Tazemetostat (TAZ) in relapsed/refractory (R/R) follicular lymphoma (FL): Propensity-score matched analysis of E7438-G000-101 trial outcomes.

Abstract

e18627 Background: Oncology drug development often requires the use of non-randomized, open-label, phase 2 basket studies to better understand the early activity and safety of a potential new therapy. As such, baseline demographics and disease characteristics may differ between cohorts which can impact the perception of efficacy between cohorts. TAZ, a first-in-class, oral enhancer of zeste homolog 2 (EZH2) inhibitor was approved by the US FDA after demonstrating single-agent, antitumor activity in a phase 2 study in adults with wild-type (WT) or mutant (MT) EZH2 R/R FL who had received ≥2 prior systemic therapies (NCT01897571). Differences between the cohorts in baseline characteristics known to be prognostic for clinical outcomes were noted, with the WT EZH2 cohort enrolling more patients with poor-risk features. This analysis assessed outcomes in the 2 groups after minimizing differences in baseline characteristics by creating a matched sample of directly comparable WT and MT patients. Methods: Propensity scores for each WT (n = 54) and MT (n = 45) EZH2 patient in the study were generated, based on the likelihood of being selected given their baseline characteristics. Characteristics identified for inclusion in the model were chosen if they were prognostic based on peer-reviewed literature and where larger differences were observed between cohorts at baseline: ECOG performance status, number of prior lines of anticancer therapy, progression of disease within 24 months, double refractory status, and prior history of hematopoietic stem cell transplant. Patients were matched 1:1 on propensity score, using a nearest-neighbor approach with caliper restrictions. Baseline covariates between the two matched groups were found to be sufficiently balanced. Objective response rate (ORR) point estimates were measured for the matched WT and MT EZH2 groups, and progression-free survival (PFS) was described using Kaplan-Meier analyses. Results: The propensity-matched sample included 56 patients (28 WT and 28 MT). Prior to matching, ORR was 35% (95% CI [22%, 48%]) in the WT and 69% (95% CI [55%, 83%]) in MT EZH2 groups; after matching, the ORR was 50% (95% CI [31%, 69%]) and 71% (95% CI [54%, 88%]), respectively. Median PFS was 11.1 months (95% CI [5.4, 16.7]) in the WT and 13.8 months (95% CI [11.1, 22.1]) in the MT EZH2 groups prior to matching, and 14.3 months (95% CI [11.1, inf]) and 14.8 (95% CI [10.7, inf]) months in the WT and MT EZH2 matched groups, respectively. Conclusions: As expected, efficacy remained higher in the MT EZH2 group; however, after adjustment, the ORR and PFS improved in the WT EZH2 group. This hypothesis-generating analysis suggests that outcomes in patients with WT EZH2 R/R FL treated with TAZ may have been more similar to those in the MT EZH2 group in the phase 2 trial had the baseline disease characteristics been more equally matched.