Automation of Bone Marrow Aspirate ExaminationUsing the XE-2100 Automated Hematology Analyzer

Yusuke Mori,1* Toshihiro Mizukami,1 Yukio Hamaguchi,1 Kazuto Tsuruda,2 Yasuaki Yamada,2 andShimeru Kamihira2

1Product Development Division, Sysmex Corporation, Kobe, Japan2Department of Laboratory Medicine, Nagasaki University Graduate School of Biomedical Sciences, Nagasaki, Japan

Background: Attempts to analyze bone marrow aspirates have been reported with the use of severalautomated blood cell counters, but sufficient accuracy in examination is not acquired yet. Major problemshave included difficulties in correctly differentiating various immature cells and interference by lipid in bonemarrow aspirates. The goal of this study was to solve these problems to attain more accurate assessmentof bone marrow aspirates with automated blood cell counters.

Methods: We modified the XE-2100 Automated Hematology Analyzer (Sysmex Corporation, Kobe, Japan)to fit it for bone marrow aspirate measurement and evaluated its performance. Measurements wereperformed with the modified XE-2100 on 81 patient samples of bone marrow aspirates; as a reference, themanual visual method was used and flow cytometric analysis were carried out.

Results: Good correlations between results with the modified XE-2100 and the manual visual method wereobtained for total nucleated cell count (TNCC; r � 0.99), erythroblast/TNCC ratio (r � 0.93), and myeloidcell/TNCC ratio (r � 0.75).

Conclusions: When this device is used, bone marrow aspirate differentials can be determined quickly andeasily. This device will be useful for preliminary examination to obtain a summary of various blood cell ratiosin bone marrow aspirates before performance of microscopic examination. © 2003 Wiley-Liss, Inc.

Key terms: automated hematology analyzer; bone marrow aspirate; total nucleated cell count; myeloid toerythroid ratio

Examination of bone marrow aspirates is indispens-able for the diagnosis of various hematologic diseases,metastasis of cancer to bone marrow, and monitoring ofthe myelogenous effects of chemotherapy. At present,bone marrow aspirate analysis is conventionally per-formed microscopically. This manual method is impre-cise due to the small number of cells counted and it islabor intensive and time consuming. If analysis of bonemarrow is possible with automated hematology analyz-ers, then faster and more objective examination will bepossible. Trials of analysis of bone marrow aspirateshave been reported with several automated hematologyanalyzers (1– 4), but sufficient accuracy has not beenobtained due to various technical problems. Majorproblems have included failure to count nucleated redblood cells (NRBCs), difficulty in correctly differentiat-ing various immature cells, and interference by lipid inbone marrow. However, remarkable advances in tech-nology has enabled automated blood cell counters todifferentiate and count immature granulocytes (IGs)and NRBCs, in addition to the white blood cell (WBC)differential. The XE-2100, which uses a flow cytometry

(FCM) technique and a nucleic acid staining dye incombination, can perform WBC differentials, IG differ-entials, and NRBC differentials. With the expectationthat this device will solve the aforementioned problemsand be available for bone marrow aspirate examination,we made appropriate modifications to it and evaluatedits performance in cell differential analysis and countingas an automated bone marrow aspirate analyzer.

MATERIALS AND METHODSPatients

The samples used in this study were 81 bone marrowaspirates obtained as routine samples at Nagasaki Univer-sity Hospital (between September 2001 and March 2002).Table 1 lists the diseases.

*Correspondence to: Yusuke Mori, Product Development Division, Sys-mex Corporation, 4-4-4 Takatsukadai, Nishiku, Kobe, 651-2271, Japan.

E-mail: yusuke_mori@sysmex.co.jpReceived 7 May 2003; Accepted 23 September 2003

DOI: 10.1002/cyto.b.10070

Cytometry Part B (Clinical Cytometry) 58B:25–31 (2003)

© 2003 Wiley-Liss, Inc.

Methods

Differential counting was carried out with the modifiedXE-2100, and its performance was assessed by comparingcell differential results with those of manual and FCMmeasurements.

