ORIGINAL ARTICLE

Effects of bacteria and yeast on WBC countingin three automated hematology counters

Hye Ryoun Kim & Bo Rae G. Park & Mi Kyung Lee

Received: 2 November 2007 /Accepted: 11 February 2008 /Published online: 27 February 2008# Springer-Verlag 2008

Abstract Bacteria or yeast may be observed on peripheralblood smears and may lead to spuriously elevated plateletcounts. They have been reported to disturb the white bloodcell (WBC) differential count if they clumped together, anda large number of such microorganisms have been shown toincrease WBC counts. The purpose of this study was toevaluate the spurious rise inWBC counts according to speciesof microorganisms and automated hematology analyzers.The species we selected were Staphylococcus aureus,Escherichia coli, Candida albicans, C. tropicalis, C. krusei,C. dubliniensis, C. glabrata, and C. parapsilosis. Weinvestigated the effects of bacteria and yeast on peripheralblood samples by the ADVIA 120/2120 HematologySystem, Sysmex XE-2100 (TOA Medical Electronics,Kobe, Japan) and Coulter LH 750 (Beckman Coulter,Miami, FL, USA). C. albicans, C. tropicalis, C. krusei, andC. dubliniensis had an overt effect on the WBC count atconcentrations of up to 1–5×107 colony-forming units(CFU)/mL in three automated cell counters, and C.glabrata and C. parapsilosis, when present at concentra-tions of 1–5×108 CFU/mL, caused a significant increase inthe WBC count obtained by the Sysmex XE-2100 but notby the ADVIA 120/2120 system and Coulter LH 750 (p<0.05). In conclusion, yeast may influence the results ofperipheral blood smears only when the yeast concentrationis unusually high. The results differed among analyzers andamong species of yeast. Hematologists should be aware thatsamples containing bacteria and yeast may give erroneously

high WBC counts and differential leukocyte counts andshould review the peripheral blood smear by microscopy.

Keywords ADVIA 120/2120 . Sysmex XE-2100 .

Coulter LH 750 . Candidemia

Introduction

Automated cell counters have a central role in thehematology section of modern clinical laboratories. Theuse of automated analyzers in hematology laboratories isnow the rule rather than the exception. These instrumentshave enhanced the precision of results and, with optimalquality control measures in the laboratory, have improvedthe accuracy of tests [1]. Therefore, the vast majority ofresults reported from most modern hematology laboratoriesare released directly from automated instruments, withoutany microscopic review of the blood smear. Most labora-tories perform microscopic examination only on samplesflagged by the automated cell counter as requiring a manualreview. However, there are a variety of conditions whereautomated parameters may be fictitious, including spuri-ously elevated platelet counts caused by microspherocytosisor by bacteremia [2]. Branda and Kratz [3] have reportedthat very high concentrations of C. glabrata and C.parapsilosis significantly cause spuriously elevated plateletcounts and white blood cell (WBC) counts by the ADVIA120/2120 hematology system (Bayer HealthCare, Diagnos-tics Division, Tarrytown, NY, USA). Thus, we investigatedthe effects of bacteria and yeast on peripheral bloodsamples by the ADVIA 120/2120 Hematology system,Sysmex XE-2100 (TOA Medical Electronics, Kobe, Japan),and Coulter LH 750 (Beckman Coulter, Miami, FL, USA).We simulated bacteremia and candidemia by adding known

Ann Hematol (2008) 87:557–562DOI 10.1007/s00277-008-0464-1

H. R. Kim : B. R. G. Park :M. K. Lee (*)Departments of Laboratory Medicine,Chung-Ang University College of Medicine,65-207, 3-Ka Hangang-Ro Yongsan-Ku,Seoul 140-757, South Koreae-mail: cpworld@cau.ac.kr

concentrations of two species of bacteria and six species ofyeast to the blood and analyzed the specimens with theautomated cell counters.

Materials and methods

Subjects

For the purpose of performing the WBC differential countand the platelet count, the peripheral blood specimens werecollected using BD vacutainer K2EDTA plus plastic tubes(Becton Dickinson, Franklin Lakes, NJ, USA). We usedone sample with normal counts (WBC count, 4.5×109/L;platelet count, 150×109/L), one sample with thrombocyto-penia (WBC count, 5.5×109/L; platelet count, 70×109/L),and two samples with pancytopenia (WBC count, <2.5×109/L; platelet count, <100×109/L), which were made bydiluting them with their own plasma.

