hemodiafiltration with Endogenous
Reinfusion
Mary Lou Wrattena, Paolo M. Ghezzib
aResearch Scientist, Sorin Group Italia, Mirandola, and bMedical Devices Consultant,
Montemassi, Italy
Abstract
Hemodiafi ltration (HDF) is well known to increase the solute convective clearance due
to increased ultrafi ltration but requires substantial amounts of high-quality reinfusion fl uid.
Initially in the early 90s, individual bags or containers of reinfusion fl uid were used and caused
many problems related to handling (storage, repeated connections) and costs. Additionally
there was an increased risk of circuit contamination. The interest in HDF pushed technological
research for online production of sterile and ultrapure reinfusion solutions. Using a 2-chamber
fi lter, it is possible to produce reinfusion fl uid from the ultrafi ltrate of the patient, which has
been ‘regenerated’ by a sorbent bed, in a closed circuit. This action eliminates any problems of
sterility and apyrogenicity, while also providing the possibility of reinfusing physiologically
important substances such as bicarbonates and essential and branched-chain amino acids. This
HDF method, called hemofi ltrate reinfusion (HFR), has been clinically demonstrated to reduce
the loss of physiological components and is associated with decreased infl ammation and oxi-
dative stress. In addition to its ease of use, the technique is also highly biocompatible. Based
on these observations, HFR appears to be a useful technique for patients with complex risk
factors such as malnutrition, infl ammation and atherosclerosis.
Copyright © 2007 S. Karger AG, Basel
Hemodiafi ltration (HDF) was initially proposed as a mixed diffusive-con-
vective technique that offered the advantages of two systems of transmembrane
transport: diffusion and convection. This combination allowed better removal
of both middle molecules, particularly with respect to hemodialysis (HD), and
small uremic toxins when compared to hemofi ltration [1, 2].
Although HDF is characterized by processes that can negatively interfere
between diffusion and convection, leading to academic and clinical arguments
over the choice between pre-, post- or pre-/postdilution, overall the develop-
ment of HDF offers, without doubt, an important positive evolution in dialytic
Hemodiafi ltration with Endogenous Reinfusion 95
strategy. Today it is one of the fastest-growing segments for the treatment of
chronic uremic patients.
Hemodiafiltration Problems
Hemodiafi ltration is associated with 3 problems:
• interference between convection and diffusion;
• quantity and quality of the reinfusion fl uid;
• loss of important physiological components in the ultrafi ltrate.
Interference between Diffusion and Convection
Convective clearance (and therefore mass transfer) of a diffusible solute in
HDF cannot be fully represented by the ultrafi ltration fl ow rate (QUF) – in that
the simultaneous process of both convection and diffusion diminish the solute’s
concentration. To resolve this problem, Ghezzi et al. [3, 4] proposed a novel
form of HDF that used a 2-stage fi lter, in series, to separate diffusion from con-
vection. The 2 stages permitted simultaneous convection and diffusion, but also
offered several benefi ts over traditional HDF combined in one fi lter unit. The
fi rst stage of the fi lter used a membrane with high hydraulic permeability for
convective solute removal, while the second stage used a membrane with low
hydraulic permeability for diffusive solute removal and to control the patient
weight. Reinfusion of externally prepared or purchased fl uid occurred between
the 2 fi lter stages. This fl uid was equal to the QUF, in order to maintain the effec-
tive blood fl ow rate. The method was called paired fi ltration dialysis.
Quantity and Quality of the Reinfusion Solutions
The choice of the QUF depends on several factors. First, from a practical point
of view, the QUF must always be considered within the limits permitted by the QUF,
the hematocrit and total protein fractional fi ltration. Elevated QUF improves the
depurative effi ciency of the treatment, but it also necessitates large quantities of
reinfusion solution that must absolutely have a guarantee of safety for the patient.
The utilization of ready-to-use reinfusion sacks produced by the pharmaceutical
industry is associated with notable problems including handling (repeated connec-
tions to the hematic lines, storage) and cost. This has led to interest in online pro-
duction of reinfusion fl uids that can guarantee sterility and allow elevated QUF, thus
leading to economic and practical handling issues to give a good cost/benefi t ratio.
Loss of Useful Substances
Every renal replacement treatment has to guarantee optimal depuration,
subtraction of excess liquid to achieve the ideal dry weight and correction of
Wratten/Ghezzi 96
electrolyte as well as acid-base abnormalities. Often however, this comes at a
price. Many times, very useful benefi cial substances are also lost – particularly
at high QUF – and this can lead to severe depletion of substances such as total,
essential and branched-chain amino acids, vitamins, hormones and growth fac-
tors. Chronic renal failure patients often have high nutritional losses during
both convective and diffusive dialytic treatments that may be closely linked to
other patient comorbidities or that may aggravate patient health and well-being.
HDF is in particular associated with notable losses of amino acids, and it is
not surprising that higher losses are found with membranes that have higher
hydraulic permeability [5–7].
Development of Hemofiltrate Reinfusion
The easy availability of isolated continuous ultrafi ltrate (UF) during paired
fi ltration dialysis led to the hypothesis that the UF could be ‘regenerated’ and
used as an endogenous reinfusion fl uid. The fi rst attempt to regenerate the UF
was done in the early 90’s with 130 ml of noncoated mineral carbon inserted
into the ultrafi ltration line [8,9]. The method was called hemofi ltrate reinfu-
sion (HFR) and is illustrated in fi gure 1. The technique proved easy to use
and offered high treatment tolerance, an optimal balance of bicarbonate (since
this is not adsorbed and therefore is reinfused) and was also associated with a
diminished infl ammatory response often associated with the endogenous rein-
fusion [10–14].
During the years 1999–2000, HFR was improved by switching to a sorbent
cartridge containing 40 ml of a hydrophobic styrenic resin with high affi nity for
several uremic toxins and middle molecules such as 2-microglobulin, homo-
cysteine, angiogenin, parathyroid hormone and several chemokines and cyto-
kines [15, 16]. Urea, creatinine, uric acid, Na⫹, K⫹, phosphate and bicarbonate
are not adsorbed and remain unchanged after passage through the cartridge.
These can be managed during the second stage of the diffusive sector of the
circuit.
Thus, the regenerated UF in the closed circuit is an endogenous reinfusion
fl uid that is characterized as sterile, ultrapure with a physiological content of
bicarbonate and amino acids. In particular with regard to amino acids, HFR
has been associated with an amino acid loss similar to that observed with low-
fl ux membranes such as Cuprophan® and surely much lower than that of other
high-permeability membranes or HDF which have associated losses as high as
33%. The amino acid loss for HFR and low-fl ux membranes is approximately
10–11% [17, 18].
Hemodiafi ltration with Endogenous Reinfusion 97
Ultrafiltrate Characteristics
The UF is a lot more than merely plasma water containing a few uremic
toxins. Studies using proteomics and other chromatographic analyses have
shown that UF contains over 18,000 proteins and peptides [19–21]. Richter et
al. [19] found that the UF, analyzed by matrix-assisted laser desorption ioniza-
tion/time-of-fl ight mass spectrometry, consisted of approximately 95% masses
that were smaller than 15 kDa. Of these, 55% were found to be fragments from
plasma protein fragments (fi brinogen, albumin, 2-microglobulin, cystatin); 7%
were hormones, growth factors and cytokines; 33% consisted of complement,
enzymes, enzyme inhibitors and transport proteins. Weissinger et al. [21] also
found a signifi cant polypeptide population in a recent study that analyzed UF
from uremic patients using either high- or low-fl ux hemodialyzers. In this study
they found a higher number of polypetides in samples obtained from uremic
HFR
Original
scheme
UF regeneration
and its use as
replacement
fluid
QUF
QR
QDi
QDo
QR⫽QUF
Cartridge
(Sorbent
bed)
QBi
QBo
Dialysate
Convection
Diffusion Weight
loss
Blood pump
UF-reinfusion
pump
QBi Blood flow at the dialyzer inlet
QBo Blood flow at the dialyzer outlet
QDi Dialysate flow at the dialyzer inlet
QDo Dialysate flow at the dialyzer outlet
UF Ultrafiltrate
QUF Ultrafiltrate flow
QR Reinfusate flow
Fig. 1. Original scheme of HFR [9]. QBi ⫽ Blood fl ow at the dialyzer inlet; QBo ⫽ blood
fl ow at the dialyzer outlet; QDi ⫽ dialysate fl ow at the dialyzer inlet; QDo ⫽ dialysate fl ow at
the dialyzer outlet; QR ⫽ reinfusate fl ow.
Wratten/Ghezzi 98
patients with high-fl ux dialyzers compared to low-fl ux dialyzers (1,394 poly-
peptides with high-fl ux dialyzers vs. 1,046 with low-fl ux dialyzers) as well as a
signifi cant difference if they obtained UF from healthy human donors by fi lter-
ing plasma through a 5- or 50-kDa fi lter (590 polypeptides for the high cutoff,
490 polypeptides for the low cutoff). Although the study focused on the charac-
terization of uremic toxins, there are certainly a lot of benefi cial substances that
are also lost during HDF with high convection. One of the advantages of HFR
over classical HDF is that the technique allows the advantages of convection
to better remove higher-molecular-weight toxins, but also reinfuses important
vitamins, hormones and other physiological compounds.
How Does Hemofiltrate Reinfusion Work?
HFR is a renal replacement therapy that utilizes convection, diffusion and
adsorption (fi g. 1). It uses a double-stage fi lter that consists of a high-fl ux poly-
ethersulfone (DiapesTM) fi lter in the fi rst convective stage and a low-fl ux poly-
ethersulfone fi lter (Diapes) in the second diffusive stage. The stages of the fi lter
allow complete separation of convection from diffusion. The convective part
of the fi rst stage allows pure UF to pass through a sorbent resin cartridge. The
resin is a hydrophobic styrenic resin consisting of many pores and channels
adding to its large surface area – approximately 700 m2/g of resin. The resin
has a high affi nity for many different uremic toxins. These toxins are adsorbed
to the resin beads, and the purifi ed UF is then reinfused to a port between the
fi rst and second stages of the fi lter. The fi rst convective/adsorption stage has no
net fl uid removal. The blood and reinfused clean UF then undergo traditional
dialysis. The second stage works by classical HD. This is also where the patient
net fl uid loss occurs.
One of the reasons that the UF was chosen to pass through the cartridge
instead of direct hemoperfusion, was that the UF has a slower fl ow rate than
the blood and this allows a longer contact time with the resin and a higher toxin
adsorption. In addition there are no problems related to hemoincompatibility
due to the absence of infl ammatory cells and platelets. The cartridge adsorp-
tion was maximized by various studies to determine the maximal adsorption
at different fl ow rates for different cartridge diameters and quantities of resin.
The treatment is usually performed on the Formula PlusTM dialysis machine
(Bellco-Sorin) which has a particular software program that automatically
determines the best QUF based initially on the maximal linear velocity (the
fl ow rate that gives the best adsorption). The machine also determines the
patient’s hematocrit and transmembrane pressure to adjust the QUF based on
these parameters. Thus, the QUF is usually higher at the start of the treatment
Hemodiafi ltration with Endogenous Reinfusion 99
and then adjusts if necessary to reduce the fl ow rate based on changes in hemo-
concentration [22].
Clinical Benefits of Hemofiltrate Reinfusion
There are several studies that show a clinical benefi t for patients using
HFR. HFR is generally indicated for end-stage renal disease patients with an
increased risk of complications related to infl ammation, malnutrition and ath-
erosclerosis. This category of patients includes patients with diabetes, high
levels of C-reactive protein, elderly patients and patients at higher risk of car-
diovascular problems.
Meloni et al. [23, 24] performed a study to determine whether HFR in a
postdilution mode would have a benefi t in uremic toxin and cytokine removal.
They observed high urea and 2-microglobulin removal, and surprisingly also
had effi cient cytokine reduction despite the high sieving coeffi cients associated
with interleukin 6 and tumor necrosis factor ␣. The cartridge is also able to
adsorb signifi cant amounts of homocysteine without signifi cant adsorption of
vitamin B12 or folate as described by Splendiani et al. [25]. They suggest that
this mechanism may also be important in reducing cardiovascular risk.
Another study by Bolasco et al. [26] also observed a signifi cant reduction
in C-reactive protein in HFR patients compared to when the patients had stan-
dard HD. Of interest they also observed improved phosphate removal which they
associated with a deceleration in bone turnover and reduced total and bone alka-
line phosphatase. Both cytokines and infl ammation have been linked to harmful
bone metabolism, and alkaline phosphatase has been associated with mortality.
Panichi et al. [27] performed a study to determine whether HFR and online
HDF had an effect on infl ammatory and nutritional markers. The study was
designed as a crossover study with a 1-month washout period of standard HD
between a 4-month period of HFR followed by a second period of online HDF
(or vice versa). Both techniques signifi cantly improved infl ammation indicated
by reduced levels of C-reactive protein and interleukin 6 compared to the start
from standard HD or during the washout period. In addition they observed
increased levels of interleukin 10.
A more recent crossover study comparing HFR to standard HD has recently
been completed by Calò et al. [28]. This study not only looked at circulating
concentrations of molecules associated with infl ammation and oxidative stress,
but also determined long-term changes in gene expression in patients under-
going HFR. They observed reduced mRNA production and protein expression
of p22phox and plasminogen activator inhibitor 1 (PAI-1) compared with HD.
Both p22phox and PAI-1 are implicated in infl ammation and oxidative stress
Wratten/Ghezzi 100
[29–32]. p22phox is an important subunit of the NAD(P)H complex that pro-
duces most of the reactive oxygen species in vascular tissue. The p22phox sub-
unit has been implicated in vascular hypertrophy and is often upregulated in
atherosclerotic vessels and vessels of diabetic individuals. PAI-1 inhibits fi bri-
nolysis and is also upregulated in insulin resistance. In fact, insulin, glucose and
very-low-density lipoprotein triglyceride also stimulate PAI-1 transcription and
secretion in endothelial cells. It is thought to be important in progression of cor-
onary syndromes and development of myocardial infarcts, in particular because
fi brinolysis can be reduced in venous occlusion. In addition they also observed
decreased levels of circulating oxidized low-density lipoprotein compared to no
change in standard HD.
Despite good removal of uremic toxins and reduction of infl ammatory
molecules, HFR is also associated with a sparing of amino acids. Typically
high-fl ux HD or HDF is associated with a 25–30% loss of amino acids which
can amount up to a loss of 3–4 g/treatment [5, 33]. Ragazzoni et al. [17] showed
a signifi cant sparing in essential, branched and total amino acids during HFR
compared to online HDF.
A more recent application has been to optimize sodium balance during
the HFR treatment by measuring the conductivity of the isolated UF [34].
Frequently, both HD and HDF are associated with sodium imbalances lead-
ing to clinical symptoms such as hypotension, headache and nausea during
the dialytic treatment. Ursino et al. [34] developed a mathematical model to
predict the solute kinetics, osmolality and fl uid volume changes during HFR.
The model was validated in a small clinical study that showed good agreement
between measured and predicted sodium concentrations. Studies are currently
under way to evaluate a biofeedback model.
Conclusions
Although HDF clearly shows increased clinical and survival benefi ts,
chronic renal failure patients continue to have problems associated with infl am-
mation, oxidative stress and cardiovascular morbidity and mortality. Diabetes as
a cause of end-stage renal disease is increasing at an alarming rate. The dialysis
patient population is also getting older. These factors may be important when
considering what type of convective therapy to use. Any type of renal replace-
ment therapy has to be a compromise between optimization of toxin removal
and eventual loss of benefi cial physiological substances. HFR may be able to
offer distinct advantages related to infl ammation and nutrition for patients with
high levels of comorbidities or syndromes associated with infl ammation and
cardiovascular complications.
Hemodiafi ltration with Endogenous Reinfusion 101
The constant increase in anagraphic patient age and comorbidities for
patients requiring renal replacement therapies has translated in a new group of
patients with a more fragile clinical profi le and, unfortunately, often a poorer
quality of life. Renal replacement therapies have made tremendous strides
but should aim even further at providing treatment that is not merely aimed
at life-sustaining, but rather towards providing an acceptable quality of life.
HFR represents a biotechnological response that offers a rationale and adequate
treatment that may be useful in limiting (or decreasing) complications related to
the new profi le of many chronic renal failure patients today.
References
1 Kunitomo T, Lowrie EG, Kumazawa S, et al: Controlled ultrafi ltration with hemodialysis: analysis
of coupling between convective and diffusive mass tansfer in a new HD-UF system. Trans Am Soc
Artif Intern Organs 1977;23:234–243.
2 Leber HW, Wizemann V, Goubeaud G, et al: Simultaneous hemodialysis/hemofi ltration: an
effective alternative to hemofi ltration and conventional hemodialysis in the treatment of uremic
patients. Clin Nephrol 1978;9:115–121.
3 Ghezzi PM, Frigato G, Fantini GF, et al: Theoretical model and fi rst clinical results of the paired
fi ltration dialysis (PFD). Life Support Syst 1983;1(suppl 1):271–274.
4 Ghezzi PM, Sanz-Moreno C, Gervasio R, et al: Technical requirements for rapid high-effi ciency
therapy in uremic patients: paired fi ltration dialysis (PFD). Trans Am Soc Artif Intern Organs
1987;33:546–550.
5 Ikizler TA, Flakoll PJ, Parker RA, Hakim RM: Amino acid and albumin losses during hemodialy-
sis. Kidney Int 1994;46: 830–837.
6 Navarro JF, Marcen R, Teruel JL, et al: Effect of different membranes on amino-acid losses during
hemodialysis. Nephrol Dial Transplant 1998;13:113–117.
7 Navarro JF, Mora C, Leon C, et al: Amino acid losses during hemodialysis with polyacrylonitrile
membranes: effect of intradialytic amino acid supplementation on plasma amino acid concentra-
tions and nutritional variables in nondiabetic patients. Am J Clin Nutr 2000;71:765–773.
8 Ghezzi PM, Gervasio R, Tessore V, et al: Hemodiafi ltration without replacement fl uid: experimen-
tal study. Trans Am Soc Artif Intern Organs 1992;38:61–65.
9 Ghezzi PM, et al: Use of the ultrafi ltrate obtained in two-chamber hemodiafi ltration as replace-
ment fl uid. Int J Artif Organs 1991;14:227–234.
10 Arese M, Cristol JP, Bosc JY, et al. Removal of constitutive and inducible nitric oxide synthase-
active compounds in a modifi ed hemodiafi ltration with on-line production of substitution fl uid: the
contribution of convection and diffusion. Int J Artif Organs 1996;19:704–711.
11 De Francisco AL, Botella J, Escallada R, et al: Hemodiafi ltration with sorbent-regenerated ultrafi l-
trate as replacement fl uid: a multicenter study. Nephrol Dial Transplant 1997;12:528–534.
12 La Greca G, Brendolan A, Ghezzi PM, et al: The concept of sorbents in hemodialysis. Int J Artif
Organs 1998;21:303–308.
13 Botella J, Ghezzi PM, Sanz-Moreno C. Adsorption in hemodialysis. Kidney Int 2000;76:S60–
S65.
14 De Francisco ALM, Ghezzi PM, Brendolan A, et al: Hemodiafi ltration with online regeneration of
the ultrafi ltrate. Kidney Int 2000;76:S66–S71.
15 De Nitti C, Giordano R, Gervasio R, et al: Choosing new adsorbents for endogenous ultrapure
infusion fl uid: performances, safety and fl ow distribution. Int J Artif Organs 2001;24:765–776.
16 Tetta C, Ghezzi PM, De Nitti C, et al: New options for on-line hemodiafi ltration; in Ronco C,
La Greca G (eds): Hemodialysis Technology. Contrib Nephrol. Karger, Basel, 2002, vol 137, pp
212–220.
Wratten/Ghezzi 102
17 Ragazzoni E, Carpani P, Agliata S, et al: HFR vs HDF on-line: valutazione della perdita amino-
acidica plasmatica. G Ital Nefrol 2004;21(suppl 30):85–90.
18 De Simone W, De Simone M, De Simone A, et al: Aspetti dell’emodiafi ltrazione online con rige-
nerazione e reinfusione dell’ultrafi ltrato (HFR). Studio multicentrico. G Ital Nefrol 2004;21(suppl
30):161–167.
19 Richter R, Schulz-Knappe P, Schrader M, et al: Composition of the peptide fraction in human blood
plasma: database of circulating human peptides. J Chromatogr B Biomed Sci Appl 1999;726:
25–35.
20 Lefl er DM, Pafford RG, Black NA, et al: Identifi cation of proteins in slow continuous ultrafi ltrate
by reversed-phase chromatography and proteomics. J Proteome Res 2004;3:1254–1260.
21 Weissinger EM, Kaiser T, Meert N, et al: Proteomics: a novel tool to unravel the patho-physiology
of uraemia. Nephrol Dial Transplant 2004;19:3068–3077.
22 Wratten ML, Sereni L, Lupotti M, et al: Optimization of a HFR sorbent cartridge for high molecu-
lar weight uremic toxins. G Ital Nefrol 2004;21(suppl 30):67–70.
23 Meloni C, Ghezzi PM, Cipriani S, et al: One year of experience in postdilution hemofi ltration with
online reinfusion of regenerated ultrafi ltrate. Blood Purif 2004;22:505–509.
24 Meloni C, Ghezzi PM, Cipriani S, et al: Hemodiafi ltration with post-dilution reinfusion of the
regenerated ultrafi ltrate: a new on-line technique. Clin Nephrol 2005;63:106–112.
25 Splendiani G, De Angelis S, Tullio T, et al: Selective adsorption of homocysteine using an HFR
online technique. Artif Organs 2004;28:592–595.
26 Bolasco PG, Ghezzi PM, Ferrara R, et al: Effect of on-line hemodiafi ltration with endogenous
reinfusion (HFR) on the calcium-phosphorus metabolism: medium-term effects. Int J Artif Organs
2006;29:1042–1052.
27 Panichi V, Manca-Rizza G, Paoletti S, et al: Effects on infl ammatory and nutritional markers of
haemodiafi ltration with online regeneration of ultrafi ltrate (HFR) vs online haemodiafi ltration: a
cross-over randomized multicentre trial. Nephrol Dial Transplant 2006;21:756–762.
28 Calò LA, Naso A, Carraro G, et al: Effect of haemodiafi ltration with online regeneration of ultra-
fi ltrate on oxidative stress in dialysis patients. Nephrol Dial Transplant 2007;22:1413–1419.
29 Bonora E: The metabolic syndrome and cardiovascular disease. Ann Med 2006;38:64–80.
30 Yamamoto K, Takeshita K, Kojima T, et al: Aging and plasminogen activator inhibitor-1 (PAI-1)
regulation: implication in the pathogenesis of thrombotic disorders in the elderly. Cardiovasc Res
2005;66:276–285.
31 Hoekstra T, Geleijnse JM, Schouten EG, Kluft C: Plasminogen activator inhibitor-type 1: its
plasma determinants and relation with cardiovascular risk. Thromb Haemost 2004;91:861–872.
32 Gill PS, Wilcox CS: NADPH oxidases in the kidney. Antioxid Redox Signal 2006;8:1597–1607.
33 Prado de Negreiros Nogueira Maduro I, Elias NM, Nonino Borges CB: Total nitrogen and free
amino acid losses and protein calorie malnutrition of hemodialysis patients: do they really matter?
Nephron Clin Pract 2007;105:c9-c17.
34 Ursino M, Coli L, Magosso E, et al: A mathematical model for the prediction of solute kinetics,
osmolarity and fl uid volume changes during hemodiafi ltration with on-line regeneration of ultra-
fi ltrate (HFR). Int J Artif Organs 2006;29:1031–1040.
Mary Lou Wratten
Sorin Group Italia
Via Camurana, 1
IT–41037 Mirandola, Modena (Italy)
Tel. ⫹39 0535 29281, Fax ⫹39 0535 29282, E-Mail marylou.wratten@sorin.com
Ronco C, Canaud B, Aljama P (eds): Hemodiafi ltration.
Contrib Nephrol. Basel, Karger, 2007, vol 158, pp 103–109
Low- (Classical) and High-Efficiency
Haemodiafiltration
Volker Wizemann
Georg-Haas-Dialysezentrum, Giessen, Germany
Abstract
The distinction between high-effi ciency haemodiafi ltration (HDF), usually applied with
online preparation of substitution fl uid, and ‘classic’ low-effi ciency HDF (using less than 15
litres of substitution fl uid) makes sense since the magnitude of convection (expressed by
substitution volume) is important for the claimed benefi ts of HDF. Many experimental and
observational data support the notion that in comparison to conventional low-fl ux haemodi-
alysis, high-effi ciency HDF might have many clinical advantages and might prolong life.
Randomized prospective trials, such as a current Dutch trial, are overdue to prove these
hypotheses. Low-effi ciency HDF is as effective as high-fl ux haemodialysis in providing con-
vection. Clinical comparisons between high-fl ux haemodialysis and HDF are sparse. The
magnitude of convection is indirectly dependent on the degree of extracorporeal blood fl ow.
With 14-gauge needles, blood fl ows of ⬎500 ml/min can be safely maintained without
haemodynamic or hyperkalaemic consequences. With regard to blood purifi cation kinetics,
high-effi ciency HDF appears ideal for performing daily short treatments.
Copyright © 2007 S. Karger AG, Basel
Haemodiafi ltration (HDF) began to be used in the 1970s [1] under the
name ‘simultaneous haemofi ltration/haemodialysis’, which today is still help-
ful to remember, since HDF aims to combine the advantages of haemofi ltra-
tion (convection and thereby increased removal of larger substances) with those
of haemodialysis (diffusion and thereby increased removal of smaller solutes).
Thus the disadvantages of both basic methods are avoided. From the point of
view of solute transport and removal, HDF is a win-win method.
In the beginning, the amount of substitution fl uid was predetermined by
the availability of 1 single brand of 4.5 litres of sterile replacement fl uid. HDF
was therefore fi rst used with 2 bags(9 litres) of substitution fl uid, and later
with 4 bags (18 litres). The buffer substance in the bags was lactate, and in the
Hemodiafiltration Techniques
Wizemann 104
dialysate acetate. When online HDF was introduced experimentally in the
1980s and the restrictive costs of infusion fl uid no longer applied, substitution
volumes in (postdilution) HDF usually surpassed 20 litres.
Definition of Low-Efficiency and High-Efficiency Haemodiafiltration
The distinction – albeit arbitrary – between the 2 types of postdilution
HDF makes sense, since the clearances of the so-called medium- and high-
molecular-weight substances primarily depend on the magnitude of convective
transport, of which the extent of necessary substitution is an indirect measure.
The diffusive part of postdilution HDF is mostly uninfl uenced by the magnitude
of convection (substitution). Since all potential benefi ts of HDF over haemo-
dialysis are based on the addition of convection, it is rational to classify HDF
according to the magnitude of convection (achieved substitution volume). In
an international observational study, high-effi ciency HDF (postdilution) was
defi ned when the volume of replacement exceeded 15 l/session [2]. A further
argument to accept such a distinction comes from an ongoing prospective rand-
omized trial [3], in which online HDF with substitution volumes of 20–30 litres
is compared to low-fl ux haemodialysis with focus on mortality and morbidity.
A third argument for distinguishing high-effi ciency HDF from classic low-vol-
ume HDF is based on the observation that high-fl ux haemodialysis (in contrast
to low-fl ux haemodialysis) is a kind of low-volume HDF, since convection and
substitution occurs internally within the dialyser, and convection can be as high
as 10 l/session [4]. This fi nding is supported by the fi ndings of Canaud et al. [2]
in the European DOPPS population that the relative risk of mortality is identical
in low-effi ciency (-volume) HDF and high-fl ux haemodialysis. Thus, subsum-
ing all forms of HDF (with substitution volumes ranging from 3 to 65 litres) as
1 homogeneous treatment form and comparing the outcome with haemodialysis
not surprisingly showed no differences [5].
In the following the focus will be on high-effi ciency HDF, defi ned as online
HDF requiring 15 litres or more of postdilution volume substitution per session.
The postdilution mode was chosen because most observational and prospective
trials use this method.
Rationale for High-Efficiency Haemodiafiltration
The reported benefi ts of HDF on amyloidosis, nutrition, renal anaemia,
cardiovascular morbidity, reduction of infl ammatory markers and survival are
reviewed elsewhere [2, 3, 6–9]. Further recent observational data [10] confi rm
Low- (Classical) and High-Effi ciency Haemodiafi ltration 105
the DOPPS fi ndings [2]. However, most evidence is class III–V, and despite
sophisticated adjustments a bias in patient selection cannot be excluded.
Furthermore, HDF patients have often been compared with haemodialysis
patients with a questionable quality of treatment, which might rather indicate
method-linked side effects such as achieving higher Kt/V scores, sterile dia-
lysate, consequences of more biocompatible membranes and suffi cient com-
pensation of acidosis rather than the benefi cial effects of convection itself. All
these described benefi cial effects could also have been achieved by using the
full potential of haemodialysis, even low-fl ux haemodialysis. The European
MPO study is a prospective randomized trial comparing high- and low-fl ux
haemodialysis groups of sicker patients (albumin less than 4 g/dl) and the pri-
mary objective is survival. Since this study includes only incident patients and
the distinction of membrane permeability is clear cut, the outcome might be
more conclusive than that of the HEMO study, and if a survival benefi t of the
high-fl ux group is found, it might be interesting to learn whether even more
convection (with HDF) could surpass the benefi ts of high-fl ux dialysis. Today,
we are in a similar situation as we have been for 20 years, and we urgently
require confi rmation of the observational studies by randomized controlled tri-
als such as the ongoing CONTRAST study [3]. The well-known technical ver-
satility of HDF to be applied also in pre- or mixed-dilution mode [for review,
see 6] and to advocate HDF in all its modes (according to the rheological situ-
ations) as standard therapy appears to be 1 step too many at the present level of
hard evidence.
Convection
Less controversial is the statement that the convective part of HDF should
be maximized since the clearance of all medium-molecular marker substances
directly depends on the magnitude of convection [11, 12]. Figure 1 shows that
clearance of 2-microglobulin can be markedly increased by increased convec-
tion (substitution volume is a surrogate), whereas small solutes are infl uenced
only marginally. Figure 2 demonstrates that diffusion is the decisive factor for
small-solute clearances.
Magnitude of Blood Flow
Convection and diffusion are both indirectly and directly dependent on the
magnitude of blood fl ow. Absolute diffusive clearances of urea increase with
increasing blood fl ow, whereas 2-microglobulin as a marker for convective
Wizemann 106
transport is indirectly related to blood fl ow, because only a fi xed proportion
(approx. 20%) of the blood entering the dialyser can be fi ltered safely. The pro-
portion of fi ltrate is also (to a lesser extent than blood fl ow) associated with
the degree of renal anaemia and the haemorheological situation. Under routine
conditions the main determinant for the magnitude of convection is the quantity
of blood fl ow.
For both parts (convection and diffusion) of HDF, high blood fl ow is a
prerequisite for high effi ciency. In some European countries, such as Germany,
comparatively low blood fl ows are traditional, and higher extracorporeal blood
fl ows are met with 2 typical arguments, the fi rst being ‘results in haemodynamic
instability during dialysis’. This topic was already studied in detail 17 years ago
by Ronco et al. [13], who demonstrated that increasing blood fl ow had no infl u-
ence at all on haemodynamics. We systematically followed 32 haemodialysis
patients in whom blood fl ows of 250 ml/min were compared to 500 ml/min, and
no differences were found regarding heart rate, blood pressure or intradialytic
0
50
100
150
200
250
300
350
400
0 50 100 150 200 250
QS (ml/min)
Clearance (ml/min)
Urea
2-Microglobulin
Postdilution
Predilution
Postdilution
Predilution
Fig. 1. Clearance as a function of substitution fl uid fl ow QS (Fresenius FX60, blood
fl ow 400 ml/min, dialysate fl ow 800 ml/min). Clearance increase by convection is more pro-
nounced for 2-microglobulin.
Low- (Classical) and High-Effi ciency Haemodiafi ltration 107
hypotension [Techert et al., in press]. Since increasing blood fl ow could alter
the thermal balance, and since this factor was identifi ed as the cause of the
observed haemodynamic stability with convective methods [14], special atten-
tion should be paid to achieving a neutral thermal balance during high-effi -
ciency HDF, which not only requires high substitution volumes, but also high
blood fl ows, as thermal factors.
The second argument against higher blood fl ows – ‘leads to haemolysis
and hyperkalaemia’ – to our knowledge lacks any confi rmation in the literature.
In our series with 32 patients, blood fl ow rates of 500 ml/min were associated
with identical plasma-free haemoglobin levels at the end of dialysis to treat-
ments with blood fl ows of 250 ml/min.
It should be noted that the vast majority of arteriovenous fi stulas can pro-
vide a blood fl ow of 400–500 ml/min when 14-G needles are used. Under these
conditions, negative ‘arterial’ pressures are even lower than with blood fl ows of
250 ml/min and use of 17-gauge needles.
0
50
100
150
200
250
300
350
400
450
200 300 400 500 600
QB (ml/min)
Clearance (ml/min)
Creatinine
Urea
2-Microglobulin
Fig. 2. Clearance as a function of blood fl ow QB (Fresenius FX100, dialysate fl ow
800 ml/min, substitution fl uid fl ow 60 ml/min, online, postdilution). Clearance of small mol-
ecules is strongly increased by higher blood fl ow rates.
Wizemann 108
Online mixed HDF (substitution pre- and postdilution), controlled by trans-
membrane pressure ultrafi ltration feedback, can further increase the clearances
of marker solutes (with the exception of urea) at a given blood fl ow [15]. The
feedback system automatically adjusts the infusion volume and site according to
blood fl ow conditions and internal pressures to maintain a constant and safe fi l-
trate fl ow. From the aspect of safety it remains open whether a 10% increase in 2-
microglobulin clearance justifi es a more than doubling of total infusion volumes.
Variation of High-Efficiency Haemodiafiltration Duration and Frequency
Due to the compartmentalization of the body and the transport properties
of larger solutes, an effective blood purifi cation method focusing on such sol-
utes must not necessarily be applied in the same fashion as conventional haemo-
dialysis, which is usually done for 4–5 h 3 times a week. Kinetic studies show
that the vascular compartment is already almost depleted of 2-microglobulin,
or substances which behave like ‘middle molecules’ such as phosphate, when
high-effi ciency HDF is applied [Beck et al., in press]. In comparison, during
conventional haemodialysis the depletion curves are shifted to the right, indi-
cating that longer duration treatment in contrast to HDF makes sense. Thus,
the application of high-effi ciency HDF of short duration (e.g. for 2–2.5 h) daily
would lead to a considerable increase in removal of ‘middle-molecular’ sub-
stances compared to an identical HDF of longer duration but applied only 3
times a week [Beck et al., in press]. Maduell et al. [16] clinically tested a daily
schedule of short high-effi ciency HDF and found a 21% reduction in pretreat-
ment 2-microglobulin compared to a conventional high-effi ciency schedule of
identical weekly duration (3 ⫻ 4 h/week).
Conclusions
High-effi ciency (postdilution) HDF is a potent blood purifi cation method,
and most published and proposed studies refer to this mode. It remains to be
clarifi ed by class I evidence whether HDF is as safe as low-fl ux and high-fl ux
haemodialysis, and whether there are mortality and morbidity advantages.
References
1 Leber HW, Wizemann V, Goubeaud G, et al: Simultaneous hemofi ltration/hemodialysis: an
effective alternative to hemofi ltration and conventional hemodialysis in the treatment of uremic
patients. Clin Nephrol 1978;9:115–120.
Low- (Classical) and High-Effi ciency Haemodiafi ltration 109
2 Canaud B, Bragg-Gresham JL, Marshall MR, et al: Mortality risk for patients receiving hemodia-
fi ltration versus hemodialysis: European results from the DOPPS. Kidney Int DOI: 10.1038/sj.ki.
5000447.
3 Penne EL, Blankestijn PJ, Bots ML, et al: Resolving controversies regarding hemodiafi ltration
versus hemodialysis: the Dutch Convective Transport Study. Semin Dial 2005;18:47–51.
4 Ronco C, Brendolan A, Lupi A, et al: Effects of reduced inner diameter of hollow fi bers in hemo-
dialyzers. Kidney Int 2000;58:809–817.
5 Rabindranath KS, Strippoli GFM, Roderick P, et al: Comparison of hemodialysis, hemofi ltra-
tion, and acetate-free biofi ltration for ESRD: systematic review. Am J Kidney Dis DOI: 10.1053/
ajkd.2004.11.008.
6 Canaud B, Levesque R, Krieter D, et al: On-line hemodiafi ltration as routine treatment of end-
stage renal failure: why pre- or mixed dilution mode is necessary in on-line hemodiafi ltration
today? Blood Purif 2004;22(suppl 2):40–48.
7 Canaud B, Morena M, Leray-Moragues H, et al: Overview of clinical studies in hemodiafi ltration:
what do we need now? Hemodial Int 2006;10:S5–S12.
8 Maduell F: Hemodiafi ltration. Hemodial Int 2005;9:47–55.
9 Kooman JP, van der Sande FM, Beerenhout CM, et al: On-line fi ltration therapies: emerging hori-
zons. Blood Purif 2006;24:159–162.
10 Jirka T, Cesare S, di Benedetto A, Perera Chang M, et al: Mortality risks for patients receiving
hemodiafi ltration versus hemodialysis. Kidney Int 2006;70:1524.
11 Wizemann V, Rawer P, Schmidt H, et al: Effi ciency of hemodialysis, hemofi ltration, hemodiafi ltra-
tion. Hemodiafi ltration – Proceedings 1st Symposium Giessen. Oberursel, Verlag Hygieneplan,
1981.
12 Wizemann V, Külz M, Techert F, et al: Effi cacy of haemodiafi ltration. Nephrol Dial Transplant
2001;16(suppl 4):27–30.
13 Ronco C, Feriani M, Chiaramonte S, et al: Impact of high blood fl ows on vascular stability in
haemodialysis. Nephrol Dial Transplant 1990;5(suppl):109–114.
14 van der Sande FM, Kooman JP, Konings CJ, et al: Thermal effects and blood pressure response
during post-dilution hemodiafi ltration and hemodialysis: the effect of amount of replacement fl uid
and dialysate temperature. Am J Soc Nephrol 2001;12:1916–1920.
15 Pedrini LA, De Christofaro V: On-line mixed hemodiafi ltration with a feedback for ultrafi ltration
control: effect on middle molecule removal. Kidney Int 2003;64:1505–1513.
16 Maduell F, Navarro V, Torregrosa E, et al: Change from thrice weekly on-line hemodiafi ltration to
short daily on-line hemodiafi ltration. Kidney Int 2003;64:305–313.
Prof. Dr. Volker Wizemann
Georg-Haas-Dialysezentrum
Johann-Sebastian-Bach-Strasse 40
DE–35392 Giessen (Germany)
Tel. ⫹49 641 92207 15, Fax ⫹49 641 29187, E-Mail wizemann.volker@phv-dialyse.de
Ronco C, Canaud B, Aljama P (eds): Hemodiafi ltration.
Contrib Nephrol. Basel, Karger, 2007, vol 158, pp 110–122
Online Hemodiafiltration
Technical Options and Best Clinical Practices
B. Canaud
Nephrology, Dialysis and Intensive Care Unit, Aider and Renal Research and
Training Institute, Lapeyronie University Hospital, Montpellier, France
Abstract
Online production of substitution fl uid by ‘cold sterilization’ (ultrafi ltration) of dialysis
fl uid gives access to virtually unlimited amounts of sterile and nonpyrogenic solution. The
incorporation of the online hemodiafi ltration (ol-HDF) module into the dialysis proportion-
ing machine hardware simplifi es the handling procedure, secures the process by keeping the
safety regulation of the monitor and offers virtually unlimited amounts of sterile and nonpy-
rogenic substitutive solution. The safety of the ol-HDF relies upon use of ultrapure water and
strict and permanent highly hygienic rules of use. The use of a specifi cally designed certifi ed
HDF machine is also mandatory. Several forms of ol-HDF have been developed and used to
cover specifi c clinical needs of chronic kidney disease patients. Conventional ol-HDF are
classifi ed according to the mode of substitution as post-, pre- and mixed dilution. Alternative-
based ol-HDF incorporate push/pull HDF, double high-fl ux HDF, paired HDF and middilu-
tion HDF. A very simple description of these methods is provided in this section. Best clinical
practices are summarized in this section to optimize performances of ol-HDF and maximize
the safety of the method. It is noteworthy to stress the important role of blood fl ow, fl uid
volume exchange, hemodiafi lter performances and duration of sessions in the overall treat-
ment effi cacy. It is also crucial to insist on the importance of strict hygienic handling, micro-
biology monitoring and the quality assurance process to ensure the safety of the method. In
addition, ol-HDF offers the best technical platform to develop new therapeutic strategies
such as daily treatment, total automation of priming and cleansing procedures and biofeed-
back volume control.
Copyright © 2007 S. Karger AG, Basel
Technical Options and Best Clinical Practices in Online Hemodiafi ltration 111
Rationale for Online Hemodiafiltration
Hemodiafi ltration (HDF) is an established treatment modality that tends to
gain in popularity since it offers now an optimal and affordable form of renal
replacement therapy in chronic renal disease patients [1–3]. By enhancing
and enlarging the molecular-weight spectrum of uremic toxins removed, HDF
improves dialysis effi ciency [4–6]. By increasing instantaneous fl uxes of various
solutes, HDF facilitates the correction of internal milieu disturbances [7]. By
improving the global hemocompatibility of the dialysis system (synthetic low-
reactive membrane, ultrapure dialysis fl uid, protein membrane coating), HDF
contributes to reduce side effects and complications of long-term dialysis [8, 9].
The behavior of solute clearances is unique in HDF, since the simultaneous
occurrence of diffusive and convective clearances in the same dialyzer tends
to cut down the power of each process [10]. Accordingly, the global effi ciency
of HDF may not be considered as a simple sum of diffusive and convective
clearances, it is a more complex relationship. In addition, the site of substitu-
tion fl uid (post-, pre-, mid- or mixed-dilution mode) affects signifi cantly HDF
performances.
Online production of substitution fl uid by ‘cold sterilization’ (ultrafi ltra-
tion) of dialysis fl uid gives access to virtually unlimited amounts of sterile and
nonpyrogenic intravenous-grade solution [11–13]. The incorporation of the
online HDF (ol-HDF) module into the dialysis proportioning machine hardware
is benefi cial: fi rst, it simplifi es the handling procedure compared to bag use;
second, it secures the process by enslaving the infusion module to the safety
regulation of the HDF monitor; third, it permits to check regularly the physi-
cal integrity of the ultrafi lters by means of a built-in air pressure test [14]. This
technically advantageous and cost-competitive production of substitution fl uid
has been employed to develop various forms of high-effi ciency HDF modalities
(postdilution, predilution, mixed-dilution, middilution). The generic term com-
monly used to characterize these modalities is ol-HDF [15–17].
Technical Requisite and Hygiene Handling
The safety of the ol-HDF is relying upon strict and permanent conditions
of use and handling. Compliance with guidelines is the only way to prevent
adverse effects and to warrant success of the ol-HDF program.
The use of ultrapure water (UPW) to feed the HDF machine is a basic
requirement for ol-HDF [18]. Several studies have updated our knowledge con-
cerning the water treatment system required [19]. UPW is a high-grade quality
water which has been developed mainly to satisfy the needs of the semiconductor
Canaud 112
industry. For HDF purposes, UPW refers to reverse-osmosis-treated water (one
or more stages of reverse osmosis in series) with a resistivity in the range of
0.1–5.0 M⍀/cm with a very low level of bacterial and endotoxin contamination,
i.e. ⬎100 CFU/l and endotoxin Limulus amebocyte lysate (LAL) ⬍0.03 endo-
toxin units (EU)/ml. Production and distribution of UPW to the HDF machine
may be achieved with several water treatment options. Distribution pipes must
be adequately designed to prevent stagnation, to eliminate dead arms and other
recontamination sites. Permanent recirculation of treated water through a closed
loop circuit with a microfi ltration system is required when a buffer tank is used.
