Adult Acute Erythroleukemia: an Analysis of 91 Patients Treated at a Single Institution

Adult acute erythroleukemia: an analysis of 91 patients treated at a single institution

FPS Santos1, S Faderl1, G Garcia-Manero1, C Koller1, M Beran1, S O’Brien1, S Pierce1, EJ Freireich1, X Huang2, G Borthakur1, C Bueso-Ramos3, M de Lima4, M Keating1, J Cortes1, H Kantarjian1 and F Ravandi1

1Department of Leukemia, University of Texas - M. D. Anderson Cancer Center, Houston, TX, USA; 2Department of Biostatistics, University of Texas - M. D. Anderson Cancer Center, Houston, TX, USA; 3Department of Hematopathology, University of Texas - M. D. Anderson Cancer Center, Houston, TX, USA and 4Department of Stem Cell Transplantation, University of Texas - M. D. Anderson Cancer Center, Houston, TX, USA

Acute erythroleukemia (AML-M6) is an uncommon subtype of erythroid (AML-M6a), but with the lower blast threshold of 20%) acute myeloid leukemia (AML); it is considered to have a poor plus the entity known as pure erythroleukemia (AML-M6b), prognosis. From 1 January 1980 to 21 May 2008, 91 patients with newly diagnosed AML-M6 were seen at the University of when more than 80% of BM cells are immature cells (with an Texas–M.D. Anderson Cancer Center (UT–MDACC). Forty-ﬁve undifferentiated or pro-normoblastic appearance) committed 8 patients (50%) had a history of myelodysplatic syndrome exclusively to the erythroid lineage. (MDS), compared with 41% in our control group (patients with AML-M6 has been considered to be a subtype of AML with other AML subtypes) (P ¼ 0.08). Poor-risk cytogenetics were a worse prognosis.9–13 It is very uncommon in children,14,15 and more common in patients with AML-M6 (61% versus 38%, it corresponds to 3–5% of adult AML cases.12 Familial and P ¼ 0.001). Complete remission rates were 62% for patients with congenital forms have been described.16,17 It is often associated AML-M6, comparing with 58% for the control group (P ¼ 0.35). Median disease free survival (DFS) for patients with AML-M6 with an antecedent diagnosis of myelodysplastic syndrome was 32 weeks, versus 49 weeks for the control group (P ¼ 0.05). (MDS) and the presence of poor-risk chromosome abnormal- Median overall survival (OS) of patients with AML-M6 was 36 ities.9,18–20 Owing to its rarity, the clinical experience with weeks, compared with 43 weeks for the control group (P ¼ 0.60). AML-M6 is limited. It is also not clear if the pathological On multivariate analysis for DFS and OS, AML-M6 was not an diagnosis of AML-M6 is associated with a worse prognosis or if independent risk factor. AML-M6 is commonly associated with a previous diagnosis of MDS and poor-risk karyotype. The the worse prognosis reported with this leukemia is because of its diagnosis of AML-M6 does not impart by itself a worse frequent association with poor-risk karyotype and other poor- prognosis, and treatment decisions on this disease should be risk features. guided by well known AML prognostic factors. The objective of this study is to describe the clinical Leukemia (2009) 23, 2275–2280; doi:10.1038/leu.2009.181; and biological features of patients with AML-M6 treated at published online 10 September 2009 the University of Texas–M.D. Anderson Cancer Center Keywords: acute erythroleukemia; clinical features; prognosis (UT–MDACC) and to compare them with patients with non- M6 AML. We also evaluated the impact of a diagnosis of AML-M6 in response rate and survival independent of other Introduction risk factors.

