Randomized trial comparing standard vs sequential high-dose chemotherapy for inducing early CR in adult AML

In adult acute myelogenous leukemia (AML), intensive induction chemotherapy is a standard approach for achieving complete remission (CR),¹  as defined by an international panel of experts.²  CR enables the delivery of consolidation chemotherapy and/or allogeneic hematopoietic stem cell transplantation (HSCT), which can prolong survival and cure the disease.^1,3-5  Because CR is the starting point for curative treatment and a strong predictor of survival, it is important to optimize the induction chemotherapy protocol.

Several groups have attempted to improve the standard “3+7” CR induction regimen, which includes 3 days of daunorubicin and 7 days of conventional-dose cytarabine.⁶  Large randomized trials have tested high-dose (HD) daunorubicin (or equivalent idarubicin),^7-12  HD cytarabine,^13-17  and nucleoside analogs,^18-20  with variable results. HD cytarabine is a mainstay of postremission consolidation,²¹  and is widely used in relapsed/refractory AML. The use of HD cytarabine-based programs improved the rates of CR and relapse-free survival (RFS) in the Australasian Leukemia Study Group trial,¹³  as well as among patients aged 45 years or younger in the European Organization for Research and Treatment of Cancer-Gruppo Italiano Malattie Ematologiche dell’Adulto AML 12 study,¹⁷  but not in 3 other trials.^14,16,20 

The German AML Study Group (AMLSG) developed an effective sequential HD regimen for relapsed/refractory AML, which includes cytarabine plus either mitoxantrone or idarubicin.^22-25  Up-front use of this regimen with dose-dense HD cytarabine-mitoxantrone (HAM)–HAM double induction (cytarabine 3 g/m² and mitoxantrone) yielded 83% CR and 7% persistent leukemia.²⁶  A trial extension in the German collaborative intergroup project yielded 76% CR and 13% refractory AML. A parallel East Germany Hematology-Oncology Group trial (NCT01414231) compared continuous vs infusional sequential cytarabine (2 g/m²) and achieved 74% CR and 14% resistance, showing no improvement over standard ICE (idarubicin-cytarabine-etoposide) induction.²⁷ 

The Northern Italy Leukemia Group performed a phase 2 study using a sequential HD regimen for patients resistant to standard ICE induction (NCT00400673). Of 95 treated patients, 57% achieved CR, with similar rates across different clinico-cytogenetic risk groups.²⁸  Long-term outcome was significantly improved among patients who entered early CR during course 1 (ie, before receiving any HD chemotherapy).²⁹  Achieving CR after a single course is considered a favorable prognostic factor,^30-34  with rare exceptions.³⁵  In 1 large retrospective study (n = 8907), late CR achieved after reinduction therapy predicted significantly poorer outcome.³¹  Early responders are typically easier to manage, curable with shorter hospitalization and fewer complications, and can receive earlier postremission therapy and HSCT.

In the present randomized trial, we compared ICE with sequential HD schedule (sHD) in untreated patients, with the aim of improving the early CR rate and evaluating any favorable effect on survival.

All patients with newly diagnosed AML or high-risk (HR) myelodysplasia (MDS; marrow blasts > 10%) who presented at the participating institutions during the study period were registered for an outcome assessment project and screened for trial eligibility. Eligible patients had AML or HR MDS that was de novo, secondary to cancer chemo-radiotherapy, or developed in a background of MDS or chronic myeloproliferative neoplasm, excluding acute promyelocytic leukemia. Only patients at least 16 years old were included. Because of the high AML prevalence among elderly individuals, we did not initially set an upper age limit for study inclusion, but rather enrolled older patients who were deemed fit enough to receive study medications. Thus, this trial included patients older than 65 years until a study amendment set an age limit of 65 years. All included patients gave their written informed consent. The study was approved by the Institutional Review Boards of the 17 participating Northern Italy Leukemia Group sites, registered as NCT00495287, and conducted in accordance with the Declaration of Helsinki.

AML diagnosis was locally confirmed by cytogenetics and immunophenotype, following the standard criteria adopted for this study (supplemental File 1). Patient enrollment required central review of diagnostic bone marrow slides and trephine biopsy. FLT3,³⁶ NPM1,³⁷ CEBPA, and MLL mutations were investigated at the Laboratorio “Paolo Belli,” Department of Hematology, Bergamo Hospital, and the Institute of Hematology, Perugia University.

