Content uploaded by Chris Harrington
Author content
All content in this area was uploaded by Chris Harrington on May 22, 2015
Content may be subject to copyright.
1 23
Analytical and Bioanalytical
Chemistry
ISSN 1618-2642
Anal Bioanal Chem
DOI 10.1007/s00216-015-8631-4
Feasibility of asymmetric flow field-flow
fractionation coupled to ICP-MS for the
characterization of wear metal particles
and metalloproteins in biofluids from hip
replacement patients
Katrin Loeschner, Chris F.Harrington,
Jacque-Lucca Kearney, David J.Langton
& Erik H.Larsen
1 23
Your article is protected by copyright and
all rights are held exclusively by Springer-
Verlag Berlin Heidelberg. This e-offprint is
for personal use only and shall not be self-
archived in electronic repositories. If you wish
to self-archive your article, please use the
accepted manuscript version for posting on
your own website. You may further deposit
the accepted manuscript version in any
repository, provided it is only made publicly
available 12 months after official publication
or later and provided acknowledgement is
given to the original source of publication
and a link is inserted to the published article
on Springer's website. The link must be
accompanied by the following text: "The final
publication is available at link.springer.com”.
RESEARCH PAPER
Feasibility of asymmetric flow field-flow fractionation coupled
to ICP-MS for the characterization of wear metal particles
and metalloproteins in biofluids from hip replacement patients
Katrin Loeschner
1
&Chris F. Harrington
2
&Jacque-Lucca Kearney
2
&
David J. Langton
3
&Erik H. Larsen
1
Received: 22 January 2015 /Revised: 4 March 2015 / Accepted: 10 March 2015
#Springer-Verlag Berlin Heidelberg 2015
Abstract Hip replacementsare used to improve the quality of
life of people with orthopaedic conditions, but the use of
metal-on-metal (MoM) arthroplasty has led to poor outcomes
for some patients. These problems are related to the generation
of micro- to nanosized metal wear particles containing Cr, Co
or other elements, but the current analytical methods used to
investigate the processes involved do not provide sufficient
information to understand the size or composition of the wear
particles generated in vivo. In this qualitative feasibility study,
asymmetric flow field-flow fractionation (AF
4
) coupled with
inductively coupled plasma mass spectrometry (ICP-MS) was
used to investigate metal protein binding and the size and
composition of wear metal particles present in serum and hip
aspirates from MoM hip replacement patients. A well-
established HPLC anion exchange chromatography (AEC)
separation system coupled to ICP-MS was used to confirm
the metal–protein associations in the serum samples. Off-line
single particle ICP-MS (spICP-MS) analysis was used to con-
firm the approximate size distribution indicated by AF
4
of the
wear particles in hip aspirates. In the serum samples,
AF
4
–ICP-MS suggested that Cr was associated with transfer-
rin (Tf) and Co with albumin (Alb) and an unidentified spe-
cies; AEC–ICP-MS confirmed these associations and also in-
dicated an association of Cr with Alb. In the hip aspirate sam-
ple, AF
4
–ICP-MS suggested that Cr was associated with Alb
and Tf and that Co was associated with Alb and two uniden-
tified compounds; AEC analysis confirmed the Cr results and
the association of Co with Alb and a second compound.
Enzymatic digestion of the hip aspirate sample, followed by
separation using AF
4
with detection by UV absorption
(280 nm), multi-angle light scattering and ICP-MS, suggested
that the sizes of the Cr-, Co- and Mo-containing wear particles
in a hip aspirate sample were in the range 40–150 nm. Off-line
spICP-MS was used to confirm these findings for the Co- and
Cr-containing nanoparticles. Whilst limited in scope, the re-
sults are sufficient to show the interaction of ions with trans-
port proteins and give an indication of particle size, providing
useful pathological indices. As such, the methods indicate a
new way forward for in vivo investigation of the processes
which lead to tissue necrosis and hip loosening in patients
with MoM hip replacements.
Keywords Nanoparticles .Asymmetric flow field-flow
fractionation .AF
4
.ICP-MS .Single particle ICP-MS .
Clinical/biomedical analysis .Anion exchange
chromatography .Metal-on-metal hip replacement
Introduction
Hip replacements have enhanced the lives of many people
across the world since their first introduction in the 1930s.
Although there are different implant systems currently in
use, based on the materials used as the bearing surface, they
fall into four main categories: metal on plastic (MoP), ceramic
Electronic supplementary material The online version of this article
(doi:10.1007/s00216-015-8631-4) contains supplementary material,
which is available to authorized users.
*Chris F. Harrington
chris.harrington1@nhs.net
1
Division of Food Chemistry, National Food Institute, Technical
University of Denmark, Mørkhøj Bygade 19, 2860 Søborg, Denmark
2
Supra-Regional Assay Service (SAS) Trace Element Centre, Surrey
Research Park, 15 Frederick Sanger Road, Guildford, Surrey GU2
7YD, UK
3
University Hospital of North Tees, Hardwick Road,
Stockton-on-Tees, Cleveland TS19 8PE, UK
Anal Bioanal Chem
DOI 10.1007/s00216-015-8631-4
Author's personal copy
on ceramic (CoC), ceramic on plastic (CoP) and metal on
metal (MoM) [1,2]. There are two main approaches to hip
replacement surgery using MoM systems: total hip replace-
ment (THR), where the ball of the femur is removed and
replaced with a stemmed femoral ball component, and
resurfacing arthroplasty, where the femur is capped or re-
surfaced with a hollow femoral ball component [1]. A variety
of materials are used in the construction of joint replacements,
but the most common is a cobalt (Co)–chromium (Cr)–mo-
lybdenum (Mo) alloy (ASTM F75) containing Co, Cr and Mo
in the ratio 60:30:7 [1]. Other materials and components of the
implant used include titanium (Ti) stems and Ti cup shells
along with plastic and ceramic parts.
All materials that are used in hip implants will wear with
use and release wear particles [3]. This is true of MoP and
MoM implants and although MoM articulations release a
greater number of metal particles compared to the number of
polyethylene particles released by MoP joints, the actual vol-
umetric wear of MoM joints is much lower. Generally, MoM
joints release particles in the nanometre range whereas MoP-
generated polyethylene particles are in the micrometre diam-
eter size range [4]. The wear debris from MoM joints consists
of nano- to micro-sized particles and soluble forms of the
metal ions that the joint is composed of, and they are generated
by a number of processes which fall into three broad catego-
ries: mechanical wear, surface corrosion or a combination of
both [4]. Wear products are found to a large extent in the
synovial fluid surrounding the joint and in tissue close to the
implant, and further accumulation has been observed in re-
gional lymph nodes, liver and spleen [4]. Free or phagocy-
tosed wear particles can be transported within the lymphatic
system, whereas wear debris as ions or particles can be dis-
tributed in the vascular system [4]. These clinical findings
highlight the need for methods which can characterize the size
and composition of wear metal particles, because this data will
provide a better understanding of their fate, transport and
effect within the body.
Blood metal concentrations of Co and Cr are used in pa-
tients with MoM hip replacements as biomarkers to screen
patients to assess proper function of the implanted device. In
2010, the UK Medicines and Healthcare Products Regulatory
Agency (MHRA) issued guidance [5] to clinicians working in
the orthopaedic area that they should measure Co and Cr in
whole blood from hip replacement patients to ascertain wheth-
er the prosthesis was failing and if the levels were greater than
7μg/L for Co or Cr. In such cases, further testing and, if
appropriate, imaging of the implant should be undertaken to
determine whether the hip is functioning correctly and wheth-
er signs of localized adverse toxicological effects are present.
This guideline originated from clinical observations of some
patients having joint pain, discomfort and swelling [6].
Further work has revealed that the wear particles generated
in the surroundings of the hip replacement can lead to
localized inflammation and loosening of the hip, due to necro-
sis of the tissue and bone [3,4].
