Flow cytometry CD45 gating for immunophenotyping of acute myeloid leukemia

Transcription

1 Leukemia (1997) 11, Stockton Press All rights reserved /97 $12.00 Flow cytometry CD45 gating for immunophenotyping of acute myeloid leukemia F Lacombe, F Durrieu, A Briais, P Dumain, F Belloc, E Bascans, J Reiffers, MR Boisseau and Ph Bernard Laboratoire d Hématologie, Service des Maladies du Sang, Hôpital du Haut-Lévêque, Pessac, France A flow cytometry method has been introduced into the routine age values within the malignant blast cell populations or investigation of whole bone marrow samples following red within the total nucleated cells present in each sample. In blood cell lysis on the basis of a primary CD45/side scatter order to give numbers of malignant cells and proportions of (SSC) gating procedure. Blast cells were first identified by CD45/SSC gating in 74 cases of acute myeloid leukemia (AML) cells within this cohort, a reliable practical discrimination and the results were compared to a conventional FSC/SSC gating procedure and to MGG-staining smears. The percentages required. In this paper, as previously recommended by between malignant blasts and normal cell types would be of blast cells in these samples as defined by the morphological Borowitz et al, 6 we suggest that such a discrimination can be analysis of MGG smears correlated better with the values readily facilitated by introducing primary gating for CD45 determined by CD45/SSC gating (r 0.94) than with the blast antigen expression and side scatter (CD45/SSC). We also sugcell counts recorded with FSC/SSC gating (r 0.76). These findings were not surprising because while CD45 expression gest that this step should replace the first gating step for for- was regularly lower on leukemic blasts than on normal lymphoid and monocytic cells, the FCS/SSC characteristics of these does not discriminate well between leukemic blasts, lympho- ward scatter and side scatter (FSC/SSC), as this latter procedure populations were overlapping. In 53 samples, the blast cell cytes and monocytes. populations were also analyzed with a panel of FITC-conju- In this paper we present the application of double and triple gated monoclonal antibodies that were utilized in double labeimmunofluorescence (IF) analysis of bone marrow samples ling with CD45-PE. We show that the CD45/SSC gating protaken from patients during the presentation of AML. With the cedure improved phenotypic determination of the blast cells in three ways: (1) by discriminating between leukemic blast cells systematic use of leukocyte common antigen (CD45) marker and residual normal cells; (2) by excluding normal cells from in combination with lineage-specific markers, a good disthe phenotypic analysis of leukemic blast cells; and (3) by crimination can be achieved between the blast cell popuidentifying blast cell heterogeneity in many cases of leukemia lations and the normal cells. This discrimination is based on on the basis of different CD45 display. Moreover, this immunothe fact that the precursor cells in the bone marrow, as well phenotyping procedure on whole bone marrow samples also allowed an efficient discrimination between the various cell linlow and intermediate values of CD45, while lymphocytes and as the leukemias which derive from these cell types, express eages and facilitated the analysis of leukemic blasts present in low proportions. monocytes express high levels of CD45. 7 Keywords: acute myeloid leukemia; immunophenotyping; CD45; We also present evidence that with conventional FSC/SSC flow cytometry gating miscalculations of blast cell percentages can occur, which can lead to an increase in both false positivity and false negativity in the percent values of leukemic blasts expressing Introduction different antigens. Furthermore, the use of CD45/SSC gating facilitates the expression of results, primarily gated on The characterization of acute leukemias is based on a multi- CD45 low blast cells, in a uniform manner that provides similar parametric analysis which includes clinical features, cell mor- observations in lysed whole bone marrow and in enriched phology, genetics and immunological markers. These markers mononuclear populations following Ficoll separation. Our have been shown to be important for the diagnosis and prog- observation adds to previous findings by Borowitz et al, 6 Shah nosis of acute myeloid leukemia (AML), particularly when the et al, 7 Stelzer et al 8 and Rainer et al 9 and indicates that precise lineage of the blast cells cannot be defined by light CD45/SSC gating may provide a common platform for microscopy morphology and cytochemistry. 