Académique Documents
Professionnel Documents
Culture Documents
Department of Pathology & transfusion count increments that significantly exceed the number of
Laboratory Medicine and Centre
for Blood Research, University of
transfused platelets. This study demonstrates that the automated
British Columbia, Vancouver, BC, hematology analyzer Sysmex XE-2100 reports erroneously high optical
Canada platelet counts when the blood sample contains particles in the size
Division of Hematopathology, blood samples were spiked with 1 lm latex beads (n = 14) to mimic
Vancouver Acute, Vancouver, BC, contaminants under controlled conditions. Optical and impedance
Canada measurements of spiked and control samples with the Sysmex XE-2100
were compared with the Advia 120 and the manual counts. The added
Correspondence:
Dr Elisabeth Maurer-Spurej, beads unexpectedly increased the automated optical platelet counts in
University of British Columbia, the Sysmex XE-2100 and, to a lesser extent, the Advia 120 (Wilcoxon
Centre for Blood Research, 2350 signed ranks test, P < 0.05), while the beads were not included in the
Health Sciences Mall, Vancouver,
BC, Canada V6T 1Z3.
impedance or the manual microscopic platelet counts. Differential
Tel.: +1 604 827 5993; interference contrast microscopy was used to investigate samples from
Fax: +1 604 822 7135; platelet concentrates for transfusion. Platelet concentrates (32/128)
E-mail: emaurer@interchange. were identified as possible sources for particles that were microscop-
ubc.ca
ically distinct from platelets but would be included in the automated
doi:10.1111/j.1751-553X.2008.01097.x optical count. Transfusion of platelet concentrates containing contam-
inating particles can lead to unexpectedly high post-transfusion platelet
Received 31 January 2008;
counts and misdiagnosis of thrombocytopenic patients.
accepted for publication 7 July
2008
Keywords
Platelets, platelet count, thrombo-
cytopenia, hematology analyzer,
platelet transfusion
(Briggs, Harrison & Machin, 2007). However, the accu- study to determine whether additional particles can
racy of automated platelet counts has seen renewed be present in platelet transfusions and could cause the
interest because of recent reports of erroneous low observed platelet count abnormalities.
platelet counts (Zandecki et al., 2007; Diquattro et al.,
2008) and the ongoing discussion to reduce the trans-
MATERIALS AND METHODS
fusion trigger for prophylactic transfusions (Heddle
et al., 2006).
Study population
Prophylactic platelet transfusions are the standard
care for patients with acute leukemia when their plate- This study was approved by the Institutional Ethics
let count reaches the transfusion trigger, generally 10 Review Committee of the University of British
or 20 109/l (Schiffer et al., 1986; Heddle et al., 2006). Columbia and Vancouver Hospitals and was con-
The determination of pretransfusion platelet counts in ducted in accordance with the Declaration of Helsinki.
patient samples is an important application for auto- Fourteen healthy volunteers gave written informed
mated hematology analyzers although it has long been consent prior to donating whole blood, which was
recognized that the accuracy of automated hematology platelet depleted and analyzed in the absence or after
analyzers decreases with decreasing platelet numbers the addition of latex beads. Forty-nine patients with
(Segal et al., 2005). It is well known that automated acute leukemia were enrolled at the Vancouver Gen-
hematology analyzers, irrespective of the measurement eral Hospital and gave written informed consent to
technique, can fail to discriminate platelets from other allow us access to their routine laboratory results.
cell fragments and debris. The introduction of the Platelet counts 16 h before (precount) and 1 h after
International Reference Method (IRM) (Harrison et al., the administration of a platelet transfusion (1 h post-
2001) has been a major improvement in the calibra- count) were recorded. We compared the measured
tion of automated platelet counts but it requires access and the predicted platelet count increments for 128
to expertise in immunofluorescence flow cytometry transfusions.
(Segal & Harrison, 2006). Adding to the discussion on
the reliability of the optical platelet counts compared
Laboratory tests
with the impedance counts (Briggs & Machin, 2006)
we report here that optical platelet counts were falsely The platelet count of normal whole blood samples
increased when 1-lm sized foreign particles were (EDTA vacutainer; Becton Dickinson, Oakville, ON,
added to blood samples. Canada) was reduced by differential removal of plate-
The maximal increase in platelet count in the reci- lets and not by dilution. Whole blood was centrifuged
pient of a platelet transfusion is expected to be equiv- at 150 g for 12 min. Platelet poor plasma was obtained
alent to the transfusion dose, which is the number of by centrifugation of the supernatant platelet rich plasma
transfused platelets corrected for the blood volume of (PRP) at 2000 g for 15 min and was added back to the
the recipient. Depending on the platelet product, the red blood cell fraction. Differences in nal platelet
platelet dose ranges from 9 109/l for three pooled counts of these articially thrombocytopenic (range 20
whole blood derived platelet units to 39 109/l for 60 109/l) or low-normal whole blood samples (range
apheresis platelets based on a 5 l blood volume. In a 133173 109/l) were probably due to donor variations
recent pilot study we found that approximately 25% and the age of the blood samples. Subsequently, platelet
of platelet transfusions (32/128) caused an increase of depleted samples were split and latex beads (Polybead,
the 1-h post-transfusion platelet count in recipients, 1.0 lm diameter, cat. #07310; Polysciences, Warring-
which significantly exceeded the expected limits ton, PA, USA) were added to one aliquot (+Beads)
(19 5 109/l). To explain these discrepancies, we while the other aliquot served as the control sample.