The use of specifi cally designed HDF and European-Community-certifi ed
machines is necessary. Several certifi ed ol-HDF machines are presently avail-
able on the European market. Basically, these ol-HDF machines share common
features that include an infusion pump with a fl ow-measuring system, a dialy-
sate ultrafi lter module (usually two certifi ed ultrafi lters in series) placed onto
the hydraulic circuit of the machine and an enslaving system feedback control
by the machine alarm detection system. The infusate module is a captive part
of the machine which is disinfected simultaneously with each process of the
HDF machine. In some machines, a built-in pressure test (bubble point) is per-
formed periodically by the HDF monitor to check the integrity of the ultrafi lter
membrane. The infusate module consists in an adjustable pump running up to
200 ml/min with a counter calculating the total amount of fl uid infused into the
patient. The safety of the infusion module is linked to the general alarms of the
HDF monitoring system. Ultrapure dialysate fl owing into the dialysate com-
partment of the hemodiafi lter is produced through an ultrafi lter (UF1) placed
just at the exit site of the dialysate [20, 21]. A fraction of the fresh dialysate
(100/800 ml/min) produced by the proportioning HDF system is diverted by
the infusion pump and infused into the blood of the patient (either postfi lter or
prefi lter infusion). Ultrapurity of the infusate is then secured by a second-stage
ultrafi ltration (UF2) before being infused into the patient. In this confi guration,
infusate fl ow diverted from the inlet dialysate is compensated by an equivalent
ultrafi ltration fl ow taken from the patient through the hemodiafi lter with the
fl uid balancing chamber. Ultrafi lters are an integral part of the HDF machine
that are disinfected after each run and changed periodically.
Online cold sterilization of biological fl uids is based on a membrane fi l-
tration process (ultrafi lter). However, it is important to recall that the reten-
tive capacity of an ultrafi lter is restricted to certain conditions of use [22, 23].
The use of UPW, sterile electrolyte concentrates and frequent disinfection of
the HDF machine reducing the bacterial contamination level are basic require-
ments to prevent ultrafi lter bacterial overfl ow.
Hygiene handling is a crucial measure to ensure a permanent safety of the
HDF system. Measures needed to maintain the bacterial contamination at a low
Technical Options and Best Clinical Practices in Online Hemodiafi ltration 113
level have two targets: one is to maintain the ultrapurity of water feeding the HDF
machines by means of frequent disinfection of the water treatment system, destruc-
tion of biofi lm by chemical agents and/or thermochemical disinfection, change of
fi lters and disposable tubings at regular intervals and by permanent recirculation
of UPW in the distribution system; the other is to prevent recontamination and
bacterial proliferation in the HDF machine by means of frequent disinfection, use
of sterile liquid concentrate or powder and periodical changes of ultrafi lters.
Quality monitoring of the dialysate and the infusate is mandatory to detect
early microbiological contamination of the system. The microbiological inven-
tory of water, dialysate and infusate should be performed according to best prac-
tice guidelines and pharmacopeia regulation. Sampling methods, culture media
and sensitive microbiological methods have been validated and published else-
where [24, 25]. Endotoxin content (infusate and dialysate) should be assessed
using a sensitive LAL assay with a threshold detection limit of 0.03 EU/ml.
Information concerning the microbiological monitoring must be stored to prove
the quality of treatment. Such rules must be considered as a part of the good
medical practices for ol-HDF.
Hemodiafiltration Treatment Options
Conventional ol-HDF relies on the combination of diffusive and forced
convective clearances in the same hemodiafi lter module. Basically, the substitu-
tion fl uid (infusate) is a sterile nonpyrogenic solution produced extemporane-
ously from fresh dialysate by double ultrafi ltration (cold sterilization process)
and infused directly into the patient’s blood on the venous side. Infusate diverted
from the inlet dialysate is extemporaneously compensated by the fl uid balancing
system of the dialysis machine which is ultrafi ltering the same amount of fl uid
from the patient’s blood. A high ultrafi ltration rate is achieved by the dialysis
machine by increasing adequately the transmembrane pressure (TMP) applied
to the hemodiafi lter. Weight loss is required to correct patient fl uid overload and
is achieved in addition by increasing consequently the ultrafi ltration rate.
Depending on the infusion site of fl uid substitution, several HDF modalities
have been described: postdilution HDF (infusion after the hemodiafi lter); predi-
lution HDF (infusion before the hemodiafi lter); mixed HDF (infusion simultane-
ously before and after the hemodiafi lter) [26]. This is presented in fi gures 1–3.
HDF requires use of a high-fl ux dialyzer and achievement of high blood
fl ow (350–450 ml/min) with high dialysate fl ow (600–800 ml/min). It is recom-
mended to couple the infusion rate to the effective blood fl ow to reduce the fi l-
tration fraction (20–30% maximum). A simple rule of prescribing infusion fl ow
is to use one third of the inlet blood fl ow in postdilution mode and half of the
Canaud 114
inlet blood fl ow in predilution HDF. Typical infusion fl ow rates are 100 ml/min
(24 liters for a 4-hour session) in postdilution HDF and 200 ml/min (48 liters for
a 4-hour session) in predilution HDF mode, respectively, to match urea clear-
ance. Mixed pre- and postdilution HDF represents a recently introduced tech-
nical option. As it is suggested by recent studies, mixed HDF optimized fl uid
exchanges by enslaving infusion rate and fl ow repartition to the TMP trends.
Accordingly, one fraction of the substitution fl uid is infused in the postdilution
site while the other fraction is infused in the predilution site. The ratio of pre-
Dialysate outlet
⫹ultrafiltrate
Dialysate inlet
⫺infusate
Fluid
balancing
module
Sterilizing
ultrafilters
Infusion
pump
Fig. 1. Postdilution ol-HDF.
Fig. 2. Predilution ol-HDF.
Infusion
pump
Dialysate outlet
⫹ultrafiltrate
Dialysate inlet
⫺infusate
Fluid
balancing
module
Sterilizing
ultrafilters
Technical Options and Best Clinical Practices in Online Hemodiafi ltration 115
to postinfusion fl ow is feedback controlled by the HDF monitor (programmed
HDF) in order to maintain the TMP in a safe range. Based on this TMP-con-
trolled HDF, it has been shown that signifi cantly higher 2-microglobulin clear-
ances and lower albumin losses can be achieved [27, 28].
Regarding alternative-based convective HDF methods, several variants of
HDF have been described over the last decade. All of them claim to be advan-
tageous as compared to the conventional ol-HDF method. They are briefl y
described in the chronological order of description.
Push/pull HDF is based on a double-cylinder piston pump (push/pull pump)
implemented on the effl uent dialysate line of the dialysis machine. Based on
this alternate pump device, 25 alternate cycles of 20 ml of ultrafi ltration (pull)
and backfi ltration (push) are performed through the hemodialyzer per minute
meaning that 120 liters of ultrafi ltered plasma water are backfi ltered from the
fresh inlet dialysate in a 4-hour treatment [29].
Double high-fl ux hemodialysis consists in two high-fl ux dialyzers assem-
bled in series while the dialysis fl uid irrigates countercurrently the two dialyzers
[30]. By means of an adjustable clamp restriction placed on the dialysis fl uid
pathway between the two dialyzers, ultrafi ltration is promoted in the fi rst dia-
lyzer and backfi ltration in the second dialyzer [31].
Paired hemofi ltration is a double-chamber HDF technique that was ini-
tially proposed to separate convective and diffusive solute fl uxes in two modules
[32] (fi g. 4). This method is based on the association of two high-fl ux dialyz-
ers in series, one with a small surface (e.g. 0.4 m2) that permits the infusion
of substitution fl uid (backfi ltration) and the second a high-fl ux hemodialyzer
Fig. 3. Mixed-dilution ol-HDF with TMP controlled by microprocessor.
Infusion
pumps
Microprocessor
Dialysate outlet
⫹ultrafiltrate
Dialysate inlet
⫺infusate
Sterilizing
ultrafilters
Fluid
balancing
module
TMP
Canaud 116
(1.8 m2) that allows convective and diffusive exchange from dialysate (fi g. 4).
The substitution fl uid produced by cold sterilization from the fresh dialysis fl uid
is infused either on predilution mode or on postdilution mode according to the
position of the dialyzer.
HDF with endogenous reinfusion derives from paired hemofi ltration. The
main feature of HDF with endogenous reinfusion is the online regeneration of
the ultrafi ltrate by an adsorbing multilayer device [33]. The regenerated ultrafi l-
trate is then reinfused as an endogenous substitution fl uid (fi g. 5).
Middilution HDF is the last option that relies on a newly designed hemo-
diafi lter consisting in two high-fl ux fi ber bundles built in one dialyzer hous-
ing. Blood is running countercurrently (ultrafi ltration in the fi rst bundle and
diffusion in the second bundle) with an infusion performed on the distal head
(middle part of the two bundles). In this version the fi rst part of the module
ensures the convective transport (ultrafi ltration) and the second part of the mod-
ule the diffusive transport countercurrently with the dialysate after blood has
been diluted in the opposite head of the dialyzer (fi g. 6).
Online Hemodiafiltration – Best Clinical Practices
Indications for Online Hemodiafiltration
All chronic kidney disease stage 5 (CKD-5) patients requiring a renal
replacement therapy support may benefi t from ol-HDF. No specifi c contraindi-
cation to ol-HDF therapy has been reported to date.
Fig. 4. Paired hemofi ltration.
Dialysate outlet
Ultrafiltrate
Infusate
Infusion
pump
Dialysate inlet
⫺infusate Sterilizing
ultrafilters
Fluid
balancing
module
Technical Options and Best Clinical Practices in Online Hemodiafi ltration 117
Two categories of CKD patients are nevertheless particularly suitable for
ol-HDF: fi rst, unstable CKD patients presenting with severe cardiovascular risk
factors, chronic hypotension, diabetics, elderly and uncompliant patients with
large interdialytic fl uid gain; second, junior and senior CKD patients requiring
a large dialysis dose covering the whole spectrum of uremic toxins or being
exposed to prolonged periods of renal replacement therapy. Selection of CKD
patients may be justifi ed by the better vascular stability and by the improved
biocompatibility provided by the method.
Ultrafiltrate
Sorbent
Infusate
Dialysate outlet
Dialysate inlet
⫺infusate Sterilizing
ultrafilters
Fluid
balancing
module
Fig. 5. Hemofi ltration with regeneration of infusate.
Fig. 6. Middilution ol-HDF.
Mixing
chamber
Fluid
balancing
module
Infusion
pump
Dialysate outlet
⫹ultrafiltrate
Dialysate inlet
⫺infusate
Sterilizing
ultrafilters
Canaud 118
Vascular Access
Patients treated with ol-HDF require an access capable of delivering an
extracorporeal blood fl ow of at least 350 ml/min, and preferably higher, on a
reliable basis. High blood rate facilitates the ultrafi ltration fl ow and reduces the
TMP problems during the session.
Hemodiafilter
A high-fl ux, high-effi ciency dialyzer is required. The membrane should
have a high hydraulic permeability (ultrafi ltration coeffi cient KUF ⬎50 ml/h/mm
Hg), high solute permeability (mass transfer-area coeffi cient KoA urea ⬎600 and
2-microglobulin ⬎60 ml/min) and large surface of exchange (1.50–2.10 m2).
Prescription and Substitution Fluid Volume per Session
The conventional ol-HDF treatment schedule is based on 3 dialysis ses-
sions per week of 4 h (12 h/week). In this relatively short treatment time, it is
of paramount importance to ensure high blood fl ows (400 ml/min) coupled
with high dialysate and/or infusate fl ow rates in order to optimize solute
exchange.
By increasing the frequency and/or duration of HDF sessions, it is also
possible to achieve a more physiological and more effective treatment.
Follow-Up and Monitoring of Patients Treated with Online
Hemodiafiltration
Follow-up and monitoring of ol-HDF-treated patients are strictly equiva-
lent to those of patients treated with regular conventional hemodialysis. Dialysis
adequacy targets as recommended by the Kidney Disease Outcome Quality
Initiative and the European Best Practice Guidelines should be equivalent in
terms of extracellular fl uid volume control, blood pressure control, minimum
dialysis dose delivered (urea Kt/Vdp ⬎1.2), uremia control, acidosis and hyper-
kaliemia correction, phosphorus, calcium and parathyroid hormone control,
and anemia correction.
On a regular basis, ol-HDF provides a higher solute removal rate as com-
pared to conventional low- and high-fl ux hemodialysis for low- and middle-size
uremic toxins including the 2-microglobulin. On a long-term basis, this higher
effi cacy translates in a reduction of the time-averaged concentration of blood
2-microglobulin concentrations meaning that this middle-size marker should
be routinely incorporated in the criteria of dialysis adequacy.
Due to the high volume of fl uid exchanged per session (25–50 l/session) it
is also recommended to follow on a monthly basis the infl ammatory profi le of
ol-HDF-treated patients (e.g. high-sensitivity C-reactive protein) and the nutri-
tional markers (albumin and transthyretin).
Technical Options and Best Clinical Practices in Online Hemodiafi ltration 119
With ol-HDF, achieving a minimum of international adequacy standards
is however quite easy due to the better hemodynamic stability and the higher
solute removal capacity of the method.
Microbiological Monitoring
The ultrapurity of the bicarbonate-based dialysate and infusate produced
by ol-HDF machines relies on three components: fi rst, a well-bioengineered
design water production and delivery system; second, a strict control of hygienic
rules aiming to maintain regular disinfection procedures of the water treatment
system and the proportioning HDF machines; third, a planned microbiological
inventory monitoring of the complete chain of treatment.
Disinfection procedures and monitoring frequency of water treatment
system and ol-HDF machines may vary from country to country according to
specifi c regulations of the health authorities. The overall aim is to ensure at
any time the quality and safety of the ol-HDF method. It is recommended to
perform a complete disinfection of the hydraulic circuit of the ol-HDF machine
(chemical, heat or mixed) after each run as stated by best clinical practice
guidelines. A new sterile tubing set for the infusate line is requested at each new
HDF session. Periodical changes of ultrafi lters installed on inlet dialysate and
infusate lines should be performed according to manufacturer instructions or
earlier in case of technical failure. Disinfection of the water treatment system
and water distribution circuit should be performed as a minimum on a monthly
basis. The type of disinfection (chemical, heat or mixed) and periodicity of dis-
infection procedures may vary from one center to another but should comply
in any cases with the manufacturer recommendations and should be adapted
to the microbiological results. More frequent disinfection procedures (daily or
weekly) of the water distribution pipe using heat or mixed heat/chemical proce-
dures appear to be the optimal way of preventing bacterial contamination and
biofi lm formation.
Monitoring the microbiology of the water treatment chain and ol-HDF
machines should comply with best practice recommendations and country-spe-
cifi c rules. All recommendations have been reported in detail in the European
Best Practice Guidelines. They represent the most comprehensive and update
guidelines that should be applied to secure the ol-HDF method [34]. Water feed-
ing the HDF machines should be checked weekly during the validation phase
and at least monthly during the surveillance and maintenance period. Dialysate
and infusate produced by proportioning ol-HDF machines should be checked
at least every 3 months. Microbiological monitoring should include the culture
of water and/or dialysate and the determination of endotoxin content. Sampling
method, culture media and delay for observation have been published else-
where. Membrane fi ltration and culture on a nutrient-poor medium (R2A) are
Canaud 120
strongly recommended [35, 36]. Cultures are maintained at room temperature
(20–22°C) and observed for 7 days. Bacterial colony count and identifi cation
should be performed with appropriate methods. Endotoxin content (infusate
and dialysate) should be assessed with a sensitive LAL assay with a threshold
detection limit of 0.03 EU/ml.
Conclusions
Online HDF modalities offer at the present time the most effective renal
replacement modality for CKD-5 patients [37, 38]. High-fl ux ol-HDF allows
delivering a high ‘dialysis dose’ based on the conventional urea marker. By
enhancing the convective fl uxes, ol-HDF enlarges the spectrum and increases
the uremic toxin mass removed. ol-HDF improves the hemocompatibility pro-
fi le of extracorporeal renal replacement modalities and reduces infl ammation
of CKD-5 patients. Online production of substitution fl uid reduces the cost of
treatment and simplifi es the technical aspect of the method. In addition, by giv-
ing access to an unlimited amount of high-quality intravenous fl uid, the online
HDF concept opens new therapeutic options (feedback control of volemia,
automation of priming and restitution). These unique properties should give the
online HDF a leading position in the CKD-5 therapeutic options to enhance the
overall effi cacy of renal replacement therapy and to improve the global care of
end-stage renal failure patients [39].
References
1 Henderson LW, Besarab A, Michaels A, Bluemle LW: Blood purifi cation by ultrafi ltration and
fl uid replacement (diafi ltration). Trans ASAIO 1967;13:216–222.
2 Leber HW, Wizemann V, Goubeaud G, Rawer P, Schutterle G: Hemodiafi ltration: a new alternative
to hemofi ltration and conventional hemodialysis. Artif Organs 1978;2:150–153.
3 Passlick-Deetjen J, Pohlmeier R: On-line hemodiafi ltration: gold standard or top therapy? In
Ronco C, La Greca G (eds): Hemodialysis Technology. Contrib Nephrol. Basel, Karger, 2002, vol
137, pp 201–211.
4 Sprenger KB: Haemodiafi ltration. Life Support Syst 1983;1:127–136.
5 Ofsthun NJ, Leypoldt JK: Ultrafi ltration and backfi ltration during hemodialysis. Artif Organs
1995;19:1143–1161.
6 Leypoldt JK: Solute fl uxes in different treatment modalities. Nephrol Dial Transplant 2000;1:3–9.
7 Ledebo I: On-line hemodiafi ltration: technique and therapy. Adv Ren Replace Ther 1999;6:195–208.
8 Canaud B, Bosc JY, Leray H, Stec F, Argiles A, Leblanc M, Mion C: On-line haemodiafi ltration:
state of the art. Nephrol Dial Transplant 1998;5:3–11.
9 Canaud B, Wizemann V, Pizzarelli F, Greenwood R, Schultze G, Weber C, Falkenhagen D: Cellular
interleukin-1 receptor antagonist production in patients receiving on-line haemodiafi ltration ther-
apy. Nephrol Dial Transplant 2001;16:2181–2187.
10 Jaffrin MY, Gupta BB, Malbrancq JM: A one-dimensional model of simultaneous hemodialysis
and ultrafi ltration with highly permeable membranes. J Biomech Eng 1981;103:261–266.
Technical Options and Best Clinical Practices in Online Hemodiafi ltration 121
11 Henderson LW, Sanfelippo ML, Beans E: ‘On-line’ preparation of sterile pyrogen-free electrolyte
solution. Trans ASAIO 1978;24:465–467.
12 Shinzato T, Sezaki R, Usuda M, Maeda K, Ohbayashi S, Toyota T: Infusion-free hemodiafi ltration:
simultaneous hemofi ltration and dialysis with no need for infusion fl uid. Artif Organs 1982;6:453–
456.
13 Canaud B, Flavier JL, Argilés A, Stec F, Nguyen QV, Bouloux C, Garred LJ, Mion C:
Hemodiafi ltration with on-line production of substitution fl uid: long-term safety and quantitative
assessment of effi cacy. Contrib Nephrol 1994;108:12–22.
14 Canaud B, Nguyen QV, Argilés A, Polito C, Polaschegg HD, Mion C: Hemodiafi ltration using
dialysate as substitution fl uid. Artif Organs 1987;11:188–190.
15 Sternby J: A decade of experience with on-line hemofi ltration/hemodiafi ltration; in Maeda K,
Shinzato T (eds): Effective Hemodiafi ltration: New Methods. Contrib Nephrol. Basel, Karger,
1994, vol 108, pp 1–11.
16 Lonnemann G, Behme TC, Lenzner B, Floege J, Schulze M, Colton CK, Koch KM, Shaldon S:
Permeability of dialyzer membranes to TNF␣-inducing substances derived from water bacteria.
Kidney Int 1992;42:61–68.
17 Canaud B, Imbert E, Kaaki M, Assounga A, Nguyen QV, Stec F, Garred LJ, Boström M, Mion
C: Clinical and microbiological evaluation of a postdilutional hemofi ltration system with in-line
production of substitution fl uid. Blood Purif 1990;8:160–170.
18 Canaud B, Peyronnet P, Armynot AM, et al: Ultrapure water: a need for future dialysis. Nephrol
Dial Transplant 1986;1:110.
19 Canaud BJM, Mion CM: Water treatment for contemporary hemodialysis; in Jacobs C, Kjellstrand
CM, Koch KM, Winchester JF (eds): Replacement of Renal Function by Dialysis. Dordrecht,
Kluwer Academic Publishers, 1996, pp 231–255.
20 Schindler R, Lonnemann G, Schaffer J, Shaldon S, Koch KM, Krautzig S: The effect of ultrafi l-
tered dialysate on the cellular content of interleukin-1 receptor antagonist in patients on chronic
hemodialysis. Nephron 1994;68:229–233.
21 Mion CM, Canaud B: Should hemodialysis fl uid be sterile? Semin Dial 1993;6:28–30.
22 Mion CM, Canaud B: ‘On-site’ preparation of sterile apyrogenic electrolyte solutions for hemo-
fi ltration and hemodiafi ltration; in Cambi V (ed): Short Dialysis. Boston, Martinus Nijhoff
Publishing, 1987, vol 12, pp 261–291.
23 Sundaram S, Barrett TW, Meyer KB, Perrella C, Neto MC, King AJ, Pereira BJ: Transmembrane
passage of cytokine-inducing bacterial products across new and reprocessed polysulfone dialyz-
ers. J Am Soc Nephrol 1996;7:2183–2191.
24 Ward RA, Luehmann DA, Klein E: Are current standards for the microbiological purity of hemo-
dialysate adequate? Semin Dial 1989;2:69–72.
25 Pass T, Wright R, Sharp B, Harding GB: Culture of dialysis fl uids on nutrient-rich media for short
periods at elevated temperatures underestimate microbial contamination. Blood Purif 1996;14:
136–145.
26 Pedrini LA, De Cristofaro V, Pagliari B, et al: Mixed predilution and postdilution online hemodia-
fi ltration compared with the traditional infusion modes. Kidney Int 2000;58:2155.
27 Pedrini LA, De Cristofaro V: On-line mixed hemodiafi ltration with a feedback for ultrafi ltration
control: effect on middle-molecule removal. Kidney Int 2003;64:1505.
28 Kim ST, Yamamoto C, Taoka M, et al: Programmed fi ltration, a new method for removing large
molecules and regulating albumin leakage during hemodiafi ltration treatment. Am J Kidney Dis
2001;38(suppl 1):S220.
29 Miwa M, Shinzato T: Push/pull hemodiafi ltration: technical aspects and clinical effectiveness.
Artif Organs 1999;23:1123.
30 von Albertini B, Miller JH, Gardner PW, Shinaberger JH: Performance characteristics of the
Hemofl ow F 60 in high-fl ux hemodiafi ltration; in Streicher E, Seyffart G (eds): Highly Permeable
Membranes. Contrib Nephrol. basel, Karger, 1985, vol 46, pp 169–173.
31 Pisitkun T, Eiam-Ong S, Tiranathanagul K, Sakunsrijinda C, Manotham K, Hanvivatvong O,
Suntaranuson P, Praditpornsilpa K, Chusil S, Tungsanga K: Convective-controlled double high
fl ux hemodiafi ltration: a novel blood purifi cation modality. Int J Artif Organs 2004;27:195–
204.
Canaud 122
32 Ghezzi PM, Botella J, Sartoris AM, et al: Use of the ultrafi ltrate obtained in two-chamber (PFD)
hemodiafi ltration as replacement fl uid: experimental ex vivo and in vitro study. Int J Artif Organs
1991;14:327.
33 de Francisco AL, Pinera C, Heras M, Rodrigo E, Fernandez G, Ruiz JC, Tetta C, Arias M:
Hemodiafi ltration with on-line endogenous reinfusion. Blood Purif 2000;18:231–236.
34 European Best Practice Guidelines for Haemodialysis. Part 1. IV. Dialysis fl uid purity. Nephrol
Dial Transplant 2002;1(suppl 7):45–62.
35 Ward RA, Luehmann DA, Klein E: Are current standards for the microbiological purity of hemo-
dialysate adequate? Semin Dial 1989;2:69–72.
36 Pass T, Wright R, Sharp B, Harding GB: Culture of dialysis fl uids on nutrient-rich media for short
periods at elevated temperatures underestimate microbial contamination. Blood Purif 1996;14:
136–145.
37 Golper TA: What technological advances will signifi cantly alter the future care of dialysis patients?
Semin Dial 1994;7:323–324.
38 Canaud B, Kerr P, Argilés A, Flavier JL, Stec F, Mion C: Is hemodiafi ltration the dialysis modality
of choice for the next decade? Kidney Int 1993;43(suppl 41):S296–S299.
39 Henderson LW: Dialysis in the 21st century. Am J Kidney Dis 1996;28:951–957.
Prof. B. Canaud
Nephrology, Dialysis and Intensive Care Unit, Lapeyronie University Hospital
371, avenue du Doyen-Gaston-Giraud
FR–34295 Montpellier (France)
Tel. ⫹33 4 67 33 84 95, Fax ⫹33 4 67 60 37 83, E-Mail b-canaud@chu-montpellier.fr
Ronco C, Canaud B, Aljama P (eds): Hemodiafi ltration.
Contrib Nephrol. Basel, Karger, 2007, vol 158, pp 123–130
Mixed-Dilution Hemodiafiltration
Luciano A. Pedrini, Simona Zerbi
Nephrology and Dialysis Department, Bolognini Hospital of Seriate, Seriate, Italy
Abstract
Mixed-dilution hemodiafi ltration (mixed HDF) controlled by the transmembrane pres-
sure (TMP) feedback, improves the depurative capacity of the more traditional HDF tech-
niques by fully exploiting the convective mechanism of small- and middle-molecular-weight
solute removal. The feedback allows the TMP to be set and profi led from patient and opera-
tional parameters recorded online by the machine. It automatically adjusts the infusion ratio
between predilution and postdilution at the maximum fi ltration fraction without reducing the
total infusion/ultrafi ltration rate and taking into account fl ow conditions, internal pressures
and hydraulic permeability of the dialyzer, and their complex interactions and changes during
the session. The application of the TMP profi le, while avoiding dangerous hydrostatic pres-
sures within the dialyzer and their negative effects, helps better preserve the permeability of
the membrane with the effect of a signifi cantly increased solute removal in a wide molecular
range and with minimal protein leakage. In the light of the more recent observations in the
literature, the high biocompatibility resulting from the use of synthetic membranes and ultra-
pure dialysate, combined with the enhanced removal of small- and middle-molecular-weight
uremic toxins obtained with high-effi ciency HDF, seems to be the best available strategy to
prevent or delay the occurrence of long-term dialysis complications and to promote improved
survival of dialysis patients. Preliminary results of its application indicate that TMP-modu-
lated mixed-dilution HDF could be one of the most powerful strategies to achieve this goal.
Copyright © 2007 S. Karger AG, Basel
Clinical Benefits and Outcome in Hemodiafiltration
With respect to standard and high-fl ux hemodialysis (HD), hemodiafi ltra-
tion (HDF) may induce a sustained increase in removal of both small- and mid-
dle-molecular-weight solutes [1–7], several of which are risk factors or markers
of severe uremic complications and causes of death in HD patients, such as
infl ammation, amyloidosis, secondary hyperparathyroidism and accelerated
atherosclerosis. Sustained improvement in the uremic toxicity profi le may be
Pedrini/Zerbi 124
induced by reducing their level. Lower plasma 2-microglobulin (2-MG) lev-
els were shown to reduce the incidence of bone amyloidosis and carpal tunnel
syndrome [8–10] and have recently been associated with reduced mortality in
dialysis patients [11]. Controlled studies reported a signifi cant improvement
in hemoglobin level and a reduced need for erythropoietin administration in
patients on online HDF [2, 12]. Improved control of secondary hyperparathy-
roidism may be attained in HDF, shown to ensure lower basal levels of phos-
phate [13, 14] in the long term, which were associated with improved survival
in 2 US Renal Data System studies [15].
The HEMO study [16] did not confi rm earlier observations of reduced
morbidity and mortality in HD patients in association with the use of high-fl ux
membranes [9, 17, 18]. Actually, the difference in 2-MG clearance between
the two groups compared in the HEMO study (low- and high-fl ux HD, 4 ⫾ 7
vs. 34 ⫾ 11 ml/min) could have been too low to bring to light a signifi cant dif-
ference in the overall outcome between groups. Indeed, one of the noteworthy
fi ndings of this study was the association between 2-MG basal levels and death
risk in dialysis patients [11]. In HDF defi nitely higher middle-molecule removal
than that in high-fl ux HD may be attained, provided that HDF is performed with
high volume exchange. In randomized studies comparing the two strategies, a
signifi cant difference in basal 2-MG levels only emerged when HDF was per-
formed with a mean fi ltration volume of 21 l/session [7], but not with a relatively
low volume of 8–12 l/session [19]. In the recently published European Dialysis
Outcomes and Practice Patterns Study [20], a signifi cantly lower death risk was
only observed in patients on HDF when high volume exchange of more than
15 l/session was applied, while no difference was observed in terms of mortality
between low-effi ciency HDF and conventional or high-fl ux HD.
Hemodiafiltration Technique and Infusion Modality
HDF is the strategy enabling the high potential of hydraulic and solute
permeability of synthetic membranes to be most properly exploited. To opti-
mize its clinical application and achieve the most effi cient convective transport,
the ultrafi ltration fl ow rate (QUF) must be forced [3, 20] while maintaining a
safe pressure regimen and fl ow within the dialyzer. The infusion mode highly
infl uences HDF performance [3, 21]. For postdilution, the most effi cient infu-
sion mode, there is a limit placed on the QUF by hemoconcentration, high blood
viscosity and resistance to fl ow, which may result in capillary and dialyzer clot-
ting [22]. Thickness of the secondary protein layer, proportional to the fi ltra-
tion pressure, results in permanent and signifi cant reduction of the membrane
permeability, which may compromise the effi ciency of the sessions and require
Mixed Hemodiafi ltration 125
the application of increasingly higher and often unpredictable transmembrane
pressure (TMP) gradients to maintain the planned ultrafi ltration [23]. Control
systems implemented on currently available HDF machines are of little help in
counteracting the events described above and are unsuited to plan and carry out
a session in which ultrafi ltration fl ow or pressure is profi led in order to maintain
safe operational conditions [24]. Predilution HDF ensures better rheological
and hydraulic conditions than postdilution, and the possibility of higher infu-
sion rates, but at the price of reduced effi ciency due to dilution of the solute’s
concentration available for diffusion and convection.
In search of new and more effi cient dialysis strategies, mixed-dilution
HDF with a feedback device for QUF control through TMP has recently been
proposed [23, 25, 26] with the aim of improving the effi ciency and safety of
the HDF technique, while at the same time reducing the shortcomings and risks
associated with the traditional HDF infusion modes.
Optimization of Convection
Some observations form the basis of the new technique. At a given blood
fl ow, the maximal effi ciency in convective solute removal occurs at the high-
est achievable fi ltration fraction (FF) [25]. This is often unpredictable, due to
the events described above and to a patient variability, presumably related to
the individual refi lling capacity as ultrafi ltration progresses. At any given blood
fl ow, TMP is exponentially related to FF, and the slope of the curve is a function
of the hydraulic permeability of the dialyzer [25]. Above a certain TMP level,
the system becomes unstable, and sudden dangerous pressure peaks are likely
to result from small changes in blood fl ow or viscosity, venous pressure or for
technical reasons [22]. These events are diffi cult to prevent or counteract with-
out a feedback system which is able to automatically ensure a constant ultrafi l-
tration and the highest possible FF while maintaining TMP within a safe range.
Principles and Configuration of Mixed-Dilution HDF with TMP
Feedback Control
Online mixed HDF was originally performed in our center on a 4008 H
online Fresenius system (Fresenius Medical Care, Germany) modifi ed with
the application of a Y-shaped infusion line and an additional pump on one
Y branch, which diverted part of the total infusion from the postfi lter to the
prefi lter infusion site. A feedback system for TMP control was used in mixed
HDF to modulate the predilution/postdilution ratio while maintaining the total
Pedrini/Zerbi 126
infusion constant throughout the session (fi g. 1). The basic concept is that
splitting the infusion between the pre- and postfi lter in order to optimize FF
guarantees the best possible rheological and hydraulic conditions within the
dialyzer at the highest fl uid exchange rate and with the most solute removal by
convection.
The mean pressure gradient between blood and dialysate compartments
along the dialyzer (TMP, mm Hg) is calculated online by means of dedicated
software analyzing signals from pressure transducers placed at the inlet and
outlet blood and dialysate ports of the dialyzer (PB in, PB out, PD in, PD out, respec-
tively) using the equation:
TMP ⫽ 0.5 ⭈ [(PB in ⫹ P B out) ⫺ (PD in ⫹ P D out)] ⫺ Ponc (1)
where Ponc (mm Hg) is the mean oncotic pressure exerted by the plasma pro-
teins, set by default to a constant value of 25 mm Hg.
In its last confi guration, implemented on the new 5008 Fresenius system,
the feedback device operates with internal pressure transducers to monitor and
compute TMP. Replacement fl uid is driven to the infusion sites of the dialyzer
QB in
QD out
QD in
Infusion
Predilution
Postdilution
T
T
T
T
T
Dialysate
QB out
P
UF
UF
P
Fig. 1. Online mixed-dilution HDF: schematic representation. Sterile infusion fl uid,
prepared online with double ultrafi ltration (UF), is driven to the infusion ports of the dialyzer
by means of 2 peristaltic pumps (P) at relative infusion rates modulated by the TMP feedback
through changes of the pumps’ speed. TMP, calculated online by means of dedicated soft-
ware analyzing signals from 4 pressure transducers (T), is forced to follow a defi nite profi le
during the session by modulating the ratio between predilution and postdilution in order to
optimize the FF.
Mixed Hemodiafi ltration 127
by means of two separate pumps at a speed modulated by the internal software
on the basis of the actual TMP value [5].
Setting the Infusion Rate
The system device operates at the start of the mixed HDF session by set-
ting the total infusion rate (QS, ml/min) proportionally to the plasma water fl ow
rate of the patient (QPW in, ml/min). The total infusion is then split into pre-
and postinfusion (QS pre , QS post, ml/min) at relative infusion rates allowing the
desired FF to be obtained.
QPW in is calculated online from the effective blood fl ow (QB eff ⫽ Q B compen-
sated by means of the arterial pressure), from hematocrit (Hct), monitored online
with an integrated device (blood volume monitor, Fresenius Medical Care), and
from the water fraction of plasma (Fp), according to the classic equation [27]:
QPW in ⫽ Q B eff ⭈ (1 ⫺ Hct/100) ⭈ Fp. (2)
FF is defi ned arbitrarily as the fraction of QPW in ultrafi ltered during the passage
through the dialyzer, in analogy with a more strict defi nition [28], as:
FF ⫽ (1 ⫺ Q PW out/QPW in) ⫽ Q UF/QPW in (3)
where QPW out is the outlet plasma water fl ow rate.
Initial QS is usually set equal to QPW in but , according to the clinical needs
and the characteristics of the dialyzer, different values for the QS/QPW in ratio
may be planned from 0.7 to 1 in steps of 0.05. Different options are also avail-
able for the initial FF from 0.3 to 0.5.
On the basis of the planned QS and FF, the relative pre- and postinfusion
rates are computed by the software device according to the following equations,
derived from equation 3:
QS pre ⫽ (QUF ⫺ FF ⭈ Q PW in)/FF (4)
QS post ⫽ Q S ⫺ Q S pre. (5)
Controlling the Infusion Rate
After the start of the session, the TMP feedback control acts by modulating
the ratio between pre- and postdilution in order to gradually achieve and then
maintain an optimal and safe TMP value for the entire session (250–300 mm
Hg), without affecting the total QS nor the planned QUF. If TMP falls below the
lowest value of the range, a small amount of fl uid (5–10 ml/min) is diverted from
pre- to postinfusion by increasing the postinfusion pump speed, increasing FF
(and thus TMP) as a result. Vice versa, the same amount of fl uid is diverted from
post- to predilution, thus reducing FF, whenever TMP rises beyond its maxi-
mum tolerated value. In short, the feedback is aimed at ensuring throughout the
Pedrini/Zerbi 128
sessions the highest FF compatible with the progressive hemoconcentration
and loss of hydraulic membrane permeability, which occurs as the session
progresses.
Transmembrane Pressure Profiling
High-fl ux membranes, having generally a cutoff up to 20 kDa, may be
responsible for massive protein leakage, mainly when high fi ltration pressure
is applied to the intact membrane in the early phase of the session, and even
large molecules such as albumin may be forced into the intact pores and either
cross them and get lost in the dialysate or be entrapped inside, with the effect of
a permanent and signifi cant reduction of membrane permeability. At a low QUF,
only small peptides and proteins are trapped by the membrane pores and adhere
to their inner surface. Compared to the intact membrane, this narrowing of the
pore size leads to larger plasma molecules, such as albumin, being rejected,
whereas permeability to middle-molecular-weight solutes, such as 2-MG, is
not substantially modifi ed. Based on these observations, in the more recent
application of mixed HDF low fi ltration pressure was applied at the begin-
ning of the session by setting a relatively low FF (0.35–0.40). Then, TMP was
allowed to increase gradually up to its optimal value according to an ultrafi ltra-
tion profi le obtained with the help of the TMP feedback with automatic shifts of
small amounts of the infusion fl uid from the postdilution to the predilution port
of the dialyzer without reducing the total QUF.
Results of Mixed HDF
Due to its experimental characteristic mixed HDF is presently performed
in selected centers and the results of its application refer to a limited number
of patients. However, the results of the randomized studies published to date
show that this infusion mode, when performed at a matched infusion rate,
ensures safe hydraulic and fl ow conditions similar to predilution HDF but may
achieve an effi ciency in removing small solutes which is defi nitely higher than
that obtained in predilution HDF and similar to that reached in postdilution
HDF performed at a maximal ultrafi ltration rate [25]. However, the advantage
of mixed HDF clearly appears also with respect to middle-molecule removal
when higher infusion rates are applied under the control of the TMP feedback,
which allows the TMP to be set and modulated according to a defi ned profi le
[5, 23]. The feedback automatically adjusts the infusion ratio between pre- and
postdilution at the maximum FF without reducing the total infusion and taking
into account fl ow conditions, internal pressures and hydraulic permeability of
the dialyzer, and their complex interactions and changes occurring during the
Mixed Hemodiafi ltration 129
sessions. Under these conditions, the highest membrane potential of convective
transport is optimized, and online mixed HDF may yield a signifi cantly higher
2-MG removal than that obtained in postdilution HDF and in middilution HDF
[29, 30]. Moreover, the application of the TMP profi le, while simultaneously
minimizing protein leakage in the fi rst part of the session and dangerous hydro-
static pressures within the dialyzer, helps better preserve the permeability of
the membrane in the remaining time of the session, favoring the signifi cantly
increased cumulative solute removal typical of this technique.
References
1 Bammens B, Evenepoel P, Verbeke K, Vanrenterghem Y: Removal of the protein-bound solute
p-cresol by convective transport: a randomized crossover study. Am J Kidney Dis 2004;44:278–
285.
2 Lin CL, Huang CC, Chang CT, Wu MS, Hung CC, Chien CC, Yang CW: Clinical improvement by
increased frequency of on-line hemodiafi ltration. Ren Fail 2001;23:193–206.
3 Lornoy W, Becaus I, Billiouw JM, Sierens L, Van Malderen P, D’Haenens P: On-line haemodiafi l-
tration: remarkable removal of 2-microglobulin. Long-term clinical observations. Nephrol Dial
Transplant 2000;15(suppl 1):49–54.
4 Lornoy W, De Meester J, Becaus I, Billiouw JM, Van Malderen PA, Van Pottelberge M: Impact of
convective fl ow on phosphorus removal in maintenance hemodialysis patients. J Ren Nutr 2006;
16:47–53.
5 Pedrini LA, Cozzi G, Faranna P, Mercieri A, Ruggiero P, Zerbi S, Feliciani A, Riva A:
Transmembrane pressure modulation in high-volume mixed hemodiafi ltration to optimize effi -
ciency and minimize protein loss. Kidney Int 2006;69:573–579.
6 Schroder M, Riedel E, Beck W, Deppisch RM, Pommer W: Increased reduction of dimethylargi-
nines and lowered interdialytic blood pressure by the use of biocompatible membranes. Kidney Int
2001;59:19–24.
7 Ward RA, Schmidt B, Hullin J, Hillebrand GF, Samtleben W: A comparison of on-line hemodiafi l-
tration and high-fl ux hemodialysis: a prospective clinical study. J Am Soc Nephrol 2000;11:2344–
2350.
8 Gejyo F, Kawaguchi Y, Hara S, Nakazawa R, Azuma N, Ogawa H, Koda Y, Suzuki M, Kaneda H,
Kishimoto H, Oda M, Ei K, Miyazaki R, Maruyama H, Arakawa M, Hara M: Arresting dialysis-
related amyloidosis: a prospective multicenter controlled trial of direct hemoperfusion with a 2-
microglobulin adsorption column. Artif Organs 2004;28:371–380.
9 Koda Y, Nishi S, Miyazaki S, Haginoshita S, Sakurabayashi T, Suzuki M, Sakai S, Yuasa Y,
Hirasawa Y, Nishi T: Switch from conventional to high-fl ux membrane reduces the risk of carpal
tunnel syndrome and mortality of hemodialysis patients. Kidney Int 1997;52:1096–1101.
10 Kuchle C, Fricke H, Held E, Schiffl H: High-fl ux hemodialysis postpones clinical manifestation of
dialysis-related amyloidosis. Am J Nephrol 1996;16:484–488.
11 Cheung AK, Rocco MV, Yan G, Leypoldt JK, Levin NW, Greene T, Agodoa L, Bailey J, Beck
GJ, Clark W, Levey AS, Ornt DB, Schulman G, Schwab S, Teehan B, Eknoyan G: Serum 2-
microglobulin levels predict mortality in dialysis patients: results of the HEMO study. J Am Soc
Nephrol 2005;17:546–555.
12 Vaslaki L, Major L, Berta K, Karatson A, Misz M, Pethoe F, Ladanyi E, Fodor B, Stein G,
Pischetsrieder M, Zima T, Wojke R, Gauly A, Passlick-Deetjen J: On-line haemodiafi ltration ver-
sus haemodialysis: stable haematocrit with less erythropoietin and improvement of other relevant
blood parameters. Blood Purif. 2006;24:163–173.
13 Ding F, Ahrenholz P, Winkler RE, Ramlow W, Tiess M, Michelsen A, Patow W: Online hemodiafi ltra-
tion versus acetate-free biofi ltration: a prospective crossover study. Artif Organs 2002;26:169–180.
Pedrini/Zerbi 130
14 Minutolo R, Bellizzi V, Cioffi M, Iodice C, Giannattasio P, Andreucci M, Terracciano V, Di Iorio
BR, Conte G, De Nicola L: Postdialytic rebound of serum phosphorus: pathogenetic and clinical
insights. J Am Soc Nephrol 2002;13:1046–1054.
15 Block GA, Hulbert-Shearon TE, Levin NW, Port FK: Association of serum phosphorus and calci-
um ⫻ phosphate product with mortality risk in chronic hemodialysis patients: a national study. Am
J Kidney Dis 1998;31:607–617.
16 Eknoyan G, Beck GJ, Cheung AK, Daugirdas JT, Greene T, Kusek JW, Allon M, Bailey J, Delmez
JA, Depner TA, Dwyer JT, Levey AS, Levin NW, Milford E, Ornt DB, Rocco MV, Schulman G,
Schwab SJ, Teehan BP, Toto R: Effect of dialysis dose and membrane fl ux in maintenance hemodi-
alysis. N Engl J Med 2002;347:2010–2019.
17 Leypoldt JK, Cheung AK, Carroll CE, Stannard DC, Pereira BJ, Agodoa LY, Port FK: Effect of
dialysis membranes and middle molecule removal on chronic hemodialysis patient survival. Am J
Kidney Dis 1999;33:349–355.
18 Port FK, Wolfe RA, Hulbert-Shearon TE, Daugirdas JT, Agodoa LY, Jones C, Orzol SM, Held PJ:
Mortality risk by hemodialyzer reuse practice and dialyzer membrane characteristics: results from
the USRDS dialysis morbidity and mortality study. Am J Kidney Dis 2001;37:276–286.
19 Locatelli F, Mastrangelo F, Redaelli B, Ronco C, Marcelli D, La Greca G, Orlandini G: Effects
of different membranes and dialysis technologies on patient treatment tolerance and nutritional
parameters. The Italian Cooperative Dialysis Study Group. Kidney Int 1996;50:1293–1302.