Instruments

The tested instruments were the ADVIA 120/2120 hema-tology system, Sysmex XE-2100, and Coulter LH 750.

The Coulter LH 750 uses Volume, conductivity, andscatter (VCS) technology to count and classify leukocytes.Cells are identified and classified by simultaneous three-dimensional analysis using volume, conductivity, and lightscatter. Volume measured by direct current is used toidentify the size of the cells. Conductivity or radiofrequency measurement provides information about theinternal characteristics of the cells. Light scatter measure-ment, obtained as the cells pass through the helium–neonlaser beam, provides information about cell surface charac-teristics and cell granularity.

In leukocyte differential analysis by the ADVIA 120/2120, two channels are used to analyze WBCs: (1) aperoxidase channel in which a peroxidase reagent and lightscatter are used to differentiate WBCs by myeloperoxidasecontent and size and (2) a lobularity/nuclear densitychannel in which differential WBC lysis is combined withlight scatter analysis to determine the WBC count,differential data, and information about the nuclear maturityof WBCs. WBC counts for each sample are performedindependently in the lobularity/nuclear density channel andthe peroxidase channel.

The Sysmex XE-2100 measures WBCs by flow cytometryusing a semiconductor laser to detect forward- and side-scattered light information. Red cell lysis is performed using areagent that selectively suppresses the degranulation ofbasophils, resulting in their separation from other forms ofWBCs. In the DIFF channel,WBCs are permeabilized to havetheir deoxyribonucleic acid and ribonucleic acid stain with a

fluorescent dye. The cells are then categorized according totheir side-scattered light and intensity of fluorescence. A five-partWBC differential is created from theWBC populations oflymphocytes, monocytes, eosinophils, neutrophils, and baso-phils. In the immature myeloid information channel, a specialreagent acts on the lipid pattern of the cell membrane toselectively protect immature WBC against disruption, where-as mature leukocytes are disrupted.

Bacteria and yeast

Isolates of S. aureus, E. coli, C. albicans, C. tropicalis, C.krusei, C. dubliniensis, C. glabrata, and C. parapsilosiswere acquired from the American Type Culture Collection(ATCC nos. 25923, 25922, 14053, 750, 6258, MYA 646,2001, and 22019; Manassas, VA, USA). Highly turbidsaline suspensions were prepared from overnight culturesby picking numerous colonies and suspending them insterile 0.9% sodium chloride by the use of an electricvortex. Serial tenfold dilutions were prepared from theundiluted stock suspensions, with four tenfold dilutions inaddition to the stock suspension for each yeast or bacteriumstrain. The cell density of each undiluted stock suspensionwas verified by colony counts of the final tenfold dilutionon Sabouraud dextrose agar or Sheep blood agar todetermine colony-forming units (CFU) per milliliter.

Simulated candidemia

We made 500-μL aliquots with 50 μL of suspensionscontaining C. albicans, C. tropicalis, C. krusei, and C.dubliniensis at concentrations of 1–5×107, 1–5×106, 1–5×105, and 1–5×104 CFU/mL, respectively, and S. aureus, E.coli, C. glabrata, and C. parapsilosis at concentrations of1–5×108, 1–5×107, 1–5×106, and 1–5×105 CFU/mL,respectively. In each case, one aliquot represented anegative control sample, containing the same amount ofnormal saline instead of bacteria or yeast suspension.

Cell count

The analyzers were operated by the hospital technologistsaccording to standard laboratory procedures. Samples werecoded, and the technologists were blinded to the meaningof the codes.

Statistical analysis

Significant increment was defined as WBC count incrementgreater than 20% compared to specimens without bacteriaor yeast. Statistical tests were performed using SPSSversion 13.0. A value of p<0.05 was considered statisticallysignificant, and values were expressed as mean±SD.

