Content uploaded by Mathie P G Leers
Author content
All content in this area was uploaded by Mathie P G Leers on Mar 19, 2014
Content may be subject to copyright.
DOI 10.1515/cclm-2012-0549Clin Chem Lab Med 2013; 51(6): 1285–1290
Jeffrey F.W. Keuren , Johannes J.M.L. Hoffmann and Mathie P.G. Leers*
Analysis of serous body fluids using the CELL-DYN
Sapphire hematology analyzer
Abstract
Background: Correct cell enumeration and differential
analysis of body fluids are important in the diagnosis and
management of several diseases. Currently, microscopic
analysis is still considered the “ gold standard ” . The aim
of the present study was to evaluate the analytical perfor-
mance of the CELL-DYN Sapphire hematology analyzer
for automated differentiation of cells in serous fluids and
to explore whether manual analysis of the raw data files
could improve the differential count compared with refer-
ence microscopy.
Methods: A total of 105 serous fluids (39 peritoneal and
66 pleural effusions) were analyzed by the CELL-DYN
Sapphire using standard whole-blood algorithm. Addi-
tionally, we performed optimized manual gating of the
Sapphire raw data file using standard flow cytometry
software.
Results: The standard Sapphire algorithm showed
substantial deviations from the reference micro scopic
differentiation: polymorphonuclear cell counts were
too high because they contained some monocytic cells.
However, when optimized manual gating strategy is used,
a good correlation and negligible bias were found.
Conclusions: We have demonstrated that with a modified
algorithm, CELL-DYN Sapphire will provide reliable iden-
tification and enumeration of blood cells in peritoneal
and pleural fluids .
Keywords: pleural effusion; mesothelial cells; microscopy .
*Corresponding author: Mathie P.G. Leers, PhD, Department of
Clinical Chemistry and Hematology, Atrium Medical Center, Henri
Dunantstraat 5, 6401 CX Heerlen, the Netherlands,
Phone: + 31-45-576-7503, Fax: + 31-45-576-6575,
E-mail: m.leers@atriummc.nl
Jeffrey F. W. Keuren: Department of Clinical Chemistry and
Hematology, Atrium Medical Center, Heerlen, the Netherlands
Johannes J.M.L. Hoffmann: Abbott Diagnostics Division, Wiesbaden-
Delkenheim, Germany
Introduction
Hematology laboratories frequently perform cellular analy-
sis of serous body fluids such as peritoneal and pleural
fluids. Usually, the total nucleated cells are enumerated and
a differential leukocyte count is done. The latter is impor-
tant for diagnosis and therapy. For example, in peritoneal
fluid, a polymorphonuclear (PMN) cell count > 250 cells/ μ L
is diagnostic for bacterial peritonitis, and this finding
requires immediate antibiotic treatment [ 1 ].
Microscopic cell differentiation and cytological
assessment of cytospin preparations are still considered
the gold standard for analyzing body fluids. Because many
laboratories need to increase their process efficiency, auto-
mated hematology analyzers are increasingly being used
for analyzing body fluids. However, these instruments are
specifically designed for measuring cells in whole blood
and may not be suited for analyzing cells in other body
fluids without modifications [ 2 , 3 ]. Serous fluids have a
different matrix from whole blood, which can affect the
properties of blood cells in a body fluid. Moreover, it is not
unusual that these fluids contain cells of nonhematologi-
cal origin, such as macrophages, mesothelial cells, and
tumor cells, which cannot be accurately classified by the
standard algorithms of hematology analyzers. This was
recently demonstrated for the first time in a study using
the CELL-DYN Sapphire [ 4 ]. These authors found that the
analyzer could be used to determine the concentration of
total nucleated cells with a functional sensitivity limit of
50 cells/ μ L. However, it appeared that CELL-DYN Sapphire
included substantial amounts of epithelial cells and mac-
rophages in the PMN count, making the automated differ-
entiation unreliable [ 4 ]. This PMN overestimation could
lead to false-positive results in the diagnosis of bacterial
peritonitis and pleural cavity infection.
The aims of the present study were to evaluate the
CELL-DYN Sapphire for automated cell differentiation in
serous fluids compared with cytospin microscopy as a
reference and to explore the modifications of the stand-
ard gating strategy for optimizing the automated differ-
ential counts. For this purpose, we analyzed the raw data
files of CELL-DYN Sapphire off-line using a standard flow
cytometry software.
