Minimally Differentiated Acute Myeloid Leukemia (AML-M0): Comparison of 25 Cases With Other French-American-British Subtypes

OVER THE PAST FEW YEARS, there has been an increasing interest for a new distinct entity, characterized by fewer than 3% myeloperoxidase-positive (MPO⁺)/Sudan Black B-positive (SBB⁺) blasts, which has been referred to as “minimally differentiated acute myeloid leukemia” (or AML-M0).1-15 As a consequence of the ever more frequent description of these cases, the members of the French-American-British (FAB) group permanently included this particular variant of myeloid leukemia into the FAB classification.16 They categorized this leukemia as AML-M0 and established strict morphological, cytochemical, and phenotypic criteria based on which diagnosis could be made.16 By definition, the diagnosis of AML-M0 requires less than 3% MPO⁺ and/or SBB⁺ blasts, expression of myeloid-associated markers, and lack of B/T-lineage–associated antigens.16 Thus, for the recognition of AML-M0 (and AML-M7), immunophenotypic evaluation is integrated in the FAB scheme as a part of the diagnostic procedure. Apart from the biological interest and the need for leukemia assignment, great clinical concerns are at stake owing to the very unfavorable prognosis frequently reported for AML-M0.1,3-9 We and others4,5-8 recently pointed out that the concurrent expression of factors having a negative prognostic impact, such as CD7, GP-170, complex karyotypes, and age over 60 years, is a common feature of AML-M0, possibly accounting for its poor outcome. Here, we present the results of a comparative analysis between 25 cases of AML-M0 and 247 AML cases with more differentiated morphology. Our data indicate the reliability of anti-MPO antibodies to detect minimal myeloid differentiation and show the high frequency of CD34 and terminal deoxynucleotidyl transferase (TdT) expression and complex karyotypes, including anomalies of chromosomes 5 and 7 in AML-M0. In this series, we could find no substantial difference with regard to the incidence of some poor prognostic features (CD7, GP-170, age, and white blood cell count) among AML-M0 and other AMLs.17 18 

Patients.The cases for this study were identified among 272 consecutive patients with newly diagnosed AML observed at our institution between 1987 and 1995. A total of 150 men and 122 women with a median age of 58 years (range, 18 to 81 years) were classified according to the FAB guidelines16,19 as follows: M0, 25 patients (9%); M1, 54 patients (20%); M2, 59 patients (22%); M3, 29 patients (11%); M4, 51 patients (19%); M5, 51 patients (19%); and M6 and M7, 3 patients (1%). No patients had a history of prior therapies with anticancer drugs or a diagnosis of myelodysplastic syndrome (MDS). Antileukemic treatment was differentiated according to age; patients over 60 years of age (n = 86) were administered a combination of cytarabine (ARA-C; 1 g/m²/d) for 6 days and mitoxantrone (6 mg/m²/d) for 6 days as remission induction therapy. Twenty-nine patients with acute promyelocytic leukemia (APL), regardless of age, received an induction course with idarubicine (10 mg/m²/d) for 6 days either alone or associated with ARA-C. A total of 112 patients less than 60 years of age received ARA-C (200 mg/m²/d) for 7 days combined with daunorubicin (45 mg/m²/d). The other 45 patients younger than 60 years of age were treated with ARA-C (100 mg/m²/d ) for 7 days and etoposide (100 mg/m²/d) for 5 days plus daunorubicin (50 mg/m²/d) or idarubicine (10 mg/m²/d) for 3 days. All patients who achieved complete remission (CR) received a postinduction therapy according to the respective protocols.20-24 CR was defined by bone marrow (BM) with less than 5% blasts and normal restored hematopoiesis. Remission failures were divided into two categories as previously described by Preisler²⁵: (1) early death, ie, patients who died of infection or hemorrhage within 7 days after the end of chemotherapy or who died while hypoplastic; and (2) resistant disease, ie, patients who survived for 13 days or more and whose leukemia cells were continuously present in their BM or whose leukemia cells were cleared from their BM during or after therapy but reappeared before the resumption of normal hematopoiesis.

Definition of AML-M0.Five cases fit the FAB criteria for the diagnosis of AML-M0 (less than 3% of blasts positive for MPO and/or SBB, positivity for myeloid-associated markers, and lack of B/T lineage-associated antigens).¹⁶ In the remaining 20 cases, the expression of surface lymphoid antigens was not regarded as an argument against the diagnosis of AML-M0, based on the positivity for anti-MPO and negativity for cCD3 and cCD22. The contemporary expression of anti-MPO and cCD3 or cCD22 identified the so-called “genuine biphenotypic” leukemia, which is likely to be a discrete entity26 and, therefore, is not included in the present study.

