Analyzing Efficacy Outcomes from the Phase 2 Study of Single-Agent Tazemetostat As Third-Line Therapy in Patients with Relapsed or Refractory Follicular Lymphoma to Identify Predictors of Response

Background: Tazemetostat, a first-in-class, oral, enhancer of zeste homolog 2 (EZH2) inhibitor was recently approved by the US Food and Drug Administration in patients with relapsed/refractory (R/R) follicular lymphoma (FL) after demonstrating single-agent, antitumor activity in patients with wild-type (WT) or mutant (MT) EZH2. Progression of disease within 24 months (POD24), exposure to multiple lines of prior therapy, and refractoriness to rituximab therapy have been shown to adversely affect the prognosis of patients receiving second- or third-line regimens for R/R FL, including chemoimmunotherapy. We performed a post hoc exploratory analysis to better understand how these factors impact the outcomes in patients receiving tazemetostat. Methods: This open-label, multicenter study (NCT01897571) evaluated tazemetostat 800 mg administered orally twice daily in patients with MT or WT EZH2 R/R FL. The primary endpoint was objective response rate (ORR; complete response + partial response) as assessed by an independent review committee (IRC); secondary efficacy endpoints included duration of response (DOR) by IRC, progression-free survival (PFS) by IRC, and overall survival (OS) by investigator assessment. Predictive modeling using baseline demographic and disease characteristic variables combined from patients with MT or WT EZH2 R/R FL was performed to identify variables predictive of response (ORR, DOR, PFS, and OS). Models were fitted for variables that were categorical and had no missing observations; they also were fitted with/without Groupe d’Etude des Lymphomes Folliculaires (GELF) criteria and with the number of prior lines of therapy (1, 2, or >2, and 1 or 2 vs >2). Due to incomplete data collection, GELF criteria were analyzed with missing observations (n=28) set to “no.” Stepwise logistic and Cox regression was used to determine possible predictors; inclusion at a specified step was based on P≤0.40. Final model inclusion was based on P≤0.20. A final model was run using the possible predictors identified from the previous stepwise regressions. Contingency tables and Kaplan-Meier plots were used to examine significant variables (P≤0.05). Results: In the phase 2 study, the efficacy outcomes by IRC in combined WT and MT EZH2 populations (N=99) were: ORR, 51% (n=50); median DOR, 11 months (95% CI: 7, 19); and median PFS, 12 months (95% CI: 8, 15). Median OS was not reached (95% CI: 38.2, not estimable). Predictive modeling using 17 baseline variables ide