Acute erythroleukemia (AML-M6) was first described in 1912 1 2 by Copelli, but it was Di Guglielmo who coined the term Patients and methods ‘eritroleucemia’, describing an abnormal proliferation of ery- 3 throid cells, myeloblasts and megakaryocytes. In 1923 Di Patients Guglielmo4 described a case of pure erythroleukemia char- Adult patients with newly diagnosed AML-M6 seen at acterized by pure proliferation of erythroid immature elements, UT–MDACC between 1 January 1980 and 21 May 2008 were akin to acute leukemia. In 1953 Dameshek proposed the term compared with patients with other AML subtypes, excluding ‘Di Guglielmo’s Syndrome’, which described three phases of those with acute promyelocytic leukemia. Charts were manually alterations in the bone marrow (BM): an erythremic phase, an reviewed. Patients were treated on studies conducted at erythromyeloblastic phase and a myeloblastic phase, similar to UT–MDACC. Studies were approved by the Institutional Review acute myeloid leukemia (AML).5 Board and conducted in accordance with the Declaration of The French–American–British classification of AML in 1976 Helsinki. All patients provided written informed consent before established the first set of criteria for the diagnosis of AML-M6.6 study entry. A diagnosis of AML-M6a (mixed erythroid/myeloid) A revision of the French–American–British classification defined was considered if more than 50% of the total nucleate BM cells AML-M6 as the presence of more than 50% erythroid cells with were of erythroid lineage and more than 20% of non-erythroid more than 30% blasts among non-erythroid BM nucleated 8 cells were blasts. AML-M6b (pure erythroleukemia) was cells.7 The World Health Organization (WHO) classification diagnosed in patients who presented with more than 80% of recognizes the French–American–British entity (mixed myeloid/ 8 BM nucleate cells consisting of immature erythroid precursors. On the basis of previous reports of similar cases we also Correspondence: Dr F Ravandi, Department of Leukemia, University considered a diagnosis of AML-M6b in patients who presented of Texas - M.D. Anderson Cancer Center, 1515 Holcombe Boulevard, with acute undifferentiated leukemia with blasts that were Unit 428, Houston, TX 77030, USA. E-mail: [email protected] positive by immunophenotyping for glycophorin A, even though 21–24 Received 8 April 2009; revised 14 July 2009; accepted 21 July 2009; these cases did not fit the current WHO criteria. Also of published online 10 September 2009 importance, no patient had received erythropoietin at the time Adult acute erythroleukemia FPS Santos et al 2276 of BM aspiration for the diagnosis to be confirmed. Pathology remaining 30 patients, 5 patients had diagnostic criteria for reports of BM biopsies were reviewed for information about AML-M6b (more than 80% of BM cells consisted of immature dysplasia in one or more hematopoietic lineages. Cytogenetic erythroid precursors) and 8 patients did not fulfill the criteria for data was stratified into three subgroups (modified from previous AML-M6b but presented with undifferentiated leukemia with reports25–27): Poor-risk (À5, À7, þ 8 and 11q involvement and blasts positive for glycophorin A. The remaining 17 patients complex karyotype), good-risk (t(8;21) or Inv(16) t(16;16)) and were diagnosed as having AML-M6 in our institution, on the intermediate-risk (neither good nor poor ). Patients received basis of morphology and/or positivity for periodic acid Schiff different treatment regimens according to the period of diagnosis (PAS) stain, but as they did not fulfill the current criteria for and prevailing studies. The treatment regimens were divided this disease, they were excluded from this analysis. All these 17 into: Group 1, regimens with cytarabine (ara-C) and anthra- patients were diagnosed before the current WHO criteria were cyclines. Group 2, regimens with ara-C and fludarabine without defined. Thus, 91 patients remained for our analysis, corre- anthracyclines. Group 3, regimens with ara-C plus topotecan. sponding to 3.0% of all patients with AML. There was a higher Group 4, other ara-C containing regimens. Group 5, miscella- proportion of patients with AML-M6 diagnosed in the decades neous non-ara-C containing intensive chemotherapy regimens. of 1990 and 2000, probably a reflection of the advances in immunophenotyping and establishment of diagnostic criteria for this entity: there were 6 (1%) of 612 patients during 1980–1989, Criteria for response and survival, and statistical analysis 34 (3.1%) of 1087 patients during 1990–1999 and 51 (3.9%) of Complete remission (CR) was defined by the presence of o5% 1276 patients from 2000–2008 (P ¼ 0.002). This can also be an blasts in the BM with more than 1 Â 109/l neutrophils and more artifact because we excluded 17 patients who were diagnosed in than 100 Â 109/l platelets in the peripheral blood.28,29 A relapse the time period from 1980 until 1999, when the WHO criteria was defined by more than 5% blasts in a BM aspirate unrelated for AML-M6b were not established. The clinical features of to recovery or by the presence of extramedullary disease. patients with AML-M6 are summarized in Table 1. They were Disease-free survival (DFS) was calculated from the time of CR different from patients with other AML subtypes in a number of until relapse or death in CR. Event free survival (EFS) was characteristics. There was a male predominance (68% versus calculated from the beginning of treatment until an event. An 55%, P ¼ 0.02), a slightly higher prevalence of antecedent event was defined as relapse, resistant disease and death. hematological disorder (AHD)/MDS (50% versus 41%, P ¼ 0.08) Induction death was defined as death occurring before and more frequent finding of dysplastic changes in BM cells achievement of CR or confirming resistant disease. Overall (80% vs 50%, Po0.001). They also had, at presentation, lower survival (OS) was calculated from the time of diagnosis until hemoglobin levels (Po0.001), lower WBC count (Po0.001), death. All patients alive at last follow-up were censored. lower platelet count (Po0.001) and a higher prevalence of Categorical and continuous variables were compared by the poor-risk cytogenetic abnormalities (61% versus 38%, w2/Fischer’s exact test and Mann–Whitney U test, respectively.30 Po0.001). All variables significantly associated with AML-M6 The survival curves were estimated using Kaplan–Meier plots in Table 1 were fitted into a logistic regression model to predict and compared by the log-rank test.31 Multivariate logistic the diagnosis of AML-M6. As data was lacking in a substantial regression models were used to evaluate the associations proportion of patients for BM dysplasia and b2-microglobulin, between multiple characteristics and diagnosis of AML-M6 or these variables were not included into the model. In the final achievement of CR. Clinical and biological characteristics were model, the following covariates predicted for the diagnosis analyzed for their association with survival using multivariate of AML-M6: intermediate-risk karyotype (versus good-risk), Cox proportional hazards models.32 All the univariately poor-risk karyotype (versus good-risk), lower percentage of significant variables are included in a multivariate model, and BM blasts and lower WBC (Supplementary Table 1S). The year then stepwise variable selection was done. For white blood cell of diagnosis was not a predicting factor for the diagnosis count (WBC), platelets and lactate dehydrogenase (LDH), as of AML-M6. some patients had extremely large values, the variables were In all, 78 patients with AML-M6 (86%) had AML-M6a. The grouped into four quantiles. In stepwise selection, if any of the median number of erythroblasts was 60% (43–87%). One group variables was significant, then all the other groups were case had 43% erythroblasts on the BM aspirate, but sheets of included in the final model even though some variables may immature erythroblasts were seen on BM biopsy. The median have P-values larger than 0.05. The survival analysis for stem number of myeloid blasts (of non-erythroid cells) was cell transplantation (SCT) for AML-M6 patients was done with a 45% (range 19–90.3%). One case had 19% blasts seen on multivariate Cox proportional hazards model with transplant aspirate, but the number of immature myeloid cells was status used as a time-dependent covariate (predictor). This estimated to be 30% by the biopsy. Thirteen patients analysis automatically matches patients who received SCT with (14%) presented with AML-M6b. The median number of patients who did not receive SCT on the selected prognostic erythroid blasts was 60% (30–86.6%). Eleven patients factors, and requires each non-SCT match to live, at least, long were tested for glycophorin A; it was positive in 10 (91%). as the time to SCT for his/her pairmate. All computations were The PAS stain was carried out in 67 patients (AML-M6a and done in SAS (Cary, NC, USA) and Statistica, version 6.0 (Tulsa, AML-M6b) and was positive in 58 cases (87%). One patient was OK, USA). positive for CD36 and in one case the diagnosis was confirmed by electron microscopy. Among the 13 patients with a diagnosis of AML-M6b, 10 (77%) had concordance of diagnosis Results by morphology, cytochemistry and flow cytometry. In two patients the diagnosis was established on morphological Clinical and biological features grounds (480% erythroblasts) and cytochemistry (globular There were 2992 patients with newly diagnosed AML seen at PAS-positivity) (flow cytometry was not carried out in these the UT–MDACC during the specified period, and 108 were patients) and in one case the diagnosis was established based on diagnosed as having AML-M6. Among these 108 patients, 78 morphology and flow cytometry (glycophorin-A positivity), with had diagnostic criteria consistent with AML-M6a. Of the negative PAS.