Modifications of the XE-2100

The XE-2100 (Sysmex Corporation, Kobe, Japan) wasmodified to enable bone marrow aspirate examination.There were two major modifications: (a) a computer pre-pared for multiparameter analysis was connected to theXE-2100 so that the optical signals (forward scatter light,side scatter light, and side fluorescence) from the XE-2100’s optical detector were sent to it and analyzed there;and (b) a special mode (bone marrow measurementmode) was added to enable immediate measurement atthe optical detector without dilution or lysing, becausethese processes were not required for myeloid cell/totalnucleated cell count (TNCC) ratio measurement in whichpretreated samples were used.

Determination of TNCC and Erythroid Cell/TNCC Ratio

Bone marrow samples were obtained by the usualmethod from the posterior iliac crest, anticoagulated withheparin, diluted with RPMI medium, and filtered with100-�m pore filters to remove several sources of interfer-ence, such as bone chips and clots. Measurement wascarried out in the XE-2100 manual mode. The Diff_WBC#value, one of the XE-2100 research parameters, was usedfor the TNCC. The erythroid cell/TNCC ratio was deter-mined by three-parameter analysis in the analyzing com-puter by using the forward scatter light, side scatter light,and side fluorescence received from the NRBC channel sothat the interference due to lipid was removed (the NRBCchannel of the current XE-2100 enables two-parameteranalysis by using forward scatter light and side fluores-cence).

Determination of Myeloid Cell/TNCC Ratio

The myeloid cell/TNCC ratio was determined by usingpretreatment samples in the bone marrow measurement

mode added for this study. A mixture of 1 ml of Stroma-tolyser IM (the reagent for the XE-2100 and the SE-9000automated hematology analyzers; Sysmex Corporation)and 20 �l of 50 ppm polymethine dye were incubated at33°C, 20 �l of a bone marrow aspirate sample was addedto it, and 7 s later measurement was performed in thebone marrow measurement mode. Figure 1 shows anexample of measurement of a bone marrow aspirate. Stro-matolyser IM, the reagent for the immature informationchannel of the XE-2100 and SE-9000, selectively damagesmembranes of cells and exposes their nuclei, except forthe immature series of granulocytes. The nucleic acidstaining dye contained in the reagent stains NRBCs, thelymphocytic series, monocytes, and mature granulocytesand leaves myeloblasts and IGs unstained. Lymphoblastsare included in the lymphocyte � NRBC cluster. Themyeloid series ratio was obtained from the sum of themyeloblast, IG, mature granulocyte, and monocyte ratios.

Manual Counting and Differential

TNCC counting was performed after diluting the bonemarrow aspirates 5 to 50 times with Turk’s solution in aBurker-Turk chamber. Differential counting was per-formed manually on May-Grunwald Giemsa-stained smearsprepared immediately after obtaining the bone marrowaspirates (300 counts). Because cell type categorizationwith manual differential counting was finer than that withthe modified XE-2100 (Fig. 1), the manual differentialresults were adjusted as follows: myeloblasts were definedas myeloblasts; promyelocytes, myelocytes, and metamy-elocytes as IGs; band forms, segmented neutrophils, eo-sinophils, and basophils as mature granulocytes; IGs andmature granulocytes as granulocytes; pronormoblasts, ba-sophilic erythroblasts, polychromatophilic erythroblasts,and orthochromatic erythroblasts as NRBCs; and lympho-

Table 1.Characteristics of Patients

Diseases n

Acute myelocytic leukemia 18Myelodysplastic syndrome 12Acute lymphocytic leukemia 13Chronic lymphocytic leukemia 1Chronic myelogenous leukemia 2Malignant lymphoma 16Adult T-cell leukemia/lymphoma 6Multiple myeloma 6Anemia 2Idiopathic thrombocytopenic purpura 1Epstein-Barr virus infection 2Macroglobulinemia 2

Total 81

FIG. 1. Scattergram of the myeloid cell counting method with themodified XE-2100.

26 MORI ET AL.

cytes, atypical lymphocytes, and lymphoblasts as lympho-cytes.