558 Ann Hematol (2008) 87:557–562

Results

The presence of bacteria and yeast in blood samples did notaffect red blood cell (RBC) counts and platelet counts onthe three automated cell counters, regardless of theconcentration of bacterium and yeast or type of species.The effects of yeast on WBC counting are shown in Fig 1.Each bacterium had no effect on the WBC count and thedifferential leukocyte count, even at the highest concentra-tion assayed (∼1–5×108 CFU/mL). C. albicans, C. tropi-calis, C. krusei, and C. dubliniensis had an overt effect onWBC counts at concentrations of up to 1–5×107 CFU/mLin three automated cell counters. C. glabrata and C.parapsilosis, when present at a concentration of 1–5×108

CFU/mL, caused a significant increase in WBC countsobtained by the Sysmex XE-2100 (6,110.0±1,993.2) butnot by the ADVIA 120/2120 system and Coulter LH 750(3,637.5±1,957.6, p<0.05). As the number of WBC countsin the samples decreased, the spurious effect of WBCcounts increased.

Yeast was misclassified mainly as a basophil by SysmexXE-2100 (concentrations, 1–5×107 and 1–5×106 CFU/mL),as a lymphocyte and a large unstained cell by ADVIA 120/

2120 (concentrations, 1–5×107 CFU/mL) and as an eosin-ophil by Coulter LH 750 (concentrations, 1–5×107 CFU/mL;Table 1, Fig. 2).

Discussion

The widespread use of automated hematology analyzers hasled to a major improvement in cellular hematology through, forthe most parts, quick and accurate results. Many hematologyanalyzers that enumerate WBCs generate WBC scattergrams,which are the basis for the automated WBC differentialcount. Although the WBC count and the WBC differentialcount are possibly erroneous in many instances, WBCscattergrams allow the detection of abnormalities related tospurious counts [4–8]. In blood cell analysis, each large-sized particle (greater than the size of platelet) that is notdestroyed by hemolytic agents can be identified as a WBCon most hematology analyzers [9, 10]. After enumeration,according to the types of particles, impedance with low-and high-frequency electromagnetic or direct current, laserlight scattering, or peroxidase-staining intensity is used,either individually or together [11, 12].

Fig. 1 The presence of Cadida tropicalis (upper row) had an overteffect on WBC counts at concentrations of up to 1–5×107 CFU/mL inthe Sysmex XE-2100, ADVIA 120/2120 system, and Coulter LH 750.Candida parapsilosis (middle row) caused a significant increase in

WBC counts obtained by the Sysmex XE-2100 but not by the ADVIA120/2120 system and Coulter LH 750 at a concentration of 1–5×108 CFU/mL. E. coli (lower row) had no effect on WBC counts in thethree automated cell counters

Ann Hematol (2008) 87:557–562 559

Our results demonstrated that candidemia models of C.glabrata and C. parapsilosis affected the WBC count andthe differential leukocyte count only at the highestconcentration (1–5×108 CFU/mL) and usually by theSysmex XE-2100 and sometimes by the ADVIA120/2120(only WBC count<2.0×109/L). C. albicans, C. tropicalis,C. krusei, and C. dubliniensis had a significant effect on theWBC count measured by all of the automated cell countersin our simulated candidemia model (concentrations, 1–5×107 CFU/mL). Because the sizes of C. albicans, C.tropicalis, and C. krusei and C. dubliniensis were biggerthan those of C. glabrata and C. parapsilosis, C. glabrataand C. parapsilosis had a smaller effect on WBC countsthan the other Candida species. The methodology used forthe differential count varies between different automatedcounters. The WBC differential count in the Sysmex XE-2100 is based on impedance with direct current, whichreflects the cell size and radio frequency and the internalstructure of cells. The spurious increment of C. albicans, C.tropicalis, C. krusei, and C. dubliniensis was misclassified

as a basophil and that of C. glabrata and C. parapsilosis asa neutrophil. A variety of sizes create differences in WBCidentification. The ADVIA 120/2120 uses two WBC-counting channels: the basophil/lobularity channel and theperoxidase channel. Because of the discrepant effect on thetwo WBC channels, the instrument flagged a comparisonerror of the WBC parameter for all samples with falselyelevated WBC counts. The majority of these spurioussignals were misidentified by the instruments as lympho-cytes because they were small and peroxidase negative. InCoulter LH 750, VCS technology, which involves volu-metric impedance, electrical conductivity/cell volume, anda white cell scatter plot, is used. We reviewed peripheralblood smears by microscopy and found that the yeastformed clusters. Eosinophils were identified on volumetricimpedance (12–14 μm) and scatter plot (high-signal cells)by Coulter LH 750. Thus, yeast clusters may be shown aslarge as eosinophils and may increase cytoplasmic granules.Our results showed that only the Sysmex XE-2100 hadfalse increments of WBC counts in candidemia of C.

