During the recent years, considerable progress has been made in deciphering the cytogenetic abnormalities and molecular genetic events occurring in acute myeloid leukemia (AML) cells. These data allowed for the assessment of new diagnostic and prognostic markers in AML, especially in normal karyotype AML (NK-AML) patients, considered as intermediate risk. Thus, it has been shown that mutations occur frequently in several genes in NK-AML cells, and two of them are now especially relevant in clinical practice for the prediction of relapse and the subsequent indication for allogeneic hematopoietic stem cell transplantation (allo-HSCT) in first complete remission (CR1). The nucleophosmin gene (NPM1) has been shown to be mutated in almost half of NK-AML patients [1, 2], whereas ∼30% of these same patients carry a fms-like tyrosine kinase 3 gene internal tandem duplication (FLT3-ITD) . Multiple studies have shown that the genotype “mutated NPM1 without FLT3-ITD” represents a favorable prognostic marker, with higher CR rates, and better relapse-free survival and overall survival (OS), reminiscent of what is seen in patients with favorable cytogenetics (i.e., inv(16)/t(16;16) or t(8;21) AML) [4–7]. Thus, an international expert panel of the European LeukemiaNet (ELN) has recently proposed to include mutant NPM1 (NPM1+) without FLT3-ITD NK-AML patients in the subset of favorable prognostic patients . Allo-HSCT would thus no longer be recommended for such patients although there is accumulating evidence that allo-HSCT is an attractive option for patients with NK-AML and unfavorable molecular markers (i.e., NPM1+/FLT3-ITD+, NPM1-/FLT3-ITD+, and NPM1-/FLT3-ITD−) . According to the ELN, autologous HSCT (auto-HSCT) represents a good alternativeoption for AML patients eligible for an allograft but lacking a human leukocyte antigen (HLA)-compatible donor and may offer an advantage in specific subsets compared with chemotherapy . However, the impact of auto-HSCT according to the NPM1/FLT3-ITD status has been poorly studied so far in NK-AML. This is what was evaluated here retrospectively in a cohort of 135 NK-AML patients treated by allo-HSCT, auto-HSCT, or chemotherapy alone as consolidation within the Groupe d'Etude des Leucémies et Autres Maladies du Sang (GOELAMS) LAM-2001 trial.
Materials and Methods
For the present analyses, we screened all the NK-AML patients of the phase III GOELAMS LAM 2001 trial who achieved CR1 and for whom samples at diagnosis were available to assess the NPM1/FLT3-ITD molecular status. The multicenter prospective randomized Phase III GOELAMS LAM-2001 trial (http://clinical.trials.gov#NCT01015196) was initiated to assess the potential benefit of adding a second auto-HSCT in non-M3-AML adults less than 60 years of age in CR1 lacking an HLA-identical sibling donor [9, 10]. This study included 832 patients (median age: 48 years old, range: 17–60). The trial received approval from the IRB of Nantes University Hospital, France. All patients provided informed consent in accordance with the Declaration of Helsinki. The design of the study is summarized in Fig. 1. Briefly, patients were first randomized at diagnosis to receive, at induction and later in the treatment course, either idarubicin (IDA) or daunorubicin (DNR) as anthracyclin. Induction chemotherapy consisted in DNR (60 mg/m2 on days 1–3) or IDA (8 mg/m2 on days 1–5) associated with continuous infusion of cytosine arabinoside (Ara-C) (200 mg/m2 on days 1–7). If there were >5% bone marrow (BM) blasts at day 15, a second induction course was given including DNR 35 mg/m2 or IDA 8 mg/m2 on days 17 and 18, according to initial randomization, and Ara-C 1000 mg/m2 bid on days 17–20. After induction therapy, all patients received a miniconsolidation (IDA 12 mg/m2 or DNR 60 mg/m2 on days 1–2 and Ara-C 50 mg/m2 bid SC on days 1–7).
