Abstract Background: Few effective treatment options for advanced NSCLC are available following frontline immune checkpoint inhibitor (ICI) -based therapy. The randomized, phase II Lung-MAP S1800A study of RP versus SOC for pts with NSCLC previously treated with ICI demonstrated benefit in overall survival (OS) with an improved toxicity profile over SOC. SWOG S2302 Pragmatica-Lung was pragmatically designed to evaluate OS while increasing representation in trial participation. The interim analysis was previously reported. Methods: S2302 is a registration-intent randomized phase III trial for pts with advanced NSCLC who previously received ICI for at least 84 days and platinum-based therapy, randomized to SOC or RP, stratified by immediate prior therapy including ICI (yes/no) and PS (0/1 v. 2). The pragmatic design led to eligibility focused on stage, prior therapy and safety. Laboratory assessment and imaging were not required. The primary objective was OS. The secondary objectives were to compare OS between the arms within key subgroups, including squamous cell carcinoma (SCC) and to summarize reports of serious and unexpected high-grade treatment-related AEs. The accrual goal was 800 based on 90% power for an HR of 0. 77 using a 1-sided 2. 5% level log-rank test. Full information was 616 OS events. Results: S2302 enrolled 838 pts with 419 in each arm from March 2023 to December 2024. Study data were released at an interim analysis for futility. Median age (range) was 68 (34-88), 22% non-white /13% Black, 15% rural, 29% SCC, 63% adenocarcinoma, 81% had ICI as the most recent treatment, 13% had PS2. With 582 deaths reported, OS was not different between the arms (HR 1. 04 0. 88-1. 22 p= 0. 66; median OS (mOS) was 10. 1 months (mos) for RP and 10. 0 mos for SOC). In SCC, the HR (95% CI) was 0. 90 0. 66-1. 22 p= 0. 25 with mOS of 10. 7 mos for RP and 9. 6 mos with SOC. In nonSCC, the HR (95% CI) was 1. 10 0. 91-1. 34 p= 0. 84 with mOS of 9. 8 mos for RP and 10. 6 mos with SOC. Conclusions: The pragmatic trial design led to rapid accrual with increased participant representativeness. RP did not improve OS overall but was not worse than SOC. Of note, for SCC, the estimated HR is 1 and the confidence interval is consistent with non-inferiority bounds. We interpret that RP for SCC may provide an option equivalent to SOC with better tolerability. Support: NIH/NCI/NCTN grants U10CA180888 and U10CA180819; and in part by Merck Sharp Part 2 (Late-Breaking, Clinical Trial, and Invited Abstracts) ; 2026 Apr 17-22; San Diego, CA. Philadelphia (PA): AACR; Cancer Res 2026;86 (8Suppl): Abstract nr CT140.
Reckamp et al. (Fri,) studied this question.