Morphology and cytochemistry.BM and peripheral blood were stained with Wright-Giemsa for morphology and differential. Cytochemical stains included MPO, SBB, naphtol-chloroacetate esterase (NASDCAE), α-naphtyl-acetate esterase (αANAE), αANAE inhibition with sodium fluoride, α-naphtyl-butyrate esterase (αNBE), periodic acid-Schiff reaction (PAS), and acid phosphatase. Cytochemical localization of MPO has involved the use of p-phenylenediamine and catechol as an indicator system.27 

Immunophenotyping.The membrane antigens were detected by double/triple-color immunofluorescence assay combining phycoerthrin (PE)-, fluorescein isothiocyanate (FITC)-, and PerCP-conjugated monoclonal antibodies (MoAbs) including: CD33-PE, CD13-PE, CD14-PE, CD19-PE, and CD20-PE purchased from Coulter (Hialeah, FL); CD33-PE, CD13-PE, CD14-PE, CD34-PE/PerCP, CD15-FITC, CD7-FITC, CD3-FITC/PerCP, CD4-FITC, CD8-PE, CD2-PE, CD5-PE, CD10-FITC, CD19-FITC/PE, CD22-PE, CD20-FITC, CD11b-PE, CD11c-PE, HLA-DR–PerCP, Leu54-PE, CD61-FITC, and CD62-PE purchased from Becton Dickinson (Mountain View, CA); and CD41-FITC and CD42-FITC purchased from Immunotech (Marseille, France). Mononuclear cells from peripheral blood and/or BM samples were separated on Fycoll-Hypaque density gradient and washed 3 times in phosphate-buffered saline (PBS) solution. The samples contained ≥80% of blasts and were analyzed within 24 hours from collection or were cryopreserved with 10% fetal calf serum and 10% dimethyl sulfoxide and analyzed later after quick thawing. Preliminary studies showed no difference in surface, cytoplasmic, and nuclear antigen expression before and after the cryopreservation (data not shown). The viability, as assessed by Trypan-blue exclusion, was greater than 95%. Before antibody staining, cells were incubated with human AB serum (100 μL/mL) for 10 minutes at room temperature to minimize the FC receptor binding. After a further washing, the cells were resuspended in PBS at a final concentration of 10 × 10⁷/mL. Each MoAb was incubated with 10⁶ cells in 100-μL volume. Optimal MoAb concentration was previously established in a series of titration experiments. Isotype-matched Ig was used as negative control. After 10 minutes of incubation at room temperature, the cells were washed 3 times in PBS with 0.2% azide and 0.2% bovine serum albumin (PBSA) and were analyzed with a flow cytometer. Either an Epics Profile (Coulter) or a FACScan (Becton Dickinson) were used to process the samples. For each sample, a minimum of 10,000 events were acquired, and, based on forward-light/orthogonal-light scatter properties of the leukemic population, a blast gate was used to determine the percentage of positive cells. To reduce contamination of residual nonleukemic myeloid cells, a positive reaction was considered when at least 20% of gated cells were more fluorescent than the negative control. To investigate cytoplasmic/nuclear antigens, the following reagents were also used: CD3, anti-MPO (αMPO), polyclonal TdT (supplied by Sera-LAB, Sussex, UK), and CD22 (obtained from Becton Dickinson). To visualize TdT, cytoplasmic CD3 and CD22 and αMPO cytocentrifuge preparations were fixed in cold methanol for 30 minutes (nuclear staining) or in acetone for 5 to 10 minutes at room temperature (cytoplasmic staining). After incubation with the primary MoAb and the second layer for 30 minutes in a moist chamber, the slides were washed in PBS and mounted with glycerol:PBS (1:1) and a coverslip. αMPO was also shown by using the alkaline phosphatase anti-alkaline phosphatase method as previously described.28 Slides were counterstained with hematoxylin Gill solution (Sigma Diagnostics, St Louis, MO). A total of 193 samples were also investigated for the expression of the multidrug-resistance P-glycoprotein (GP-170; C219-FITC obtained from Centocor, Malvern, PA). The MoAb against this protein recognizes an epitope on the inner surface of the cell membrane; accordingly, the blasts were fixed and permeabilized in 3% paraformaldehyde/PBS and 50% cold acetone/PBS. Samples were then incubated at 4°C for 30 minutes with 10 μL of C219-FITC solution and were washed 3 times in PBSA. Analysis was performed by flow cytometry as mentioned above. Given the heterogeneous quantitative and qualitative expression of GP-170, in accord with Campos et al,29 the threshold of positivity was set to a conventional 20%.