20 Canaud B, Bragg-Gresham JL, Marshall MR, Desmeules S, Gillespie BW, Depner T, Klassen P,
Port FK: Mortality risk for patients receiving hemodiafi ltration versus hemodialysis: European
results from the DOPPS. Kidney Int 2006;69:2087–2093.
21 Pedrini LA, Mercieri A: Pre- and post-dilution hemodiafi ltration compared. G Ital Nefrol 2004;
21(suppl 30):S12–S16.
22 Henderson LW: Biophysics of ultrafi ltration and hemofi ltration; in Maher JF (ed): Replacement of
Renal Function by Dialysis, ed 3. Dordrecht, Kluwer Academic, 1989, pp 300–326.
23 Pedrini LA, De Cristofaro V: On-line mixed hemodiafi ltration with a feedback for ultrafi ltration
control: effect on middle-molecule removal. Kidney Int 2003;64:1505–1513.
24 Pedrini LA: On-line hemodiafi ltration: technique and effi ciency. J Nephrol 2003;16(suppl 7):
S57–S63.
25 Pedrini LA, De Cristofaro V, Pagliari B, Sama F: Mixed predilution and postdilution online hemo-
diafi ltration compared with the traditional infusion modes. Kidney Int 2000;58:2155–2165.
26 Pedrini LA, De Cristofaro V, Pagliari B, Filippini M, Ruggiero P: Optimization of convection on
hemodiafi ltration by transmembrane pressure monitoring and biofeedback; in Ronco C, La Greca
G (eds): Hemodialysis Technology. Contrib Nephrol. Basel, Karger, 2002, vol 137, pp 254–259.
27 Sargent JA, Gotch FA: Principles and biophysics of dialysis; in Jacobs C, Kjellstrand CM, Koch
KM, Winchester JF (eds): Replacement of Renal Function by Dialysis. Dordrecht, Kluwer
Academic, 1996, pp 34–102.
28 Bosch JP, Geronemus R, Glabman S, Lysaght R, Kahn T, von Albertini B: High fl ux hemofi ltra-
tion. Artif Organs 1978;2:339–342.
29 Krieter DH, Falkenhain S, Chalabi L, Collins G, Lemke HD, Canaud B: Clinical cross-over com-
parison of mid-dilution hemodiafi ltration using a novel dialyzer concept and post-dilution hemo-
diafi ltration. Kidney Int 2005;67:349–356.
30 Feliciani A, Riva MA, Zerbi S, Ruggiero P, Plati AR, Cozzi G, Pedrini LA: New strategies in hae-
modiafi ltration (HDF): prospective comparative analysis between on-line mixed HDF and mid-
dilution HDF. Nephrol Dial Transplant 2007. DOI:10.1093/ndt/gfm023.
Dr. Luciano A. Pedrini
UO Nefrologia e Dialisi, Ospedale Bolognini
Via Paderno 21
IT–24068 Seriate, Bergamo (Italy)
Tel. ⫹39 035 306 3259, Fax ⫹39 035 306 3375, E-Mail nefrologia.seriate@bolognini.bg.it
Ronco C, Canaud B, Aljama P (eds): Hemodiafi ltration.
Contrib Nephrol. Basel, Karger, 2007, vol 158, pp 131–137
Paired Hemodiafiltration
F. Pizzarelli
Nephrology and Dialysis Unit, SM Annunziata Hospital, Florence, Italy
Abstract
The feasibility of obtaining low-cost high-quality online reinjection fl uids was fi rst
explored almost 30 years ago, but regulatory conservatism delayed adoption of the technique
for almost 20 years. Online treatments are now commonplace in Europe. The competitive
advantages of this treatment modality compared to standard convective treatments include
lower costs, better quality assurance, a lower environmental burden and better clinical out-
comes. The very high volumes of re-infusion fl uids peculiar to online treatment allow a better
removal of 2-microglobulin, and there are claims that survival and anemia are better
improved by online treatments than by standard convective treatments. In contrast, the ace-
tate burden and its attendant potential hazards are relevant in patients under online treatment,
given the considerable quantity of dialysis fl uid injected. Acetate-free paired hemodiafi ltra-
tion, a new online technique, may further ameliorate performances and clinical outcomes,
and may actually cut the gordian knot of the safety of online treatments owing to the imple-
mented safeguards.
Copyright © 2007 S. Karger AG, Basel
Online Treatments
The feasibility of obtaining a sterile and apyrogenic reinfusion fl uid using the
cold-fi ltration technique was demonstrated by Henderson and Beans in 1978 [1],
and the fi rst report of a clinical application was published a few years later [2].
Many more studies have verifi ed the feasibility and safety of online tech-
nology, in both the short [3–7] and the long term [8–12].
These techniques are now rapidly replacing traditional methods in many
dialysis centers in Italy and Europe. This spread is mainly due to the facts that
sophisticated techniques have ruled out the possibility that the infused fl uid
could induce cytokine production or infl ammation in the patient [11–13], and
that the legislative issues involved in online treatments have been resolved
[14].
In our center, online treatment has completely replaced standard hemodia-
fi ltration since 1991. In these 15 years, we have performed more than 15,000
treatments, reinfusing almost 500,000 liters of fi ltered dialysate. Thirty-eight
patients have been treated safely for at least 1 year (3 years on average, with a
range of 1–10 years).
Although ours is a monocentric clinical experience, and as such liable to
errors in case selection, online techniques confi rm economic, management and
organizational, environmental and clinical advantages as compared to conven-
tional convective techniques using commercially available fl uids.
The environmental advantages are not negligible. Online dialysis elimi-
nates the production, transport (almost always by road, in increasingly crowded
highways) and disposal of plastic bags. This has particular importance at a time
when the problem of environmental pollution is assuming planetary dimen-
sions. With regard only to our center, if we consider that the commercial bags
contain 5 liters of solution for reinfusion, we have avoided the production of
almost 100,000 plastic containers.
Several comparative studies have indicated that online hemodiafi ltration
provides an improvement in 2-microglobulin removal compared with standard
hemodialysis, and, consequently, less risk of amyloid-related disease and death
[15–18].
In all the studies analyzing survival and comorbidity there is a trend favoring
convection, but the data are highly variable, with the effect being more evident in
the older studies [19]. A more recent analysis of the European DOPPS database
[20] shows a signifi cant reduction in mortality. After adjustment for potential con-
founders, patients on high-effi ciency hemodiafi ltration had a 35% lower mortality
risk than those receiving low-fl ux hemodialysis (relative risk ⫽ 0.65, p ⫽ 0.01).
The HEMO study, the only randomized prospective study addressing the ques-
tion [21], while not demonstrating statistically signifi cant differences in terms of
mortality and morbidity between high- and low-fl ow membranes, did however
attribute a marginal advantage to the former. It must be asked whether a possible
effect, estimated at 9%, while not statistically signifi cant, would not however be
biologically relevant in view of the very hard outcome tested, that is, the survival
of the patient. A post-hoc analysis of the HEMO study also provides another inter-
esting piece of data: the link between mortality and levels of 2-microglobulin;
the mortality-related risk grows by 13% for every increase of 10 mg/dl in predia-
lytic levels of 2-microglobulin, and this correlation is highly signifi cant [22].
With regard to anemia, the results are more heterogeneous, with some
papers against [23, 24] and some in favor of a benefi cial effect of online hemo-
diafi ltration [25–27].
To summarize, treatments based on convective fl uxes, in contrast to stan-
dard hemodialysis, offer confi rmed advantages in terms of less dialysis-related
Pizzarelli 132
amyloidosis and some degree of improvement in patient survival and anemia.
Although online treatments fall squarely within the defi nition of ‘device’ as
set forth in the European standards [12, 14], the main concern is still there: in
online treatments only a post-hoc assessment of sterility and apyrogenicity of
the reinfusion fl uid can be made and nephrologists have concerns about the
impossibility of testing the purity of the fi ltered dialysis fl uid infused during the
course of the treatment.
Paired Hemodiafiltration
Paired hemodiafi ltration (PHF; Bellco®) is a new online method offering a
response to the above-mentioned concerns [28, 29].
In this technique, which uses a dual stage fi lter with high-fl ux membranes
in both chambers, diffusion and convection are performed in the larger (1.9 m2)
chamber. Part of the dialysis fl uid is diverted by the infusion pump into the infu-
sion line and injected into the patient by backfi ltration in the smaller (0.7 m2)
fi lter chamber (fi g. 1).
Interestingly enough, PHF is based on backfi ltration, but is derived from
paired fi ltration dialysis (Bellco), a technique conceived by Ghezzi et al. [30]
to avoid backfi ltration, and its attendant potential microbiological hazards, by
physical separation of convection and diffusion. Technological improvements in
monitors and quality of the treated water changed the scenario. Backfi ltration is
no longer a challenge for nephrologists but instead a means to obtain high-qual-
ity reinjection fl uid.
From this point of view PHF offers some special and unique characteristics.
First, in PHF, the last stage of dialysis fl uid ultra-fi ltration takes place inside
the dialyzer, i.e. exactly where the dialysis fl uid mixes with blood. Lack of
connections or infusion lines beyond the last dialysis fl uid ultrafi ltration point
minimizes any possible source of contamination (fi g. 1).
Second, PHF incorporates an online dialyzer membrane integrity check.
By reversing the direction of rotation of the infusion pump, instead of inject-
ing fl uids into the patient, blood ultrafi ltrate is obtained. A standard blood leak
detector will detect any red blood cells in that ultrafi ltrate, thereby giving an
idea of the integrity of the membrane devoted to the last stage of dialysis fl uid
fi ltration (fi g. 1). We have performed experiments, both in vitro and in vivo,
which tested fi lters in which 1 single fi ber was deliberately broken [28]. The
system demonstrated high sensitivity to even such minimal damage. The online
integrity check is so reliable because it works with leaking blood not diluted
with dialysis fl uid. The membrane integrity test is performed at the beginning of
the treatment and while it is under way, and as often as deemed necessary.
Paired Hemodiafi ltration 133
Pizzarelli 134
Fiber
integrity
check
BLD
P
Dialysate in
Dialysate out⫹UF
Reinfusion
Dialysate
after 2
previous
stages of
ultrafiltration
Bidirectional pump
for reinfusion or
fiber integrity test
Blood in
Blood out
P
Stage for reinfusion or
fiber integrity check
Stage for convection
and diffusion
Fig. 1. Graphical representation of PHF circuit. See text for details. BLD ⫽ Blood leak
detector; UF ⫽ ultrafi ltrate.
As for clinical outcomes, we have demonstrated for PHF the same pecu-
liarities as other online methods: optimum removal of 2-microglobulin and no
cytokine-inducing capability [28, 29].
We recently tested acetate-free PHF [31]. The background for doing so is
that in patients on online convective treatments, given the considerable quantity
of dialysis fl uid reinfused, the small amount of acetate present in the bicar-
bonate dialysis fl uid as a pH-stabilizing factor may allow a signifi cant transfer
of acetate into the patient, which could induce cytokine activation. We found
that in comparison with the pretreatment values, plasma acetate levels were
unchanged during and after acetate-free PHF, while they were 5–6 times higher
in the course of PHF containing acetate in the dialysis fl uid; plasma acetate
Paired Hemodiafi ltration 135
levels returned to basal values 2 h after the end of the procedure. The total
increase in bases in the patient attributable to acetate was 36%. IL-6 plasma
levels were superimposable at the beginning and in the course of the 2 methods
compared, but there was a tendency towards a greater increase at an interval of
2 h following PHF with acetate.
Conclusions
Online treatments make up an emerging form of dialytic replacement
therapy, as is demonstrated by their continually increasing use and the interest
shown in the method by the major producers of electromedical devices in the
fi eld. We believe that there are clinical, economic and environmental reasons
for preferring online treatments to conventional techniques. The analysis of our
data and of the literature allows us to conclude that online treatments are just
as safe as standard methods in terms of sterility and apyrogenicity of the fl uid
infused. These methods are in line with applicable standards, provided that the
producer clearly defi nes the limits of safe use of the methods and that the users,
the physicians and nurses, ensure that such limits are effectively respected in
routine clinical practice.
Among the many different online treatments available today, PHF offers some
special and unique characteristics. For the same performance and sterility, PHF is
the only online method currently available that makes it possible to verify, in the
course of the treatment, the integrity of the ultrafi lter membrane in which the
fi nal fi ltration of the dialysis fl uid occurs. Acetate-free PHF reduces both acetate
burden and cytokine activation. Clinical advantages due to these effects should be
evaluated in properly designed prospective studies. Should our data be corrobo-
rated in a broader population observed over an appropriate period of time, then
acetate-free PHF may become the top standard of convective online treatments.
References
1 Henderson LW, Beans E: Successful production of sterile pyrogen-free electrolyte solution by
ultrafi ltration. Kidney Int 1978;14:522–525.
2 Ramperez P, Beau MC, Deschodt G, et al: Economic preparation of sterile pyrogen free infusate
for haemofi ltration. Proc EDTA 1981;18:293–296.
3 Luehmann D, Hirsch D, Collins A, Shapiro F, Keshaviah P: Central on-site preparation of substitu-
tion fl uid for hemofi ltration. Trans ASAIO 1984;30:195–198.
4 Canaud B, Nguyen QV, Lagarde C, Stec F, Polaschegg HD, Mion C: Clinical evaluation of a
multipurpose dialysis system adequate for hemodialysis or for postdilution hemofi ltration/hemo-
diafi ltration with on-line preparation of substitution fl uid from dialysate; in Streicher E, Seyffart G
(eds): Highly Permeable Membranes. Contrib Nephrol. Basel, Karger, 1985, vol 46, pp 184–186.
Pizzarelli 136
5 Rindi P, Pilone N, Ricci V, Cioni L: Clinical experience with a new hemodiafi ltration (HDF) sys-
tem. Trans ASAIO 1988;34:765–768.
6 David S, Caserta C, Cambi V: Preparazione estemporanea di soluzioni sterili ed apirogene di sosti-
tuzione: Liquido di dialisi liquidi di sostituzione. Milano, Wichtig, 1989, pp 59–64.
7 Pizzarelli F, Cerrai T, Dattolo P, Neri V, Maggiore Q: L’emodiafi ltrazione con preparazione on-line
del reinfusato è fattibile e sicura; in La Greca G, Petrella E, Cioni A (eds): I liquidi nella dialisi.
Milano, Ghedini, 1992, pp 131–137.
8 Sternby J: A decade of experience with on-line hemofi ltration/hemodiafi ltration; in Maeda K,
Shinzato T (eds): Effective Hemodiafi ltration: New Methods. Contrib Nephrol. Basel, Karger,
1994, vol 108, pp 1–11.
9 Canaud B, Flavier JL, Argils A, Stec F, N’Guyen QV, Bouloux C, Garred LJ, Mion C:
Hemodiafi ltration with on-line production of substitution fl uid: long-term safety and quantitative
assessment of effi cacy; in Maeda K, Shinzato T (eds): Effective Hemodiafi ltration: New Methods.
Contrib Nephrol. Basel, Karger, 1994, vol 108, pp 12–22.
10 von Albertini B, Mishkin G, Lew SQ, Velasquez M, Bosch JP: Long-term survival outcome with
high effi ciency dialysis: effects of time, membrane and reuse. J Am Soc Nephrol 1996;7:1468.
11 Pizzarelli F, Cerrai T, Dattolo P, Tetta C, Maggiore Q: Convective treatments with on-line produc-
tion of replacement fl uid: a clinical experience lasting 6 years. Nephrol Dial Transplant 1998;13:
363–369.
12 Pizzarelli F, Maggiore Q: Clinical perspectives of on-line hemodiafi ltration. Nephrol Dial
Transplant 1998;13(suppl 5):34–37.
13 Canaud B, Wizemann V, Pizzarelli F, Greenwood R, Schultze G, Weber C, Falkenhagen D: Cellular
interleukin-1 receptor antagonist in patients receiving on-line haemodiafi ltration therapy. Nephrol
Dial Transplant 2001;16:2181–2187.
14 Pirovano D: Regulatory issues for on-line haemodiafi ltration. Nephrol Dial Transplant 1998;
13(suppl 5):21–23.
15 van Ypersele de Strihou C, Jadoul M, Malghem J, Maldague B, Jamart J: Effect of dialysis mem-
brane and patient’s age on signs of dialysis-related amyloidosis. The Working Party on Dialysis
Amyloidosis. Kidney Int 1991;39:1012–1019.
16 Hakim RM, Wingard RL, Husni L, Parker RA, Parker TF: The effect of membrane biocompatibil-
ity on plasma 2-microglobulin levels in chronic hemodialysis patients. J Am Soc Nephrol 1996;
7:472–478.
17 Locatelli F, Marcelli D, Conte F, Limido A, Malberti F, Spotti D: Comparison of mortality in
ESRD patients on convective and diffusive extracorporeal treatments. The Registro Lombardo
Dialisi e Trapianto. Kidney Int 1999;55:286–293.
18 Lornoy W, Becaus I, Billiouw JM, Sierens L, van Malderen P: Remarkable removal of 2-micro-
globulin by on-line hemodiafi ltration. Am J Nephrol 1998;18:105–108.
19 Locatelli F, Manzoni C, Di Filippo S: The importance of convective transport. Kidney Int Suppl
2002;80:115–120.
20 Canaud B, Bragg-Gresham JL, Marshall MR, et al: Mortality risk for patients receiving hemodiafi l-
tration versus hemodialysis: European results from the DOPPS. Kidney Int 2006;69:2087–2093.
21 Eknoyan G, Beck GJ, Cheung AK, Hemo Study Group: Effect of dialysis dose and membrane fl ux
in maintenance hemodialysis. N Engl J Med 2002;347:2010–2019.
22 Cheung AK, Levin N, Greene T, et al: Effects of high fl ux hemodialysis on clinical outcomes:
results of the HEMO study. J Am Soc Nephrol 2003;14:3251–3263.
23 Ward RA, et al: A comparison of on-line hemodiafi ltration and high-fl ux hemodialysis: a prospec-
tive clinical study. J Am Soc Nephrol 2000;11:2344–2350.
24 Wizemann V, et al: On-line haemodiafi ltration versus low-fl ux haemodialysis: a prospective ran-
domized study. Nephrol Dial Transplant 2000;15(suppl 1):43–48.
25 Maduell F, et al: Change from conventional haemodiafi ltration to on-line haemodiafi ltration.
Nephrol Dial Transplant 1999;14:1202–1207.
26 Bonforte G, et al: Improvement of anemia in hemodialysis patients treated by hemodiafi ltration
with high-volume on-line prepared substitution fl uid. Blood Purif 2002;20:357–363.
27 Vaslaki L, et al: On-line haemodiafi ltration versus haemodialysis: stable haematocrit with less eryth-
ropoietin and improvement of other relevant blood parameters. Blood Purif 2006;24:163–173.
Paired Hemodiafi ltration 137
28 Pizzarelli F, Tetta C, Cerrai T, Maggiore Q: Double chamber online hemodiafi ltration: a novel
technique with intra-treatment monitoring of dialysate ultrafi lter integrity. Blood Purif 2000;18:
237–241.
29 Pizzarelli F, Cerrai T, Tetta C: Paired hemodiafi ltration: technical assessment and preliminary
clinical results; in Locatelli F, Ronco C, Tetta C (eds): Polyethersulfone: Membranes for Multiple
Clinical Applications. Contrib Nephrol. Basel, Karger, 2003, vol 138, pp 99–105.
30 Ghezzi PM, Frigato G, Fantini GF: Theoretical model and fi rst clinical results of the paired fi ltra-
tion dialysis (PFD). Life Support System 1983;1(suppl 1):271–274.
31 Pizzarelli F, Cerrai T, Dattolo P, Ferro G: On-line haemodiafi ltration with and without acetate.
Nephrol Dial Transplant 2006;21:1648–1651.
Francesco Pizzarelli
Direttore UO Nefrologia e Dialisi, Ospedale SM Annunziata
Via dell’Antella 58
IT–50011 Antella, Firenze (Italy)
Tel./Fax ⫹39 055 2496 223, E-Mail fpizzarelli@yahoo.com
Ronco C, Canaud B, Aljama P (eds): Hemodiafi ltration.
Contrib Nephrol. Basel, Karger, 2007, vol 158, pp 138–152
Acetate-Free Biofiltration
Antonio Santoroa, Francesco Guarnierib, Emiliana Ferramoscaa,
Fabio Grandib
aDepartment of Nephrology, Dialysis and Hypertension, Malpighi Hospital, and
bHospal SpA, Bologna, Italy
Abstract
Acetate-free biofi ltration (AFB) is a hemodiafi ltration technique that, technically as
well as biologically speaking, has all the premises for being a perfectly biocompatible tech-
nique capable of satisfying even the demands of critical patients laden with comorbidities.
Important clinical benefi ts to patients have been reported, such as a better correction of acid-
base balance, an improved nutritional status and a better hemodynamic stability. In particu-
lar, as far as the cardiovascular instability is concerned, several studies have shown that the
rationale behind a better hemodynamic stability is the overall absence of acetate usually pres-
ent in the dialysis bath, which often leads to an impaired vascular tone and a reduced cardiac
contractility. One of the powerful features of AFB is its adaptability to new devices and tools
which can be easily and safely used. In AFB, potassium modulation in the dialysate is easily
achieved. Thus, patients with elevated levels of predialysis potassium and a tendency to
develop both intra- and interdialysis arrhythmias benefi t most. Lastly, the possibility to asso-
ciate AFB with devices like Hemocontrol (which allows for a feedback conditioning of blood
volume) broadens its practical scope, not only for use with hypotension-prone patients, but
also with hypertensive patients with massive increases in their interdialysis body weight. In
this category of patients, avoiding the risk of dangerous hypovolemias allows for the achieve-
ment of dry body weight, thereby facilitating the control of arterial blood pressure and mini-
mizing the clinical consequences of a chronic fl uid overload.
Copyright © 2007 S. Karger AG, Basel
Over the past few years, both the progressive increase in the mean age of
patients on chronic dialysis and greater comorbidity, particularly cardiovascular
pathologies and diabetes, have signifi cantly increased the critical clinical status
of patients undergoing extracorporeal dialysis.
The major repercussions of this clinical complexity can be observed in
the symptoms of the hemodialysis (HD) sessions that have seen the return of
Acetate-Free Biofi ltration 139
hemodynamically unstable events and a greater incidence of intradialytic car-
diac arrhythmias.
A technological response has followed chiefl y by the development of new
dialysis techniques and so-called online intradialytic monitoring that seeks to
prevent critical situations and to continuously measure various physiological
parameters of the patient, both of a hemodynamic and a biochemical kind.
Acetate-free biofi ltration (AFB) is a hemodiafi ltration technique that,
technically as well as biologically speaking, has all the premises for being a
perfectly biocompatible technique capable of satisfying the demands even of
critical patients laden with comorbidities.
The History of Acetate-Free Biofiltration
The idea of an acetate-free dialysis technique, with no buffer at all, fi rst
originated about 25 years ago. The inventors of dialysis with intravenous bicar-
bonate infusion published their feasibility study in the USA in 1980 [1]. Just 2
years later, in Europe, Zucchelli et al. [2] developed biofi ltration, a technical
and conceptual modifi cation of acetate dialysis, with the aim of decreasing the
frequency and severity of some side effects related to acetate, such as intra-
dialytic metabolic acidosis and hemodynamic intolerance to acetate. In 1984,
simultaneously in Italy and France, AFB was launched [3].
Currently, almost all dialysis techniques contain some acetate in the dialy-
sis fl uid in order to keep it chemically stable. Acetate mainly has a chemical
role, allowing for the improvement of the dialysis fl uid’s electrolytic stability.
However, together with bicarbonate, acetate is also a source of buffers that
restore the patient’s acid-base balance. The proportion of acetate transferred
by this mechanism may theoretically reach as much as one third or more of
the total buffers transferred by diffusion to the patient [4]. This result has been
confi rmed by a clinical observation by Agliata et al. [5]. Acetate anions from
the dialysis fl uid in HD cross the dialysis membrane and the wall of the cell by
diffusion. According to the Krebs cycle, acetate may lead to the generation of
bicarbonate, increasing this ion’s intracellular concentration. Bicarbonate anions
from the dialysis bath in HD cross the dialysis membrane and restore the extra-
cellular buffer level, until a concentration gradient is maintained. Meanwhile,
bicarbonate from the intracellular space moves to the extracellular space due to
the concentration gradient. Despite the small proportion of acetate in bicarbon-
ate dialysis (BD), the level of plasma acetate may rise, as it has been shown by
Higuchi et al. [6] in a study comparing AFB and BD. In that study, each patient
was his/her own control and the pre and post dialysis plasma acetate levels were
Santoro/Guarnieri/Ferramosca/Grandi 140
0.24 ⫾ 0.08 and 0.28 ⫾ 0.06 mEq/l in AFB, respectively, and 0.3 ⫾ 0.03 mEq/l
and 0.42 ⫾ 0.12 mEq/l in BD (p ⬍ 0.05).
Indeed, acetate is directly and indirectly involved in generating a number
of side effects. Among these are hypoxia, vasodilatation and the increased pro-
duction of infl ammatory mediators, such as cytokines. All factors increase the
risk of cardiovascular instability.
Technical Aspects of Acetate-Free Biofiltration
AFB is a diffusive-convective dialysis technique characterized by a com-
pletely buffer-free dialysis fl uid, a highly biocompatible hemodialyzer, i.e.
AN69ST, and a sodium bicarbonate substitution fl uid infused to the patient in
postdilution mode (fi g. 1).
Unlike other diffusive-convective dialysis therapies, the absence of a buffer
in the dialysis bath makes the correction of the acid-base balance very simple and
controllable. The acid-base balance correction in AFB is achieved by the intra-
venous infusion of sodium bicarbonate. The overall bicarbonate mass balance is
straightforward, because the process of bicarbonate removal from the fi lter and
the bicarbonate restoration from the substitution fl uid are completely separate so
that the steady-state plasma bicarbonate can be computed as the ratio between
the infusion fl ow rate and the bicarbonate clearance multiplied by the bicarbon-
ate concentration of the substitution fl uid. In practice, steady-state plasma bicar-
bonatemia can be preset at the beginning of the dialysis session and controlled
during the treatment simply by the infusion and the blood fl ow rates. The dialysis
monitor can be equipped with a surveillance system to supervise the bicarbonate-
mia recovery process during dialysis, thereby preventing accidental errors, which
could potentially lead to patient acidosis or alkalosis states by the treatment end.
Buffer-free
dialysate
NaHCO3
Patient
Fig. 1. The AFB technique setting.
Acetate-Free Biofi ltration 141
In order to better personalize the correction of metabolic acidosis in AFB,
the importance of the patient buffer system should be considered during the
intravenous administration of bicarbonate. Some authors have studied this sys-
tem and developed a mathematical model based on the apparent bicarbonate dis-
tribution space (ABS), which is the volume of distribution of the bicarbonate.
This model explains the relationship between blood pH and acid-base status in
chronic dialysis patients. In particular, Fernandez et al. [7] have reckoned that
the lower the bicarbonate serum concentration, the larger the ABS (and that the
higher the bicarbonate serum level, the smaller the ABS). This model allows for
a quantifi cation of the amount of bicarbonate needed to modify the pH and the
bicarbonate serum level to a given extent just by knowing the initial bicarbona-
temia [8]. This fact then suggests that knowing the predialysis bicarbonate levels
is essential when administering bicarbonate to the patient, so as to quantify the
ABS and calculate the infusion fl ow rate of bicarbonate suitable for the patient.
The absence of acetate in the AFB bath is offset by adding an extra amount of
chloride to make the bath electrochemically stable. Special attention ought to be
paid to the role of chloride anion (Cl⫺) concentration on conductivity, since chlo-
ride contributes to conductivity as much as the sodium cation. Dialysis fl uid con-
ductivity is indeed related to all cations and anions dissolved in the dialysis fl uid,
and not to sodium alone. A more elevated chloride concentration in the dialysis
fl uid, as well as for the AFB, affects the conductivity value, thereby triggering its
increase. Furthermore, as chloride ions contribute to conductivity more than bicar-
bonate and acetate anions, the conductivity of the dialysis fl uid in AFB is higher
than that in BD. In order to achieve the same concentration of sodium in both BD
and AFB, a higher value of dialysis fl uid conductivity must be set in AFB.
AN69ST Dialysis Membrane for Acetate-Free Biofiltration
AFB, thanks to the highly biocompatible AN69ST dialysis membrane, is
different from the other diffusive-convective therapies. The AN69ST membrane
is superior in terms of absorption clearances, particularly of high-molecular-
weight molecules, which are responsible for severe pathologies. The recent
innovative treatment (ST stands for ‘surface-treated’) applied to the membrane
surface by using macromolecules, such as polyethylene imines, allowed for the
creation of a more neutral surface, characterized by areas with a high density
of electrically positive charges that can absorb heparin molecules in a stable
way [9]. This allows for dialysis treatments with a reduced heparin regimen, or
with a total absence of heparin as a systemic anticoagulant [10, 11]. AN69ST
maintains the same characteristics as the original AN69 membrane in terms of
cytokine absorption.
Santoro/Guarnieri/Ferramosca/Grandi 142
Clinical Results
Since its fi rst clinical application [12, 13], important clinical benefi ts to
patients have been reported, such as a better correction of acid-base balance, an
improved nutritional status and a better hemodynamic stability.
Correction of Metabolic Acidosis and Nutritional Status
Metabolic acidosis leads to several important complications, amongst
which there is the muscular catabolism stimulus accompanied by a reduction
in lean body mass, a progressive loss of calcium and sodium carbonate from
the bone and cardiovascular instability, mainly due to an increase in peripheral
vascular resistances secondary to a sympathetic refl ex. The ease with which
uremic acidosis can be corrected is one of the prerogatives of AFB. Santoro
and numerous colleagues [4–14] have established statistical models that rapidly
allow us to obtain, from the patient’s bedside, the desired target values of post-
dialysis bicarbonate levels. In a study on 81 patients followed for 67 months,
Movilli et al. [15] found that bicarbonatemia seems to be directly correlated
with serum albumin (p ⫽ 0.001) and inversely correlated with the protein cata-
bolic rate (p ⫽ 0.027). In particular, by clustering the patients into two groups
(HCO3⫺ ⬍20 mmol and HCO3⫺ ⬎23 mmol) the level of serum albumin was
around 3.95 and 4.17 g/dl, respectively. The clinical results were very important
as far as the so-called fragile patients are concerned, such as the elderly. In this
group of patients, AFB offers a better correction of metabolic acidosis, allow-
ing us to achieve a better predialysis blood pH level as compared with standard
BD and hemodiafi ltration.
Galli et al. [16] have investigated the role of dialysis therapies on cardiovas-
cular stability and nutritional status by comparing AFB and BD in a 1-year lon-
gitudinal study on 18 patients. As far as the nutritional status is concerned, they
found that the serum albumin level increased from 3.8 to 4.1 g/dl (p ⫽ 0.013),
while postdialysis bicarbonatemia and pH were 30.2 ⫾ 3.6 mmol in AFB and
26.1 ⫾ 3.2 mmol in BD (p ⫽ 0.017) and 7.44 ⫾ 0.05 in AFB and 7.48 ⫾ 0.04 in
BD (p ⫽ 0.038). Also, Chiappini et al. [17] found that AFB seems to be respon-
sible for a better nutritional status with higher prealbumin values compared to
conventional dialysis (118.8 ⫾ 14.3 in AFB vs. 97.8 ⫾ 15.8% of normal value in
BD) and lower levels of interleukin 1 (10.3 ⫾ 7.4 in AFB vs. 15.6 ⫾ 7.0 pg/ml
in BD).
Acetate-Free Biofiltration and Improvement of Hemodynamic Stability
Both AFB and hemodiafi ltration, as compared with standard BD, offer a
better dialysis tolerance in elderly patients, as reported by Movilli et al. [18,
19]. Very important are the clinical results related to dialysis patients with
Acetate-Free Biofi ltration 143
cardiovascular and cerebrovascular disease. In such a group, a lower frequency
of neurological symptoms in AFB as compared with BD has been reported
[20]. The clinical benefi ts obtained in the diabetic patients were considered by
the Italian Society of Nephrology, which set AFB as the gold standard for the
treatment of diabetic patients with marked intolerance to conventional dialysis
[21].
At the beginning of dialysis, hypotension can be seen as being linked to
both nonautonomic and autonomic causes. As said before, one of the causes of
cardiovascular instability is intolerance to the acetate present in the dialysis bath.
By different mechanisms, the acetate can provoke cardiovascular instability. In
particular, increased nitric oxide synthesis provokes a relaxation in the vessel
smooth muscles, causing an increase in the internal diameter of the vessel.
The different potential of various dialysis fl uids in provoking nitric oxide
generation was reported in 1997 by an in vitro study by Amore et al. [22].
Furthermore, the effect of AFB on peripheral vascular disease, in comparison to
standard dialysis, was demonstrated by Bufano et al. [23] in 2000. Nitric oxide
is also involved in the pathogenesis mechanisms of vascular damage. Amore
et al. [24] further assessed the benefi cial effects of AFB in reducing the risk of
vascular damage.
In that study, the infl ammatory status was evaluated by means of the mea-
surement of nitric oxide synthase activity on lymphocytes and monocytes,
endothelial cells and smooth muscle cells coming from the blood of healthy
donors, after incubation following different simulations of dialysis treatment.
All the studied markers showed a higher nitric oxide synthase activity in BD.
Furthermore, lymphomonocytes treated with BD activate the smooth muscle
cells. Bearing in mind that the different treatments had all been performed using
the same dialysis membrane and the same biological quality of the dialysis bath
(endotoxin-free), any clinical difference highlighted by this study can only be
attributed to the presence or the absence of acetate in the standard dialysis and
AFB baths, respectively.
We have analyzed 9 clinical studies on AFB, focusing particularly on
cardiovascular stability, specifi cally on the capacity of AFB to prevent dialy-
sis-related hypotension. Table 1 presents the main qualitative and quantitative
information related to these studies [25–30]. The overall population is made up
of around 200 patients, and the studies’ follow-up is between 4 and 12 months.
The pooled analysis started calculating the ‘odds ratio’ and respective confi -
dence interval at the 95% level in each of the 9 studies. The odds ratio was
calculated by computing the proportion of dialyses complicated by hypotension
from the data reported in each study. The pooled odds ratio was then calculated
by the Peto-Yusuf method (fi g. 2). The ratio means the odds ratio of hypotensive
symptoms in AFB in relation to BD; hence, values below 1 favor to AFB, while
Santoro/Guarnieri/Ferramosca/Grandi 144
values above 1 favor to BD. The probability of intradialysis hypotension in AFB
is about 40% of the probability of dialysis hypotension in BD.
The Evolution of Acetate-Free Biofiltration
Potassium-Profiled and Blood-Volume-Tracking in Acetate-Free
Biofiltration
Potassium-profi led AFB (AFBK) is a recently introduced modifi ed AFB
whose purpose is to prevent sudden variations in plasma potassium that can be
potentially risky for the onset of arrhythmias during dialysis.
Table 1. Main characteristics of the studies used in the systematic analysis of hypotension in AFB versus BD
Author Year Multi-
center
Centers Study
length
months
Experi-
mental
design
Random-
ized
PRV Patients Male/
female
Age
years
Time on
dialysis
months
Ronco
et al. [25]
1988 no 1 6 p-g yes no 6 55
Briganti
et al. [20]
1991 yes 7 8 c-o yes yes 48 29/19 54 62.8
Galli
et al. [16]
1992 no 1 6 c-o no yes 18 7/11 64.5 22
SCS [26] 1992 yes 12 c-o no yes 33 15/18 49
Kuno
et al. [27]
1994 no 1 6 c-o no yes 6 3/3 48 158
Movilli
et al. [19]
1996 no 1 18 c-o yes yes 12 7/5 76 18
Verzetti
et al. [28]
1998 yes 13 12 c-o yes yes 41 24/17 60 25
Schrander-
van der
Meer
et al. [29]
1999 no 1 12 p-g yes no 24 11/9 65 68
Cavalcanti
et al. [30]
2004 no 1 4 c-o yes no 12 4/8 73.8 4
Overall 200
PRV ⫽ Primary response variable; patients ⫽ number of enrolled patients; p-g ⫽ parallel group; c-o ⫽ cross-
over; SCS ⫽ Spanish Cooperative Study.
Acetate-Free Biofi ltration 145
AFBK is based on conventional AFB, but the dialysate potassium decreases
over time from an initial to a fi nal value in a exponential-like pattern (fi g. 3).
This potassium profi le can be established by the use of 2 independent concen-
trate fl uid pumps connected with the dialysis monitor and a double-compart-
ment concentrate bag, containing potassium-rich AFB concentrate in the small
compartment and potassium-free concentrate in the large one. During AFBK
the potassium level is reduced while the other electrolytes (sodium, calcium,
magnesium) remain constant in the course of the treatment. This facilitates the
achievement of the desired potassium mass balance in the patient by a smoother
potassium removal rate since the concentration gradient between the blood and
the dialysis fl uid is kept more constant over the treatment time. Thus, equiva-
lent overall potassium removal has been shown to be obtained by adjusting the
initial and fi nal dialysate potassium levels [31]. A more constant potassium con-
centration gradient also limits the negative effects of extracellular potassium
removal on the electrophysiology of cardiac cells.
The clinical indications for end-stage renal disease patients on dialysis to
undergo AFBK is hyperkalemia, and patients who are highly prone to intradial-
ysis and interdialysis arrhythmias, such as those with diabetes mellitus, hyper-
tension and cardiomyopathy, may also benefi t signifi cantly from this therapy.
The major prevalence of diabetes, anemia, hyperparathyroidism and hyper-
tension among chronic dialysis patients engenders structural heart disease.
Moreover, fl uid overload and metabolic abnormalities such as metabolic aci-
Ronco et al. [25] (1988)
Briganti et al. [20] (1991)
Galli et al. [16] (1992)
SCS [26] (1992)
Kuno et al. [27] (1994)
MoviIIi et al. [19] (1996)
Verzetti et al. [28] (1998)
Schrander et al. [29] (1999)
Cagnoli [30] (2002)
Overall
0 0.1
Favor AFB Favor BD
1.0 10 100
Odds ratio with 95% confidence interval
Fig. 2. Systematic analysis of the peer-reviewed data from the literature on hypoten-
sion in AFB versus BD. The value of the overall odds ratio calculated by Mantel-Haenszel
statistics is 0.4.
Santoro/Guarnieri/Ferramosca/Grandi 146
dosis, dyskalemia or dysmagnesemia lead to an increased risk of clinically sig-
nifi cant ventricular arrhythmias and sudden cardiac death. During the dialysis
procedure, the dialysis patients present a nonhomogeneous repolarization, and
this is confi rmed by the increase in Q–T duration and Q–T dispersion [32]. The
dialysis-related sudden variations in extracellular potassium, calcium and pH
may be corroborating factors in the genesis of an electrical disequilibrium in
myocardial cells. One of the potential therapeutic options is indeed the adjust-
ment of the dialysis fl uid composition. A multicenter crossover clinical study
aimed at investigating the electrical behavior of two different K⫹ removal rates
upon myocardial cells (risk of arrhythmia and ECG alterations) has recently
been performed [31]. Conventional AFB and AFBK were used in a patient sam-
ple to understand the effect on premature ventricular contraction (PVC) and
on repolarization indices. The study was divided into two phases: phase 1 was
a pilot study to evaluate K⫹ kinetics and to test the effect on the electrophysi-
ological response of the two procedures. The second phase was set up as an
extended crossover multicenter trial in a patient subset prone to arrhythmias
during dialysis.
The main result of the phase 1 of the study was that serum potassium
showed a marked decrease in conventional AFB during the fi rst half of dialy-
sis, greater than in AFBK. The greatest difference in serum potassium was
achieved at 60 min after the start of dialysis (4.1 ⫾ 0.4 mEq/l in AFBK vs.
3.8 ⫾ 0.4 mEq/l in AFB) while the fi nal values were equivalent in both treat-
ments. Despite this difference in serum potassium, the fi nal potassium removed
by AFB and AFBK was comparable (88 ⫾ 15 mEq in AFBK vs. 92 ⫾ 19 mEq
in AFB).
Fig. 3. Dialysate potassium profi les in AFBK. The upper and lower limits assure that
the patient cannot be exposed to harmful concentrations of dialysate potassium. By adjusting
the values at the beginning and at the end of the dialysis, the overall potassium removal can
be set prior to the beginning of the dialysis.
0
1
2
3
4
5
6
Time
Potassium (mmol/l)
Acetate-Free Biofi ltration 147
The ANOVA for repeated measures applied to the PVC per hour did not
show any statistical differences (p ⫽ 0.428), but in a patient subgroup (at least
within grade II of Lown grading, i.e. number of PVC per hour ⬎30) these were
consistently lower in AFBK than in AFB (p ⫽ 0.02, Wilcoxon test).
During phase 2 the PVC again increased in both AFB and AFBK, although
less so in the latter halfway through dialysis (fi g. 4). By plotting the PVC per
hour in AFBK against the those in AFB, most of the values lie below the bisec-
tor indicating a marked reduction in arrhythmias in AFBK. By means of this
study, it was possible to show that it is not the K⫹ removal rate alone that may
be destabilizing from an electrophysiological standpoint, but rather its removal
dynamics. This is all the more evident in patients with arrhythmias who benefi t
from the K⫹ profi ling during their dialysis treatment.
A possible explanation for this fi nding has recently been published by
Buemi et al. [33], who investigated the clinical effects of AFBK as compared
with conventional AFB on the repolarization indexes (Q–T interval corrected,
Q–Tc, and Q–Tc dispersion) in 28 patients (fi g. 5). The indexes were assessed in
the midweek dialysis at the times 0, 15, 45, 90, 120 and 240 min during dialysis.
It is worth noting that the time patterns of the parameters were different between
the two therapies with a rapid increase in AFB and a stable behavior in AFBK.
Again, as previously reported by Santoro et al. [31], the discriminant was the
different plasma potassium time pattern. In fact, plasma potassium decreased
more slowly in AFBK than in BD. The highest difference was found at 15, 45
and 90 min after the start of dialysis (p ⬍ 0.01), which could partly explain the
related effect on the ventricular repolarization.
The blood volume tracking (BVT) system is a tool designed to prevent
a severe fall in blood volume during HD that potentially exposes patients to
Fig. 4. Scatter plot of PVC per hour in AFBK against standard AFB. The thin line is
the identity line, the thick one is the regression line [32].
y⫽0.4x⫹25.9
R2⫽0.4
p⫽0.001
0
100
200
300
0 100 200 300
AFB (PVC/h)
AFBK (PVC/h)
Santoro/Guarnieri/Ferramosca/Grandi 148
volume-dependent hypotension. The main goal of BVT is to drive the actual
patient’s blood volume along an ideal blood volume trajectory by a continuous
adjustment of two dialysis parameters: the ultrafi ltration rate and the dialysis
fl uid conductivity. The ideal trajectory is chosen in a way that the blood volume
will decrease less during dialysis with BVT than during a number of test dialy-
ses in which dialysis hypotension developed. BVT has two further goals: the
achievement of total weight loss and the prevention of sodium overload in the
patient. It is clear that the three different goals are mutually contradictory, so the
BVT system continuously searches for the best solution or compromise among
these three goals. To do so effi ciently, the system is given some tolerances, with
adjustable width, as regards each of the three goals.
A number of publications highlight that BVT dialysis leads to a decrease
in both the frequency and the severity of dialysis hypotension, particularly in
the hypotension-prone patient. In the meantime, there have also been some
reports on the benefi ts of BVT in patients experiencing dialysis hypotension
only incidentally.
The above-mentioned studies were all aimed at understanding the rela-
tive role played by BVT alone in preventing dialysis-related hypotension.
Nevertheless, BVT only acts on one factor involved in intradialytic symptoms,
such as volume-dependent hypotension. Then, to further improve dialysis toler-
ance, the idea of combining BVT and AFB seems to be promising, as AFB acts
on other factors, namely vascular resistance and cardiac output.
The fi rst attempt to show an improvement in the treatment tolerance was
made by Ronco et al. [34] who compared AFB and BVT-AFB in 12 patients
Fig. 5. Repolarization index Q–Tc dispersion in AFBK compared to AFB [34].
Signifi cant differences were found at 45, 90 and 120 min (p ⬍ 0.05). However, AFB showed a
fast increase in the fi rst part of dialysis not observed in AFBK, the values of which remained
stable throughout the session.
