Evaluation of clinical endpoints as surrogates for overall survival in patients treated with immunotherapies.

Abstract

e14557 Background: In immuno-oncology (IO), the correlation between clinical trial endpoints, specifically, objective response rate (ORR), disease control rate (DCR) or progression free survival (PFS), and overall survival (OS) is poorly understood. The effect of IO agents as opposed to chemotherapy, is not on tumor cells, but on immune cells and OS benefit has been observed in absence of PFS benefit. However, decisions on registration trials often rely on PFS from Phase II trials. Methods: We conducted a systematic literature review with PubMed and Embase (Jan. 2005–Nov. 2016), supplemented with oncology conference proceedings (2014–2016). Eligible studies were randomized controlled trials (RCT) that investigated ≥1 immune checkpoint blockers (CBs) targeting programmed death proteins (PD-1/PD-L1) and cytotoxic T-lymphocyte-associated protein 4 (CTLA-4), and reported their relative effect on OS and on ≥1 of the clinical endpoints, DCR, ORR, and PFS. Log-transformed hazard ratios were fitted using weighted regression models to determine the power of relative effects on clinical endpoints to predict OS effects, with correlation coefficients estimated and presented with adjusted R2 (high values indicating better goodness-of-fit). Results: This analysis included 18 RCTs involving 7140, patients. Most studies (10/18) evaluated the efficacy of CBs vs conventional chemotherapy, whereas 8 studies compared the efficacy of ≥1 CB. Among trials that evaluated anti-CTLA-4, the adjusted R2 for the relative efficacy of CBs on DCR, ORR, PFS, and relative efficacy of CBs on OS were 0.160 ( P= 0.156), 0.016 ( P= 0.332), and 0.000 ( P= 0.623) respectively. Among trials that evaluated either anti-PD-1 or anti-PD-L1, the adjusted R2 were 0.038 ( P= 0.401), 0.066 ( P= 0.251), and 0.432 ( P= 0.032), for DCR, ORR and PFS respectively. Among trials that evaluated CBs in melanoma, the adjusted R2 were 0.030 ( P= 0.267), 0.028 ( P= 0.279), and 0.192 ( P= 0.154), for DCR, ORR, and PFS respectively. Conclusions: No clear correlations were observed between relative effects of conventional clinical endpoints and OS for CBs. New surrogate endpoints may be needed to better predict OS benefit for CBs and other forms of cancer immunotherapy.