Whereas typical blood levels of Co and Cr in patients with
well-functioning hip implants are approximately 1.7 and
2.3 μg/L, respectively [1], the experience from the Supra-
Regional Assay Service (SAS) laboratory at Guildford is that
approximately 20 % of Cr and 7 % of Co blood results are
greater than the MHRA cut-off and blood concentrations
range from undetected to 150 and 75 μg/L for Co and Cr
respectively; the higher level for Co is thought to be due to
the composition of the alloy used. Very high concentrations of
CoandCrhavebeenreported[1], with some patient samples
containing upwards of 387 μg/L Co and 179 μg/L Cr in blood.
In comparison, much greater concentrations of 400 mg/L Cr
and 22 mg/L Co have been determined in the synovial fluid
surrounding the hip replacement. As both Co and Cr are
known toxicants, their systemic toxicity, genotoxic and
mutagenic effects have been called into question at these
concentrations [4].
Some work on the binding of Co and Cr to serum proteins
using anion exchange (AE) separation of the serum proteins
via HPLC-inductively coupled plasma mass spectrometry
(ICP-MS) has been published [7]; however, this work relates
to patient samples where the concentration in the blood of Co
and Cr was low (<3 μg/L) for a hip implant patient and sig-
nificantly below the 7-μg/L MHRA cut-off value. Analysis of
the association of Cr with serum proteins was not possible in
the patient samples due to the low concentrations present, but
Cr spiking showed it was associated with transferrin (Tf). In
the case of Co, this element was associated with albumin (Alb)
in the patient samples. In the current work, this AE separation
procedure was used to independently confirm the metal–pro-
tein associations determined using the protein asymmetric
flow field-flow fractionation (AF
4
)–ICP-MS separation
conditions.
Although the wear particles from the hip implant are
known to cause inflammation and loosening of the implant,
little information is available on the particle size distribution
or composition in vivo that causes these clinical effects. The
information that is available shows particles in the size range
6–800 nm, which includes larger agglomerates, differently
shaped particles with the majority containing Cr either as an
oxide, hydroxide or phosphate (the latter occurs in non-
synovial environments only) [4,8,9], but also particles com-
posed of Co, Cr, Fe, Ni and Mo were detected [9]. The latter
findings were based on in vitro studies involving particles
generated from the mechanical testing of the implants and
from examination of patient histopathological slides using
transmission and scanning electron microscopy (TEM and
SEM) and X-ray spectroscopy.
In order to further investigate the binding of wear metal
ions to proteins or their existence as particles in vivo, analyt-
ical methodologies that allow their separation by size and
K. Loeschner et al.
Author's personal copy
selective detection by atomic spectrometric techniques are
needed. One possible separation technique is AF
4
which is a
flow-based separation technique, well suited to the separation
of proteins and nanosized particles in complex matrices. The
theory of AF
4
is described elsewhere [10], but practically, the
sample is injected in a channel which consists of a closed plate
at the top (depletion wall) and a semi-permeable membrane
(accumulation wall) supported by a frit at the bottom. Particles
and proteins are driven through the channel by the primary
channel flow, and the separation is achieved by a cross flow
which is applied perpendicularly to the primary channel flow.
The cross flow originates within the channel due to a pressure
difference across the membrane/frit assembly. Separation oc-
curs for sub-micrometre-sized particles in the so-called normal
or Brownian mode. More details regarding the AF
4
cell can be
found in reference [11].
A major advantage of AF
4
is that it can separate metals
bound to proteins and metal nanoparticles using the same
separation apparatus, although different separation conditions
are necessary. The current study focuses ontwo proteinslikely
to be involved in the transport of Co and Cr in vivo: Alb, the
protein with the highest concentration in human serum and
known to bind trace elements as well as other compounds,
and Tf, another important metal (Fe)-transporting protein, in-
volved in the formation of red blood cells in bone marrow. By
using ICP-MS as the detector, it is possible to characterize
metals binding to serum proteins and the metal composition
of differently sized wear particles, including the composition
of particles corroded post-wear. The use of AF
4
coupled to
ICP-MS for the separation of Alb and Tf standards has been
reported previously [12].
In this study, we apply for the first time the hyphenated
AF
4
–ICP-MS system for separation and selective detection
of Co, Cr and Mo associated with proteins in serum and hip
aspirate samples, as well as wear particles in hip aspirate sam-
ples. Hip aspirate samples are highly heterogeneous fluids and
contain a mixture of tissue, cells, haemolyzed red blood cells,
proteins and high concentrations of wear metals, particularly
Co and Cr. Originating from intimate contact with the pros-
theses, they provide an important insight into the interface
between the hip replacement and human tissue. A well-
proven AEC–ICP-MS method was used to confirm the met-
al–protein associations found using AF
4
–ICP-MS. By using
the same AF
4
–ICP-MS apparatus with different fractionation
conditions and including an enzymatic sample preparation
step, the particle size and composition in the hip aspirate sam-
ples were investigated. The particle size range was confirmed
using off-line spICP-MS measurements. The characterization,
both of how ions of these metals interact with proteins in
human serum samples and the size and composition of the
Co–Cr–Mo wear particles, is an important step in understand-
ing the processes leading to the negative effects some patients
suffer. This feasibility study highlights the potential for using
AF
4
–ICP-MS approaches to achieve a better understanding of
MoM wear particles in human samples.
Materials and methods
Materials, reagents and samples
Serum samples were supplied to the SAS laboratory in
Guildford in secondary tubes after sample preparation in the
hospital requesting the analysis; this involved taking the
whole blood sample into red-topped BD Vacutainer® blood
collection tubes (Becton, Dickinson and Co., Plymouth, UK)
and allowing the blood to clot, before centrifugation of the
sample for 5 min at 3000 rpm to separate the red blood cells
from the serum, which was thenwithdrawnusing a pipette and
placed in a secondary sample tube. Hip aspirate samples were
collected in the clinic from the fluid accumulated around the
hip prosthesis using a syringe, and the obtained sample was
stored in a large-diameter clear plastic container or white-
topped screw top Universal vials (Teklab, UK).
All the patient samples used in this work were from a single
hospital location, University Hospital of North Tees, Stockton,
Cleveland, TS19 8PE, UK, and mailed overnight at ambient
temperature to the SAS laboratory by first class delivery; on
arrival, the samples were refrigerated (4–8 °C) until analysis.
After routine analysis for Co and Cr, the samples were
anonymized and samples covering a suitable concentration
range for Co and Cr were chosen for further investigation.
CertiPUR-certified metal calibration standards of Cr, Co
and Mo at a concentration of 1000 μg/mL in 2 % (v/v)nitric
acid and traceable to NIST (Merck, Darmstadt, Germany)
were used for spiking and total elemental measurements.
Germanium 1000 μg/mL (Aristar, BDH, UK) was used as
the internal standard for the total metal assay and prepared
by dilution with 0.5 % (v/v) nitric acid. Multi-elemental tune
solution A 10 μg/L (Thermo Fisher, Hemel Hempstead, UK)
containingBa, Li, In, Pb and Tl was used to set up the ICP-MS
instrument. Two whole blood and two urine internal quality
control (IQC) materials were used for the total elemental as-
say: Seronorm trace elements whole blood L-2 and L-3 (Sero,
Billingstad, Norway) and Lyphochek urine materials control
L1 and L2 (Bio-Rad, Hemel Hempstead, UK).
Human Tf (iron content 300–600 ppm, purity by agarose
electrophoresis, T3309-500MG), human serum Alb (lyophi-
lized powder, A3782-1G Lot#090M7001V, >99 % purity) and
bovine Alb (lyophilized powder, monomer≥97 %, A1900-
250MG) were obtained from Sigma-Aldrich (St. Louis, MO,
USA, or Dorset, UK). The mobile phase for HPLC consisted
of Trizma® buffer: HCl >99.0 % and base ≥99.9 % (Sigma-
Aldrich, Dorset, UK) and ammonium acetate analytical re-
agent grade (Fisher Scientific, UK). All deionized reverse os-
mosis water used in the generation of buffers and standards
Characterization of MoM hip wear metal particles by AF
4
Author's personal copy
was >18.2 MΩcm resistivity (Millipore, Synergy Water
Purification System).