Leukemia Adult acute erythroleukemia FPS Santos et al 2277 Table 1 Characteristics of AML-M6 versus AML-other patients Table 2 Response and survival data (excluding APL) Parameter Median (Range) or No. (%) P Characteristics Median (Range) or No. (%) P AML-M6 AML-other AML-M6 AML-other (n ¼ 91) (n ¼ 2884) CR rate 57 (62) 1666 (58) 0.35 DFS, weeks 32 (3–822+) 49 (0–1.281+) 0.05 Male Sex 62 (68) 1600 (55) 0.01 EFS, weeks 18 (0–831+) 20 (0–1.285+) 0.70 Age (years) 59 (17–85) 59 (14–89) 0.89 OS, weeks 36 (0–831+) 43 (0–1.285+) 0.61 PS 3–4 2 (2) 259 (9) 0.02 Prior AHD/MDS 45 (50) 1169 (41) 0.08 Abbreviations: AML, acute myeloid leukemia; AML-M6, acute ery- Prior chemotherapy 17 (19) 423 (15) 0.28 throleukemia; CR, complete remission; DFS, disease free survival; or XRT EFS, event free survival; OS, overall survival. WBC, Â 109/l 2.3 (0.5–45.1) 9.7 (0.2–433) o0.001 Hb, g/100 ml 7.7 (2.5–12.6) 8.3 (2–16.1) o0.001 Platelets, Â 109/l 42 (3–222) 49 (1–2.292) 0.02 1.0 BM blasts, % 19 (2–87) 53 (0–99) o0.001 b2-microglobulin, mg/l 2.2 (1.2–12.3) 2.8 (0–31.3) 0.001 LDH, IU/l 737 (226–25.980) 916 (66–58.586) 0.01 0.8 Total Deaths AML-Other 2884 2288 Cytogeneticsa o0.001 AML-M6 91 69 Poor 55 (61) 1042 (38) 0.6 Intermediate 35 (39) 1416 (52) p = 0.70 Favorable 0 (0) 255 (10) 0.4 BM dysplasiab 62 (80) 916 (50) o0.001 Protected environment 58 (64) 1713 (59) 0.4