Search for an HLA identical sibling donor was performed for each patient at diagnosis and was pursued up to the documentation of CR1 after induction. Unrelated HLA-matched donors were not considered in this study. Patients with a sibling donor were programmed to receive allo-HSCT. Patients up to 50 years of age were planned to receive a BM allo-HSCT with a myeloablative conditioning regimen (MAC), whereas patients over 50 years of age received a peripheral blood stem cells (PBSC) allo-HSCT with a reduced intensity conditioning regimen (RIC). Patients lacking a donor were secondly randomized to receive either one auto-HSCT or two auto-HSCTs. When transplantation (auto or allo-HSCT) was not feasible after this randomization, patients underwent the chemotherapy-based consolidation therapy planned by the protocol (one course of high-dose Ara-C 3 g/m2 bid over 4 days and one course of amsacrine 150 mg/m2 per day and etoposide 150 mg/m2 per day both over 5 days if included in the two auto-HSCTs arm).
For this study, available diagnosis samples from patients included in the GOELAMS LAM-2001 trial were screened for NPM1 mutations and FLT3 internal tandem duplication. CEBPA mutations were not considered in this study. Two central reference molecular laboratories of the GOELAMS (at Reims and Angers) analyzed these samples according to the reported methods [11, 12].
Descriptive statistics were reported as frequencies, or medians and range. Normality of samples' distribution was checked with the Kolmogorov–Smirnov test. Comparisons of median values were performed using the Mann–Whitney rank sum test. The chi-squared test or Fisher's exact test was used to test for differences between groups.
OS and leukemia-free survival (LFS) were calculated from the date of the first randomization until the date of death from any cause and from the date of CR1 until the date of relapse or death of any cause, respectively, censoring other patients at the date of last follow-up. Estimates of LFS and OS were calculated by the Kaplan–Meier method, and comparisons of survivals were achieved with the log rank test. A multivariate analysis was performed according to the Cox proportional-hazards regression model. All statistical analyses were performed with the MedCalc® software (Mariakerke, Belgium). Final analyses were performed in December 2011.
Although molecular studies for NPM1 or FLT3 had not been planned initially, BM or peripheral blood samples collected at diagnosis from 135 of the 337 NK-AML patients of the multicenter prospective randomized Phase III GOELAMS LAM-2001 trial could be retrieved retrospectively. We did not note any discrepancies in terms of clinical or biological features nor outcome between these 135 patients analyzed here and the whole NK-AML cohort (data not shown). Forty-one patients received chemotherapy alone as consolidation therapy according to the design of the trial. Fifty-seven patients received one (n = 40) or 2 (n = 17) autografts. As recently reported, we did not see any advantage to perform a double over a single autograft in our series . Consequently, the whole auto-HSCT cohort was tested as one group for the significance of the molecular status. Thirty-seven patients received an allo-HSCT (MAC: n = 27; RIC: n = 10). The outcome of allo-HSCT patients was also not different in this trial whatever they received a RIC or a MAC regimen, and the whole allo-HSCT cohort was therefore also tested as one group .
In line with the ELN classification , the impact of the favorable NPM1+/FLT3-ITD− status was compared to all other molecular combinations. The characteristics of patients are summarized in Table I. A total of 46 patients presented the NPM1+/FLT3-ITD− profile, whereas 89 patients presented one of the other combinations (NPM1+/FLT3-ITD+ in 25, NPM1-/FLT3-ITD− in 49, and NPM1-/FLT3-ITD+ in 15). There was no significant difference between the two groups regarding the characteristics of patients. Both molecular subgroups appeared also well balanced concerning the type of consolidation treatment received (i.e., chemotherapy, auto-HSCT, or allo-HSCT). According to initial randomization, 44% of patients with the NPM1+/FLT3-ITD− profile received IDA as induction compared with 35% for the other profiles (P = 0.025). However, as we have previously shown, the CR rate was not significantly different between the two induction treatment arms .