The study protocol (supplemental File 2), its background (supplemental File 3.1), and the trial amendments (supplemental File 3.2) are reported online. Amendment 3 limited enrollment to patients 65 years old or younger, as insufficient accrual of older patients (n = 35) led to inconclusive results in randomization 1. For CR induction (randomization 1), patients were stratified according to age (≤60 or >60 years) and were randomly assigned to undergo standard ICE or sHD. Patients who achieved CR underwent risk-adapted postremission therapy (Figure 1). Both induction groups received a cumulative idarubicin dose of 36 mg/m², previously defined in a dose-escalation study.³⁸  All patients received granulocyte colony-stimulating factor (G-CSF), which is safe and effective in reducing absolute neutropenia duration and related complications after chemotherapy.^39,40  In both study groups, G-CSF administration started on day 11 (ie, after the second sHD block). The current sHD regimen was patterned after the administration schedule of cytarabine 3 g/m² (more activity than 1 g/m²) developed by the AMLSG for relapsed/refractory AML.²³  To limit toxicity, particularly the risk for early death because of infection associated with cytarabine 3 g/m², patients in the sHD group received cytarabine 2 g/m² (1 g/m² if >65 years of age).²⁴  Adopting a 2-step induction strategy derived from the previous trial, patients who were unresponsive to course 1 received an augmented sHD course with cytarabine 3 g/m² (2 g/m² when >60 years of age) or other salvage therapy at the discretion of the treating physicians. Patients who did not achieve CR after 2 courses were removed from the study, as were patients who were assigned to HSCT as salvage instead of chemotherapy.

Patients who achieved CR were risk stratified according to cytogenetics,^41-43 FLT3 mutations, and selected clinical risk factors (HR MDS, AML secondary to MDS/myeloproliferative neoplasm, therapy-related AML, minimally differentiated AML, erythroleukemia, megakaryoblastic leukemia, primary isolated granulocytic sarcoma, white blood cell count, >50 × 10⁹/L, and late CR). The HR group comprised patients with unfavorable or unknown cytogenetics or with intermediate/normal cytogenetics plus either FLT3 and MLL mutations or another clinical risk feature. This group was scheduled to receive HSCT from siblings or unrelated donors. The standard-risk (SR) group comprised patients with favorable cytogenetics and those with intermediate/normal cytogenetics and lacking FLT3 mutations or any clinical risk factors. SR patients, along with HR patients unable to receive allogeneic HSCT, underwent randomization 2. Some patients received intravenous busulfan (BU; 0.8 mg/kg every 6 hours for 4 days) and cyclophosphamide (CY; 60 mg/kg daily for 2 days), followed by blood stem cell autotransplantation. Others underwent repetitive blood stem cell-supported HD chemotherapy courses (cumulative cytarabine 20 g/m² plus idarubicin 16 mg/m²). Blood stem cells were mobilized with intermediate-dose (ID) cytarabine and G-CSF. Patients who displayed poor mobilization (<2 × 10⁶/kg CD34⁺) and patients older than 65 years were consolidated with age-adapted ID chemotherapy courses (Figure 1) or were considered for HSCT despite SR status.

The primary study endpoint was whether sHD significantly reduced the risk for chemoresistance and increased the early CR rate compared with standard ICE chemotherapy. In the prior study, 22% of patients exhibited ICE-resistant AML (n = 129/581), of whom 57% were effectively rescued by sHD (n = 54/95)²⁸  (supplemental File 3.1). Extrapolating these data for the sample size calculation for the current study, we estimated that we needed at least 250 patients per group to demonstrate a 38% relative risk reduction (RRR) for unresponsive disease, with sHD given as investigational course 1, with 80% power and 0.05 α error. We used standard endpoint definitions. CR was defined as more than 1.0 × 10⁹/L neutrophils and at least 100 × 10⁹/L platelets, regenerating marrow with evidence of trilineage hematopoiesis, blast cells less than 5% (no Auer rods), and no extramedullary leukemia. CR included CR with incomplete hematological recovery, defined as fewer than 1.0 × 10⁹/L neutrophils and/or fewer than 100 × 10⁹/L platelets.⁵  Resistant AML was defined as less than CR/CR with incomplete hematological recovery, with persistence of AML blasts in the bone marrow and/or blood smears. Response was assessed by evaluation of bone marrow on day 28, or later if clinically indicated. Day 14 bone marrow examination was not planned in this study.

