Fludarabine-treosulfan compared to thiotepa-busulfan-fludarabine or FLAMSA as conditioning regimen for patients with primary refractory or relapsed acute myeloid leukemia: a study from the Acute Leukemia Working Party of the European Society for Blood and Marrow Transplantation (EBMT)

Allogeneic hematopoietic stem cell transplant is the only potentially curative option for patients with acute myeloid leukemia (AML) in primary induction failure or refractory relapse. This population represents a big challenge for transplant physicians; nevertheless, according to recent evidence [1, 2], long-term survival can be achieved in about one third of patients undergoing transplant with active leukemia, and recent recommendations support prompt transplant in this setting, avoiding further chemotherapy [3]. The choice of the conditioning regimen is of vast importance in these fragile patients, as the need for powerful cytoreduction should not negate an acceptable toxicity profile of the protocol [4]. Historically, regimens employed in this setting included mainly standard myeloablative protocols based on alkylators or total-body irradiation (TBI) [5‐7]. More recently, alternative strategies have been developed. The sequential fludarabine, intermediate dose Ara-C, amsacrine, total body irradiation/busulfan, cyclophosphamide (FLAMSA) regimen, designed by Kolb and colleagues in the early 2000s [8, 9], has shown promising outcome and currently represents one of the most widely employed protocols in this setting. On the other hand, the relentless effort of transplant physicians to temper conditioning toxicity while retaining a significant myeloablative power recently prompted the design of novel regimens, taking advantage in time of rather old drugs like thiotepa or treosulfan or combining two alkylators at reduced doses. The combination of thiotepa, busulfan, and fludarabine (TBF) was initially proposed as a preparative regimen for cord blood transplant [10]; subsequently, it has demonstrated excellent anti-leukemic activity in haploidentical [11, 12], matched sibling donor (MSD) and unrelated donor (UD) transplant [13, 14]. An additional option is represented by the association of fludarabine and treosulfan (FT), which has been intensely investigated in the last decade [15‐18]. Preliminary results of a prospective randomized trial demonstrated promising outcome following FT conditioning in patients with AML and MDS [19]. Furthermore, in a recent retrospective study comparing treosulfan with busulfan-based regimens in patients with active leukemia at the time of transplant, FT protocol resulted in improved outcome [20]. Given the lack of available reports analyzing and comparing the alternative conditioning protocols in patients with refractory or relapsed AML, we designed the current study to compare outcome of FT, TBF, and FLAMSA regimens in this particularly challenging setting.

Limited data is available to guide the choice of the conditioning regimen for patients with primary refractory or relapsed AML. We thus analyzed and compared the outcome of three commonly used conditioning regimens for active AML namely fludarabine-treosulfan, thiotepa-busulfan-fludarabine, and FLAMSA sequential regimen. Our results indicate global survival of 34% at 2 years; the type of conditioning protocol did not significantly affect survival, which was mostly determined by patient characteristics.

A major obstacle in transplanting patients with active leukemia is the high risk of non-relapse mortality; in fact, historical trials employing standard busulfan- or TBI-based regimens report a NRM rate of approximately 30–40% at day 100 after transplant [24‐26]. In our study including patients up to 76 years of age, NRM at day 100 was around 5% following FT and FLAMSA and 13% after TBF, this difference being not statistically significant. Similarly, NRM at 2 years did not differ among the three regimens. It is important to highlight that the FT cohort included significantly older patients as compared to TBF and FLAMSA groups; in fact, 70% of FT patients were older than 50 years (median age of FT group, 58 years). Different strategies have been followed by researchers aiming to reduce mortality and improve the outcome of patients undergoing transplant with persistent leukemia. The design of the sequential FLAMSA regimen by the German group has represented a major breakthrough in this setting, combining promising anti-leukemic activity with acceptable NRM (about 22% at 2 years) [9]. On the other hand, some recent evidence suggests that redesigning standard myeloablative regimens could be an alternative strategy [1]. In fact, the Gruppo Italiano Trapianto di Midollo Osseo (GITMO) selected the TBF protocol as conditioning regimen for the GANDALF prospective trial, whose results have been recently presented, reporting a NRM rate of 35% at 2 years [27]. An alternative regimen is represented by the combination of fludarabine and treosulfan, which was shown to provide an interesting safety profile and promising outcome in patients undergoing transplant in remission [17] or with persistent leukemia [18, 20].