Table I. Characteristics of the 135 NK-AML Patients
Median follow-up was 86 months for alive patients (range: 16–118) in the whole cohort. In univariate analysis, 4-year LFS and 4-year OS were significantly higher for NPM1+/FLT3-ITD− patients compared with patients presenting another molecular profile (NPM1+/FLT3-ITD+, NPM1-/FLT3-ITD−, or NPM1-/FLT3-ITD) [(4-year LFS: 61 vs. 43% and 4-year OS: 72 vs. 48%, P = 0.02 and P = 0.01, respectively].
Outcomes according to the type of consolidation therapy received were then analyzed in each molecular profile subgroup. In the NPM1+/FLT3-ITD− subgroup, the chemotherapy, the auto-HSCT, or the allo-HSCT received as consolidation provided similar outcome for the patients [4-year LFS: 60, 56, and 71% (P = 0.77) and 4-year OS: 60, 71, and 73% (P = 0.73), respectively] (Fig. 2). For patients with other NPM1/FLT3-ITD molecular profiles, chemotherapy alone provided a clearly worse outcome than allo-HSCT but also than auto-HSCT. In this adverse molecular subgroup, 4-year LFS was 36, 44, and 68% (P = 0.004) and 4-year OS was 29, 52, and 68% (P = 0.02), respectively, for the chemotherapy, auto-HSCT, and allo-HSCT subgroups (Fig. 3).
In multivariate analysis including age, white blood count, consolidation therapy (chemotherapy, auto-HSCT, or allo-HSCT), and mutational status (NPM1+/FLT3-ITD− or others), only the type of consolidation strategy showed a significant prognostic impact on LFS (HR: 1.89, 95% confidence interval: 1.28–2.57; P = 0.0006) and OS (HR: 1.68, 95% confidence interval: 1.17–2.42; P = 0.004).
This study took advantage of an existing retrospective cohort of NK-AML patients in CR1, treated in an original trial that started in 2001 and aimed at comparing one vs. two auto-HSCT(s) in cases without sibling donor for allo-HSCT. Molecular studies for NPM1 or FLT3 could be performed on 135 of the 337 NK-AML patients initially included in the Phase III GOELAMS LAM-2001 trial. We could check beforehand that the series of patients for whom we could use stored material was astonishingly matching the whole group for all clinical and biological features. Moreover, skipping molecular profiles, survival curves in this subgroup exactly matched those of the whole series (data not shown).
On the basis of the rationale of this specific protocol, our data confirm the favorable prognostic impact of the NPM1+/FLT3-ITD− molecular profile in a homogeneous population of de novo NK-AML patients below 65 years of age in CR1. Indeed, the impact of the NPM1/FLT3-ITD molecular status is now well established [1–5, 8]. It was recently evaluated in 99 autografted NK-AML patients showing significantly longer OS and LFS in the subgroup of 19 patients with a favorable molecular status (OS: not reached vs. 25 months, P = 0.02; LFS: not reached vs. 16 months, P = 0.007) .
So far, one German team has compared outcomes after allo-HSCT vs. other consolidation strategies in NK-AML patients according to their molecular profile . This donor vs. no donor analysis showed a benefit for allo-HSCT in the adverse molecular subgroup but none in the favorable NPM1+/FLT3-ITD− subgroup. Nevertheless, it appears that no clear comparison between chemotherapy, auto-HSCT, and allo-HSCT was assessed in this series as reported here. Our analyses showed that patients with the favorable NPM1+/FLT3-ITD− molecular profile had a similar outcome whether they received chemotherapy, autologous transplantation, or allogeneic transplantation as consolidation therapy. The demonstration of a very good outcome and the lack of benefit for allo-HSCT in the favorable molecular subgroup is of course a strong argument to include these patients in the non-M3 AML subgroup with favorable cytogenetics, as recommended recently . Indeed, it is now well demonstrated that non-M3 AML patients in CR1 with favorable cytogenetics have a favorable outcome and do not benefit from allo-HSCT as consolidation. Moreover, comparison of auto-HSCT vs. allo-HSCT in this favorable population showed no significant differences in terms of survival [14, 15]. Currently, high-dose Ara-C stands as the consolidation of choice in these patients, although the results of auto-HSCT are very good, reaching, for example, a 5-year LFS of 66% in the study of Gorin et al.  and a 4-year LFS of 72% in our own LAM-2001 trial . These data raise the question of the role of auto-HSCT vs. chemotherapy alone as consolidation therapy either for NPM1+/FLT3-ITD− patients or non-M3 AML patients with favorable cytogenetics in prospective studies.
