Increased LFS Following Hematopoietic Cell Transplantation As Compared to Conventional Consolidation Therapy in Patients >60 Years with AML in First Complete Remission and a Matched Donor: Results of a Randomized Phase III Study

Background: Acute myeloid leukemia (AML) has a dismal prognosis especially in elderly patients where the incidence is highest. In particular the reduced complete remission (CR) and the very high relapse rates (>80%) result in long-term survival of 20% or less. Allogeneic stem cell transplantation (SCT) is the treatment modality with the highest anti-leukemic potential, combining immunological anti-leukemia effects with preparative regimens of variable intensities. Retrospective analyses and comparisons with alternative treatments have indicated a possible advantage of SCT, but selection biases remain a concern. Since randomized trials have been lacking, we investigated the role of SCT in comparison to non-SCT consolidation in an international, multicenter, open label, and randomized phase III trial (EudraCT-Number: 2007-003514-34).

Methods: Newly diagnosed AML patients 60-75 years of age were treated with induction therapy with initiation of a donor search on reaching CR1. Following one consolidation therapy, patients underwent cytogenetic/molecular and comorbidity evaluation and patients with a related or unrelated donor were randomized to SCT or non-SCT (2:1 ratio). SCT was performed after conditioning with fludarabine/200 cGy total body irradiation and cyclosporine /mycophenolate mofetil immunosuppression as previously published (Hegenbart U et al JCO 2006). Non-SCT consolidation was administered according to local protocols. Primary endpoint was leukemia free survival (LFS), secondary endpoints included cumulative incidence of relapse (RI), treatment related mortality (TRM), overall survival (OS) and complications including graft-versus-host disease (GVHD). At the planned first interim analysis based on data from 78 patients, we noted that the LFS was markedly worse than expected, and that the proportional hazard assumption was clearly wrong. The original sample size target of n = 231 was reset to n = 150. Statistical analysis was refocused on the difference in Restricted Mean (RM)-LFS within a time frame of 5 years. The analysis followed the intent to treat principle.

Results: From 2010 to 2017, 245 patients in CR1 (median age 67 years) were registered. After consolidation, 66 patients (29%) failed screening because of relapse/noCR1/no hematological recovery (40.9%), morbidity/informed consent withdrawal/unknown (39.4%), no donor/protocol violation (10.6%) and death (9.1%). Of the remaining 179 patients, 135 (75.4%) had an HLA identical donor and 125 were randomized to SCT (n=83) or non-SCT (n=42). Ten patients were not randomized because they rejected trial participation (n=9) or exposed significant comorbidity (n=1). The recommended time intervals from randomization to SCT (4 weeks) or non-SCT (2 weeks) as defined in the protocol were achieved by 40% and 48% of patients, respectively. The proportions of patients not receiving treatment according to randomization was 20.5% in the SCT arm (relapse n=7; morbidity n=4; withdrawal n=3; unavailable donor n=3) and 16.7% in the non-SCT (relapse n=2; morbidity n=2; withdrawal n=3). After a follow-up of 62 months, the 5-year LFS rates were 28.8% (95%CI: 20.4-40.6) in the SCT and 8.9% (95%CI: 3.1-25.7) in the non-SCT arm. There was a statistically significant difference of 8.9 months (95%CI: 1.3-16.6; p = 0.02) in the primary endpoint RM-LFS favoring SCT (p = 0.02) in multivariate analysis adjusted for donor type and cytogenetic risk (p= 0.01; Figure 1A). RI amounted to 37.8% (95%CI: 27.2-48.4) and 91.1% (95%CI: 80.7-100.0) at 60 months as shown in Figure 1B [HR: 3.1 (1.93-4.98) p<0.0001] and TRM to 26.8% (95%CI: 17.1-36.5) and 0% (95%CI: n.a.-n.a) at 24 months (Figure 1C) in the SCT and non-SCT arm, respectively. Causes of death included infections (39.4%) and GVHD (21.5%). Overall, 12.1% of transplanted patients showed grade III/IV acute GVHD and 37.9% extensive chronic GVHD. OS curves crossed at about 18 months. 5-years OS was 31.3% (95%CI: 22.6-43.2) in the SCT arm and 27.1% (95%CI [15.9-46.4]) in the non-SCT arm. However, 23 of 34 (68%) relapsed non-SCT patients received SCT as part of second line treatment.

Conclusion: This randomized trial demonstrates that SCT is superior to non-SCT consolidation treatment for LFS in elderly AML patients in CR1 on a time horizon of five years. SCT substantially reduces RI that outweighs the associated increased TRM.