Ultrapure water was obtained from a Millipore Element
apparatus (Millipore, Milford, MA, USA) and used through-
out the AF
4
work, and carrier liquids were produced by dis-
solving either ammonium nitrate or sodium dodecyl sulphate
(SDS) ≥98.5 % purity (both ReagentPlus, Sigma-Aldrich, St.
Louis, MO, USA) in the ultrapure water. The AF
4
accumula-
tion wall used was a NADIR® regenerated cellulose mem-
brane with a molecular weight cut-off of 10 kDa (Wyatt
Technology, Dernbach, Germany).
Prior to characterization of the wear metal particles, the hip
aspirate sample was enzymatically digested to liberate any
organically bound particles and remove any protein aggrega-
tions. Proteinase K from Engyodontium album (Sigma-
Aldrich, St. Louis, MO, USA), a serine protease with a mo-
lecular mass 28,930 Da, a broad pH range of 7.5–12 and a
temperature optimum of 37 °C for enzyme activity, was used
for this purpose. The enzyme solution was prepared [13], by
dissolving proteinase K to a final concentration of 3 mg/mL in
50 mM ammonium bicarbonate buffer (pH 7.4) containing
5 mg/mL SDS and 0.2 mg/mL sodium azide (≥98 % purity,
Sigma-Aldrich, St. Louis, MO, USA) used as antimicrobial
agent.
Instrumental analysis
Measurement of the total metal concentration by ICP-MS
The concentrations of Co and Cr in the serum and hip aspirate
samples were measured using an X2 series ICP-MS instru-
ment (Thermo Fisher Scientific, Hemel Hempstead, UK) in
collision cell mode using helium containing 7 % hydrogen
(4 mL/min) in kinetic energy discrimination (KED) mode
using an ESI Fast SC autosampler (ESI, Omaha, NE), and
the settings are detailed in Table 1. In some samples, Mo
was measured using the same methodology. The instrument
was tuned for sensitivity (cps), mass accuracy and resolution,
using a 10-μg/L tune solution, so as to achieve the instrument
manufacturer’s specifications. Standards in the range 0.1–
10 μg/L spiked with 200 μL horse blood tomatrix match them
to the samples were prepared using a solution of Ge, the in-
ternal standard. Internal quality control (IQC) solutions of
blood and urine were used throughout the analytical run to
ensure accurate performance, and the calibration data was
only accepted if the IQC values were within the certificated
ranges. No hip aspirate IQCs are available.
Protein and particle separation and detection
by AF
4
–ICP-MS
The samples were centrifuged prior to AF
4
analysis using a
micro-centrifuge (Eppendorf MiniSpin, Eppendorf AG,
Hamburg, Germany). For enzymatic digestion of the hip as-
pirate sample prior to nanoparticle analysis, a water bath with
a Heto HMT 200 RS thermostat and a Telesystem HP15 mag-
netic stirrer (Holm & Halby, Allerød, Denmark) were used.
The AF
4
system consisted of a HPLC pump (G1311A), an
Agilent 1200 series autosampler (G1329A) equipped with
900 μL injection loop (Agilent Technologies, Santa Clara,
CA, USA), an Eclipse™3AF
4
flow control module and a
short channel-type AF
4
separation channel (Wyatt
Technology Europe GmbH, Dernbach, Germany). The chan-
nel had a trapezoid shape with a length from inlet to outlet of
172 mm, a length from inject port to outlet of 152 mm and a
width of 24 mm at the inlet and 3 mm at the outlet port. The
width at the sample inject port was 21.5 mm. The length of the
injection capillary tubing was 280 mm with an inner diameter
of 0.12 mm. The area of the accumulation wall was
2363 mm
2
.
After changing the membrane or when the carrier liquid
composition was changed, the system was flushed with a flow
of 1 mL/min for 60 min to allow equilibration. The time t
extra
required for the sample to travel through external tubing was
subtracted from the apparent retention time. The value t
extra
was determined based on the length and inner diameter of the
external tubing and the applied detector flow rate.
Following separation by AF
4
, various detectors were used
to collect information about the fractioned eluting material. A
series 1200 diode array detector (DAD, Agilent G1315A) was
used to record the absorbance signal at pre-defined wave-
lengths. A DAWN® HELEOS™(Wyatt Technology Europe
GmbH, Dernbach, Germany) multi-angle light scattering
(MALS) detector with a linear polarized laser light at
658 nm was used to record the light scattering signal at an
observation angle of 90°. The MALS detector was set to a
sampling time interval of 1 s per data point. The final detector
in the hyphenated system was an ICP-MS instrument (7500ce,
Agilent Technologies, Japan). The typical instrumental
settings are presented in Table 1.
Protein separation by AF
4
The applied AF
4
programme for the separation of serum pro-
teins, including the composition of the carrier liquid, is pre-
sented in Table 2. They represent typical conditions for the
separation of proteins [10], and similar conditions have been
used previously to separate Alb and Tf standards [12].
Standards of Alb and Tf were prepared by dissolving the pro-
teins in aqueous carrier liquid (50 mM NH
4
NO
3
,pH7.2–7.4)
to a final concentration ofeach protein of approximately 2 mg/
mL. Serum samples were diluted fivefold with carrier liquid,
and a volume of V
inj
=5 μL was injected in the AF
4
channel.
Basedonanestimatedproteinconcentrationinserumof
50 mg/mL, the total protein mass injected was 50 μg. Before
dilution, the hip aspirate samples were centrifuged at
K. Loeschner et al.
Author's personal copy
5000 rpm (1677×g) for 15 min to remove coarse organic con-
stituents that would block the system. Afterwards, the samples
were diluted fivefold with carrier liquid, and a volume of V
inj
=
5μL was injected in the AF
4
channel. When required for
confirmation of binding, spiked serum samples were prepared
by incubating them with an aliquot of a Cr and Co standard
(50 ng/mL). Results were processed off-line and plotted in
OriginPro 8.1 (OriginLab Corporation, MA, USA).
Protein separation by AEC–ICP-MS
HPLC AE separations were performed with a Waters Alliance
2690 separation module (Waters, Herts, UK) and an AE col-
umn: Mono Q 5/50 GL, 50×5 mm, mono-beads matrix,
10 μm particle size, glass casing, 460 psi max pressure (GE
Healthcare Sciences, Bucks, UK). For ICP-MS detection, the
same type of instrument (7500ce, Agilent Technologies,
Japan) as for AF
4
was used and the typical instrumental set-
tings are presented in Table 1. Samples were prepared by a 1:5
dilution with 50 mM Tris buffer (pH 7.4). HPLC parameters
were as follows: gradient elution; buffer A (50 mM Tris (pH
7.4)) and buffer B (buffer A with 1 M ammonium acetate (pH
7.4)), linear increase of buffer B (0–100 %), for 10 min, flow
rate 1.0 mL/min. The column was interfaced to the spray
chamber by a short piece of PEEK tubing (0.25 mm ID). An
injection volume of 50 μL was used by the autosampler, and
the results were processed off-line using MS Excel and the
chromatograms plotted in OriginPro 8.1 (OriginLab
Corporation, MA, USA).