1 3 Recently, both uniform data processing during the immunodiagnosis of AML. the European Group for the Immunological Characterization of Leukemias (EGIL) 4 and the European Working Group on Flow Cytometry and Image Analysis 5 have identified the most Materials and methods important lineage-specific markers for the myeloid and lymphoid lineages. Nevertheless, there is still disagreement in two Specimens areas. First, it is not yet decided whether the blast cells should be enriched by Ficoll-density gradient centrifugation prior to Seventy-four bone marrow samples were consecutively colphenotypic analysis or should be studied in the whole bone lected over a 12-month period from adult patients with a susmarrow following red cell lysis. While opinions tend to favor pected diagnosis of de novo AML. Bone marrow (2 3 ml) was the study of lysed whole bone marrow, samples are still fre- aspirated into EDTA K3 tubes (Vacutainer; Becton Dickinson, quently Ficolled for cryopreservation and storage. Thus, Pont de Claix, France) and received in the laboratory within phenotypic methods which are suitable for analyzing leu- 1 h of collection. Heparinized samples were not suitable for kemic blast cells following both sample handling procedures the bone marrow lysis method used in our laboratory. could be very useful. The second related issue is that it is not yet agreed whether the results should be expressed as percent- Correspondence: F Lacombe, Laboratoire d Hématologie, Hôpital du Haut-Lévêque, Avenue Magellan, Pessac, France; Fax: Received 7 March 1997; accepted 1 August 1997 Patient population The patients included 38 women and 36 men and their age ranged from 16 to 84 years (median 58.5). The morphological analysis and differential count (excluding erythroid cells) were

2 performed on May Grünwald Giemsa-stained smears by two experienced readers. Diagnosis was made according to the FAB classification, 10 cytochemistry and immunophenotyping, as AML0 (n = 8), AML1 (n = 9), AML2 (n = 23), AML3 (n = 10), AML4 (n = 9), AML5 (n = 10), AML6 (n = 1), AML7 (n = 1), biphenotypic acute leukemia (n = 3) as defined by GEIL. 11 Flow cytometry immunophenotyping of AML and CD45/SSC gating were adjusted using a CD4-FITC/CD8-PE staining of whole blood lymphocytes and levels for positivity were determined according to PE- and FITC-conjugated isotypic Ig control (IgG1-FITC and IgG1-PE from Coulter). A first step (gate A) was to exclude debris and fat cells on a FSC/SSC histogram; cells were counted in gating A and data were recorded in list mode Immunofluorescence (IF) staining Data analysis IF staining was performed on fresh cells by direct IF as follows. List mode data on viable cells (gate A; see above) were analyzed Samples were diluted cells/ml with AB-positive pooled using the Coulter Elite (version 4.01) or Coulter System plasma (instead of PBS; see below) and 50 l were incubated II programs. On the CD45/SSC histograms four gates were set for 30 min at room temperature in the dark with 5 l of each up: gate L for lymphocytes, M for monocytes, G for granulo- FITC (fluorescein isothiocyanate)-conjugated antibody and cytes and B for blast cells (Figure 1b). As the CD45 reagent 5 l of PE (phycoerythrin)-conjugated CD45 (Table 1). Diagnosis was labeled with phycoerythrin (CD45-PE) in each gate, the was confirmed by cytoplasmic markers for myeloperoxi- analysis of IgG1-FITC fluorescence intensity distribution was dase, CD3c and CD22c 4,5 or megakaryocytic and erythroid also performed. Fewer than 1% cells were found in the linear markers (CD61, CD41b and glycophorin-a; not shown). regions shown in each monoparametric histogram of Figures Samples were then treated using the Coulter ImmunoPrep 1c and 3. Histograms for a panel of antibodies used were anareagent kit and Multi-Q-Prep System (Coutronics, Margency, lyzed using the same settings (ie gate A in Figures 1a or 5a, France) according to the manufacturer s instructions. The c and gates L, M, G, B in Figures 1b or 5b, d). An example Multi-Q-Prep worked well only in samples diluted with in Figure 1c for the CD33 antigen shows the differential plasma as the admixture of PBS caused cell aggregation. In expression of the four cell sub-populations determined in the the first 21 patients, the FSC/SSC and CD45/SSC histograms CD45/SSC histogram. Other examples from the same sample were analyzed and in the other 53 patients all additional are given in Figure 3 for the CD3, CD19, CD14 and CD7 double IF tests were also performed. antigens. In some cases, the gates could be repositioned using backgating from the staining with the FITC-labeled lineagespecific reagents (eg CD3 for T lymphocytes, CD19 for B lymphocytes, Flow cytometry CD14 for monocytes, as shown in Figure 3). However, these adjustments were minor. In order to keep the setting and histograms uniform during the Blast cell