0
20
40
60
80
100
120
140
0 15 45 90 120 240
Time (min)
Q–TC dispersion (ms)
AFB
AFBK
Acetate-Free Biofi ltration 149
with frequent dialysis hypotensions. That study showed that in the presence
of similar weight loss rates (3,646 ⫾ 684 ml without BVT vs. 3,710 ⫾ 710
with BVT, p ⫽ n.s.), cardiovascular tolerance signifi cantly improved in the
biofeedback-driven sessions (hypotension episodes 59/72 without BVT vs.
24/72 with BVT, p ⬍ 0.001). That study also showed a possible positive effect
of the improved cardiovascular stability on urea kinetics and treatment effi -
ciency (equilibrated Kt/V 1.03 ⫾ 0.01 without BVT vs. 1.12 ⫾ 0.05 with BVT,
p ⬍ 0.001), accounted for as the result of an improved tissue perfusion high-
lighted by a reduced urea rebound (14.2 ⫾ 2.7% without BVT vs. 6.4 ⫾ 2.3%
with BVT, p ⬍ 0.001).
Further evidence emerged from a study by Santoro et al. [31], aimed at
investigating the cardiovascular compensatory response to hypovolemia in a
population of 12 HD patients, by means of a model-based computer analysis
of BVT-AFB against conventional dialysis. A novel result was that the criti-
cal blood volume threshold (i.e. the relative blood volume change at which
acute hypotension appeared) was signifi cantly lower in BD than in BVT-AFB,
that is to say less negative (⫺7.9 ⫾ 2%) in BD than BVT-AFB (⫺10.9 ⫾ 2.6%,
p ⬍ 0.05). As direct consequence, hypotensive events occurred earlier in BD
than in BVT-AFB (collapse time 123 ⫾ 41 min in BD vs. 183 ⫾ 25 min in BVT-
AFB, p ⬍ 0.01).
Differences in the effectiveness of compensation to hypovolemia were evi-
dent in the control of microvascular resistance and the effi cacy of inotropic con-
trol on cardiac contractility. The result of the computer-based analysis is that the
estimated total peripheral resistance increased twice as much in BVT-AFB than
in BD (change in total peripheral resistance 9.1 ⫾ 9.4% in BD vs. 18.9 ⫾ 6.6%
in BVT-AFB, p ⬍ 0.05), whereas the decrease in the estimated stroke volume
was twice as high in BD (change in stroke volume ⫺19.1 ⫾ 8.2% in BD vs.
⫺10.7 ⫾ 7.8% in BVT-AFB, p ⬍ 0.01).
A deeper insight into the same experiment was gained by the analy-
sis of heart rate variability [35]. This revealed that the low-frequency com-
ponent in BD, after an initial increase, slowly became depressed (from
0.56 ⫾ 0.19 nU at the beginning of dialysis to 0.56 ⫾ 0.22 at the end in BD, yet
from 0.59 ⫾ 0.14 nU at the beginning to 0.67 ⫾ 0.12 nU at the end of dialysis in
BVT-AFB, p ⬍ 0.05).
The low-frequency component of heart rate variability refl ects the auto-
nomic activation of the mechanisms controlling the cardiovascular functions.
The lowering of the low-frequency component in BD is therefore consistent
with a worsened effectiveness of vasoconstriction and cardiac contractility even
in the presence of signifi cant adrenergic activation (heart rate increase from
72 ⫾ 9 at the beginning to 75 ⫾ 10 beats/min at the end in BVT-AFB, vs. 73 ⫾ 10
at the beginning to 85 ⫾ 10 beats/min at the end in BD, p ⬍ 0.05).
Santoro/Guarnieri/Ferramosca/Grandi 150
Conclusions
AFB is defi nitely one of most user-friendly hemodiafi ltration techniques.
Indeed, the chance to modify the composition of the dialysis bath and the bicar-
bonate solution to be infused either alternatively or simultaneously gives it
a versatility that is not at all typical of the online techniques. As a matter of
fact, in the online techniques the composition of the dialysis liquid cannot be
separated from that of the infusion liquid. This is a major drawback that often
undermines their adaptability to the patient’s needs. In AFB, the potassium
modulation in the dialysate is easily achieved and is starting to bear its fruits in
clinical terms. Patients with elevated levels of predialysis potassium and a ten-
dency to develop both intra- and interdialysis arrhythmias benefi t most. Lastly,
the chance to associate AFB with devices such as Hemocontrol (which allows
for a feedback conditioning of the blood volume) broadens its practical scope,
not only for hypotension-prone patients, but also hypertensive patients with
massive increases in their interdialysis body weight. In this category of patients,
avoiding the risk of dangerous hypovolemias allows for the achievement of dry
body weight, thereby facilitating the control of arterial blood pressure and mini-
mizing the clinical consequences of a chronic fl uid overload.
References
1 Van Stone JC, Mitchell A: Hemodialysis with base-free dialysate. Proc Dialysis Transplant Forum
1980;10:268–271.
2 Zucchelli P, Santoro A, Raggiotto G, Degli Esposti E, Sturani A, Capecchi V: Biofi ltration in ure-
mia: preliminary observations. Blood Purif 1984;2:187–195.
3 Buoncristiani U, Ragaiolo M, Petrucci V, Bruni F, Pala E, Damiani C, Brugnano R: Biofi ltration
with buffer-free dialysate. Int J Artif Organs 1986;9(suppl 3):9–14.
4 Santoro A, Spongano M, Ferrari G, Bolzani R, Augella F, Borghi M, Briganti M, Cagnoli L, Docci
D, Feletti C, et al: Analysis of the factors infl uencing bicarbonate balance during acetate-free bio-
fi ltration. Kidney Int 1993;41(suppl):S184–S187.
5 Agliata S, Atti M, Fortina F, Airoldi G, Baroni A, Ragazzoni E, Schweiger C, Cavagnino A: Acetate
in the dialysate in bicarbonato dialysis (abstract); in Ward RA, Golper TA (eds): Abstracts of
the Tenth Annual Meeting of the International Society of Blood Purifi cation, Louisville, October
1992. Blood Purif 1992;10(special issue).
6 Higuchi T, Kuno T, Takahashi S, Kanmatsuse K: Chronic effect of long-term acetate-free biofi ltra-
tion in the production of interleukin 1 and interleukin-1 receptor antagonist by peripheral blood
mononuclear cells. Am J Nephrol 1997;17:428–434.
7 Fernandez PC, Cohen RM, Feldman GM: The concept of bicarbonate distribution space: the cru-
cial role of body buffers. Kidney Int 1989;36:747–752.
8 Pacitti A, Casino FG, Pedrini l, Santoro A, Atti M: Prescription and surveillance of the acetate-free
biofi ltration sessions: the bicarbonate cycle. Int J Artif Organs 1995;18:722–725.
9 Renaux JL, Atti M: The AN69ST dialysis membrane; in Ronco C, La Greca G (eds): Hemodialysis
Technology. Contrib Nephrol. Basel, Karger, 2002, vol 137, pp 111–119.
10 Chanard J: Assessment of heparin-binding to the AN69 ST hemodialysis membrane. I. Preclinical
study. ASAIO J 2005;51:342–347.
Acetate-Free Biofi ltration 151
11 Lavaud S: Assessment of the heparin binding AN69 ST hemodialysis membrane. II. Clinical stud-
ies without heparin administration. ASAIO J 2005;51:348–351.
12 Bandiani G, Camaiora E, Nicolini MA, Perotta U: Tolerance and adequacy of acetate-free bio-
fi ltration (AFB): two years clinical experience (abstract). Proc EDTA Congr, Madrid, 1988, p
148.
13 Bandiani G, Camaiora E, Nicolini MA, Perotta U: The acid-base balance during acetate-free bio-
fi ltration (AFB) (abstract). Proc EDTA Congr, Madrid, 1988, p 148.
14 Santoro A, Ferrari G, Spongano M, Badiali F, Zucchelli P: Acetate-free biofi ltration: a viable alter-
native to bicarbonate dialysis. Artif Organs 1989;13:476–479.
15 Movilli E, Bossini N, Viola BF, Camerini C, Cancarini G, Feller P, Strada A, Maiorca R: Evidence
for an independent role of metabolic acidosis on nutritional status in hemodialysis patients.
Nephrol Dial Transplant 1998;13:125–131.
16 Galli G, Bianco F, Panzetta G: Acetate free biofi ltration: an effective treatment for high-risk dialy-
sis patients; in Man NK, Rotella J, Zucchelli P (eds): Blood Purifi cation in Perspective: New
Insight and Future Trend. Cleveland, ICAOT Press, 1992, No 320, vol 2.
17 Chiappini MG, Moscatelli M, Batoli R: Effects of different hemodialysis methods on the nutri-
tional status of HD patients. Ren Fail 1990;12:277–278.
18 Movilli E, Gaggiotti M, Maiorca R: Hemodiafi ltration in hypotension-prone elderly uremic
patients. Geriatr Nephrol Urol 1995;4:183–187.
19 Movilli E, Camerini C, Zein H, D’Avolio G, Sandrini M, Strada A, Maiorca R: A prospective
comparison of bicarbonate dialysis, hemodiafi ltration and acetate-free biofi ltration in the elderly.
Am J Kidney Dis 1996;27:541–547.
20 Briganti M, Borghi M, Cagnoli L, Feletti C, Fusaroli M, Sanna G, Santoro A, Stallone C, Zucchelli
P: Valutazione clinica e neurologica della biofi ltrazione senza acetato: studio multicentrico. G Ital
Nefrol 1991;8:291–297.
21 Fuiano G, Zoccali C: Linee guida per la diagnosi e la terapia della nefropatia diabetica. G Ital
Nefrol 2003;20(suppl):S96–S108.
22 Amore S, Cirina P, Mitola S, Peruzzi L, Bonaudo R, Gianoglio B, Coppo R: acetate intolerance
is mediated by enhanced synthesis of nitric oxide by endothelial cells. J Am Soc Nephrol 1997;8:
1431–1436.
23 Bufano G, Grandi F, Ariano R, Atti M: Plasma levels of nitric oxide (NO) and peripheral vascular
resistances (PVR) during hemodialysis (abstract 1358). J Am Soc Nephrol 2000, Abstr 33rd Annu
Meet ASN, Toronto, p 258A.
24 Amore A, Cirina P, Bonaudo R, Conti G, Chiesa M, Coppo R: Bicarbonate dialysis, unlike ace-
tate-free biofi ltration, triggers mediators of infl ammation and apoptosis in endothelial and smooth
muscle cells. J Nephrol 2006;19:57–64.
25 Ronco C, Fabris A, Chiaromonte S, De Dominicis E, Feriani M, Brendolan A, Brigantini L, Milan
M, Dell’Aquila R, La Greca G: Comparison of four different short dialysis techniques. Int J Artif
Organs 1988;11:169–174.
26 Junco E, Aljama P, Arias M, Bocal J, Rotella J, Caralps A, Castello D, Luno J, Martin de Francisco
AL, Martin Malo A, Montenegro J, Perez R, Sanz C, Saracco J, Teixido J, Valderrabano F, Madero
R: Acetate free biofi ltration: Spanish Cooperative Study; in Man NK, Rotella J, Zucchelli P (eds):
Blood Purifi cation in Perspective: New Insight and Future Trend. Cleveland, ICAOT Press, 1992,
No 320, vol 2.
27 Kuno T, Kikuchi F, Yanai M, Nagura Y, Takahashi S: Clinical advantages of acetate-free biofi ltra-
tion; in Maeda K, Shinzato T (eds): Effective Hemodiafi ltration: New Methods. Contrib Nephrol.
Basel, Karger, 1994, vol 108, pp 123–130.
28 Verzetti G, Navino C, Bolzani R, Galli G, Panzetta G: Acetate-free biofi ltration versus bicarbon-
ate hemodialysis in the treatment of patients with diabetic nephropathy: a cross-over multicentric
study. Nephrol Dial Transplant 1998;13:955–961.
29 Schrander-van der Meer AM, Ter Wee PM, Kan G, Donker AJ, Van Dorp WT: Improved cardio-
vascular variables during acetate-free biofi ltration. Clin Nephrol 1999;51:304–309.
30 Cavalcanti S, Ciandrini A, Severi S, Badiali F, Bini S, Gattiani A, Cagnoli L, Santoro A: Model
based study of the effects of the hemodialysis technique on the compensatory response to hypovo-
lemia. Kidney Int 2004;65:1499–1510.
Santoro/Guarnieri/Ferramosca/Grandi 152
31 Santoro A, Mancini E, Gaggi R, Cavalcanti S, Severi S, Cagnoli L, Badiali F, Perrone B, London
G, Fessy H, Mercadal L, Grandi F: Electrophysiological response to dialysis: the role of dialy-
sate potassium content and profi ling; in Ronco C, Brendolan A, Levin NW (eds): Cardiovascular
Disorders in Hemodialysis. Contrib Nephrol. Basel, Karger, 2005, vol 149, pp 295–305.
32 Cupisti A, Galetta F, Caprioli R, Morelli E, Tintori GC, Franzoni F, lippi A, Meola M, Rindi
P, Barsotti G: Potassium removal increases the QTc interval dispersion during hemodialysis.
Nephron 1999;82:122–126.
33 Buemi M, Aloisi E, Coppolino G, et al: The effect of two different protocols of potassium haemo-
diafi ltration on QT dispersion. Nephrol Dial Transplant 2005;20:1148–1154.
34 Ronco C, Brendolan A, Milan M, Rodeghiero MP, Zanella M, La Greca G: Impact of biofeedback-
induced cardiovascular stability on hemodialysis tolerance and effi ciency. Kidney int 2000;58:
800–808.
35 Severi S, Ciandrini A, Grandi E, Cavalcanti S, Bini S, Badiali F, Gattini A, Cagnoli L: Cardiac
response to hemodialysis with different cardiovascular tolerance: heart rate variability and QT
interval analysis. Hemodial Int 2006;10:287–293.
Antonio Santoro, MD
Department of Nephrology, Dialysis and Hypertension, Malpighi Hospital
Via P. Pelagi, 9
IT–40138 Bologna (Italy)
E-Mail santoro@aosp.bo.it
Ronco C, Canaud B, Aljama P (eds): Hemodiafi ltration.
Contrib Nephrol. Basel, Karger, 2007, vol 158, pp 153–160
Mid-Dilution: An Innovative High-Quality
and Safe Haemodiafiltration Approach
Jacky Potier
Centre Hospitalier Louis-Pasteur, Cherbourg, France
Abstract
Mid-dilution (MidD) is a new concept allowing post- and predilution in the same dia-
lyser (Olpur® MD 190). The aim of the study was to compare, in 6 patients, MidD with post-
and predilution wird regard to purifi cation tests, such as reduction ratios and instantaneous
whole-blood clearances, of urea, creatinine and phosphorus as examples of low-weight mol-
ecules and of 2-microglobulin as an example of a middle molecule (MM). The aim was also
to indicate directions for the use of this new dialyser, taking into account our own experience
and the few observations already published. It was concluded that MidD, under excellent
safety conditions, in spite of increased intradialyser pressures, offers a very high purifi cation
performance, particularly for MM, because of high convective volumes exceeding the rec-
ommended objectives for a better survival in dialysis.
Copyright © 2007 S. Karger AG, Basel
Traditionally, there are 2 methods of haemodiafi ltration (HDF): post-
dilution (PostD) and predilution (PreD), each with counter-indications deter-
mined by their potential risks. PostD, because of haemoconcentration on the
outlet side of the dialyser, increases the risks of coagulation and increase in the
transmembrane pressure (TMP), the more so when the blood fl ow (QB) is low
or the haematocrit (Hct) is high. PreD decreases the purifi cation of the small
molecules because of haemodilution in the dialyser, unless a high substitution
fl ow (QS) is used. The idea of a joint use of the 2 methods is the origin of mixed
dilution [1]. The variation of the 2 substitution fl uid fl ow rates, PostD and PreD,
throughout the session is subjected to an objective of TMP and leads to use
fi rst the purifying capacity of PostD, then at the end of dialysis the protective
advantage of PreD. The joint association of the 2 methods is also at the origin
of mid-dilution (MidD), a new concept based on the Olpur MD 190 dialyser
(Nephros Inc., New York, N.Y., USA), already available and usable on all moni-
tors equipped with an online production of dialysate for HDF.
Potier 154
Din
Dout
Substitution
Bin
Bout
Fig.1. Origin of Olpur MD 190 from 2 dialysers in series, the fi rst (PostD) without its
peripheral fi bres and the second (PreD) without its central fi bres. The 180-degree rotation of
the second on the fi rst results in a single dialyser with 2 blood stages, in which the dialysate
fl ows with the blood in the second PreD stage and against the blood in the PostD stage.
Bin ⫽ Inlet blood; Bout ⫽ outlet blood; Din ⫽ inlet dialysate; Dout ⫽ outlet dialysate.
Concept
The initial principle (fi g. 1) is to use in series 2 dialysers, with a fi rst stage
in PostD and a second in PreD. For practical reasons, a single dialyser was used
thanks to a 180-degree rotation of the PreD component.
The Olpur 190 MD is thus the manufactured product resulting from this new
concept. It is composed of polyethersulphone fi bres (wall thickness ⫽ 35 μm,
inner diameter ⫽ 210 μm) with a total surface of 1.90 m2. In its initial confi gu-
ration, 1.10 m2 were devoted to PostD (peripheral fi bres) and 0.80 m2 to PreD
(central fi bres).
In fact, the fi rst clinical trials in this confi guration highlighted an instabil-
ity of the pressures in the dialyser, manifested by a high TMP and a neces-
sary reduction of the total convection volume, manifestly lower than the target
expected by the system’s designer.
The fi rst modifi cation [2] consisted in inverting the blood connections –
an operation called simple reverse (SR) – the PreD stage becoming PostD and
vice versa. The internal haemodynamic dialyser conditions were thus mani-
festly improved, allowing a Qs ⬎200 ml/min. It is also possible to perform
a reverse dialysis in a double reverse (DR) modality, with identical purifi ca-
tion results, in particular for a better adaptation to the Fresenius 5008 monitor
(unpubl. data).
Principles of Use
In this 2-stage system, the PostD stage is the more effective, but also the
limiting factor in terms of QS, since it is responsible for the haemoconcentration
phenomenon.
Mid-Dilution 155
If it is assumed that the PostD/PreD ratio of the convection is close to
the ratio of their surfaces, then the PostD convection is 0.8/1.9, i.e. 42% of
the total convection. In addition, it is generally recommended in PostD not to
exceed a fi ltration fraction (FF) of 50% [3] of the plasmatic water fl ow (QPW).
If QB ⫽ 350 ml/min with Hct ⫽ 33% and total protides (Pt) ⫽ 70 g/l, then QPW is
estimated as QPW ⫽ Q B ⫻ (QB ⫺ Ht ⫻ 0.01) ⫻ (QB ⫺ Pt ⫻ 0.00017), i.e. 200 ml/
min. QS authorized in PostD thus equals 200 ⫻ 0.5 ⫽ 100 ml/min, and total QS
including PostD and PreD could then be 100 ⫻ (100/42) ⫽ 238 ml/min.
We published a table [4] allowing QS to be determined for each QB and
according to Hct, and were able to check in vivo the relevance of these recom-
mendations. More simply, with QB ⬎320 ml/min and Hct ⬍36%, QS can be
fi xed at 225 ml/min without exceeding a monitor TMP of 250 mm Hg.
The complexity of diffusion and convection in such a dialyser makes their
modelling delicate [5], and for the moment leads to more assumptions than cer-
tainties, thus placing paramount importance on clinical experimentation.
Methods
Six stable patients on renal replacement therapy were submitted to 1 session of MidD in
its DR confi guration, 1 of PreD and 1 of PostD. All sessions were carried out under similar
operating conditions with a Fresenius 5008 dialysis system (Fresenius Medical Care, Bad
Homburg, Germany). The hollow-fi bre dialysers employed were an Olpur MD 190 in MidD
and an FX80 (Fresenius Medical Care) in PostD and PreD. Effective QB as calculated by the
machine was set at 320 ml/min (equivalent to a QB pump of 360 ml/min), dialysate fl ow (QD)
at 500 ml/min and QS at 225 ml/min for MidD, 100 ml/min for PostD and 200 ml/min for
PreD. The same usual patient low-molecular-weight heparin anticoagulation therapy was
applied.
Treatment effi cacy was determined by measuring reduction ratios (RR), equilibrated
Kt/V (EqKt/V) and instantaneous whole-blood clearances (K) after 60 min.
RR ⫽ (1 ⫺ C Post/CPre) ⫻ 100
with CPre and CPost being the plasma concentrations before the start and at the end of each
treatment session. RR was determined for the small solutes urea, creatinine and phosphorus,
and for the middle molecule (MM) 2-microglobulin (2-MG).
K ⫽ Q B ⫻ (CAr t ⫺ C Ve n /CArt) ⫹ QUF ⫻ (CVen /CArt)
with CArt and CVen being the plasma concentrations of the blood samples obtained from the
arterial and venous blood lines, respectively, of the extracorporeal circuit. QUF is the ultrafi l-
tration rate, which was set at 600 ml/min during sampling. K was determined for urea, cre-
atinine and 2-MG. It was impossible to evaluate phosphorus clearance because of the too
low value of the venous sample, under the lowest limit of the laboratory method used.
Comparative statistical analyses were assessed with the t test for paired data
(StatView).
Potier 156
Results
Results are presented in table 1.
Low-Weight Molecules
Generally, for high-fl ux (HF) membranes, the performances of diffusive
purifi cation of low-weight molecules (LM) depend above all on QB and the gra-
dient of concentration between blood and dialysate. The simultaneous convec-
tion improves further the performances in PostD but tends to diminish them
proportionally in PreD because of haemodilution [6].
In addition, with MidD, dialysate is at co-current in PreD and even if it
is with countercurrent in PostD, it is also partially ‘polluted’ by PreD. Thus, it
is diffi cult to predict the resultant between these diffusive conditions, a priori
unfavourable, and the important favourable convective participation up to 50
liters.
For urea, K is intermediate in MidD (271.6 ml/min), between PostD
(288.8 ml/min) and PreD (254.5 ml/min). It is correlated with RR: MidD ⫽ 78.3%,
PostD ⫽ 80.3% and PreD ⫽ 76.8%, and with EqKt/V: MidD ⫽ 1.55%, PostD ⫽
1.66% and PreD ⫽ 1.52%.
For creatinine, we found similar K values, with MidD (215.5 ml/min)
between PostD (232.5 ml/min) and PreD (199.8 ml/min), but also RR with
MidD ⫽ 69.0%, PostD ⫽ 70.3% and PreD ⫽ 66.3%.
Table 1. Comparison of RR (%) and instantaneous whole-blood clear-
ance (ml/min) between MidD, PostD and PreD
MidD PostD PreD
Total convection, litres 54.5⫾2.3 25.2⫾1.2 47.7⫾1.4
RR urea 78.3⫾5.6a80.3⫾4.9a, b 76.8⫾7.4b
RR creatinine 69.0⫾5.7a70.3⫾4.6b66.3⫾6.1a, b
RR phosphorus 63.8⫾7.8 61.1⫾6.3 57.3⫾13.3
RR 2-MG 82.3⫾3.8a80.9⫾4.1b74.6⫾4.1a, b
K urea 273.6⫾12.7a288.8⫾10.0b264.5⫾5.3a, b
K creatinine 215.5⫾20.9a232.5⫾17.8b199.8⫾9.3a, b
K 2-MG 172⫾11.5a165.6⫾10.3b125.5⫾9.2a, b
EqKt/V 1.55⫾0.25a1.66⫾0.26a, b 1.52⫾0.27b
Values with same superscripts within rows indicate signifi cant differ-
ences, with p ⬍ 0.05.
Mid-Dilution 157
For phosphorus, RR, under the same conditions, was more favourable in
MidD (63.8%) than in PostD (61.1%) and PreD (57.3%).
Middle Molecules
The purifi cation of the MM depends above all on the convective fl ow and,
to a lesser degree, on the surface of the membrane, with PostD providing an
advantage. With the Olpur MD 190 and the FX80 having comparably sized
surface areas (1.9 and 1.8 m2, respectively), and taking into account the increase
in the total convective fl ow in MidD, compensating for the pejorative effect of
the predominant PreD participation, it is logical to fi nd comparable values for
the RR of 2-MG between MidD (82.3) and PostD (80.9), rather higher than
PreD (74.6).
Safety of Use
As regards safety, it should be recognized that the evaluation known as ‘2
points’ by the monitors of the TMP, using the ingoing blood pressure (PB in) and
the outgoing dialysate pressure (PD out) cannot be used with this type of dialyser.
In the initial confi guration, the TMP was found to be, depending on various
conditions of use [2, 7], between 611 and 713 mm Hg in the PostD stage and
between 293 and 307 mm Hg in the PreD stage, with PB in between 731 and 902
and PB out between 116 and 189 mm Hg.
In SR [2], these results drop to 422 ⫾ 90 in the PostD stage and 188 ⫾ 54 in
the PreD stage. However, there are neither clinical nor technical consequences
with these high modes of pressure. We had, in our experience, neither mem-
brane rupture nor abnormal frequency of coagulation.
On the other hand, the use of the Fresenius 5008 monitor, with convective
fl ows ⬎200 ml/min, and the joint use of blood temperature monitor (BTM) and
online clearance monitor (OCM) modules, makes it necessary to use the DR
mode to avoid unforeseen alarms or monitoring module failures.
The impact of a high-pressure system on albumin losses is nil, with an
average of 1.6 g/session with SR or DR confi gurations (unpubl. data with partial
collected dialysate) with QS between 150 and 225 ml/min.
Discussion
MidD has the advantage of offering within the same dialyser, without any
specifi c machine, the complementary methods of PostD and PreD, allowing
adequately large QS favourable to the MM and compensating for the diffusive
conditions which are largely rather unfavourable for purifi cation of the LM.
Potier 158
Besides, it is diffi cult to compare results from different papers, given the
different protocols used, notably in terms of QB (derived from the pump speed or
estimated by the monitor), QDd (with fl ow entirely or partially passing through
the dialyser, depending on the substitution fl ow monitor management), sam-
pling time, ultrafi ltration (standardized or not during sampling) etc.
In any case, the results obtained are well beyond the requirements in terms
of purifi cation of the PM, and even promising for phosphorus.
For MM, the penalizing effect of PreD, more important in the SR and DR
confi gurations than in the initial confi guration, has undoubtedly no clinical con-
sequences [8, 9].
In fact, the amount of convective volumes needed for their purifi cation are
well beyond the 15 litres recommended [10] and, anyway, the 2-MG plasma
water clearance exceeds its intercompartmental clearance estimated at 82 ml/
min, suggesting that only a different strategic approach in duration and fre-
quency will allow a better purifi cation [11].
Some data give an indication of a better per dialysis purifi cation of cystatin
C (CyC), retinol-binding protein (RBP) or leptin compared to PostD [9, 10] or
angiogenin and RBP compared to high-fl ux haemodialysis [12].
More interesting will be measurements of purifi cation of these potential
uraemic toxins, not only in terms of clearance, but also in residual blood con-
centration in the long term.
A multicentric study [13] has already confi rmed that MidD had a more
benefi cial effect on CyC and RBP than PostD.
A particular observation is that PostD and PreD association in the same
dialyser could allow patients with limited vascular access [14] to benefi t from
adequate convective fl uid.
The particular confi guration of the Olpur MD 190 also leads to an increase
in pressure, in particular in the blood compartment and thus of the TMP, irre-
spective of the values of the various monitors. However, the 2 years’ experience
in various dialysis centres does not reveal any noxious effect. With a larger
surface of 2.2 m2, the Olpur MD 220, soon to be available, will have 1 m2 (45%
of the total surface) allotted to the fi rst PostD stage, which is supposed to mini-
mize the TMP by allowing the same QS in a more signifi cant number of fi bres,
but also to increase purifi cation, notably of MM.
Anyway, in order to achieve a high-quality convective transfer with com-
plete peace of mind, we strongly recommend choosing the SR or DR confi gu-
ration of the Olpur MD 190, depending on the monitor used, knowing that the
studies, still unpublished, have not shown any difference in purifi cation between
these two modes.
Mid-Dilution 159
Conclusion
The Olpur MD 190 is a dialyser of innovating design, allowing the joint
and successive use, within the same dialyser, of both PostD and PreD HDF,
determining the MidD modality. It offers very high purifi cation performance
with, for MM, an apparent advantage compared to PostD, but which remains
to be confi rmed in the long term for other potentially noxious substances still
insuffi ciently studied.
Therefore, one can only recommend this HDF modality which guaran-
tees, under excellent safety conditions, a quality of dialysis which largely
exceeds the objectives of purifi cation recommended for a better survival in
dialysis.
References
1 Pedrini LA, Cozzi G, Faranna P, et al: Transmembrane pressure modulation in high-volume mixed
hemodiafi ltration to optimize effi ciency and minimize protein loss. Kidney Int 2006;69:573–579.
2 Santoro A, Ferramosca E, Mancini E, Monari C, Varasani M, Sereni L, Wratten M: Reverse mid-
dilution: new way to remove small and middle molecules as well as phosphate with high intrafi lter
convective clearance. Nephrol Dial Transplant 2007 April 3, E-pub ahead of print. DOI: 10.1093/
ndt/gfm101.
3 Henderson LW: Biophysics of ultrafi ltration and hemofi ltration; in Maher JF (ed): Replacement of
Renal Function by Dialysis. Dordrecht, Kluwer Academic, 1989, pp 300–326.
4 Potier J, Renaux JL: OLPUR 190MD: practical user recommendations for optimal performances.
Am Soc Nephrol Congr, 863A–864A, 2006.
5 Garred LJ: Mathematical modeling of the fl uid mechanics in a mid-dilution hemodiafi ltration
device. Am Soc Nephrol Congr, 861A, 2006.
6 Wizemann V, Külz M, Techert F, Nederlof B: Effi cacy of hemodiafi ltration. Nephrol Dial
Transplant 2001;16(suppl 4):27–30.
7 Feliciani A, Riva MA, Zerbi S, Ruggiero P, Lati AR, Cozzi G, Pedrini LA: New strategies in
haemodiafi ltration (HDF): prospective comparative analysis between on-line mixed HDF and
mid-dilution HDF. Nephrol Dial Transplant, 2007 March 8, E-pub ahead of print. DOI: 10.1093/
ndt/gfm023.
8 Krieter DH, Falkenhain S, Chalabi L, Collins G, Lemke HD, Canaud B: Clinical cross-over com-
parison of mid-dilution hemodiafi ltration using a novel dialyzer concept and post-dilution hemo-
diafi ltration. Kidney Int 2005;67:349–356.
9 Krieter DH, Collins G, Summerton J, Spence E, Leray Moragues H, Canaud B: Mid-dilution
on-line haemodiafi ltration in a standard dialyser confi guration. Nephrol Dial Transplant 2005;20:
155–160.
10 Canaud B, Bragg-Gresham JL, Marshall MR, Desmeules S, Gillespie BW, Depner T, Klassen P,
Port FK: Mortality risk for patients receiving hemodiafi ltration versus hemodialysis: European
results from the DOPPS. Kidney Int 2006;69:2087–2093.
11 Ward RA, Greene T, Hartmann B, Samtleben W: Resistance to intercompartmental mass transfer
limits 2-microglobulin removal by post-dilution hemodiafi ltration. Kidney Int 2006;69:1431–
1437.
12 Santoro A, Conz PA, De Cristofaro V, Acquistapace I, Gaggi R, Ferramosca E, Renaux JL, Rizziolo
E, Wratten ML: Mid-dilution: the perfect balance between convection and diffusion; in Ronco C,
Potier 160
Brendolan A, Levin N (eds): Cardiovascular Disorders in Hemodialysis. Contrib Nephrol. Basel,
Karger, 2005, vol 149, pp 107–114.
13 Krieter DH, Nicoud P, Christensson A, Fadel B, Wambergue F, Valentin R, Potier J, Collins G,
Canaud B: Long-term multicenter trial comparing post- to mid-dilution online HDF: effects on
low-molecular weight (LMW) proteins. Am Soc Nephrol Congr, 410A, 2006.
14 Renaux JL, Graziani G, Borlandelli S, Badalamenti S, Della Cà C, Alli A, Imbasciati E, Varasani
M, Mandolfo S: Evaluation of dialytic effi ciency of mid dilution hemodiafi ltration for patients
with limited vascular access. Am Soc Nephrol Congr, 722A, 2006.
Dr. Jacky Potier
Service d’Hémodialyse et de Néphrologie
Rue du Val de Saire
FR–50102 Cherbourg (France)
Tel. ⫹33 6 80 92 12 68, E-Mail j.potier@ch-cherbourg.fr
Ronco C, Canaud B, Aljama P (eds): Hemodiafi ltration.
Contrib Nephrol. Basel, Karger, 2007, vol 158, pp 161–168
Double High-Flux Hemodiafiltration
Beat von Albertini
Clinique Cecil et Division de Néphrologie, Centre Hospitalier Universitaire Vaudois,
Lausanne, Suisse
Abstract
Hemodiafi ltration (HDF) can augment the effi ciency of removal of small and large sol-
utes for renal replacement therapy. Double high-fl ux HDF was developed in 1984 with the
objective to shorten treatment time compared to conventional therapy. It consists of a serial
pair of dialyzers in the extracorporeal circuit for optimal diffusion and fi ltration, with substi-
tution by backfi ltration of bicarbonate dialysate under volumetric control. Used in conjunc-
tion with high blood and dialysate fl ow rates and high-fl ux membranes, unmatched high rates
of simultaneous diffusive and convective solute transport can be obtained clinically with
double HDF. A favorable long-term clinical outcome in comparison with conventional hemo-
dialysis was observed with this therapy, despite shorter treatment times.
Copyright © 2007 S. Karger AG, Basel
Hemodiafi ltration (HDF) describes an intermittent renal replacement
therapy of combined simultaneous diffusive and convective solute transport, in
which the total volume of ultrafi ltration largely exceeds the desired weight loss
for the patient and is in part replaced with a physiological solution. Substitution
occurs either by administration of extraneous parenteral solutions, as intro-
duced by Leber et al. [1], or by online fi ltered dialysate [2]. The treatment pre-
sented here, termed high-fl ux HDF in the original report of 1984, is based on a
serial pair of fi lters for optimized diffusion and fi ltration, with substitution by
controlled backfi ltration of bicarbonate dialysate in the extracorporeal circuit
[3, 4]. It was designed to introduce an unprecedented effi ciency of total sol-
ute transport and thereby to provide simultaneously a better renal replacement
therapy with shorter treatment times. It can be performed with standard modern
dialysis equipment with volumetric ultrafi ltration control and is better known
in routine clinical use among patients and staff by the colloquial term ‘double
high-fl ux’.
von Albertini 162
Prior Work
Since the successful demonstration of its feasibility, improvement of
renal replacement therapy, in terms of effi ciency of small- and large-solute
transport, has been a central preoccupation of investigators and has fostered
progress in clinical understanding and technical development in this fi eld.
Given the limitations of diffusive hemodialysis with cellulosic membranes for
large-solute transport, hemofi ltration, a purely convective renal replacement
therapy approximating the range of solutes fi ltered in the human glomerulus,
was introduced in 1967 [5]. An important innovation with this therapy was
online fi ltration for the production of sterile and pyrogen-free solutions for
substitution [6]. More effi cient simultaneous small- and large-solute removal
in dialysis was demonstrated in 1972 with polyacrylonitrile, a membrane with
enhanced diffusive and hydraulic permeability. To prevent excessive ultrafi l-
tration with the clinical use of this membrane, an apparatus for closed-cir-
cuit volume control of dialysate was developed, which forms an integral part
of modern dialysis machines with programmable weight loss for the patient
[7]. The clinically obtainable effi ciency of solute transport in dialysis at that
time was nevertheless primarily limited by the rate at which the acetate buf-
fer, gained from the then used dialysate, could be metabolized by the patients,
resulting in clinical intolerance and vascular instability at blood fl ow rates
higher than 250–300 ml/min [8]. The development of bicarbonate dialysis in
1976 [9] effectively removed this barrier, but the potential for increasing treat-
ment effi ciency clinically remained unrecognized for some time [10, 11]. A
serendipitous discovery, made with hemofi ltration, was that high blood fl ow
rates (up to 500 ml/min), introduced in an effort to obtain small-solute removal
similar to control hemodialysis without prolonging treatment time, were
observed not only to be available from vascular access, but also to be well tol-
erated by the patients [12]. The critical importance of appropriate small-solute
removal was highlighted in 1983 by the outcome of the National Cooperative
Dialysis Study, incidentally demonstrating the validity of the concept of reci-
procity of effi ciency versus time for prescription of small-solute removal with
dialysis [13]. Shorter treatments with HDF were introduced in 1983 but later
abandoned [14].
Method
A confi guration of 2 high-fl ux dialyzers is used in the extracorporeal circuit, as depicted
schematically in fi gure 1 (on the left). Its key functional features for HDF are a fl ow restrictor
between the serial dialyzers in the countercurrent dialysate circuit (in the middle) and the
Double HDF 163
volumetric control of the standard equipment (on the right), which also contains a fi lter for
online sterile and pyrogen fi ltration of fresh bicarbonate dialysate.
The volumetric control of dialysate, a standard feature of modern dialysis machines,
consists of a closed hydraulic circuit, in which the volume of degassed dialysate entering the
extracorporeal circuit is precisely matched to that leaving the circuit by volume displace-
ment. This is technically achieved by 2 identical loops within the machine, which are alter-
nately opened and closed by a set of control valves and pumps. Each loop contains a
fi xed-volume chamber which is separated into 2 compartments by a fl exible membrane. In
one circuit, one of these 2 compartments is fi lled with infl owing freshly prepared dialysate,
while the other compartment, fi lled in a previous cycle with spent dialysate, is thereby emp-
tied into the drain. Simultaneously, in the other loop, one compartment is fi lled with spent
dialysate and thereby empties the other compartment, fi lled with fresh dialysate, into the
dialyzer circuit. To achieve the desired weight loss during the treatment, a predetermined
volume is continuously removed from the effl uent dialysate by a programmable pump and
bypasses the volume control mechanism. The resulting volume defi cit lowers the hydrostatic
pressure in the closed circuit and thereby augments the pressure gradient across the dialyzer
membrane (transmembrane pressure, TMP). Ultrafi ltration of plasma water occurs in
response to the self-adjusting TMP, at a rate matching the volume of dialysate removed from
the closed circuit.
The principle of self-adjusting TMP under volumetric control is exploited for HDF in
this confi guration of 2 serial dialyzers with high hydraulic permeability. Figure 2 depicts the
pressure profi le for blood and dialysate in the countercurrent extracorporeal circuit, as typi-
cally encountered during clinical treatments. Pressure in the blood circuit (from left to right)
800
825
To drain
Polysulfone
Serial pair of dialyzers and
dialysate flow restrictor
30ml/cycle
Volumetric control
(Fresenius 4008 and 5008)
Sterile/pyrogen filter
(Diasafe® plus)
Net UF pump
Volumetric
control
Active cycle
Inactive cycle
Dialysate
pump
800
Fresh
dialysate
Closed
loop
Extracorporeal circuit Equipment
30ml/cycle
Ultrafiltration 150
600
450
825
Flow restrictor
I
Polyamide
2.1m2
II
Polyamide
2.1m2
575
800
Blood pump
675
675
125 Backfiltration 2.2m2
25
With representative flow rates (ml/min) for blood ( ) and dialysate ( )
Fig. 1. Confi guration for double HDF with representative fl ow rates occurring during
clinical operation.
von Albertini 164
diminishes gradually from a high initial pressure, generated by the high rate of the blood
pump and the fl ow resistance of the serial dialyzers, to the pressure in the venous blood line
returning to the patient. Pressures in the countercurrent dialysate circuit (from right to left)
are modifi ed by the fl ow restrictor between the serial dialyzers, resulting in higher pressures
upstream than downstream of the restriction. The fl ow restrictor is a very simple device, con-
sisting of a clamp and a calibrated bore in the dialysate line. Once the clamp is manually
closed, all dialysate is forced through the restrictive bore until the clamp is opened at the end
of treatment.
During treatment, an important hydrostatic pressure gradient exists between blood and
dialysate compartments of the fi rst and in part of the second dialyzer, resulting in high ultra-
fi ltration of plasma water to dialysate and thereby increasing the blood oncotic pressure. The
higher dialysate and lower blood compartment hydrostatic pressures result in a reverse TMP
towards the end of the second dialyzer and transfer of fl uid from dialysate to blood. This
backfi ltration is facilitated by oncotic pressure, which diminishes as the concentrated blood
is rediluted. The pressures in the confi guration are continuously self-adjusting by the volu-
metric control of the equipment, maintaining optimal transmembrane fl uxes during treat-
ment. The magnitude of these fl uxes is evident from the representative clinical fl ow rates
given in fi gure 1. Blood fl ow diminishes during passage of the fi rst fi lter (in this example
from 600 to 450 ml/min), due to ultrafi ltration (150) and is rediluted in the second fi lter (to
575) by backfi ltation (125), the difference being the net weight loss rate (25), programmed
for the treatment in the conventional fashion. Conversely, countercurrent dialysate fl ow
diminishes in the second fi lter (from 800 to 675 ml/min), due to backfi ltation (125) and
increases again in the fi rst fi lter (to 825), due to addition of ultrafi ltered plasma water (150).
The entire system is self-contained and functions automatically: all that is required for opera-
tion is that the blood fl ow and weight loss rate be set and the fl ow restrictor clamp be closed
at the onset of treatment and the patient and equipment be monitored in the conventional
manner.
Blood in
Dialysate
out Blood out
Dialysate
in
0
200
400
600
800
1,000
Pressure (mmHg)
Dialyzer I Dialyzer IIArterial VenousMiddle
Ultrafiltration
Backfiltration
Oncotic pressure
Dialysate flow restrictor
Blood flow
Dialysate flow
Fig. 2. Pressure profi le and transmembrane fl uxes in the extracorporeal circuit of dou-
ble HDF, typically encountered during clinical operation.
Double HDF 165
Results
Used in conjunction with high blood and dialysate fl ow rates and dialyzer
membranes with high diffusive and hydraulic permeability, substantial rates of
simultaneous diffusive and convective solute transport can be obtained with
this confi guration. A variety of membranes have been used for this purpose
[15], made of cellulose acetate, polymethylmethacrylate, polysulfone, acryloni-
trile and polyamide, which is now routinely used in our center. A recent clear-
ance study, made with two Gambro Polyfl ux® 21R dialyzers (2.1 m2) in series,
revealed at an effective blood fl ow rate of 533 ml/min and a dialysate fl ow rate
of 808 ml/min (with ultrafi ltration in the fi rst fi lter of 167 ml/min, backfi ltration
in the second of 150 ml/min and net weight loss rate of 25 ml/min) an average
of whole-blood and dialysate determined clearances of 449 ml/min for urea,
420 ml/min for creatinine and 370 ml/min for phosphate, respectively, and an
average of plasma and dialysate clearances for 2-microglobulin of 168 ml/min
in the extracorporeal circuit [unpubl. data].
From its inception, good clinical tolerance was observed with double HDF,
despite higher weight loss rates during treatment. Shortened treatment times did
not negatively affect blood pressure control in the patients [16]. The long-term
clinical outcome with double HDF has recently been analyzed in comparison
with other high-effi ciency hemodialysis treatments in a total of 183 patients over
6 years [17]. Patient survival was overall favorable with all high-effi ciency ther-
apy in comparison to the national registry (US Renal Data System), matched for
patients’ age, race and etiology of end-stage renal disease. The best outcome was
observed in the group treated with double HDF, which incidentally was the one
with the shortest treatment times. Analysis by standardized mortality rate revealed
a 59% better outcome in comparison to the registry for this group, which was the
only one reaching a statistically signifi cant difference by this analysis [18].
Discussion
The development of this treatment must be placed in the historical context
in which it occurred, which was towards the conclusion of a period which can
now be considered to have probably been the ‘golden years’ for the develop-
ment of renal replacement therapy. Based on earlier fundamental contributions
establishing feasibility, therapy was particularly moved forward during this time
by a large number of seminal contributions, which enhanced clinical under-
standing and led to most of the technical developments incorporated into mod-
ern dialysis. In this stimulating period, interaction within the relatively small
von Albertini 166
international community of clinicians and investigators was intense and open-
minded, and clinical research in this fi eld was encouraged by funding agencies,
namely the NIH.
