Leukemia (1998) 12, 1056–1063  1998 Stockton Press All rights reserved 0887-6924/98 $12.00 http://www.stockton-press.co.uk/leu Prognostic relevance of the expression of Tdt and CD7 in 335 cases of acute myeloid leukemia A Venditti1, G Del Poeta1, F Buccisano1, A Tamburini1, MC Cox-Froncillo1, G Aronica1, A Bruno1, B Del Moro1, AM Epiceno1, A Battaglia2, L Forte1, M Postorino1, V Cordero1, S Santinelli1 and S Amadori1 ` 1Cattedra di Ematologia, Universita ‘Tor Vergata’, Divisione di Ematologia, Ospedale S Eugenio, Rome; and 2Cattedra di Ginecologia e ` Ostetricia, Policlinico A Gemelli, Universita Cattolica Sacro Cuore, Rome, Italy

We have analyzed the expression of Tdt and CD7 in 335 cases of LyAg in 335 unequivocal cases of AML. We have found of unequivocal acute myeloid leukemia (AML). Tdt was that Tdt+ and/or CD7+ AML represents a distinct entity with expressed in 80 (25%) of 321 evaluable cases. Twenty-six of 77 (34%) Tdt+ patients assessable for response, entered complete an unfavorable outcome, possibly due to the coordinate remission (CR) vs 121 of 209 (58%) Tdt− cases (P Ͻ 0.001). CD7 occurrence of GP-170 and poor prognostic chromosomal was expressed in 102 of 332 (30%) evaluable cases; 37 of 93 abnormalities. assessable (40%) CD7+ patients attained a CR as compared to 114/204 (56%) CD7− (P = 0.013). Duration of survival was sig- nificantly shorter for patients with CD7+ or Tdt+ AML (P = 0.006 and 0.001, respectively). In a multivariate analysis, Tdt was found to significantly adverse achievement of CR (P = 0.018), while CD7 affected duration of CR (P = 0.037). Overall the Patients and methods expression of either Tdt or CD7 correlated with a relatively high expression of CD34 (P Ͻ 0.001), GP-170 (P = 0.003), lymphoid antigens (LyAg) (P Ͻ 0.001), t(9;22) or anomalies of chromo- Patients and treatments some 5/7 (P Ͻ 0.001). Finally, we pooled the patients into four phenotypic classes, according to the presence of Tdt, CD7 or − − + − − + + + both: [Tdt CD7 ], [Tdt CD7 ], [Tdt CD7 ] and [Tdt CD7 ]. The 31 category [Tdt+CD7+] was characterized by a more unfavorable The 335 consecutive patients with newly diagnosed AML outcome as suggested by a lower rate of CR (P Ͻ 0.001) and a included in this study were admitted to our institution shorter duration of survival as compared to cases [Tdt−CD7−], between June 1987 and May 1997. No patients had a history [Tdt+CD7−] and [Tdt−CD7+](P = 0.002). This figure is consistent + + of prior therapies with anticancer drugs or a diagnosis of myel- with the frequent convergence in the subset [Tdt CD7 ] of GP- odysplastic syndrome. Antileukemic treatment was differen- 170 positivity (P = 0.003), translocation t(9;22), anomalies of Ͻ tiated according to age. Among patients aged over 60 years, 5 and/or 7 (P 0.001) and signs of lineage infi- 2 delity (deviant expression of lymphoid antigens) (P Ͻ 0.001). 98 were given a combination of Ara-C (1 g/m /day) for 6 days We conclude that the expression of Tdt or CD7 is associated and mitoxantrone (6 mg/m2/day) for 6 days as remission with an unfavorable outcome and that the combination of both induction therapy and 20 received Ara-C (100 mg/m2/day) for defines a clinical subset with a poorer prognosis due to the 7 days, etoposide (100 mg/m2/day) for 3 days and mitoxan- significantly higher association with MDR phenotype, and ‘poor trone (7 mg/m2/day) for 3 days, according to the prognostic’ chromosomal abnormalities. Keywords: AML; CD7; Tdt; GP-170; karyotype EORTC/GIMEMA protocol AML-13. Among patients aged less than 60 years, 122 received Ara-C (200 mg/m2/day) for 7 days combined with daunorubicin (45 mg/m2/day), while 61 were treated with Ara-C (100 mg/m2/day) for 7 days, etoposide Introduction (100 mg/m2/day) for 5 days and daunorubicin (50 mg/m2/day), either idarubicin (10 mg/m2/day) or mitoxantrone Immunophenotyping represents a widely used method to (7 mg/m2/day) for 3 days according to the Amgen protocol G- diagnose and classify acute leukemias complementing mor- CSF 91134 or EORTC/GIMEMA protocol AML-10. Patients phology and cytochemistry.1 Evaluation of surface and intra- with promyelocytic leukemia (APL), regardless of age, cellular antigens allows identification of lineage commitment received an induction course either with idarubicin and the stage of maturation of the cell involved in the leu- (10 mg/m2/day) for 6 days alone or associated with Ara-C kemic process.2,3 The systematic use of monoclonal anti- (n = 24), or idarubicin 12 mg/m2/day for four doses plus all- bodies (moAbs) has also allowed detection of inappropriate trans retinoic acid 45 mg/m2/day (n = 10).32 All patients expression of antigens from more than one lineage, thus rais- achieving CR were given an anthracyline-based post-induc- ing the possibility of ‘lineage infidelity’ or ‘lineage promis- tion regimen, even including cytosine-arabinoside and etopo- cuity’.4–19 Recognition of lymphoid antigen (LyAg)-positive side. CR was defined by a bone marrow (BM) with less than AML is potentially useful for immunological detection of mini- 5% blasts and restored normal hematopoiesis. Remission fail- mal residual disease,20–24 but the prevalence and prognostic ures were divided into three categories as previously importance of LyAg expression in AML remain contro- described.33 (1) early death, ie patients who died of infection versial.5,10,11,13–15,17,25–30 In a previous study of 154 patients, or hemorrhage within 7 days after the end of chemotherapy; we found that certain phenotypic subgroups had a worse (2) death in hypoplasia; and (3) resistant disease, ie patients prognosis based on the expression of CD7.13 In the present who survived for 13 days or more and whose leukemic cells study, we extend these findings and report on the expression were continuously present in their BM, or whose leukemic cells were cleared from their BM during or after