Treatment Regimens 0.13 Overall Survival Probability 0.2 Group 1 47 (51) 1276 (44) Group 2 6 (7) 320 (12) Group 3 11 (12) 284 (10) 0.0 Group 4 17 (19) 788 (27) 0 200 400 600 800 1000 1200 1400 Group 5 10 (11) 216 (7) Time (Weeks) Abbreviations: AHD, antecedent hematological disorder; AML, acute 1.0 myeloid leukemia; AML-M6, acute erythroleukemia; APL, acute promyelocytic leukemia; BM, bone marrow; Chemo, chemotherapy; Hb, hemoglobin; LDH, lactate dehydrogenase; MDS, myelodysplastic Total Relapse/Death in CR syndrome; PB, peripheral blood; PS, performance status; WBC, white 0.8 blood cell count; XRT, radiotherapy. AML-Other 1666 1258 aCytogenetic data was available for 90 patients in the AML-M6 group AML-M6 57 41 and for 2713 patients in the AML-other group. 0.6 bBM dysplasia data was available for 77 patients in the AML-M6 group P = 0.05 and for 1830 patients in the AML-other group. 0.4

In all, 45 patients (50%) had a previous history of MDS or 0.2 other AHD. A minimum period of 4 weeks with abnormal blood Disease Free Survival Probability counts was needed for a diagnosis of AHD. Forty-one (45%) 0.0 patients had MDS and/or cytopenias, with a median duration of 0 200 400 600 800 1000 1200 1400 5 months (1–72) before the diagnosis of AML-M6. Three patients Time (Weeks) had a previous diagnosis of myeloproliferative disorder (one patient with polycythemia vera, one with essential thrombo- Figure 1 Kaplan–Meier curves for overall survival (OS) (a) and cythemia and one with both polycythemia vera and MDS), and disease free survival (DFS) (b) for patients with acute erythroleukemia (AML-M6) versus acute myeloid leukemia (AML)-other. Actuarial one patient had chronic lymphocytic leukemia as well as MDS. survival data is presented in Table 2. A total of 17 patients had previously received chemotherapy and/or radiation therapy, and two additional patients had a history of solid tumors, but had only received surgical treatment. Treatment response and survival Cytogenetic data was available for 90 patients (98%) (one Patients were treated using different regimens of intensive case had insufﬁcient metaphases). Cytogenetic abnormalities induction chemotherapy, according to the time of diagnosis and were detected in 61 patients (68%). The most common the prevailing protocols (Table 1). There was no signiﬁcant cytogenetic abnormality was a complex karyotype, present in difference in the proportions of patients with AML-M6 and non- 49 patients (54%), including 40 (44%) patients who presented M6 AML in the different treatment groups (P ¼ 0.45). with abnormalities of chromosomes 5 and/or 7. Isolated trisomy Response and survival outcomes for all patients are summar- 8 was present in 5.5% of cases. No patient with AML-M6 ized in Table 2 and Figure 1. The median follow-up for the presented with good-risk cytogenetics. There were four entire cohort is 321 weeks. The CR rate for AML-M6 was 62% patients (4.4%) with 19q13 abnormalities, including one trans- (versus 58% for non-M6 AML, P ¼ 0.35). In all, 18 patients with location t(11;19)(q11;q13.3). All of these patients had complex AML-M6 were refractory (20%) and 16 (18%) patients died at karyotypes. induction. All characteristics in Table 1 were evaluated for their

Leukemia Adult acute erythroleukemia FPS Santos et al 2278 Table 3 Multivariate analysis for disease free survival Table 4 Multivariate analysis for overall survival

Characteristics HR (95% CI) P Characteristics HR (95% CI) P

AML-M6 1.07 (0.78–1.47) 0.64 AML-M6 1.03 (0.81–1.32) 0.76 Age (years) 1.01 (1.00–1.015) o0.0001 Age 1.022 (1.020–1.024) o0.0001

Cytogenetics Cytogenetics Intermediate (versus favorable) 1.41 (1.20–1.66) o0.0001 Intermediate (versus favorable) 1.47 (1.28–1.68) o0.0001 Poor (versus favorable) 2.39 (2.00–2.86) o0.0001 Poor (versus favorable) 2.62 (2.26–3.02) o0.0001 Hb 0.93 (0.90–0.96) o0.0001 Hb 0.94 (0.92–0.96) o0.0001

Platelets (groups by quantiles) WBC (groups by quantiles) 27–49 versus o27 0.93 (0.81–1.06) 0.32 2.5–9.1 versus o2.5 0.99 (0.88–1.12) 0.93 49–89 versus o27 1.04 (0.91–1.19) 0.47 9.1–33.3 versus o2.5 1.15 (1.01–1.31) 0.02 489 versus o27 1.34 (1.11–1.63) 0.002 4 33.3 versus o2.5 1.24 (1.08–1.42) 0.002