Particle separation by AF
4
The applied AF
4
programme for the separation of particles,
including the composition of the carrier liquid, is presented in
Table 3which includes SDS, a common surfactant for the
separation of particles in AF
4
[10]. Before injection into the
AF
4
channel, enzymatic digestion of the hip aspirate samples
to release any particles associated with this heterogeneous
matrix was performed as follows: a volume of 4.5 mL enzyme
solution was added to 500 μL hip aspirate sample, and the
mixture was incubated at 50 °C in a water bath using contin-
uous stirring until no solids were visible in the sample, which
took approximately 30 min. Avolume of 5 μLofthisdigestate
was injected into the AF
4
channel. Results were processed off-
line and plotted in OriginPro 8.1 (OriginLab Corporation,
MA, USA).
Tabl e 2 AF
4
separation programme for proteins
Carrier liquid 50 mM NH
4
NO
3
,pH7.2–7.4
Membrane material Regenerated cellulose
Membrane cut-off 10 kDa
Spacer height 350 μm
Detector flow rate 1.0 mL/min
Injection flow rate 0.2 mL/min
Focus flow rate 1.5 mL/min
Step Duration
(min)
Mode Cross-flow rate
(mL/min)
12 Elution 4.0
21 Focus –
31 Focus+injection–
42 Focus –
5 20 Elution 4.0
6 4 Elution –
Tabl e 1 Typical ICP-MS settings used for total elemental analysis, spICP-MS and coupling to AF
4
or AEC
Total meta l AF
4
AEC spICP-MS
RF power (W) 1350 1550 1550 1550
Plasma gas flow
rate (L/min)
13 15 15 14
Carrier gas flow
rate (mL/min)
0.74 0.90 0.99 0.96
Makeup gas flow
rate (mL/min)
n/a 0.45 0.20 0.80
Cell gas and flow
rate (mL/min)
7%H
2
in He n/a n/a n/a
Nebulizer Glass concentric (ESI) Micro Flow (Agilent, G3139A-100) MicroMist (Glass Expansion) Low-flow concentric
Spray chamber Cyclonic Scott double-pass Scott double-pass Cyclonic
Isotopes monitored
53
Cr,
59
Co,
98
Mo,
74
Ge (IS)
48
Ti (for
32
S
16
O),
53
Cr,
57
Fe,
59
Co,
60
Ni,
65
Cu,
66
Zn,
98
Mo
48
Ti (for
32
S
16
O),
50
Cr,
54
Fe,
57
Fe,
58
Ni,
59
Co,
60
Ni,
63
Cu,
64
Zn
65
Cu,
66
Zn
53
Cr,
59
Co,
98
Mo
Integration time
per isotope
a
(ms)
1000 Co, Cr, Mo (1000);
Ge (250)
100 100 3
n/a not applicable
a
In some cases, a shorter dwell time was used as indicated in the text
Characterization of MoM hip wear metal particles by AF
4
Author's personal copy
Particle detection by spICP-MS
A Thermo Scientific iCAP Q ICP-MS instrument (Thermo
Fisher Scientific GmbH, Bremen, Germany) was used for all
spICP-MS experiments. Instrument tuning was performed pri-
or to analysis by using a tuning solution according to the
manufacturer’s recommendation. Instrument settings are giv-
en in Table 1. Before spICP-MS analysis of nanoparticles in
the hip aspirate, the samples were enzymatically digested to
liberate the particles from the heterogeneous matrix: a volume
of 4.9 mL enzyme solution was added to 100 μL hip aspirate
sample, and the mixture was incubated at 37 °C in a water bath
using continuous stirring until no solids were visible in the
sample, which took approximately 40 min. After enzymatic
digestion, the samples were diluted 1000 times with ultrapure
water.
The theory of ICP-MS used in single particle mode for
characterization of colloids in water has been introduced by
Degueldre et al. [14]. Data processing was performed accord-
ing to the methodology described by Pace [15]usingaspread-
sheet routine. The sample flow rate was 0.34 mL/min, which
was accurately determined by weighing. Quantification of
particle size was based on a calibration curve constructed from
an acid blank and five concentration levels of certified stan-
dard solutions of Cr, Co and Mo ranging from 0.2 to 5.0 ng/L,
in 2 % nitric acid. The isotopes
53
Cr,
59
Co and
98
Mo were
detected in separate runs using 3 ms dwell time for a total
measurement time of 120 s (equivalent to 40,000 data points
per isotope). The ICP-MS signal intensity for each standard
solution was acquired by averaging the signal intensity record-
ed during the 120-s measurement time. The transport efficien-
cy of the liquid samples through the sample introduction sys-
tem was determined according to the Bparticle frequency^
method [14] by measuring a NIST AuNP suspension with a
known average particle diameter (56.0±0.5 nm as determined
by TEM) and gold mass concentration (51.86±0.64 μg/g)
diluted 10
6
times with ultrapure water. The transport efficien-
cy was calculated as the percentage of all Au nanoparticles
detected by spICP-MS versus the theoretical (calculated) par-
ticle number in the introduced sample volume, derived from
information on average size (by TEM) published in the certif-
icate, the measured uptake rate and time of introduction of the
sample suspension into the ICP-MS instrument.
The recorded signal intensity data were plotted (in cps)
versus number of Bevents^, to create a signal distribution his-
togram using OriginPro 8.1 (OriginLab Corporation, MA,
USA). Low, stable and relatively noise-free signal intensities
were considered to be caused by instrument background and
polyatomic interferences or, for slightly higher intensity
values, also caused by dissolved metal-containing molecules.
An iterative algorithm was applied where particle events were
distinguished as outliers from the polyatomic background and
dissolved metal signal if the measured intensity was more than
five times the standard deviation of the whole data set [16].
This criterion ensured that only particle events and no back-
ground or signal from metal ions were included in the data set,
but also excluded low-intensity events corresponding to the
smallest nanoparticles.
Results and discussion
Total wear metal concentrations by ICP-MS
The total concentrations of Co and Cr in the serum and hip
aspirate samples were measured as part of the routine moni-
toring of UK National Health Service patients with MoM hip
replacements as recommended by the MHRA in 2010 [5]. In
some cases, further measurements, particularly of Mo, were
made depending on the results from the speciation analysis
described below; however, in some cases, an insufficient sam-
ple volume was available so further testing was not possible.
The results are shown in Table 4.
Protein separation and detection in serum samples
by AF
4
–ICP-MS
Separation of a mixture of Alb (66.5 kDa) and Tf (79.5 kDa)
both at a concentration of 2 mg/mL was achieved using AF
4
coupled to ICP-MS, as demonstrated by the signals at m/z
65
Cu,
66
Zn and
57
Fe (Fig. 1b–d), and a suitable cross-flow rate
was selected based on the most selective separation. Cross-
flow rates of 3, 4 and 5 mL/min were tested, but the channel
back-pressure at the 5-mL/min cross-flow rate was too high
and no further improvement of the separation could be
achieved in comparison to 4 mL/min. The separation at
4 mL/min was superior to that corresponding to 3 mL/min
Tabl e 3 AF
4
separation programme for particles
Carrier liquid 0.05 % SDS, pH 6
Membrane material Regenerated cellulose
Membrane cut-off 10 kDa
Spacer height 350 μm
Detector flow rate 1.0 mL/min
Injection flow rate 0.2 mL/min
Focus flow rate 1.0 mL/min
Step Duration
(min)
Mode Cross-flow rate
(mL/min)
12 Elution 1.0
21 Focus –
31 Focus+injection–
42 Focus –
5 20 Elution 0.3→0.1
6 4 Elution –
K. Loeschner et al.
Author's personal copy
and was therefore chosen for the final separation method. The
achieved selectivity of the separation of the proteins was iden-
tical to that predicted by modelling based on AF
4
theory. This
underscored that, because using a higher cross-flow rate was
not possible, no further improvement of the separation was
practically achievable. Because the best possible selectivity
of the separation of the proteins was prioritized, their recovery
was not included as an optimization parameter of the AF4
separation method.
