Analysis of response-related endpoints in trials of first-line medical treatment of metastatic colorectal cancer.

BACKGROUND: Tumor radiologic response after systemic chemotherapy has been used as endpoint of trials of patients with metastatic colorectal cancer (mCRC), which can report the best overall response rate (ORR) and the disease control rate (DCR) by RECIST criteria as well as the early tumor shrinkage (ETS). The present study perform a trial-level analysis to verify whether such response-related endpoints are predictive of overall survival (OS). METHODS: After a systematic search, randomized clinical trials (RCTs) were selected each time they evaluated the three response endpoints and progression-free survival (PFS). Two arms per trial were selected, and the correlation between the difference in each endpoint and the difference in OS was calculated. The analysis then evaluated the effects of treatment on ∆ORR, or ∆DCR, ∆ETS, ∆PFS, and on ∆OS, using separate linear regressions for each of them, and the proportion of variability explained (R2trial) on OS for each of the four endpoints was calculated. RESULTS: The systematic review of the literature led to the selection of 12 RCTs, 7 phase-3 and 5 phase-2. ETS reported a different performance in the entire sample compared to phase-3 trials (R2trial = 0.172 vs. 0.842), differently from DCR (R2trial = 0.541 vs. 0.816) and ORR (R2trial = 0.349 vs. 0.740). Surprisingly, PFS predicted OS with a weak correlation, which was not significant in the subgroup of phase-3 studies (R2trial = 0.455 vs. 0.466). CONCLUSION: The results of the present trial-level analysis report a good performance of two response-related endpoints, DCR and ETS, and suggest that they could be differently used depending on the setting of disease and the type of medical treatment.