PS42 1.34 (1.04–1.71) 0.02 LDH (groups by quantiles) AHD/MDS 1.21 (1.07–1.36) 0.001 570–904 versus o570 1.14 (1.00–1.30) 0.05 904–1732 versus o570 1.38 (1.21–1.58) o0.0001 Year of diagnosis 4 1732 versus o570 1.65 (1.42–1.92) o0.0001 1990–1999 (versus 1980–1989) 0.75 (0.64–0.86) 0.0001 2000–2008 (versus 1980–1989) 0.69 (0.59–0.81) o0.0001 PS 4 2 2.28 (1.99–2.62) o0.0001 Prior chemotherapy/radiation therapy 1.41 (1.25–1.57) o0.0001 Abbreviations: AHD, antecedent hematologic disorder; AML-M6, acute AHD/MDS 1.26 (1.16–1.37) o0.0001 erythroleukemia; CI, confidence interval; Hb, hemoglobin; HR, hazard ratio; MDS, myelodysplastic syndrome; PS, performance status. Treatment Group Group 2 (vs Group 1) 0.98 (0.85–1.14) 0.85 Group 3 (vs Group 1) 1.07 (0.93–1.23) 0.30 Group 4 (vs Group 1) 1.24 (1.11–1.37) o0.0001 association with achievement of CR by multivariate logistic Group 5 (vs Group 1) 1.49 (1.28–1.75) o0.0001 regression model. In the final model a diagnosis of AML-M6 was an independent factor for achievement of CR (Odds Ratio (OR) Abbreviations: AHD, antecedent hematologic disorder; AML-M6, acute erythroleukemia; CI, confidence interval; Hb, hemoglobin; HR, 1.82 (95% CI 1.11–2.98); P ¼ 0.01) (Supplementary Table 2S). hazard ratio; LDH, lactate dehydrogenase; MDS, myelodysplastic For those patients who achieved CR, the median DFS was 32 syndrome; PS, performance status; WBC, white blood cell count. weeks for patients with AML-M6 versus 49 weeks for non-M6 AML (P ¼ 0.05). Patients with AML-M6 tended to have a worse OS, but the difference was not statistically significant (P ¼ 0.61), of AML-M6 was not an independent predictive factor for worse with a median OS of 36 weeks for AML-M6 versus 43 weeks for DFS (HR 1.07; P ¼ 0.64). No treatment group was predictive of non-M6 AML. The median EFS was 18 weeks for AML-M6 worse DFS by multivariate analysis. versus 20 weeks for non-M6 AML (P ¼ 0.70). The following factors were associated with worse OS by Twenty-one patients with AML-M6 received a SCT. Their univariate analysis: diagnosis in the 1980s, older age, inter- median age was 50 years (versus 62 years for those who did not mediate-risk cytogenetics, poor-risk cytogenetics, lower hemo- receive an SCT, Po0.001). The stem cell sources were matched globin, WBC 433.3 Â 109/l, low platelets (o27 Â 109/l) and related donors (N ¼ 11), matched unrelated donors (N ¼ 8), high platelets (489 Â 109/l), high LDH (4904 IU/l), poor haploidentical related donor and cord blood unit in one patient performance status, history of AHD/MDS, history of previous each. Transplants were carried out in first CR (N ¼ 9), second CR chemotherapy and treatment group 5. Prognostic factors or first relapse (N ¼ 7) and refractory disease (N ¼ 4). Median associated with a worse OS by multivariate analysis are time to SCT was 25 weeks (range 13–86 weeks). The median OS summarized in Table 4. The diagnosis of AML-M6 was not an was 82 weeks (range 30–468 þ ) for AML-M6 patients who independent predictive factor for worse OS (HR 1.03; P ¼ 0.76). received an SCT versus 23 weeks (range 0–831 þ ) for AML-M6 patients who did not receive an SCT (P ¼ 0.001). Similar differences were seen for EFS (median 34 weeks (range 2– Discussion 323 þ ) versus 11 weeks (range 0–831 þ ), P ¼ 0.02). A Cox- proportional hazards model using transplant status as a time In this report we analyzed the clinical and laboratory dependent covariate did not show that transplantation was characteristics and survival outcomes of 91 patients with AML- an independent factor for OS in patients with AML-M6 M6 and compared them with other patients with AML diagnosed (hazard ratio (HR) 1.31 (95% CI 0.63–2.73); P ¼ 0.46) and treated during the same period. Patients were treated with (Supplementary table 3S). several different regimens, but there was no significant imbalance in treatment group distribution, and thus we did not stratify our results by therapy. There was no difference in CR Prognostic factors associated with survival rate, but in a multivariable analysis a diagnosis of AML-M6 was By univariate analysis, the following factors were associated with associated with a higher probability of achieving CR. This is worse DFS: older age, intermediate risk-cytogenetics, poor-risk surprising in view of the association between AML-M6 and other cytogenetics, lower hemoglobin, lower platelets, poor perfor- poor risk features that usually lead to lower response rates. mance status, history of AHD/MDS and previous chemotherapy/ However, because of its retrospective nature, these results must radiation therapy. Treatment group 2 was associated with better be considered with caution. In addition, most patients were DFS on univariate analysis. Prognostic factors associated with a treated with ara-C and received higher doses than what is worse DFS by multivariate analysis are in Table 3. The diagnosis conventionally used (1.5 g/m2 per day), and a high percentage

Leukemia Adult acute erythroleukemia FPS Santos et al 2279 Table 5 Selected reports of AML-M6