Two peaks were visible in the UVabsorbance signal for the
Alb–Tf standard mixture (Fig. 1a). The absence of any peak in
the void volume of this fractogram suggested that degradation
or non-ideal elution behaviour of the analyte proteins did not
happen. Importantly, the specificity of ICP-MS for character-
ization of different elements and the selective binding of cop-
per and zinc to Alb allowed for the detection of Alb based on
the
65
Cu and
66
Zn ICP-MS signals (Fig. 1b, c; black curves).
In the commercial Alb standard, monomers, dimers and tri-
mers of this protein were identified, as reported previously
[17,18]. The specificity of ICP-MS and the selective binding
of iron to Tf facilitated the detection of Tf based on the
57
Fe
ICP-MS signal (Fig. 1d). Whereas the absorbance signal could
not resolve the Alb dimer and Tf, the detection of the selec-
tively bound elements by ICP-MS was able to distinguish the
two proteins. The separation achieved using this method was
in agreement with previous work using AF
4
–ICP-MS for the
separation of standards of Alb and Tf [12].
Patient serum samples were separated with the same AF
4
method, and comparison of a representative sample (S4) and a
mixture of Alb and Tf (2 mg/mL) standard substances is
shown in Fig. 1(red curve). The retention time (Tr) of the
Alb monomer peak was confirmed by the
65
Cu and
66
Zn sig-
nals for the Alb standard (Fig. 1b, c).TheTfpeakinserum
was confirmed by the
57
Fe signal for the Tf standard (Fig. 1d).
The second peak recorded by the
65
Cu trace for the blood
serum sample (Fig. 1c) could be ceruloplasmin, another Cu-
binding protein. Therefore, the
66
Zn ICP-MS signal was
chosen as a selective indicator for the elution of Alb in further
analyses.
The fractograms for a variety of analysed hip patient serum
samplesaswellasacontrol patient sample (non-hip
replacement) spiked with Co and Cr are shown in Fig. 2.
The
66
Zn signal represents the elution of Alb (left vertical line
at Tr 6.6 min) and the
57
Fe signal the elution of Tf (middle
vertical line at Tr 7.7 min). In this way, the specificity of the
ICP-MS signals for Zn and Fe ensured the selective determi-
nation of the two proteins and allowed for assignment of their
identities. Because these two proteins in human serum are
highly abundant, they were potential binding sites also for
trace elements originating from the patients’hip replacements.
Elution of a Cr-containing peak at the same retention time as
that for Tf was observed for sample S3 (Fig. 2c), which was
the sample with the highest total Cr content (2050 nmol/L
corresponding to 121 μg/L) in Table 4. Furthermore, spiking
of an ionic Cr standard solution to serum from the control
patient also led to a Cr-containing peak co-eluting with Tf.
This suggested the association of Cr with Tf and not with
Alb. For the other serum samples, the intensity of the
53
Cr
signal was below the limit of detection. For similar reasons,
the results suggested that Co was associated with Alb (Tr
6.6 min) and with an additional unidentified protein eluting
at Tr 9.3 min (Fig. 2d). No peaks were detected when moni-
toring
98
Mo, probably because the concentrations measured in
the patient serum samples (Table 4) were too low, about ten
times lower than the concentrations for Co and Cr.
Protein separation and detection in hip aspirate samples
by AF
4
–ICP-MS
The AF
4
–ICP-MS system was also used to study the associa-
tion of metals with proteins in a hip aspirate sample (Fig. 3),
which is considerably more complex and heterogeneous in
composition than human serum and provides an important
insight into the interface between the hip replacement and
human tissue. It could be shown (Fig. 3a) that Cr was bound
to Alb (Tr ~6.8 min) and Tf (Tr ~7.7 min), Co (Fig. 3b)was
bound to Alb (Tr ~6.8 min) and to two unknown fractions
eluting at Tr 9.7 and Tr 14.7 min and Mo (Fig. 3c) was asso-
ciated with Alb (Tr ~6.8 min) and an unidentified species
eluting at Tr 9.7. The results suggest that investigation of the
association of Cr and Co with Alb and Tf in serum and also in
the more complex hip aspirate sample is feasible using AF
4
–
ICP-MS. These results were confirmed by the well-
established AEC–ICP-MS method.
Confirmation of metal–protein binding in serum samples
by AEC–ICP-MS
A previously published separation method [7] used for the
characterization of the binding of wear metal ions to proteins
Tabl e 4 Total concentration of wear metals in the samples studied
Sample name Cr nmol/L
(μg/L)
Co nmol/L
(μg/L)
Mo nmol/L
(μg/L)
Serum samples
S1 153 (9.0) 69.6 (3.6) 7.3 (0.7)
S2 253 (14.9) 183 (9.5) 22.4 (2.2)
S3 2050 (121.0) 1860 (96.7) insufficient
S4 603 (35.6) 486 (25.3) 10.8 (1.1)
S5 264 (15.6) 177 (9.2) 19.5 (1.9)
S6 49.5 (2.9) 351 (18.3) 10.9 (1.1)
S7 259 (15.3) 246 (12.8) 9.6 (0.9)
S8 (control) Not detected 3.4 (0.2) Insufficient
Hip aspirate sample 292,300 (15,200) 134,000 (7900) 18,200 (1800)
Characterization of MoM hip wear metal particles by AF
4
Author's personal copy
in hip replacement patients was used to confirm the AF
4
–ICP-
MS results. The AEC–ICP-MS results for the Tf and Alb
standards (shown in the Electronic Supplementary Material
(ESM), Fig. S1) show the same elution characteristics as pre-
viously reported [7]. The retention times for Tf and Alb vary
slightly depending on the isotope chosen, such that for Tf the
peak maximum is at Tr 2.9 min using detection of
54
Fe and Tr
3.1 min using
48
SO. This slight difference in retention time is
because of the wide peak profiles for these proteins. This
effect also occurs for Alb with a slight difference in peak
retention times depending on the isotope monitored: Tr
(min) of 4.3 (
66
Zn), 4.5 (
48
SO) and 4.8 (
65
Cu). Our studies
have shown this effect to be consistent for a large number of
human patient samples, and a possible explanation is that it is
due to the wide peak profiles and different metal loadings in
the binding sites for Zn and Cu, which change the charge on
the molecule sufficiently for the Alb isoforms contain-
ing these elements to elute at slightly different times.
From a practical respect, this slight difference has no
bearing because when identifying the proteins in the
samples by comparing the retention times for the Alb
and Tf standards to the patient samples (ESM Fig. S2
(a) and (b)), they are consistent for each element. This
makes the identification more robust because the sample
and standard peaks have to agree for all of the isotopes
monitored, rather than relying on a single isotope. This
is further shown in ESM Fig. S3 which shows the re-
sults for the same four isotopes in serum samples S1,
S2 and S4 to 7. The second peak present in the
65
Cu
chromatogram in these figures, and in ESM Fig. S1 (b),
is due to ceruloplasmin, which is an important Cu-
containing protein found in human serum.