populations were identified by using either gate studies, a Coulter XL flow cytometer was checked daily with S (Figure 1a) in the FSC/SSC histogram, or gate B in the calibration beads (DNA-Check; Coulter). With 488 nm excitation, CD45/SSC histogram (Figure 1b). CD antigen expression on FITC green fluorescence was measured through a the blast cell population was compared according to these 525 PB filter and orange PE fluorescence through a 575 PB two gating procedures, and more than 20% positive blasts filter. The four parameters simultaneously collected were lin- were recorded as positive. ear forward-angle light scatter (FSC), linear side scatter (SSC), log FITC and log PE. Fluorochrome compensation settings Results Table 1 Panel of monoclonal antibodies used for acute leukemia Clustering of bone marrow cell lineages immunophenotyping in this series Figure 1 shows the advantage of the CD45/SSC gating pro- CD number Reactivity Source cedure over FSC/SSC for identifying leukemic cells. When the FSC/SSC gate is used without an immunological marker, the CD3, FITC T lineage Ortho blast cells in AML partially overlap with the gates for normal CD4, FITC T lineage Coulter CD5, FITC T lineage Dako lymphocytes and monocytes (Figure 1a). As a contrast, CD7, FITC T lineage Coulter CD45/SSC gating clearly separates the four cell categories of CD8, PE T lineage Coulter lymphocytes, monocytes, granulocytes and blast cells (Figure CD10, FITC B lineage Coulter 1b). Mature lymphocytes show the highest CD45 fluorescence CD19, FITC B lineage Immunotech intensity and the lowest SSC signal (gate L). Mature monocytes CD20, FITC B lineage Coulter express slightly lower but still high amounts of CD45 and are CD22, FITC B lineage Dako HLA-DR, FITC Ortho easily distinguished from lymphocytes by their higher SSC sig- CD11b, FITC Granulocytes, monocytes, Ortho nal (gate M). Granulocytic lineage expresses low CD45 and NK cells very high SSC (gate G). In a few samples within gate G, two CD13, FITC Myeloid cells, monocytes Dako populations could be seen: the CD45 low population rep- CD14, FITC Monocytes Coulter resenting immature granulocytes and the CD45 medium popu- CD33, FITC Myeloid cells, monocytes Immunotech lation exhibiting mature features. 8 Blasts cells had the lowest CD34, FITC Immature precursors Immunotech CD45, PE Pan-leukocyte Coulter CD45 intensity and a low SSC, reflecting features of normal immature bone marrow precursor cells (gate B). Erythrocytes Ortho, Ortho Diagnostic System, Raritan, NJ, USA; Coulter, Coulexcluded. and platelets, which do not express CD45 antigen, are tronics, Margency, France; Dako, Glostrup, Denmark; Immunotech, Marseille, France. In 74 consecutive samples, the morphological bone marrow

3 1880 Figure 1 Representation of flow cytometry histograms of an acute myeloid leukemia. (a) Bivariate histogram FSC/SSC with two gates: gate A to eliminate debris and fat cells, gate S on presumed blast cells. Two gates (dotted lines) show the impossibility of defining a specific area for monocytes and blast cells. (b) Bivariate histogram CD45/SSC with four gates: gate G including granulocytes, gate M monocytes, gate L lymphocytes and gate B blast cells. Corresponding cell percentages are indicated (the sum is not 100% since rare cells are found outside the gates). (c) Univariate histograms of CD33 expression intensity conditioned on gate L (histogram 4 of lymphocytes), gate M (histogram 5 of monocytes), gate G (histogram 6 of granulocytes), gate B (histogram 7 of blast cells defined in (b), gate S (histogram 8 of blast cells defined in (a). Corresponding light microscopic counts are indicated. differential counts were compared with the results of the CD45/SSC gating procedure. A strong positive correlation was found for the four cell subpopulations (Figure 2b e): blast cells (r = 0.94), lymphocytes (r = 0.94), neutrophils (r = 0.90) and monocytes (r = 0.95). By contrast, a similar comparison between the morphological blast cell count and their corresponding count performed on the FSC/SSC histogram showed a greater spread of results (r = 0.76; Figure 2a). Moreover, the analysis of lysed whole bone marrow by the CD45/SSC gating showed an excellent correlation with the immunological identification of normal and malignant cells (Figure 3). The cells were first gated on gate A to exclude red cells, debris and fat cells (Figure 1a). Then it was confirmed that the cells classified as lymphocytes, monocytes, granulocytes and blast

4 1881 Figure 2 Correlations between a light microscopic bone marrow cell differential and the flow differential for hematopoietic cell evaluation. (a) FSC/SSC gating for blast cells (b e) CD45/SSC gating for (b) blast cells (c) lymphocytes (d) granulocytes (e) monocytes.

