Durvalumab plus carboplatin/paclitaxel followed by durvalumab with or without olaparib as first-line treatment for newly diagnosed advanced or recurrent endometrial cancer: Japan subset from the phase III DUO-E trial

Abstract

Objective: DUO-E/GOG-3041/ENGOT-EN10 (NCT04269200) demonstrated statistically significant and clinically meaningful progression-free survival (PFS) improvement with durvalumab plus carboplatin/paclitaxel, followed by durvalumab with or without olaparib, vs. carboplatin/paclitaxel alone (intention-to-treat [ITT] population) in patients with newly diagnosed advanced or recurrent endometrial cancer. We evaluated efficacy and safety in the Japan subset of DUO-E.

Methods: Patients with newly diagnosed International Federation of Gynecology and Obstetrics stage III/IV or recurrent endometrial cancer were randomized 1:1:1 to control arm (carboplatin/paclitaxel + durvalumab placebo [6 cycles] followed by durvalumab placebo + olaparib placebo), durvalumab arm (carboplatin/paclitaxel + durvalumab [1,120 mg every 3 weeks] [6 cycles] followed by durvalumab [1,500 mg every 4 weeks] + olaparib placebo), or durvalumab + olaparib arm (carboplatin/paclitaxel + durvalumab [6 cycles] followed by durvalumab + olaparib [300 mg twice a day]). Dual primary endpoints were investigator-assessed PFS for durvalumab and durvalumab + olaparib arms vs. control. This prespecified exploratory analysis evaluated PFS and safety in the Japan subset.

Results: In the Japan subset (n=88) PFS favored durvalumab (hazard ratio=0.61, 95% confidence interval [CI]=0.32-1.12) and durvalumab + olaparib (0.44, 95% CI=0.22-0.85) vs. control; median PFS was 9.9 and 15.1 vs. 9.5 months, and the 18-month PFS rate was 37.0% and 42.1% vs. 22.2%, respectively. The safety profile in the Japan subset was generally consistent with the full safety analysis set and the established profiles of the individual agents.

Conclusion: Efficacy and safety in the Japan subset were generally consistent with outcomes in the DUO-E ITT population. This Japanese subset analysis of DUO-E supports carboplatin/paclitaxel + durvalumab followed by durvalumab with or without olaparib as new treatment options in patients with advanced or recurrent endometrial cancer and is the first to report on these regimens in Japanese patients alone.

Keywords: Carboplatin; Durvalumab; Endometrial Cancer; Immunotherapy; Japan; Olaparib.

Fig. 1. Patient disposition for the DUO-E Japan subset.

DCO, data cutoff.

Fig. 2. Analyses of PFS according to RECIST version 1.1 in the subset of patients from Japan. Crosses indicate a censored observation. Patients without an event were censored at the latest evaluable RECIST assessment.

CI, confidence interval; HR, hazard ratio; NR, not reached; PFS, progression-free survival; RECIST, Response Evaluation Criteria in Solid Tumors.

Fig. 3. Analyses of (A) ORR and (B) DoR in the subset of patients from Japan.

CI, confidence interval; DoR, duration of response; IQR, interquartile range; NR, not reached; ORR, objective response rate; RECIST, Response Evaluation Criteria in Solid Tumors. ^*Data obtained until starting subsequent cancer therapy or up to RECIST progression or last RECIST assessment. ^†Odds of a response (complete or partial). Odds ratio >1 favors durvalumab or durvalumab + olaparib arm over control. ^‡Time from randomization to onset of response. ^§Time from first confirmed response until progression or censoring for progression-free survival.

Fig. 4. Analyses of (A) TDT, (B) TFST, (C) PFS2 and (D) TSST in the subset of patients from Japan. Time to event rates were estimated by the Kaplan-Meier method.

CI, confidence interval; HR, hazard ratio; NR, not reached; PFS2, time from randomization to second progression or death; TDT, time to treatment discontinuation or death; TFST, time to first subsequent therapy or death; TSST, time to second subsequent therapy or death. ^*CIs for medians were derived using the Brookmeyer–Crowley method. ^†HRs were estimated from an unstratified Cox proportional hazards model. CIs were calculated using a profile likelihood approach.