0 5 10 15 20
0
5
10
15
20
25
Alb + Tf
Serum
Retention time (min)
a
0 5 10 15 20
0
20,000
40,000
60,000
8.3 min
Alb + Tf
Serum
Retention time (min)
b
65Cu (cps)
Alb monomer
6.3 min
Alb trimer
10.3 min
Alb dimer
8.8 min
Alb
6.6 min
Cu
0 5 10 15 20
0
5,000
10,000
15,000
20,000 Zn
Alb + Tf
Serum
Retention time
(
min
)
c
66Zn (cps)
Alb monomer
6.3 min
Alb trimer
10.3 min
Alb dimer
8.8 min
Alb
6.6 min
0 5 10 15 20
15,000
20,000
25,000
30,000
35,000 Fe
Tf
7.7 min
Retention time (min)
Alb + Tf
Serum
d
57Fe (cps)
Tf
7.7 min
Absorbance @280 nm (mAU)
Fig. 1 Fractograms of a mixture of standards of Alb and Tf (black curves) and patient serum sample S4 (red curves) based on detection by aUV
absorbance at 280 nm and b–dICP-MS of
65
Cu,
66
Zn or
57
Fe, respectively
K. Loeschner et al.
Author's personal copy
The AEC–ICP-MS results for the association of Co and Cr
with Tf and Alb in serum are shown in ESM Fig. S2 (a) and
(b). The gradient elution programme used led to a rising base-
line for m/z53, caused by the interference from
40
Ar
13
C
+
due
to the increase in ammonium carbonate concentration over the
course of the gradient (buffer B). To overcome this effect and
because there was no detectable
50
Ti or
50
V in the samples, the
signal for
50
Cr was used to show the elution of Cr, and this
effectively overcame the problem with the elevated baseline
signal. The results for
50
Cr in ESM Fig. S2 (a) show that Cr
was associated with Tf and Alb, whereas the results using
AF
4
–ICP-MS only showed Cr associated with Tf. This differ-
ence is probably due to the greater capacity of AEC which
allows for more sample to be injected and thus the association
betweenCrandAlbtobediscerniblefromthebackground.
When a higher concentration of Cr is present, as seen in the hip
aspirate sample (see below and Fig. 3a), AF
4
–ICP-MS indi-
cates Cr binding to both Tf and Alb. For
59
Co (ESM Fig S2
(b)), AEC shows this element to be associated with Alb except
in one sample (S4), where an additional second peak occurs
before Alb. This confirms the results obtained by the AF
4
–
ICP-MS method of analysis, where a second compound is also
present in the fractogram for sample S4. Two samples, S3 and
S4, contain a second peak using AF
4
–ICP-MS; unfortunately,
there was insufficient sample to analyse S3 by AEC.
Confirmation of metal–protein binding in hip aspirate
samples by AEC–ICP-MS
As shown in ESM Fig. S4 (g), problems were encountered
with the AEC-ICP-MS method when running the hip aspirate
sample from the same patient as that shown in Fig. 3;dueto
0 5 10 15 20
0
5,000
10,000
15,000
20,000
25,000 Zn Alb
Retention time (min)
S1
S2
S3
S4
S5
S6
S7
S8 (control)
a
66Zn (cps)
0 5 10 15 20
20,000
30,000
40,000
50,000 Fe Tf S1
S2
S3
S4
S5
S6
S7
S8 (control)
Retention time (min)
b
57Fe (cps)
offset 4000 cps
0 5 10 15 20
0
200
400
600
800
1,000
1,200 Cr
offset 500 cps
Retention time (min)
S3
S8 spiked
(control subtracted)
c
53Cr (cps)
Alb Tf
0 5 10 15 20
0
500
1,000
1,500
2,000
2,500
3,000
3,500
4,000 Co
unknown
Alb Tf
Retention time (min)
S1
S2
S3
S4
S5
S6
S7
S8 spiked
(control subtracted)
d
59Co (cps)
Fig. 2 Fractograms based on the a
66
Zn, b
57
Fe, c
53
Cr and d
59
Co ICP-
MS signals for serum samples S1–S7from hip replacement patients and a
spiked control sample (S8). For the
53
Cr and
59
Co ICP-MS response, the
signal intensities recorded for the control patient sample were subtracted.
Vertical lines represent the elution of Alb (Tr 6.6 min), Tf (Tr 7.7 min) and
an unidentified species (Tr 9.3 min)
Characterization of MoM hip wear metal particles by AF
4
Author's personal copy
the high concentration of Co, which affected the electron mul-
tiplier detector in the ICP-MS, it was not possible to discern all
of the binding proteins in the sample. Binding of the other
elements, Cu, Zn and Fe, to the serum proteins is also shown
in ESM Fig. S4 (b–e), and the binding of Cr is shown in ESM
Fig. S4 (f). A similar hip aspirate sample from another patient
was analysed to provide information on Co and Cr binding,
and the results are shown in ESM Fig. S2 (c) and (d). AEC–
ICP-MS analysis confirmed the findings by AF
4
–ICP-MS that
Co was bound to Alb (ESM Fig. S2 (d)) and at least one other
compound which elutes after Alb, whereas the results for Cr
(ESM Fig. S2 (c)) showed an association with Tf and Alb.
The results show that the selectivity of the separation of
proteins by AEC was superior to that of AF
4
and that AEC had
a greater sample capacity for these serum proteins. Because
the mechanism of separation by AF
4
relies on the diffusion
coefficient of the analytes, it was not possible to separate
proteins with similar molecular weights. The AEC separation
of proteins is related to their different isoelectric points and the
buffer pH used and is influenced to a lesser extent by size.
Because a stationary phase is not required for separations
using AF
4
–ICP-MS, it has the advantage that non-covalently
bound metals do not interact with the stationary phase, which
can occur with HPLC methods.
Wear metal particle separation and detection in hip
aspirate samples by AF
4
–ICP-MS
For detection of the possible presence of particles in the hip
aspirate, enzymatic digestion with the broad-spectrum prote-
ase proteinase K was chosen based on a protocol which had
been previously developed for chicken meat containing silver
nanoparticles [13]. Enzymatic digestion of this complex and
heterogeneous sample was used because other separation
methods for the removal of coarse organic material, such as
filtration or extensive high-speed centrifugation, would in-
crease the risk of losing particles due to their possible associ-
ation with organic material. The fractograms based on the
65
Cu,
53
Cr,
59
Co and
98
Mo ICP-MS signals for a patient hip
aspirate sample are shown in Fig. 4.Thisparticularsample
was chosen for AF
4
analysis because of its pale grey appear-
ance suggesting the presence of particulate matter, and be-
cause of the highly elevated levels of Cr at 292,300 nmol/L
(15,200 μg/L), Co at 134,000 nmol/L (7900 μg/L) and Mo at
18,200 nmol/L (1800 μg/L) in the sample.
The peak at Tr 1 min in Fig. 4(Bmolecules peak^)couldbe
attributed to dissolved metal ions or metals bound to organic
molecules, which remained after enzymatic digestion; Cr, Co
and Mo were detected in this peak. For Co, the peak was
0 5 10 15 20
0
2,000
4,000
6,000
8,000
10,000 Cr
unknown 2
unknown 1
Alb Tf
Retention time (min)
a
53Cr (cps)
0 5 10 15 20
0
50,000
100,000
150,000
200,000
250,000 Co
unknown 2
unknown 1
Alb Tf
Retention time (min)
b
59Co (cps)
0 5 10 15 20
0
1,000
2,000
3,000
Mo
unknown 2
unknown 1
Alb Tf
Retention time
(
min
)
c
98Mo (cps)
Fig. 3 Fractograms of the hip aspirate sample based on the a
53
Cr, b
59
Co
or c
98
Mo ICP-MS signals. The dashed vertical lines represent the elution
of Alb, Tf and two additional fractions at Tr 9.7 and Tr 14.7 min
K. Loeschner et al.
Author's personal copy
significantly higher in relation to the nanoparticle peak (Tr
~5 min) compared to Cr or Mo. This indicates an enrichment
of Co ions in the hip aspirate compared to the composition of
the alloy, which could be due to corrosion of particles in the
physiological environment. The broad peak in the retention
time range of approximately 4 to 12 min was caused by elut-
ing nanoparticles (Bparticle peak^).Asthepeakappearedin
the Cr, Co and Mo ICP-MS signals, it can be concluded that
nanoparticles consisting of all three elements were released
from the alloy. The
65
Cu ICP-MS signal represents organic
molecules, more specifically Cu-containing peptides released
by the enzymatic digestion step. The Brelease peak^appeared
because the cross-flow rate (=separation force) was set to zero
after 20 min elution at which point particles that had not eluted
during the 20 min elution time eluted from the channel. This
indicated that this fraction of larger nanoparticles was not
included in the area under the curve of the fractogram.