tested (i) the possibility that the automated hematol- The automated hematology analyzer Sysmex XE-2100
ogy analyzer Sysmex XE-2100 includes particles other (Sysmex, Mississauga, ON, Canada) was used in the
than platelets in the platelet counts and (ii) whether impedance as well as the optical mode. For the man-
platelet concentrates could be the source for these ual platelet counts, EDTA-anticoagulated whole blood
particles. We report the results of a proof-of-principle was diluted with 1% ammonium oxalate (1 : 20),
2008 The Authors
Journal compilation 2008 Blackwell Publishing Ltd, Int. Jnl. Lab. Hem.
E. MAURER-SPUREJ ET AL. ERRONEOUS AUTOMATED PLATELET COUNTS 3
mixed well and incubated for 5 min at room tempera- (ml) = 70body weight (kg). For obese patients the
ture, applied to the hemocytometer (VWR, Toronto, blood volume was calculated according to (Lemmens,
ON, Canada) and allowed to settle for 1520 min Bernstein & Brodsky, 2006):
prior to cell counting on the light microscope (40
dry objective). Samples were also counted with the 70
BV q
3
Advia 120 hematology analyzer (Bayer, Tarrytown, weight= 22 height2
NY, USA), which uses an optical detection method
that is different from the Sysmex XE-2100 utilizing The unit for weight is (kg) and for height is (cm).
two scattering detectors at 23 and 515.
One hundred and twenty-eight post-transfusion
Statistical analysis
remnants of platelet concentrates were sampled using
aseptic technique. Samples were fixed with buffered Statistical analysis was performed using the statistics
paraformaldehyde (2% final concentration) and software SPSS for Windows, version 16.0 (SPSS Inc.,
imaged on a Leica DMRA2 microscope (Leica, Rich- Chicago, IL, USA). The nonparametric Wilcoxon
mond Hill, ON, Canada) with a differential interfer- signed ranks test for paired data was used to deter-
ence contrast (DIC) attachment, 100 oil-immersion mine the statistical significance of the differences
objective (NA = 1.4) and a digital camera (Retiga EXi between automated platelet counts of control and
Fast Cooled Mono 12-bit; QImaging, Surrey, BC, Can- spiked blood samples. This test gives exact probabili-
ada). This setup was also used to image whole blood ties for small sample sizes without knowledge of their
samples in the absence and presence of latex beads distribution. For the calculation of the statistical differ-
because this microscopy technique yields high resolu- ence between measured and predicted 1 h post-
tion images suitable for publication. transfusion platelet counts from thrombocytopenic
leukemia patients the Students t-test for paired sam-
ples was used. The normal distribution of the data sets
Calculation of the transfusion dose
was verified with Q-Q plots.
The following equations were used to calculate the
predicted 1 h post-transfusion platelet counts
RESULTS
(Table 2).
predicted 1 h postcount 109 =l precount platelet dose Automated and manual platelet counts in control and
1 spiked samples
VT platelet content 100
platelet dose 109 =l 2 Figure 1a shows a representative optical scatter plot of
BV VT
a thrombocytopenic whole blood sample (control)
VT is the approximate volume of the platelet transfu- obtained with the Sysmex XE-2100. The optical scat-
sion. For a transfusion of pooled random donor platelet ter plots of thrombocytopenic whole blood samples
rich plasma (PRP) units VT = n60 ml, with n being spiked with latex beads showed that nonfluorescent
the number of single units pooled. For apheresis beads were counted as platelets (Figure 1b). The
platelets or buffy coat pools VT = 360 ml. The mini- addition of approximately 20 109 latex beads /l
mally required platelet content of platelet concentrates increased the platelet counts of 10 artificially throm-
derived from whole blood by the PRP method is bocytopenic and four samples with low-normal plate-
5.5 1010 platelets/unit (60 ml) and 2.8 1011 plate- let counts (Figure 1c, black bars) compared with the
lets/pool (360 ml) for buffy coat platelets (Maurer- control samples (white bars). Latex beads were used
Spurej & Chipperfield, 2007). The platelet content of to mimic natural particles that could affect the optical
single donor platelets derived by apheresis was mea- platelet count due to their size and shape. As
sured and ranged from 2.3 to 5.9 1011 platelets/unit expected, the relative error caused by the added parti-
(360 ml). VT/(BV + VT) is the dilution factor for the cles decreased with increasing platelet counts.