MDACC Colita et al.33 Olopade et al.11 Atkinson et al.19

Time Period 1980–2008 1976–1999 1969–1991 1977–1990 Number 91 54 26 15 Median age, years (range) 59 (17–85) 59 (19–84) 66 (6–85) 65 (16–85) Male, no. (%) 62 (68) 40 (74) 16 (62) 7 (47) AHD/MDS, no. (%) 45 (50) NR NR 3 (20) Treatment-related, no. (%) 17 (19) NR NR 4 (27) Median WBC, Â 109/l (range) 2.3 (0.5–45.1) 3.24 (0.4–367) 3.7 (0.8–10.6) 2.9 (1.2–6.2) Median Hb, g/100 ml (range) 7.7 (2.5–12.6) 8.5 (4.7–13.4) 6.5 (1.0–10.1) 7.2 (2.4–14.7) Median platelets, Â 109/L (range) 42 (3–222) 47 (6–334) 48 (12–819) 48 (3–259) Poor-risk karyotype, no. (%) 55 (61) 18 (33) 17 (65) 4/6 (66) CR rate, % 62 54 62 30 Median OS, months (range) 9 (0–208+) 9 (NR) 5.7 (0.5–73.5) 3.2 (1–84+) Abbreviations: AHD, antecedent hematologic disorder; AML-M6, acute erythroleukemia; CR, complete remission; Hb, hemoglobin; MDACC, M. D. Anderson Cancer Center; MDS, myelodysplastic syndrome; NR, not reported; OS, overall survival; WBC, white blood cell count. of patients with AML-M6 (64%) received induction chemo- represents a ‘true erythroleukemia’, where only erythroid cells therapy inside protected environment rooms; a higher are involved in the malignant clone.36 percentage than what is commonly found in other institutions. In our cohort we found 13 patients with a diagnosis of pure These limitations might affect the validity and general applic- erythroleukemia. Not all patients that were diagnosed as AML- ability of these results. M6b in our series fulfilled the WHO criteria,8 as 8 patients had Despite being associated with the achievement of CR, there less than 80% immature erythroblasts in the BM, but were was a worse DFS for patients with AML-M6 compared with nonetheless considered as having AML-M6b based on positivity other subtypes of AML. The EFS and OS were also shorter for for glycophorin A, a specific red cell marker, which can also be patients with AML-M6, but this was not statistically significant. negative in some cases of AML-M6.24 The erythroblast threshold However, our study may have been underpowered to detect of the WHO criteria for AML-M6b is arbitrary, and several cases such differences. In the multivariate analysis for DFS, the of AML-M6b may be missed by setting such a high threshold for diagnosis of AML-M6 was not an independent prognostic factor. diagnosis, as has already been reported by others.37 In our Thus, patients with AML-M6 had a worse prognosis due to the cohort, patients with AML-M6b commonly had dysplastic frequent association of this subtype of AML with other poor-risk features in the erythroid series and were usually positive features (such as monosomal karyotypes and history of MDS). for the PAS stain in blast cells. PAS is not specific for Other case series of adult AML-M6 are summarized in AML-M6, and can be positive in other leukemias including Table 5.11,19,33 Our series confirm some of the findings in acute lymphoblastic leukemia.38 However, the presence of previous reports. The presence of pancytopenia, with low coarse block-like PAS staining in blasts with absence of circulating blasts is typical of AML-M6, with very few patients other lineage-specific markers (such as MPO and TdT) is presenting with elevated WBC (only six patients in our series suggestive of AML-M6.19,37,39 had WBC more than 12 Â 109/l). No recurrent chromosomal The role of allogeneic SCT in AML-M6 has been the subject of abnormalities have been reported. Abnormalities of chromo- two reports which suggest that allogeneic SCT should be offered some 5 and/or 7 with a complex karyotype were the most to all patients with AML-M6 in first CR.40,41 In our report, common cytogenetic aberration detected in our series, as well patients with AML-M6 who underwent allogeneic SCT had as in previous reports.9,18,20,34 A recent study reported a high longer EFS and OS compared with patients who did not receive rate (10%) of chromosome 19 aberrations, in band 19q13.35 an allogeneic SCT. However, patients who received an Four patients in our series had abnormalities of chromosome 19 allogeneic SCT were younger, and this suggests possible at the same band, confirming this data. In our series we fitted a selection biases in these results. In a multivariate model multivariate logistic regression model to evaluate characteristics analyzing SCT as a time-dependent covariate there was no associated with a diagnosis of AML-M6. The OR for a diagnosis effect of transplant in OS. Nevertheless, allogeneic SCT should of AML-M6 decreased with an increase in WBC (OR 0.03 for be considered in patients with AML-M6 who present with other WBC 433.3 Â 109/l, P ¼ 0.0016) and increased in the presence poor-risk features. of poor-risk karyotype (OR 16.09, P ¼ 0.006). In conclusion, AML-M6 is a rare subtype of AML that is What can be gleaned from our study and previous reports is frequently related to MDS. Our results confirm the importance of that most cases of AML-M6 are strongly related to MDS, with classical AML prognostic factors in the outcome of patients with patients usually presenting with a previous history of MDS or AML-M6. The pathological diagnosis of AML-M6 does not with de novo AML with MDS-like clinical features, including impart by itself a worse outcome. Thus, even though the French– pancytopenia, unbalanced chromosomal abnormalities and American–British classification has historical significance, in dysplastic morphology. The increase in both myeloid blast cells general clinicians should refrain from making treatment decisions and erythroblasts in AML-M6a indicates the involvement of an based on morphological grounds alone for patients with AML-M6. early stem cell in the pathogenesis of this disease. Previous Instead, they should rely on cytogenetics and other prognostic reports have demonstrated multilineage involvement in cases features when deciding the best therapeutic approach. of AML-M6a, with the presence of a myelomonocytic and megakaryocytic component besides the erythroid component.34 Thus, it could be argued that AML-M6a is a clonal myeloid Conflict of interest disorder similar to MDS that has multilineage involvement and presents with a major erythroid component, whereas AML-M6b The authors declare no conflict of interest.