Future work will focus on the optimization of the separation
parameters, so that large particles are separated by the elution
programme. Future quantitative work will allow the determi-
nation of the elemental ratio between Cr, Co and Mo as a
function of particle size.
The intense signal spikes for
53
Cr,
59
Co and
98
Mo recorded
in the retention time range >4 min in Fig. 4were caused by the
elution of particles from the AF
4
channel. In contrast, the
65
Cu
signal did not show such spikes and the recorded smoother
signal corresponds to non-particulate Cu, possibly Cu bound
to small molecular fragments left after the enzymatic diges-
tion. The spikes in the ICP-MS signals for
53
Cr,
59
Co and
98
Mo increased in intensity with increasing retention time,
which is in accordance with the AF
4
theory predicting that
larger particles elute at longer retention times.
To obtain further information about the hip aspirate sample,
the
53
Cr ICP-MS, light scattering and 280 nm absorbance
signals were studied (Fig. 5). The intensity of the light scat-
tering signal in the retention time range <4 min was low,
which confirms the absence of scattering entities and conse-
quently that the enzymatic digestion of proteins and other
biomolecules present was efficient. The scattered light inten-
sity is proportional to the diameter of scattering molecules or
particles to the power of six, so the intensity of the scattered
light of fragments of proteins was therefore much less than
that of the intact proteins. This was demonstrated for the elu-
tion of the intact enzyme (proteinase K), which was detected
by light scattering and occurred before the retention time
range of the Bparticle peak^. Injection of the pure enzyme
alone confirmed the observed retention time of 3 min
(fractogram not presented). The pronounced light scattering
signal at retention times >4 min confirmed the presence of
large light scattering entities in the injected sample, and the
53
Cr peak supported the proposition that particles eluted in the
5–12-min retention time range. The interpretation of the 280-
nm absorbance signal was more difficult as organic molecules
0
50,000
0
50,000
0
50,000
0 5 10 15 20
0
5,000
53Cr
59Co
Cr
65Cu
release peak
Mo
Co
Cu
98Mo
molecules particles
Retention time (min)
Fig. 4 Fractograms of the enzymatically digested hip aspirate sample
based on the
65
Cu,
53
Cr,
59
Co and
98
Mo ICP-MS signal using the AF
4
“particle method”(V
inj
=5 μL). A relatively short dwell time of 10 ms was
used for detection
0 5 10 15 20
0
2E4
4E4
6E4
8E4
53
Retention time (min)
0
2
4
6
ICP-MS
Absorbance
Light scattering
t0
0
10
20
30
90°
enzyme
light scattering signal (mV)
Absorbance @280 nm (mAU)
Cr (cps)
Fig. 5 Fractograms of the enzymatically digested hip aspirate sample
presented in Fig. 4directly comparing the
53
Cr ICP-MS, the 280-nm
absorbance and the 90° light scattering signals
Characterization of MoM hip wear metal particles by AF
4
Author's personal copy
(amino acids and peptides remaining from the proteolysis) as
well as the studied metals Co and Cr absorb light at the 280-
nm wavelength.
The approximate size range of the nanoparticles in Fig. 5
was estimated by spiking the sample with 80-nm AuNPs as
calibrant for size, this being the best available metallic NP
with known size in the absence of any Co–Cr-containing
NPs. The semi-quantitative size range (hydrodynamic diame-
ter) estimated this way was 50–140 nm. Obtaining quantita-
tive size information by AF
4
is challenging and will re-
quire an intensive optimization of the separation condi-
tions especially with respect to carrier liquid composition
and membrane material [19].
Confirmation of the presence of wear metal particles
in the nanometre size range by spICP-MS
The presence of particles in the enzymatically digested hip
aspirate samples was confirmed by spICP-MS analysis.
Figure 6shows the time-resolved
53
Cr and
59
Co ICP-MS sig-
nals measured in single particle mode. The dashed vertical
lines in Fig. 6b represent the limit of detection (LOD) for
particle size determination, which were based on filtration of
particle events using the 5 standard deviations criterion as
explained in the BInstrumental analysis^section. The size de-
tection LOD for Co was lower (approximately 30 nm) than
that for Cr (approximately 100 nm), which was caused by a
poorer signal response for
53
Cr (9.5 % abundant) than that for
59
Co (100 % abundant).
According to the theory of spICP-MS [15], the frequency
of the time-resolved intense signal Bspikes^is directly related
to the particle number concentration in the sample and the
height of these Bspikes^is directly related to the particle di-
ameter assuming a spherical shape. In the same sample, more
Bspikes^were recorded for the
53
Cr signal (1069) and conse-
quently more Cr-containing particles were detected than those
corresponding to the
59
Co signal (455) in the same volume of
aspirated enzymatic digest. This cannot be explained by the
different LODs for size detection for Cr and Co. The results
indicated that there were more particles present in the hip
aspirate that contained only Cr and no Co despite the fact that
the hip replacement alloy originally contained both Co and Cr.
This could be explained by findings described in the literature
that Cr as Cr(III) was the most abundant metal in the wear
debris found in tissues surrounding failed metal-on-metal hip
replacements [8,9]. This was confirmed by in vitro hip sim-
ulator studies of these replacements [20]. Within the synovial
fluid, Cr
2
O
3
and Cr(OH)
3
were the predominant species,
whereas CrPO
4
occurred in non-synovial environments [4].
The observed (near) absence of Co in wear particles could
be explained by rapid corrosion of the particles in the physi-
ological environment; whilst protected in the bulk alloy by the
passivating effect of Cr, Co
2+
is soluble under physiological
conditions [8]. Assuming that such a dissolved fraction occurs
as low-molecular compounds containing Co (atomic clusters,
ions or Co-containing proteins (Fig. 3)) in the hip fluid, this
might explain the recorded elevated baseline level of the time-
resolved Co signal from the spICP-MS analysis (red line) in
comparison to that in ultrapure water (black line in Fig. 6a).
The finding is in agreement with the observed high Co peak
0
200
400
600
800
1000
Co ultrapure water
hip aspirate
53
Cr
012345678910
0
200
59
Time (s)
elevated baseline
0 100 200 300 400 500
0
100
200
300
b
Based on
53Cr signal;
assumed particle compositions
ASTM F75
Cr2O3
Based on
59Co signal;
assumed particle compositions
ASTM F75
CoO
LOD
Particle diameter
(
nm
)
Number of particles Cr (counts/event)
Co (counts/event)
Fig. 6 Analysis of the enzymatically digested hip aspirate sample by
spICP-MS. aTime-resolved ICP-MS signal for
53
Cr and
59
Co in single
particle mode, bparticle number size distributions based on spICP-MS
analysis of
53
Cr, calculated for two possible compositions of the Cr-
containing particles: ASTM F75 alloy and Cr
2
O
3
and
59
Co signal,
calculated for two possible compositions of the Co-containing particles:
ASTM F75 alloy and CoO
K. Loeschner et al.
Author's personal copy
(Tr 1 min) in AF
4
–ICP-MS (Fig. 4)whichwasattributedto
dissolved metal ions or metals bound to organic molecules.
The transition from newly formed wear particles to Bdegraded^
particles and the changes this brings to their composition lead-
ing to effective enrichment of some elements over the others is
clearly important from a toxicological perspective.
To determine the approximate particle diameter and
the number-based particle size distribution (Fig. 6b),
particles were assumed to be spherical in shape with
one of two potential particle compositions: (1) as that
in the frequently used implant alloy ASTM F75 (~66 %
Co, ~28 % Cr, ~6 % Mo) assuming no corrosion of the
particles [4]or(2)asseparateCr
2
O
3
and CoO particles
[8]. An assumption of the particle composition was nec-
essary, as the applied spICP-MS method did not provide
this information. The calculated diameters of the Cr-
containing particles were in the size range from the size
LODofapproximately100to550nmbasedonthe
53
Cr signal. The differences between the sizes obtained
when assuming the ASTM F75 alloy composition or
Cr
2
O
3
were small (median diameters 141 or 119 nm,
respectively). The relatively high LOD for particle size
for
53
Cr restricted the minimum detectable particle size
to about 100 nm. Smaller particles are, however, most
likely present in the hip aspirate, as in vitro hip simu-
lation experiments demonstrated that wear particles con-
taining Cr were in the size range of 5 to 70 nm and
consisted mainly of Cr
2
O
3
[20]. The mean diameter of
particles, which were detected in tissue surrounding
failed MoM hip implants, was 30 nm [8].