Brought to you by | Abbott Labs Library
Authenticated | hans.hoffmann@abbott.com
Download Date | 5/24/13 8:38 AM
1286Keuren et al.: Hemocytometric analysis of serous body fluids
Materials and methods
Patient samples
A total of 105 serous uids (39 peritoneal e usions and 66 pleural
uids) that were collected for routine diagnostic purposes were ex-
amined in this study. Only the residual materials were used once all
routine tests were completed. Samples were collected in sterile tubes
with K
2 -EDTA as an anticoagulant (Becton Dickinson, Plymouth, UK).
The analyses were performed as soon as possible but always within
3 h from the time of sample collection. The study was performed
according to the Declaration of Helsinki and was approved by the
Ethics Committee of our hospital.
Analysis of body fluids in CELL-DYN Sapphire
standard mode
Before the start of the study, the CELL-DYN Sapphire (Abbott Diag-
nostics Division, Santa Clara, CA, USA) was calibrated with a com-
mercial calibration material (CELL-DYN HemCal Plus) following the
procedure described in the operators manual. CELL-DYN 29 Plus tri-
level controls were used daily in monitoring the performance of the
analyzer.
The CELL-DYN Sapphire optically counts and classi es leuko-
cytes using four angles of laser light scattering: forward (0 ° ), inter-
mediate (7 ° ), polarized, and depolarized side scatter (90 ° and 90 ° D,
respectively). The total leukocyte counts and the di erential counts
of neutrophils, eosinophils, basophils, monocytes, and lympho-
cytes are reported [ 5 ]. The scatter signals of all leukocytes measured
are stored as raw data in a list mode le (see below). The body uid
samples were measured in the so-called Extended Count mode of
the CELL-DYN Sapphire, in which the counting time is prolonged
to 32 s (instead of 8 s in standard CBC mode) to increase counting
precision.
Analysis of body fluids using CELL-DYN
Sapphire raw data files
For every sample measured, CELL-DYN Sapphire stores all raw
measurement data in a list mode le in Flow Cytometry Standard
(FCS) format, which can be easily downloaded from the analyzer for
additional assessment. These raw data les contain the light scat-
ter signals of all cells measured, and in case uorescent antibod-
ies are used, they also contain uorescence data. We loaded the
raw data les of the body uid samples into the FCS Express so -
ware package (version 3; DeNovo So ware, Los Angeles, CA, USA).
Following standard ow cytometry practices, we created in this so -
ware a body uid protocol with optimized gate setting based on ve
randomly chosen test samples using the light scatter signals (see
Figure 1 for an example). First, we identi ed the cellular debris and
noise, which was disregarded. Then, the protocol identi ed the PMN
cells, lymphocytes, monocytic cells, and other cells. A er the proto-
col was established, the gates were stored in the FCS Express so -
ware and retrieved without changes for the analysis of the remaining
100 serous uid samples.
Reference microscopic method
Microscopic di erential cell counts were performed a er the cyto-
centrifugation of the samples (8 min at 400 g ), followed by May-
Gr ü nwald Giemsa staining. Experienced observers microscopically
reviewed the cytospin preparations by counting 200 nucleated
cells. To assess the di erential count, the percentages of the neu-
trophils, basophils, and eosinophils were added and regarded as
PMN cells. Furthermore, the mononuclear cells were classi ed as
lymphocytes and monocytic cells (including macrophages and
mesothelial cells). If present, the (suspected) tumor cells were
noted separately.
Reference immunophenotypic method
The CELL-DYN Sapphire is equipped with a 488-nm solid-
state blue laser that allows three-color fluorescence measure-
ments, similar to regular flow cytometry principles. The instru-
ment has an operational mode in which T-cell subsets are
determined using CD3/CD4 and CD3/CD8 antibody mixtures.
After a timed incubation period, the multiangle optical scatter
and fluorescence signals are measured [ 6 ]. For the present study,
we adapted this procedure by substituting the monoclonal anti-
bodies. All antibodies used were obtained from IQ-Products
(Groningen, the Netherlands). In the first tube, 100 μ L
of serous fluid was mixed with 10 μ L CD3-FITC (clone
UCHT1) and 10 μ L CD19-PE (clone LT19) to analyze the
T- and B-lymphocytes, respectively. In the second tube, 1 μ L of CD45-
FITC (clone ML2) was added to 100 μ L of serous fluid to distinguish
the leukocytes from other cells. In the immunophenotypic mode, the
CELL-DYN Sapphire can collect up to 10,000 white blood cell events
and store the data in a list mode file in FCS format. As indicated
above, the FCS files were processed using the FCS Express software,
and results were reported as percentages of lymphocytes (CD3 + plus
CD19 + ), other leukocytes (CD45 + minus CD3/CD19), and unclas-
sified cells (CD45−). These percentages were used as the reference
immunological differential count.