platelet content of the product after transfusion. BV is The Advia 120 also uses an optical method to enu-
the patient blood volume and is calculated as BV merate platelets. However, the measured parameters
2008 The Authors
Journal compilation 2008 Blackwell Publishing Ltd, Int. Jnl. Lab. Hem.
4 E. MAURER-SPUREJ ET AL. ERRONEOUS AUTOMATED PLATELET COUNTS
(c) (d)
(c)
Table 1. Automated platelet counts (109/l) and Wilcoxon signed ranks test
Thrombocytopenic samples 20 47 27 24 24 0 24 28 4
22 42 20 22 24 2 22 24 2
25 52 27 28 29 1 23 29 6
25 53 28 25 26 1 24 31 7
26 52 26 31 30 )1 28 34 6
31 60 29 38 37 )1 33 39 6
39 64 25 44 45 1 43 47 4
40 58 18 45 45 0 46 55 9
48 74 26 50 54 4 49 61 12
60 82 22 65 64 )1 58 63 5
N 10 10 10 10 10 10
Mean 33.6 58.4 37.2 37.8 35 41.1
SD 12.9 12.2 13.8 13.8 13 14.4
Significance (2-tailed) 0.005
0.272 0.005
Low-normal platelet count 133 146 13 161 155 )6 152 141 )11
138 157 19 164 156 )8 157 148 )9
164 176 12 186 180 )6 175 160 )15
173 186 13 200 197 )3 191 185 )6
N 4 4 4 4 4 4
Mean 152 166.25 177.75 172 168.75 158.5
SD 19.5 18.1 18.6 20.3 17.8 19.3
Significance (2-tailed)
0.066 0.066 0.068
(a) (b)
Figure 3. Differential interference contrast (DIC) micrographs of a post-transfusion platelet concentrate sample rep-
resentative for 32 similar samples that caused unexpectedly high post-transfusion platelet counts. Arrowheads mark
particles that differ from platelets to illustrate their relative abundance (a) and the difference in size and morphol-
ogy (b). The scale bars represent 5 lm (a) and 1 lm (b).
patients with acute leukemia cannot be explained counting method for samples from thrombocytopenic
with platelet redistribution triggered by platelet trans- leukemia patients because the risk for erroneously
fusion, or by marrow recovery. While intravenous high platelet counts by the optical method increases
immunoglobulin treatment of thrombocytopenic with decreasing transfusion trigger. The introduction
patients is known to release platelets from the liver of a gate in the Sysmex scatter plot could reduce the
and other stores (Heyns Adu et al., 1986; Tanaka et al., spurious counts. Our data also highlight the necessity
2005), unexpectedly high post-transfusion platelet for pretransfusion quality determinations of platelet
counts are likely caused by erroneous counts of addi- concentrates. New technology for noninvasive, pre-
tional particles contained in the transfusion. These transfusion testing exists but is not yet commercially
particles could be blood cell fragments, bacterial con- available (Maurer-Spurej et al., 2006; Maurer-Spurej
taminants or protein aggregates (von Ahsen et al., & Chipperfield, 2007; Maurer-Spurej, Labrie & Brown,
2001; Latif et al., 2003; Sysmex Europe Science, 2008).
2005; Branda & Kratz, 2006). Their characterization is
a focus of our ongoing research.
ACKNOWLEDGEMENTS
Our data support the previously reported finding
by Briggs, Kunka and Machin (2004) that optical We thank Prof. Paul Harrison for helpful comments
platelet counts obtained with the Sysmex XE-2100 are on the size exclusion of particles in automated hema-
less accurate than the impedance counts. We recom- tology analyzers. This work was supported by a
mend the use of the Sysmex XE-2100 impedance research grant from Canadian Blood Services to EMS.