5 1882 Figure 3 Univariate histograms of some CD expression in lymphocytes, monocytes and blast cells from the same sample as in Figure 1. Histograms are all conditioned on gate A. From top to bottom, the distributions of fluorescence are shown for: CD3, CD19, CD14, CD7. Histograms (a d) are conditioned on gates L or M of Figure 1b. Histograms (e h) are conditioned on gate S of Figure 1a and histograms (i l) on gate B of Figure 1b. Arrows in histograms e, f and h show the contamination of lymphocytes in blast cell immunophenotyping. Arrow in histogram l shows the CD7 + blast cell population.

6 cells (Figure 1b) did indeed express the appropriate immunological markers (Figure 3). While monocytes were positive for CD33 (Figure 1c) and CD14 (Figure 3c), this particular case of AML expressed only CD33 (Figure 1c) and was negative for CD3 (Figure 3i), CD19 (Figure 3j) and CD14 (Figure 3k) if blast cells were determined on gate B of the CD45/SSC histogram. By contrast, the contamination of the blast cell population by T lymphocytes (Figure 3e) and B lymphocytes (Figure 3f) is obvious if gate S on the FSC/SSC histogram is chosen to isolate blast cells. The sensitivity of the method is documented with a study of CD7 antigen. T lymphocytes were CD7 +,as expected (Figure 3d). The blast cells had two populations: one CD7-negative and one CD7-positive (Figure 3l). The analysis of CD7 positivity based on a FCS/SSC gating alone would have led to the erroneous interpretation that CD7 positivity is due to lymphocyte contamination. Moreover, this example also illustrates the trivial error introduced by rigidly applying an artificial value of 20% threshold positivity, since in this case CD7 expression was present in low proportions of AML blast cells Determination of surface antigens of blast cell populations We used a double labeling method with CD45-PE against each CD-FITC antigen in the second part of the study where 53 consecutive samples were included in order to appreciate the gain of information in blast cell immunophenotyping. The positivity of CD-FITC antigen expression on blast cells was evaluated using FSC/SSC and CD45/SSC gating. Since in the first part of the analysis, CD45/SSC gating procedure showed a better determination of blast cells, we have chosen this protocol as the reference procedure. If the positivity threshold for each CD/FITC antigen expression was strictly fixed at 20%, false positive (FP) samples would have been recorded in Figure 4 Antigenic expression discrepancies between FSC/SSC and CD45/SSC gating. CD45/SSC gating was considered as the refer- ence procedure (see text). (a) Histograms of false positive/negative results given by the FSC/SSC procedure. A threshold value of 20% gave the antigen positivity. (b) For positive samples only (ie with samples which were positive (CD expression 20%) with the 20% positivity), histograms of CD expression discrepancies 15% FSC/SSC gating but negative (CD 20%) using CD45/SSC gat- between FSC/SSC and CD45/SSC gating. Numbers of positive cases are indicated for each CD. ing. By contrast, false negative (FN) samples were also seen negative using FSC/SSC gating but positive with CD45/SSC gating. In Figure 4a all FP cases were found for lymphoid (CD10, CD3, CD5, CD7) and monocytic (CD13, CD14, CD45/SSC gating. These findings are illustrated in Figure 1c CD11b) antigens. In every case the FP was due to lymphocytic where differences in blast cell CD33 positivity are seen or monocytic contamination of samples by peripheral blood. between CD45/SSC gating (60% positivity) and FSC/SSC gating On the other hand, FN cases were found only for myeloid (40% positivity). With CD45/SSC gating the counts of (CD33, CD13, CD11b), DR and CD34 antigens. Among these negative cells decreased (the upper arrow on Figure 1c) and samples, most leukemias with myeloid and/or monocytic the CD33 expression blasts were properly discriminated maturation have been cases of AML2 (n = 9), AML4 or AML5 (lower arrow). (n = 7). Thus, CD45/SSC gating excludes peripheral blood cell con- Even when FP and FN cases were absent, in many leukem- tamination from blast cell antigen analysis, especially in cases ias significant differences were noted in the degree of CD-FITC of heterogeneous bone marrow sub-populations. Moreover, in antigen expression when the CD45/SSC and FSC/SSC gating