Model‐Based Meta‐Analysis of Objective Response Rate and Survival Endpoints to Compare PD ‐1 and PD ‐ L1 Treatment Outcomes in Non‐Small Cell Lung Cancer

Programmed cell death (PD) protein (ligand [L])-1 inhibitors are established treatments for metastatic non-small cell lung cancer (mNSCLC). In oncology, progression-free survival (PFS) and objective response rate (ORR) are often used as surrogates for overall survival (OS) to inform clinical development; however, there remains uncertainty in the concordance between these endpoints. This study evaluated the impact of a broad set of PD-(L)1 inhibitors on efficacy, explored the relationship between ORR and survival endpoints, and compared PD-1 and PD-L1 treatments for mNSCLC. A dataset of 114 studies was used to conduct a sequential two-stage model-based meta-analysis (MBMA). Firstly, an MBMA with mixed-effects logistic regression was applied to evaluate treatment-specific and clinical covariate effects on ORR. Secondly, MBMAs for OS and PFS were conducted with a mixed-effects semi-parametric proportional hazard approach using digitized Kaplan-Meier curves, with the treatment type, covariates, and ORR as inputs. ORR was demonstrated to be a significant predictor of OS and PFS. Simulations of head-to-head comparisons of treatment types were conducted using these models. Trends in predicted outcomes numerically favored PD-1 over PD-L1 treatments, but differences were not statistically significant. These findings support evidence-based decision-making for late-stage trial designs using ORR data from earlier phase trials, enabling benchmarking of emerging data by adjusting for explained and unexplained sources of variability in existing and emerging data.