therapy, but ` reappeared before the resumption of normal hematopoiesis. Correspondence: A Venditti, Cattedra di Ematologia, Universita ‘Tor Vergata’, Divisione di Ematologia, Ospedale S Eugenio, Piazzale Relapse was defined as reappearance of blasts in the blood or Umanesimo 10, 00144 Rome, Italy; Fax: 39 6 5915965 the finding of more than 5% blasts in the BM, not attributable Received 11 August 1997; accepted 20 March 1998 to other causes (eg BM regeneration). Prognostic relevance of the expression of Tdt and CD7 in AML A Venditti et al 1057 Diagnostic criteria cells in 100 ␮l volume. Optimal Moab concentration was pre- viously established in a series of titration experiments. Isotype- The 335 patients (186 men and 149 women) enrolled in this matched Ig was used as a negative control. After 10 min incu- study were diagnosed to have de novo AML. Diagnosis of bation at room temperature the cells were washed three times AML was confirmed by MPO expression and lack of cCD3 in PBS with 0.2% azide and 0.2% bovine serum albumin and cCD22. Other biologic and clinical features are summar- (PBSA) and analyzed with a flow cytometer. Either an Epics ized in Table 1. Profile (Coulter), or a Facsorter (Becton Dickinson, Mountain View, CA, USA) were used for flow cytometry. A minimum of 10 000 events were acquired in each sample and, based on Immunophenotyping forward-/orthogonal light scatter properties of the leu- kemic population, a blast gate was used to determine the per- Bone marrow and peripheral blood samples for this study centage of positive cells. A positive reaction was considered were obtained under local institutional review board proto- when at least 20% of gated cells were more fluorescent than cols. Surface antigens were detected by double/triple-color the negative control. The expression of lymphoid antigens was immunofluorescence assay combining phycoerthrin (PE), investigated in double/triple staining assays to exclude lym- fluorescein isothyocyanate (FITC) and peridin chlorophyll pro- phocyte contamination. For CD34-positive cases the double tein (PerCP) conjugated monoclonal (Moabs). staining was performed combining CD34 plus the lymphoid These included: CD33 PE, CD13 PE, CD14 PE, CD19 PE, antigen which tested positive. The availability since 1991 of CD20 PE purchased from Coulter (Hialeah, FL, USA); CD33 CD34 and other PerCP-conjugated Moabs has allowed us to PE, CD13 PE, CD14 PE, CD34 FITC/PE/PerCP, CD15 FITC, introduce a third marker which was usually a myeloid one. CD7 FITC, CD3 FITC/PerCP, CD4 FITC, CD8 PE, CD2 PE, By comparing the two techniques (double vs triple staining) CD5 PE, CD10 FITC, CD19 FITC/PE, CD22 PE, CD20 FITC, no changes were observed in the percentage of positivity for CD11b PE, CD11c PE, HLA-DR PerCP, Leu54 PE, CD56 a given lymphoid marker. For the cases which did not express FITC/PE, CD61 FITC, CD62 PE purchased from Becton CD34, the double staining was performed associating the Dickinson (San Jose, CA, USA); CD41 FITC, CD42 FITC pur- lymphoid marker which tested positive with a myeloid one. chased from Immunotech (Marseille, France). Mononuclear The double/triple stained samples were analyzed with the cells from peripheral blood and/or bone marrow samples were Paint-a-Gate program. This program provides multidimen- separated on Fycoll–Hypaque density gradient and washed sional, multicolour analysis of FACS-acquired list-mode data three times in phosphate-buffered saline solution (PBS). The files. It allows classification of events by painting them differ- samples contained у80% of blasts, as confirmed with May– ent colors and then quantifying these populations as a per- ¨ Grunwald–Giemsa preparation, and were analyzed within centage of the total events. This is in order to visualize in dif- 24 h from collection or cryopreserved with 10% fetal calf ferent plots cell populations which are difficult to see in two serum and 10% dimethyl sulfoxide and analyzed later after dimensions. To investigate cytoplasmic/nuclear antigens the quick thawing. The viability, as assessed by trypan blue following reagents were also used: CD3, anti-MPO (␣MPO), exclusion, was Ͼ95%. Before staining, cells were polyclonal TdT supplied by Technogenetics (Milan, Italy), and incubated with human AB serum (100 ␮l/ml) for 10 min at CD22 obtained from Becton Dickinson. To visualize TdT, room temperature to minimize the Fc receptor binding. After cytoplasmic CD3 and CD22 and ␣MPO cytocentrifuge prep- a further wash, cells were resuspended in PBS at a final con- arations were fixed in cold methanol for 30 min (nuclear centration of 10 × 107/ml. Each Moab was incubated with 106 staining) or acetone for 5–10 min at room temperature (cytoplasmic staining). After incubation with the primary Moab and the second layer for 30 min, the slides were washed Table 1 Patient population in PBSA and mounted with glycerol:PBS 1:1 and a coverslip. ␣MPO was also revealed by using the alkaline phosphatase Age median (range) 60 (18–86) anti-alkaline phosphatase (APAAP) method as previously 14,15 Sex M/F 186/149 described. Since 1994, intracellular antigens have also been investigated by flow cytometry using Ortho PermeaFix × 9 WBC 51 10 /l (OPF) according to the manufacturer’s recommendations.34 median range 0.7–400 × 109/l Tdt was regarded as positive when at least 20% of gated cells were more fluorescent than the isotypic control. Cases with a FAB (%) simultaneous expression of MPO, cCD3 and/or cCD22 were M0 32 (9) M1 68 (20) classified as genuine ‘biphenotypic’ leukemias and not M2 70 (21) included in the present study. We also tested expression of M3 34 (10) the multidrug-resistance P-glycoprotein (GP-170) (C219 FITC M4 65 (19) obtained from Centocor, Malvern, PA, USA) in 224 samples. M5 58 (17) As C219 reacts with an epitope located on the inner surface M6 5 (3) of the cell membrane, cells were fixed and permeabilized