Statistics
Statistical processing was performed using Analyse-it software
(Leeds, UK). Passing and Bablok analysis was used to determine the
intermethod agreement of the different cell percentages as deter-
mined by reference and automated analyses. In addition, the Spear-
man correlation coefficients between methods were calculated as
well as the mean bias with respect to reference microscopic analysis.
The Wilcoxon test for paired samples was used to determine differ-
ences between methods.
Results
Initially, we compared the CELL-DYN Sapphire automated
leukocyte differential percentages with reference micro-
scopic results of cytocentrifuged serous fluid prepara-
tions. A total of 105 pleural and peritoneal fluids were
Brought to you by | Abbott Labs Library
Authenticated | hans.hoffmann@abbott.com
Download Date | 5/24/13 8:38 AM
Keuren et al.: Hemocytometric analysis of serous body fluids1287
analyzed. As depicted in Figure 2 A – C and Table 1 , there
was a rather poor agreement between the automated and
reference microscopic analyses.
Subsequently, we manually analyzed the FCS files
containing the raw light scatter data using FCS Express
software; in this software, a gating protocol was generated
that was optimized for cells in serous body fluids ( Figure
1 ). This reanalysis substantially improved the line of best
fit for all nucleated cells ( Figure 2 D – F). The mean differ-
ences between the CELL-DYN Sapphire and the reference
microscopic differential counts decreased and the corre-
lation increased when manual FCS file analysis was used
( Table 1 ). Overall, the agreement between automated and
reference microscopic differentiation improved when the
optimized algorithm was used instead of standard CELL-
DYN Sapphire settings for whole-blood analysis.
With respect to monocytic cells, Figure 2 C and F
demon strate that the CELL-DYN Sapphire measured fewer
of these cells compared with the reference microscopic
method. This suggested that at least a subcategory of the
monocytic cells was not identified as such by CELL-DYN
Sapphire.
In addition, we compared the CELL-DYN Sapphire
differential obtained using the improved body fluid
protocol with the reference immunological differential.
As Table 2 shows, we found a good agreement for lym-
phocytes and a satisfactory agreement for the other cell
types.
WBC Differential
A
B
%N 60.8*
%L 32.9*
%E 0.00*
%B 0.00*
32,768
24,576
16,384
All
8192
00
32,768
24,576
16,384
PSS
8192
0 8192 16,384 24,576 32,768
IAS
Gate
None 1153
Lymphocytes 202
Polymorphonuclear cells 32
Mononuclear cells 579
Other cells 47
Debris 189
100.0
17.52
2.78
50.22
4.08
16.39
100.0
17.52
2.78
50.22
4.08
16.39
# of Events % of Gated cells % of All cells
0 8192 16,384 24,576 32,768
All
Band
IG
%M 6.28*
7°-Complexity
0 Size
Figure 1 The effect of the optimized gating strategy on the results of the cell differentiation.
(A) An example of a pleural fluid sample as analyzed by the CELL-DYN Sapphire. Due to a certain amount of debris, a suboptimal separation
of the polymorphic and mononuclear cells is made (%N = % neutrophilic granulocytes, %L = % lymphocytes, %M = % monocytes, %E = %
eosinophilic granulocytes, and %B = % basophilic granulocytes). (B) An optimized, fixed-gating strategy applied on the FCS file generated
by the CELL-DYN Sapphire gave much better separation between the polymorphic and mononuclear cells. Note that this is the same pleural
fluid sample as depicted in (A). ALL represents ° ; IAS, ° ; and PSS, polarized ° scatter. The results generated by the optimized gating
strategy are in concordance with the microscopic analysis of the cytospin: % PMN cells, % lymphocytes, and % mononuclear cells
(monocytes, mesothelial cells, and histiocytes).