The primary goal of intermittent renal replacement therapy is to improve
the human condition of patients with end-stage renal disease by maintaining
relative well-being and enabling the pursuit of a functional life. Well-being is
largely proportional to the quantity of and quality of dialytic therapy, relating
to the absence of intra- and interdialytic side effects. Quantity of dialysis is
determined by the total clearance volumes of metabolic solutes approximating
those excreted by the natural kidneys; technically it is effected by the product
of treatment effi ciency and treatment time. Time spent on treatment, on the
other hand, impacts negatively on the one available for more enjoyable activi-
ties, which also defi ne quality of life for a patient. It was in this perspective
that the attempt was made in 1984 to combine all known elements of the state-
of-art of the time, relevant for effi ciency and clinical tolerance for optimiza-
tion of treatment, with the set objective to substantially shorten the treatment
time over conventional therapy without a reduction of removal of small and
large solutes.
Double HDF effectively more than doubled the rates of solute transport
over conventional therapy. A confi guration of paired dialyzers had previously
been used for hemodialysis [19] and, in a serial confi guration, for HDF with
self-generation of substitution fl uid [20, 21]. The progress made with double
HDF stems from the conjunction of this confi guration with the unprecedented
high blood fl ow rates and bicarbonate dialysate during clinical treatments. The
former were found to be readily available from most patients’ vascular access,
and the latter contributed to the observed good clinical tolerance of high treat-
ment effi ciency.
It is estimated that close to 100,000 treatments with double HDF have
been performed up to now in the USA, Europe and Asia without notable
complications. The treatment is well liked by patients and staff for the good
clinical tolerance it provides, even at high net ultrafi ltration rates (up to
30 ml/min), for its shorter treatment times and ease of operation. For the
clinician, it is a valued tool to provide patients with high body weight with
a treatment meeting guidelines for treatment adequacy. Further application
is limited by the need for double fi lters, which is economically unfavorable
without reuse.
Acknowledgements
The author wishes to acknowledge the stimulation and encouragement he was privi-
leged to receive from the contributions of and the personal contact with many investigators,
Double HDF 167
who cannot all be listed here. He wishes in particular to express his gratitude to his mentors
and collaborators, namely the late James H. Shinaberger and Joseph H. Miller of Los Angeles,
without whom the development of this treatment could not have been realized, and Juan
P. Bosch and Viroj Barlee in Washington, D.C. and Jacky Berger in Lausanne, without whom
it could not have been implemented clinically.
References
1 Leber HW, Wizemann V, Goubeaud G, Rawer P, Schuetterle G: Hemodiafi ltation: a new alterna-
tive to hemofi ltration and conventional hemodialysis. Artif Organs 1978;2:150–153.
2 Canaud B, Nguyen A, Argiles C, Polito C, Polaschegg HD, Mion C: Hemodiafi ltration using dialy-
sate as substitution fl uid. Artif Organs 1987;11:188–190.
3 von Albertini B, Miller JH, Gardner PW, Shinaberger JH: High-fl ux hemodiafi ltration: under six
hours/week treatment. Trans Am Soc Artif Intern Organs 1984;30:227–231.
4 Miller JH, von Albertini B, Gardner PW, Shinaberger JH: Technical aspects of high-fl ux hemodia-
fi ltration for adequate short (under two hours) treatment. Trans Am Soc Artif Intern Organs 1984;
30:377–381.
5 Henderson LW, Besarab A, Michaels AS, Bluemle LW: Blood purifi cation by ultrafi ltration and
fl uid replacement (diafi ltration). Trans Am Soc Artif Intern Organs 1967;16:216–222.
6 Henderson LW, Beans E: Successful production of sterile pyrogen-free electrolyte solution by
ultrafi ltration. Kidney Int 1978;14:522–525.
7 Funck-Brentano JL, Sausse A, Man NK, Grangier A, Roudon-Nucete M, Zingraft J, Jungers P:
A new method for hemodialysis combining a high permeability membrane for the medium mol-
ecules and a dialysis bath in a closed circuit. Proc Eur Dial Transplant Assoc 1972;9:55–66.
8 Kveim M, Nesbakken R: Utilization of exogenous acetate during hemodialysis. Trans Am Soc
Artif Intern Organs 1975;21:138–143.
9 Graefe U, Multinovitch J, Follete WC, Vizzo JE, Babb AL, Scribner BH: Less dialysis-induced mor-
bidity and vascular instability with bicarbonate in the dialysate. Ann Intern Med 1978;88:332–336.
10 von Albertini B, Petersen J: Comparison of high blood fl ows in bicarbonate vs acetate hemodi-
alysis (abstract). National Kidney Foundation Scientifi c Meeting Program and Abstracts. Am J
Kidney Dis 1983;p32.
11 Keshaviah P, Collins A: Rapid high-effi ciency bicarbonate hemodialysis. Trans Am Soc Artif
Intern Organs 1986;32:17–23.
12 Geronemus R, von Albertini B, Glabman S, Kahn T, Moutoussis G, Bosch JP: High-fl ux hemodia-
fi ltration: further reduction in treatment time. Proc Clin Dial Transplant Forum 1979;9:125–127.
13 Parker TF, Laird NM, Lowrie EG: Comparison of the study groups in the National Cooperative
Dialysis Study and a description of morbidity, mortality and patient withdrawal. Kidney Int Suppl
1983;13:S42–S49.
14 Wizemann V, Kramer W, Knopp G, Rawer P, Mueller K, Schuetterle G: Ultrashort hemodiafi ltra-
tion: effi ciency and hemodynamic tolerance. Clin Nephrol 1983;19:24–30.
15 von Albertini, B, Miller JH, Gardner PW, Shinaberger JH: Performance characteristics of high-
fl ux hemodiafi ltration. Proc Eur Dial Transplant Assoc 1984;21:447–453.
16 Velasquez M, von Albertini B, Lew SQ, Mishkin G, Bosch JP: Equal levels of blood pressure
control in ESRD patients receiving high-effi ciency hemodialysis and conventional hemodialysis.
Am J Kidney Dis 1998;31:618–623.
17 Bosch JP, Lew SQ, Barlee V, Mishkin GJ, von Albertini B: Clinical use of high-effi ciency hemodi-
alysis treatments: long-term assessment. Hemodial Int 2006;10:73–81.
18 Wolfe RA, Gaylin DS, Port FK, Held PJ, Wood CL: Using USRDS generated mortality tables to
compare local ESRD mortality rates to national rates. Kidney Int 1992;42:991–996.
19 Kuruvila KC, Cadnapaphornchai P, Leasor G, Popovitzer M, Alfrey A, Schrier RW: A model for
evening home hemodialysis. Am J Med 1974;57:706–713.
von Albertini 168
20 Shinzato T, Sezaki R, Matsatune U, Maeda K, Ohbayashi S, Toyota T: Infusion-free hemodiafi ltra-
tion: simultaneous hemofi ltration and dialysis with no need for infusion fl uid. Artif Organs 1982;
6:453–456.
21 Cheung AC, Kato Y, Leypolt JK, Henderson LW: Hemodiafi ltration using a hybrid membrane
system for self-generation of diluting fl uid. Trans Am Soc Artif Organs 1982;28:61–65.
Dr. Beat von Albertini
Centre de dialyse Cecil
Avenue de Savoie 10
CH–1003 Lausanne (Switzerland)
Tel. ⫹41 21 343 01 82, Fax ⫹41 21 343 01 81, E-Mail Beat.vonalbertini@hirslanden.ch
Ronco C, Canaud B, Aljama P (eds): Hemodiafi ltration.
Contrib Nephrol. Basel, Karger, 2007, vol 158, pp 169–176
Push/Pull Hemodiafiltration
Toru Shinzato, Kenji Maeda
Daiko Medical Engineering Research Institute, Nagoya, Japan
Abstract
Push/pull hemodiafi ltration is characterized by alternate fi ltration and backfi ltration,
while sterile pyrogen-free dialysate is fl owing through a hemodiafi lter. During the fi ltration
phase, uremic substances are eliminated not only by diffusive, but also by convective trans-
port. During the backfi ltration phase, dialysate is quickly pushed to the blood side (i.e. back-
fi ltration) so as to make up for the excessive reduction in body fl uid that has developed
during the immediately preceding fi ltration phase. In the most recently improved version of
push/pull hemodiafi ltration, the body fl uid replacement volume is over 120 liters during a 4-
hour treatment. This replacement of a large amount of body fl uid may be due to the increased
fi ltration rate in the hemodiafi lter resulting from failure of the complete formation of a pro-
tein gel layer on the blood side surface. The fi ltration time in push/pull hemodiafi ltration is so
short that the also short backfi ltration to follow may take over before the protein gel layer is
completely formed on the membrane surface. Since the fi ltration and backfi ltration times are
much shorter in push/pull hemodiafi ltration than the time for blood to pass through the hemo-
diafi lter, it is concentrated and diluted many times (approx. 25 times) before it leaves the
hemodiafi lter. Therefore, push/pull hemodiafi ltration is functionally similar to a predilution
hemodiafi ltration. The reduction rate of 2-microglobulin was greater by push/pull hemodia-
fi ltration than by hemodialysis, when a high-fl ux polysulfone hemodiafi lter was employed.
However, the difference in the reduction rate was rather small between them, because of the
improved hemodiafi lters, which remove so much 2-microglobulin only by dialysis.
Nevertheless, restless legs syndrome, irritability, insomnia and pruritus were alleviated after
switching the treatment modality from hemodialysis to push/pull hemodiafi ltration. This may
indicate that these symptoms are caused by the accumulation of uremic substances larger
than 2-microglobulin.
Copyright © 2007 S. Karger AG, Basel
In online hemodiafi ltration, a hemodiafi ltration method gaining wide-
spread acceptance, a part of the dialysate is infused into blood tubing as a sub-
stitution fl uid through a line from the dialysate pathway connected to the blood
tubing [1]. On the other hand, in push/pull hemodiafi ltration, which has been
Shinzato/Maeda 170
in use in Japan for more than 20 years, fi ltration and backfi ltration are repeated
alternately in the hemodiafi lter; uremic substances are eliminated by diffusion
and convection during the fi ltration phase, while during the backfi ltration phase
the dialysate is quickly pushed to the blood side as a substitution fl uid in the
hemodiafi lter [2].
This chapter reviews the mechanical and functional characteristics of push/
pull hemodiafi ltration and discusses the clinical effectiveness of this treatment
modality.
Principles
By defi nition, push/pull hemodiafi ltration is an alternate fi ltration and
backfi ltration that takes place while a sterile pyrogen-free dialysate is fl owing
through a hemodiafi lter. During the fi ltration phase, a certain volume of fl uid
is fi ltered from the blood side to the dialysate side, so as to eliminate uremic
substances by not only diffusive, but also convective transport. During the
backfi ltration phase, almost the same volume of dialysate is quickly pushed to
the blood side to compensate for the excessive reduction in body fl uid that has
developed during the fi ltration phase just before.
Push/Pull Machines
The alternate repetition of fi ltration and backfi ltration by pushing and pull-
ing dialysate in and out of the dialysate fl ow pathway is done by straightforward
push/pull machine action. The push/pull machine itself was developed in 1982
[2] and improved largely in 1994 [3]. In the original version of this push/pull
machine, the push/pull volume of the dialysate was 200 ml as compared with
the mere 16.7 ml with the more recently improved machine.
With each alternate fi ltration and backfi ltration, the blood effl ux from the
hemodiafi lter decreases or increases, respectively, when either the original or
a more recently improved machine is used. To assure a constant fl ow of blood
back to the patient’s body, with the original machine, a blood volume equivalent
to the volume of alternately occurring fi ltration and backfi ltration is pulled out
from the blood tubing into the plastic bag reservoir and then pushed back from
the reservoir into the blood tubing downstream of the hemodiafi lter, in syn-
chrony with the backfi ltration and fi ltration. With the more recently improved
machine, on the other hand, air is also pulled out from the venous chamber and
then pushed back into the chamber alternately, in synchrony with the backfi ltra-
tion and fi ltration, to assure a constant fl ow of blood back to the patient’s body.
Push/Pull Hemodiafi ltration 171
Mechanism of the Push/Pull Machine
Since the original one is rarely used today, we focus here on the mecha-
nism of the recently improved push/pull machine.
Constant Flow of Blood Back to the Body
The key component of the push/pull machine is a double-cylinder piston
pump with a discharge volume of 16.7 ml in either direction. This 2-way pump
pulls dialysate out from the dialysate removal pathway so as to develop fi ltra-
tion in the hemodiafi lter and simultaneously pushes air into the venous chamber
so as to lower the air-fl uid level therein. Next, the piston pump pushes the dialy-
sate back into the dialysate removal pathway so as to develop backfi ltration and
simultaneously pulls air from the venous chamber so as to elevate the air-fl uid
level in the chamber, when a volumetric ultrafi ltration controller is employed.
With push/pull hemodiafi ltration performed using this approach, the variation
in the blood fl ow returned to the patient’s body, which is due to alternate fi ltra-
tion and backfi ltration, is made constant by lowering and elevating the air-fl uid
level in the venous chamber.
Control of Transmembrane Pressure
The membrane water permeability changes depending on the kind of
membrane material and decreases progressively during hemodiafi ltration treat-
ment. In the push/pull hemodiafi ltration machine, the pump operation of the
double-cylinder piston pump is automatically controlled so that the transmem-
brane pressure (TMP) is consistently maintained at a preset level (i.e. usually
the highest level in the safety range) throughout treatment according to the fol-
lowing mechanism.
The pressure difference between the blood side and dialysate side cylin-
ders of the double-cylinder piston pump is almost equivalent to the TMP in the
hemodiafi lter, as shown in fi gure 1, because the pressure (PD) on the dialysate
side cylinder is equal to the pressure in the dialysate tubing, and the pressure
(PB) in the blood side cylinder is equal to the pressure in the venous chamber.
TMP ⫽ P B ⫺ P D. (1)
We may express the force (F) opposing the piston movement as the product of
the cross-sectional area (S) and the pressure difference between the blood side
and the dialysate side cylinders which is almost equal to the TMP:
F ⫽ (PB ⫺ P D) ⫻ S. (2)
Since the reciprocation of the double-cylinder piston pump is translated from the
cam rotation, by appropriately controlling the force which rotates the cam (i.e.
Shinzato/Maeda 172
the torque of the cam), one can apply a constant force on the piston. Figure 1
also shows the cam and double-cylinder piston pump. The torque of the cam
can be expressed by the following equation:
T ⫽ I ⫻ F (3)
where T is the cam torque and I is the distance from the straight line passing
through the center of the cam (which is parallel to the reciprocation direction of
the piston pump) to the connecting point of the cam and piston.
In the following equation, I is obtained:
I ⫽ L sin (4)
where L indicates the distance from the center of the cam to the connecting
point of the cam and piston, and denotes the angle of the straight line passing
through the center of the cam (which is parallel to the reciprocation direction of
the piston), and the straight line passing through the center of the cam and the
connecting point of the cam and piston.
Venous
chamber
Air filter Hemodiafilter
Ultrafiltration
controller
Blood pump
Rotating direction Piston
Motor
Cam Dialysate
side cylinder
Pb
Pd
I
L
Blood side
cylinder
Double-cylinder
piston pump
S
Fig. 1. Schematic diagram of the push/pull hemodiafi ltration system. PD and PB indi-
cate the pressures in the dialysate and blood compartments of the cylinder, respectively, S
the cross-sectional area of the piston, and L the distance from the center of the cam and I the
distance from the straight line passing through the center of the cam to the connecting point
of the cam and piston.
Push/Pull Hemodiafi ltration 173
The combination of equations 1–4 yields the following equation, by which
the cam torque is obtained:
T ⫽ TMP ⫻ S ⫻ L sin . (5)
The cam torque is proportional to the voltage applied to the DC motor con-
nected to the cam. Therefore, in a push/pull machine, the DC motor has been
improved so that the cam torque can be controlled at TMP ⫻ S ⫻ L sin at any
moment by changing the voltage applied to the motor in relation to the continu-
ously monitored value. The TMP is thus maintained at a preset constant value
during both the fi ltration and backfi ltration phases, usually 400 mm Hg during
the fi ltration phase and ⫺400 mm Hg during the backfi ltration phase.
A rigid, braid-reinforced silicone tubing must be used instead of the con-
ventional dialysate tubings to prevent the dialysate tubings from absorbing a
signifi cant portion of the push/pull volume of dialysate with this machine.
Functional Characteristics of Push/Pull Hemodiafiltration
Replacement Volume
The times for fi ltration and backfi ltration are approximately 0.8 and 0.7 s,
respectively, and the fi ltration rate from the blood is around 2.8 ml/mm Hg/min
when the TMP is set at the maximum permissible level within the safety range
throughout treatment, whether during the fi ltration or backfi ltration phase. Due
to this high fi ltration rate, which is almost equivalent to the fi ltration rate of tap
water (3.3 ml/mm Hg/min), the body fl uid replacement volume will exceed 120
liters during a 4-hour push/pull hemodiafi ltration.
The resulting increased fi ltration rate may be due not only to the maxi-
mum permissible TMP during the fi ltration and backfi ltration phases, but also
to the failure of the protein gel layer [4–6] to form completely on the blood
side surface of the hemodiafi lter membrane during the fi ltration phase. As men-
tioned earlier, in push/pull hemodiafi ltration, the fi ltration time is so short that
the short backfi ltration to follow may take over before the protein gel layer is
completely formed on the membrane surface.
Replacement Mode
Since the fi ltration and backfi ltration times in push/pull hemodiafi ltration
are much shorter at 0.8 and 0.7 s, respectively, than the time (40 s) for extracor-
poreally circulating blood to pass through the hemodiafi lter, it is concentrated
and diluted many times (approx. 25 times) before it leaves the hemodiafi lter.
Therefore, the push/pull hemodiafi ltration is functionally similar to a predilu-
tion hemodiafi ltration.
Shinzato/Maeda 174
Preparation of Dialysate
Since dialysate is infused as a substitution solution in push/pull hemodiafi l-
tration, just as with online hemodiafi ltration, the fi nally infused dialysate must
be of intravenous quality. Therefore, dialysate is prepared so as to be sterile and
nonpyrogenic in the push/pull hemodiafi ltration system.
The dialysate preparation process in the push/pull hemodiafi ltration sys-
tem is virtually the same as for the online hemodiafi ltration system [7].
Solute Removal
According to our recent unpublished data, the reduction rate of 2-micro-
globulin was greater by push/pull hemodiafi ltration than by hemodialysis, when
a high-fl ux polysulfone hemodiafi lter (i.e. PS-1.9UW®, Kawasumi Co. Ltd.,
Tokyo, Japan) was employed with a volumetric ultrafi ltration controller (DCS-
26®, Nikkiso Co. Ltd., Tokyo, Japan). However, the difference in the reduc-
tion rate was rather small (71.5 ⫾ 6.5% with push/pull hemodiafi ltration against
62.5 ⫾ 5.8% with hemodialysis; p ⬍ 0.05). This is due to the vastly improved
hemodiafi lter, which even in conventional hemodialysis can remove a huge
amount of 2-microglobulin.
We wish to remove great amounts of low-molecular-weight protein, typi-
fi ed by 2-microglobulin, with a minimized loss of albumin. However, although
albumin (60,000 Da) and 2-microglobulin (12,000 Da) have a great difference
in molecular weight, there is only a small difference in their molecular radius.
At the time when the original or present push/pull machine was developed, no
membrane was available which could screen albumin and 2-microglobulin
very effectively. At that time, when the membrane albumin sieving coeffi cient
(SC) was kept at 0.01, the SC for 2-microglobulin could only be increased
to 0.3. However, nowadays membranes have been developed with which the
albumin SC can be kept to 0.006, while the SC for 2-microglobulin can be
increased to over 0.8. Of course, this type of membrane also shows a marked
increase in its ability to eliminate 2-microglobulin by diffusion. Moreover,
when such membranes are incorporated in a recent hemodiafi lter, the removal
of 2-microglobulin is more remarkable due to enhanced internal fi ltration
[8].
Therefore, as long as the recent hemodiafi lter incorporating such mem-
branes is used, there may not be much difference in 2-microglobulin removal
between hemodialysis and push/pull hemodiafi ltration, or between hemodialy-
sis and online hemodiafi ltration.
Push/Pull Hemodiafi ltration 175
Although the reduction rate of 2-microglobulin is greater by push/pull
hemodiafi ltration to some extent than by hemodialysis, the urea reduction rate
is comparable between them.
Clinical Effectiveness
Although the reduction rate of 2-microglobulin was greater by push/pull
hemodiafi ltration than by hemodialysis, the difference was rather small between
them, because of the new and vastly improved hemodiafi lters, which remove so
much 2-microglobulin only by dialysis. Nevertheless, restless legs syndrome,
irritability, insomnia and pruritus were clearly alleviated after switching the
treatment modality from conventional hemodialysis to push/pull hemodiafi ltra-
tion according to our observations. This may indicate that these symptoms could
be due to accumulation of uremic substances larger than 2-microglobulin.
Treatment Cost
The fi nal pyrogen fi lter on the substitution fl uid line branching out from
the dialysate pathway and connected to the blood tubing is essential to online
hemodiafi ltration. In push/pull hemodiafi ltration, on the other hand, the hemo-
diafi lter also serves as the fi nal pyrogen fi lter. Therefore, when the fi nal pyrogen
fi lter must be disposable in online hemodiafi ltration, push/pull hemodiafi ltra-
tion is more cost-effective.
References
1 Canaud B, N’Guyen QV, Lagarde C, Stec F, Polaschegg HD, Mion C: Clinical evaluation of a
multipurpose system adequate for hemodialysis, for postdilution hemofi ltration/hemodiafi ltra-
tion with on-line preparation of substitution fl uid from dialysate; in Streicher E, Seyffart G (eds):
Highly Permeable Membranes. Contrib Nephrol. Basel, Karger, 1985, vol 46, pp 184–186.
2 Usuda M, Shinzato T, Sezaki R, Kawanishi A, Maeda K, Kawaguchi S, Shibata M, Toyoda T,
Asakura Y, Ohbayashi S: New simultaneous HF and HD with no infusion fl uid. Trans Am Soc
Artif Organs 1982;28:24–27.
3 Shinzato T, Fujisawa K, Nakai S, Miwa H, Kobayakawa H, Takai I, Morita H, Maeda K: Newly
developed economical and effi cient push/pull hemodiafi ltration; in Maeda K, Shinzato T (eds):
Effective Hemodiafi ltration: New Methods. Contrib Nephrol. Basel, Karger, 1994, vol 108, pp
79–86.
4 Dorson WJ, Pizziconi VB, Allen JM: Transfer of chemical species through a protein gel. Trans Am
Soc Artif Organs 1971;17:287–292.
5 Porter MC: Concentration polarization with membrane ultrafi ltration. Ind Eng Chem Prod Res
Dev 1972;11:234–248.
Shinzato/Maeda 176
6 Colton CK, Henderson LW, Ford CA, Lysaght MJ: 1. In vitro transport characteristics of a hollow-
fi ber blood ultrafi lter. J Lab Clin Med 1975;85:355–371.
7 Canaud B, Flavier JL, Argil SA, Stec F, N’Guyen QV, Bouloux CH, Garred LJ, Mion C:
Hemodiafi ltration with on-line production of substitution fl uid: long-term safety and quantitative
assessment of effi cacy; in Maeda K, Shinzato T (eds): Effective Hemodiafi ltration: New Methods.
Contrib Nephrol. Basel, Karger, 1994, vol 108, pp 12–22.
8 Dellanna F, Wuepper A, Baldamus CA: Internal fi ltration-advantage in haemodialysis? Nephrol
Dial Transplant 1996;11(suppl 2):83–86.
Toru Shinzato
Daiko Medical Engineering Research Institute
4–16–23, Daiko, Higashi-ku
Nagoya-shi, Aichi-ken 461-0043 (Japan)
Tel. ⫹81 52 711 8889, Fax ⫹81 52 711 8808, E-Mail shinzato@xj8.so-net.ne.jp
Ronco C, Canaud B, Aljama P (eds): Hemodiafi ltration.
Contrib Nephrol. Basel, Karger, 2007, vol 158, pp 177–184
Principles and Practice of Internal
Hemodiafiltration
Gianfranco Beniamino Fiorea, Claudio Roncob
aDipartimento di Bioingegneria, Politecnico di Milano, Milan, and
bDepartment of Nephrology, St. Bortolo Hospital, Vicenza, Italy
Abstract
It has recently been suggested that the potentials of modern high-fl ux membranes could
be exploited with the so-called internal hemodiafi ltration (iHDF) technique. In principle,
iHDF works just as high-fl ux hemodialysis but requires the convective dose to be clinically
relevant, quantifi able and possibly adjustable by the operator. In this chapter, we briefl y sur-
vey the theoretical, technological and practical aspects of iHDF, focusing on the mechanism
ensuring its convective potential, i.e. the internal fi ltration/backfi ltration (IF/BF) phenome-
non. Based on theory, it is highlighted that the enhancement of the convective dose during
iHDF relies upon a wise design of the hemodialyzer, both in terms of membrane perfor-
mance and of hydrodynamics, whereas the adjustment of convection is feasible by proper
regulation of the treatment parameters. IF/BF measurements appear to be feasible by indirect
means; however, investments are needed to bring technology from the mere research fi eld to
the clinical practice. An alternative approach to IF/BF quantifi cation is using mathematical
models provided that the developed calculation tools are handy enough for the clinician, or
even be implemented in the dialysis machine itself. IF/BF ‘calculators’ also represent a
means to make the clinical staff conscious of the fi ltration phenomena that take place inside
a high-fl ux hemodialyzer. It is concluded that iHDF is a possible complementary, simplifi ed
technique which could increase the clinical diffusion of high-fl ux convective treatments in
the near future.
Copyright © 2007 S. Karger AG, Basel
It is a widely accepted concept that an appropriate combination of con-
vective solute transport to diffusive solute transport enhances the potentials
of hemodialysis. The growing success of the clinical use of hemodiafi ltration
(HDF) [1–3] is evidence of this: HDF is the convective treatment par excel-
lence, because it allows the convective dose to reach very high levels (up to
4.5–5.0 l/h in postdilution) and to be under the operator’s direct control. On the
Fiore/Ronco 178
other hand, HDF suffers from some practical disadvantages (basically, its cost
and the complication of the hardware and procedures involved) which poten-
tially limit its clinical spread.
Convective transport is the leading benefi t of high-fl ux dialysis, too. But
high-fl ux dialysis is based on letting convective transport happen spontaneously
within the dialyzer, thanks to the internal fi ltration/backfi ltration (IF/BF) mech-
anism [4, 5]. As detailed below, in the usual countercurrent setting, direct cross-
fi ltration (DF) happens in the proximal part of the device, which provides for
convective solute removal since DF is discharged with the exhausted dialysate.
Distal BF of fresh dialysate acts as a ‘spontaneous reinfusion’ phenomenon,
intrinsically ensuring fl uid balance. Exploiting BF implies a need for good water
quality, just as for the online version of HDF, but it introduces an additional
screen for the reinfused fl uid (i.e. the dialyzer’s membrane itself); moreover, it
avoids the need for substitution fl uids or additional technology. High-fl ux dialy-
sis thus has enough practical advantages to be appealing to the clinician as a
‘hidden HDF’. We proposed that high-fl ux dialysis could be reconsidered as
a simplifi ed, ‘internal’ HDF technique (iHDF) [6], provided that the following
conditions are met: (1) the attainable convective dose should be clinically rel-
evant; (2) the convective dose should be under the operator’s control or, at least,
be quantifi able in the clinical theater.
Increasing the Amount of Convection in Internal Hemodiafiltration
IF/BF is governed by hydraulic and oncotic pressures, as sketched in
fi gure 1. Locally, the amount of membrane fi ltration depends on local trans-
membrane pressure (TMP) and on the membrane’s water permeability, accord-
ing to the relation:
JUF(x) ⫽ LP (pb(x) ⫺ pd(x) ⫺ ⌸(x)) ⫽ LP (TMP(x)) (1)
where JUF is the local ultrafi ltration fl ux, LP is the membrane’s hydraulic perme-
ability, p is hydrostatic pressure, ⌸ is the blood’s oncotic pressure and x is the
axial coordinate along the device; subscripts b and d refer to the blood and dial-
ysate compartments, respectively. In the countercurrent arrangement, pressures
in the two compartments decay with opposite slopes. Hence BF may occur at
distal portions of the device if TMP becomes negative [or the total pressure on
the blood side (pb[x] ⫺ ⌸[x]) falls below the pressure on the dialysate side].
Thus, in the presence of BF, the device’s net ultrafi ltration rate QUF is due to the
difference between the DF rate QDF and the BF rate QBF:
QUF ⫽ QDF ⫺ QBF or QDF ⫽ QUF ⫹ QBF. (2)
Internal Hemodiafi ltration 179
X
X0
PressuresTMP
X
pb – ⌸
pb
pd
TMP ⬎0
TMP ⬍0
UF fluxes
X
DF
Flows
X
Dialysate
Blood
Concentrations
X
CP
Hct
DF
a
b
c
d
e
BF
Fig. 1. Schematic diagrams of relevant quantities, plotted against the axial coordinate
of the hemodialyzer, x, in the presence of IF/BF. a The pressure behavior; pb ⫽ Hydraulic
pressure in the blood compartment; pd ⫽ hydraulic pressure in the dialysate compartment; ⌸:
oncotic pressure. b The behavior of transmembrane pressure (TMP). c The behavior of ultra-
fi ltration (UF) fl uxes. d The behavior of fl uid fl ow (blood fl ows left to right; dialysate fl uid
fl ows right to left). e The behavior of blood concentrations; Hct ⫽ hematocrit; CP ⫽ plasma
protein concentration.
Fiore/Ronco 180
The term IF is associated with BF to signify that the real fi ltration behavior
occurring inside the device is not visible: the total convection amount (QDF)
may not be evaluated measuring the net QUF only.
For a zero-balance condition (QUF ⫽ 0), it is QDF ⫽ QBF. But, for a perme-
ability high enough (as for high-fl ux membranes), IF persists even when draw-
ing a net QUF (such as in fi g. 1). The overall QBF is given by:
QBF ⫽ LTMPxp x
x
L
Pf
() d
0
∫ (3)
where pf is the total fi ber perimeter along a cross-section, L is the device’s active
length and x0 is the axial position where TMP ⫽ 0 (fi g. 1). The integral on the
right-hand side of equation 3 is represented by the BF area of fi gure 1c: the
overall amount of IF/BF therefore increases with the TMP at the extremities of
the dialyzer (which in turn is infl uenced by the pressure drops along the fi lter
both in the blood and dialysate compartments), and on the membrane’s water
permeability: the amount of convection obtainable by the IF/BF mechanism is
limited by oncotic effects but may be enhanced by a wise exploitation of the
hydrodynamic pattern.
The matter of increasing the convective dose has therefore to be tackled by
properly orienting the dialyzer design in terms of membrane permeability and
hydraulic pressure drops. As for the former, the technical challenge is fi nding
a compromise between enhancing LP and maintaining an appropriate sieving
effect. High-fl ux polysulfone membranes are normally reported to have a pure-
water permeability in the range of 100–400 ml/(h ⭈ mm Hg ⭈ m 2), with peaks of
700 ml/(h ⭈ mm Hg ⭈ m 2) for enhanced high-fl ux membranes [6].
As for the pressure behavior, in the past researchers have already proposed
to enhance pressure drops by acting on the dialyzer’s design. For the dialysate
side, the inclusion of spacer yarns or path constrictors or the increase in fi ber
density were proposed [7–9]. For the blood side, fi ber diameter shrinking [7, 8]
or fi ber length increase were proposed. At a careful analysis, the latter shows to
be a safer choice for the limitation of the hemolytic potential [6].
Measurement of Internal Filtration
Quantifi cation of the convective dose during clinical iHDF is not trivial.
IF takes place within the dialyzer: at the current state of the technology, there
is no way to measure DF or BF directly (even if their difference, i.e. the net
QUF, is measured). Methods for indirect measurement have been proposed, but
their application was confi ned to in vitro studies. One chance is measuring the
local concentration of a marker molecule in the blood path [10, 11] or in the
Internal Hemodiafi ltration 181
dialysate compartment [12]. Indeed, one ‘side effect’ of the IF/BF phenomenon
is a relevant concentration/dilution behavior taking place along the dialyzer
(fi g. 1, bottom panel). Hence, when the inlet value and the peak value for a
marker molecule concentration are measured, a simple formula [10] allows one
to deduct the actual DF and/or BF rates taking place inside the device. Amounts
of convection of 1,800–3,000 ml/h were measured in vitro by such methods in
polysulfone dialyzers [10, 11]. However, the direct transfer of such methods
to the clinical use is unfeasible, either due to safety reasons (e.g. when using
radioactive or potentially harmful molecules) or to practical reasons (e.g. cum-
bersome equipment, complex procedures).
One second chance is using Doppler measurements of blood velocity
within the dialyzer fi bers [13]. The DF of blood in the proximal part of the
dialyzer causes erythrocytes to travel more slowly around the x0 section than
at the entrance (fi g. 1d), and particularly the peak decrease in blood fl ow rate
(estimated measuring the peak percent decrease in Doppler velocity) equals
DF. With this method, amounts of convection up to 2,300 ml/h were indirectly
measured in vitro in a polysulfone high-fl ux dialyzer [13]. The use of such a
method during real treatments appears more feasible; however, it is not credible
that a common hemodialysis unit would invest so much in the iHDF concept as
to be supplied with Doppler equipment and with personnel skilled for its daily
use.
Other measuring principles might be explored for indirect quantifi cation of
IF/BF with the state-of-the-art sensors. But apart from technical considerations,
it seems clear that IF/BF measurement may become a clinical practice in the
near future only if this challenge is accepted by the manufacturing companies
as an option to provide their dialysis equipment with a technologically oriented
added value.
Mathematical Estimation of Internal Filtration
An alternative approach is resorting to the mathematical quantifi cation
of convection based on a mathematical model, designed to accept clinically
measured quantities as input values and to yield the dialyzer’s fi ltration behav-
ior as its output. An extensive literature exists on the mathematical modeling
of fl uid and/or solute transport in hollow-fi ber dialyzers: a comprehensive
review was reported by Eloot et al. [14]. A sound engineering approach con-
sists in tackling the complete continuum problem by means of computational
fl uid dynamics techniques, which allow for very comprehensive, 3-dimen-
sional, non-Newtonian descriptions of local fl ow/fi ltration dynamics and rhe-
ology. However, this approach cannot be widely applied for the systematic
Fiore/Ronco 182
quantifi cation of real clinical cases, due to the technical competences, the
computational times and the costs requested by computational fl uid dynamics
analyses. At the opposite end, in the past, estimation of BF was proposed by
means of elementary calculations, based on neglecting the changes in blood
viscosity and protein concentration that take place with blood concentration
and dilution [15]. Such early attempts, justifi ed by the need to avoid BF in
a period when bad water quality was a threat, were reasonable for low-fl ux
dialyzers. With modern high-fl ux devices, having a simple analytical formula
available for IF/BF estimation is mere fancy, because neglecting oncotic lim-
itation causes greater and greater estimation errors at increasing hydraulic
permeability.
The ideal solution in today’s scenario would be to supply the clinician
with an advanced simulation tool, as easy as a pocket calculator in its use
but strong enough in its theoretical basis to yield trustworthy estimations. In
an attempt towards this direction, we have recently pursued a 1-dimensional
theoretical approach, coupled with the possibility of precharacterizing the
devices in vitro [16]. Modeling choices were driven by the aim of building a
lightweight calculation tool, usable by a clinician as a support in the quantita-
tive interpretation of practical cases and as a tool to help dialysis prescription.
A semiempirical mathematical model describing dialyzer hydrodynamics and
membrane fi ltration was hence developed, and the necessary hydraulic char-
acteristics of commercial hemodialyzers were derived experimentally with 3
hollow-fi ber polysulfone devices of different sizes. Both the model and the in
vitro characterization were implemented into an easy-to-use PC-based soft-
ware tool aimed at estimating cross-fi ltration and IF/BF as a function of the
machine settings as well as blood hematocrit and plasma protein concentra-
tion. In practice, the clinical user would run simulations related to a patient
when the hematochemical data for that patient are available, so as to esti-
mate the possible convective dose achievable for that subject with different
devices and/or settings for the dialysis machine. Prospectively, a similar tool
could be implemented in future dialysis machines with appropriate software
adaptations.
Model results obtained for an IF/BF-prone device (the Toray BS-1.8UL),
at normal operating conditions and blood parameters, showed an overall con-
vection around 2,850 ml/h which may be obtained by virtue of IF/BF at zero net
ultrafi ltration; values up to 3,600 ml/h are reached when the net QUF is increased
to 1,200 ml/h. Results also showed that options exist for the operator to enhance
convection by adjusting either blood or dialysate fl ow rates, with blood fl ow
rate achieving the greater effect. Moreover, model results showed an excellent
agreement with experimental results obtained in purposely performed in vitro
scintigraphic tests, with only a 3% prediction error.
Internal Hemodiafi ltration 183
Internal Hemodiafiltration: A Competitor of Hemodiafiltration?
Most of the considerations summarized above lead to the conclusion that
technology is ripe enough to introduce iHDF as a well-grounded technique.
The convective effect achieved cannot be compared with the high convec-
tion rates achieved in online HDF, since the amount of convection is limited by
the mechanics of the fl uids involved in the treatment and the intrinsic nature of
the hemodialyzer. But convection fi gures are high enough to possibly yield vis-
ible clinical effects, as recently reported [17].
Quantifi cation of the convective dose appears to be feasible. Even if its clin-
ical measurement is not available for now, the use of appropriate IF/BF calcula-
tors could represent a temporary bridge to a future setting in which investments
from the industry will have made the necessary technology a reality. Moreover,
mathematical modeling has a role in clarifying the mechanisms of fi ltration and
BF, making the capacity of convective transport by a specifi c device and mem-
brane fully understood and possibly exploited at best, with machine settings
used as regulation tools.
In conclusion, rather than considering iHDF a competitor for HDF, it should
be included as a possible alternative to extend the conscious clinical application
of the convective principle, with lesser potentials than HDF but with lower costs
and easier requirements.
References
1 Canaud B, Bragg-Gresham JL, Marshall MR, Desmeules S, Gillespie BW, Depner T, Klassen P,
Port FK: Mortality risk for patients receiving hemodiafi ltration versus hemodialysis: European
results from the DOPPS. Kidney Int 2006;69:2087–2093.
2 Canaud B, Morena M, Leray-Moragues H, Chalabi L, Cristol JP: Overview of clinical studies in
hemodiafi ltration: what do we need now? Hemodial Int 2006;10(suppl 1):S5–S12.
3 Penne EL, Blankestijn PJ, Bots ML, van den Dorpel MA, Grooteman MP, Nube MJ, van der Tweel
I, ter Wee PM, the CONTRAST study group: Effect of increased convective clearance by on-
line hemodiafi ltration on all cause and cardiovascular mortality in chronic hemodialysis patients
– The Dutch Convective Transport Study (CONTRAST): rationale and design of a randomised
controlled trial (ISRCTN38365125). Curr Control Trials Cardiovasc Med 2005;6:8.
4 Leypoldt JK, Schmidt B, Gurland HJ: Net ultrafi ltration may not eliminate backfi ltration during
hemodialysis with highly permeable membranes. Artif Organs 1991;15:164–170.
5 Baurmeister U, Travers M, Vienken J, Harding G, Million C, Klein E, Pass T, Wright R: Dialysate
contamination and back fi ltration may limit the use of high-fl ux dialysis membranes. ASAIO Trans
1989;35:519–522.
6 Fiore GB, Ronco C: Internal hemodiafi ltration (iHDF): a possible option to expand hemodiafi ltra-
tion therapy. Int J Artif Organs 2004;27:420–423.
7 Ronco C, Brendolan A, Lupi A, Metry G, Levin NW: Effects of a reduced inner diameter of hol-
low fi bers in hemodialyzers. Kidney Int 2000;58:809–817.
8 Mineshima M, Ishimori I, Ishida K, Hoshino T, Kaneko I, Sato Y, Agishi T, Tamamura N, Sakurai
H, Masuda T, Hattori H: Effects of internal fi ltration on the solute removal effi ciency of a dialyzer.
ASAIO J 2000;46:456–460.
Fiore/Ronco 184
9 Ronco C, Orlandini G, Brendolan A, Lupi A, La Greca G: Enhancement of convective transport
by internal fi ltration in a modifi ed experimental hemodialyzer: technical note. Kidney Int 1998;54:
979–985.
10 Ronco C, Brendolan A, Feriani M, Milan M, Conz P, Lupi A, Berto P, Bettini M, La Greca G: A
new scintigraphic method to characterize ultrafi ltration in hollow fi ber dialyzers. Kidney Int 1992;
41:1383–1393.
11 Sakai Y, Wada S, Matsumoto H, Suyama T, Ohno O, Anno I: Nondestructive evaluation of blood
fl ow in a dialyzer using X-ray computed tomography. J Artif Organs 2003;6:197–204.
12 Leypoldt JK, Schmidt B, Gurland HJ: Measurement of backfi ltration rates during hemodialysis
with highly permeable membranes. Blood Purif 1991;9:74–84.
13 Sato Y, Mineshima M, Ishimori I, Kaneko I, Akiba T, Teraoka S: Effect of hollow fi ber length
on solute removal and quantifi cation of internal fi ltration rate by Doppler ultrasound. Int J Artif
Organs 2003;26:129–134.
14 Eloot S, De Wachter D, Van Tricht I, Verdonck P: Computational fl ow modeling in hollow-fi ber
dialyzers. Artif Organs 2002;26:590–599.
15 Soltys PJ, Ofsthun NJ, Leypoldt JK: Critical analysis of formulas for estimating backfi ltration in
hemodialysis. Blood Purif 1992;10:326–332.
16 Fiore GB, Guadagni G, Lupi A, Ricci Z, Ronco C: A new semiempirical mathematical model for
prediction of internal fi ltration in hollow fi ber hemodialyzers. Blood Purif 2006;24:555–568.
17 Lucchi L, Fiore GB, Guadagni G, Perrone S, Malaguti V, Caruso F, Fumero R, Albertazzi A:
Clinical evaluation of internal hemodiafi ltration (iHDF): a diffusive-convective technique per-
formed with internal fi ltration enhanced high-fl ux dialyzers. Int J Artif Organs 2004;27:414–419.
Gianfranco B. Fiore, PhD
Dipartimento di Bioingegneria, Politecnico di Milano
Piazza Leonardo da Vinci, 32
IT–20133 Milano (Italy)
Tel. ⫹39 02 2399 3337, Fax ⫹39 02 2399 3360, E-Mail gianfranco.fi ore@polimi.it
Ronco C, Canaud B, Aljama P (eds): Hemodiafi ltration.
Contrib Nephrol. Basel, Karger, 2007, vol 158, pp 185–193
Clinical Aspects of Haemodiafiltration
Francesco Locatelli, Salvatore Di Filippo, Celestina Manzoni
Department of Nephrology and Dialysis, A. Manzoni Hospital, Lecco, Italy
Abstract
Standard haemodialysis is not a very effi cacious treatment of chronic uraemia and
patient mortality rate is still very high. The 2002 results of the HEMO study showed that
alternative treatments such as ‘high-effi ciency haemodialysis’ and ‘high-fl ux haemodialysis’
are associated with a non-signifi cant reduction in the relative risk of mortality (4 and 8%,
respectively). In an attempt to defi ne the clinical impact of haemodiafi ltration, we review
some of the effi cacy data from clinical studies in light of a number of factors that may be
related to the high mortality among haemodialysis patients.
Copyright © 2007 S. Karger AG, Basel
Uraemia is a pathological condition caused by the retention of solutes that
are normally excreted by the kidneys. The aim of haemodialysis (HD) is to
remove these solutes, but standard HD is not very effi cacious, and patient mor-
bidity and mortality rates are still very high (15–25% per year).
Nearly 20 years ago, the hypothesis that the extremely high morbidity
and mortality rates were associated with inadequate removal of ‘middle mol-
ecules’ led to proposals for two alternative methods: high-effi ciency HD [1]
and high-fl ux HD [2]. At the same blood and dialysate fl ows, and using mem-
branes that have the same low permeability as those used for standard HD
but a larger surface area, high-effi ciency HD increases vitamin B12 clearance
by about 50% in vitro; high-fl ux HD uses high-permeability membranes that
increase the clearance of solutes with a molecular weight of about 1,500 Da
and also remove solutes with a molecular weight of about 11,000 Da, such as
2-microglobulin.