in M7 3 (1) 3% paraformaldehyde/PBS and 50% cold acetone/PBS. Overall CR rate 152/299 (51%) Samples were then incubated at +4°C for 30 min with 10 ␮l Duration of SV (weeks) of unconjugated C219 (final concentration: 10 ␮g/ml). After median 51 two washings in PBSA, cells were further incubated at +4°C range 1–396 with an FITC-conjugated F(ab)2 fragment of goat antimouse Ig Duration of CR (weeks) (dilution 1:20; Technogenetics, Milan, Italy). Negative con- median 61 trols were performed by incubating cells with an isotype anti- range 4–367 body (IgG2a). Analysis was performed by flow cytometry as mentioned above. In parallel, 130 of our cases were tested for Prognostic relevance of the expression of Tdt and CD7 in AML A Venditti et al 1058 the expression of GP-170 using other MoAbs (4E3, MRK16 Tdt expression (Table 2): The expression of Tdt was het- and UIC2) and no relevant differences were noted in terms of erogeneous in terms of number of cells positively stained rang- positivity compared to C219. ing between 21 and 80%. Moreover, its mean fluorescence intensity was quite uniformly dim and markedly weaker than that observed in lymphoblastic leukemia. In this view no cases Cytogenetics with bright Tdt expression could be identified, and the analy- sis was based on a positive-negative fashion model. Tdt was positive in 80/321 evaluable cases (25%). It was significantly Procedures for cytogenetic evaluation have been previously 35 associated with more immature FAB classes such as M0 and described in detail. We performed a methotrexate cell M1 (P Ͻ 0.001), accordingly a significant correlation between synchronization technique and a direct preparation, and Tdt and CD34 was found (P = 0.001). Sixty-six percent of Tdt examined the after Giemsa stain. Whenever positive cases also expressed GP-170 but the association was possible, at least 20 metaphases were analyzed. Karyotypes not statistically significant (P = 0.08). No significant corre- were classified in agreement with the guidelines of the Inter- 36 lations were observed between Tdt and lymphoid antigens national System for Human Cytogenetic nomenclature. The such as CD2, CD4, CD5, CD10, CD19, CD20 and CD56, as detection of a minimum of two metaphases with an identical well as between Tdt and sex, age, extramedullary disease or rearrangement or extra chromosome was regarded as evi- WBC higher than 50 × 109/l. dence of the existence of an abnormal clone. Complex kary- otypes were defined by the presence of three or more events 37 of translocation and nondisjunction in the same clone or by CD7 expression (Table 2): CD7 was expressed in 102/332 the presence of multiple unrelated clones. (31%) cases tested. It was significantly correlated with the FAB classes M0, M1 and M5 (P = 0.018), with CD34 (P Ͻ 0.001) and GP-170 (P Ͻ 0.001). Forty-five percent of CD7-positive Statistical analysis patients had a WBC у50 × 109/l (P = 0.009), while no corre- lations were found with lymphoid antigens such as CD2, CD4, Relationship of the expression of biological data to quantitat- CD5, CD10, CD19, CD20 and CD56 or with other clinical ive parameters (age, blood counts, percentage of marrow parameters such as sex, age or extramedullary disease. blasts) was studied by analysis of variance (ANOVA). Relationship of marker expression to patient’s characteristics and response to treatment were estimated by two-sided ␹2 test Cytogenetic analysis (or Fisher’s exact test when either group included fewer than 20 cases). A P value of 0.05 or less was considered to be Karyotypes were available in 221 of the 335 cases (66%) and significant. Since each patient category was assigned to a spe- revealed chromosomal defects in 170 (77%) of them. As cific therapeutic protocol, the effects of treatment were expected, a significant correlation (P = 0.001) was observed specifically factored before significance was assigned. The between certain chromosomal abnormalities and specific FAB Kaplan–Meier product-limit was used for estimation of sur- subsets (Table 3): t(15;17) was exclusively associated with M3 vival (SV) and continuous CR (CCR) duration. SV was calcu- subtype; t(8;21) was seen in M2 cases but 2; 6 of 13 t(9;22) lated from the date of diagnosis to the date of death or last belonged to the M1 subset. Twenty-four of 30 cases of anom- follow-up. CCR was measured from achievement of CR until alies of chromosome 5 and/or 7 were detected in the FAB relapse or death in remission. Patients undergoing bone mar- categories M0, M1 and M2, and finally 11/15 rearrangements row transplantation were censored at the time of marrow of chromosome 11 at band q23 were consistent with infusion. For comparison of survival and remission duration myelomonocytic/monocytic differentiation. In order to gain of two or more groups, the log- test was applied. To prognostic indications linked to the expression of Tdt or CD7, evaluate the simultaneous impact of different variables on dur- correlations of these markers with specific karyotypic abnor- ation of SV and CCR, a multivariate analysis was performed malities were investigated. Eight of 13 cases (61%) with t(9;22) using the Cox regression model. Conversely, a stepwise and 11/26 (42%) with anomalies of chromosome 5 and/or 7 Ͻ regression model was used to assess the effect of different vari- were Tdt positive (P 0.0001). Only one (7%) case of ables on achievement of CR. Only variables for which univari- rearrangement at 11q23 was Tdt-positive. Eight of 13 cases ate analysis had shown a significant association were (61%) with t(9;22), 12/30 (40%) with anomalies of chromo- considered in the multivariate analysis. some 5 and/or 7 and 5/15 (33%) with rearrangements 11q23 were CD7 positive (P = 0.025). Conversely, cases with t(8;21) were Tdt and CD7 negative; cases with t(15;17) were Tdt negative and only 9% (3/34) were CD7 positive (Table 4). Results

