Surrogacy of intermediate endpoints for overall survival in randomized controlled trials of first-line treatment for advanced soft tissue sarcoma in the pre- and post-pazopanib era: A meta-analytic evaluation

Abstract

Background: Overall survival is the true endpoint for most randomized controlled trials (RCTs) of malignant tumors, whereas progression-free survival (PFS) is considered the most reliable surrogate endpoint for overall survival (OS). The present study aimed to evaluate the correlation between surrogate endpoints and OS in randomized trials of first-line chemotherapy with doxorubicin (DOX), the standard treatment for advanced and metastatic soft tissue sarcomas (ASTS), using a meta-analytic approach. Methods: In a systematic review, we identified RCTs of first-line chemotherapy for ASTS that compared single-agent doxorubicin (DOX) with other chemotherapy regimens, and were published in English during January 1974-December 2017. A meta-analysis was performed to evaluate the efficacy of first-line treatments for ASTS. Surrogacy of the intermediate endpoints for OS was investigated using weighted linear regression analysis. Correlation strength was examined using the coefficient of determination (R 2 ). Results: Twenty-seven randomized trials, comprising 6156 patients (3371 patients in the experimental arm and 2785 patients in the DOX arm) were identified. The hazard ratios for OS and PFS showed that the efficacy of treatment for ASTS was not significantly different between standard DOX and experimental treatments. The median OS was significantly prolonged in RCTs published after 2012 when pazopanib was approved for treating ASTS. The median PFS, however, did not differ significantly. The correlation between PFS and OS was moderate (R 2 = 0.557), but better than that between OS and 3-month PFS, 6-month PFS, and response rate (R 2 = 0.200, 0.073, and 0.278, respectively). The correlation between PFS and OS tended to be more favorable in RCTs published after 2012 (R 2 = 0.586 and 0.459, respectively). Conclusions: The trial-level correlation between PFS and OS was only modest; it tended to be better in RCTs published after 2012. While the effective lines of chemotherapy and the introduction of new drugs prolonged OS but not PFS, PFS is a better surrogate than other intermediate endpoints in the first-line ASTS trials even in the post-pazopanib era. Although this does not negate the need for more reliable surrogate endpoints for OS.