Brought to you by | Abbott Labs Library
Authenticated | hans.hoffmann@abbott.com
Download Date | 5/24/13 8:38 AM
1288Keuren et al.: Hemocytometric analysis of serous body fluids
020 40 60 80 100
0
20
40
60
80
100
A
Reference microscopy PMC, %
CD-Sapphire PMC, %
020 40 60 80 100
0
20
40
60
80
100
B
Reference microscopy lymphocytes, %
CD-Sapphire lymphocytes, %
020 40 60 80 100
0
50
100
Reference microscopy monocytic cells, %
CD-Sapphire monocytic cells, %
C
020 40 60 80 100
0
20
40
60
80
100
Reference microscopy PMC, %
CD-Sapphire PMC, %
D
020 40 60 80 100
0
20
40
60
80
100
Reference microscopy lymphocytes, %
CD-Sapphire lymphocytes, %
E
020 40 60 80 100
0
50
100
Reference microscopy monocytic cells, %
CD-Sapphire monocytic cells, %
F
Standard gating
Adapted gati ng
Figure 2 CELL-DYN Sapphire differential cell percentages compared with reference microscopy (n = 105).
On the x-axis, the reference microscopy results are shown, and on the y-axis, the cells as determined by automated differentiation with
CELL-DYN Sapphire standard settings (A – C) and reanalyzed manually in the FCS files (D – F). The black lines correspond to line of best fit
(Passing and Bablok), and the dotted lines represent the line of identity.
Cell type Slope (% CI) Intercept (% CI) Mean bias (% CI) r p
PMN (S) . (. to .) . (. to .) . (. to .) . < .
PMN (BF) . (. to .) . (–. to .) –. (–. to .) . NS
Lymphocyte (S) . (. to .) . (. to .) . (. to .) . < .
Lymphocyte (BF) . (. to .) . (. to .) . (. to .) . < .
Monocytic (S) . (. to .) –. (–. to .) –. (–. to − .) . < .
Monocytic (BF) . (. to .) . (–. to .) –. (–. to –.) . < .
Table 1 Agreement between automated CELL-DYN Sapphire and manual microscopic cell differentiation for PMN, lymphocytes, and mono-
cytic cells.
Differential counting was performed with the CELL-DYN Sapphire standard algorithm (S) and manual FCS file analysis using an optimized
body fluid (BF) protocol. CI, confidence interval; PMN, polymorphonuclear cells; r, Spearman correlation; p, significance (by the Wilcoxon
test for paired samples); NS, not significant.
Discussion
This study was undertaken to explore whether an
improvement of the CELL-DYN Sapphire leukocyte differ-
ential count was possible, using a dedicated body fluid
algorithm instead of the standard approach, which was
designed and validated for cells in whole blood only.
When the standard CELL-DYN Sapphire mode was
used for analyzing cells in serous body fluids, an overesti-
mation of PMNs and lymphocytes and an underestimation
of monocytic cells were seen, compared with the reference
microscopic method. These results are fully in line with the
data recently published by De Smet et al. [ 4 ]. These authors
already suggested that the CELL-DYN Sapphire presumably
Brought to you by | Abbott Labs Library
Authenticated | hans.hoffmann@abbott.com
Download Date | 5/24/13 8:38 AM
Keuren et al.: Hemocytometric analysis of serous body fluids1289
included macrophages and mesothelial cells in the PMN
count rather than in the mononuclear cell count.
When manually reanalyzing the FCS files of the body
fluid samples using an optimized, fixed-gating proto-
col in a standard flow cytometry analysis software, the
agreement between CELL-DYN Sapphire and reference
micro scopy strongly improved. In our opinion, the obser-
vation that PMN percentages obtained with manual FCS
file analysis are no longer significantly different from
reference microscopic results validates its clinical appli-
cability for diagnosing bacterial peritonitis or pleurisy.
Although the correlation of lymphocytic and monocytic
cell percentages between reference microscopy and CELL-
DYN Sapphire analysis also improved when using the
optimized body fluid gating protocol, it was apparent that
CELL-DYN Sapphire was still unable to identify all mono-
cytic cells correctly. Apparently, a subpopulation of body
fluid monocytic cells fall outside the scatter gates of the
standard Sapphire algorithm for blood monocytes and are
not correctly classified. The clinical relevance of missing
these cells is probably limited because clinical decisions
are generally not made on monocyte percentages in
serous fluids.
As a second step, we validated the CELL-DYN Sap-
phire differential count obtained using the optimized
body fluid gating protocol against a limited immunologi-
cal differential count. As Table 2 shows, the agreement
was satisf actory for routine application.