Immunophenotype Prognostic features: Clinical data are reported in Table 1. CR was achieved in 152 patients out of 299 assessable for Overall 41% (137/333) of our cases expressed lymphoid anti- response (51%). Median duration of survival and CR was 51 gens; this figure did not take into account Tdt and CD7 which and 61 weeks, respectively. When considering the CR rate in were analyzed separately as no longer considered specific for relation to Tdt and CD7 expression, a significant difference lymphoid lineage. Incidence of CD2, CD4, CD5, CD10, was observed between negative and positive cases; Tdt-nega- CD19, CD20 and CD56 and their relationship with FAB tive patients had a CR rate of 58 vs 34% of Tdt-positive classes are showed in Table 2. Only CD4 and CD56 showed (P Ͻ 0.001), while CR rate for CD7-negative patients was 56 a statistically significant association with M4 and M5 AML vs 40% of CD7-positive (P = 0.013). Even survival rates were (P = 0.013 and 0.022, respectively. significantly shorter for patients expressing Tdt or CD7 Prognostic relevance of the expression of Tdt and CD7 in AML A Venditti et al 1059 Table 2 Incidence of Tdt, CD7 and lymphoid antigens according to FAB classes

AML-M0 AML-M1 AML-M2 AML-M3 AML-M4 AML-M5 AML-M6 AML-M7 Total P value

Tdt 19/32 (60) 23/65 (35) 9/64 (14) 0/34 (0) 15/61 (25) 13/57 (23) 0/5 (0) 1/3 (33) 80/321 (25) Ͻ0.001 CD7 11/32 (34) 26/68 (38) 18/70 (26) 3/34 (9) 17/63 (27) 25/57 (44) 2/5 (40) 0/3 (0) 102/332 (31) 0.018 CD2 2/31 (6) 4/66 (6) 9/70 (13) 8/34 (23) 9/63 (14) 11/57 (19) 2/5 (40) 0/3 (0) 45/329 (14) NS CD4 5/22 (23) 4/37 (11) 6/46 (13) 3/20 (15) 11/44 (25) 20/48 (42) 0/5 (0) 1/3 (33) 50/225 (22) 0.013 CD5 2/26 (8) 1/41 (2) 2/44 (4) 1/20 (5) 0/45 (0) 6/50 (12) 0/5 (0) 0/2 (0) 11/201 (5) NS CD10 3/31 (10) 3/68 (4) 4/70 (6) 0/34 (0) 2/63 (3) 5/58 (9) 1/5 (20) 0/3 (0) 18/332 (5) NS CD19 3/32 (9) 6/68 (9) 5/70 (7) 2/34 (6) 3/63 (5) 2/58 (3) 0/5 (0) 0/3 (0) 21/333 (6) NS CD20 0/29 (0) 2/54 (4) 1/55 (2) 0/28 (0) 2/54 (4) 3/49 (6) 0/4 (0) 0/3 (0) 8/278 (3) NS CD56 6/25 (24) 10/41 (24) 11/46 (24) 5/20 (20) 18/45 (40) 27/51 (53) 1/5 (20) 0/3 (0) 78/236 (33) 0.022

Values are number of positive cases/number of cases evaluated (percentage).

Table 3 Chromosomal pattern according to different FAB classes

AML-M0 AML-M1 AML-M2 AML-M3 AML-M4 AML-M5 AML-M6 AML-M7 Total

Normal karyotype 4/23 (20) 9/44 (20) 15/50 (30) 0/34 (0) 16/40 (40) 4/25 (16) 2/4 (50) 0/1 (0) 50/221 (22) t(15;17) 0/23 (0) 0/44 (0) 0/50 (0) 34/34 (100) 0/40 (0) 0/25 (0) 0/4 (0) 0/1 (0) 34/221 (15) t(9;22) 1/23 (4) 6/44 (14) 2/50 (4) 0/34 (0) 1/40 (2) 2/25 (8) 0/4 (0) 1/1 (100) 13/221 (6) t(8;21) 0/23 (0) 1/44 (2) 4/50 (8) 0/34 (0) 0/40 (0) 1/25 (4) 0/4 (0) 0/1 (0) 6/221 (3) Anomalies of chromosome 5/7 5/23 (22) 8/44 (18) 11/50 (22) 0/34 (0) 4/40 (10) 2/25 (8) 0/4 (0) 0/1 (0) 30/221 (14) Trisomy 8 2/23 (9) 4/44 (9) 1/50 (2) 0/34 (0) 2/40 (5) 0/25 (0) 0/4 (0) 0/1 (0) 9/221 (4) Trisomy 4 2/23 (9) 0/44 (0) 1/50 (2) 0/34 (0) 1/40 (2) 0/25 (0) 0/4 (0) 0/1 (0) 4/221 (2) Inv(16) and other abnormalities 0/23 (0) 3/44 (7) 1/50 (2) 0/34 (0) 4/40 (10) 1/25 (4) 1/4 (25) 0/1 (0) 10/221 (4) of chromosome 16 Rearrangements at band 11q23 1/23 (4) 2/44 (4) 0/50 (0) 0/34 (0) 4/40 (10) 7/25 (28) 1/4 (25) 0/1 (0) 15/221 (7) Other abnormalities 8/23 (34) 11/44 (24) 15/50 (30) 0/34 (0) 8/40 (20) 8/25 (32) 0/4 (0) 0/1 (0) 50/221 (22)

Values are number of positive cases/number of cases evaluated (percentage).

Table 4 Incidence of Tdt, CD7 and lymphoid antigens according to chromosomal classes

Tdt CD7 CD2 CD4 CD5 CD10 CD19 CD20 CD56

Normal karyotype 8/48 (17) 13/50 (26) 4/50 (8) 11/35 (31) 0/34 (0) 4/50 (8) 3/50 (4) 0/43 (0) 14/36 (39) t(15;17) 0/34 (0) 3/34 (9) 8/34 (23) 3/20 (15) 1/20 (5) 0/34 (0) 2/34 (6) 0/28 (0) 5/20 (25) t(9;22) 8/13 (61) 8/13 (61) 1/13 (8) 2/10 (20) 0/10 (0) 2/13 (15) 2/13 (15) 1/11 (9) 4/10 (40) t(8;21) 0/6 (0) 0/6 (0) 0/6 (0) 1/6 (17) 0/5 (0) 1/6 (17) 2/6 (33) 0/6 (0) 3/5 (60) Anomalies of chromosome 5/7 11/26 (42) 12/30 (40) 3/30 (10) 1/13 (8) 0/15 (0) 0/30 (0) 1/30 (3) 0/26 (0) 2/15 (13) Trisomy 8 2/9 (22) 3/9 (33) 1/9 (11) 1/3 (33) 0/3 (0) 1/9 (11) 0/9 (0) 0/9 (0) 1/3 (33) Trisomy 4 1/4 (25) 0/4 (0) 0/4 (0) 1/1 (50) 0/2 (0) 0/4 (0) 1/4 (25) 0/3 (0) 1/1 (50) Inv(16) and other abnormalities 3/10 (30) 2/10 (20) 4/10 (40) 2/8 (25) 0/8 (0) 0/10 (0) 0/10 (0) 0/7 (0) 2/8 (25) of chromosome 16 Rearrangements at band 11q23 1/15 (7) 5/15 (33) 5/10 (33) 2/7 (29) 1/6 (17) 2/15 (13) 0/15 (0) 3/10 (30) 3/9 (33) Other abnormalities 13/47 (28) 21/29 (42) 7/49 (14) 3/25 (12) 5/27 (18) 2/50 (4) 5/50 (10) 4/45 (9) 6/26 (23) P value Ͻ0.001 0.025 NS NS NS NS NS 0.004 NS

Values are number of positive cases/number of cases evaluated (percentage).

