Post-Allograft Romidepsin Maintenance Mitigates Relapse Risk and Stimulates the Graft-Versus-Malignancy Effect through Enhanced NK-Cell Cytotoxicity in Patients with T-Cell Malignancies: Final Results of a Phase I/II Trial

Abstract

Background: For patients with high risk, and relapsed/refractory T-cell malignancies allogeneic stem cell transplant (allo-SCT) is the only available potentially curative therapy. The efficacy of allo-SCT is limited in this population by high rates of relapse, with rates up to 55-60% post-allo-SCT. We designed a phase I/II trial, evaluating the combination of the histone deacetylases inhibitor romidepsin (rom) with busulfan/fludarabine (BuFlu) conditioning, followed by romidepsin maintenance (m-rom) in patients receiving allo-SCT for T-cell malignancies (NCT02512497). Here we present final clinical results of this therapeutic approach, including an evaluation of the stimulatory effects of m-rom on the graft-versus-malignancy effect through NK-cells post-allo-SCT. Methods: This was a phase I/II clinical trial. Eligible patients had: a diagnosis of T-cell leukemia (including T-acute lymphoblastic leukemia) or T-cell lymphoma (TCL, cutaneous or peripheral) in at least a partial remission requiring an allo-SCT, <70 years of age, with a matched sibling/unrelated donor. Patients received myeloablative (20K) or reduced intensity (16K) BuFlu with rom, followed by standard tacrolimus/methotrexate GVHD prophylaxis and anti-thymocyte globulin for unrelated donors (MUD). An expansion cohort of up to 30 patients (total) was included. M-rom was initiated between day +28 and +100 for 1 year (2 nd year optional), with built in dose reductions for toxicity. An efficacy endpoint of 20% reduction in relapse risk at 1-year in the whole cohort or any individual disease cohort was set based upon historical/CIBMTR (Center for Blood and Marrow Transplant Research) controls was pre-specified. The effect of m-rom on NK-cell cytotoxicity was assessed on samples taken pre-transplant, and 1, 3, 6, 12 months post allo-SCT. NK cytotoxicity was assessed by isolating mononuclear cells from patient samples at each timepoint, comparing those on m-rom (n=13) versus those who did BuFlu controls who did not receive m-rom (n=16). Cells were targeted against K562 targets using the calcein-AM assay. K-M and Fine-Gray models were used to estimate PFS, OS, and cumulative incidence, and compare survival across groups. Results: 28 patients were enrolled on this trial (Table). With a median follow-up time of 15 months, the median OS is 3.3 years (95% CI: 0.85-not reached), with a 1 and 3 year OS probability of 68% and 54%. The median PFS is 2.3 years (95% CI: 0.6-not reached), with 1 and 3-year PFS of 61% and 41%. Cumulative incidence (CI) of NRM at day 100 and 1 year were 14.3% and 21.4%. CI of grade II-IV aGHVD and extensive cGVHD were 46.4% and 38%. The CI of relapse (CIR) was 18.1% at 1 year and 33% at 3 -years. There was no difference between PFS among patients with MRD versus those without MRD prior to transplant (p=0.43). The CIR of patients with TCL was 9.3% at 1 and 3 years, whereas the CIR of patients with leukemia was 25% at 1 year and 48% at 3 years. No patients with PTCL relapsed, 1/2 patients with CTCL relapsed, and 4/6 patients with T-PLL are alive, disease free. Pre-specified general CIR was 55% for all patients. With an overall CIR of 18% at 1 year, and 33% at 3-years, this trial met its pre-specified endpoint of decreasing relapse by 20% at 1-year, and performed well compared to published CIBMTR control populations. Matched propensity score analysis across disease subsets with CIBMTR controls will be presented. 18/28 (64%) of patients received m-rom with a median number of 13 cycles (range 1-47). 8 patients experienced grade 3/4 adverse events (AE), and 1 patient discontinued m-rom due to toxicity, unlikely related to romidepsin (loss of CD3 chimerism). When NK-cytotoxicity was assessed between the two groups after starting maintenance, NK-cytotoxicity in the m-rom group was significantly higher than in those without m-rom (p<0.0001) (Figure). Conclusions: BuFluRom with m-rom is effective at decreasing relapse in patients with T-cell malignancies, with 1-year CI relapse below expected relapse rates for this set of diseases, meeting the pre-specified efficacy endpoint. NK-cell cytotoxicity data in a large sample set of trial patients demonstrates that m-rom enhances NK-cell cytotoxicity post allo-SCT, augmenting the GVL effect and likely accounting for at a minimum, some of the decrease in relapse seen on this trial. M-rom should be considered a new option post allo-SCT to mitigate relapse in patients with T-cell malignancies.