Importantly, the tolerability of the conditioning regimen should not compromise powerful antitumor activity in patients undergoing transplant with active AML. In the current study, we observed a complete remission rate at day 100 of about 90% following FT and FLAMSA, while CR rate was 80% after TBF, this difference being not statistically significant. Similarly, survival did not significantly differ among the three groups. Two-year survival rates were around 35% following FT and FLAMSA protocols; the latter was in accordance with the original report by the Munich group [8]. Conversely, TBF regimen was associated with a survival rate of 24% at 2 years, consistently with recent evidence from the GITMO trial employing the same protocol (OS at 2 years, 18%) [27]. Incidence of acute and chronic GVHD did not differ between the three regimens. In previous reports including patients with AML in remission, FT protocol has been associated with lower rates of GVHD as compared to busulfan-based regimens; we could not confirm this finding in our population of patients undergoing transplant with persistent leukemia [18, 28]. In fact, previous literature indicates a higher incidence of GVHD in patients transplanted with active disease in comparison to patients undergoing transplant in remission [29]. Nevertheless, we observed a tendency towards better GRFS following the FT protocol.

The rather large cohort included in the present study has allowed us to perform a multivariate analysis on factors predicting transplant outcome in the setting of active AML. Karnofsky performance score below 80% and patient positive CMV serology were strongly associated with poor survival. Further, patients with relapsed AML showed significantly higher risk of disease recurrence after transplant as compared to primary refractory AML. These findings are in accordance with a great body of previous evidence [30‐32]. Transplant from mismatched UD predicted higher risk of grade III–IV aGVHD and non-relapse mortality, with no significant impact on survival. Interestingly, the use of ATG was associated with significantly lower incidence of both acute and chronic GVHD and a strong tendency towards better GRFS, with no influence on relapse rates. The benefit of ATG in terms of reduced incidence of GVHD is well established [33, 34]. However, as graft-versus-leukemia correlates with GVHD, there is a theoretic concern that the use of ATG could result in increased relapse rates, especially in patients undergoing transplant with active leukemia. Interestingly, in the historical ATG trial by GITMO, which included a high proportion of patients with active AML receiving transplant from unrelated donors, the use of ATG was associated with lower incidence of acute and chronic GVHD with no impact on relapse [35]. The findings of the present study are in line with these data. Furthermore, a recent report on patients with high risk AML undergoing transplant following a reduced intensity conditioning regimen including ATG confirmed no increased relapse risk [36]. Finally, it is of interest that in our cohort of patients aged up to 76 years, age at transplant did not influence survival, indicating that older age should not be taken as a criterion to withhold transplant in patients with active AML.

The limitations of the present study are mostly related to its retrospective design; in fact, limited information is available on the reason why a specific patient was allocated to a certain regimen. Further, scarce data on minimal residual disease status after transplant was available in the database; similarly, information on treatment administered after transplant other than cellular therapies (i.e., chemotherapy, target therapies, hypometilating drugs) was incomplete. Nevertheless, since a prospective randomized study comparing the three different conditioning protocols in patients with active AML has not been conducted yet and it is unlikely to be performed in the near future, we believe the results of the present analysis might serve to guide physicians practice in this very high-risk patients.

In conclusion, allogeneic transplant should be strongly considered in primary refractory and relapsed AML, as it is able to provide long-term survival in about one third of these patients. FT, TBF, and FLAMSA represent three possible alternative conditioning options in this setting, providing similar efficacy, toxicity, and survival. In fact, outcome was strongly affected by patient characteristics including Karnofsky performance score and CMV serology, while age should not be taken per se as a criterion to select patients for transplant. The use of ATG was associated with reduced incidence of severe acute and chronic GVHD without influencing relapse risk. Relapse remains the major cause of transplant failure; novel post-transplant strategies are thus in need to prevent disease recurrence in this extremely high-risk population.