(P = 0.006 and 0.001, respectively) (Figure 1a and b). No sig- were found to adversely affect CCR (P = 0.001 and 0.037, nificant differences were observed as regard to CCR between respectively) (Table 5). In order to evaluate to what extent the patients Tdt or CD7 positive or negative. Multivariate analysis combination of both Tdt and CD7 might further worsen the was also performed to test the simultaneous effect of age (Ͻ60 prognosis, four phenotypic subclasses were identified for or Ͼ60 years), WBC, FAB, karyotype, CD7, Tdt and GP-170 additional analysis: [Tdt−CD7−], [Tdt+CD7−], [Tdt−CD7+] and on CR rate, overall survival and CCR. Variables were entered [Tdt+CD7+] (Table 6). Patients [Tdt+CD7+] had the lowest rate into the model and non-significant factors were removed in a of CR (30%) (P Ͻ 0.001); even duration of survival was mark- stepwise fashion followed by a re-estimation of effects and edly shortened as compared to cases [Tdt+CD7−], [Tdt−CD7+] significance at each step. Tdt emerged as an independent fac- and [Tdt−CD7−] (Figure 2a, b, and c). No significant reduction tor affecting CR achievement (P = 0.018) together with WBC of CCR was found for cases [Tdt+CD7−] or [Tdt−CD7+] com- (P = 0.001), age (P Ͻ 0.001), FAB (P = 0.012), karyotype pared to negative ones (P = 0.5 for both). Conversely, com- (P Ͻ 0.001) and GP-170 (P Ͻ 0.001). With regard to survival, parison of cases [Tdt+CD7+] and [Tdt−CD7−] resulted in a dif- the independent role of GP-170 (P = 0.001), FAB (P = 0.001) ference of CCR almost approaching a statistical significance and karyotype (P = 0.018) was confirmed. Karyotype and CD7 (P = 0.07). As for biological features, [Tdt+CD7+] cases are Prognostic relevance of the expression of Tdt and CD7 in AML A Venditti et al 1060

Figure 1 The overall survival of cases CD7+ (a) and Tdt+ (b) as compared to negative ones.

Table 5 The results of multivariate analysis

CR SV CCR

Age Ͻ0.001 NS NS FAB 0.012 0.001 NS WBC 0.001 NS NS Karyotype Ͻ0.001 0.018 0.001 GP-170 Ͻ0.001 0.001 NS Tdt 0.018 NS NS CD7 NS NS 0.037

The results of multivariate analysis, aimed at evaluating the simul- taneous impact on CR rate, overall SV and CCR of age, FAB, WBC, karyotype, GP-170, Tdt and CD7, are shown. Tdt and CD7 emerged as independent factors affecting CR achievement and duration of SV, respectively. Figure 2 The overall survival of cases [Tdt−CD7−] as compared to + − − + + + characterized by the greatest concentration of poor prognosis- [Tdt CD7 ] (a), [Tdt CD7 ] (b) and [Tdt CD7 ] (c). related factors and which we have already mentioned above to be associated with Tdt or CD7, although at lower frequency (Table 6). In fact, cases [Tdt+CD7+] show the most frequent Discussion association with MDR phenotype (77% of the cases) (P = 0.003), deviant expression of LyAg (69% of the cases) The purpose of this study was to analyze the pattern of LyAg (P Ͻ 0.001), chromosome PhЈ or anomalies of chromosome 5 expression in a large series of AML and determine its clinical and/or 7 (50% of the cases) (P Ͻ 0.001). No association was implications. The incidence of LyAg in our series (41%) was found with rearrangements at band 11q23. comprised within the ranges reported in the literature.1,11,21,38–42 Prognostic relevance of the expression of Tdt and CD7 in AML A Venditti et al

− − + − − + + + 1061 Table 6 Clinical, phenotypical and karyotypic features of the subsets [Tdt CD7 ], [Tdt CD7 ], [Tdt CD7 ] and [Tdt CD7 ]

[Tdt−CD7−] [Tdt+CD7−] [Tdt−CD7+] [Tdt+CD7+] P value

CR 93/147 (63) 17/47 (36) 27/61 (44) 9/30 (30) Ͻ0.001 CD34 68/155 (44) 42/47 (89) 49/66 (74) 27/33 (84) Ͻ0.001 GP-170 49/117 (42) 21/40 (52) 37/58 (64) 20/26 (77) 0.003 LyAg 51/173 (29) 20/48 (42) 43/67 (64) 22/32 (69) Ͻ0.001 PhЈ or defects 5/7 12/117 (10) 9/27 (33) 8/44 (18) 10/20 (50) Ͻ0.001 11q23 abnormalities 8/119 (7) 1/27 (4) 5/44 (11) 0/20 (0) NS

Values are number of positive cases/number of cases evaluated (percentage).