FCM Differential

The flow cytometer used in this study was a FACSCalibur (BD Corporation; San Jose, CA). The NRBC ratiowas determined by the NRBC reference measurementmethod (5) using anti-CD45 antibody and propidium io-dide.

The blast, IG, lymphocyte, monocyte and granulocyteratios were determined by the IG reference measurementmethod (6) using anti-CD11b, anti-CD16, and anti-CD45antibodies. Because NRBC was not included in the celldifferential results of the IG reference measurementmethod, the ratio including NRBC was obtained by thefollowing conversion formula:

ratio of each cell type (%) � ratio of the cell type

obtained from IG reference measurement

� (100 � NRBC ratio obtained from

NRBC reference measurement)/100 (%)

RESULTSFigure 2 shows some examples of modified XE-2100

measurements for normal peripheral blood and bone mar-row aspirates; they are scattergrams of the WBC/BASOchannel for the WBC/basophil ratio, the Diff channel forthe WBC differential excluding basophils, and the NRBCchannels for the NRBC ratio (from left). NRBC channel 1was identical to the NRBC channel of the current XE-2100and designed to provide the side fluorescence minus for-

ward scatter light scattergram. NRBC channel 2 wasunique to the modified XE-2100 and designed to providethe side fluorescence minus side scatter light scattergramin which the NRBC channel 1 scattergram was developedalong different axes. As indicated by the white arrows onthe scattergrams, the bone marrow samples exhibiteduniquely shaped clusters that were not observed on pe-ripheral blood measurement. The bone marrow aspirateswere centrifuged to remove lipid content and then mea-sured by the modified XE-2100. These clusters were con-firmed to be associated with the lipid content becauseonly they appeared on the scattergrams. The lipid particle-associated clusters were found in the same regions as

FIG. 3. Correlation for total nucleated cell count between the manualmethod and the modified XE-2100.

FIG. 2. Comparison for XE-2100 scattergram between normal PB sample and bone marrow (BM) sample. The sigmoid clusters in the WBC/BASO andNRBC channels are due to fat particles.

27AUTOMATED BONE MARROW ASPIRATE EXAMINATION WITH THE XE-2100

blood cells on the WBC/BASO channel and NRBC channelscattergrams, but those on the Diff channel scattergramsappeared in the ghost region.

Figure 3 shows the correlation between results ob-tained with the modified XE-2100 and the manual methodfor TNCC. The values for the modified XE-2100 wereDiff_WBC# from the Diff channel, which did not exhibitinterference by lipid. An excellent correlation betweenthe two methods was obtained for TNCC (r � 0.99).

Figure 4 shows correlations between results obtainedwith the modified XE-2100 and the FCM method. Corre-lations were good for monocytes (Fig. 4b; r � 0.89), IGs(Fig. 4d; r � 0.73), granulocytes (Fig. 4e; r � 0.97), anderythroid cell/TNCC ratio (Fig. 4f; r � 0.96). However,several samples exhibited deviations in the other catego-ries (Fig. 4a, 4c, 4g). These samples with deviation werefrom the same patients, two cases of M1 acute myeloid

leukemia (AML) and two cases of acute lymphocytic leu-kemia (ALL), and all exhibited high myeloblast ratios inthe FCM results and high lymphocyte � NRBC ratios withthe modified XE-2100. Their presence limited the correla-tion coefficient for the myeloid cell/TNCC ratio to r �0.65 (Fig. 4g).

Figure 5 shows the correlation between the resultsobtained with the modified XE-2100 and the manualmethod. Deviations were found in the same categories asthose shown in Figures 4a, 4c, and 4g. However, thenumber of cases with deviation was two, both AML M1patients. Both exhibited high myeloblast ratios on themanual examination and high lymphocyte � NRBC ratioswith the modified XE-2100 (Fig. 5a and 5c). Although anexcellent correlation was found for the erythroid cell ratio(Fig. 5f; r � 0.93), the presence of samples with deviationlimited the correlation coefficient for the myeloid cell/

FIG. 4. a–h: Correlation between the FCM method and the modified XE-2100.