Cytogenetic analysis. Procedures for cytogenetic evaluation have been previously described in detail.30 Briefly, both a methotrexate cell synchronization technique and a direct preparation were performed. Chromosomes were examined with conventional Giemsa stain. Whenever possible, at least 20 mitoses were studied. Karyotypes were classified in agreement with the guidelines of the International System for Human Cytogenetic nomenclature.³¹ The detection of a minimum of two mitoses with an identical rearrangement or extra chromosome was regarded as evidence of the existence of an abnormal clone. Finally, complex karyotypes were defined by the presence of three or more events of translocation and nondisjunction in the same clone32 or by the presence of multiple unrelated clones.

Fluorescence in situ hybridization (FISH) analysis.Of 25 cases of AML-M0, 11 were investigated by the FISH technique for the presence of bcr/abl rearrangement. We used commercially available BCR-ABL double-color probes for hybridization and kits for the detection and amplification (biotin/fluorescein and digoxigenin/rhodamine) steps (both purchased from Oncor, Inc, Gaithersburg, MD). The cytogenetic pellet was stored at 4°C in fixative solution (ethanol to acetic acid, 3:1) until spreading on cleaned slides. The slides were aged at 37°C overnight before pepsin digestion (50 μg/mL in H₂0 [pH, 2.3] for 8 minutes at 37°C). Then, the slides were fixed in formaldeyde/PBS-MgCl₂ solution for 10 minutes and dehydrated in a series of ethanol solution (70%, 80%, and 95%). The denaturation step was performed in 70% formamide/2× standard saline citrate (SSC) at 70°C to 73°C for 2 minutes and 30 seconds. After this step, the slides were immediately dipped in a series of ice-cold ethanol (70%, 80%, 95% and 100%) for 3 minutes each. After drying on a slide warmer, 10 μL of ready-to-use probe mixture was dropped on a marked area of the slide and covered with a 25×25 coverslide. Hybridization was performed overnight at 37°C in a humidified chamber. Posthybridization washing was performed at 45°C in 50% formamide/2× SSC for 15 minutes and then at 37°C in 2× SSC for 8 minutes. Detection and amplification steps were performed as suggested by the manufacturer but with longer incubation times (20 minutes for biotinylated probes and 35 minutes for digoxigenated probes). The slides, after DAPI counterstaining, were analyzed with a Zeiss Axioskop (Zeiss, New York, NY) equipped with a triple-band pass filter (Oncor Inc). We considered evaluable only those cells showing either four signals (two red spots for the BCR gene and two green spots for the ABL gene), or three signals when the BCR-ABL gene juxtaposition yielded a single yellow spot. We considered as BCR-ABL⁺ cells showing a yellow spot (most common) as well as those cells with a green and a red spot linked together (a very rare finding). Sometimes, signals were split into doublets because of DNA replication.

Statistical analysis.Relationship of marker reactivity (considered as positive of negative) to patient's characteristics and response to treatment were estimated by Fisher's exact test. A P value of .05 or less was considered to be significant. Because age is the most relevant prognostic factor in AML, P values were computed with age strata to correct for any dependence on age of the prognostic factor. The Kaplan-Meier product-limit was used for estimation of survival (SV) and continuous CR (CCR) duration. SV was calculated from the date of diagnosis to the date of death or last follow-up examination. CCR was measured from achievement of CR until relapse. Patients undergoing BM transplantation procedures were censored at the time of BM infusion. For comparison of survival and remission duration of two or more groups, the log-rank test was applied.

Morphology and cytochemistry.Of 272 cases investigated, 25 (9%) met the FAB diagnostic criteria for AML-M0. In all 25 cases, BM was hypercellular with blast cells typically L2-shaped. Leukemic cells also presented agranular basophilic cytoplasm, high nucleocytoplasmic ratio, finely dispersed chromatine, and prominent nucleoli. No AML-M0 case had dimorphic blast population or showed Auer rods. In 2 cases, 40% to 50% of the blasts were heavily vacuolated and monocytoid-shaped. Trilineage displasia was not observed in any case because of the high degree of BM blast infiltration (≥80%). All cases had less than 3% MPO/SBB⁺ blasts. NASDCAE was negative in all cases, whereas periodic acid-Schiff reaction showed a positive block pattern in 4 cases. Five cases were faintly αNBE⁺ and αANAE⁺ (sodium fluoride-resistant) in no more than 30% to 40% of blasts.

