Meta-analysis for the association between overall survival and progression-free survival in gastrointestinal stromal tumor.

PURPOSE: Gastrointestinal stromal tumor (GIST) is a relatively rare tumor that is treated with targeted therapies in advanced stages. Randomized clinical trials (RCT) often require long follow-up and large sample sizes to evaluate overall survival (OS), the gold-standard measure of treatment efficacy. However, changes in therapy following disease progression may complicate survival assessments. Establishing surrogate endpoints may facilitate the drug approval and availability of new efficacious treatments; however, no published studies have investigated this topic in unresectable and/or metastatic GIST. EXPERIMENTAL
DESIGN: A systematic literature review identified 14 RCTs and five observational studies of sufficient methodologic quality published between January 1995 and December 2013 (29 treatment arms; 2,189 patients). Weighted linear regression was used to evaluate the relation between median OS and median progression-free survival (PFS) for all arms combined and stratified by treatment line, treatment type, and quality score.
RESULTS: Median OS and PFS were positively related with a correlation of 0.91. The association was still moderate (correlation 0.72) after eliminating four influential data points. In stratified analyses, correlation of OS and PFS was greater in later lines of therapy (first line = 0.52; second line = 0.80; third- and later-line = 0.70) and imatinib showed a stronger association (0.91) than other evaluated treatments (-0.26 to 0.69).
CONCLUSION: This analysis identified a strong relationship between median OS and PFS, especially in later lines of therapy. Findings suggest that PFS could serve as a surrogate marker for OS; however, analyses of patient-level data are needed to establish its validity in GIST. ©2014 American Association for Cancer Research.