OBJECTIVES: This study was designed to evaluate differences in ORR and to evaluate the relationship with longer-term outcomes of PFS and OS among patients with MCL. This information would inform the value of ORR as a surrogate endpoint for longer-term outcomes in this disease.

METHODS: Randomized trials included in this analysis were identified from the literature that reported ORR with either PFS or OS. Trial data were manually extracted, and survival curves for PFS and OS were digitized to obtain hazard ratios. Multiple approaches were undertaken to evaluate the association between outcomes: bivariate random effects meta-analysis models with non-informative prior, Wishart prior, and product normal formulation, which include heterogeneity for both target and surrogate endpoint, as well as Pearson product (r) correlation. Correlation statistics were generated using R and JAGS.

RESULTS: A total of 14 studies were identified in the literature. PFS and ORR data were reported for 2,196 patients in 12 trials, and OS and ORR data reported for 2,015 patients in 13 trials. Pearson correlation statistics demonstrated the strongest relationship between ORR and logPFS (r = -0.62; 95% CI: -0.88 to -0.07). The relationship between ORR and logOS (r = -0.20; 95% CI: -0.68 to 0.39) and logPFS and logOS (r = 0.28; 95% CI: -0.42 to 0.77) were low. Bayesian bivariate analyses were limited by the lack of individual patient-level data and trial-to-trial variability.

CONCLUSIONS: In this study, a moderate relationship was identified between ORR and PFS. While both ORR and PFS are events that occur within the treatment period of clinical trials, OS occurs later, and can be influenced by future events. Therefore, the lack of a direct relationship between ORR and survival events with less temporal proximity is not unexpected. Findings should be interpreted with caution due to the small number of studies.