Fig. 2 Scattergram generatedby a Sysmex XE-2100 hematol-ogy analyzer. a A negativecontrol sample, containing thesame amount of normal salineinstead of bacteria or yeast sus-pension. b A sample fromCadida tropicalis had an overteffect on WBC counts at con-centrations of up to 1–5×107

CFU/mL (clusters: sky blue,neutrophils and basophils; red,eosinophils; pink, lymphocytes;green, monocytes; blue, ghost)

Table 1 Data of complete blood cell counts in three automated hematology counters

Cell counter Sysmex XE-2100 ADVIA 120/2120 Coulter LH750

No bacteria or yeast Candidemia No bacteria or yeast Candidemia No bacteria or yeast Candidemia

Hemoglobin (g/dL) 7.3 7.0 7.5 7.6 7.2 7.3WBC (×109/L) 2.21 5.25 2.42 4.48 2.30 3.2Platelet (×109/L) 102 110 117 119 112 117Differential count (per μL)Segmented neutrophil 900 1,922 1,089 1,344 1,044 1,405Lymphocyte 999 1,922 961 2,285 1,014 954Monocyte 210 599 213 385 198 163Eosinophil 91 226 90 90 28 675Basophil 11 583 7 367 16LUC 61 376

WBC White blood cell, LUC large unstained cell

560 Ann Hematol (2008) 87:557–562

glabrata and C. parapsilosis. The scattergrams generatedby the Sysmex XE-2100 hematology analyzer showed anatypical distribution (the increment of RBC debris andneutrophils) of the WBC/baso channel. This might beattributed to the result of the moving threshold, whichincompletely separates the WBC population from neutro-phil granulocytes and from RBC remnants. RBCs may notbe destroyed by lysis reagents under certain circumstances,which can lead to a false increase in WBC counts. Thus, itis conceivable that different lysis reagents used in threeinstruments may give different measurements.

Bacteria and yeast may induce falsely elevated plateletcounts because of their presence in vivo. Although it is arare situation, even in septic patients, some bacteria may beobserved on the peripheral blood smear and are associatedwith positive blood cultures. The Candida species mayshow the same size as platelets and may be observed onperipheral blood smears. They have recently been reportedas increased platelet counts in thrombocytopenic patientsinfected by Candida [3, 13–15].

Branda and Kratz [3] have reported that extremely highconcentrations (1–5×108 CFU/mL) of C. glabrata and C.parapsilosis cause spuriously elevated platelet counts. Nosuch effect was observed with C. albicans. All of the threeCandida species significantly increased automated WBCcounts and in a dose-dependent manner at concentrations ofgreater than or equal to 1–5×106 CFU/mL. The Candidaspecies were mainly misidentified as lymphocytes in asimulated candidemia model. Furthermore, a linear rela-tionship between yeast concentrations and WBC countswas noted [3].

However, our study showed that the presence of micro-organisms in blood samples did not affect platelet counts inthe three automated cell counters, regardless of theconcentration of bacterium and yeast or type of species.Because most microorganisms tended to clump, the size ofthe microorganism clumps was bigger than that of platelets.For this reason, the microorganisms were classified asWBCs. In addition, another cause might be the number ofplatelets in mild thrombocytopenic samples (>100×109/L),leaving the platelet count unaffected by bacteria and yeast.Since the size of yeast ranged from 2 to 4 μm, the inabilityof the analyzer to distinguish yeast from platelets is readilyexplained. Subsequent analysis of larger yeast, including C.albicans and C. parapsilosis, did not show the increase,which was similar to that of our study. Moreover, WBCstended to increase in a dose-dependent manner, but a linearrelationship between yeast concentrations and WBC countswas not observed, which was different from the results of aprevious study [3].

Pseudoeosinophilia sometimes presents with no incre-ment of WBC counts in the Coulter LH 750, probablybecause yeast might show atypical light-scattering patterns.