the present series, 27 samples (36%) contained fewer than procedures were compared. Thus, for each CD-FITC antigen 50% of blast cells and these would not have been properly expression, we accepted the value of antigen expression when immunophenotyped without the help of CD45/SSC gating. No the difference using the two gating procedures was below an correlation was found between morphological count and arbitrary value of 15% (Figure 4b). The discrepancies with FSC/SSC gating of the blast cells (r = 0.1), whereas an excellent 15% difference were mostly seen in the myeloid antigenic correlation remained when CD45/SSC gating was applied (r determination of blast cells. Frequently the intensity of CD- = 0.92). FITC antigen expression was lower using FSC/SSC gating than CD45/SSC gating for CD13 (38.5%), CD33 (26.3%) and CD34 (23.3%). This was mostly due to the contamination of bone Utility of CD45/SSC gating in frozen samples marrow samples by peripheral blood cells which were present within the FSC/SSC blast cell gate. On the other hand, in many When Ficoll gradient preparations are needed for storage of cases we also frequently found an increased expression of DR frozen samples for cytogenetics or molecular biology, (35%) and CD11b (40%) (Figure 4b) when studied with CD45/SSC gating remains a valuable and important step to

7 1884 Figure 5 Representation of flow cytometry histograms of the same case of AML2. (a and b) Bivariate histograms of the fresh sample with gates similar to Figure 1. (c and d) Bivariate histograms of the frozen sample after Ficoll preparation. Corresponding cell percentages are indicated. identify blast cells. An example of AML2 shown in Figure 5 of CD45 low blast cells in the fresh sample and the corresponding demonstrates the better separation of cell populations using frozen one are very similar, showing the reliability of the the CD45/SSC gating when compared to FSC/SSC gating. It CD45/SSC gating approach. It can also be noted that, in is clear that after freezing the conservation of CD45 antigen frozen samples, Multi-Q Prep can only be reliably applied expression is excellent. Table 2 indicates that the percentages after washings, and thus its methodological advantage over other methods is lost. Table 2 Percentages of CD + blast cells in the same case of AML2 (fresh and frozen samples) using the CD45/SSC gating Discussion CD expression Fresh sample Frozen sample In this study, we demonstrate the usefulness of a simple and reliable flow cytometry method to guide the classification of CD CD AML. In AML, peripheral blood or bone marrow blast cells CD are frequently admixed to other cell lineages and the blast CD cells themselves are also heterogeneous. To avoid some of CD these problems, Ficoll density gradient centrifugation is frequently used to enrich samples in leukemic cells. Neverthe-

8 less, these steps have three major disadvantages: (1) a differen- The introduction of CD45/SSC primary gating helps define tial count of the leukemic sub-populations becomes unreliable the prognostic values of other antigens without the danger of and diverges from the differentials seen on smears; counting normal populations within the leukemic gate. 15 (2) valuable information on the granulocytic cells is lost; and Consequently inter-laboratory variations are reduced and the (3) the selective elimination of granulocytes from the sample misdiagnosis of biphenotypic leukemias is avoided where this also prevents their use as internal controls for staining mature result was due to the presence of normal T and B cells in the myeloid cells. Thus, we prefer to proceed with whole bone leukemic gate. The increased sensitivity of the CD45/SSC gating marrow samples and to preserve all the nucleated cells in the inevitably leads to a suggestion that cut-off values (eg arbimarrow sample by performing an erythrocyte lysis with the Multi-Q- trary 20%) lose their significance as aberrant blast cell populations Prep method (Coulter). This technique, originally recommended can now be identified at lower detection limits ( 5%). for peripheral blood cell immunophenotyping, is In addition, the sensitivity of detecting minimal residual disease adaptable to bone marrow immunophenotyping. The results is improved. In this context is is essential to emphasize are reproducible and time is saved since no further washings that normal precursor cells are also CD45 low, and therefore or centrifugations are needed. Two limitations to the use of CD45/SSC per se does not discriminate between normal and these methods have, however, been noted: (1) immature malignant cells. 