Immunophenotyping.All cases were studied with a large panel of MoAbs with the exception of 2 AML-M6 and 1 AML-M7, which had too low of a cellularity. Phenotypic diagnosis of AML-M7 required confirmation by testing platelet peroxidase on BM biopsy tissue specimens by electron microscopy. Stem cell antigen CD34 was strongly associated with AML-M0; 96% of M0 cases were CD34⁺, whereas the percentage of its positivity was significantly lower for the other FAB groups (M1, 79%; M2, 71%; M3, 14%; M4, 71%; M5, 47%; P = .001 for all comparisons). HLA-DR was similarly expressed in all FAB groups, with the exception of M3 (Tables 1 and 2).

All AML-M0 cases were αMPO⁺. Over the same period of study, 70 acute lymphoblastic leukemia cases were also observed, and all but 2 showed negativity for αMPO; the latter were excluded from the analysis because they were considered true biphenotypic leukemias (simultaneous expression of αMPO and cCD3).

CD13 and CD33 were expressed in 68% (17 of 25 cases) and 64% (16 of 25 cases) of cases, respectively. Notably, 4 cases (16%) expressed neither CD13 nor CD33, with CD15 being the only surface myeloid-associated antigen expressed in 1 case. Conversely, only 1 (2%) of the other 244 cases (AML-M1) lacked both CD13 and CD33. As expected, CD14 was expressed in most cases of AML-M4/M5 (90% and 86%, respectively), but also was expressed in other FAB classes including M0. CD15 was significantly associated with FAB M2/M4/M5 (81%, 76%, and 72%, respectively); however, again, it was not a distinctive aspect, being found in the remaining FAB classes as well. CD11b and CD11c were positive in 45% (9 cases) and 28% (5 cases) of AML-M0, respectively, and were also strictly linked to FAB-M4/M5. Finally, C-kit appeared to be equally distributed among FAB subtypes, with the exception of M3 and M5, which showed the lowest frequency of expression (14% and 33%, respectively; Tables 1 and 2).

Despite FAB guidelines that require lymphoid antigens to be negative for a diagnosis of AML-M0, 20 (80%) of the 25 M0 cases expressed lymphoid-associated antigens. In all cases, multiparametric analysis confirmed the coexpression of lymphoid and myeloid antigens on the same cells rather than the existence of different leukemic populations within the “blast gate.” Indeed, the expression of lymphoid-associated antigens occurred more frequently in AML-M0 as compared with that in the remaining FAB subtypes (P = .002 for all comparisons). TdT was more frequently expressed in AML-M0 and M1 (P = .001); CD4 was more frequently expressed in AML-M0, M4, and M5 (P = .014); and CD7 was more frequently expressed in AML-M0, M1, and M5 (P = .003). CD2, CD5, CD10, and CD19 were similarly distributed among the different FAB classes without any significant correlations. We also analyzed the concurrent expression of two or more lymphoid-associated antigens, and it was more frequently observed in AML-M0 and AML-M5 than in the other FAB subsets (P = .002; Table 3). Even the coexpression of TdT and other lymphoid-associated antigens was associated with AML-M0 (P = .032; Tables 1, 2 and 3).

GP-170 tested positive in 37% of the AML-M0 samples evaluated. Surprisingly, this incidence was the lowest along with that of APL (20%); in fact, AML-M1, M2, M4, and M5 showed a superior level of GP-170 positivity (P = .02; Tables 1 and 2).