Randomization 2 compared long-term RFS (time from date of CR to relapse or death in remission) between SR treatment groups. Secondary objectives included overall survival (OS; from study enrollment to death), event-free survival (from enrollment to induction failure, relapse, or death in remission),²  cumulative incidence of relapse, and the feasibility and efficacy of study treatments in different age and risk groups. To evaluate prognostic effects not assessable by the original Northern Italy Leukemia Group risk classification, we performed post hoc prognostic analysis, using the European LeukemiaNet (ELN) 2010 genetic risk stratification,⁴⁴  including the results of centralized analysis of CEBPA and MLL mutations in cryopreserved AML cells. Therapy-related complications were graded using common toxicity criteria (CTC; http://ctep.cancer.gov/protocolDevelopment/electronic_applications/docs/ctcaev3.pdf).

Statistical analyses were performed following the intention-to-treat principle. Baseline patient characteristics were compared between treatment groups, using a χ² or Fisher’s exact test (categorical variables) or the Mann-Whitney U test (continuous variables). We assessed the relationship between CR achievement and study group, using the χ² test. The RRR indicated the proportion of ICE failures (resistance/death) that could have been avoided with sHD. Survival was estimated using the Kaplan-Meier method, and long-term outcomes compared using the log-rank test. Relapse incidence was assessed using cumulative incidence function, considering death as a competing event. Subgroup and multivariate analyses with logistic regression and Cox models were performed to estimate odds ratios with 95% confidence intervals (CIs) for CR achievement, and hazard ratios with 95% CIs for survival outcomes. In multivariate analysis, HSCT was considered a time-dependent variable. P values were 2-sided and not adjusted for multiple comparison. The significance level was fixed at 5%. Statistical analyses were performed with SAS software version 9.4.

Figure 2 shows patient disposition and study flow. Between January 2007 and March 2012, we registered and screened 1354 patients for eligibility, of whom 574 were enrolled and randomized. Among these patients, 572 were evaluable: 286 in each study group. Two trial patients were excluded from analysis: 1 withdrew consent and 1 had Philadelphia-positive AML and was switched to tyrosine kinase inhibitors. Five patients did not complete the allocated treatment, including 2 in the ICE group (1 early death, 1 medical decision) and 3 in the sHD group (1 early death; 2 other chemotherapy), but were included in the analysis based on treatment intention. The median patient age was 52 years (range, 16-73 years). Diagnostic characteristics and HR proportions were balanced between study groups, except that the sHD group included a higher incidence of patients with an optimal performance score (Table 1). Of the 1354 patients registered in the prospective outcome study project, 780 were excluded from the trial for the following reasons: diagnosis of acute promyelocytic leukemia (n = 118), trial ineligibility (n = 264), medical decisions related to higher age and/or clinical unfitness (n = 209; median age, 64 years; range, 19-89 years), refusal (n = 43), early death (n = 6), unspecified reason (n = 5), and (starting September 2009) age older than 65 years (n = 135; supplemental File 3.2). The outcomes of all 1354 registered patients are available online (supplemental File 3.3).

In randomization 1, the CR rate was significantly higher with sHD (n = 233, 81.5%) than ICE (n = 198, 69.2%; P = .0007) because of a markedly lower incidence of refractory AML (11.2% vs 25.2%; P < .0001) and a comparable early death rate (P = .39; Table 2). sHD treatment yielded a RRR of 0.40 (95% CI, 0.19-0.55), matching the predicted primary study endpoint (RRR, 0.38). Based on the course 1 refractory responses, course 2 reinduction chemotherapy was administered to 46/72 patients in the ICE-resistant group and 17/32 patients in the sHD-resistant group, achieving late CR in 33 (71.7%) and 6 (35.3%) patients, respectively, with augmented sHD (n = 25/36, 69.4%) or other chemotherapy (n = 14/27, 51.9%). After 2 induction courses, the overall CR rates were 80.8% in the ICE group and 83.6% in the sHD group (P = .38). In addition, CR was achieved in 12 of 13 course-1-resistant patients who underwent off-study allogenic HSCT.