16,17 On the other hand, the CD45/SSC gate erythroid cells are damaged and cannot be accurately analyzed focuses attention on the strict normal vs malignant cell combe for glycophorin-a; and (2) debris and fat cells need to parison utilizing the other two channels of antibody labe- excluded by FSC/SSC pre-gating (gate A) in Figure 1a. ling. 18 These reagents may include CD7 and CD19 in ident- We confirm earlier reports suggesting that the blast cell ifying aberrantly expressing AML blasts while documenting population can be clearly identified by virtue of their low antigen that normal precursors are CD7, CD19. In these gated popu- membrane CD45 antigen expression and their light scatter lations the features of normal precursor cells guide, as internal properties. 6 9,12,13 Terstappen et al 12,13 have already emphasized controls, the quantitative characterization of aberrant marker that, when analyzed by their FSC and SSC light proper- expression on the residual leukemic blasts. ties, blast cells cannot be properly distinguished from lymphocytes This technique applied to AML in our series can be used in and monocytes. Other groups have recommended other acute leukemias such as acute lymphoblastic leukemia. 6 CD45/SSC gating to provide a better definition of all the cell The real ease of defining blast cells with CD45/SSC also led sub-populations of the samples in normal bone marrow analy- us to test this gating procedure in myelodysplasia, which often sis 7 9 as well as in acute leukemia. 6 CD45 was selected as the presents low percentages of blast cells. Preliminary results are basis for the gating procedure because it is found in different showing the reliability of this approach (data not shown). amounts in immature and mature hematopoietic cells. 7 In We are finding therefore that CD45/SSC gating is the most combination with CD45, SSC makes it possible to discrimi- reliable and easy way to perform acute leukemia immunophenotyping, nate all major lineages in separate populations in a single especially for samples with low proportions of bivariate analysis. Moreover, debris and mature erythrocytes blast cells. This question could well become a point for debate with low FSC signals are excluded from a CD45 gate. 7,8 And in consensus conferences. We suggest that CD45/SSC gating finally, the combination of CD45 expression and propidium should replace FSC/SSC gating, and that the cost of tests could iodide staining has also been found to be a good technique be decreased as the systematic use of this method is a practi- to evaluate peripheral blood contamination in the bone cal way to decrease the number of monoclonal antibodies marrow. 14 necessary to diagnose a leukemia, since the blast cell popu- In this work, we used PE-conjugated CD45 combined with lation is definitively isolated in most cases. This could contribute other FITC-conjugated antibodies in parallel samples to define to decreasing costs without affecting diagnostic quality. the exact phenotypic expression of AML blast cells. One of In conclusion, a clear discrimination of blast cells improves the main findings of this paper is that this technique gives an the measurement of their antigenic expression and their separ- excellent correlation with the light microscopic bone marrow ation from the accompanying hematopoietic cells: CD45/SSC differential counts in smears (r = 0.94 in Figure 2). Furthermore, gating seems to be the method of choice. A full understanding there are similarly high correlations between the per- of the leukemic process calls upon the global immunological centages of lymphocytes, granulocytes and monocytes and the appreciation of all the cells present in the leukemic sample corresponding populations defined by the CD45/SSC gates (r by using a lysed whole bone marrow analysis. Such was the = 0.94, 0.90 and 0.95, respectively, in Figure 2). We then case when the FAB classification was set up. 10 In the future, used this technique to compare the antigenic expression of the method described here might become a useful technique the blast cells determined