28 MORI ET AL.

TNCC ratio to r � 0.75 (Fig. 5g). As seen from the smearphoto (Fig. 6), these cells had atrophied greatly.

DISCUSSIONSome attempts have been made to apply automated

blood cell counters to bone marrow aspirate measure-ment (1–4), with an increasing number of reports beingpresented especially after the appearance of blood cellcounters capable of NRBC counting. However, satisfac-tory results have yet to be obtained. Major problems haveincluded interference by lipid in bone marrow aspiratesand difficulties with correct differentiation of various im-mature cells. To solve these problems, we modified theXE-2100 and carried out bone marrow aspirate differentialcounting. Previous studies have compared results of dif-ferential counting with automated hematology analyzers

with only manual measurements. In this study, cell surfaceantigen analysis using a flow cytometer was adopted as anadditional reference for comparison.

Conventional automated blood cell counters count thelipid particles contained in bone marrow aspirates asblood cells, thereby falsely increasing the TNCC or caus-ing errors in cell differential results (1,7). The same wasfound in the XE-2100 WBC/BASO channel and NRBCchannel scattergrams (Fig. 2), which included the lipidparticle clusters in the same regions as the blood cells.The XE-2100 WBC/BASO channel differentiates variouscells according to size and morphological informationobtained from the forward and side scatter light signals.This suggested that the lipid particles were not correctlydifferentiated from blood cells because of their similarityin optical information.

FIG. 5. a–h: Correlation between the manual method and the modified XE-2100.

29AUTOMATED BONE MARROW ASPIRATE EXAMINATION WITH THE XE-2100

The fluorescence intensity, obtained by staining thenucleic acid within a cell with a polymethine dye, is usedas a parameter in the NRBC channel and the Diff channel.The lipid particle cluster appearance of the NRBC channelis presumably due to the nonspecific adsorption of thedye on the lipid particles. However, the cluster of lipidparticles was not observed in the Diff channel. Hence, theDiff channel, which is not affected by the presence oflipid particles, correlated well with the manual cell count(Fig. 3).

To measure the erythroid cell ratio in bone marrowaspirates, it was necessary to eliminate the effect of lipidparticles on the NRBC channel. The interference by lipidparticles observed in the NRBC channel was excluded byperforming a three-parameter analysis, where side scatterlight intensity was added to the conventional two param-eters, forward scatter light and side fluorescence. Whereasthe lipid particle and the NRBC clusters overlap in theNRBC channel 1 (side fluorescence minus forward scatterlight) scattergram, the lipid particle cluster has higher sidescatter intensity than does the NRBC cluster in the NRBCchannel 2 (side fluorescence minus side scatter light)scattergram. Because the two clusters appear in differentpositions on the NRBC channel 2 scattergram, we classi-fied the NRBC proportion after eliminating the lipid par-ticle cluster. We consider two possibilities for higher sidescatter light intensity of lipid particles: (a) lipid has higherrefractive index than blood cells and (b) the aggregationof many small lipid particles causes the side scatter inten-sity to increase.

As shown in Figures 4 and 5, excellent correlationswere observed between the modified XE-2100 cell differ-ential results and those for the FCM and manual methods,with the exception of several samples exhibiting devia-

tion. The results obtained with the modified XE-2100 alsoexhibited improvements from results noted in previousreports. The samples exhibiting deviation were found inthe categories of lymphocyte � NRBC, blast, and myeloidcell/TNCC ratio. Two of them were cases of ALL whoseresults from the FCM method did not match the modifiedXE-2100 or manual results. The blast gating for the FCMCD45/SCC cytograms allows lymphoblasts and other he-matopoietic tumor cells to be extracted together withmyeloblasts (8,9). However, with the modified XE-2100,only myeloblasts are included in the myeloblast area (Fig.1), and lymphoblasts are included in the lymphocyte �NRBC area (Fig. 1). Therefore, measurements of highlymphoblast samples differed between the two methods(in both samples lymphoblasts accounted for more than80% in the manual differential). This was proven by thefact that no problems attributable to the ALL cases wereobserved when results obtained with the modified XE-2100 were compared with results obtained with the man-ual method (Fig. 5).