Cytogenetics.As previously shown by us and others,1,4,5,7-9,16 32-39 AML-M0 stands out for the expression of a high incidence of abnormalities such as complex karyotypes, −5/5q, −7/7q, +8, and +13. In this study, we aimed at determining the frequency of this feature in AML-M0 as compared with that in other FAB classes. APL cases were not included because all had the t(15; 17) and/or the rearrangement of PML/RARα. In contrast to 16% (P = .003) among other cases, 53% of AML-M0 cases expressed complex karyotypes (P = .005). Anomalies of chromosome 5 and/or 7 were noted in a similar fashion among the various FAB classes. Trisomy 13 was reported in 7% (1 of 15) of AML-M0 and 4% (1 of 27) of AML-M1 cases; trisomy 4 was detected in 13% (2 of 15) of AML-M0, 3% (1 of 34) of AML-M2, and 5% (1 of 20) of AML-M4 cases. Trisomy 8 occurred in 13% (2 of 15) of AML-M0, 15% (4 of 27) of AML-M1, 3% (1 of 34) of AML-M2, and 10% (2 of 20) of AML-M4 cases. Notably, 80% of anomalies of chromosome 5 and/or 7 in AML-M0 cases were included within a complex karyotype, whereas this pattern was detected in only 13% of other FAB groups. In 31% of AML-M0 cases, TdT and complex karyotypes were associated; the same association occurred in 5% of AML-M2 and 9% of AML-M4 cases. In 26% of AML-M0 cases, TdT was combined with aberrations of chromosome 5 and/or 7; AML-M1, M4, and M5 cases presented a similar pattern in 9%, 3%, and 10% of the cases, respectively. The triple combination of TdT, complex karyotypes, and abnormalities of chromosome 5 and/or 7 was a finding exclusively belonging to AML-M0, being undetectable in any other FAB subset (Tables 2 and 4). No evidence of the presence of bcr/abl rearrangement was detected by FISH analysis in 11 of 25 cases of AML-M0 examined.

Clinical findings.Patients with AML-M0 did not have unusual clinical features, abnormalities on physical examination or routine laboratory parameters. The median age was 60 years (range, 27 to 81), and 13 patients were women and 12 were men. The presenting median white blood cell count was 27.5 × 10⁹/L (range, 0.6 to 185 × 10⁹/L). The overall CR rate for all FAB subtypes except for M0 was 56%, with a median SV ± SEM and CCR ± SEM of 21 ± 4 and 34 ± 6 weeks, respectively. AML-M0 had a consistently shorter duration of SV and CR (Fig 1). The highly significant P value (<.001) with regard to SV and CCR persisted even after removing the APL cases from the analysis; this removal of the APL cases was performed to avoid an overestimation of the poor SV and CCR of M0 because of the excellent outcome associated with APL. Of the 25 patients with AML-M0, 7 (28%; P = .016) achieved a CR; the median durations of SV and CR were 12 ± 3 and 27 ± 4 weeks, respectively. One patient died in CR because of complications of surgery for excision of pulmonary aspergillosis; the other patient underwent allogeneic BM transplantation. Five patients had a relapse of disease, and no second remissions were obtained with salvage therapy. Interestingly, in 2 of 5 patients with relapse, the leukemia reappeared with typical M4/M5 features; the same result was reported in 4 refractory patients who, after chemotherapy, experienced a short aplasia followed by the resumption of leukemia with myelomonocytic aspects. In examining the presenting phenotypic and karyotypic characteristics of these cases, we found that 1 of them expressed CD4; 1 coexpressed CD4 and CD11b; 1 carried CD14, CD11b, and CD11c; 1 bore both CD4 and CD14; 1 showed CD14 and CD11c; and, finally, the last one had a t(6; 11)(q15;q23), which is known to be strongly associated with monoblastic/monocytic leukemia.39 In all cases, CD14 and CD4 were dimly expressed. This pattern of positivity was regarded as very unusual to be consistent with an initial diagnosis of M4/M5 leukemia that normally much more brightly expresses CD14 and CD4.

Reports of treatment of AML-M0 are limited, and adequate therapy has not yet been defined. In general, consensus exists that conventional chemotherapy yields disappointing results.1,5-9 Some factors that may be responsible for this unfavorable outcome include the high frequency of chromosome abnormalities, the cellular immaturity as indicated by the high expression of CD34 and CD7, and the frequent expression of multidrug resistance-associated GP-170.4,5-8 In this study, we determined the incidence of those factors in AML-M0 and in a large series of AML with other morphology. We observed that αMPO is a sensitive and reliable reagent in distinguishing AML-M0; this conclusion is reinforced by the evidence that, in 16% of our cases, CD13 and CD33 were both negative, whereas only in 1 case of AML-M1 was the concurrent absence of CD13 and CD33 observed. In cases with αMPO positivity by electron microscopy or immunoassay and with cCD3/cCD22 negativity, the expression of lymphoid-associated antigens should not rule out a diagnosis of AML-M0. In fact, 80% of the AML-M0 in our series expressed lymphoid markers, and 48% were positive for two or more of them. CD2, CD5, CD7, CD10, and CD19 were expressed in a similar fashion among the different FAB classes. TdT expression in AML was tightly linked to AML-M0; 68% of these cases were TdT⁺, and the association was statistically significant when compared with the other FAB subsets (P < .001). This finding is in line with previous reports documenting a higher incidence of TdT in poorly differentiated AML.4-6,8,40-46 In fact the inappropriate expression of TdT suggests the engagement of a multipotent stem cell retaining multiple differentiation antigens or reflects a misprogramming of the differentiation pathway leading to “lineage infidelity.” TdT also showed a significant prognostic importance. In fact, in a multivariate analysis, it emerged as an independent variable influencing achievement of CR (P = .031). CD4 was frequently expressed on blasts from AML-M0, M4, and M5. We can speculate that CD4 positivity is a “stem cell” feature, a notion supported by consistent expression of CD4 in immature AML^45-47 and on CD34⁺ hematopoietic progenitors.48 Alternatively, it may be that certain cases of AML-M0 are very immature monoblastic proliferations. This view is also suggested by the fact that 3 of 6 cases having a relapse or leukemic regrowth as M4/M5 forms, were weakly CD4⁺ at presentation; in 1 case, CD4 was combined with CD14, and, in another one, CD4 was combined with CD11b. Two other cases expressed CD14 along with CD11b and/or CD11c, and, in another case, a possible monocytic nature was suggested by the presence of t(6; 11)(q15;q23).

