Surrogate endpoints and outcomes in modern small cell lung cancer trials.

person Seren Durer University Hospitals Seidman Cancer Center, Case Western Reserve University, Cleveland, OH info_outline Seren Durer, Pingfu Fu, Zhengyi Chen, Afshin Dowlati

Authors person Seren Durer University Hospitals Seidman Cancer Center, Case Western Reserve University, Cleveland, OH info_outline Seren Durer, Pingfu Fu, Zhengyi Chen, Afshin Dowlati Organizations University Hospitals Seidman Cancer Center, Case Western Reserve University, Cleveland, OH, Case Western Reserve University School of Medicine, Cleveland, OH, Case Western Reserve University School of Medicine, Department of Population and Quantitative Health Sciences, Cleveland, OH Abstract Disclosures Research Funding No funding sources reported Background: Overall Survival (OS) may not always be a feasible primary endpoint. Endpoints such as Overall Response Rate (ORR), Disease Control Rate (DCR=CR+PR+SD), Progression Free Survival (PFS), are frequently used in the assessment of treatments. However, improvements in DCR, ORR, PFS may not translate into OS benefit. Resilient trial showed a doubling of ORR to PEGylated liposomal irinotecan 1 compared to control but no improvement in PFS/OS. Furthermore, these associations may differ between 1 st line (1L) and relapsed (2L) disease settings. This study analyzes the relationship between ORR, DCR, PFS, and OS in extensive-stage SCLC. Methods: MEDLINE, Embase, ClinicalTrials.gov were searched to identify randomized, phase III trials between 2014-2024. PRISMA guideline was followed. DCR, ORR, OS, PFS data were extracted. Weighted linear regression models investigated the association between treatment effects on DCR, ORR, OS, PFS. Weighted Pearson correlation analyses were performed under logarithmic transformation of DCR/ORR odds ratio besides DCR/ORR difference between treatment arms, with weights equal to the inverse of variance of DCR/ORR associated with the log PFS HR and the log OS HR. Logistic regression models were used to derive DCR/ORR odds ratios. Hazard ratios (HR) estimated from Cox proportional hazards regression models were used. All tests are two-sided; p-value < 0.05 is significant. Results: 115 trials were identified. 22 met the criteria. DCR: 1) 1L and 2L: DCR difference between arms and the log odds ratio and log PFS HR were strongly correlated: –0.63 (P = 0.001) for the DCR difference model; –0.64 (P = 0.0008) for the log odds ratio model. No correlation was found between DCR and OS; 2) 1L: No correlation between DCR and PFS & OS was found; 3) 2L: DCR difference between arms and the log odds ratio and log PFS HR were correlated: –0.84 (P = 0.01) for the DCR difference model; –0.84 (P = 0.01) for the log odds ratio model. ORR:1) 1L and 2L: ORR difference between arms and the log odds ratio and log PFS HR were correlated: –0.43 (P = 0.036) for the ORR difference model; –0.51 (P = 0.011) for the log odds ratio model; 2) 1L: No correlation between ORR and PFS & OS was found; 3) 2L: No correlation between ORR and PFS & OS was found. PFS:1)1L and 2L: The Pearson correlation coefficients between log HR of PFS and log HR of OS: 0.6 (p = 0.002); 2) 1L: The Pearson correlation coefficient between log HR of PFS and log HR of OS: 0.77 (p = 0.0005) 3) 2L: The Pearson correlation coefficients between log HR of PFS and log HR of OS: -0.14 (p = 0.732). Conclusions: In the 1L setting, neither DCR nor ORR predict PFS and OS. However, PFS predicts OS. In the 2L, ORR does not predict either OS or PFS; however, DCR predicts PFS and shows a trend with OS. Unlike in the frontline setting, PFS does not predict OS in the second-line setting. Reference: 1 Rudin CM, Dowlati A, Chen Y, et al.161O RESILIENT part 2: A randomized, open-label phase III study of liposomal irinotecan versus topotecan in adults with relapsed SCLC.