Expression of CD2, CD5, CD10, CD19, and CD20 lacked any The contribution of CD7 in this dual model seems strictly significant prognostic impact or classificative relevance, while related to the expression of GP-170. All together these data CD4 and CD56 were significantly associated with M4 and M5 indicate that Tdt+ and/or CD7+ AML are prone to retain signs subtypes cases (Table 2). In particular we have not observed of immaturity and infidelity and that the presence of Tdt and the association between immunophenotype and karyotype CD7 as a single marker or in combination may correspond to reported in the CALGB and EORTC studies,43,44 probably different degrees of such immaturity/infidelity. The data because of the smaller number of our cases (Table 4). Tdt was presented are also of some interest since biologic similarities significantly associated with AML-M0, M1 and CD34 thus between de novo CD7+ and/or Tdt+ AML and secondary leu- confirming its link with immature leukemias.7–11,14,15 CD7 was kemia are evident. Secondary AML frequently occurs in the significantly associated with AML-M0, M1, M5, with CD34 elderly (over 60 years) and expresses an MDR phenotype and GP-17045 in line with other reports which have also pos- within the context of multiple and non-specific abnormalities tulated the expansion of a pluripotent CD7+ precursor in that are cytogenetically complex and associated with gain and immature AML.12,24,46,47 A significant proportion of patients loss of certain chromosomal regions (eg abnormalities of chro- whose blasts expressed Tdt or CD7 failed to enter a CR mosome 5/7).49–52 Our phenotypic subgroups fit exactly this (P Ͻ 0.001 and 0.013, respectively) and experienced a shorter criteria. Our findings are also consistent with the results pub- survival (P = 0.006 and 0.001, respectively) (Figure 1a and b); lished by Cuneo et al53 and Carbonell et al54 who demon- in addition, in a multivariate analysis, Tdt and CD7 expression strated that among biphenotypic leukemia, in the context of were found to adversely affect achievement and duration of a general heterogeneity, some chromosomal abnormalities CR (P = 0.018 and 0.037, respectively). Based on this obser- appeared to be more common: t(9;22), 11q23 rearrangements vation, four phenotypic classes of AML were distinguished for and defects of chromosome 5/7. Therefore, we conclude that further analysis: [Tdt−CD7−], [Tdt+CD7−], [Tdt−CD7+] and the expression of typical LyAg in AML lacks prognostic value, [Tdt+CD7+]. Overall, the results indicated that the subsets while Tdt and/or CD7, no longer regarded as genuine lymph- [Tdt+CD7+], [Tdt+CD7−] and [Tdt−CD7+] were characterized oid markers, design a ‘discrete’ entity which closely resembles by a greater incidence of GP-170, t(9;22) and defects of chro- secondary leukemias and whose dismal outcome may be attri- mosome 5/7 compared to [Tdt−CD7−] (Table 6). In addition, buted to the convergence of factors such as GP-170, t(9;22), the category [Tdt+CD7+] was characterized by a more unfavor- defects of chromosome 5/7, immaturity and marked degree of able prognosis as suggested by a lower rate of CR and a lineage infidelity. This observation also indicates that certain shorter survival as compared to cases [Tdt+CD7−], [Tdt−CD7+] chromosomal defects may lead to treatment failure since they and [Tdt−CD7−] (Figure 2a, b and c). These results reaffirm the are intrinsically associated with specific mechanisms of chem- poor prognostic role of CD7 expression in AML,5,13 and high- oresistance, such as GP-170 and abrogation of apoptosis. light the prognostic importance of Tdt expression.48 We Banker et al55 have recently demonstrated that some cyto- speculate that the expression of Tdt, CD7 or both reflect vari- genetic groups such as del(7) show consistently low apoptosis, able stages of leukemic immaturity. Indeed, the [Tdt+CD7+], both spontaneously or after exposure to chemotherapeutic [Tdt+CD7−] and [Tdt−CD7+] leukemias show: (1) a variable agents in vitro. In this view, the expression of Tdt and/or CD7 percentage of incidence of t(9;22) and defects of chromosome may represent the phenotypic counterpart of cytogenetic and 5/7, with cases [Tdt+CD7+] having the highest and those chemoresistance patterns which need to be searched for in [Tdt−CD7+] the lowest occurrence; (2) a variable percentage order to deliver therapies as tailored as possible, even of incidence of LyAg, with the group [Tdt+CD7+] having the including MDR modulators, apoptosis-inducers and stem highest and those [Tdt+CD7−] the lowest occurrence; (3) occur cell transplantation. more frequently in AML-M0, M1 and M5. This is opposite to cases [Tdt−CD7−] which have a significant lower incidence of PhЈ, defects of chromosome 5/7, LyAg and involve basically Acknowledgements AML-M2, M3 and M4. Of note, is the significant presence of t(9;22) or defects in chromosomes 5 and/or 7 linked to the The authors are grateful to Dr Dario Campana for critically dual expression of CD7 and Tdt (Table 5). This raises the issue reviewing the manuscript. This work was supported in part by of the relative contribution of the diverse molecular factors to CNR, special project ACRO, contract no. 96.00498.PF39 and net clinical outcome. While t(9;22) and defects in chromo- by MURST 40%. somes 5 and/or 7 are themselves powerful predictors of poor clinical outcome it is conceivable that additional mechanisms References of chemoresistance operationally connected to Tdt may be ¨ active in the leukemic cells. One of these mechanisms might 1 Terstappen LWMM, Safford M, Konemann S, Loken RL, Zurlutter ¨ ¨ be bcl-2-related (Dr G Del Poeta, unpublished observation). K, Buchner T, Hiddemann W, Wormann B. Flow cytometric Prognostic relevance of the expression of Tdt and CD7 in AML A Venditti et al 1062 characterization of acute myeloid leukemia. Part II. Phenotypic otidyl transferase positive subpopulation occur in the majority of heterogeneity at diagnosis. Leukemia 1991; 5: 757–767. ANLL: implications for the detection of minimal disease. Leukemia 2 Bennet JM, Catovsky D, Daniel MT, Flandrin G, Galton DAG, 1990; 4: 404–410. Gralnick HR, Sultan C (FAB Cooperative Group). Proposals for the 21 Campana D, Pui CH. Detection of MRD in acute leukemia: recognition of minimally differentiated acute myeloid leukaemia methodologic advances and clinical significance. Blood 1995; 85: (AML-M0). Br J Haematol 1991; 78: 325–329. 1418–1434. 3 Buccheri V, Matutes E, Dyer MJS, Catovsky D. Lineage commit- 22 Drach J, Drach D, Glassl H, Gattringer C, Huber H. Flow cytome- ment in biphenotypic acute leukemia. Leukemia 1993; 7: 919– try determination of atypical antigen expression in acute leukemia 927. for the study of minimal residual disease. Cytometry 1992; 13: 4 Greaves MF, Chan LC, Furley AJW, Watt SM, Molgaard HV. Lin- 893–901. eage promiscuity in hemopoietic differentiation and leukemia. 