The other samples exhibiting deviation were from AMLM1 cases. Their modified XE-2100 results did not matchthose of the manual and FCM methods. In these samples,although myeloblasts accounted for more than 60% in themanual differential, the lymphocyte � NRBC cluster ob-tained from the modified XE-2100 accounted for morethan 80%. As shown in Figure 6, the myeloblasts in thisclinical case atrophied extensively to assume the maturelymphocyte structure. Because these cells atrophied, it issupposed that these cells were denatured.

The reagent used to determine the myeloid cell ratiodamages cell membranes, except those of immature my-eloid cells, and the dye contained in the reagent stains thenucleic acids in them for cell differential counting. We

FIG. 6. Smear photo of the deviationsample.

30 MORI ET AL.

believe that, in this clinical case, because the cell mem-brane of myeloblasts, which are not normally damaged,were damaged due to cell denaturation, they were stainedwith nucleic acid staining dye and appeared as a clusterwithin the lymphocyte � NRBC region.

This device was found to be capable of quickly andeasily determining the TNCC and the myeloid to erythroidM/E ratio in bone marrow aspirates. It will be useful forpreliminary examination to obtain a summary of bonemarrow aspirates before performance of in-depth manualexamination.

We believe the cell differentiation function of a hema-tology analyzer can be improved by using the device thatwe report here. When this method is applied to peripheralblood measurement, myeloblasts could be counted rou-tinely, and the sensitivity of lymphoblast detection, aweakness in the hematology analyzers currently on the mar-ket, could be made better. The automation of cell differen-tiation could be made more objective than morphologicobservation, which depends on a medical technician’s sub-jectivity. We believe that our proposal could make a signifi-cant advancement in the function of a hematology analyzerand play an important role in day-to-day clinical use.

LITERATURE CITED1. Bentley SA, Taylor MA, Killian DE, et al. Correction of bone marrow

nucleated cell counts for the presence of lipid particles. Am J ClinPathol 1995;104:60–64.

2. Tatsumi J, Tatsumi Y, Tatsumi N. Counting and differential of bonemarrow cells by an electronic method. Am J Clin Pathol 1986;86:50–54.

3. Sakamoto C, Yamane T, Ohta K, Hino M, Tsuda I, Tatsumi N. Auto-mated enumeration of cellular composition in bone marrow aspiratewith the CELL-DYN4000TM automated hematology analyzer. ActaHaematol 1999;101:130–134.

4. D’Onofrio G, Zini G, Tommasi M, Laurenti L, Vergine G, Van Hove L.Quantitative bone marrow analysis using the Abbott CELL-DYN 4000hematology analyzer. Lab Hematol 1997;3:146–152.

5. Tsuji T, Sakata T, Hamaguchi Y, et al. New rapid flow cytometricmethod for the enumeration of nucleated red blood cells. Cytometry1999;37:291–301.

6. Fujimoto H, Sakata T, Hamaguchi Y, et al. Flow cytometric method forenumeration and classification of reactive immature granulocyte pop-ulations. Commun Clin Cytometry 2000;42:371–378.

7. Zini G, Mistretta G, Giordano G, Laurenti L, D’Onofrio G. Automatedanalysis of bone marrow aspirate with the Sysmex XE-2100 TM bloodcell counter. Infus Ther Transfus Med 2001;28:277–279.

8. Stelzer GT, Shults KE, Loken MR. CD45 gating for routine flowcytometric analysis of human bone marrow specimens. Ann NY AcadSci 1993;677:265.

9. Borowitz MJ, Guenther KL, Shults KE, Stelzer GT. Immunophenotyp-ing of acute leukemia by flow cytometry analysis. Use of CD45 andright-angle light scatter to gate leukemic blasts in three-color analysis.Am J Clin Pathol 1993;100:534–540.

31AUTOMATED BONE MARROW ASPIRATE EXAMINATION WITH THE XE-2100