Observational studies have consistently shown that high-fl ux treatments
have positive effects on the morbidity and survival of HD patients. However, the
2002 results of the HEMO study [3], a prospective, randomized study aimed at
Clinical Aspects of Hemodiafiltration
Locatelli/Di Filippo/Manzoni 186
Table 1. Cardiovascular risk factors in chronic kidney
disease
Traditional risk factors Non-traditional risk factors
Older age Albuminuria/proteinuria
Male gender Homocysteine
Hypertension Anaemia
Higher low-density
lipoprotein cholesterol
levels
Abnormal calcium-
phosphate metabolism
Low high-density
lipoprotein cholesterol
levels
Extracellular fl uid overload
Diabetes Oxidative stress
Smoking Infl ammation
Physical inactivity Malnutrition
Menopause
Family history of
cerebrovascular disease
Left ventricular hypertrophy
verifying the advantages of high-effi ciency and high-fl ux HD over standard HD,
were very surprising and in some way disappointing insofar as they showed that
greater urea removal non-signifi cantly reduces the relative risk of mortality by
only 4% and that high-fl ux HD was associated with a non-signifi cant reduction
of 8%. The major criticisms of the HEMO study design were that it included
prevalent instead of only incident patients, reused dialysers and used high-fl ux
dialysers with low convective clearances.
The preliminary results of the Membrane Permeability Outcome (MPO)
study [4] seem to be very promising in terms of the fi rst 2 aspects (on the basis
of a presentation by the principal investigator at the investigators’ meeting); as
far as the third is concerned, it is possible to increase convection by means of
haemodiafi ltration (HDF) as Schneider and Streicher [5] have already shown
that, even when using the same membrane, it signifi cantly increases middle-
molecules clearance in comparison with HD.
In an attempt to defi ne the clinical impact of HDF, we will review some
of the effi cacy data from clinical studies in the light of a number of factors
that may be related to the high mortality among HD patients. It is well known
that cardiovascular disease is the major cause of death, and we will analyse
the impact of HDF on some of the main cardiovascular risk factors listed in
table 1.
Clinical Aspects of Haemodiafi ltration 187
Hyperphosphataemia
Hyperphosphataemia has been associated with an increased risk of all-
cause mortality, including cardiovascular mortality [6].
Zehnder et al. [7] compared the transmembrane solute mass removal and
clearance of phosphate in 16 patients who underwent high-fl ux HD for 1 week
followed by 1 week’s postdilutional online HDF. The results strongly suggested
that HDF increases phosphate clearance, and the authors concluded that it
should be considered an additional treatment option for dialysis patients with
uncontrolled hyperphosphataemia. However, because it was so short, this study
does not give any information concerning the possible difference in long-term
predialysis phosphataemia levels of the two treatments.
Anaemia
Together with hypertension, anaemia is the main cause of ventricular
hypertrophy in dialysis patients.
Maduell et al. [8] evaluated the difference between conventional HDF (mean
fl uid replacement 4 l/session), in which the extent of convection is roughly com-
parable with that of high-fl ux HD, and online HDF (mean fl uid replacement
22.5 l/session) in 37 patients over a period of 1 year. The most interesting result
was that online HDF led to the better correction of anaemia with lower eryth-
ropoietin doses, possibly because the greater elimination of medium-sized mol-
ecules reduced the erythropoietin response, although the role of the better quality
of dialysate due to online treatment cannot be ruled out. This possibility is further
suggested by the results of a study by Schiffl et al. [9] which clearly support the
hypothesis that the use of ultrapure (fi ltered, pyrogen-free and sterile) dialysate
reduces the recombinant human erythropoietin doses required to maintain hae-
moglobin levels as a result of a reduction in systemic infl ammatory processes.
Cardiovascular Stability
Cardiovascular instability is the most frequent clinical problem that occurs
during both acute and long-term HD. Preventing intradialytic hypotension is
of great importance not only to deliver an adequate dialysis dose, but also to
achieve the target patient dry body weight, as hypertension in dialysis patients
is largely due to fl uid overload.
A retrospective study by Pizzarelli et al. [10] compared the results obtained
during online HDF with those obtained during standard bicarbonate HD. Online
Locatelli/Di Filippo/Manzoni 188
HDF led to a better cardiovascular tolerance to fl uid removal, with a signifi -
cantly lower incidence of episodes of symptomatic hypotension requiring the
administration of saline and/or hypertonic solutions.
A prospective, randomized trial by Lin et al. [11] also found that online
HDF led to better haemodynamic stability. They treated 111 patients, who were
randomly divided into 4 groups receiving different frequencies of online HDF
and/or high-fl ux HD: HDF 3 times a week; HDF twice and high-fl ux HD once
a week; HDF once and high-fl ux HD twice a week; high-fl ux HD 3 times a
week. There were fewer episodes of symptomatic hypotension and lower mean
saline infusion volumes at greater frequencies of online HDF, which also sig-
nifi cantly reduced the amount of erythropoietin required, and improved intra-
and interdialysis symptoms. It is interesting that higher predialysis natraemia
levels (2.3 mEq/l) were observed in the patients receiving more frequent online
HDF, thus suggesting that a reduced sodium removal during HDF was at least
partially responsible for the better cardiovascular stability. The same is true for
the results of Maduell et al. [8].
Altieri et al. [12] compared the effects of online haemofi ltration and online
HDF on cardiovascular stability and blood pressure in a randomized trial involv-
ing 39 patients and concluded that both treatments allow the good control of
intrasession symptoms and blood pressure in stable patients.
According to the original observation by Maggiore [13] that a better
haemodynamic protection is provided by a dialysate temperature of about 35°C
in comparison with the standard dialysate temperature of 37–38°C, an alterna-
tive hypothesis to explain the decrease in hypotensive episodes during online
HDF has been suggested by Donauer et al. [14], who identifi ed blood cooling
as the main factor. They found that enhanced energy loss occurred within the
extracorporeal system despite identical dialysate and substitution fl uid tempera-
ture settings, which meant that the blood returning to the patient was cooler
during online HDF than during HD. The use of cooler, temperature-controlled
HD led to an incidence of symptomatic hypotension that was similar to that
observed during online HDF.
2-Microglobulin
In order to verify the impact of online HDF, Wizemann et al. [15] conducted
a 24-month controlled prospective study in which 44 chronic dialysis patients
were randomized to low-fl ux HD or online HDF. There were no differences in
morbidity, blood pressure, dialysis-associated hypotensive episodes, haematocrit
or erythropoietin dose between the groups, nor any differences in body weight
and nutrition parameters. As expected, plasma 2-microglobulin concentrations
Clinical Aspects of Haemodiafi ltration 189
did not change in the HD group throughout the 2 years but decreased from simi-
lar values to 18 mg/l before dialysis (p ⬍ 0.01) during the fi rst 6 months of HDF
treatment, and then remained constant until the end of the study. However, it is
possible that the clinical reversal of the situation by convective methods takes a
long time, including the effects of a reduction in 2-microglobulin levels.
Ward et al. [16] carried out a prospective clinical trial involving 44 patients
randomized to online postdilution HDF or high-fl ux HD for 12 months, and
found a similar decrease in pretreatment plasma 2-microglobulin levels despite
the apparent difference in the removal of 2-microglobulin indicated by the sig-
nifi cantly greater pre- to posttreatment reduction in the HDF group. It should
be remembered that a change in the concentration of a solute is a good indica-
tor of removal only in the case of solutes distributed in a single pool including
plasma, whereas the fact that a substantial rebound in posttreatment plasma
2-microglobulin levels has been reported suggests that a single-pool model is
inadequate to describe 2-microglobulin kinetics. Intrabody mass transfer rates
limit 2-microglobulin removal, and so the pre- to posttreatment change in con-
centration overestimates the actual removal.
Emerging Cardiovascular Risk Factors
It has been shown that hyperhomocysteinaemia is independently associ-
ated with an increase in cardiovascular risk in dialysis patients.
Arnadottir et al. [17] studied the effects of standard HD on the plasma
concentrations of total homocysteine (tHcy) and creatinine in 56 patients, and
found that the dialysis-induced reduction in tHcy was less than the reduction
in creatinine, despite their similar molecular weight and distribution volumes
(0.45 l/kg [18] and 0.48 l/kg [19]). An alternative explanation is that tHcy is par-
tially protein bound and the bound fraction cannot be removed by either low- or
high-fl ux HD; furthermore, as the problem lies in the permeability of the mem-
brane, it cannot be expected that different methods will lead to different results.
Support for this hypothesis comes from a 3-month randomized trial by House
et al. [20], who examined the effect of maintenance HD with high-fl ux polysul-
phone versus low-fl ux polysulphone on predialysis tHcy levels in 48 patients.
More permeable superfl ux dialysers designed to maximize convective transport
are the only ones capable of signifi cantly removing tHcy [21].
Chronic infl ammation and oxidative stress are highly prevalent in patients
with chronic kidney disease and end-stage renal disease, and may contribute to
the high mortality rates associated with cardiovascular disease [22].
In addition, advanced glycation end products (AGEs) may represent
a novel class of uraemic toxins with signifi cant implications for long-term
Locatelli/Di Filippo/Manzoni 190
dialysis-related pathological states. A study by Lin et al. [23] analysed long-
term changes in serum AGE levels using various dialysis modalities. Eighty-one
patients with chronic uraemia were divided into 3 groups receiving conventional
HD, high-fl ux HD or online HDF. During the 6-month study period, predialysis
serum AGE levels were signifi cantly lower in the patients treated with online
HDF than in those treated with conventional or high-fl ux HD. In line with this,
Gerdemann et al. [24] found that predialysis AGE levels in patients treated with
HDF and haemofi ltration are signifi cantly lower than those of patients treated
with high-fl ux HD using standard dialysis fl uid, but the difference was not sig-
nifi cant when ultrapure dialysis fl uid was used. This suggests that factors other
than removal (including ultrapure dialysis fl uid and water quality) are respon-
sible for the lower pretreatment AGE levels found in patients treated with HDF
compared to HD.
Survival
It is a matter of fact that survival, together with the quality of life, is the
most important outcome.
The characteristics and outcomes of patients from 5 European countries
receiving HDF or HD in the Dialysis Outcomes and Practice Patterns Study
[25] were published in 2006. From 1998 to 2001, the study analysed 2,165
patients stratifi ed into 4 groups: low- and high-fl ux HD (63.1 and 25.2% of all
patients), and low- and high-effi ciency HDF (7.2 and 4.5% of all patients). The
patients undergoing high-effi ciency HDF had a 35% lower relative mortality
risk (⫽ 0.65; p ⫽ 0.01) than those receiving low-fl ux HD, whereas those on low-
effi ciency HDF showed a non-signifi cant 7% reduction (relative risk ⫽ 0.93;
p ⫽ 0.68). These results are very impressive but only demonstrate an associa-
tion: as the authors themselves acknowledged, the benefi ts of HDF must be
tested in controlled clinical trials before any recommendations can be made for
clinical practice.
This is particularly true when considering discrepancies between the
results of observational and randomized studies. One prospective rand-
omized trial [26] involving 380 patients compared low-fl ux HD, high-fl ux
HD and HDF in order to evaluate possible advantages in terms of treatment
tolerance, nutritional parameters and pretreatment 2-microglobulin levels.
However, it was not found that convection and/or membrane biocompati-
bility improved cardiovascular stability during dialysis, mainly because the
incidence of intradialytic hypotension was very low in the study population
as a whole, i.e. the study was underpowered to fi nd possible differences.
Moreover, the same trial did not fi nd any difference in survival related to
Clinical Aspects of Haemodiafi ltration 191
membrane biocompatibility or fl ux, but it was not primarily designed and
powered for that.
An observational study by Hornberger et al. [27] showed that patients
treated by high-fl ux HD had a 65% lower relative risk of mortality than those
treated with standard HD, and other observational studies indicate that HD with
high-fl ux dialysers is associated with less morbidity and mortality than HD with
low-fl ux dialysers. In a large observational study comparing convective and dif-
fusive treatments, a non-signifi cant 10% better survival rate was observed in
favour of convective treatments [28]. The very large, prospective randomized
HEMO study of fl ux and survival found a non-signifi cant 8% better survival
rate in the patients treated with high-fl ux membranes, although a statistical
reduction in cardiovascular morbidity in favour of high-fl ux dialysis was found
in a post hoc analysis.
A systematic review of randomized controlled trials comparing HD,
haemofi ltration, HDF and acetate-free biofi ltration to assess their clinical
effectiveness has been published [29], but, because the assessed trials were
not powered adequately and their methodological quality was suboptimal, no
defi nite conclusions can be drawn as to which is the best replacement therapy
[30].
As the number of randomized prospective trials comparing HDF and
standard HD is still limited, no conclusive data are available concerning the
effect of HDF on survival and morbidity in patients with end-stage renal
disease. However, 2 further studies are currently exploring the potential ben-
efi cial effect of convection. An Italian prospective multicentre study [31]
is comparing online convective treatments (haemofi ltration and HDF) with
standard low-fl ux HD, taking cardiovascular stability and blood pressure con-
trol as the primary end points, and impact on symptoms, morbidity and mor-
tality as the secondary end points. The Dutch Convective Transport Study,
which was started in the second quarter of 2004 [32], is being conducted in
more than 20 centres in the Netherlands and will randomize approximately
800 incident and prevalent HD patients to low-fl ux HD or online HDF for
3 years in order to investigate the effect of increased convective transport
by online HDF on all-cause and cardiovascular mortality in chronic HD
patients.
At present, some results of the HEMO study (decreased cardiac mortality)
[33] and the preliminary data of the MPO study (principal investigator presen-
tation at MPO investigator meeting [4]) with a signifi cant reduction in mortality
in patients with serum albumin levels of less than 4 g/dl (primary end point) as
well as in diabetic patients (secondary analysis) make a strong case in favour
of high-fl ux treatments. A large, randomized controlled study is now needed to
evaluate the clinical advantages of online HDF.
Locatelli/Di Filippo/Manzoni 192
References
1 Keshaviah P, Collins A: Rapid high-effi ciency bicarbonate hemodialysis. Trans Am Soc Artif
Intern Organs 1986;32:17.
2 Von Albertini B, Miller JH, Gardner PW, Shinaberger JH: High-fl ux hemodiafi ltration: under six
hours/week treatment. Trans Am Soc Artif Intern Organs 1984;30:227–231.
3 Eknoyan G, Beck GJ, Cheung AK, Daugirdas JT, Greene T, et al, for the Hemodialysis (HEMO)
Study Group: Effect of dialysis dose and membrane fl ux in maintenance hemodialysis. N Engl J
Med 2002;347:2010–2019.
4 Locatelli F, Hannedouche T, Jacobson SH, La Greca G, Loureiro A, Martin-Malo A, Papadimitriou
M, Vanholder R: The effect of membrane permeability on ESRD: design of a prospective ran-
domised multicentre trial. J Nephrol 1999;12:85–88.
5 Schneider H, Streicher E: Mass transfer characterization of a new polysulfone membrane. Artif
Organs 1985;2:180–183.
6 Block GA, Klassen PS, Lazarus JM, Ofsthun N, Lowrie EG, Chertow GM: Mineral metabolism,
mortality and morbidity in maintenance hemodialysis. J Am Soc Nephrol 2004;15:2208–2218.
7 Zehnder C, Gutzwiller JP, Renggli K: Hemodiafi ltration: a new treatment option for hyperphos-
phatemia in hemodialysis patients. Clin Nephrol 1999;52:152–159.
8 Maduell F, del Pozo C, Garcia H, Sanchez L, Hdez-Jaras J, Albero MD, Calvo C, Torregrosa I,
Navarro V: Change from conventional haemodiafi ltration to on-line haemodiafi ltration. Nephrol
Dial Transplant 1999;14:1202–1207.
9 Schiffl H, Lang SM, Bergner A: Ultrapure dialysate reduces dose of recombinant human erythro-
poietin. Nephron 1999;83:278–279.
10 Pizzarelli F, Cerrai T, Dattolo P, Tetta C, Maggiore Q: Convective treatments with on-line produc-
tion of replacement fl uid: a clinical experience lasting 6 years. Nephrol Dial Transplant 1998;13:
363–369.
11 Lin CL, Huang CC, Chang CT, Wu MS, Hung CC, Chien CC, Yang CW: Clinical improvement by
increased frequency of on-line hemodiafi ltration. Ren Fail 2001;23:193–206.
12 Altieri P, Sorba G, Bolasco P, Ledebo I, Ganadu M, Ferrara R, Menneas A, Asproni E, Casu D,
Passaghe M, Sau G, Cadinu F, Sardinian Group on Hemofi ltration On-Line: Comparison between
hemofi ltration and hemodiafi ltration in a long-term prospective cross-over study. J Nephrol 2004;
17:414–422
13 Maggiore Q: Blood temperature and vascular stability during hemodialysis and hemofi ltration.
Trans Am Soc Artif Organs 1982;28:523–537.
14 Donauer J, Schweiger C, Rumberger B, Krumme B, Bohler J: Reduction of hypotensive side effects
during online-haemodiafi ltration and low temperature haemodialysis. Nephrol Dial Transplant
2003;18:1616–1622.
15 Wizemann V, Lotz C, Techert F, Uthoff S: On-line haemodiafi ltration versus low-fl ux haemodialy-
sis: a prospective randomized study. Nephrol Dial Transplant 2000;15(suppl 1):43–48.
16 Ward RA, Schmidt B, Hullin J, Hillebrand GF, Samtleben W: A comparison of on-line hemodia-
fi ltration and high-fl ux hemodialysis: a prospective clinical study. J Am Soc Nephrol 2000;11:
2344–2350.
17 Arnadottir M, Berg AL, Hegbrant J, Hultberg B: Infl uence of haemodialysis on plasma total homo-
cysteine concentration. Nephrol Dial Transplant 1999;14:142–146.
18 Guttormsen AB, Ueland PM, Svarstad E, Refsum H: Kinetic basis of hyperhomocysteinemia in
patients with chronic renal failure. Kidney Int 1997;52:495–502.
19 Jones JD, Burnett PC: Creatinine metabolism in humans with decreased renal function: creatinine
defi cit. Clin Chem 1974;20:1204–1212.
20 House AA, Wells GA, Donnelly JG, Nadler SP, Hébert PC: Randomized trial of high-fl ux vs
low-fl ux haemodialysis: effects on homocysteine and lipids. Nephrol Dial Transplant 2000;15:
1029–1034.
21 Van Tellingen A, Grooteman MPC, Bartels PCM, Van Limbeek J, Van Guldener C, ter Wee PM,
Nubé MJ: Long-term reduction of plasma homocysteine levels by super-fl ux dialyzers in hemodi-
alysis patients. Kidney Int 2001;59:342–347.
Clinical Aspects of Haemodiafi ltration 193
22 Locatelli F, Andrulli S, Memoli B, Maffei C, Del Vecchio L, Aterini S, De Simone W, Mandalari
A, Brunori G, Amato M, Cianciaruso B, Zoccali C: Nutritional-infl ammation status and resistance
to erythropoietin therapy in hemodialysis patients. Nephrol Dial Transplant 2006;21:991–998.
23 Lin CL, Huang CC, Yu CC, Yang HY, Chuang FR, Yang CW: Reduction of advanced glycation end
product levels by on-line hemodiafi ltration in long-term hemodialysis patients. Am J Kidney Dis
2003;42:524–531.
24 Gerdemann A, Wagner Z, Solf A, Bahner U, Heidland A, Vienken J, Schinzel R: Plasma levels of
advanced glycation end products during haemodialysis, haemodiafi ltration and haemofi ltration:
potential importance of dialysate quality. Nephrol Dial Transplant 2002;17:1045–1049.
25 Canaud B, Bragg-Gresham JL, Marshall MR, Desmeules S, Gillespie BW, Depner T, Klassen P,
Port FK: Mortality risk for patients receiving hemodiafi ltration versus hemodialysis: European
results from the DOPPS. Kidney Int 2006;69:2087–2093.
26 Locatelli F, Mastrangelo F, Redaelli B, Ronco C, Marcelli D, La Greca G, Orlandini G, the Italian
Cooperative Dialysis Study Group: Effects of different membranes and dialysis technologies on
patient treatment tolerance and nutritional parameters. Kidney Int 1996;50:1293–1302.
27 Hornberger JC, Chernew M, Petersen J, Garber AM: A multivariate analysis of mortality and hos-
pital admission with high-fl ux dialysis. J Am Soc Nephrol 1992;3:1227–1237.
28 Locatelli F, Marcelli D, Conte F, Limido A, Malberti F, Spotti D: Comparison of mortality in
ESRD patients on convective and diffusive extracorporeal treatments. The Registro Lombardo
Dialisi e Trapianto. Kidney Int 1999;55:286–293.
29 Rabindranath KS, Strippoli GF, Roderick P, Wallace SA, MacLeod AM, Daly C: Comparison of
hemodialysis, hemofi ltration and acetate-free biofi ltration for ESRD: systematic review. Am J
Kidney Dis 2005;45:437–447.
30 Locatelli F: Comparison of hemodialysis, hemodiafi ltration and hemofi ltration: systematic review
or systematic error? Am J Kidney Dis 2005;46:787–788.
31 Bolasco P, Altieri P, Andrulli S, Basile C, Di Filippo S, Feriani M, Pedrini L, Santoro A, Zoccali
C, Sau G, Locatelli F: Convection versus diffusion in dialysis: an Italian prospective multicentre
study. Nephrol Dial Transplant 2003;18(suppl 7):50–54.
32 Penne EL, Blankestijn PJ, Bots ML, Van den Dorpel MA, Grooteman MPC, Nubé MJ, ter Wee
PM, on behalf of the CONTRAST Group: Resolving controversies regarding hemodiafi ltration
versus hemodialysis: the Dutch Convective Transport Study. Semin Dial 2005;18:47–51.
33 Cheung AK, Sarnak MJ, Yan G, Berkoben M, Heyka R, et al: Cardiac diseases in maintenance
hemodialysis patients: results of the HEMO Study. Kidney Int 2004;65:2380–2389.
Prof. Francesco Locatelli, MD, FRCP
Divisione di Nefrologia e Dialisi, Azienda ospedaliera A. Manzoni
Via dell’Eremo 9–11
IT–23900 Lecco (Italy)
Tel. ⫹39 034 148 9850, Fax ⫹39 034 148 9860, E-Mail f.locatelli@ospedale.lecco.it
Ronco C, Canaud B, Aljama P (eds): Hemodiafi ltration.
Contrib Nephrol. Basel, Karger, 2007, vol 158, pp 194–200
The Biological Response to Online
Hemodiafiltration
V. Pa n i c h i a, C. Tettab
aInternal Medicine Department, University of Pisa, Pisa, Italy; bFresenius Medical
Care, International Research and Development, Bad Homburg, Germany
Abstract
The biologic response to uremia and to the associated chronic infl ammation is an active
area of research. Among the different modalities developed in the technology fi eld of chronic
renal replacement, hemodialfi ltration has evolved consistently. On-line production of substi-
tution fl uid by ‘cold sterilization’ of dialysis fl uid by ultrafi ltration gives access to virtually an
unlimited amount of sterile and non-pyrogenic intravenous grade solution. Today, on line
HDF is already a widespread, accepted treatment. Here, we will review the main mechanisms
through which on line hemodialfi ltration acts and the biological response observed in rela-
tion to the immune system dysfunction and the anemia associated to chronic kidney disease.
Copyright © 2007 S. Karger AG, Basel
Chronic kidney disease represents a complex mosaic of interwoven altera-
tions at several levels: the cell, the microenvironment, the vasculature and the
organ. Anemia, atherosclerosis and cardiovascular disease, immune dysfunc-
tion and alterations in bone metabolism are the clinical hallmarks. The biologi-
cal response to uremia and to renal replacement therapy is at present an active
area of research. The uremic syndrome is characterized by retention of solutes
over a very large range of molecular weights. Despite its being elusive for sev-
eral years, the uremic syndrome is the target of a major effort by the European
Society for Artifi cial Organs, Uremic Toxins Group, to categorize uremic ‘tox-
ins’ on the basis of molecular weight, protein binding and biological activity
[1, 2]. These uremic retention solutes can be conveniently grouped into small
water-soluble compounds of low molecular weight (less than 500 Da), middle
molecules (greater than 500 Da), protein-bound solutes and low-molecular-
weight proteins (5–35 kDa). Increasingly, evidence has been accumulating that
both middle molecules (such as 2-microglobulin) and largely protein-bound
Biological Response to Online Hemodiafi ltration 195
solutes (such as p-cresol and advanced glycosylation end products) may be
important mediators of uremic toxicity. Standard hemodialysis (HD), which
relies on diffusive solute clearance, has at best a moderate effect on removal of
these larger-molecular-weight uremic retention products.
In recent years, it has become clear that the most important result of ure-
mic toxicity not fully corrected by HD is vascular damage, characterized by
the extraordinarily high rate of cardiovascular events in chronic kidney disease
patients [3] and in the HD population [4]. While the pathogenesis of cardiovas-
cular risk associated with uremia is not fully understood, it is clear that at least
a component of this risk is due to ‘nontraditional’ cardiovascular risk factors,
which include acute-phase infl ammation, endothelial dysfunction, oxidative
stress and insulin resistance [5–8]. Despite continuous technical improvement
and better global patient care management, the annual mortality rate of patients
with end-stage renal disease managed with 3 times weekly HD remains unac-
ceptably high (10–22%) [9, 10], Factors affecting mortality include advanced
age and comorbid conditions at the start of dialysis [11], the effi cacy and quality
of renal replacement therapy [12], practice patterns that may vary from region
to region [13], different background atherosclerosis in the general population
[14] and the intensity of chronic systemic infl ammation. Chronic infl ammation
may play an important role in early morbidity and mortality in HD patients [15–
18]. Several studies have attempted to address the question whether the type of
the dialysis membrane, the quality of the dialysate or the uremic state may be
responsible for the induction of a chronic infl ammatory state [19–21].
It is known that among other nontraditional risk factors, the acute-phase
reactants represent a class of proteins – mainly C-reactive protein and serum
amyloid A – that are secreted primarily by hepatocytes under various appropriate
stimuli such as interleukin 6 (IL-6). They are not merely biochemical markers of
infl ammation, but also act as modulators of the infl ammatory response [22].
For years, standard low-fl ux HD has been the only treatment available. The
introduction of high-permeability membranes has paved the way to high-fl ux
HD and to hemodiafi ltration (HDF).
Since the 80s there has been a steady growth in the technical development
and clinical appraisal of HDF. Online production of substitution fl uid by cold
sterilization of dialysis fl uid by ultrafi ltration gives access to virtually an unlim-
ited amount of sterile and nonpyrogenic intravenous-grade solution. Today,
online HDF (OL-HDF) is already a widespread, accepted treatment. It is also
the dialysis modality for which the most technology and inventiveness have
been produced [23].
At present, HDF is a well-recognized treatment modality that offers a
means of optimizing renal replacement therapy in chronic kidney disease
patients [24]: by enhancing and enlarging the molecular-weight spectrum of
Panichi/Tetta 196
uremic toxins removed, the convective clearance improves dialysis effi ciency
[25, 26]; by increasing the instantaneous solute fl ux of solutes including ions,
HDF facilitates the restoration of the internal milieu; by improving the global
hemocompatibility of the dialysis system (synthetic low-reactive membrane,
ultrapurity of the dialysis fl uid, protein coating of the membrane), HDF con-
tributes to reducing the side effects and complications of long-term standard
HD [for a review, see 27]. The incorporation of the infusion module into the
dialysis proportioning machine hardware is also benefi cial: fi rst, it simplifi es
the handling procedure compared to bag HDF; second, it secures the process by
enslaving the infusion module to the safety regulation of the HDF monitor, and
third, it allows the physical integrity of the ultrafi lters to be checked regularly
by means of a built-in air pressure test. This low-cost production of substitution
fl uid has allowed the excellent convective and diffusive clearances of today’s
dialyzers to be exploited.
Parallel to further improvement in technical optimization for ever better effi -
ciency, a great deal of published evidence has shown that OL-HDF induces a bio-
logical response from the host. This is intriguing, since it may be the link between
some of the reported clinical effects. How strong this link might be is in general
not easy to state, simply because of the fact that studies on the biological response
were forcedly performed on relatively small patient cohorts. However, it has been
clearly shown that from basic knowledge, the later extension of the determination
of biomarkers, e.g. C-reactive protein, in large patient cohorts has spurred new
trends in predictive risk analysis as it has been the case for the impact of chronic
infl ammation on overall mortality and that from cardiovascular disease. Figure 1
illustrates the possible aspects where convective-diffusive treatments could pro-
vide signifi cant changes in patients with chronic kidney disease. We will focus on
some fi ndings that could provide the rational bases for future studies.
Immune Dysfunction and Online Hemodiafiltration
Studies on the ability of OL-HDF to modulate the immune response and
systemic, chronic infl ammation have disclosed an important potential for this
technique. Ward et al. [26] and Beerenhout et al. [28] observed a reduction in
plasma levels of complement D by postdilution HDF and predilution hemo-
fi ltration, respectively. Although the link to a clinically relevant outcome has
yet to be shown, it should be remembered that complement D is a stimulant
of the alternative route of complement and an inhibitor of the degranulation of
polymorphonuclear leukocytes. More recent observations come from in-depth
studies on the modulating role of OL-HDF on cells of the monocytic lineage.
Circulating monocytes are heterogeneous in normal individuals. However, in
Biological Response to Online Hemodiafi ltration 197
infl amed states such as in HD, the CD14⫹CD16⫹ subpopulation markedly
increases, and this correlates with C-reactive protein levels. In a prospec-
tive, crossover study, Carracedo et al. [29] demonstrated that compared with
high-fl ux HD, OL-HDF markedly reduced the number of proinfl ammatory
CD14⫹CD16⫹ cells and the production of tumor necrosis factor ␣ and IL-6.
Future studies are needed to assess the possible therapeutic effect of convective
transport on chronic infl ammation that is associated with HD. In a subsequent
study from the same group [30], the authors showed that the CD14⫹CD16⫹
subpopulation was correlated with the number of endothelial progenitor cells
and microparticles. OL-HDF was able to reduce the levels of endothelial pro-
genitor cells and microparticles and this reduction was paralleled by a signifi cant
decrease in the CD14⫹CD16⫹ subpopulation. OL-HDF attenuates endothelial
dysfunction possibly by decreasing chronic infl ammation. This effect may be
directly caused by a modulatory effect of OL-HDF on proinfl ammatory cells or
by a complex interaction that encompasses a wider removal of uremic toxins.
Microinflammation
Uremia Membrane
Comorbid state Dialysate contamination
ROS
production
AntiOxidative
defense
Oxidative stress
Therapy
–
–
–
X
–
Fig. 1. Biological response to OL-HDF. Attenuation of microinfl ammation may occur
due to the improved removal of retention solutes and of the uremic microenvironment, to
improved biocompatibility of the membrane and the use of ultrapure reinfusion fl uids lead-
ing to reduced monocyte activation and radical oxygen generation and fi nally to innovative
OL-HDF monitors for absolute patient safety and long-term patient surveillance. ROS ⫽
Reactive oxygen species.
Panichi/Tetta 198
Anemia and Online Hemodiafiltration
In HD patients, erythropoietin resistance may be caused either by absorp-
tion of recombinant human erythropoietin by the membrane or an increased
release of cytokines that inhibit erythropoiesis, such as IL-1, tumor necrosis
factor ␣ and ␥-interferon, or by a decrease in stimulatory cytokines such as IL-
3, IL-6 and IL-10 [for a review, see 31]. These negative phenomena are reversed
by the use of biocompatible dialysis techniques such as HDF. Other possible
pathophysiological mechanisms for an improved regulation of anemia with
OL-HDF have not yet been fully established. Uremic serum inhibits erythro-
poiesis, possibly by accumulation of protein-bound polyamines such as sperm-
ine or spermidine or the tetrapeptide mdN-acetyl-seryl-aspartyl-lysyl-proline
(487 Da). Theoretically, protein-bound solutes are better removed during OL-
HDF. It would be interesting to compare the effects of ultrafi ltrate obtained
by HD and HDF on erythropoiesis, and the effect of convective techniques on
these potential inhibitors of erythropoiesis. To the best of our knowledge, such
a study has not yet been performed.
In recent years, evidence has accumulated for the role of infl ammatory
cytokines in the inhibition of erythropoiesis in the anemia of HD patients. The
most important mechanism for cytokine-induced anemia is the suppression of
bone marrow erythropoiesis, but the extent to which increased cytokine lev-
els and acute-phase response may contribute to resistance to erythropoietin-
stimulating agent treatment is still not clear. Erythroid colony-forming units are
inhibited by soluble factors in the sera from uremic patients. However, available
results are confl icting, mainly because of differences in treatment modalities or
membranes, lack of control groups and small numbers of enrolled patients.
Conclusions
There is accruing evidence for a biological response to OL-HDF. The
increased awareness of its multiple advantages and technical advances have led
to its widespread use. More studies will in the future provide the rational bases
for its application in sicker and older patients.
References
1 Vanholder R, De Smet R, Glorieux G, et al: Review on uremic toxins: classifi cation, concentra-
tion, and interindividual variability. Kidney Int 2003;63:1934–1943.
2 Vanholder R, De Smet R, Lameire N: Protein-bound uremic solutes: the forgotten toxins. Kidney
Int 2001;78(suppl):S266–S270.
Biological Response to Online Hemodiafi ltration 199
3 Levey AS, Coresh J, Balk E, et al: National Kidney Foundation practice guidelines for chronic
kidney disease: evaluation, classifi cation, and stratifi cation. Ann Intern Med 2003;139:137–147.
4 Stenvinkel P, Pecoits-Filho R, Lindholm B: Coronary artery disease in end-stage renal disease: no
longer a simple plumbing problem. J Am Soc Nephrol 2003;14:1927–1939.
5 Zoccali C, Tripepi G, Mallamaci F: Predictors of cardiovascular death in ESRD. Semin Nephrol
2005;25:358–362.
6 Himmelfarb J, Ikizler TA, Stenvinkel P, Hakim RM: The elephant in uremia: oxidant stress as a
unifying concept of cardiovascular disease in uremia. Kidney Int 2002;62:1524–1538.
7 Morena M, Delbosc S, Dupuy AM, et al: Overproduction of reactive oxygen species in end-stage
renal disease patients: a potential component of hemodialysis-associated infl ammation. Hemodial
Int 2005;9:37–46.
8 Brancaccio D, Tetta C, Gallieni M, Panichi V: Infl ammation, CRP, calcium overload and a high
calcium-phosphate product: a ‘liaison dangereuse’. Nephrol Dial Transplant 2002;17:201–203.
9 Rayner HC, Pisoni RL, Bommer J, Canaud B, Hecking E, Locatelli F, et al: Mortality and hospital-
ization in haemodialysis patients in fi ve European countries: results from the Dialysis Outcomes
and Practice Patterns Study (DOPPS). Nephrol Dial Transplant 2004;19:108–120.
10 Ganesh SK, Hulbert-Shearon T, Port FK, et al: Mortality differences by dialysis modality among
incident ESRD patients with and without coronary artery disease. J Am Soc Nephrol 2003;14:
415–424.
11 Goodkin DA, Bragg-Gresham JL, Koenig KG, et al: Association of comorbid conditions and mor-
tality in hemodialysis patients in Europe, Japan, and the United States: the Dialysis Outcomes and
Practice Patterns Study (DOPPS). J Am Soc Nephrol 2003;14:3270–3277.
12 Wolfe RA, Hulbert-Shearon TE, Ashby VB, et al: Improvements in dialysis patient mortality are
associated with improvements in urea reduction ratio and hematocrit, 1999 to 2002. Am J Kidney
Dis 2005;45:127–135.
13 Goodkin DA, Mapes DL, Held PJ: The Dialysis Outcomes and Practice Patterns Study (DOPPS):
how can we improve the care of hemodialysis patients? Semin Dial 2001;14:157–159.
14 Yoshino M, Kuhlmann MK, Kotanko P, et al: International differences in dialysis mortality refl ect
background general population atheroslerotic cardiovascular mortality. J Am Soc Nephrol 2006;
17:3510–3519.
15 Ridker PM, Cushman M, Stampfer MJ, et al: Infl ammation, aspirin and the risk of cardiovascular
disease in apparently healthy men. N Engl J Med 1997;336:973–979.
16 Vasan RS, Sullivan LM, Roubenoff R, et al: Infl ammatory markers and risk of heart failure in
elderly subjects without prior myocardial infarction: the Framingham Heart Study. Circulation
2003;107:1486–1491.
17 Liuzzo G, Biasucci LM, Gallimore JR, et al: The prognostic value of C-reactive protein and serum
amyloid A protein in severe unstable angina. N Engl J Med 1994;331:417–424.
18 Lagrand WK, Visser CA, Hermens WT, et al: C-reactive-protein as a cardiovascular risk: more
than an epiphenomenon? Circulation 1999;100:96–102.
19 Carracedo J, Ramirez R, Madueno JA, et al: Cell apoptosis and hemodialysis-induced infl amma-
tion. Kidney Int Suppl 2002;80:89–93.
20 Lonnemann G: When good water goes bad: how it happens, clinical consequences and possible
solutions. Blood Purif 2004;22:124–129.
21 Stenvinkel P, Ketteler M, Johnson RJ, et al: IL-10, IL-6, and TNF-␣: central factors in the altered
cytokine network of uremia – The good, the bad, and the ugly. Kidney Int 2005;67:1216–1233.
22 Schwendler SB, Filep JG, Galle J, et al: C-reactive protein: a family of proteins to regulate cardio-
vascular function. Am J Kidney Dis 2005;47:212–222.
23 Canaud B, Levesque R, Krieter D, et al: On line hemodiafi ltration as routine treatment of end-
stage renal failure: why pre- or mixed dilution mode is necessary in on line haemodiafi ltration
today? Blood Purif 2004;22(suppl 2):40–48.
24 Passlick-Deetjen J, Pohlmeier R: On-line hemodiafi ltration: gold standard or top therapy? In
Ronco C, La Greca G (eds): Hemodialysis Technology. Contrib Nephrol. Basel, Karger, 2002, vol
137, pp 201–211.
25 Kerr PB, Argiles A, Flavier JL, Canaud B, Mion CM: Comparison of hemodialysis and hemodia-
fi ltration: a long-term longitudinal study. Kidney Int 1992;41:1035–1040.
Panichi/Tetta 200
26 Ward RA, Schmidt B, Hullin J, Hillebrand GF, Samtleben W: A comparison of on-line hemodia-
fi ltration and high-fl ux hemodialysis: a prospective clinical study. J Am Soc Nephrol 2000;11:
2344–2350.
27 Van Laecke S, De Wilde K, Vanholder R: On line hemodiafi ltration. Artif Organs 2006;30:
579–585.
28 Beerenhout CH, Luik AJ, Jeuken-Mertens SG, et al: Pre-dilution on-line haemodiafi ltration
vs low-fl ux haemodialysis: a randomized prospective study. Nephrol Dial Transplant 2005;20:
1155–1163.
29 Carracedo J, Merino A, Nogueras S, et al: CD14⫹CD16⫹ monocyte-derived dendritic cells: a
prospective, crossover study. J Am Soc Nephrol 2006;17:2315–2321. DOI: 10.1681/.
30 Ramirez R, Carracedo J, Merino A, et al: Microinfl ammation induces endothelial damage in
hemodialysis patients: the role of convective transport. Kidney Int 2007, E-pub ahead of print.
31 Locatelli F, Del Vecchio L, Pozzoni P, Andrulli S: Dialysis adequacy and response to erythopoi-
esis-stimulating agents: what is the evidence base? Semin Nephrol 2006;26;269–274.
Dr. Vincenzo Panichi
Internal Medicine Department, University of Pisa
Via Roma 57
IT–56100 Pisa (Italy)
Tel. ⫹39 050 992 887, Fax ⫹39 050 553 414, E-Mail vpanichi@med.unipi.it
Ronco C, Canaud B, Aljama P (eds): Hemodiafi ltration.
Contrib Nephrol. Basel, Karger, 2007, vol 158, pp 201–209
Clearance of Beta-2-Microglobulin and
Middle Molecules in Haemodiafiltration
James Tattersall
Department of Renal Medicine, St. James’s University Hospital, Leeds, UK
Abstract
Middle molecules, consisting mostly of peptides and small proteins with molecular
weight the range of 500–60,000 Da, accumulate in renal failure and contribute to the uraemic
toxic state. 2-Microglobulin (2-MG) with a molecular weight of 11,000 is considered rep-
resentative of these middle molecules. These solutes are not well cleared by low-fl ux dialysis.
High-fl ux dialysis will clear middle molecules, partly by internal fi ltration. This convective
component of high-fl ux dialysis can be enhanced in a predictable way by haemodiafi ltration
(HDF). The convective and diffusive clearance rates of any middle molecule across any hae-
modiafi lter can be predicted from known or measurable factors such as its sieving coeffi -
cient, bound fraction and molecular weight. The removal of middle molecules is also
infl uenced by factors within the patient. 2-MG is distributed within the extracellular fl uid.
During HDF, 2-MG must transfer into the intravascular compartment across the capillary
walls. This transcapillary transfer at a rate of approximately 100 ml/min slows 2-MG
removal from the body. Continuing transfer after the end of a treatment session results in a
signifi cant rebound of 2-MG levels. This intercompartment transfer and its effect on 2-MG
clearance and concentration can be predicted by a 2-compartment model. By extrapolation,
the behaviour of other middle molecules can be predicted. The 2-compartment model, which
takes non-dialytic 2-MG clearance at a rate of 3 ml/min and 2-MG generation at a rate of
0.1 mg/min into account, can predict the effect of any HDF schedule on 2-MG levels. Low-
fl ux dialysis results in a 2-MG level of around 40 mg/l. Three times weekly, 4-hour HDF can
reduce 2-MG levels to around 20 mg/l. Long (nocturnal) HDF can reduce 2-MG levels to
around 10 mg/l, compared to physiological levels of less than 5 mg/l.
Copyright © 2007 S. Karger AG, Basel
One of the main motivations in performing haemodiafi ltration (HDF)
instead of haemodialysis is to enhance the removal of middle-molecular-weight
toxins. To this end, new membranes have been developed to target removal of
material of molecular weight in the range of 500–60,000 Da. These so-called
Tattersall 202
middle molecules consist mostly of peptides and low-molecular-weight proteins
with evidence of toxicity [1]. 2-Microglobulin (2-MG) at 11,000 Da is con-
sidered representative of this material and is the most extensively studied [2].
This chapter reviews the techniques for quantifying and predicting the effect of
HDF on the plasma levels of middle molecules.
Clearance of Middle Molecules in the Haemodiafilter
Diffusive Clearance
The rate of diffusion of solute through an aqueous fl uid is inversely propor-
tional to the square root of molecular weight (MW) [3]. Therefore, the diffusive
clearance (KD) can be estimated for any middle molecule from the equation below:
KK MWref
MW
DDref
=×
where KDref is the diffusive clearance of a reference middle molecule of molecu-
lar weight MWref (e.g. inulin), which can be read from the dialyser data sheet.
For middle molecules, the relatively slow rate of diffusion and transfer through
the membrane pores are the main limiting factors on KD. Therefore, KD is rela-
tively independent of blood or dialysate fl ow rates as long as these are more
than 200 ml/min. For 2-MG (11,000 Da) KD cannot exceed 12% of the mass
transfer area coeffi cient (KoA) for urea (60 Da). KD may be further reduced by
resistance to passage through the pores, dependent on pore size and molecular
weight.
Convective Clearance
In haemofi ltration or HDF, the spontaneous fi ltration/backfi ltration which
occurs in high-fl ux dialysis is replaced by a controlled fi ltration and re-infusion
which can be quantifi ed precisely. As long as the solute is small enough to pass
through the haemodiafi lter membrane pores unimpeded, the clearance due to
the fi ltration (convective clearance KC) is equal to the fi ltration rate and will be
the same for any solute.
In practice, larger solutes are somewhat impeded by the limited size of
the membrane pores, resulting in a ‘sieving’ effect, where solute is retained
on the blood side of the membrane, reducing clearance. This sieving effect is
quantifi ed for a specifi c solute and membrane as the sieving coeffi cient (SC). A
sieving coeffi cient of 1 indicates no impediment, so clearance is equal to the fi l-
tration rate (QF). A sieving coeffi cient of zero means that the solute cannot pass
through the membrane and the clearance is zero, regardless of fi ltration rate.
