LBA64 Overall survival from a phase II randomised double-blind trial (PERLA) of dostarlimab (dostar) + chemotherapy (CT) vs pembrolizumab (pembro) + CT in metastatic non-squamous NSCLC

The Phase II PERLA trial (NCT04581824) was the first global randomized, double-blind, head-to-head study of two programmed death (PD)-1 inhibitors in NSCLC. In the primary analysis, dostar + CT met its primary endpoint of equivalence, with a favorable numerical trend in overall response rate (ORR) and progression-free survival compared to pembro + CT [1]. Here we report overall survival (OS) analyses from PERLA. Patients (pts) with metastatic NSCLC, documented PD-L1 status, absence of EGFR, ALK or other actionable genomic aberrations determined locally, ECOG 0–1, and no prior systemic treatment were randomised 1:1 to dostar 500 mg or pembro 200 mg Q3W IV up to 35 cycles, combined with CT (4 cycles pemetrexed [pem; 500 mg/m2] + carboplatin [AUC 5 mg/mL/min] or cisplatin [75 mg/m2], followed by pem up to 35 cycles) Q3W IV. The primary endpoint was ORR by blinded independent central review (BICR). OS (secondary endpoint) and duration of response (DoR) by BICR (exploratory endpoint) were determined by Kaplan-Meier method, 95% confidence intervals (CIs) by Brookmeyer-Crowley method and hazard ratio by stratified Cox proportional hazard model. At data cut-off on 07 July 2023, 121 and 122 pts in the dostar + CT and pembro + CT arms respectively, were randomised and treated. OS maturity was 55% (134/243). Median OS was 19.4 mo (95% CI 14.5, NR) in the dostar + CT arm vs 15.9 mo (95% CI 11.6, 19.3) in the pembro + CT arm (Table), after median follow-up (IQR) of 20.7 (17.3, 24.0) and 21.6 (18.3, 24.1) mo respectively, with a similar trend across PD-L1 subgroups. ORR and DoR are shown in the table. No new safety signals were reported.Table: LBA64Dostar+CT (N=121)Pembro+CT (N=122)OS events (death), n (%)59 (49)75 (61)Median OS (95% CI)*, mo19.4 (14.5, NR)15.9 (11.6, 19.3)Hazard Ratio (95% CI)†0.75 (0.53,1.05)ORR by BICR‡, % (95% CI)§ CR, n (%) PR, n (%)45 (36.4, 54.8) 4 (3) 51 (42)39 (30.6, 48.6) 6 (5) 42 (34)Difference in ORR, % (80% CI)¶6.04 (−1.95, 14.02)Median DoR‡ (95% CI)*, mo12.4 (8.3, 17.9)14.4 (9.7, NR)*Brookmeyer-Crowley method; †based on profile-likelihood confidence limits; ‡according to RECIST v1.1; §Clopper-Pearson method; ¶Mantel and Haenszel method with Sato variance estimator. CI, confidence interval; CR, complete response; mo, months; NR, not reached; PR, partial response. Open table in a new tab *Brookmeyer-Crowley method; †based on profile-likelihood confidence limits; ‡according to RECIST v1.1; §Clopper-Pearson method; ¶Mantel and Haenszel method with Sato variance estimator. CI, confidence interval; CR, complete response; mo, months; NR, not reached; PR, partial response. In this follow-up analysis, dostar + CT continues to demonstrate strong clinical efficacy with no unexpected safety signals. In addition, a numerical trend in OS favoring dostar + CT vs pembro + CT was observed. 1. Peters, S et al. IOTECH 2022;16(S1):100162 Funding:GSK (213403).