In randomization 1, sHD more frequently led to blood cytopenia, fever, and infections related to treatment-emergent toxicity, as well as hepatic and metabolic complications, but did not increase early mortality (Table 3). From treatment day 1, recovery to a neutrophil count higher than 0.5 × 10⁹/L took 21 days with ICE vs 27 days with sHD (P < .0001), and recovery to a platelet count higher than 20 × 10⁹/L took 21 days with ICE vs 29 days with sHD; P < .0001). Infection rates were 82% with ICE and 90.5% with sHD (P = .003), with sHD associated with high incidences of bacterial and Aspergillus spp. etiology, sepsis, and pneumonia. Most hepatic, metabolic, and cutaneous adverse events were CTC grade 2 or lower. Only 2 to 5 episodes in each CTC category scored higher with sHD than ICE.

As a result of greater myelosuppression and related toxicities, significantly fewer patients with sHD completed early postremission consolidation (n = 166/239, 69.5%) compared with ICE patients (186/231, 80.5%; P = .005). Compared with ICE, sHD required longer intercycle intervals and displayed inferior CD34⁺ blood cell mobilization (supplemental File 3.4). Of the 470 total CR patients, 193 (41%; 20% of SR and 50% of HR; 102 ICE vs 91 sHD, P = .18) received allogeneic HSCT as their final treatment. Another 103 patients (22%; 41% of SR and 13% of HR) received HD treatment under randomization 2, of whom 83 were randomized and 80 actually underwent either autologous HSCT or repetitive blood stem cell-supported HD courses. Of the remaining patients, 97 (21%) received ID chemotherapy courses, 60 (12.8%) underwent only early consolidation because of early relapse or toxicity, and 17 (3.8%) received no postremission therapy.

After a median follow-up of 4.9 years (range, 0.2-8.4 years), 204 patients (35.6%) were alive in first CR (84 ICE [29.3%] and 120 sHD [41.9%]; P = .002), 42 (7.3%) survived in second/later CR, 313 (54.7%) died, and 13 (2.3%) were lost to follow-up. The 5-year cumulative incidences of relapse were 55% (ICE) and 44% (sHD; P = .046). Compared with ICE, sHD was associated with improved 5-year OS (39% [median, 2.2 years] vs 49% [median, 4.5 years]; P = .045), 5-year event-free survival (29% [median, 1.2 years] vs 40% [median, 1.6 years]; P = .019), and 5-year RFS (36% [median, 1.5 years] vs 48% [median, 3.4 years]; P = .028; Table 2; Figure 3). We further examined the survival effects of allogeneic HSCT or randomization 2 therapy. When the 193 patients who underwent allogeneic HSCT in first CR were censored at the time of transplant, sHD no longer conveyed a detectable significant survival advantage (supplemental File 3.5). However, when separately considering the 2 study groups, sHD-treated patients (n = 91) exhibited significantly better posttransplantation outcomes than ICE-treated patients (n = 102): 5-year survival of 69% vs 55% (P = .06) and RFS of 66% vs 49% (P = .03; supplemental File 3.6). Finally, among randomization 2 patients (n = 83 evaluable; 44 ICE and 39 sHD), 5-year OS and RFS did not significantly differ between the 2 groups (supplemental File 3.7.1). Projected RFS was 50% (median, 2.6 years) in the blood stem cell-supported HD chemotherapy group (n = 42, 103 total HD courses; median, 3 courses; range, 1-3 courses) and 36% (median, 1.48 years) in the BU-CY autotransplantation group (n = 41; P = .31). We observed a trend of better survival with repetitive HD courses than with autotransplantation in patients with sHD (P = .05; supplemental File 3.7.2), but no superiority compared with ICE-treated patients. Compared with HR patients, SR patients had a higher 5-year RFS (51% vs 30%; P = .03). Pancytopenic death rates were 2.4% with autotransplantation vs 2.5% with HD courses (1% among 103 total HD courses). The cumulative incidence of bacterial infections and metabolic, neurologic, and cutaneous toxicity was greater in the repetitive HD chemotherapy group (supplemental File 3.8).