REFERENCES Palma B. & Menozzi I. (2008) Relation- Latif S., Veillon D.M., Brown D., Kalten-
ships between platelet counts, platelet bach J., Curry S., Linscott A.J., Oberle
von Ahsen N., Ehrlich B., Scott C.S., volumes and reticulated platelets in A. & Cotelingam J.D. (2003) Spurious
Riggert J. & Oellerich M. (2001) Cryo- patients with ITP: evidence for automated platelet count. Enumeration
globulins interfere with platelet counts significant platelet count inaccuracies of yeast forms as platelets by the cell-
by optical and impedance methods but with conventional instrument methods. dyn 4000. American Journal of Clinical
not with the cd61 immunoplatelet International Journal of Laboratory Pathology 120, 882885.
count. Clinical Chemistry 47, 1858 Hematology doi: 10.1111/j.1751- Lemmens H.J., Bernstein D.P. & Brodsky
1860. 553X.2007.01025.x. J.B. (2006) Estimating blood volume in
Branda J.A. & Kratz A. (2006) Effects of Harrison P., Ault K.A., Chapman S., obese and morbidly obese patients.
yeast on automated cell counting. Charie L., Davis B., Fujimoto K., Hou- Obesity Surgery 16, 773776.
American Journal of Clinical Pathology wen B., Kunicka J., Lacombe F., Maurer-Spurej E. & Chipperfield K. (2007)
126, 248254. Machin S., Raynor R., van Hove L. & Past and future approaches to assess the
Briggs C., Harrison P. & Machin S.J. van Assendelft O.W. (2001) An inter- quality of platelets for transfusion. Trans-
(2007) Continuing developments with laboratory study of a candidate refer- fusion Medicine Reviews 21, 295306.
the automated platelet count. Interna- ence method for platelet counting. Maurer-Spurej E., Labrie A. & Brown K.
tional Journal of Laboratory Hematol- American Journal of Clinical Pathology (2008) Routine quality testing of blood
ogy 29, 7791. 115, 448459. platelet transfusions with dynamic light
Briggs C., Kunka S. & Machin S.J. (2004) Heddle N.M., Cook R.J., Sigouin C., Slich- scattering. Particle & Particle Systems
The most accurate platelet count on the ter S.J., Murphy M. & Rebulla P. (2006) Characterization 25, 99104.
Sysmex XE-2100. Optical or imped- A descriptive analysis of international Maurer-Spurej E., Brown K., Labrie A.,
ance? Clinical and Laboratory Haema- transfusion practice and bleeding out- Marziali A. & Glatter O. (2006) Portable
tology 26, 157158. comes in patients with acute leukemia. dynamic light scattering instrument and
Briggs C.J. & Machin S.J. (2006) Discrep- Transfusion 46, 903911. method for the measurement of blood
ancy between impedance and immuno- Heyns Adu P., Badenhorst P.N., Lotter platelet suspensions. Physics in Medi-
fluorescence platelet counting has M.G., Pieters H., Wessels P. & Kotze cine and Biology 51, 37473758.
implications for clinical decision making H.F. (1986) Platelet turnover and Moroff G. (2008) Transfusion of Platelets:
in patients with idiopathic thrombocy- kinetics in immune thrombocytopenic Current Issues. American Red Cross
topenic purpura. British Journal of Hae- purpura: results with autologous 111In- Biomedical Services, Medical and Scien-
matology 135, 416417. labeled platelets and homologous tific Updates, Number 86, http://chap-
51
Diquattro M., Gagliano F., Calabro G.M., Cr-labeled platelets differ. Blood 67, ters.redcross.org/ca/norcal/phys/trans-
Tommasi M., Scott C.S., Mancuso G., 8692. fus_transplat.html.
Schiffer C.A., Lee E.J., Ness P.M. & Reilly ing haematology analysers in severe transient thrombocytopenia after eptifi-
J. (1986) Clinical evaluation of platelet thrombocytopenia and potential impact batide injection in baboons. Thrombosis
concentrates stored for one to five days. on platelet transfusion. British Journal Research 115, 7987.
Blood 67, 15911594. of Haematology 128, 520525. Zandecki M., Genevieve F., Gerard J. &
Segal H.C. & Harrison P. (2006) Methods Sysmex Europe Science (2005) Sepsis; Godon A. (2007) Spurious counts and
for counting platelets in severe throm- Streptococcus Pyogenes Septicaemia. spurious results on haematology analy-
bocytopenia. Current Hematology http://www.sysmex-europe.com/casefo- sers: a review. Part II: white blood cells,
Reports 5, 7075. rum/may05/start.asp. red blood cells, haemoglobin, red cell
Segal H.C., Briggs C., Kunka S., Casbard Tanaka K.A., Katori N., Kelly A.B., Kotze indices and reticulocytes. International
A., Harrison P., Machin S.J. & Murphy H., Marzec U. & Hanson S.R. (2005) In Journal of Laboratory Hematology 29,
M.F. (2005) Accuracy of platelet count- vivo platelet redistribution and acute 2141.