True bacteremia and candidemia usually results in a smallnumber of bacteria and yeast in the blood smear, unlike thesimulated samples presented here, which had numerousclumps of bacteria or yeast. However, the presence ofbacteria or yeast sometimes interferes with the differentialleukocyte count regardless of its concentration.

In summary, we demonstrated a cause and effectrelationship between the presence of bacteria and yeastand the differences of automated hematology analyzers.Although the presence of yeast can influence the WBCcounts generated by automated hematology analyzers,spurious results will be flagged by the analyzers for manualreview. Yeast may influence the results of peripheral smearsonly when the concentrations are unusually high. Theresults differ in analyzers and among yeast species.Hematologists should be aware that samples containingbacteria and yeast may give erroneously high WBC countsand differential leukocyte counts and should review theperipheral blood smear by microscopy.

References

1. Bentley SA, Johnson A, Bishop CA (1993) A parallel evaluationof four automated hematology analyzers. Am J Clin Pathol100:626–632

2. Zandecki M, Genevieve F, Gerard J, Godon A (2007) Spuriouscounts and spurious results on haematology analysers: a review.Part I: platelets. Int J Lab Hematol 29:4–20

3. Branda JA, Kratz A (2006) Effects of yeast on automated cellcounting. Am J Clin Pathol 126:248–254

4. Hur M, Lee YK, Lee KM, Kim HJ, Cho HI (2004) Pseudobaso-philia as an erroneous white blood cell differential count with adiscrepancy between automated cell counters: report of two cases.Clin Lab Haematol 26:287–290

5. Zandecki M, Genevieve F, Gerard J, Godon A (2007) Spuriouscounts and spurious results on haematology analysers: a review.Part II: white blood cells, red blood cells, haemoglobin, red cellindices and reticulocytes. Int J Lab Hematol 29:21–41

6. Grimaldi E, Carandente P, Scopacasa F, Romano MF, PellegrinoM, Bisogni R, De Caterina M (2005) Evaluation of the monocytecounting by two automated haematology analysers compared withflow cytometry. Clin Lab Haematol 27:91–97

7. Huh J, Jung J, Yoon H, Chung W (2005) Pseudoeosinophiliaassociated with malaria infection determined in the Sysmex XE-2100 hematology analyzer. Ann Hematol 84:400–402

8. Ducrest S, Meier F, Tschopp C, Pavlovic R, Dahinden CA(2005) Flowcytometric analysis of basophil counts in humanblood and inaccuracy of hematology analyzers. Allergy 60:1446–1450

9. Sandhaus LM, Osei ES, Agrawal NN, Dillman CA, MeyersonHJ (2002) Platelet counting by the Coulter LH 750, SysmexXE 2100, and Advia 120: a comparative analysis using theRBC/platelet ratio reference method. Am J Clin Pathol 118:235–241

10. Elghetany MT, Hudnall SD (1996) Spurious automated white cellcount with Coulter STKS in the myelodysplastic syndromessuggests the presence of a red cell membrane defect. Am JHematol 52:69

Ann Hematol (2008) 87:557–562 561

11. Bourner G, Dhaliwal J, Sumner J (2005) Performance evaluationof the latest fully automated hematology analyzers in a large,commercial laboratory setting: a 4-way, side-by-side study. LabHematol 11:285–297

12. Nakul-Aquaronne D, Sudaka-Sammarcelli I, Ferrero-Vacher C,Starck B, Bayle J (2003) Evaluation of the Sysmex Xe-2100hematology analyzer in hospital use. J Clin Lab Anal 17:113–123

13. Kakkar N (2004) Spurious rise in the automated platelet countbecause of bacteria. J Clin Pathol 57:1096–1097

14. Latif S, Veillon DM, Brown D, Kaltenbach J, Curry S, LinscottAJ, Oberle A, Cotelingam JD (2003) Spurious automated plateletcount. Enumeration of yeast forms as platelets by the cell-DYN4000. Am J Clin Pathol 120:882–885

15. Arnold JA, Jowzi Z (1999) Images in haematology. Candidaglabrata in a blood film. Br J Haematol 104:1

562 Ann Hematol (2008) 87:557–562