Early CR rates were significantly higher with sHD than ICE in several groups, including the large reference group of patients aged 60 years or less with de novo AML, the HR AML group, adverse genetic/cytogenetic subsets, and the NPM1 wild-type AML group (Figure 4). Although the sHD group had a higher incidence of favorable Eastern Cooperative Oncology Group performance score (ECOG PS 0-1), this regimen was associated with an improved early CR rate even among patients with an ECOG PS higher than 1. Some smaller patient populations showed nearly significant (P = .05-.09) CR differences between treatments, including patients with HR morphological variants, high leukocyte counts, age older than 60 years, and/or non de novo AML. The assumption that an improved early CR rate would positively affect OS was validated in both SR and HR groups (supplemental File 3.9), excluding center-related effects on early CR rate, OS, and RFS (Table 2). sHD improved OS and/or RFS in patients aged 60 years or younger with de novo AML, with SR AML, and with core-binding factor-positive AML. Post hoc analysis with the ELN 2010 genetic risk stratification revealed that sHD improved OS and/or RFS in the favorable group, including patients with the NPM1-mutated/FLT3 wild-type genotype or mutated CEBPA with a normal karyotype other than core-binding factor AML (supplemental File 3.10).

Multivariate analysis including randomization 1 chemotherapy and the allogeneic HSCT results described earlier confirmed that both sHD induction chemotherapy and allogeneic HSCT had independent, favorable, prognostic effects on survival. This effect was not detectable with autotransplantation in patients entering randomization 2 (supplemental File 3.7.2). Older age, poor ECOG performance score, intermediate/adverse cytogenetics, and FLT3 mutations negatively affected survival probability, and NPM1 mutations were consistently associated with better outcome (Table 4). We repeated this analysis with the study risk model and the ELN 2010 genetic risk stratification. After removing the single risk factors included in either risk classification, both sHD regimen and the SR or ELN favorable risk profile were associated with better CR, OS, and RFS.

In this randomized study, we compared the sHD regimen with standard ICE chemotherapy, which was considered at least as effective as a HD daunorubicin “3+7” regimen for inducing CR in adult AML.^9,11  This controlled study evolved from a prior AML investigation in which we tested sequential HD reinduction for ICE-unresponsive patients,^29,38  obtaining similar results across all cytogenetic risk groups,²⁸  and evaluated blood stem cell-supported HD consolidation courses (NCT00400673) in lieu of autotransplantation.^45-47 

Our present results indicated that sHD yielded a significantly better early induction response and survival than ICE. This study included 572 evaluable patients aged at least 16 years (median age, 52 years). More than 20% had very HR AML or HR MDS and/or were aged older than 60 years, and more than 70% were classified as HR according to the mixed functional clinico-cytogenetic prognostic model used to distinguish between patients eligible for allogeneic HSCT (HR) and others (SR). The primary study endpoint (CR after course 1) showed therapeutic benefit with sHD (early CR of 81.5% vs 69.2%; P = .0007) that was attributable to a reduction of chemo-resistant AML from 25% to 11%. The sHD and ICE groups exhibited comparable early mortality, consistently below 10% and lower than reported in German trials with similar regimens.^14,26  After confirming these results in most HR subsets, including patients with adverse ECOG PS, we further demonstrated the feasibility of the current sHD schedule. Although sHD lacked double induction with recycling on day 21, as in the AMLSG trial,⁴⁸  it did not favor higher resistance rates. After 23 ICE-resistant patients entered CR with augmented sHD, the gap between study groups (resulting from early CR) was eventually closed. The prognostic advantage related to early CR was confirmed by comparative survival analysis including late CR patients. Retreatment of sHD-resistant patients was more difficult, with poor results after augmented sHD. These very HR patients should be considered for new experimental therapies and/or immediate HSCT salvage.

Induction mortality was not higher with sHD, but sHD showed higher hematologic and infectious toxicity than ICE, which hampered postremission therapy, contributing to deficient patient accrual for randomization 2. However, despite the trend of inferior transplantability, sHD provided a lower relapse incidence and better RFS than the control group, consistent with the idarubicin/HD cytarabine/fludarabine group of the Medical Research Council 15 trial.⁴⁹  These results suggested that early treatment intensity had an overall higher effect than the number and interval of subsequent chemotherapy courses. Both the current sHD and idarubicin/HD cytarabine/fludarabine regimens included idarubicin plus cytarabine 2 g/m², with G-CSF, and the latter regimen added fludarabine.^49,50  Together, these findings suggest that sHD may be a valid alternative to the classic “3+7” or ICE regimens for inducing early CR, particularly in HR patients, and to standard reinduction for patients unresponsive to a conventional first course, in keeping with recent ELN recommendations.¹ 