Leukemia Adult acute erythroleukemia FPS Santos et al 2280 References 22 Greaves MF, Sieff C, Edwards PA. Monoclonal antiglycophorin as a probe for erythroleukemias. Blood 1983; 61: 645–651. 1 Copelli M. Di una emopatia sistemizzata rapresentata da una 23 Shichishima T. Minimally differentiated erythroleukemia: recogni- imperplasia eritroblastica (eritronatosis). Pathologica 1912; 4: 460. tion of erythroid precursors and progenitors. Intern Med (Tokyo, 2 Di Guglielmo G. Richerche di ematologia. I. Un caso di Japan) 2000; 39: 761–762. eritroleucemia. Megacariociti in circolo e loro funzione piastrino- 24 Villeval JL, Cramer P, Lemoine F, Henri A, Bettaieb A, Bernaudin F poietico. Folia Medica (Pavia) 1917; 13: 386. et al. Phenotype of early erythroblastic leukemias. Blood 1986; 68: 3 Bain BJ. Di Guglielmo and his syndromes. Br J Haematol 2003; 1167–1174. 120: 939–943. 25 Grimwade D, Walker H, Oliver F, Wheatley K, Harrison C, 4 Di Guglielmo G. Eritremie acute. XXIX Congr Italiano Medical Harrison G et al. The importance of diagnostic cytogenetics on International; Roma, 1923. outcome in AML: analysis of 1,612 patients entered into the MRC 5 Dameshek W, Baldini M. The Di Guglielmo syndrome. Blood AML 10 trial. The Medical Research Council Adult and Children’s 1958; 13: 192–194. Leukaemia Working Parties. Blood 1998; 92: 2322–2333. 6 Bennett JM, Catovsky D, Daniel MT, Flandrin G, Galton DA, 26 Slovak ML, Kopecky KJ, Cassileth PA, Harrington DH, Theil KS, Gralnick HR et al. Proposals for the classification of the acute Mohamed A et al. Karyotypic analysis predicts outcome of leukaemias. French–American–British (FAB) co-operative group. preremission and postremission therapy in adult acute myeloid Br J Haematol 1976; 33: 451–458. leukemia: a Southwest Oncology Group/Eastern Cooperative 7 Bennett JM, Catovsky D, Daniel MT, Flandrin G, Galton DA, Oncology Group Study. Blood 2000; 96: 4075–4083. Gralnick HR et al. Proposed revised criteria for the classification of 27 Byrd JC, Mrozek K, Dodge RK, Carroll AJ, Edwards CG, Arthur DC acute myeloid leukemia. A report of the French–American–British et al. Pretreatment cytogenetic abnormalities are predictive of Cooperative Group. Ann Intern Med 1985; 103: 620–625. induction success, cumulative incidence of relapse, and overall 8 Brunning RD, Matutes E, Flandrin G, Vardiman JW, Bennett JM, survival in adult patients with de novo acute myeloid leukemia: Head DR et al. Acute myeloid leukemia not otherwise categorized. results from Cancer and Leukemia Group B (CALGB 8461). Blood In: Jaffe ES, Harris NL, Stein H, Vardiman JW (eds). World Health 2002; 100: 4325–4336. Organization Classification of Tumours: Pathology and Genetics of 28 Cheson BD, Cassileth PA, Head DR, Schiffer CA, Bennett JM, Tumours of Haematopoietic and Lymphoid Tissues. IARC Press: Bloomfield CD et al. Report of the National Cancer Institute- Lyon, 2001, pp 91–105. sponsored workshop on definitions of diagnosis and response in 9 Davey FR, Abraham Jr N, Brunetto VL, MacCallum JM, Nelson DA, acute myeloid leukemia. J Clin Oncol 1990; 8: 813–819. Ball ED et al. Morphologic characteristics of erythroleukemia 29 Cheson BD, Bennett JM, Kopecky KJ, Buchner T, Willman CL, (acute myeloid leukemia; FAB-M6): a CALGB study. Am J Hematol Estey EH et al. Revised recommendations of the International 1995; 49: 29–38. Working Group for Diagnosis, Standardization of Response 10 Peterson BA, Levine EG. Uncommon subtypes of acute nonlympho- Criteria, Treatment Outcomes, and Reporting Standards for cytic leukemia: clinical features and management of FAB M5, M6 Therapeutic Trials in Acute Myeloid Leukemia. J Clin Oncol and M7. Semin Oncol 1987; 14: 425–434. 2003; 21: 4642–4649. 11 Olopade OI, Thangavelu M, Larson RA, Mick R, Kowal-Vern A, 30 Snedecor G, Cochran W. Statistical Methods, 7th edn, Iowa State Schumacher HR et al. Clinical, morphologic, and cytogenetic University Press: Armes, IA, 1980. characteristics of 26 patients with acute erythroblastic leukemia. 31 Kaplan EL, Meire P. Nonparametric estimation from incomplete Blood 1992; 80: 2873–2882. observations. J Am Stat Assoc 1958; 53: 457–481. 12 Bennett JM, Begg CB. Eastern Cooperative Oncology Group study 32 Cox D. Regression models and life tables. J R Stat Soc 1972; 34: of the cytochemistry of adult acute myeloid leukemia by 187–202. correlation of subtypes with response and survival. Cancer Res 33 Colita A, Belhabri A, Chelghoum Y, Charrin C, Fiere D, Thomas X. 1981; 41: 4833–4837. Prognostic factors and treatment effects on survival in acute 13 Lowenberg B. Prognostic factors in acute myeloid leukaemia. Best myeloid leukemia of M6 subtype: a retrospective study of 54 cases. Pract Res 2001; 14: 65–75. Ann Oncol 2001; 12: 451–455. 14 Barnard DR, Alonzo TA, Gerbing RB, Lange B, Woods WG. 34 Cuneo A, Van Orshoven A, Michaux JL, Boogaerts M, Louwagie A, Comparison of childhood myelodysplastic syndrome, AML FAB Doyen C et al. Morphologic, immunologic and cytogenetic studies M6 or M7, CCG 2891: report from the Children’s Oncology in erythroleukaemia: evidence for multilineage involvement and Group. Pediatr Blood Cancer 2007; 49: 17–22. identification of two distinct cytogenetic-clinicopathological types. 15 Malkin D, Freedman MH. Childhood erythroleukemia: review of Br J Haematol 1990; 75: 346–354. clinical and biological features. Am J Pediatr Hematol Oncol 35 Cigudosa JC, Odero MD, Calasanz MJ, Sole F, Salido M, Arranz E 1989, Fall 11: 348–359. et al. De novo erythroleukemia chromosome features include 16 Lazure T, Beauchamp A, Croisille L, Ferlicot S, Feneux D, Fabre M. multiple rearrangements, with special involvement of chromosomes Congenital anerythremic erythroleukemia presenting as hepatic 11 and 19. Genes Chromosomes Cancer 2003; 36: 406–412. failure. Arch Pathol Lab Med 2003; 127: 1362–1365. 36 Park S, Picard F, Dreyfus F. Erythroleukemia: a need for a new 17 Novik Y, Marino P, Makower DF, Wiernik PH. Familial definition. Leukemia 2002; 16: 1399–1401. erythroleukemia: a distinct clinical and genetic type of familial 37 Domingo-Claros A, Larriba I, Rozman M, Irriguible D, Vallespi T, leukemias. Leuk Lymphoma 1998; 30: 395–401. Aventin A et al. Acute erythroid neoplastic proliferations. A 18 Nakamura H. Cytogenetic heterogeneity in erythroleukemia biological study based on 62 patients. Haematologica 2002; 87: defined as M6 by the French–American–British (FAB) Cooperative 148–153. Group criteria. Leukemia 1989; 3: 305–309. 38 Hayhoe FGJ, Quaglino D. Haematological Cytochemistry, 2nd 19 Atkinson J, Hrisinko MA, Weil SC. Erythroleukemia: a review of 15 edn, Churchill Livingstone: Edinburgh, 1988. cases meeting 1985 FAB criteria and survey of the literature. Blood 39 Frydecka I, Brodzka W, Lawinska B, Sciborski R. Clinical course Rev 1992; 6: 204–214. and results of treatment of erythroleukaemia. Folia Haematol Int 20 Lessard M, Struski S, Leymarie V, Flandrin G, Lafage-Pochitaloff M, Mag Klin Morphol Blutforsch 1984; 111: 283–289. Mozziconacci MJ et al. Cytogenetic study of 75 erythroleukemias. 40 Fouillard L, Labopin M, Gorin NC, Polge E, Prentice HG, Meloni G Cancer Genet Cytogenet 2005; 163: 113–122. et al. Hematopoietic stem cell transplantation for de novo 21 Garand R, Duchayne E, Blanchard D, Robillard N, Kuhlein E, erythroleukemia: a study of the European Group for Blood and Fenneteau O et al. Minimally differentiated erythroleukaemia Marrow Transplantation (EBMT). Blood 2002; 100: 3135–3140. (AML M6 ’variant’): a rare subset of AML distinct from AML M6. 41 Killick S, Matutes E, Powles RL, Min T, Treleaven JG, Rege KP et al. Groupe Francais d’Hematologie Cellulaire. Br J Haematol 1995; Acute erythroid leukemia (M6): outcome of bone marrow 90: 868–875. transplantation. Leuk Lymphoma 1999; 35: 99–107.

Supplementary Information accompanies the paper on the Leukemia website (http://www.nature.com/leu)

Leukemia