Compared to chromium, the
59
Co ICP-MS signal allowed
detection of significantly smaller Co-containing particles,
which were in the size range of 32 to 136 nm and with a
median diameter of 38.5 nm, when assuming a composition
of the particles as in the ASTM F75 alloy. However, as
discussed before, Co-containing particles could also be pres-
ent in other compositions such as CoO. In either case, the
calculated sizes were similar with median diameters of
39 nm. The presented spICP-MS results proved the existence
of wear particles in the nanometre size range in the hip aspirate
sample and provided an indication of their approximate size
range. Obtaining quantitative size information for Cr-
containing particles was limited by the relatively high LOD
for this element and by the unknown composition of the
particles.
Conclusions
AF
4
–ICP-MS was shown to be a promising instrumental tech-
nique for the characterization of hip patient samples as it gave
information on the binding of released metals to important
serum transport proteins, such as Alb and Tf, and also
provided information on particle size and elemental composi-
tion of wear particles present in the hip aspirate. The separa-
tion power and sample capacity of AF
4
for the metal-binding
proteins were inferior to those of AEC, but still enough to
provide a sufficient separation and detection when relying
on the selective binding of metals to proteins and the selectiv-
ity of the on-line metal detection by ICP-MS.
The potential advantages of AF
4
–ICP-MS for characteriza-
tion of the wear debris in comparison to other atomic spectro-
scopic or microscopy techniques are the following: (1) it could
allow quantitative determination of wear metal particles in
terms of mass concentration; (2) from the same fractogram,
molecule-bound metal could be quantified in terms of mass
concentration (first peak of the fractogram); (3) information
on the particle size distribution could be obtained, which can
be used to trigger more elaborate analysis of nanoparticle size;
and (4) metal binding to proteins in hip aspirate samples could
be studied and quantified with the same equipment and re-
quires only adjustment of the separation conditions. To obtain
an optimum AF
4
separation method with respect to recovery,
separation power, analysis time and accurate alignment of
sample separation and (size) standard separation, a detailed
study of critical parameters such as the carrier liquid compo-
sition, membrane material, cross-flow rate, spacer height, fo-
cus flow and focus time will be required [19].
Analysis using spICP-MS was shown to be a suitable
method for detection of nanoparticles in hip aspirate samples
from patients with MoM hip replacements. The relatively high
LOD for size of Cr however limited the accessible size range
of nanoparticles. Further information about the particle com-
position could be obtained with modern instrumentation like,
for example, multi-collector instruments, by which detection
of several isotopes during a single particle event is possible.
A recent 2013 European multidisciplinary consensus
statement [21] on the use and monitoring of MoM
joints includes in its list of unmet medical needs for
future research the necessity to determine the mecha-
nisms creating particles/ions/metallo-organic compounds
or aggregates in MoM bearings and their distribution
and to investigate the local and systemic distribution
and pathological effects of particles/ions/metallo-organic
compounds produced in MoM bearings. The use of
spectromicroscopy of tissues surrounding MoM replace-
ments has been shown [8] to provide some of these
answers, but the full picture of the processes and spe-
cies involved will require the development of analytical
tools such as those presented here. Whilst a preliminary
study such as this does not provide a fully validated
and quantitative method for the measurement of the size
and composition of wear particles, the presented AF
4
–
ICP-MS methodology is clearly a significant new tool
for future characterization of wear metal particles in
human samples.
Characterization of MoM hip wear metal particles by AF
4
Author's personal copy
Acknowledgments We thank Marianne Hansen for her support during
the experimental work, Thermo Fisher Scientific for providing the iCAP
Q instrument, the Biotechnology and Biological Sciences Research
Council (UK) for funding to CH for the Agilent 7500ce ICP-MS used
in the HPLC experiments and AstraZeneca in Macclesfield, UK, for
donation to CH of the Waters HPLC system.
Conflict of interest All the authors express no conflict of interest
relating to this work.
References
1. Sampson B, Hart A (2012) Ann Clin Biochem 49:118–131
2. Sedrakyan A, Normand SL, Dabic S, Jacobs S, Graves S, and
Marinac-Dabic D (2011) BMJ 343:d7434
3. Campbell JR, Estey MP (2013) Clin Chem Lab Med 51:213–220
4. Keegan GM, LearmonthID, CaseCP (2007) J Bone Joint Surg (Br)
89-B:567–573
5. Medicines and Healthcare products Regulatory Agency (MHRA).
All metal-on-metal (MoM) hip replacements, MDA/2010/033, is-
sued on 22nd April 2010
6. Fary C, Thomas GER, Taylor A, Beard D, Carr A, and Glyn-Jones
S (2011) BMJ 343:d7441
7. Nuevo Ordonez Y, Montes-Bayon M, Blanco-Gonzalez E, Paz-
Jimenez J, Maria Tejerina-Lobo J, Miguel Pena-Lopez J, Sanz-
Medel A (2009) J Anal At Spectrom 24:1037–1043
8. Goode AE, Perkins JM, Sandison A, Karunakaran C, Cheng H,
Wall D, Skinner JA, Hart AJ, Porter AE, McComb DW, Ryan
MP (2012) Chem Commun 48:8335–8337
9. Hart AJ, Quinn PD, Sampson B, Sandison A, Atkinson KD,
Skinner JA, Powell JJ, Mosselmans JF (2010) Acta Biomater 6:
4439–4446
10. Schimpf ME, Caldwell KD, Giddings JC (2000) Field-flow frac-
tionation handbook. Wiley-Interscience, New York
11. Von der Kammer F, Legros S, Larsen EH, Loeschner K, Hofmann T
(2011) Trends Anal Chem 30:425–436
12. Heroult J, Havard T, Goenaga Infante H (2012) LC:GC The
Column 8:11–15
13. Loeschner K, Navratilova J, Købler C, Mølhave K, Wagner
S, Kammer F, Larsen E (2013) Anal Bioanal Chem 405:
8185–8195
14. Degueldre C, Favarger PY (2003) Colloids Surf A Physicochem
Eng Asp 217:137–142
15. Pace HE, Rogers NJ, Jarolimek C, Coleman VA, Higgins CP,
Ranville JF (2012) Anal Chem 84:4633
16. Tuoriniemi J, Cornelis G, Hassellöv M (2012) Anal Chem 84:
3965–3972
17. Wen J, Arakawa T, Philo JS (1996) Anal Biochem 240:155–166
18. Fraunhofer W, Winter G (2004) Eur J Pharm Biopharm 58:
369–383
19. Loeschner K, Navratilova J, Legros S, Wagner S, Grombe R,
Snell J, Kammer F, Larsen E (2013) J Chromatogr A 1272:
116–125
20. Pourzal R, Catelas I, Theissmann R, Kaddick C, Fischer A (2011)
Wear 271:1658–1666
21. Hannemann F, Hartmann A, Schmitt J, Lutzner J, Seidler A,
Campbell P, Delaunay CP, Drexler H, Ettema HB, Garcia-
Cimbrelo E, Huberti H, Knahr K, Kunze J, Langton DJ, Lauer W,
Learmouth I, Lohmann CH, Morlock M, Wimmer MA, Zagra L,
Gunther KP (2013) Orthop Traumatol Surg Res 99:263–271
K. Loeschner et al.
Author's personal copy