both on FSC/SSC gating and for the immunophenotyping of acute leukemia at diagnosis CD45/SSC gating. and in treatment monitoring. In the majority of leukemias with intraclonal heterogeneity such as AML2, 4 and 5, the evaluation of blast cell populations would have been significantly interfered with by the FSC/SSC primary gating because the positively gated populations References would have included contaminating normal CD45 high 1 Foon KA, Todd RF. Immunological classification of leukemia and lymphocytes and monocytes (Figure 3 e l). Furthermore, as lymphoma. Blood 1986; 68: has also been pointed out by Rothe et al 5 and Borowitz et al, 6 2 Neame P, Soamboosrup P, Browman G, Meyer R, Benger A, Wilson W, Walker I, Seed N, McBride J. Classifying acute leukemia 3-colour IF analysis using CD45 as the third colour is even more discriminating than 2-color IF; the clone J-33 (labeled by immunophenotyping: a combined FAB immunologic classi- with cyanine-5y-5) performs well in this respect (Figure 5). We fication of AML. Blood 1986; 68: Drexler HG. Classification of acute myeloid leukemias a comtherefore conclude that the most informative reference histoparison of FAB and immunophenotyping. Leukemia 1987; 1: gram for blast cell identification is the CD45/SSC that can be optimally combined with two additional reagents in 3-colour 4 Bene MC, Castoldi G, Knapp W, Ludwig WD, Matutes E, Orfao IF for precise leukemia diagnosis. A, Van t Veer MB. European Group for the Immunological charac- 1885

9 1886 terisation of Leukemias. Proposals for the immunological classi- D, Büchner T, Hiddemann W, Wörmann B. Flow cytometric fication of acute leukemias. Leukemia 1995; 9: characterization of acute myeloid leukemia. Part I. Significance of 5 Rothe G, Schmitz G for the Working Group On Flow Cytometry light scattering properties. Leukemia 1991; 5: And Image Analysis. Consensus protocol for the flow cytometric 13 Tersteppen LWMM, Safford M, Könemann S, Loken MR, Zurlutter immunophenotyping of hematopoietic malignancies. Leukemia D, Büchner T, Hiddemann W, Wörmann B. Flow cytometric 1996; 10: characterization of acute myeloid leukemia. Part II. Phenotypic 6 Borowitz M, Guenther KL, Shults KE, Stelzer GT. Immunopheno- heterogeneity at diagnosis. Leukemia 1991; 5: typing of acute leukemia by flow cytometry analysis. Use of CD45 14 Abrahamsen JF, Lund-Johansen F, Laerum OD, Schem BC, Slet- and right-angle light scatter to gate on leukemic blasts in three- vold O, Smaaland R. Flow cytometric assessment of peripheral colour analysis. Am J Clin Pathol 1993; 100: blood contamination and proliferative activity of human bone 7 Shah VO, Civin CI, Loken MR. Flow cytometric analysis of human marrow cell populations. Cytometry 1995; 19: bone marrow: IV. Differential quantitative expression of T Dubosc-Marchenay N, Lacombe F, Dumain P, Marit G, Montas- common leukocyte antigen during normal hematopoiesis. J Immu- truc M, Belloc F, Reiffers J. Role of blast cell immunophenotyping nol 1988; 140: for the diagnosis and prognosis of acute myeloid leukemia. Hema- 8 Stelzer G, Shults K, Loken MR. CD45 gating for routine flow cyto- tol Oncol 1992; 10: metric analysis of human bone marrow specimen. Ann NY Acad 16 Syrjälä M, Anttila V-J, Ruutu T, Jansson S-E. Flow cytometric detec- Sci 1993; 677: tion of residual disease in acute leukemia by assaying blasts coexpressing 9 Rainer RO, Hodges L, Seltzer GT. CD45 gating correlates with myeloid and lymphatic antigens. Leukemia 1994; 8: bone marrow differential. Cytometry 1995; 22: Bennett JM, Catovsky C, Daniel MT, Flandrin G, Galton DAG, 17 Sutherland DR, Anderson L, Keeney M, Nayar R, Chin-Yee I. The Gralnick HR, Sultan C. Classification of acute myeloid leukemia. ISHAGE guidelines for CD34+ cell determination by flow cytometry. A report of the French American British Cooperative Group. Ann J Hematother 1996; 5: Intern Med 1985; 103: Macedo A, Orfao A, Gonzalez M, Vidriales MB, Lopez-Berges 11 Garand R, Béné MC and the GEIL. A new approach of acute lymphoblastic MC, Martinez A, San Miguel JF. Immunological detection of blast leukemia immunophenotypic classification. Leuk Lym- cell subpopulations in acute myeloblastic leukemia at diagnosis: phoma 1994; 13 (Suppl. 1): Terstappen LWMM, Könemann S, Safford M, Loken MR, Zurlutter implications for minimal residual disease studies. Leukemia 1995; 9: