Context.—Acute leukemia displays characteristic patterns of surface antigen expression (CD antigens), which facilitate their identification and proper classification and hence play an important role in instituting proper treatment plans. In addition to enzyme cytochemical analysis, multiparameter flow cytometric analysis has become commonplace in most laboratories for that purpose. The essential role and caveats of flow cytometry in that regard, however, have received little scrutiny.
Objective.—To evaluate the expression of commonly used immunomarkers and patterns in various acute leukemias to help define the best use and role of multiparameter flow cytometry in the diagnosis and proper classification of acute leukemias.
Design.—We have retrospectively analyzed the immunophenotypic data from 508 de novo adult and pediatric acute leukemia patients, as studied using multiparameter flow cytometry in addition to routine morphologic and enzyme cytochemical analysis. Cytogenetic and/or molecular data were correlated in all 41 cases of acute promyelocytic leukemia (APL) and in 203 other cases of acute leukemia where those data were available. We have also determined the positive and negative predictive values of a combined CD34 and HLA-DR expression pattern for the differentiation of APL from other myeloid leukemias.
Results.—In acute myeloid leukemia (AML) other than APL, expression of CD34 was seen in 62% and expression of HLA-DR in 86% of all cases. Twenty-six (10%) of 259 cases of non-APL AML were negative for both CD34 and HLA-DR as opposed to 33 (80%) of 41 cases of APL. None of the cases of APL were positive for both CD34 and HLA-DR in contrast to 149 (58%) of 259 cases of non-APL AML. Fifty-three cases were found to be examples of minimally differentiated AML (AML M0) based on the lack of expression of cytoplasmic CD3 and cytoplasmic CD79a and expression of one or more myelomonocytic-associated antigens and/or myeloperoxidase. Expression of CD20 was seen in 40 (24%) of 168 cases of precursor B-cell acute lymphoblastic leukemia (pB-ALL) and 52 (29%) lacked CD34 expression. Five of 180 cases of pB-ALL and 2 cases of precursor T-cell ALL (pT-ALL) were negative for terminal deoxynucleotidyl transferase (TdT). Aside from cytoplasmic CD3, CD5 and CD7 were the most sensitive antigens present in all 21 cases of pT-ALL. CD33 was more sensitive but less specific than CD13 for myeloid lineage.
Conclusion.—Aside from identification of blasts, flow cytometry was found to be especially useful in the correct identification of AML M0, differentiation of APL from AML M1/M2, and correct identification of TdT-negative ALL and unusual variants, such as transitional B-cell ALL and undifferentiated and biphenotypic acute leukemias.
The immunophenotypic analysis of acute leukemia by flow cytometry has become a powerful tool for proper identification of myeloid or lymphoid lineage and in this capacity has enjoyed widespread use in clinical laboratories. Although many acute leukemias can be correctly identified morphologically with or without enzyme cytochemical analysis, immunophenotyping remains indispensable for proper identification of myeloid lineage in minimally differentiated acute myeloid leukemia (AML M0) and determination of B-cell or T-cell lineage in acute lymphoblastic leukemia (ALL).1 Also, expression patterns of CD34 and HLA-DR help in distinguishing acute promyelocytic leukemia (APL/AML M3) from AML M1/M2.2 The immunophenotypic patterns of acute leukemias are well known; however, much less known is the diagnostic usefulness and specificity of immunophenotypic markers for various acute leukemia subtypes. Interpretation of complex data can be a daunting task not only for the individual with limited experience with flow cytometry but also for the expert. This study critically evaluates the diagnostic usefulness of flow cytometric immunophenotyping of acute leukemias and provides guidelines for the judicious use of immunomarkers and proper interpretation of results. The results are derived from more than 500 well-characterized cases of acute leukemias studied at a single institution using multiparameter flow cytometry.
MATERIAL AND METHODS
A total of 542 cases of acute leukemia were reviewed at the Lauren V. Ackerman Laboratory of Surgical Pathology at Washington University Medical Center (St Louis, Mo) between January 1, 1993, and August 31, 2001, using 2- and 4-color flow cytometric analysis. This included samples of peripheral blood and/or bone marrow from 508 de novo acute leukemia patients and 34 cases of blast crises of chronic myelogenous leukemia. All cases included in this study were classified according to French-American-British Cooperative Group (FAB) criteria or later using the World Health Organization guidelines at the time of initial review. The classification of acute leukemia into FAB M1, M2, M4, M5, and M6 types was based on morphologic and enzyme cytochemical analysis only (myeloperoxidase [MPO] and α-naphthyl butyrate esterase [NBE]). Cases of FAB M0, FAB M7, precursor B-cell ALL (pB-ALL), precursor T-cell ALL (pT-ALL), undifferentiated, biphenotypic, and other rare types were diagnosed using morphologic, enzyme cytochemical, and flow cytometric data. The diagnosis of APL was made using these modalities, but for this study only those cases were included that showed t(15;17) (or other variants) by routine cytogenetics and/or showed PML-RARα fusion by fluorescent in situ hybridization (FISH) or polymerase chain reaction. Cases of relapsed or recurrent acute leukemia were not included in this study. Also not included were cases reviewed as consulting cases. All previously published cases of AML M0, however, were included in this study.3 All except 7 cases, for which slides were unavailable, were verified on a second review by one of the authors (Z.K.). Four cases were misclassified initially as FAB M1/M2 based on morphologic and enzyme cytochemical analysis but were correctly classified as APL based on the demonstration of t(15;17) and/or positive FISH for PML-RARα and are included as APL in this study. For another 3 cases, the impression on the second review was in conflict from the initial diagnosis based on review of the original cytochemical analysis results, and those cases were excluded from the study. Cytogenetics and/or molecular data were available for most of the cases and correlated with the diagnosis where indicated. Immunohistochemical analysis was also performed on most cases classified as FAB M0 (as part of the study) and a few other cases, including examples of pB-ALL, M6, and M7.
All specimens were obtained and prepared for morphologic examination using standard techniques. The core biopsy specimens were fixed in 10% buffered formaldehyde, embedded in paraffin, and processed routinely, and the sections were stained with hematoxylin-eosin, Leder (chloroacetate esterase), iron (Prussian blue), and reticulin stains for light microscopy (most cases). Bone marrow aspirate smears and peripheral blood specimens (when available) were air dried and stained with Wright-Giemsa technique and examined under light microscopy.
Enzyme Cytochemical Analysis
For the detection of MPO, the air-dried bone marrow or peripheral blood smears were fixed for 1 minute at room temperature in 10% formalin and ethanol followed by washing with tap water for 15 to 30 seconds. The wet slides were placed for 30 seconds at room temperature in an incubator mixture containing ethyl alcohol (30%, 500 mL), benzidine dihydrochloride (1.5 g), zinc sulphate solution (5 mL), sodium acetate (5 g), hydrogen peroxide (3%, 3.5 mL), sodium hydroxide (1.0 N, 7.5 mL), and safranin O (1.0 g). The slides were then washed for 5 to 10 seconds in running tap water, counterstained in working Giemsa for 10 minutes followed by another wash with tap water, and air dried and mounted.
For the detection of nonspecific esterase using NBE, the air-dried bone marrow or peripheral smear slides were fixed in cold formalin and acetone fixative for 30 seconds followed by one washing with tap water and a second washing with distilled water. Slides were incubated for 45 minutes at room temperature in a mixture containing M/15 phosphate buffer, pH 6.0 to 6.3 (18.8 mL); hexazotized pararosanilin (0.2 mL); and NBE solution (1.0 mL). The slides were first washed with tap water and then with distilled water, counterstained with methyl green and Alcian blue for 6 minutes followed by a wash with tap water, and air dried and mounted.
Bone marrow aspirates or peripheral blood were immediately transported in sodium heparin tubes to the flow cytometry laboratory. Mononuclear cells were isolated using Ficoll-Hypaque and stained with various combinations of fluorescein isothiocyanate (FITC)–, phycoerythrin (PE)–, and phycoerythrin–cyanine 5 (PC5)–labeled monoclonal antibodies against the following antigens: CD1, CD2, CD3, CD4, CD5, CD7, CD8, CD10, CD13, CD14, CD19, CD20, CD33, CD34, CD41a, HLA-DR, nuclear terminal deoxynucleotidyl transferase (TdT), and immunoglobulin kappa (Igκ) and lambda (Igλ) light chains. In several cases, additional markers were used, including CD11c, CD15, CD16/CD56, CD61, CD64, CD117, cytoplasmic CD3 (cyCD3), cytoplasmic CD79a (cyCD79a), MPO, glycophorin A, T-cell receptor α-β, and T-cell receptor γ-δ proteins. The typical combinations included CD34-FITC/CD33-PE, CD34-FITC/CD13-PC5, CD33-PE/CD13-PC5, CD34-FITC/CD117-PE, CD33-PE/CD56-PC5, CD14-FITC/CD64-PE, Kappa-PE/CD19-PC5, Lambda-FITC/CD19-PC5, CD10-PE/CD19-FITC, CD34-Pool-PE/CD19-FITC, CD7-FITC/CD1-PE, CD4-FITC/CD8-PE, CD3-PC5/CD4-FITC, CD3-PC5/CD8-PE, CD19-FITC/HLA-DR-PE, CD19-FITC/CD5-PC5, nuclear TdT-FITC/cyCD79a-PE, nuclear TdT-FITC/cyCD3-PC5, MPO-FITC, CD15-PC5, CD61-FITC, CD20-PC5, CD41-FITC, and GlyA-PE. Custom combinations of antibodies were also used when needed.
For the detection of cytoplasmic (MPO, cyCD3, and cyCD79a) and nuclear TdT antigens, surface staining of CD45 for gating purposes (for 4-color analysis only) and for any other surface antigens was performed before processing the cells for cytoplasmic and nuclear antigens. The cells were then fixed with Reagen 1 from the Beckman-Coulter-Immunotech (Miami, Fla) Intraprep Permeabilization reagent kit. After 15 minutes of fixation, the cells were washed and permeabilized for 5 minutes with Reagent 2 from the kit. After a 5-minute permeabilization, appropriate antibodies were added and incubation time was adjusted for the antibody that required the longest reaction time, 20 minutes for cytoplasmic antigens and 1 hour for nuclear antigens. The cells then were washed with Hanks balanced salt solution and fixed in the usual manner with 1% methanol-free formaldehyde before flow cytometric analysis.
Two- and 4-color flow cytometric immunophenotyping was performed on FACScan (2-color only, Becton-Dickenson, San Jose, Calif) or on the Coulter XL cytometer (2- and 4-color, Coulter, Miami, Fla) by collecting 10 000 ungated list mode events, selecting an appropriate blast gate on the combination of forward and side scatter, and analyzing cells with the most appropriate blast gate. Many of the cases were gated on CD45dim versus side scatter to isolate the blast population, and 10 000 list mode events were collected on the blast gate. An antigen was considered positively expressed when at least 20% of the gated cells expressed that antigen; however, the arbitrary cutoff limit for a positive TdT expression was 10% or more.
Immunohistochemical analysis was also performed on formalin-fixed, paraffin-embedded bone marrow biopsy specimens on several cases as deemed necessary. The antigens that were sought included cyCD3, cyCD79a, MPO, and others where indicated (CD20 [L-26], CD43 [MT-1], CD45RO [OPD-4], TdT, glycophorin A, and factor VIII). Five-micrometer-thick sections were cut and collected on lysine-coated slides and dried in a 60°C paraffin oven for 45 minutes. Sections were deparaffinized in xylene, incubated for 30 minutes in methanolic hydrogen peroxide (0.3% [vol/vol]) to quench for endogenous peroxidase, and rehydrated in graded ethanol solution followed by rinsing in distilled water and phosphate-buffered saline (pH 7.4). Heat-mediated antigen retrieval (epitope retrieval) was performed for 12 minutes in a microwave oven in the presence of citrate buffer (pH 6.0). The sections were cooled for 20 minutes followed by rinsing in water and incubation in phosphate-buffered saline. A protein block (Dako Corporation, Carpinteria, Calif) was performed for polyclonal antibodies with a 5-minute incubation period. Primary antibodies against MPO (polyclonal, 1:4000 dilution, Dako), CD3 (polyclonal, 1:40 dilution, Dako), and CD79a (monoclonal, 1:100 dilution, Dako) were applied, and the sections were incubated for 18 hours at 4°C. Antibody bridge assembly by the avidin-biotin-peroxidase complex (ABC) method using the Elite ABC kit (Vector Laboratories, Burlingame, Calif) was performed the next day by 2 sequential 1-hour incubations. Chromogenic development was accomplished by immersion of the sections in 3,3′-diaminobenzidine solution (0.25 mg/mL with 0.003% hydrogen peroxide). The slides were immersed in 0.125% osmium tetroxide to enhance chromogenic precipitation, followed by light counterstaining with Harris hematoxylin. The sections were dehydrated in graded ethanol, cleared in xylene, and mounted with Cytoseal medium (Electron Microscopy Sciences, Fort Washington, Pa).
For routine karyotyping, the bone marrow and peripheral blood specimens were transported in RPMI 1640 culture medium with 15% fetal calf serum and were cultured for 24 hours at 37°C. Cells were exposed to Colcemid (0.05 mg/μL) for 2.5 hours at 4°C and harvested routinely. Routine slide preparation and Giemsa banding at 450-band resolution were performed.
For FISH analysis, bone marrow cells or peripheral blood lymphocytes were cultured unstimulated overnight according to standard methods. Cells were examined by FISH using commercially available probes for PML/RARα and BCR/ABL rearrangements (Vysis, Downers Grove, Ill). Slides were dehydrated in a series of 70%, 85%, and 100% ethanol. Ten-microliter probes were added to each slide and covered with a 22 × 22-mm coverslip. Probe and chromosomal DNA was codenatured at 72°C for 2 minutes then hybridized overnight at 37°C in a Hybrite hybridization chamber (Vysis). Slides were washed for 2 minutes with 0.4× SSC and 0.3% NP40 at 72°C; the hybridization was then stopped by immersing for 30 seconds in 2× SSC at room temperature. Slides were air dried and counterstained with 10 μL of DAPI before scoring on an Olympus BX60 fluorescent scope. Images were captured with the Smart Capture image system (Applied Imaging, Santa Clara, Calif). A minimum of 100 cells were scored on each assay.
Statistical analysis was performed to determine the usefulness of the combined expression pattern of CD34 and HLA-DR for the differentiation of APL from AML M1/M2 only because other classes of AMLs can effectively be differentiated from APL based on the results of MPO and NBE reactivity. The positive predictive value (PPV) and negative predictive value (NPV) of the combined use of CD34 and HLA-DR for the diagnosis of APL were deduced based on our results using the expression data from APL and AML M1/M2 only. The following simple formulae were used:
The flow cytometric immunophenotypic data correlated with the pattern of MPO and NBE expression by cytochemical analysis in cases classified as FAB M0, M1, M2, M3, M4, and M5. Two cases were classified as undifferentiated, which expressed only CD34 and HLA-DR, but no other surface markers were examined (CD13, CD33, CD14, CD15, CD41, CD56, CD61, CD64, CD117, MPO, CD1, CD2, CD3, cyCD3, CD4, CD5, CD7, CD8, CD10, CD19, CD20, cyCD79a, and Igκ and Igλ light chains). Both cases, however, showed expression of TdT. One of these cases had 46,XX karyotype, whereas the other one showed 47,XY,+13. The diagnosis of biphenotypic acute leukemia was reserved for cases that showed definitive dual-lineage differentiation as previously described.3 Only one case fulfilled the criteria and showed expression of both MPO and cyCD3 in addition to CD34, HLA-DR, TdT, CD117, CD2, and CD7. No other antigens examined were expressed. Cases of ALL that expressed CD13 and/or CD33 without MPO were classified as examples of myeloid antigen–positive ALL rather than as true biphenotypic leukemia. Similarly, cases of MPO-positive or NBE-positive AML that expressed lymphoid-associated (CD7, CD2, CD4, CD19, CD10) but not lymphoid-specific (cyCD3, cyCD79a) antigens were best regarded as examples of lymphoid antigen–positive AML. The true biphenotypic acute leukemias must show dual differentiation using lineage-specific antigens. One case was classified as transitional B-ALL, which showed clonal expression of surface Igκ light chain (κ:λ = 86:1), HLA-DR, CD10, CD19, TdT (but not CD34 or CD20), and FAB L1 morphologic findings. This case had the following karyotype: 46,XY,t(3:6)(q12;q22),add(9)(p12),der(19),t(1;19)(q23;p13)/46,XY. Because of their rarity, information on transitional B-ALL is limited. The surface expression of IgM or light chains does not appear to alter their response to chemotherapy since the patient described in this report responded to chemotherapy for pB-ALL. Two cases were found to be examples of TdT-positive Burkitt leukemia (FAB L3). We consider these cases distinct from transitional B-ALL, which do not show involvement of c-myc. These 2 cases are described in detail in another report because of their diagnostic and therapeutic importance.4
Thirty-four cases of blast crises of chronic myelogenous leukemia (CML; 21 myeloid blast crises and 13 lymphoid blast crises) were assigned FAB groups. This included FAB M0 (2), M1 (5), M2 (12), and M7 (2). Eleven cases were classified as pB-ALL and 2 as pT-ALL. The immunophenotype of these “transformed” acute leukemias was similar to de novo acute leukemia with respect to CD34, HLA-DR, CD13, and CD33 expression, and hence no further comments will be made. For the purpose of this report, cases of FAB M1 and M2 were combined because in several cases the morphologic distinction is arbitrary such that in some cases the peripheral blood showed less than 90% blasts, whereas the bone marrow contained more than 90% blasts and vice versa. It is practical to combine M1 and M2 because no prognostic differences exist based on morphologic differentiation alone. The prognosis in such cases depends mainly, among other factors, on the results of cytogenetic studies, age, white blood cell count, and the presence or absence of myelodysplasia. The detailed antigen expression for each class of acute leukemia is provided in Table 1, and only salient features will be mentioned in the following sections.
HLA-DR, CD34, and CD117 Expression
Expression of HLA-DR was seen in 222 (86%) of 259 patients with non-APL AML, whereas it was seen, although only dimly and partially, in only 1 (microgranular variant [M3v]) of 41 APLs. Not as rare as HLA-DR, however, expression of CD34 was seen in 7 (17%) (all microgranular variants [M3v]) of 41 cases of APL and 161 (62%) of the non-APL AMLs. The expression of CD34 in APL was found to be more heterogeneous than in non-APL AMLs, a feature that may further aid in the distinction of APL from other AMLs. The lower frequency of CD34 than HLA-DR expression in non-APL AMLs was mainly due to lack of CD34 expression in leukemic cells of monocytic origin (AML M5 [4/34] and AML M4 [11/30]). The combined use of CD34 and HLA-DR is much more helpful in distinguishing APL from non-APL AMLs than either of these antigens alone (Table 2). None of the 41 cases of APL were positive for both CD34 and HLA-DR in contrast to 149 (58%) of 259 non-APL AMLs. However, lack of both CD34 and HLA-DR expression does not always suggest a diagnosis of APL, since 26 (10%) of our 259 non-APL AMLs also did not express either of these antigens. The lack of both HLA-DR and CD34 expression in an MPO-positive AML must cause a search for t(15;17) and/or PML-RARα or their variants, because we have seen cases with morphologic findings that are more compatible with AML M1/M2 than APL. In pB-ALL, expression of HLA-DR was seen in almost all cases (176/180 [98%]), whereas none of the 21 cases of pT-ALL showed HLA-DR expression. In contrast to HLA-DR, expression of CD34 was seen in only 128 (71%) of 180 pB-ALL cases and in only 1 of 21 cases of pT-ALL.
Expression of CD117, although analyzed in fewer cases than CD34 and HLA-DR, did not prove to be of diagnostic significance in the differentiation of APL from AML M1/M2. Its expression was seen in 8 (80%) of 10 cases of APL and 35 (80%) of 44 cases of AML M1/M2. None of the cases of ALL (pB-ALL, 0/14; pT-ALL, 0/2) showed expression of CD117.
Myelomonocytic Antigens (CD13, CD14, CD15, CD33, and CD64)
The expression of CD13 and CD33 was seen, respectively, in 229 (76%) and 266 (88%) of 301 AML cases of all subtypes. CD33 expression was much more sensitive for myelomonocytic origin of cells than CD13 present in all 41, 30, and 34 cases of APL, AML M4, and AML M5, respectively. In pB-ALL, CD13 and CD33 expression was seen in 16 (9%) and 18 (9%) of 180 cases, respectively, but only 5 cases (3%) showed dual expression. Only 1 of 21 cases of pT-ALL showed CD33 expression and none expressed CD13.
Expression of CD14 is usually considered a marker of monocytic origin, thus seen in 34 (53%) of 64 cases of AML M4 and M5. Its expression, however, was also seen in 10 (7%) of 115 AML M1/M2 cases but not in APL. Only 12 (15%) of 78 cases of AMLs showed CD15 expression, including M1/M2 (2/40), M3 (3/10), M4 (3/4), M5 (2/6), and M7 (1/4). None of the 3 cases of CD15+ APL showed CD34 expression, a feature that is of diagnostic importance since expression of both is not a feature of APL as noted previously.5 CD64, which reacts usually with cells of monocytic origin, was seen in 7 (58%) of 12 AML M4/M5 cases but also in 7 (17%) of 40 AML M1/M2 cases. Similar to CD14, however, no expression of CD64 was seen in APL.
Flow cytometric evaluation of MPO was performed in a total of 71 cases. Seventeen of 24 cases of AML M0 showed MPO expression in more than 3% of blasts. All except 1 of 22 cases of AML M1/M2 where MPO was analyzed showed positive expression in more than 10% of the gated cells. The one case with no expression of MPO by flow cytometry showed positive expression by routine cytochemical analysis for MPO. The lack of MPO expression was considered a technical problem. The expression pattern in other leukemias is shown in Table 1.
Megakaryocyte and NK Cell–Associated Antigens (CD41, CD61, and CD56)
The expression of megakaryocyte-associated antigens, CD41 and CD61, although seen in all cases of AML M7 tested (CD41, 10/10; CD61, 6/6), was also seen in some cases of AML M0, M1/M2, M4, and M5; the significance of such an expression is unclear, but nonspecific binding of platelets through Fc receptors appears to be the most likely explanation (Figure 1). However, aberrant expression of these platelet-associated antigens may be an alternative explanation, since in some cases expression of only one and not both of these antigens was seen. Expression of CD56 was seen in 20 (25%) of 81 cases of AMLs tested and more commonly in monocytic lineage (AML M4 and M5) than myeloid lineage (AML M0, M1, M2, and M3) (58% vs 17%).
Lymphoid-Associated Antigens and TdT
All cases of pB-ALL showed expression of pan B-cell markers CD19 (180/180) and cyCD79a (25/25), but only 161 (89%) of 180 and 40 (24%) of 168 expressed CD10 and CD20, respectively. Interestingly 4 (25%) of 21 cases of pT-ALL also showed expression of CD10, but none expressed CD19 or CD20. Besides cyCD3, only CD5 and CD7 were present on all pT-ALL. CD2 was lacking in 3 of 21 cases. Six of 21 cases of pT-ALL showed expression of both CD4 and CD8, whereas 8 of 21 cases expressed neither antigen. Only one case had expression of CD4 without CD8 in contrast to 6 cases that showed expression of CD8 only. Except for CD2 (2/154), none of the T-cell–associated antigens was expressed on any pB-ALL.
The expression of lymphoid antigens on myeloid leukemias was more promiscuous than expression of myeloid antigens on lymphoid leukemias. The B-cell–associated antigen CD19 was seen in 4 of 52 AML M0 cases and 3 of 123 AML M1/M2 cases. None of the myeloid leukemias tested were positive for CD20. Among T-cell–associated antigens, aside from CD3 (including cyCD3), CD1, CD5, and CD8 were found exclusively expressed on pT-ALL and none on any myeloid leukemia or pB-ALL. The most promiscuous of all T-cell–associated antigens was CD7, which was seen in 23 (44%) of 52 AML M0 cases, 31 (25%) of 123 AML M1/M2 cases, 2 (6%) of 35 APL cases, 5 (25%) of 20 AML M5 cases, 1 (12%) of 8 AML M6 cases, and 4 (50%) of 8 AML M7 cases. The next most commonly expressed T-cell–associated antigen on AMLs was CD2, which was seen in 30 (19%) of the 154 myeloid leukemias tested for this antigen followed by CD4 seen in 22 (14%) of the 154 cases. Expression of TdT was seen in all except one case of pT-ALL (95%) and 175 (97%) of 180 pB-ALL cases. Thus, lack of TdT expression is unusual but should not preclude a diagnosis of ALL in otherwise typical cases (Figure 2). Expression of TdT in myeloid leukemias was seen only in AML M0 and AML M1/M2. Except for APL, other FAB classes may express TdT, however.6
With contemporary treatment plans, an acute leukemia must be distinguished as being of lymphoid or myeloid origin; a lymphoid leukemia is further identified as precursor B-cell or T-cell origin, and for an AML, a distinction is made between an APL and all other FAB subtypes. In most clinical situations, this can be done using morphologic and enzyme cytochemical analysis alone. There are, undoubtedly, some cases that defy correct identification based on these simple techniques and require immunophenotyping for proper characterization. These are the situations where multiparametric flow cytometry plays an indispensable role. Despite its widespread use in clinical laboratories, however, the diagnostic usefulness and limitations of flow cytometric immunophenotyping have received little scrutiny. We have analyzed the usefulness of a variety of markers for various acute leukemia categories to facilitate a correct diagnosis and define the critical role of flow cytometry.
One of the best of all roles of flow cytometry is in the identification of myeloid lineage and exclusion of lymphoid lineage in minimally differentiated acute leukemias. Not only does AML M0 show minimal myeloid differentiation, but it often shows an FAB L1/L2 morphologic structure and expresses TdT and various lymphoid-associated antigens that may lead to an erroneous diagnosis of myeloid antigen–positive ALL.7 Although a diagnosis of AML M0 can be made using immunohistochemical analysis for MPO, cyCD3, and cyCD79a, flow cytometry confers a much better evaluation of blast cells since multiparametric analysis is uniquely offered by flow cytometry. Aside from cyCD3, CD5 is the best marker to distinguish AML M0 from pT-ALL because its expression was seen in all (21/21) T-ALL but in none of the 39 cases of AML M0 tested. Although none of our cases of AML M0 expressed CD1, its expression was also seen in some but not all cases of pT-ALL. The expression of MPO is not seen in all cases of AML M0, so a diagnosis must rest on other criteria as described previously.3 We have not seen expression of cyCD3 in acute leukemias other than pT-ALL, but we would like to stress that dim but not bright expression of cyCD79a has been observed in 2 of our T-cell lymphoblastic lymphomas (not included in this study) and in 1 AML M1 case. The expression of CD2, CD4, CD7, CD10, and CD19 cannot be taken as evidence of lymphoid differentiation because we have seen, not infrequently, expression of these antigens (T-cell associated more often than B-cell associated) in AML M0. It is the lack of both cyCD3 and cyCD79a (bright) that dictates the diagnosis of AML M0 more often than expression of MPO.
We have found HLA-DR to be the single best marker for distinguishing APL from other AMLs because its expression was seen in only 1 of 41 cases of APL. The specificity of this distinction is further enhanced if expression pattern of CD34 is also taken into account. Thus, the expression of both CD34 and HLA-DR in an MPO-only expressing AML (M1/M2) can effectively exclude a diagnosis of APL in the correct clinical context. The lack of expression of both CD34 and HLA-DR, however, can be seen in a small number (10%) of AML M1/M2 cases. The expression pattern of CD33 additionally contributes to the distinction of APL from M1/M2 because none of our 41 cases of APL lacked this antigen in contrast to approximately 10% of cases of M1/M2 that lacked this antigen. We suggest that a profile of CD34+, HLA-DR+, and CD33− is incompatible with a diagnosis of APL based on our data. The only other antigen of interest in this context is CD14, whose expression was not seen in any of the cases of APL but seen in approximately 9% of cases of M1/M2. In contrast to CD34, expression of CD117 was not found useful in this distinction because its expression was seen in all FAB subtypes of AMLs.
The diagnosis of AML M4, M5, and M6 can be made without flow cytometry using routine enzyme cytochemical analysis. The diagnosis of AML M7 can be a challenge given the fact that most of these cases have extensive bone marrow fibrosis that causes pancytopenia with only a few circulating blasts.8 The diagnostic difficulty can be compounded by the fact that in several cases the megakaryoblasts are small, resembling lymphoblasts or even lymphocytes. In such cases, flow cytometric immunophenotyping can identify circulating rare “atypical” cells as megakaryoblasts expressing CD41 (gpIIb) and/or CD61 (gpIIIa) but not MPO. A diagnosis of AML M7 should not shy away from this just because the peripheral blood does not meet the 20% blast criterion since that is often the case with AML M7. We have seen expression of CD41 and CD61 in otherwise typical cases of AML M1/M2 that do not qualify as AML M7, but such a distinction is often of academic rather than practical interest.
Overall, CD33 is a much more sensitive marker than CD13 for myeloid lineage, but by the same virtue, it is less specific than CD13. Not only in AML M7 but also in lymphoid leukemias, it is more common than CD13. Expression of CD64 is touted as a marker of monocytic lineage, although we have seen its expression in MPO-expressing AML M1/M2. As for CD14, in contrast to other reports, we have seen it in slightly less than 50% of our cases of AML M4/M5. It is, however, a specific marker for monocytic lineage, and more importantly we have not seen its expression in APL (0/35). This feature can be helpful, since a distinction between AML M5 and AML M3v can sometimes be difficult on morphologic examination alone, especially in cases where enzyme cytochemical analysis technically yields poor results.
The lack of TdT is unusual but can be seen in less than 5% of both pB-ALL and pT-ALL. In unusual cases of CD10+ follicular lymphomas involving peripheral blood, which do not express surface immunoglobulin light chains, a distinction from CALLA-positive pB-ALL may become difficult. CD34 may not be expressed in pB-ALL, and CD20 expression may be seen in an otherwise typical pB-ALL. Although the probability of such a scenario is low, the possibility remains. The features that may help differentiate the 2 entities include usually dim-to-moderate CD45 expression by lymphoblasts and usually bright CD45 expression by mature lymphoma cells and, if present, expression of CD13 and/or CD33 by lymphoblasts. The lack of TdT in pT-ALL can pose a serious diagnostic challenge. Peripheral T-cell lymphoma or leukemia can express or lack both CD4 and CD8 similar to pT-ALL.9 Furthermore, CD34 is also not expressed by almost all pT-ALL. The pT-ALL, however, uniformly expresses CD5 and CD7 (in our study), both of which may be partially or totally lacking from peripheral T-cell lymphoma or leukemia.9 In addition, expression of CD1, if present, confirms a thymic phenotype and hence is diagnostic of pT-ALL. Surface expression of CD3 can be seen in pT-ALL and is thus of limited value in distinguishing TdT-negative T-ALL from peripheral T-cell lymphoma or leukemia. Less commonly considered but equally important in this distinction is HLA-DR, whose expression is rare in T-ALL but common in peripheral T-cell lymphoma or leukemia.
Unusual expression of an antigen may be observed infrequently but must be interpreted in the context of the entire immunophenotypic profile, since we have seen expression of CD2 in pB-ALL and expression of dim cyCD79a in AML M1/M2. The true biphenotypic or undifferentiated nature of some leukemia is almost always determined by multiparameter flow cytometry, thus effectively excluding a diagnosis of a lymphoid leukemia. In summary, in addition to its routine role in the identification and enumeration of blasts in the clinical specimen, flow cytometric immunophenotyping can be uniquely useful in the diagnosis of AML M0, differentiation of APL from AML M1/M2, correct identification of TdT-negative pT-ALL, and the diagnosis of undifferentiated and biphenotypic leukemias.
Reprints: Zahid Kaleem, MD, Department of Pathology, Creighton University School of Medicine, 601 N 30th St, Omaha, NE 68131