Considering NK-AML patients with no favorable molecular profile, and as already reported [5, 8], it is clear from our results that they benefit from allo-HSCT as consolidation. Such patients have therefore to be still considered as belonging to the intermediate-risk subgroup where allo-HSCT is indicated if an HLA-matched donor is available . Conversely, auto-HSCT may still represent a good alternative therapy compared with chemotherapy alone as consolidation, as suggested here. In published AML randomized studies, all using BM as stem cell support so far, auto-HSCT failed to demonstrate a clear advantage compared with intensive chemotherapy with OS ranging from 43 to 56% and DFS between 38 and 53% at 3–7 years [16–21]. Indeed, although two meta-analyses [22, 23] have suggested a modest advantage for auto-HSCT, because of a higher relapse rate after intensive chemotherapy (55–68% vs. 31–52%), the higher toxicity of the auto-HSCT procedure explains the lack of significantly different outcomes between these two schedules. Reducing the toxicity of auto-HSCT would therefore be of crucial importance to improve this strategy and patients' outcome. One option could be to use PBSC instead of BM, as PBSC allow for a faster hematologic recovery with a transplant-related mortality of less than 10% [24–26]. An alternative means would be to use granulocyte colony-stimulating factor as it can also shorten significantly the duration of neutropenia and hospitalization after intensive consolidation therapy and might permit the delivery of chemotherapy at a higher dose intensity while still allowing appropriate stem cell products to be collected [27, 28]. These two conditions were fulfilled within the prospective GOELAMS LAM-2001 trial, but overall comparison of chemotherapy vs. auto-HSCT as consolidation showed no significant differences in terms of outcomes . Here, we demonstrate retrospectively that auto-HSCT may benefit to some subgroups of NK-AML patients, providing a significantly longer LFS and OS than chemotherapy alone.
In conclusion, molecular analysis of available samples from the prospective randomized GOELAMS LAM-2001 trial confirmed the major prognostic impact of the NPM1/FLT3-ITD molecular status in younger NK-AML patients. Auto-HSCT provided a similar outcome as allo-HSCT or chemotherapy alone as consolidation therapy for good prognosis of NPM1+/FLT3-ITD− patients. For patients with adverse molecular profiles, allo-HSCT remains strongly recommended but auto-HSCT has to be considered if no HLA-matched donor is available. Such results provide necessary preliminary data with the aim to design future prospective trials evaluating the indications of auto-HSCT for specific subsets of AML patients.
RG and PC analyzed data, recruited patients, provided clinical care, performed bibliographic search, and wrote the manuscript. MCB performed data collection management, validation, and statistical analyses and wrote the manuscript. PCL and OB produced and validated molecular data. BL recruited patients, provided clinical care, and performed data collection management and validation. AP, CR, BW, NF, CEB, DB, NV, MD, PT, CH, JLH, FD, and NI recruited patients and provided clinical care. All the above authors approved the manuscript for publication purposes.
The authors thank Cindy Grandjenette for her help in collecting and verifying the raw data and all the data managers (CRA) of the GOELAMS, especially Roselyne Delepine, who all contributed to construct an exquisitely comprehensive database of the trial.