23 Drach J, Gattringer C, Huber H. Combined flow cytometry assess- Blood 1986; 67: 1–11. ment of cell surface antigens and nuclear Tdt for the detection of 5 Kita K, Miwa H, Nakase K, Kawakami K, Kobayashi T, Shirakawa minimal residual disease in acute leukemia. Br J Haematol 1991; S, Tanaka I, Ohta C, Tsutani H, Oguma S, Kyo T, Dohy H, Kamada 77: 37–42. N, Nasu K, Uchino H (The Japan Cooperative Group of 24 Venditti A, Del Poeta G, Stasi R, De Fabritiis P, Coppetelli U, Leukemia/). Clinical importance of CD7 expression in Bruno A, Simone MD, Papa G. Triple immunofluorescence evalu- acute myelocytic leukemia. Blood 1993; 81: 2399–2405. ation of CD15, CD34 and class II expression by flow cytometry 6 Second MIC Cooperative Study Group. Morphologic, immuno- in normal and leukemic bone marrows. Haematologica 1993; 78: logic, and cytogenetic (MIC) working classification of the acute 359–363. myeloid leukemias. Br J Haematol 1988; 68: 487–494. 25 Ball ED, Davis RB, Griffin JD, Mayer RJ, Davey FR, Arthur DC, 7 Parreira A, Pombo de Oliveira MS, Matutes E, Foroni L, Morilla R, Wurster-Hill D, Noll W, Elghetany MT, Allen SL, Rai K, Lee EJ, Catovsky D. Terminal deoxynucleotidyl transferase positive acute Schiffer CA, Bloomfield CD. Prognostic value of sur- myeloid leukemia: an association with immature myeloblastic face markers in acute myeloid leukemia. Blood 1991; 77: leukemia. Br J Haematol 1988; 69: 219–224. 2242–2250. 8 Bradstock KF, Hoffbrand AV, Ganeshaguru K, Liewellin P, Pat- 26 Reading CL, Estey EH, Huh YO, Claxton DF, Sanchez G, Ter- terson K, Wonke B, Prentice AG, Bennett M, Pizzolo G, Bollum stappen LWMM, O’Brien MC, Baron S, Deisseroth AB. Expression FJ, Janossy G. Terminal deoxynucleotidyl transferase expression in of unusual immunophenotype combination in acute myelogenous acute non-: an analysis by immunofluoresc- leukemia. Blood 1993; 81: 3083–3090. ´ ence. Br J Haematol 1981; 47: 133–143. 27 Macedo A, Orfao A, Vidriales MB, Lopez-Berges MC, Valverde B, ´ ´ ´ 9 San Miguel JF, Gonzales M, Canizo MC, Anta JP, Portero JA, Gonzalez M,´ Caballero MD, Ramos F, Martnez M, Fernandez- Lopez-Borrasca A. Tdt activity in acute myeloid leukemias defined Calvo J, Martnez A, San Miguel JF. Characterization of aberrant by monoclonal antibodies. Am J Hematol 1986; 23: 9–17. phenotypes in acute myeloblastic leukemia. Ann Hematol 1995; 10 Lee EJ, Yang J, Leavitt RD, Testa JR, Civin CI, Forrest A, Schiffer 70: 189–194. CA. The significance of CD34 and Tdt determination in patients 28 Kurtzberg J, Waldmann TA, Davey MP, Bigner SH, Moore JO, with untreated de novo acute myeloid leukemia. Leukemia 1992; Hershfield MS, Haynes B. CD7+, CD4−, CD8− acute leukemia: a 6: 1203–1209. syndrome of malignant pluripotent lymphohematopoietic cells. 11 Drexler HG, Thiel E, Ludwig WD. Acute myeloid leukemias Blood 1989; 73: 381–390. expressing lymphoid-associated antigens: diagnostic incidence 29 Cross AH, Goorha RM, Nuss R, Behm F, Murphy SB, Kalwinsky and prognostic significance. Leukemia 1993; 7: 489–498. DK, Raimondi S, Kitchingman R, Mirro J. Acute myeloid leukemia 12 Vinante F, Pizzolo G, Rigo A, Vincenzi C, Sorio C, Cassatella M, with T-lymphoid features: a distinct biologic and clinical entity. Berton G. The CD4 molecule belongs to the phenotypic repertoire Blood 1988; 72: 579–587. of most cases of acute myeloid leukemia. Leukemia 1992; 6: 30 Launder TM, Bray RA, Stempora L, Chenggis ML, Farhi DC. 1257–1262. Lymphoid-associated antigen expression by acute myeloid leuke- 13 Del Poeta G, Stasi R, Venditti A, Suppo G, Aronica G, Bruno A, mia. Am J Clin Pathol 1996; 106: 185–191. Masi M, Tabilio A, Papa G. Prognostic value of cell marker analy- 31 Bennett JM, Catovsky D, Daniel MT, Flandrin G, Galton DAG, sis in de novo acute myeloid leukemia. Leukemia 1994; 8: Gralnick HR, Sultan C. Proposed revised criteria for the classi- 388–394. fication of acute myeloid leukemia. Ann Intern Med 1985; 103: 14 Venditti A, Del Poeta G, Stasi R, Masi M, Bruno A, Buccisano F, 620–629. Cox C, Coppetelli U, Aronica G, Simone MD, Tribalto M, Amadori 32 Avvisati G, Lo Coco F, Diverio D, Falda M, Ferrara F, Lazzarino S, Papa G. Minimally differentiated acute myeloid leukaemia M, Russo D, Petti MC, Mandelli F. AIDA (all-trans retinoic (AML-M0): cytochemical, immunophenotypic and cytogenetic acid + idarubicin) in newly diagnosed promyelocytic leukemia: a analysis of 19 cases. Br J Haematol 1994; 88: 784–793. Gruppo Italiano Malattie Ematologiche Maligne dell’Adulto 15 Venditti A, Del Poeta G, Buccisano F, Tamburini A, Cox-Froncillo (GIMEMA) pilot study. Blood 1996; 88: 1390–1398. MC, Stasi R, Bruno A, Aronica G, Maffei L, Suppo G, Simone MD, 33 Preisler HD. Treatment failure in AML. Blood Cells 1982; 8: Forte L, Cordero V, Postorino M, Tufilli V, Isacchi G, Papa G, 585–590. Amadori S. Minimally differentiated acute myeloid leukemia 34 Pizzolo G, Vincenzi C, Nadali G, Veneri D, Vinante F, Chilosi M, (AML-M0): comparison of 25 cases with other FAB subtypes. Basso G, Connelly MC, Janossy G. Detection of membrane and Blood 1997; 89: 621–629. intracellular antigens by flow cytometry following Ortho Perme- 16 Shetty V, Chitale A, Matutes E, Buccheri V, Morilla R, Catovsky D. aFix fixation. Leukemia 1994; 8: 1302–1307. Immunological and ultrastructural studies in acute biphenotypic 35 Stasi R, Del Poeta G, Masi M, Tribalto M, Venditti A, Papa G, leukaemia. J Clin Pathol 1993; 46: 903–907. Nicoletti B, Vernole P, Tedeschi B, Delaroche I, Mingarelli R, Dal- 17 Bradstock K, Matthews J, Benson E, Page F, Bishop J and the Aus- lapiccola B. Incidence of chromosome abnormalities and clinical tralian Leukaemia Study Group. Prognostic value of immunophen- significance of karyotype in de novo acute myeloid leukaemia. otyping in acute myeloid leukemia. Blood 1994; 84: 1220–1225. Cancer Genet Cytogenet 1993; 67: 28–34. 18 Paietta E, Van Ness B, Bennett J, Gucalp R, Cassileth P, Wiernik 36 Mitelman F (ed). ISCN. Guidelines for Cancer Cytogenetics. Sup- PH. Lymphoid lineage-associated features in acute myeloid leu- plement to an International System for Human Cytogenetic kaemia: phenotypic and genotypic correlations. Br J Haematol Nomenclature. Karger: Basel, Switzerland, 1995. 1992; 82: 324–329. 37 Sandberg AA. Cytogenetic features of special hematologic dis- 19 Lo Coco F, De Rossi G, Pasqualetti D, Lopez M, Diverio D, Latag- orders, including some aspects of acute leukemia. In: Sandberg liata R, Fenu S, Mandelli F. CD7-positive acute myeloid leu- AA (ed). The Chromosomes in Human Cancer and Leukemia (ed kaemia: a subtype associated with cell immaturity. Br J Haematol 2). Elsevier: NY, 1990, pp. 374–380. 1989; 73: 480–485. 38 Pui CH, Raimondi SC, Head DR, Schell MJ, Rivera GK, Mirro J ¨ 20 Adriaansen HJ, van Dongen JJM, Kappers-Klunne MC, Hahlen K, Jr, Crist WM, Behm FG. Characterization of childhood acute leu- van’t Veer MB, Wijdenes-de Bresser JHFM, Holdrinet ACJM, Har- kemia with multiple myeloid and lymphoid markers at relapse. thoorn-Lasthuizen EJ, Abels J, Hooijkaas H. Terminal deoxynucle- Blood 1991; 78: 1327–1337. Prognostic relevance of the expression of Tdt and CD7 in AML A Venditti et al