Although we have demonstrated that the standard
algorithm of CELL-DYN Sapphire can be readily improved
for measuring cells in serous body fluids, some limit-
ations have to be considered. First, in this study, we have
only investigated serous body fluids, and therefore, our
results cannot be extrapolated to other body fluids such
as cerebrospinal and dialysis fluids without additional
validation. Second, the current technology of hematology
analyzers does not allow for recognizing or flagging the
presence of tumor cells in serous fluids because tumor
cells have extremely variable morphological properties
and sometimes closely resemble nonmalignant cells, and
thus, efficient flagging remains elusive. Unfortunately,
CELL-DYN Sapphire is not an exception; however, this
analyzer offers the possibility of immunological investiga-
tion of tumor cells using specific monoclonal antibodies.
Finally, automated analysis of body fluids can still not
fully replace expert microscopic assessment. However, in
the diagnostic process of, for instance, bacterial peritoni-
tis and pleural cavity infection, automated analysis with
our improved algorithms can replace microscopic exami-
nation. Furthermore, automated analysis with improved
algorithms can certainly enhance the precision of tra-
ditional microscopic differential leukocyte count and
increase the efficiency of the laboratory workflow and
reduce turnaround time.
In conclusion, we have shown that it is feasible to
make an adaptation in the gating strategy, which enabled
us to obtain a reliable differential count of blood cells in
peritoneal and pleural fluids. When this modified algo-
rithm is incorporated in the future version of the CELL-
DYN Sapphire software, the analyzer would be better fit
for a fully automated analysis of serous body fluids.
Conflict of interest statement
Authors ’ conflict of interest disclosure: The authors stated that there
are no conflicts of interest regarding the publication of this article.
Research funding: None declared.
Employment or leadership: None declared.
Honorarium: None declared.
Received August 28, 2012; accepted November 19, 2012; previously
published online December 15, 2012
Slope (% CI) Intercept (% CI) Mean bias (% CI) r p
Lymphocytes . (. to .) –. (–. to .) . (–. to .) . NS
Other leukocytes . (. to .) –. (–. to .) –. (–. to − .) . < .
Unclassified cells . (. to .) –. (–. to .) . (. to .) . < .
Table 2 Agreement between CELL-DYN Sapphire differential count (obtained from the manual FCS file analysis using the optimized body
fluid protocol) and the reference immunological differential.
CI, confidence interval; r, Spearman coefficient of correlation; p, significance (by Wilcoxon test for paired samples); NS, not significant.
Brought to you by | Abbott Labs Library
Authenticated | hans.hoffmann@abbott.com
Download Date | 5/24/13 8:38 AM
1290Keuren et al.: Hemocytometric analysis of serous body fluids
References
1. Rimola A, Garcia-Tsao G, Navasa M, Piddock LJ, Planas R,
Bernard B, et al. Diagnosis, treatment and prophylaxis of
spontaneous bacterial peritonitis: a consensus document.
International Ascites Club. J Hepatol 2000;32:142 – 53.
2. Kleine TO, Nebe CT, L ö wer C, Lehmitz R, Kruse R, Geilenkeuser
W-J, et al. Modifications of haematology analyzers to improve
cell counting and leukocyte differentiating in cerebrospinal fluid
controls of the Joint German Society for Clinical Chemistry and
Laboratory Medicine. Cytometry 2009;75A:688 – 91.
3. Rabinovitch A. Hematology analyzers and body fluid analysis.
Am J Clin Pathol 2010;134:167 – 8.
4. De Smet D, Van Moer G, Martens GA, Nanos N, Smet L,
Jochmans K, et al. Use of the Cell-Dyn Sapphire hematology
analyzer for automated counting of blood cells in body fluids.
Am J Clin Pathol 2010;133:291 – 9.
5. M ü ller R, Mellors I, Johannessen B, Aarsand AK, Kiefer P, Hardy J,
et al. European multi-center evaluation of the Abbott Cell-Dyn
Sapphire hematology analyzer. Lab Hematol 2006;12:15 – 31.
6. Molero T, Lemes A, De la Iglesia S, Scott CS. Monoclonal antibody
fluorescence for routine lymphocyte subpopulation analysis with
the Abbott CELL-DYN Sapphire haematology analyser. Int J Lab
Hematol 2007;29:446 – 53.
Brought to you by | Abbott Labs Library
Authenticated | hans.hoffmann@abbott.com
Download Date | 5/24/13 8:38 AM