Separate analysis was required to assess suitability for allogeneic HSCT and its effects. HSCT was applicable to 50% of HR patients, and multivariate analysis confirmed that HSCT was an essential component of curative treatment of many of these patients. However, sHD patients had a significantly better posttransplantation course than ICE patients, indicating a positive interaction between HD induction and postinduction HSCT. Recent HSCT studies have demonstrated the usefulness of MRD analysis for predicting the risk for transplantation failure.^51-53  Our present randomization 1 results might suggest that sHD-treated patients may harbor less residual AML cells before HSCT compared with ICE-treated patients. Although a formal pretransplantation MRD study was not planned, at randomization 1, we performed an immunophenotypic study of early peripheral blast cell clearance.⁵⁴  These data will be separately reported, and could confirm that the sHD regimen had greater activity in achieving early blast cell reduction.

One main limitation of this study was that induction course 2 was not delivered as homogeneously as planned. Among course-1-resistant patients, 43% did not receive the augmented sHD regimen because of medical decisions. However, this finding did not affect the analysis of randomization 1 results and did not hamper the outcome comparison between early vs late CR patients. In addition, CR was not centrally assessed but was evaluated using standard criteria.⁵  Another limitation was that randomization 2 failed because of lower-than-expected patient accrual. This step required the harvest of 2 to 6 × 10⁶/kg CD34⁺ blood cells, which was successful in only 45.9% of CR cases. We previously adopted the use of repetitive HD consolidation courses, supported with 1 to 2 ×10⁶/kg CD34⁺ blood stem cells (NCT00400673), to avoid prolonged pancytopenia-related complications and ensure dose-dense consolidation^55,56  compared with similarly effective schedules.⁵⁷  Compared with BU-CY-conditioned autotransplantation, we found that the HD consolidation modality was feasible, with minimal mortality risk, although infectious morbidity was increased as a result of the cumulative effects of multiple HD courses. This type of consolidation preserved fertility and showed noninferior outcomes compared with autotransplantation. As a consequence, we consider this protocol for younger patients who are excluded from allogeneic HSCT.

Another limitation of this study was that we lacked details regarding new highly relevant genetic markers.^58,59  Nevertheless, our univariate and multivariate prognostic analyses revealed factors associated with the best outcomes. We defined HR categories according to cytogenetics (with results comparable to other adult AML series), adverse clinical characteristics, and FLT3 mutations. We also studied NPM1, MLL, and CEPBA gene mutations according to the ELN 2010 genetic risk stratification, which allowed analysis refinements in many cases. Both approaches confirmed that sHD was superior to ICE in achieving CR (particularly in HR subsets) and prolonging OS and RFS. The best results were observed in SR or favorable-risk (ELN) groups, and in the typical subset of patients aged 60 years or younger with de novo AML. The improved CR rate with sHD in the HR group did not affect survival in the general intention-to-treat analysis. However, this combination provided high RFS and survival rates in the subgroup of CR patients proceeding to HSCT.

Overall, our present findings highlighted the inappropriateness of a uniform postremission therapy, with HSCT as the sole therapeutic option, among patients with aggressive disease behavior. In this HR setting, and for AML in general, we might further improve results by targeting specific molecular lesions.^60,61 

This work was partially supported by grants from Agenzia Italiana del Farmaco (Rome, Italy, Project FARM6YMY2N/2006), Fondazione Guido Berlucchi-Onlus (Brescia, Italy, 2006), and Associazione Italiana per la Ricerca sul Cancro (grant IG 2016 n. 18568) (B.F.).

Contribution: R.B. and A.R. designed and performed research, analyzed and interpreted data, and wrote the manuscript; T.I., G.G., F.M., I.C., D.M., E.T., L.D.P., C.C., E.B., F.C., M.B., A.M.S., E.T., L.C., P.C., A.C., D.F., P.Z., O.S., E.A., S.C., A.B., B.F., and E.M.P. performed research and collected data; A.M., C.B., and E.O. analyzed and interpreted data and performed statistical analysis; and C.P. analyzed and interpreted data, performed statistical analysis, and wrote the manuscript.

Conflict-of-interest disclosure: The authors declare no competing financial interests.

Enrico M. Pogliani died on 15 July 2015.

Correspondence: Renato Bassan, Unità Operativa Complessa Ematologia, Ospedale dell’Angelo, Via Paccagnella 11, 30174 Mestre-Venezia, Italy; e-mail: renato.bassan@aulss3.veneto.it.