+ 1063 39 Ludwig WD, Bartram CR, Ritter J, Raghavachar A, Hiddemann W, 47 Zutter MM, Martin PJ, Hanke D, Kidd PG. CD7 acute non-lym- Heil G, Harbott J, Seibt-Jung H, Teichmann JV, Riehm H. Ambigu- phocytic leukemia: evidence for an early multipotential progeni- ous phenotypes and genotypes in 16 children with acute leukemia tor. Leukemia Res 1990; 14: 23–26. as characterized by multiparametric analysis. Blood 1988; 71: 48 Lo Coco F, Lopez M, Pasqualetti D, Montefusco E, Cafolla A, 1518–1524. Monarca B, Sgadari C, De Rossi G. Terminal transferase positive 40 Kaplan SS, Penchansky L, Stolc V, Contis L, Krause JR. Immuno- acute myeloid leukemia: immunophenotypic characterization and phenotyping in the classification of acute leukemia in adults. response to induction therapy. Hematol Oncol 1989; 7: 167–174. Interpretation of multiple lineage reactivity. Cancer 1989; 63: 49 Dustugue N, Payen C, Lafage-Pochitaloff M. Prognostic signifi- 1520–1526. cance of karyotype in the de novo adult acute myeloid leukemia. 41 Smith FO, Lampkin B, Versteeg C, Flowers D, Dinndorf P, Buckley Leukemia 1995; 9: 1491–1496. J, Woods W, Hammond D, Bernstein ID. Expression of lymphoid- 50 Leith CP, Kopecky KJ, Godwin J. AML in the elderly: stratification associated cell surface antigens by childhood acute myeloid leu- by MDR phenotype, secondary vs de novo status, and cytogen- kemia cells lack prognostic significance. Blood 1992; 79: 2415– etics identifies patients with high complete remission rate. A 2420. SWOG study. J Clin Oncol 1996; 15: 359 (Abstr.). 42 Bradstock KF, Kirk J, Grimsley PG, Kabral A, Hughes WG. 51 Taylor PRA, Reid MM, Stark AN. De novo acute myeloid leukemia Unusual immunophenotypes in acute leukaemias: incidence and in patients over 55 years old: a population-based study of inci- clinical correlations. Br J Haematol 1989; 72: 512–518. dence, treatment and outcome. Leukemia 1995; 9: 231–237. 43 Baer MR, Stewart CC, Lawrence D, Powell BL, Schiffer CA, Davey FR, Arthur DC, Bloomfield CD. Multiparameter flow cytometry 52 Johansson B, Mertens F, Mitelman F. Primary vs secondary neo- analysis of de novo acute myeloid leukemia reveals low fre- plasia associated chromosomal abnormalities – balanced quencies of lymphoid antigens expression and strong association rearrangements vs genomic imbalance? Chromosom Can- between immunophenotype and karyotype (CALGB 8361). Blood cer 1996; 16: 155–161. 1995; 86: 266a (Abstr.). 53 Cuneo A, Michaux JL, Ferrant A, Van Hove L, Bosly A, Stul M, 44 Bernier M, Suciu S, Stryckmans P, de Witte Th, Dardenne M, Dal Cin P, Vandenberghe E, Cassiman JJ, Negrini M, Piva N, Cas- Solbu G, Vegna ML, Willemze R, Mandelli F, Zittoun R. Immuno- toldi G, Van Den Berghe H. Correlation of cytogenetic pattern and phenotype of the EORTC/GIMEMA AML 10 protocol: clinical rel- clinicobiological features in adult acute myeloid leukemia evance and cytogenetic correlation. Blood 1996; 88: 561a expressing lymphoid markers. Blood 1992; 79: 720–727. (Abstr.). 54 Carbonell F, Swansbury J, Min T, Matutes E, Farahat N, Buccheri 45 Del Poeta G, Stasi R, Germano G, Venditti A, Cox MC, Bruno A, V, Morilla R, Secker-Walker L, Catovsky D. Cytogenetic findings Buccisano F, Masi M, Tribalto M, Amadori S, Papa G. Clinical in acute biphenotypic leukemia. Leukemia 1996; 10: 1283–1286. relevance of P-glycoprotein expression in de novo acute myeloid 55 Banker DE, Groudine M, Norwood T, Appelbaum FR. Measure- leukemia. Blood 1996; 87: 1997–2004. ment of spontaneous and therapeutic agent-induced apoptosis 46 Chabannon C, Wood P, Torok-Storb B. Expression of CD7 on nor- with bcl-2 expression in acute myeloid leukemia. Blood mal human myeloid progenitors. J Immunol 1992; 149: 2110– 1997; 89: 243–255. 2113.