Middle-Molecule Clearance 203
In the general case, for postdilution haemofi ltration, convective clearance
is given by the equation:
Kc Qf S=×
Values for SC for representative middle molecules are known for any haemo-
diafi lter [4]. SC is largely dependent on molecular weight and can be predicted
or extrapolated. For 2-MG, SC ranges from 0.6 to 0.9. The higher value is
obtained from the so-called ‘ultrafl ux’ dialysers. It is worth noting that con-
vective clearance is independent of blood fl ow, dialysate fl ow or membrane
geometry.
Combining Convective and Diffusive Clearance
Convection and diffusion interact in a predictable way, depending on blood
plasma fl ow (QBP) which is calculated from blood fl ow (QB), Hct and total pro-
tein in grams per litre (Pt) [5]:
Qbp Qb Hct Pt=×−
()
×− ×
()
1 1 0 0107.
The clearance of any solute by HDF (K) in postdilution mode is given by the
equation [6]:
KKd Qbp Kd
Qbp Qf S=+ −
⎛
⎝
⎜⎞
⎠
⎟××
For online HDF in predilution mode, dialysate fl ow (QD) is taken into account:
KKd
Qbp Qd
Qbp Qd Kd
Qd Qf S=+
+
×−
⎛
⎝
⎜
⎜
⎜
⎞
⎠
⎟
⎟
⎟××
In the case of middle molecules at the higher end of the molecular-weight range,
where KD is negligible, K is equal to KC in postdilution mode. For other middle
molecules, the contribution of KD is fairly small, so any errors in its estimation
have little impact on the calculation of K.
The above equations can be used to calculate small-solute clearance; in
this case, blood water fl ow, instead of plasma water fl ow, should be used.
In practice, 2-MG clearance of up to 100 ml/min is possible in a high-
intensity postdialysis HDF with QB ⫽ 450 ml/min, QF ⫽ 100 ml/min, KD for urea
320 ml/min and SC ⫽ 0.8.
Tattersall 204
Effect of Adsorption on Clearance
Peptides and low-molecular-weight proteins in the middle-molecule range
are hydrophobic and are attracted to any non-aqueous material such as a dialyser
membrane. This has the effect of causing some of the solute to be adsorbed onto
the haemodiafi lter membrane as it passes through. Most of this adsorption occurs
in the supporting matrix material downstream from the rate-limiting pores so
does not affect the clearance rate. Adsorption of solute onto the blood side of the
membrane will tend to increase clearance, but by a limited and variable extent
due to the relatively small area of the blood side surface and rapid saturation.
Effect of Protein Binding on Clearance
Some middle molecules bind to plasma proteins, particularly albumin [7,
8]. Only the unbound fraction (B) is able to cross the membrane either by diffu-
sion or fi ltration. If there is signifi cant binding, the clearance predictions should
be multiplied by this unbound fraction.
Measuring Middle-Molecule Clearance
Adsorption of middle molecules will reduce solute concentration in the
fi ltrate and prevent quantifi cation of clearance by fi ltrate measurements. Cell
membranes are effectively impermeable to middle molecules so plasma water
fl ow (QBP) rather than blood water fl ow is used to calculate clearance on blood
side measurements. Solute concentrations at the dialyser inlet (Cin) and down-
stream of the infusion line (Cout) are used in the following equation:
KQbp Cin Cout
Cin
=× −
⎛
⎝
⎜⎞
⎠
⎟
The Two-Compartment Model
In order to remove solute from the body, the solute must transfer from the
far recesses of the body into the haemodiafi lter. Factors delaying solute trans-
fer within the body may have almost as much infl uence on the mass of solute
removed as clearance at the haemodiafi lter.
These factors within the body can be quantifi ed and predicted using a 2-
compartment model [9]. Solute is considered to be distributed within 2 aque-
ous compartments, the central and peripheral compartments. HDF clears solute
from the central compartment only. Solute from the peripheral compartment
must transfer into the central compartment before it can be cleared. After the
end of the HDF, the concentration in the central compartment rebounds upwards
due to continuing intercompartment transfer.
Middle-Molecule Clearance 205
The rate of solute transfer between compartments is proportional to the
difference in concentrations between compartments and a patient- and solute-
specifi c intercompartment clearance rate (Ki).
During HDF the mass of solute (Mc, Mp), concentrations (Cc, Cp) and vol-
umes (Vc, Vp) of the central and peripheral compartments can be computed by
repeated solution of the following equations in small time increments (t). Gp
and Gc are the solute generation rates, WGp and WGc are the rates of volume
change in the peripheral and central compartments. The value K is the total
clearance due to non-dialytic clearance plus HDF as appropriate. WGc and WGp
will be negative during the HDF sessions and positive at other times.
Mp Mp Cc Cp Ki Gp t
Mc Mc Cp Cc Ki Gc Cc K t
Cc Mc
Vc
=+ −
()
×+
()
×
=+ −
()
×+−×
()
×
=
CCp Mp
Vp
Vc Vc WGc t
Vp Vp WGp t
=
=+ ×
=+ ×
The 2-compartment model has been well studied for urea and creatinine. For
these small solutes, Vc and Vp have volumes of approximately 25% of the body
weight each and are considered to be the intracellular and extracellular water
volumes. The mechanism of intercompartment transfer is diffusion across the
cell membranes. Ki has a value of approximately 1,000 for urea and 500 for cre-
atinine. Other small solutes such as phosphate and methylguanidine seem to be
actively transported across the intercompartment boundary so their behaviour
during dialysis is not well described by the 2-pool model [10].
2-Microglobulin Two-Compartment Kinetics
2-MG kinetics has now been well characterized and also reliably con-
forms to a 2-pool model [11, 12]. In the case of 2-MG, the total volume of the
2 compartments (V) is approximately 14% of body weight and is considered to
be the extracellular water. The 2 compartments are considered to be the plasma
water and the extracellular, extravascular water. Ki for 2-MG is approximately
100 ml/min. The plasma water compartment consists of about 25% of the total
extracellular water volume [13]. Cell membranes are relatively impermeable to
2-MG, and intracellular water appears to play no part in 2-MG kinetics dur-
ing HDF.
Tattersall 206
The generation rate for 2-MG appears to be around 0.1 mg/min and is
similar to subjects with normal renal function [14]. The non-dialytic 2-MG
clearance is around 3 ml/min in patients without renal function [15].
Kinetics of Other Middle Molecules
For middle molecules, the intercompartment boundary is the blood ves-
sel walls, probably mostly the capillaries. The mechanism of transfer here is
fi ltration and determined by capillary hemodynamics. This is likely to occur at
a similar rate for all middle molecules (at least those which can be cleared by
HDF) but this requires confi rmation.
Equilibrated Kt/V
Urea removal is traditionally quantifi ed as Kt/V where K is the total urea
clearance, t is the treatment time and V the urea distribution volume. European
guidelines recommend that Kt/V should take the postdialysis rebound and inter-
compartment effects into account. An equilibrated or rebound-corrected Kt/V
(eKt/V) can be calculated using a postdialysis blood sample taken after the
rebound is complete, at least 30 min after the end of dialysis [16].
For middle molecules (or at least 2-MG), the rebound is more pronounced
and takes longer. It is even more important to include rebound and intercom-
partment effects when considering middle molecules [17].
The 2-compartment model can be approximated by more simple mathe-
matics using the patient clearance time (tp) which is closely related to Ki/V. The
value tp can be used to calculate the postrebound concentration using the dialy-
sis time in minutes (t), pre- and immediate postdialysis concentrations [18]:
0
2
4
6
8
10
12
14
16
18
20
1 1,441 2,881
Time (min)
3⫻4 h weekly
6⫻2 h weekly
6⫻4 h weekly
6⫻8 h weekly
2-MG (mg/l)
Fig. 1. Predicted blood 2-MG concentrations over a 2-day cycle with different sched-
ules. The model parameters are the same as indicated for table 1.
Middle-Molecule Clearance 207
rebound pre post
pre
t
ttp
=×
⎛
⎝
⎜⎞
⎠
⎟
+
A value for eKt/V can be calculated from the regular Kt/V using the immediate
postdialysis blood sample:
eKt V Kt V t
ttp
//=×
+
For urea, tp is 35 min, for creatinine, 70 min.
In theory, this method should be equally suitable for predicting the rebound
for 2-MG and other middle molecules; in this case, a value for tp of 110 min is
suggested by data in the literature and mathematical analysis.
Equivalent Renal Clearance
The equivalent renal clearance was originally described for urea as EKR
[19]. It is defi ned as the generation rate (G) divided by the time-averaged con-
centration over the weekly cycle (TAC). EKR has the familiar units of millili-
tres per minute and can be normalized to a body water volume of 40 litres as
EKRc. Values for 2-MG G and TAC can be calculated using the 2-compart-
ment model and used to calculate a 2-MG or middle-molecule EKR.
EKR calculated in this way is particularly suitable for quantifying middle-
molecule removal since it can be related to a physiological reference – nor-
mal renal function – and takes non-dialytic clearance, frequency and duration
of dialysis into account. As far as we know, the toxicity of middle molecules
relates to their TAC [20].
The Effect of Treatment Duration and Frequency
It is possible to use a 2-compartment model to predict the 2-MG peak
concentration and TAC for any dialytic schedule. Values for various dialysis
schedules are shown in table 1. The results of the predictions in the table are in
general agreement with 2-MG levels reported in the literature.
With low-fl ux dialysis, the EKR for 2-MG is 3 ml/min due to non-dialysis
clearance, despite no 2-MG clearance at the dialyser. HDF or high-effi ciency,
high-fl ux haemodialysis can reduce the peak 2-MG levels to around 50% of
the level seen with low-fl ux haemodialysis [21]. The reduction in 2-MG TAC
levels are not routinely measured but will be more impressive.
Tattersall 208
Due to the considerable rebound associated with short treatments, moving
from 3 ⫻ 4 to 6 ⫻ 2 h per week results in only minimal further decrease in the
peak 2-MG level [22].
It is possible to approach physiological TAC levels for both urea and 2-MG
found with normal renal function using extended, high-intensity daily treatments.
This would have to be delivered by nocturnal or continuous dialysis [23].
References
1 Winchester JF, Audia PF: Extracorporeal strategies for the removal of middle molecules. Semin
Dial 2006;19:110–114.
2 Winchester JF, Salsberg JA, Levin NW: Beta-2 microglobulin in ESRD: an in-depth review. Adv
Ren Replace Ther 2003;10:279–309.
3 Ofsthun NJ: Limitations of membrane structure and dialyzer design on large solute removal in
dialysis. Blood Purif 2000;18:264–266.
4 Klingel R, Ahrenholz P, Schwarting A, Rockel A: Enhanced functional performance characteris-
tics of a new polysulfone membrane for high-fl ux hemodialysis. Blood Purif 2002;20:325–333.
5 Padrini R, Canova C, Conz P, Mancini E, Rizzioli E, Santoro A: Convective and adsorptive
removal of beta2-microglobulin during predilutional and postdilutional hemofi ltration. Kidney Int
2005;68:2331–2337.
6 Ficheux A, Argiles A, Mion H, Mion CM: Infl uence of convection on small molecule clearances in
online HDF. Kidney Int 2000;57:1755–1763.
Table 1. Comparison between different dialysis strategies
Schedule Duration
h/treatment
QF ml/min 2-MG
peak mg/l
2-MG
TAC mg/l
EKR of
2-MG
ml/min
EKR of
urea ml/
min
3 times/week low-fl ux
haemodialysis 4 0 40 40 3 14
3 times/week HDF 4 50 22 17 6 18
3 times/week HDF 4 100 19 14 7 18
6 times/week HDF 2 100 15 13 8 21
6 times/week HDF 4 100 12 9 11 37
6 times/week HDF 8 50 10 7 14 60
6 times/week HDF 8 100 10 6 17 60
Normal renal function ⬍5⬍5⬎20 80
Data were computed using a 2-compartment model as described in the text. For HDF, a
2.2-m2 surface area dialyser with a urea dialytic clearance of 340 ml/min at 450 ml/min blood
fl ow and an SC of 0.8 for 2-MG is used. For low-fl ux HD, there is no dialytic 2-MG clear-
ance and urea clearance of 250. Volumes of 10.2 and 40 litres were assumed for 2-MG and
urea, respectively.
Middle-Molecule Clearance 209
7 Bammens B, Evenepoel P, Verbeke K, Vanrenterghem Y: Removal of the protein-bound sol-
ute p-cresol by convective transport: a randomized crossover study. Am J Kidney Dis 2004;44:
278–285.
8 Dhondt A, Vanholder R, Van BW, Lameire N: The removal of uremic toxins. Kidney Int Suppl
2000;76:S47–S59.
9 Sargent JA, Gotch FA: Mathematic modeling of dialysis therapy. Kidney Int Suppl 1980;10:
S2–S10.
10 Eloot S, Torremans A, De Smet R, Marescau B, De Wachter D, De Deyn PP, Lameire N, Verdonck
P, Vanholder R: Kinetic behavior of urea is different from that of other water-soluble compounds:
the case of the guanidino compounds. Kidney Int 2005;67:1566–1575.
11 Ward RA, Greene T, Hartmann B, Samtleben W: Resistance to intercompartmental mass transfer
limits beta2-microglobulin removal by post-dilution HDF. Kidney Int 2006;69:1431–1437.
12 David S, Bottalico D, Tagliavini D, Mandolfo S, Scanziani R, Cambi V: Behaviour of beta2-micro-
globulin removal with different dialysis schedules. Nephrol Dial Transplant 1998;13(suppl 6):
49–54.
13 Leypoldt JK, Cheung AK, Deeter RB, Goldfarb-Rumyantzev A, Greene T, Depner TA, Kusek J:
Kinetics of urea and beta-microglobulin during and after short hemodialysis treatments. Kidney
Int 2004;66:1669–1676.
14 Floege J, Bartsch A, Schulze M, Shaldon S, Koch KM, Smeby LC: Clearance and synthesis rates
of beta 2-microglobulin in patients undergoing hemodialysis and in normal subjects. J Lab Clin
Med 1991;118:153–165.
15 Xu XQ, Gruner N, Al-Bashir A, Trutt-Ibing CH, Melzer H, Fassbinder W, Stiller JS, Manm H:
Determination of extra renal clearance and generation rate of beta2-microglobulin in hemodialysis
patients using a kinetic model. ASAIO J 2001;47:623–627.
16 Leypoldt JK, Cheung AK: Revisiting the hemodialysis dose. Semin Dial 2006;19:96–101.
17 Leypoldt JK, Cheung AK, Deeter RB: Rebound kinetics of beta2-microglobulin after hemodialy-
sis. Kidney Int 1999;56:1571–1577.
18 Tattersall JE, De Takats D, Chamney P, Greenwood RN, Farrington K: The post-hemodialysis
rebound: predicting and quantifying its effect on Kt/V. Kidney Int 1996;50:2094–2102.
19 Casino FG, Lopez T: The equivalent renal urea clearance: a new parameter to assess dialysis dose.
Nephrol Dial Transplant 1996;11:1574–1581.
20 Maduell F, del Pozo C, Garcia H, Sanchez L, Hdez-Jaras J, Albero MD, Calvo C, Torregrosa I,
Navarro V: Change from conventional haemodiafi ltration to on-line haemodiafi ltration. Nephrol
Dial Transplant 1999;14:1202–1207.
21 Pickett TM, Cruickshank A, Greenwood RN, Taube D, Davenport A, Farrington K: Membrane
fl ux not biocompatibility determines beta-2-microglobulin levels in hemodialysis patients. Blood
Purif 2002;20:161–166.
22 Canaud B, Assounga A, Kerr P, Aznar R, Mion C: Failure of a daily haemofi ltration programme
using a highly permeable membrane to return beta 2-microglobulin concentrations to normal in
haemodialysis patients. Nephrol Dial Transplant 1992;7:924–930.
23 Raj DS, Ouwendyk M, Francoeur R, Pierratos A: Beta-2-microglobulin kinetics in nocturnal hae-
modialysis. Nephrol Dial Transplant 2000;15:58–64.
James Tattersall
Department of Renal Medicine, St. James’s University Hospital
Leeds LS9 7TF (UK)
Tel. ⫹44 113 206 4119, Fax ⫹44 113 206 4696, E-Mail jamestattersall@doctors.org.uk
Ronco C, Canaud B, Aljama P (eds): Hemodiafi ltration.
Contrib Nephrol. Basel, Karger, 2007, vol 158, pp 210–215
Inflammation and Hemodiafiltration
Rafael Ramirez, Alejandro Martin-Malo, Pedro Aljama
Department of Nephrology, Hospital Reina Sofi a, Cordoba, Spain
Abstract
Atherosclerosis and the subsequent cardiovascular diseases are the most important
causes of morbidity and mortality in patients with chronic kidney disease (CKD). CKD ath-
erosclerosis is closely associated with the infl ammatory status which is chronically present in
these patients. Hemodiafi ltration (HDF) is a highly effective dialysis modality expanding the
spectrum of removed uremic toxins from small to middle-sized molecular solutes. In addi-
tion, the online (OL) HDF using high fl uid substitution allows a greater clearance of large
uremic toxins. We have shown that OL-HDF markedly reduces the number of CD14CD16
monocyte-derived dendritic cells and their proinfl ammatory potential in CKD patients with-
out clinical evidence of infl ammatory disease. Moreover, we have also reported that OL-
HDF improved endothelial dysfunction with a decrease in the number of endothelial
microparticles (EMP) in peripheral blood and an increase in the percentage of endothelial
progenitor cells (EPC) as compared with high-fl ux hemodialysis (HF-HD). The results
obtained showed in both studies a reduction of CD14CD16 dendritic cells, EMP and
ECP in comparison with HF-HD. These data strongly suggest that the microinfl ammation
status observed in CKD patients is associated with endothelial damage and that amelioration
of the chronic microinfl ammation using high convective transport appears to reduce endothe-
lial damage and promote endothelial repair. Future studies will have to assess the mecha-
nisms of these immunological changes and their relevance in the reported improved survival
of patients treated with OL-HDF.
Copyright © 2007 S. Karger AG, Basel
Infl ammation is a physiological phenomenon necessary to maintain tissue
integrity against injury. However, when the infl ammatory response is inappro-
priate or excessively activated, it may lead to a chronic infl ammatory state with
harmful consequences. In fact, an activated acute-phase response has been shown
to be a predictor of cardiovascular disease in the general population. Chronic
kidney disease (CKD) has been associated with a high prevalence of cardiovas-
cular complications [1]. However, classic risk factors for atherosclerosis such as
Infl ammation and Hemodiafi ltration 211
hypertension, dyslipidemia, obesity and smoking seem to be less important than
other additional factors such as uremic toxins or infl ammation. Several papers
have reported that most CKD patients have a subclinical microinfl ammatory
state with high serum levels of some proinfl ammmatory/proatherogenic cyto-
kines and accumulation in peripheral blood of activated mononuclear cells that
prolong their lifespan. Markers of infl ammation such as C-reactive protein and
cytokines are independent predictors of all-cause and cardiovascular mortality in
these patients [2, 3].
The potential for reversal of established vascular disease is unknown in
hemodialysis (HD) patients since cardiovascular disease in CKD stage V is
already well established. The quality and technology of HD have improved in
relation to the biocompatibility of the materials used and the ability to remove
larger uremic toxin molecules. The question of whether the use of convective
transport may have an impact on the combined process of infl ammation, endo-
thelial damage and repair is of paramount importance. By combining ultrafi l-
tration (convective clearances for removing larger solutes) with diffusion (for
removal of small solutes), hemodiafi ltration (HDF) offers a highly effective
dialysis modality expanding the spectrum of removed uremic toxins from small
to middle-sized molecular solutes. The online (OL) HDF using high fl uid sub-
stitution allows a greater clearance of large uremic toxins [4, 5].
The mortality of dialysis patients remains elevated despite advances
in dialysis technology, signifi cant improvement in dialysis quality and better
global care of patients. On one side, it is interesting to note that a preliminary
report from the international Dialysis Outcomes and Practice Patterns Study
has shown that patients undergoing HDF had a reduced risk of death compared
to those treated by conventional HD [6]. This is a unique report that deserves
further analysis showing for the fi rst time that high-effi ciency convective thera-
pies are associated with a reduced death risk accounting for comorbid condi-
tions of patients and dialysis dose. The spectrum of eliminated uremic toxins
together with the adoption of ultrapure dialysate may all contribute to explain
the reduction of the chronic infl ammation. The latter has been associated with
an elevated number of circulating monocytes, an increased percentage of mature
proinfl ammatory monocytes and an overproduction of interleukin (IL) 1, tumor
necrosis factor and IL-6, without the ability to synthesize the anti-infl am-
matory cytokine IL-10 [7]. Overt signs of chronic infl ammation as evidenced
by high plasma levels of the acute-phase response proteins such as C-reactive
protein have been recognized as an independent risk factor of gross and cardio-
vascular mortality in CKD patients [7–9].
In a recent paper of our group we have shown for the fi rst time that the
convective component of OL-HDF reduces the number of proinfl ammatory
CD14CD16 cells (fi g. 1), the intracellular production of tumor necrosis
Ramirez/Martin-Malo/Aljama 212
factor and IL-6 as compared with high-fl ux HD (HF-HD) [10]. Both dialysis
modalities HF-HD and OL-HDF were in fact equal for the membrane used and
water quality. The study was performed in patients treated in one single center
to insure uniform procedures and standards of ultrapure dialysis fl uids during
the whole duration of the study. The patients included in this study showed low
plasma levels of C-reactive protein, had normal plasma albumin and showed
no evidence of otherwise clinically evident infl ammatory disease. Therefore,
the observed changes in the CD14CD16 cell population were observed in
well-nourished patients with very mild infl ammation. Obviously, further stud-
ies will be needed to assess whether these observations may be confi rmed in an
HD population with overt infl ammation as well. In addition, we have observed
a shortened telomere in CD14CD16 activated cells. The percentage of
CD14CD16 activated cells with a short telomere was higher in HF-HD than
in OL-HDF. These data strongly suggest that there was a higher percentage of
senescent cells in HF-HD as compared with OL-HDF. The main drawback of
this study was that we did not evaluate the biological response with respect to
different volume exchanges, which might give a more precise information on
the relative role of convective transport.
Several studies have shown evidence of endothelial damage and endothe-
lial dysfunction in HD patients [11]. Research in the biology of the endothelium
has provided new targets to examine endothelial damage and repair in human
pathology. Circulating endothelial cells have recently been recognized as use-
ful markers of vascular injury and angiogenesis [12]. Upon a noxious insult,
Short telomeres
High expression of adhesion
molecules
High amount of cytoplasm cytokines
High expression of TLR-2 snd TLR-9
Expression of dendritic markers
Long telomeres
Low or moderate expression of
ICAM-1
No cytoplasm cytokines
Low or moderate expression of
TLR-2 and TLR-9
1,000
CD16
800
600
400
200
0
1,000800600400
CD14
2000
Fig. 1. Characteristics of CD14CD16 cells in HD patients. TLR Toll-like recep-
tor; ICAM intercellular adhesion molecule.
Infl ammation and Hemodiafi ltration 213
endothelial cells may undergo vesiculation releasing ‘endothelial microparticles’
(EMP) into the bloodstream. EMP harbor cell surface proteins and cytoplasmic
elements and express endothelial-specifi c surface markers. In addition, EMP
are currently viewed as a new pathway that can be used by endothelial cells to
exchange information. Endothelial progenitor cells (EPC) which originate from
the bone marrow, rather than from vessel walls, can be identifi ed through their
expression of CD34 (a surface marker common to hematopoietic stem cells and
mature endothelial cells) and vascular endothelial cell growth factor receptor 2
(or kinase-domain-related receptor). EPC are considered to have a role in the
repair of vascular injury, angiogenesis and tissue vascularization [13].
Patients undergoing HD show a high number of circulating CD31/annexin
V EMP as compared to healthy human subjects. Taking into account that the
elevated number of EMP was associated with a remarkably high number of
EPC [14], we designed a prospective crossover study to examine whether OL-
HDF may not only improve microinfl ammation, as indicated by CD14CD16
dendritic cells, but also EMP and EPC numbers, as biomarkers of endothelial
injury and repair in comparing OL-HDF with HF-HD (table 1). Both circulat-
ing EMP and EPC were reduced after 4 months of OL-HDF [15]. However,
EPC did not reach the circulating levels observed in the control group (table 1).
Interest has emerged from recent human studies underscoring the tight asso-
ciation of EMP with endothelial dysfunction and arterial dysfunction in CKD.
By a linear regression analysis, we found in this report a signifi cant correla-
tion between proinfl ammatory CD14CD16 cells and the number of EMC
and EPC. Therefore, it is possible to speculate that CD14CD16 monocyte-
derived dendritic cells may have an additional effect on endothelial activation
and injury leading to release of EMP into the circulation, and stimulation of the
bone marrow to increase the production of circulating EPC. However, we were
unable to claim a direct cause-and-effect relationship between microinfl amma-
tion, formation of EMP and increased numbers of circulating EPC.
There was no increase in EMP in postdialysis blood samples from OL-
HDF. In contrast, a marked increase in EMP was observed in all patients during
Table 1. Biomarkers (%) of endothelial injury and repair in HF-HD and OL-HDF
EMP EPC CD14CD16 cells
Healthy subjects 9 4.3 5.8 4.7 5 3
HF-HD 21.2 6.8a13.2 3.4b23 7.8a
OL-HDF 14.9 3.6a, b 9.7 1.9a12 9.2a, b
n 16. a p 0.05 versus healthy subjects; b p 0.05 versus HF-HD.
Ramirez/Martin-Malo/Aljama 214
the HF-HD period. However, in both dialysis modalities, EPC decreased signifi -
cantly in the postdialysis blood samples compared with the predialysis values.
Why EPC increase in HF-HD and diminish in OL-HDF is not clear. Recent
studies also suggest that circulating detached endothelial cells and stimulated
EPC from the bone marrow may refl ect endothelial injury. In our paper [15], we
did not assess the plasma levels of vascular endothelial growth factor, a stimu-
lus for EPC release from bone marrow. Obviously, the potential role played
by vascular endothelial growth factor could not be excluded from our study.
Another potential confounding factor is the effect of erythropoietin therapy, a
strong stimulus for EPC release; however, it is important to highlight that there
was no signifi cant change in the erythropoietin dose in the course of the study.
In summary, OL-HDF seems to improve microinfl ammation and endothe-
lial injury and repair, as indicated by a reduction of CD14CD16 dendritic
cells, EMP and ECP in comparison with HF-HD. Future studies will have to
assess the mechanisms of these immunological changes, and their relevance in
the reported improved survival of patients treated with OL-HDF.
Acknowledements
This work was supported by grants from: Instituto de Salud Carlos III (FIS 03/0946, 05/0896,
06/0724, 06/0747 and RETIC RD06/006L/0007 – FEDER) and Fundación Nefrologica.
References
1 Reiss AB, Glass AD: Atherosclerosis: immune and infl ammatory aspects. J Invest Med 2006;54:
123–131.
2 Roberts MA, Hare DL, Ratnaike S, Ierino FL: Cardiovascular biomarkers in CKD: pathophysi-
ology and implications for clinical management of cardiac disease. Am J Kidney Dis 2006;48:
341–360.
3 Zoccali C: Traditional and emerging cardiovascular and renal risk factors: an epidemiologic per-
spective. Kidney Int 2006;70:26–33.
4 Ward RA, Schmidt B, Hullin J, Hillebrand GF, Samtleben W: A comparison of on-line hemodia-
fi ltration and high-fl ux hemodialysis: a prospective clinical study. J Am Soc Nephrol 2000;11:
2344–2350.
5 Locatelli F, Manzoni C, Di Filippo S: The importance of convective transport. Kidney Int 2002;
61(suppl 80):S115–S120.
6 Canaud B, Bragg-Gresham JL, Marshall MR, Desmeules S, Gillespie BW, Depner T, Klassen P,
Port FK: Mortality risk for patients receiving hemodiafi ltration versus hemodialysis: European
results from the DOPPS. Kidney Int 2006;69:2087–2093.
7 Kaysen GA: The microinfl ammatory state in uremia: causes and potential consequences. J Am
Soc Nephrol 2001;12:1549–1557.
8 Rostaing L, Peres C, Tkaczuk J, Charlet JP, Bories P, Durand D, Ohayon E, de Preval C, Abbal M:
Ex vivo fl ow cytometry determination of intracytoplasmic expression of IL-2, IL-6, IFN- and
TNF- in monocytes and T lymphocytes, in chronic hemodialysis patients. Am J Nephrol 2000;
20:18–26.
Infl ammation and Hemodiafi ltration 215
9 Carracedo J, Ramirez R, Madueño JA, Soriano S, Rodríguez-Benot A, Rodriguez M, Martin-Malo
A, Aljama P: Cell apoptosis and hemodialysis-induced infl ammation. Kidney Int 2002;80:89–93.
10 Carracedo J, Merino A, Nogueras S, Carretero D, Berdud I, Ramirez R, Tetta C, Rodriguez M,
Martin-Malo A, Aljama P: On-line hemodiafi ltration reduces the proinfl ammatory CD14CD16
monocyte-derived dendritic cells: a prospective, crossover study. J Am Soc Nephrol 2006;17:
2315–2321.
11 Locatelli F, Pozzoni P, Tentori F, del Vecchio L: Epidemiology of cardiovascular risk in patients
with chronic kidney disease. Nephrol Dial Transplant 2003;18(suppl 7):vii2–9.
12 Jimenez JJ, Jy W, Mauro LM, et al: Endothelial microparticles (EMP) as vascular disease markers.
Adv Clin Chem 2005;39:131–157.
13 Masuda H, Kalka C, Asahara T: Endothelial progenitor cells for regeneration. Hum Cell 2000;13:
153–160.
14 Herbrig K, Pistrosch F, Oelschlaegel U, et al: Increased total number but impaired migratory
activity and adhesion of endothelial progenitor cells in patients on long-term hemodialysis. Am J
Kidney Dis 2004;44:840–849.
15 Ramirez R, Carracedo J, Merino A, Nogueras S, Alvarez-Lara MA, Rodriguez M, Martin-Malo A,
Tetta C, Aljama P: Microinfl ammation induces endothelial damage in hemodialysis patients: the
role of convective transport. Kidney Int 2007, in press.
Prof. Pedro Aljama
Department of Nephrology, Hospital Reina Sofi a
Avda. Menendez Pidal, 1
ES–14004 Cordoba (Spain)
Tel. 34 57010143, E-Mail Pedro.aljama.sspa@juntadeandalucia.es
Ronco C, Canaud B, Aljama P (eds): Hemodiafi ltration.
Contrib Nephrol. Basel, Karger, 2007, vol 158, pp 216–224
Effect of Online Hemodiafiltration on
Morbidity and Mortality of Chronic Kidney
Disease Patients
Bernard Canaud
Nephrology, Dialysis and Intensive Care Unit, Aider and Renal Research and
Training Institute, Lapeyronie University Hospital, Montpellier, France
Abstract
Conventional diffusion-based dialysis modalities including high-fl ux hemodialysis
(HD) are limited in their capacity to clear uremic toxins. Moreover they are associated with a
relatively high incidence of morbidity and mortality. Online hemodiafi ltration (ol-HDF)
combining the use of a high-fl ux membrane dialyzer, ultrapure dialysis fl uid and high con-
vective fl uid exchange is highly effi cient with the lowest bioreactive profi le in renal replace-
ment therapy methods. Regular use of high-effi ciency ol-HDF is associated with reduced
morbidity (hypotension incidence, better blood pressure control, improved hemocompatibil-
ity, reduced infl ammation profi le, improved lipid profi le, improved anemia correction,
reduced incidence of 2-microglobulin amyloidosis and hospitalization). More recently, sev-
eral cohort studies have shown that high-effi ciency ol-HDF is associated with a 35% reduced
risk of mortality in an unselected dialysis population. ol-HDF has been proven to be a safe
and very effi cient renal replacement therapy. ol-HDF has come of age and should be consid-
ered now as the new standard for highly effi cient renal replacement therapy.
Copyright © 2007 S. Karger AG, Basel
Introduction
Conventional diffusion-based dialysis modalities including high-fl ux
hemodialysis (HD) are limited in their capacity to clear middle and large-
size uremic toxins [1, 2]. Moreover, dialysis-related complications, includ-
ing 2-microglobulin amyloidosis, accelerated atherosclerosis, left ventricular
hypertrophy, infl ammation, malnutrition and ageing, are as many clinical mani-
festations showing the limits of conventional HD methods. Mixed diffusive and
convective methods such as hemodiafi ltration (HDF), mimicking glomerular
Morbidity and Mortality in Online Hemodiafi ltration 217
fi ltration of native kidneys, are then required for enlarging the molecular-weight
spectrum of uremic toxins removed by dialysis and for enhancing the overall
effi cacy of the renal replacement therapy in order to improve the outcome for
patients with chronic kidney disease stage 5 (CKD-5) [3–6].
Worldwide clinical experience has proven for several thousand CKD
patients treated with online (ol) HDF that the method was safe, reliable and eco-
nomically viable [7]. Safety and reliability of ol-HDF machines is now certifi ed
by European notifi ed bodies under the EC label (European Community).
The enhanced effi ciency of high-fl ux convective therapies is one of the
best-documented aspects of the method. HDF provides signifi cantly higher
instantaneous clearances than high-fl ux HD both for small- and middle-mol-
ecule solutes [8–10]. Inorganic phosphate and 2-microglobulin are two major
uremic markers with high clinical relevance that can be used to support this fact.
Phosphate removal is increased with HDF methods reaching up 30–35 mmol/
session [11]. Based on 3 sessions a week, total phosphate removal is still not
adequate to restore phosphate balance in CKD patients. Phosphatemia control
requires however a reduced dose of phosphate binders [12]. 2-Microglobulin
removal is also increased with high-fl ux ol-HDF [13, 14]. The reduction ratio
of 2-microglobulin averages usually 70–80%. 2-Microglobulin mass removal
ranges between 150 and 200 mg/session [15]. Several prospective controlled
studies have confi rmed that HDF treatment was accompanied by a signifi cant
decline in blood 2-microglobulin concentrations on a mid-term period [16, 17].
High circulating 2-microglobulin concentrations have been shown recently to
be associated with an increased relative risk of mortality in CKD-5 HD-treated
patients.
ol-HDF combining the use of a high-fl ux membrane dialyzer, ultrapure
dialysis fl uid and high convective fl uid exchange provides the lowest bioreac-
tive profi le in the fi eld of extracorporeal renal replacement therapy. This prop-
erty is quite benefi cial for CKD patients, since by reducing dialysis-induced
cell and protein activation, ol-HDF contributes also to the prevention or cure of
infl ammation, a major actor in the dialysis-related pathology.
Online Hemodiafiltration Reduces Morbidity in Patients with Chronic
Kidney Disease Stage 5
The regular use of high-effi ciency ol-HDF is associated with several clini-
cally benefi cial effects contributing to reduce the overall morbidity of CKD-5
patients. They are summarized in this section.
The incidence of hypotensive episodes is reduced with HDF methods
permitting to achieve more easily dry weight while restoring the sodium fl uid
Canaud 218
balance [18]. Indeed, this is particularly interesting in cardiac-compromised
patients and/or in hypotension-prone patients. This benefi cial effect has been
related to a vasomodulation effect involving a negative thermal balance (increased
peripheral vascular resistance and venous tone), a high sodium concentration of
the substitution fl uid (increased osmolality) and removal of vasodilating media-
tors (reduced endothelial dysfunction) [19, 20]. HD intolerance (nausea, vomit-
ing, cramps and headache) is reduced with high-effi ciency HDF compared to
conventional HD. Postdialysis fatigue is less frequently observed with convec-
tive therapies [21]. Such properties are particularly suitable in elderly, diabetics
and cardiac patients.
Better blood pressure control is generally observed in HDF-treated
patients. This benefi cial cardiac effect is mainly due to the preservation of intra-
dialytic hemodynamic stability permitting to correct easily the extracellular
fl uid overload. Increased treatment time or frequency combined with a better
compliance to sodium diet restriction may facilitate the achievement of this
goal [22]. Regular application of high-effi ciency ol-HDF has been associated
with a reduction of left ventricular hypertrophy contributing to a better preser-
vation of cardiac function [23].
Recent studies have suggested that high-fl ux therapies including ol-HDF
might contribute to a longer and better preservation of residual renal function.
Interestingly, this positive effect appears now comparable to that observed in
peritoneal dialysis patients [24]. Although this phenomenon is not completely
understood or proved, it might result from a reduction of the infl ammation state
of HD patients and from a reduced incidence of intradialytic hypotension epi-
sodes [25].
Improved hemocompatibility and reduced infl ammation is commonly
reported with ol-HDF. Based on sensitive markers of the acute-phase reaction
(C-reactive protein, interleukins 1 and 6, interleukin 1 and 6 receptor antibodies,
albumin) or on proinfl ammatory cell populations such as monocytes-derived
CD14⫹ and CD16⫹ cells, several prospective studies have shown that the
behavior of these markers remains stable over time in ol-HDF [26–29]. More
recently in a prospective randomized study, it has been shown that ol-HDF was
associated with less damaging and better repair effects on the endothelial cells
[30]. These positive effects result from the combined use of a synthetic biocom-
patible membrane, ultrapure dialysis fl uid and increased removal of some puta-
tive uremic toxins. Prevention of infl ammation is a crucial concern for reducing
the incidence of dialysis-related complications in long-term dialysis patients
[31].
Renal anemia commonly observed in HD patients requires erythropoietin
use in 80–100% of patients. Although this fact remains still controversial, high-
effi ciency ol-HDF has been shown to improve anemia control and to reduce
Morbidity and Mortality in Online Hemodiafi ltration 219
erythropoietin needs in HD patients [32]. This positive effect has been particu-
larly noted when patients were switched from low-fl ux HD to high-fl ux HDF or
to HD with protein-leaking high-fl ux membranes [21]. These observations sug-
gest that convective methods might remove some protein-bound erythropoietic
inhibitor substance. It is also worth noting that the improvement in anemia is
associated with a reduced infl ammation state of the patient [33].
Caloric and/or protein malnutrition is observed in about one third of dialy-
sis patients. Several recent studies have shown that the use of high-fl ux methods
including HDF may have a positive impact on the nutritional state when com-
pared to low-fl ux membranes [17, 34]. Anthropometric parameters, such as dry
weight and body mass index, and albumin tend to increase over time in patients
treated with convective therapies [35]. This is associated with an increase in
dietary protein intake as evaluated by the urea generation rate [8]. One must
recognize that this positive effect might result from combined effects of the use
of high-fl ux membranes with ultrapure dialysate and more speculatively with
the removal of anorexia-inducing uremic toxins [33].
Dyslipidemia profi le, oxidative stress and advanced glycation end products
reported in dialysis patients contribute to accelerate the atheromatosis and ath-
erosclerotic process. The regular use of convective therapies has been shown to
improve lipid profi le [36] and to reduce oxidative stress and advanced glyca-
tion end product concentrations [37, 38]. Such a benefi cial effect may be partly
due to the improved biocompatibility of the dialyzer and the ultrapurity of the
dialysate [39]. Note that the increased loss of natural antioxidant substances
(vitamin C, vitamin E, selenium) may abolish in part the benefi cial effect of
high-effi ciency convective modalities [40]. In contrast, a recent study has shown
that asymmetric dimethylarginine concentrations were similar in high-fl ux and
HDF-treated patients [41]. The regular oral supplementation in natural antioxi-
dants appears highly desirable in both HD- and ol-HDF-treated patients.
2-Microglobulin amyloidosis has become a major complication of long-
term HD therapy. Using carpal tunnel syndrome as crude and fi rst manifesta-
tion of 2-microglobulin amyloidosis in HD patients, it is commonly accepted
that its incidence reaches 50% at 10 years and 100% at 20 years when conven-
tional low-fl ux HD treatment is applied. Several cohort studies indicate that
the extended use of high-fl ux membranes has a benefi cial impact on the devel-
opment of 2-microglobulin amyloidosis reducing its incidence [42]. Indeed,
almost all studies report a 50% reduction of the incidence of carpal tunnel syn-
drome when combining the use of convective methods and ultrapure dialysis
fl uid [43].
Growth retardation is a major concern in children with CKD-5.
Conventional HD alone has not been able to reverse this development retarda-
tion. A recent study based on daily Ol-HDF has shown that this schedule is able
Canaud 220
to correct growth retardation in children with CKD [44]. This benefi cial effect
is achieved by combining the improvement of treatment effi cacy, the enhance-
ment of dietary and caloric intakes and the better correction of internal milieu
disturbances (acidosis, calcium and phosphate control) [45]. The combined use
of growth hormone, erythropoietic stimulating agents and ol-HDF provides
now the opportunity to normalize growth rate in CKD kids [46, 47].
Hospitalization, used as a global marker of renal replacement therapy mor-
bidity, is similar to conventional-HD-treated patients [48]. Few cohort studies
addressing this issue have not found signifi cant differences of morbidity and
mortality in the convective-mode-treated group of patients [49]. Large cohort
studies such as the United States Renal Data System have shown a signifi cant
reduction of morbidity and mortality in patients treated with high-fl ux meth-
ods as compared to those treated with low-fl ux dialyzers [50, 51]. These posi-
tive results were not confi rmed in the large prospective and controlled HEMO
study recently reported [52]. Indeed, it must be underlined that in the HEMO
study dialyzer reuse may have blunted benefi cial effects of high-fl ux methods
by altering some performances. In summary, no signifi cant difference in terms
of frequency and duration of hospitalization stay has been reported neither in
high-fl ux methods nor in high-effi ciency ol-HDF. In other words, this observa-
tion means that high-effi ciency ol-HDF is not harmful for CKD patients.
Online Hemodiafiltration Improves Survival in Patients with Chronic
Kidney Disease Stage 5
Mortality is the most robust primary endpoint used to compare the effi cacy
of renal replacement therapy modalities. Few cohort studies have shown that
mortality was reduced in HDF-treated groups [49, 53–55]. Meanwhile, the neg-
ative results of the HEMO study reported recently with patients either receiving
a high dialysis dose or using a high-fl ux membrane were relatively disappoint-
ing [56]. It is however interesting to note that in the high-fl ux-treated group
the incidence of cardiovascular events was nevertheless reduced. A recent reap-
praisal analysis of the HEMO study has shown that high predialysis 2-micro-
globulin concentrations were strong and independent predictors of mortality
in HD patients [57]. This observation strongly supports the fact that enhanced
removal of middle uremic toxins is benefi cial for CKD patients.
The European section of the international Dialysis and Outcomes Practice
Patterns Study has shown that high-effi ciency-HDF-treated patients had a bet-
ter survival than regularly HD-treated patients accounting for age, sex, dialysis
dose, comorbid conditions and country specifi cities [58]. Despite the fact that
the relative risk of death was reduced in both HDF-treated groups, only the
Morbidity and Mortality in Online Hemodiafi ltration 221
high-effi ciency HDF group (substitution volume 15–25 l/session) had a sig-
nifi cant benefi t marked by a 35% reduction of death risk compared to patients
treated with low-fl ux HD. Independent investigators have recently confi rmed
this positive fi nding when analyzing a large European database (Euclid). The
relative risk of death was reduced in this study by 36% in HDF-treated patients
[59]. It has also been observed in a retrospective US study that high-effi ciency
HDF based on 2 fi lters in series was able to reduce mortality in HDF by 65%
[60].
Conclusions
High-effi ciency ol-HDF has been proven to be the most effi cient renal
replacement therapy for CKD-5 patients. By combining diffusive and convective
clearances, ol-HDF offers the highest removal rates for both small- and middle-
molecule uremic toxins. By combining ultrapure dialysis fl uid and synthetic
hemocompatible membranes, ol-HDF improves considerably the hemocompat-
ibility of the HD system. Clinical studies indicate that regular use of ol-HDF
tends to reduce dialysis-related morbidity. Large cohort studies have shown that
the survival of CKD-5 HDF-treated patients was signifi cantly improved. By
customizing the dialysis schedule to the patient’s needs and tolerance, either
by increasing session frequency (e.g. daily HDF) or by lengthening treatment
duration (longer dialysis session), it has to be proven that ol-HDF may again
expand this benefi cial effect on dialysis outcomes [61, 62].