The incidence of expression of C-kit appeared lower in AML-M0 than in some more mature morphological AML phenotypes. It may be explained by the low number of AML-M0 samples tested in our series. On the other hand, the role of C-kit in defining more immature leukemias still remains a subject for debate. Some investigators point out the heterogeneity of C-kit expression among different FAB subtypes.49 In fact, immunophenotypical analysis showed no restriction of C-kit to immature, CD34⁺ leukemias; however, C-kit was also expressed on CD4⁺, CD34⁻ leukemic cells differentiating towards the monocyte lineage.49 

GP-170 was heterogeneously distributed throughout the FAB classes; however, in spite of previous demonstrations of an association between GP-170 and stem cell disease, such as MDSs,50 the lowest frequency of GP-170 was in AML-M0 and AML-M3. This observation led us to reconsider the role of GP-170 in determining chemoresistance of AML-M0, and future biological studies should concentrate on alternative mechanisms of drug resistance in immature leukemias.

AML-M0 was characterized by a more frequent occurrence of complex karyotypes. Moreover, there was a convergence, in approximately 20% of cases, of TdT positivity, complex karyotype, and abnormalities of chromosome 5 and/or 7. This pattern was not observed among the other FAB subsets. In addition, 80% of abnormalities of chromosome 5 and/or 7 were comprised in complex karyotypes, whereas only 13% of the remaining FAB cases carried this feature.

A close resemblance between AML-M0 and MDS, as also proposed by others,7 became evident even from our karyotypic analysis. In a recent study of karyotype in 179 cases of MDS, Haase et al51 reported an overall incidence of complex karyotypes of 9.5% (17 of 179); such changes were observed in 16%, 21%, and 56% of patients with refractory anemia with excessive blasts (RAEB), RAEB in transformation (RAEB-t), and secondary MDS. Eleven (65%) of them carried as the primary lesion a 5q− deletion, and, also, abnormalities of choromosome 7 were frequently present. Notably, pathologies of chromosome 5 and/or 7 have been shown in the very immature CD34⁺/CD38⁻ stem-cell–like population in AML.52 In addition, cell clones bearing these abnormalities may be prone to a preferential stimulation by cytokines and to a subsequent acquisition of growth advantage.^53-56 All these defects could induce karyotype instability that, in turn, represents the basis for the emergence and accumulation of additional chromosome breakages, ultimately defining the “complex karyotype.”

In conclusion, the results generated from our analysis show a possible “stem cell” pattern that is more frequently associated with, but not restricted to, AML-M0 and that is defined by the combination of CD34, TdT, complex karyotypes, and anomalies of chromosome 5 and/or 7. This pattern may account for the poor prognosis of AML-M0 and also indicates that the positivity of markers such as CD7 and GP-170 does not justify, in itself, the inferior prognosis of this particular variant of leukemia. However, although AML-M0, because of its dismal outcome, can be clinically considered a “discrete entity,” from a biological point of view, we still lack specific markers supporting this assumption. Therefore, future studies, especially molecular ones, should aim at filling this gap by showing the connection between clinical behavior and unique biological features in AML-M0.

The authors wish to thank Dr D. Campana (St Jude Children's Research Hospital, Memphis, TN) for his critical reading of the manuscript and his helpful suggestions.