References
1 Henderson LW: Dialysis in the 21st century. Am J Kidney Dis 1996;28:951–957.
2 Clark WR, Gao D. Low-molecular weight proteins in end-stage renal disease: potential toxicity
and dialytic removal mechanisms. J Am Soc Nephrol 2002;13:S41–S47.
3 Vanholder R, Glorieux G, De Smet R, et al: New insights in uremic toxins. Kidney Int Suppl 2003;
84:S6.
4 Vanholder R, Glorieux GL, De Smet R: Back to the future: middle molecules, high fl ux mem-
branes, and optimal dialysis. Hemodial Int 2003;7:52.
5 Henderson LW, Clark WR, Cheung AK: Quantifi cation of middle molecular weight solute removal
in dialysis. Semin Dial 2001;14:294–299.
6 Leypoldt JK, Cheung AK, Carroll CE, Stannard DC, Pereira BJ, Agodoa LY, Port FK: Effect of
dialysis membranes and middle molecule removal on chronic hemodialysis patient survival. Am J
Kidney Dis 1999;33:349–355.
7 Canaud B, Bosc JY, Leray-Moragues H, Stec F, Argiles A, Leblanc M, Mion C: On-line haemodia-
fi ltration: safety and effi cacy in long-term clinical practice. Nephrol Dial Transplant 2000;15(suppl
1):60–67.
8 Canaud B, Bosc JY, Leblanc M, Garred LJ, Vo T, Mion C. Evaluation of high-fl ux hemodiafi ltra-
tion effi ciency using an on-line urea monitor. Am J Kidney Dis 1998;31:74–80.
Canaud 222
9 Maduell F: Hemodiafi ltration. Hemodial Int 2005;9:47–55.
10 Shinzato T, Kobayakawa H, Maeda K: Comparison of various treatment modes in terms of beta
2-microglobulin removal: hemodialysis, hemofi ltration, and push/pull HDF. Artif Organs 1989;13:
66–70.
11 Lornoy W, De Meester J, Becaus I, Billiouw JM, Van Malderen PA, Van Pottelberge M. Impact of
convective fl ow on phosphorus removal in maintenance hemodialysis patients. J Ren Nutr 2006;
16:47–53.
12 Zehnder C, Gutzwiller JP, Renggli K: Hemodiafi ltration – A new treatment option for hyperphos-
phatemia in hemodialysis patients. Clin Nephrol 1999;52:152.
13 Krieter DH, Lemke HD, Canaud B, Wanner C: Beta-2-microglobulin removal by extracorporeal
renal replacement therapies. Biochim Biophys Acta 2005;1753:146–153.
14 Ward RA, Greene T, Hartmann B, Samtleben W. Resistance to intercompartmental mass trans-
fer limits beta2-microglobulin removal by post-dilution hemodiafi ltration. Kidney Int 2006;69:
1431–1437.
15 Lornoy W, Becaus I, Billiouw JM, et al: On-line haemodiafi ltration: remarkable removal of beta2-
microglobulin – Long-term clinical observations. Nephrol Dial Transplant 2000;15(suppl 1):49.
16 Locatelli F, Mastrangelo F, Redaelli B, et al: Effects of different membranes and dialysis technolo-
gies on patient treatment tolerance and nutritional parameters. The Italian Cooperative Dialysis
Study Group. Kidney Int 1996;50:1293.
17 Basile C: The effect of convection on the nutritional status of haemodialysis patients. Nephrol Dial
Transplant 2003;18(suppl 7):vii46–49.
18 Altieri P, Sorba G, Bolasco P, Ledebo I, Ganadu M, Ferrara R, Menneas A, Asproni E, Casu D,
Passaghe M, Sau G, Cadinu F, Sardinian Study Group on Hemofi ltration On-line. Related com-
parison between hemofi ltration and hemodiafi ltration in a long-term prospective cross-over study.
J Nephrol 2004;17:414–422.
19 Donauer J, Schweiger C, Rumberger B, Krumme B, Bohler J: Reduction of hypotensive side effects
during online-haemodiafi ltration and low temperature haemodialysis. Nephrol Dial Transplant
2003;18:1616–1622.
20 van der Sande FM, Rosales LM, Brener Z, Kooman JP, Kuhlmann M, Handelman G, Greenwood
RN, Carter M, Schneditz D, Leunissen KM, Levin NW: Effect of ultrafi ltration on thermal vari-
ables, skin temperature, skin blood fl ow, and energy expenditure during ultrapure hemodialysis. J
Am Soc Nephrol 2005;16:1824–1831.
21 Maduell F, del Pozo C, Garcia H, Sanchez L, Hdez-Jaras J, Albero MD, Calvo C, Torregrosa I,
Navarro V: Change from conventional haemodiafi ltration to on-line haemodiafi ltration. Nephrol
Dial Transplant 1999;14:1202–1207.
22 Stegmayr BG: Ultrafi ltration and dry weight – What are the cardiovascular effects? Artif Organs
2003;27:227–229.
23 Maduell F, Navarro V, Torregrosa E, Rius A, Dicenta F, Cruz MC, Ferrero JA: Change from
three times a week on-line hemodiafi ltration to short daily on-line hemodiafi ltration. Kidney Int
2003;64:305–313.
24 McKane W, Chandna SM, Tattersall JE, et al: Identical decline of residual renal function in high-
fl ux biocompatible hemodialysis and CAPD. Kidney Int 2002;61:256.
25 Pozzoni P, Filippo S, Manzoni C, Locatelli F: The relevance of convection in clinical practice: a
critical review of the literature. Hemodial Int 2006;10(suppl 1):S33–S38.
26 van Tellingen A, Grooteman MP, Schoorl M, et al: Intercurrent clinical events are predictive of
plasma C-reactive protein levels in hemodialysis patients. Kidney Int 2002;62:632.
27 Canaud B, Wizemann V, Pizzarelli F, et al: Cellular interleukin-1 receptor antagonist production in
patients receiving on-line haemodiafi ltration therapy. Nephrol Dial Transplant 2001;16:2181.
28 Olsson J, Dadfar E, Paulsson J, Lundahl J, Moshfegh A, Jacobson SH: Preserved leukocyte CD11b
expression at the site of interstitial infl ammation in patients with high-fl ux hemodiafi ltration.
Kidney Int 2007;71:582–588.
29 Carracedo J, Merino A, Nogueras S, Carretero D, Berdud I, Ramirez R, Tetta C, Rodriguez M,
Martin-Malo A, Aljama P: On-line hemodiafi ltration reduces the proinfl ammatory CD14⫹CD16⫹
monocyte-derived dendritic cells: a prospective, crossover study. J Am Soc Nephrol 2006;17:
2315–2321.
Morbidity and Mortality in Online Hemodiafi ltration 223
30 Ramirez R, Carracedo J, Merino A, Nogueras S, Alvarez-Lara MA, Rodriguez M, Martin-Malo A,
Tetta C, Aljama P. Microinfl ammation induces endothelial damage in hemodialysis patients: the
role of convective transport. Kidney Int 2007, E-pub ahead of print.
31 Deppisch RM, Beck W, Goehl H, et al: Complement components as uremic toxins and their poten-
tial role as mediators of microinfl ammation. Kidney Int Suppl 2001;78:S271.
32 Vaslaki L, Major L, Berta K, Karatson A, Misz M, Pethoe F, Ladanyi E, Fodor B, Stein G,
Pischetsrieder M, Zima T, Wojke R, Gauly A, Passlick-Deetjen J: On-line haemodiafi ltration ver-
sus haemodialysis: stable haematocrit with less erythropoietin and improvement of other relevant
blood parameters. Blood Purif 2006;24:163–173.
33 Schiffl H, Lang SM, Stratakis D, et al: Effects of ultrapure dialysis fl uid on nutritional status and
infl ammatory parameters. Nephrol Dial Transplant 2001;16:1863.
34 Lindsay RM, Spanner E, Heidenheim AP, et al: A multicenter study of short hour dialysis using
AN69S: preliminary results. ASAIO Trans 1991;37:M465.
35 Savica V, Ciolino F, Monardo P, Mallamace A, Savica R, Santoro D, Bellinghieri G: Nutritional
status in hemodialysis patients: options for on-line convective treatment. J Ren Nutr 2006;16:
237–240.
36 Blankestijn PJ, Vos PF, Rabelink TJ, et al: High-fl ux dialysis membranes improve lipid profi le in
chronic hemodialysis patients. J Am Soc Nephrol 1995;5:1703.
37 Chun-Liang L, Chiu-Ching H, Chun-Chen Y, et al: Reduction of advanced glycation end products
levels by on-line hemodiafi ltration in long-term hemodialysis patients. Am J Kidney Dis 2003;42:
524.
38 Calo LA, Naso A, Carraro G, Wratten ML, Pagnin E, Bertipaglia L, Rebeschini M, Davis PA,
Piccoli A, Cascone C: Effect of haemodiafi ltration with online regeneration of ultrafi ltrate on oxi-
dative stress in dialysis patients. Nephrol Dial Transplant 2007, E-pub ahead of print.
39 Ward RA, McLeish KR: Oxidant stress in hemodialysis patients: what are the determining fac-
tors? Artif Organs 2003;27:230.
40 Morena M, Cristol JP, Bosc JY, et al: Convective and diffusive losses of vitamin C during haemo-
diafi ltration session: a contributive factor to oxidative stress in haemodialysis patients. Nephrol
Dial Transplant 2002;17:422.
41 Kalousova M, Kielstein JT, Hodkova M, Zima T, Dusilova-Sulkova S, Martens-Lobenhoffer J,
Bode-Boger SM: No benefi t of hemodiafi ltration over hemodialysis in lowering elevated levels of
asymmetric dimethylarginine in ESRD patients. Blood Purif 2006;24:439–444.
42 Locatelli F, Manzoni C, Di Filippo S: The importance of convective transport. Kidney Int Suppl
2002;80:115.
43 Lonnemann G, Koch KM: Beta(2)-microglobulin amyloidosis: effects of ultrapure dialysate and
type of dialyzer membrane. J Am Soc Nephrol 2002;13(suppl 1):S72.
44 Fischbach M, Terzic J, Laugel V, Dheu C, Menouer S, Helms P, Livolsi A: Daily on-line haemodia-
fi ltration: a pilot trial in children. Nephrol Dial Transplant 2004;19:2360–2367.
45 Fischbach M, Terzic J, Menouer S, Dheu C, Soskin S, Helmstetter A, Burger MC: Intensifi ed and
daily hemodialysis in children might improve statural growth. Pediatr Nephrol 2006;21:1746–
1752.
46 Kari JA, Rees L: Growth hormone for children with chronic renal failure and on dialysis. Pediatr
Nephrol 2005;20:618–621.
47 Boehm M, Riesenhuber A, Winkelmayer WC, Arbeiter K, Mueller T, Aufricht C: Early erythropoi-
etin therapy is associated with improved growth in children with chronic kidney disease. Pediatr
Nephrol 2007, E-pub ahead of print.
48 Locatelli F, Pozzoni P, Manzoni C, Di Filippo S: High-fl ux hemodialysis and hemodiafi ltration:
impact on outcome; in Ronco C, La Greca G (eds): Hemodialysis Technology. Contrib Nephrol.
Basel, Karger, 2002, vol 137, pp 193–200.
49 Locatelli F, Marcelli D, Conte F, Limido A, Malberti F, Spotti D. Comparison of mortality in
ESRD patients on convective and diffusive extracorporeal treatments. The Registro Lombardo
Dialisi E Trapianto. Kidney Int 1999;55:286–293.
50 Port FK, Wolfe RA, Hulbert-Shearon TE, Daugirdas JT, Agodoa LY, Jones C, Orzol SM, Held PJ:
Mortality risk by hemodialyzer reuse practice and dialyzer membrane characteristics: results from
the USRDS dialysis morbidity and mortality study. Am J Kidney Dis 2001;37:276–286.
Canaud 224
51 Chauveau P, Nguyen H, Combe C, Chene G, Azar R, Cano N, Canaud B, Fouque D, Laville M,
Leverve X, Roth H, Aparicio M, French Study Group for Nutrition in Dialysis: Dialyzer mem-
brane permeability and survival in hemodialysis patients. Am J Kidney Dis 2005;45:565–571.
52 Eknoyan G, Beck GJ, Cheung AK, et al: Effect of dialysis dose and membrane fl ux in maintenance
hemodialysis. N Engl J Med 2002;19:347.
53 Santoro A, Panzetta G, Tessitore N, Atti M, Mancini E, Esteban J, London G, Ara JM, Miguel
JL, Neumann KH, Opatrny K, Perez R, Perrone B, Wizemann V, Zucchelli P: A prospective ran-
domised European multicentre study of medium-long run mortality and morbidity comparing
acetate-free biofi ltration and bicarbonate dialysis. J Nephrol 1999;12:375–382.
54 Wizemann V, Lotz C, Techert F, Uthoff S: On-line haemodiafi ltration versus low-fl ux haemodialy-
sis: a prospective randomized study. Nephrol Dial Transplant 2000;15(suppl 1):43–48.
55 Schiffl H. Prospective randomized cross-over long-term comparison of online haemodiafi ltration
and ultrapure high-fl ux haemodialysis. Eur J Med Res 2007;12:26–33.
56 Eknoyan G, Beck GJ, Cheung AK, et al: Effect of dialysis dose and membrane fl ux in maintenance
hemodialysis. N Engl J Med 2002;19:347.
57 Cheung AK, Rocco MV, Yan G, Leypoldt JK, Levin NW, Greene T, Agodoa L, Bailey J, Beck
GJ, Clark W, Levey AS, Ornt DB, Schulman G, Schwab S, Teehan B, Eknoyan G: Serum beta-2
microglobulin levels predict mortality in dialysis patients: results of the HEMO study. J Am Soc
Nephrol 2006;17:546–555.
58 Canaud B, Bragg-Gresham JL, Marshall MR, Desmeules S, Gillespie BW, Depner T, Klassen P,
Port FK: Mortality risk for patients receiving hemodiafi ltration versus hemodialysis: European
results from the DOPPS. Kidney Int 2006;69:2087–2093.
59 Jirka T, Cesare S, Di Benedetto A, Perera Chang M, Ponce P, Richards N, Tetta C, Vaslaky L:
Mortality risk for patients receiving hemodiafi ltration versus hemodialysis. Kidney Int 2006;70:
1524.
60 Bosch JP, Lew SQ, Barlee V, Mishkin GJ, von Albertini B: Clinical use of high-effi ciency hemodi-
alysis treatments: long-term assessment. Hemodial Int 2006;10:73–81.
61 Canaud B, Morena M, Leray-Moragues H, Chalabi L, Cristol JP: Overview of clinical studies in
hemodiafi ltration: what do we need now? Hemodial Int 2006;10(suppl 1):S5–S12.
62 Penne EL, Blankestijn PJ, Bots ML, van den Dorpel MA, Grooteman MP, Nube MJ, ter Wee PM,
CONTRAST Group: Resolving controversies regarding hemodiafi ltration versus hemodialysis:
the Dutch Convective Transport Study. Semin Dial 2005;18:47–51.
Prof. B. Canaud
Department of Nephrology, Dialysis and Intensive Care Unit, Lapeyronie Hospital,
University of Montpellier
371, avenue du Doyen-Gaston-Giraud
FR–34295 Montpellier Cedex 5 (France)
Tel. ⫹33 4 67 33 89 55, Fax ⫹33 4 67 60 37 83, E-Mail b-canaud@chu-montpellier.fr
Ronco C, Canaud B, Aljama P (eds): Hemodiafi ltration.
Contrib Nephrol. Basel, Karger, 2007, vol 158, pp 225–231
Optimizing the Prescription of
Hemodiafiltration
Francisco Maduell
Department of Nephrology, Hospital Clínic Barcelona, Barcelona, Spain
Abstract
Hemodiafi ltration with larger amounts of substitution fl uid offers an optimal way to
remove uremic substances. Hemodiafi ltration could be indicated for all hemodialysis patients.
Large observational studies have shown an association of a lower mortality risk with hemo-
diafi ltration using more than 15 liters of substitution fl uid. Specifi c indications should be
considered because hemodiafi ltration has been reported to be effective against hyperphospa-
temia, malnutrition, insomnia, irritability, restless-leg syndrome, polyneuropathy, anemia,
itching and joint pain, and may prevent dialysis-associated amyloidosis. In this chapter,
hemodiafi ltration prescriptions concerning blood and dialysate fl ow, infusion rate, vascular
access and frequency are detailed.
Copyright © 2007 S. Karger AG, Basel
Hemodialysis can be considered a routine renal replacement therapy that
guarantees reasonable short-term outcomes. The long-term clinical outcomes,
however, could be improved. Malnutrition is common, hyperphosphatemia,
hypertension and heart failure control is poor, rehabilitation and quality of life
are suboptimal, and hospitalization and mortality rates are high. The most com-
mon cause of mortality in chronic hemodialysis patients is cardiovascular dis-
ease, which is the attributed cause of death in approximately 50% of patients.
Depner [1] defi ned this situation as ‘residual syndrome’, which includes sus-
ceptibility to infection, reduced maximal oxygen consumption during exercise,
sleep disturbances, depression, impaired mental concentration, reduced stamina
and markedly increased susceptibility to cardiovascular complications. Possible
causes of residual syndrome are accumulation of dialyzable solutes that are
incompletely removed and aggregation of large-molecular-weight solutes that
are diffi cult to remove by dialysis.
Maduell 226
Various complications in hemodialysis patients may be related to an accu-
mulation of larger uremic substances diffi cult to remove by conventional hemo-
dialysis. Hemodiafi ltration with larger amounts of substitution fl uid offers an
optimal way to remove uremic substances ranging widely in molecular size
from small solutes to low-molecular-weight proteins [2, 3].
Hemodiafiltration Prescriptions: Patient Indication
Hemodiafi ltration could be indicated in all hemodialysis patients. High-volume
hemodiafi ltration techniques mark a new step towards mimicking the blood purifi ca-
tion of the native kidney. These techniques offer superior uremic substance removal
over a wider range of molecular sizes, yet require the use of biocompatible mem-
branes and ultrapure dialysate, which has been related to additional clinical benefi ts.
Recent large observational studies with robust adjustments for demographic and
comorbid confounding factors have shown an association of a lower mortality risk
with hemodiafi ltration using more than 15 liters of substitution fl uid [4, 5].
A number of studies have addressed the potential role played by larger
solutes or low-molecular-weight proteins in dialysis-related complications and
the potential clinical advantages offered by high-convection therapies. Specifi c
indications should be considered because hemodiafi ltration has been reported
to be effective against hyperphosphatemia, malnutrition, insomnia, irritability,
restless-leg syndrome, polyneuropathy, anemia, itching and joint pain, and may
prevent dialysis-associated amyloidosis.
Hyperphosphatemia
Hemodiafi ltration improves phosphate elimination and could be consid-
ered as a treatment option for hyperphosphatemia [6]. Lornoy et al. [7] reported
that treatment with online hemodiafi ltration in postdilution mode resulted in
higher phosphorus removal than hemodialysis.
Malnutrition
Anorexia in uremic patients has been related to an accumulation of uremic
substances. In uremic rats, Anderstam et al. [8] identifi ed toxins in the range of
1,000–5,000 Da, which led to suppression of food intake. Leptin, with a molec-
ular weight of 16,000 Da, may have an appetite-suppressing effect and could
accumulate in dialysis patients [9].
Anemia
Online hemodiafi ltration could improve erythropoietin response as a result
of the increased removal of medium- and large-sized molecules. Bonforte et al.
Optimizing the Prescription of Hemodiafi ltration 227
[10] showed an improvement in anemia in 32 patients with high-volume online
replacement fl uid. Osawa et al. [11] were able to lower the erythropoietin dose
with pull/push hemodiafi ltration. Maduell et al. [12] reported improved cor-
rection of anemia in 37 patients with lower erythropoietin doses when conven-
tional hemodiafi ltration (4 liters) was switched to online hemodiafi ltration (24
liters). Ward et al. [13] and Wizemann et al. [14] could not confi rm these obser-
vations in 24 and 23 patients, respectively, treated with online hemodiafi ltration
compared with 21 patients treated with high-fl ux hemodialysis and 21 patients
treated with low-fl ux hemodialysis.
Infectious Complications
Uremic patients have a signifi cant risk of infectious complications. Indeed,
these complications are the fi rst cause of hospitalization and the second cause
of death in hemodialysis patients. Several granulocyte-inhibiting proteins
are present in uremic patients, which may contribute to the high incidence
of infectious complications. Degranulation-inhibiting protein I and granulo-
cyte inhibitory protein II inhibit in vitro glucose uptake and chemotaxis of
polymorphonuclear leukocytes. Factor D decreases the complement-mediated
clearance of immune complexes and inhibits granulocyte degranulation. All
these uremic toxins are better removed with high-volume hemodiafi ltration
[13, 15].
Joint Pain
Maeda et al. [16] observed a signifi cant increase in upper arm movement
range and alleviation of shoulder joint pain in 30 patients after switching treat-
ment from hemodialysis to push/pull hemodiafi ltration (30 liters of convective
volume). The clinical observations of Kim et al. [17] support the hypothesis
that joint-pain-related substances may have a molecular size larger than 2-
microglobulin. These authors investigated the relationship between joint pain
relief and the removal pattern of low-molecular-weight proteins, and observed
higher removal rates of ␣1-microglobulin and ␣1-acid glycoprotein with online
hemodiafi ltration than with high-fl ux hemodialysis. Sato and Koga [18] also
observed a decrease in joint pain and signifi cant improvements in the range of
upper arm adduction and abduction movements when 6 patients undergoing
hemodialysis were changed to online hemodiafi ltration.
Dialysis-Related Amyloidosis
Using data from a Japanese dialysis patient registry, Nakai et al. [19] inves-
tigated which of a number of treatment modes was most effective for the treat-
ment of dialysis-related amyloidosis in 1,196 patients. When the risk of a worse
therapeutic effect for low-fl ux hemodialysis was stated as 1, the risk for patients
Maduell 228
using high-fl ux hemodialysis was 0.489, while that for online hemodiafi ltration
was 0.013.
Cardiovascular Stability
Convective treatments have been reported to provide superior cardiovascular
stability, thereby reducing intradialysis hypotension even in patients with increased
cardiovascular risk [20]. Donauer et al. [21] reported a reduction of hypotensive
side effects during online hemodiafi ltration and low-temperature hemodialysis.
In some patients with severe hypotension, we have observed [unpubl. data] an
improvement in predialysis blood pressure with high convective treatments.
Neurological Complications
Insomnia, irritability, restless-leg syndrome, polyneuropathy and pru-
ritus can be due to accumulation of medium-sized or large molecules.
Hemodiafi ltration with high-volume replacement fl uid improves these symp-
toms due to improved blood purifi cation [22, 23].
Blood and Dialysate Flow
The main limiting factors for infusion fl ow (QI) are blood fl ow (QB) and
transmembrane pressure, which rise in proportion to QI. Although some moni-
tors have a QI one third of the QB value, the maximum recommended QI is 25%
of the QB value in postdilution mode. Although online hemodiafi ltration can
be performed with every QB, a prescription of QB between 360 and 500 ml/min
allows a QI of between 80 and 125 ml/min.
In a routine 3 times weekly session, the recommended dialysate fl ow is
700–800 ml/min, although lower fl ows are possible.
Infusion Rate
Postdilution hemodiafi ltration is the most effi cient infusion mode for
obtaining maximum clearances of small and larger solutes, even though this
mode can increase the frequency of technical problems (hemoconcentration and
high transmembrane pressure). The predilution mode, while partially prevent-
ing technical problems, reduces the cumulative solute transfer as a consequence
of the diluted concentration [14, 24].
Simultaneous pre- and postdilution, i.e. mixed, hemodiafi ltration, could be
a highly effective technique to remove uremic toxins, while avoiding the disad-
vantages of the traditional infusion modes [25].
Optimizing the Prescription of Hemodiafi ltration 229
Vascular Access
A native fi stula is the best option for all hemodialysis modalities as well as
for hemodiafi ltration. However, the use of a native fi stula or graft has decreased
over the last decade, due to greater patient age and the increased prevalence of
cardiovascular disease and diabetes. For this reason, the use of permanent tun-
neled catheters has increased in the last few years (20–25% in our dialysis unit),
and the possibility of performing hemodiafi ltration with permanent catheters
has been considered.
In 1998, Canaud et al. [26] reported 7 patients with double-lumen perma-
nent catheters who underwent online hemodiafi ltration. During the last year,
we have treated 8 patients, aged 65 years or older, 3 men and 5 women, with
online hemodiafi ltration with a new generation of permanent tunneled cath-
eters. Dialysis parameters were a helixone dialyzer of 1.8 m2 surface area, blood
fl ow of 372 ml/min (range 300–450), dialysate fl ow of 800 ml/min, dialysis time
of 289 min and bicarbonate-buffered dialysate. The mean reinfusion volume
was 20.1 ⫾ 4 liters (range 18–25 liters).
Frequency
Daily dialysis experiences have shown excellent clinical results because a
higher frequency of dialysis is more physiological and decreases fl uctuations in
liquids, solutes and electrolytes. Improvements in comfort during and between
dialysis, as well as in clinical and biochemical parameters, anemia correction,
hypertension control, nutritional status and quality of life have been reported.
In a previous study, we combined the more physiological and effective
dialysis schedule (daily dialysis) with online hemodiafi ltration. Patients on stan-
dard 4- to 5-hour 3 times weekly online hemodiafi ltration were changed to 2-
to 2.5-hour 6 times weekly online hemodiafi ltration. The principal advantages
observed were excellent patient clinical tolerance, disappearance of postdialy-
sis fatigue, improvement of sleep disorders, reduction of phosphate binders,
improvement of nutritional status (body weight increased by more than 3 kg
after 1 year), better control of blood pressure without antihypertensive medica-
tions and partial regression of left ventricular hypertrophy [27].
In a pilot trial in children with daily online hemodiafi ltration [28], 5 chil-
dren were switched from standard online hemodiafi ltration (4 h, 3 times/week)
to daily online hemodiafi ltration (3 h, 6 times/week). This strategy led to a
reduction in blood pressure and an improvement in left ventricular size and
function, normalization of predialytic plasma phosphorus and improvements in
general well-being and dialysis acceptance.
Maduell 230
References
1 Depner TA: Uraemic toxicity: urea and beyond. Semin Dial 2001;14:246–251.
2 Clark WR, Gao D: Low-molecular weight proteins in end-stage renal disease: potential toxicity
and dialytic removal mechanism. J Am Soc Nephrol 2002;13(suppl 1):S41–S47.
3 Henderson LW, Clark WR, Cheung AK: Quantifi cation of middle molecular weight solute removal
in dialysis. Semin Dial 2001;14:294–299.
4 Canaud B, Bragg-Gresham JL, Marshall MR, Desmeules S, Gillespie BW, Depner T, Klassen P,
Port FK: Mortality risk for patients receiving hemodiafi ltration versus hemodialysis: European
results from the DOPPS. Kidney Int 2006;69:2087–2093.
5 Jirka T, Cesare S, Di Benedetto, Chang MP, Ponce P, Richards N, Tetta C, Vaslaky L: Mortality risk
for patients receiving hemodiafi ltration versus hemodialysis. Kidney Int 2006;70:1524.
6 Zehnder C, Gutzwiller JP, Renggli K: Hemodiafi ltration – A new treatment option for hyperphos-
phatemia in hemodialysis patients. Clin Nephrol 1999;52:152–159.
7 Lornoy W, De Meester JD, Becaus I, Billiouw JM, Van Malderen PA, Van Pottelberge: Impact of
convective fl ow on phosphorus removal in maintenance hemodialysis patients. J Ren Nutr 2006;
16:47–53.
8 Anderstam B, Mamoun A, Södersten P, Bergström J: Middle-sized molecule fractions isolated
from uremic ultrafi ltrate and normal urine inhibit ingestive behavior in the rat. J Am Soc Nephrol
1996;7:2453–2460.
9 Merabet E, Dagogo-Jack S, Coyne DW, Klein S, Santiago JV, Hmiel SP, Landt M: Increased plasma
leptin concentrations in end stage renal disease. J Clin Endocrinol Metab 1997;82:847–850.
10 Bonforte G, Grillo P, Zerbi S, Surian M: Improvement of anemia in hemodialysis patients treated
by hemodiafi ltration with high-volume on-line prepared substitution fl uid. Blood Purif 2002;20:
357–363.
11 Osawa S, Sakuraba N, Yamamoto H, Hisajima S. Clinical evaluation of HDF: especially effects on
EPO administration in HDF patients. Clin Pharmacol Ther 1997;7:1159–1162.
12 Maduell F, Pozo C, Garcia H, Sanchez L, Hdez-Jaras J, Albero D, Calvo C, Torregrosa l, Navarro
V: Change from conventional haemodiafi ltation to on-line haemodiafi ltration. Nephrol Dial
Transplant 1999;14:1202–1207.
13 Ward RA, Schmidt B, Hullin J, Hillebrand GF, Samtleben W: A comparison of on-line hemodia-
fi ltration and high-fl ux hemodialysis: a prospective clinical study. J Am Soc Nephrol 2000;11:
2344–2350.
14 Wizemann V, Lotz C, Techert F, Uthoff S: On-line haemodiafi ltration versus low-fl ux hemodialy-
sis: a prospective randomized study. Nephrol Dial Transplant 2000;15(suppl 1):S43–S48.
15 Haag-Weber M, Cohen G, Hörl WH: Clinical signifi cance of granulocyte-inhibiting proteins.
Nephrol Dial Transplant 2000;15(suppl 1):15–16.
16 Maeda K, Kobayakawa H, Fujita Y, Takai I, Morita H, Emoto Y, Miyazaki T, Shinzato T:
Effectiveness of push/pull hemodiafi ltration using large-pore membrane for shoulder joint pain in
long-term dialysis patients. Artif Organs 1990;14:321–327.
17 Kim ST, Yamamoto C, Asabe H, Sato T, Takamiya T: Online haemodiafi ltration: effective removal
of high molecular weight toxins and improvement in clinical manifestations of chronic haemodi-
alysis patients. Nephrology 1996;2(suppl 1):S183–S186.
18 Sato T, Koga N: Centralized on-line hemodiafi ltration system utilizing purifi ed dialysate as substi-
tution fl uid. Artif Organs 1998;22:285–290.
19 Nakai S, Iseki K, Tabei K, Kubo K, Masakane I, Fushimi K, Kikuchi K, Shinzato T, Sanaka T,
Akiba T: Outcomes of hemodiafi ltration based on Japanese dialysis patient registry. Am J Kidney
Dis 2001;38(suppl 1):S212–S216.
20 Mion M, Kerr PG, Argiles A, Canaud B, Flavier JL, Mion CM: Hemodiafi ltration in high-cardio-
vascular-risk patients. Nephrol Dial Transplant 1992;7:453–455.
21 Donauer J, Schweiger C, Rumberger B, Krumme B, Böhler J: Reduction of hypotensive side effects
during online-haemodiafi ltration and low temperature haemodialysis. Nephrol Dial Transplant
2003;18:1616–1622.
Optimizing the Prescription of Hemodiafi ltration 231
22 Mucsi I, Molnar MZ, Ambrus C, Szeifert L, Kovacs AZ, Zoller R, Barotfi S, Remport A, Novak
M: Restless legs syndrome, insomnia and quality of life in patients on maintenance dialysis.
Nephrol Dial Transplant 2005;20:571–577.
23 Zakrzewska-Pniewska B, Jedras M: Is pruritus in chronic uremia patients related to peripheral
somatic and autonomic neuropathy? Study by R-R interval variation test (RRIV) and by sympa-
thetic skin response (SSR). Neurophysiol Clin 2001;31:181–193.
24 Maduell F, García H, Hdez-Jaras J, Calvo C, Navarro V: Comparación de la infusión predilucional
versus postdilucional en HDF en línea. Nefrología 1998;18(suppl 3):49.
25 Pedrini L, De Cristofaro V, Pagliari B, Samà F: Mixed predilution and postdilution online hemo-
diafi ltration compared with traditional infusion modes. Kidney Int 2000;58:2155–2165.
26 Canaud B, Bosc JY, Leray H, Stec F, Argiles A, Leblanc M, Mion C: On-line haemodiafi ltration:
state of the art. Nephrol Dial Transplant 1998;13(suppl 5):3–11.
27 Maduell F, Navarro V, Torregrosa E, Rius A, Dicenta F, Cruz MC, Ferrero JA: Change from thrice
weekly on-line hemodiafi ltration to short daily on-line hemodiafi ltration. Kidney Int 2003;64:
305–313.
28 Fischbach M, Terzic J, Laugel V, Dheu C, Menouer S, Helms P, Livolsi A: Daily on-line haemodia-
fi ltration: a pilot trial in children. Nephrol Dial Transplant 2004;19:2360–2367.
Francisco Maduell Canals, MD
Servicio de Nefrología, Hospital Clínic Barcelona
C/Villarroel, 170
ES–08006 Barcelona (Spain)
Tel. ⫹34 93 227 5400, Fax ⫹34 93 454 6033, E-Mail fmaduell@clinic.ub.es
Author Index
232
Aljama, P. X, 210
Bonucchi, D. 80
Canaud, B. X, 110, 216
Cappelli, G. 80
Clark, W.R. 20
De Amicis, S. 80
Di Filippo, S. 185
Ferramosca, E. 138
Fiore, G.B. 177
Ghezzi, P.M. 94
Grandi, F. 138
Guarnieri, F. 138
Henderson, L.W. 1
Hoenich, N.A. 57
Ledebo, I. 87
Locatelli, F. 185
Maduell, F. 225
Maeda, K. 169
Manzoni, C. 185
Martin-Malo, A. 210
Panichi, V. 194
Pedrini, L.A. 123
Pizzarelli, F. 131
Polaschegg, H.-D. 68
Potier, J. 153
Ramirez, R. 210
Ricardi, M. 80
Rocha, E. 20
Ronco, C. X, 9, 20, 34,
177
Roy, T. 68
Santoro, A. 138
Shinzato, T. 169
Tattersall, J. 201
Tetta, C. 194
von Albertini, B. 161
Wizemann, V. 103
Wratten, M.L. 94
Yamashita, A.C. 50
Zerbi, S. 123
Subject Index
233
Acetate-free biofi ltration (AFB)
AN69ST dialysis membrane 141
blood volume tracking 148–150
historical perspective 14, 139, 140
outcomes
hemodynamic stability 142–144
metabolic acidosis correction 142
nutritional status 142
potassium-profi led acetate-free
biofi ltration 144–148
prospects 150
technical aspects 140, 141
Adsorption
effects on 2-microglobulin
clearance 204
membranes 23, 24
Advanced glycation end products (AGEs),
hemodiafi ltration fi ndings 189, 190
Albumin, membrane polyvinylpyrrolidione
content and loss 51–55
Amyloidosis, hemodiafi ltration
prescription 227, 228
Anemia, hemodiafi ltration fi ndings 187,
218, 219, 226, 227
Arteriovenous hemodiafi ltration, historical
perspective 17
B. Braun online hemodiafi ltration system,
features 74
Bellco online hemodiafi ltration system,
features 74, 133
Biofi ltration, historical perspective 11
Blood pressure, hemodiafi ltration
fi ndings 218
Blood volume tracking (BVT), acetate-free
biofi ltration 148–150
Cardiovascular risk factors
hemodiafi ltration fi ndings 189, 190, 219
hemodialysis fi ndings 195
Cardiovascular stability
acetate-free biofi ltration 142–144
hemodiafi ltration fi ndings 187, 188, 217,
218, 228
Classical hemodiafi ltration, see Low-
effi ciency hemodiafi ltration
Complement D, hemodiafi ltration
effects 196
Continuous venovenous hemodiafi ltration,
historical perspective 17
Convection
diffusion interference in
hemodiafi ltration 95
history of study 2, 10
internal hemodiafi ltration 178, 180, 183
2-microglobulin removal 105, 106, 202,
203
optimization 125
C-reactive protein (CRP)
hemofi ltrate reinfusion and reduction of
levels 99
secretion 195
Cross-fi ltration, hollow-fi ber
hemodialyzers 41–47
Subject Index 234
Deionizer, water purifi cation 83
Dialysate fl ow, hemodiafi ltration
prescription 228
Diffusion
convection interference in
hemodiafi ltration 95
2-microglobulin removal 202
Double high-fl ux hemodiafi ltration
historical perspective 16, 161, 162
outcomes 165, 166
principles 116
safety 166
technique 162–164
Dyslipidemia, hemodiafi ltration
fi ndings 219
Electrolyte solution, online production 5, 6,
16
Endothelial function, hemodiafi ltration
fi ndings 213, 214
Endotoxin, limits 76, 77
End-to-pressure drop, calculation 46
Equivalent renal clearance (EKR),
calculation 207
European Dialysis Outcomes and Practice
Patterns Study, hemodiafi ltration
benefi ts 124
Fluid mechanics
blood compartment 35–39
dialysate compartment 39–41
FME online hemodiafi ltration system,
features 71–74
Frequency
hemodiafi ltration prescription 229
high-effi ciency hemodiafi ltration 108
Gambro online hemodiafi ltration system,
features 69–71, 78
Growth retardation, hemodiafi ltration
fi ndings 219, 220
Hagen-Poiseuille formula 46
Hemodialysis, historical perspective 1, 2
Hemodynamic stability, see Cardiovascular
stability
Hemofi ltrate reinfusion (HFR)
benefi ts 99–101
historical perspective 96
mechanisms 98, 99
principles 96
ultrafi ltrate characteristics 97, 98
HEMO Study
high-fl ux versus conventional
hemodialysis 185, 186
survival outcomes 191
High-effi ciency hemodiafi ltration
defi nition 104
duration 108
fl ow rate complications 106, 107
frequency 108
rationale 104, 105
High-volume hemodiafi ltration, historical
perspective 14
Hollow-fi ber membranes
adsorption 23, 24
classifi cation
cellulosic membranes 21
synthetic membranes 21
design 59, 60
performance by device 64
permeability 22
pore size distribution 22, 23
solute clearance
blood water and plasma clearance 25,
26
diffusive solute clearance
determinants
blood compartment 27–29
dialysate compartment 30–32
mass transfer resistance 26, 27
mass removal rate versus solute
clearance 26
whole-blood clearance 24, 25
Hospitalization, hemodiafi ltration morbidity
fi ndings 220
Hydraulic permeability, hemodiafi lter
performance 63, 65
Hyperkalemia, potassium-profi led acetate-
free biofi ltration 146–148
Hyperphosphatemia, hemodiafi ltration
fi ndings 187, 226
Hypotension, hemodiafi ltration
fi ndings 217, 218
Subject Index 235
Infection, hemodiafi ltration
prescription 227
Infl ammation
chronic kidney disease 210, 211
hemodiafi ltration fi ndings
CD14+CD16+ cells 211, 212
cytokines 197
endothelial function 213, 214
hemocompatibility 218
hemodialysis patients 195
Infusion rate, hemodiafi ltration
prescription 228
Interleukin-6 (IL-6), hemodiafi ltration
effects 197
Internal hemodiafi ltration
applications 183
convection 178, 180, 183
historical perspective 15
internal fi ltration
mathematical estimation 181, 182
measurement 180, 181
internal fi ltration/backfi ltration 178
Joint pain, hemodiafi ltration
prescription 227
Kt/V, equilibrated 206, 207
Low-effi ciency hemodiafi ltration,
defi nition 104
Malnutrition, hemodiafi ltration
fi ndings 219, 226
Mass removal rate versus solute
clearance 26
Membrane Permeability Outcome (MPO)
Study, fi ndings 186, 191
Membranes, see also Hollow-fi ber
membranes
acetate-free biofi ltration 141
historical perspective 3, 10
leak safety in online
hemodiafi ltration 76, 77
materials 58, 59
performance parameters
biocompatibility 63
hydraulic permeability 63, 65
solute transport 60, 62, 63–65
polyvinylpyrrolidione content,
biocompatibility and solute loss
effects 51–55
Membrane ultrafi ltration coeffi cient,
ultrafi ltration rate and transmembrane
pressure relationship 45
2-Microglobulin
adsorption effects on clearance 204
clearance measurement 204
convection and removal 105, 106, 202,
203
convective plus diffusive clearance 203
diffusive clearance 202
equivalent renal clearance 207
hemodiafi ltration fi ndings 188, 189, 207,
208
mid-dilution hemodiafi ltration
effects 157, 158
two-compartment kinetics 205, 206
Mid-dilution hemodiafi ltration
benefi ts 158, 159
fi lter design 60
historical perspective 15, 16
outcomes
low-weight-molecule removal 156
middle-molecule removal 157, 158
safety 157
study design 155
principles 117, 154, 155
transmembrane pressure 155, 157
Middle molecules, see 2-Microglobulin
Mixed-dilution hemodiafi ltration
convection optimization 125
infusion rate 127, 128
outcomes 128, 129
principles and confi guration 125–128
rationale 124, 125
transmembrane pressure feedback control
and profi ling 125–128
Nephros online hemodiafi ltration system,
features 74
Neurological complications,
hemodiafi ltration prescription 228
Nikkiso online hemodiafi ltration system,
features 74
Subject Index 236
Online hemodiafi ltration, see also specifi c
techniques
B. Braun system 74
Bellco system 74, 133
benefi ts 120, 121, 132
conventional technique 113–115
FME system 71–74
follow-up and monitoring 119
Gambro system 69–71, 78
hemodiafi lter 118
historical perspective 14, 68, 69
hygiene handling 112, 113
indications 117, 118
microbiological monitoring 119, 120
Nephros system 74
Nikkiso system 74
prescription 118
quality monitoring 113
rationale 111
safety aspects 74–78
technical requirements 111, 112
vascular access 118
water quality 111, 112
Oxidative stress, hemodiafi ltration
fi ndings 219
Paired hemodiafi ltration
benefi ts 135
historical perspective 3, 4, 11, 15
outcomes 134, 135
principles 116, 133
Polyvinylpyrrolidione (PVP),
biocompatibility and solute loss effects in
membranes 51–55
Potassium-profi led acetate-free biofi ltration,
see Acetate-free biofi ltration
Prevalence, hemodiafi ltration use by
country 6, 7
Protein binding, middle-molecule clearance
effects 204
Push-pull hemodiafi ltration
blood fl ow 171
costs 175
dialysate preparation 174
effectiveness 175
equipment 170
historical perspective 16
principles 116, 170
replacement mode 173
replacement volume 173
solute removal 174
transmembrane pressure control
171–173
Reinfusion, see Hemofi ltrate reinfusion
Replacement fl uid
bagged fl uids 91, 92
characteristics for various forms of
hemodiafi ltration 88, 89
composition 88–90
quality 90, 95
volume 91, 95
Reverse osmosis (RO), water
purifi cation 83
Rhodial 75 system, historical
perspective 10
Solute clearance
blood water and plasma clearance 25, 26
cross-fi ltration 41–47
determinants of diffusive solute clearance
blood compartment 27–29
dialysate compartment 30–32
mass transfer resistance 26, 27
fl ow rate dependence 106, 107
fl uid mechanics
blood compartment 35–39
dialysate compartment 39–41
mass removal rate versus solute
clearance 26
polyvinylpyrrolidione content effects in
membranes 51–55
push-pull hemodiafi ltration 174
two-compartment model 204, 205
whole-blood clearance 24, 25
Solute transport, hemodiafi lter
performance 60, 62, 63–65
Survival, hemodiafi ltration outcomes 190,
191, 211, 220, 221
Transmembrane pressure (TMP)
calculation 41
gradient distribution 37
internal hemodiafi ltration 178
Subject Index 237
mid-dilution hemodiafi ltration 155, 157
mixed-dilution hemodiafi ltration
feedback control and profi ling
125–128
push-pull hemodiafi ltration control
171–173
ultrafi ltration rate and membrane
ultrafi ltration coeffi cient
relationship 45
Transmittance coeffi cient, calculation 25
Treatment adequacy, hemodiafi ltration 65,
66
Tumor necrosis factor-␣ (TNF-␣),
hemodiafi ltration effects 197
Ultrafi ltration rate
hemodialysis versus
hemodiafi ltration 87, 88
selection factors 95
transmembrane pressure and membrane
ultrafi ltration coeffi cient
relationship 45
Vascular access, online
hemodiafi ltration 118, 229
Water fl ux, calculations 42
Water quality
harmonization of standards 81
online hemodiafi ltration 111, 112
quality assurance 85
recommendations 82
ultrapurifi cation 82–84
