Utility Of Effect Size To Define Populations With Durable Clinical Outcomes Across Trials Of Metastatic Colorectal Cancer.

e15104 Background: ASCO defined clinically meaningful trial endpoints in colorectal cancer (CRC) as absolute overall survival (OS) of 3-5 months and OS hazard ratio (HR) ≤0.67. Few tools predict clinically distinct subgroups from heterogeneous populations. Effect size (Glass’s delta) calculates the absolute difference in clinical outcomes relative to the standard deviation of control group. We hypothesized that effect sizes ≥1 would be useful in predicting subgroup populations of clinical significance for trials with an indeterminate HR. Methods: Prospective phase II-III trials in metastatic CRC were queried from clinicaltrials.gov and cataloged by clinical outcomes of PFS and OS. Effect size was back-calculated from trials reporting 95% confidence intervals and compared with absolute difference in clinical outcome and hazard ratio. Results: 46 prospective trials were evaluable including 49% with biomarker selection, 57% in the first-line setting, and more commonly studied with targeted therapy over chemotherapy (70 v. 30%; p < 0.001). EGFR inhibitors were studied in 34% and VEGF inhibitors in 23%. Both effect size and HR correlated similarly with PFS (correlation coefficients, R = 0.65 vs. 0.76) and OS (R = 0.83 vs. 0.84) across these studies. Of studies with an indeterminate HR (0.69-0.86; n = 24), 10 studies had a significant effect size (Glass’s delta > 1), indicating the potential for sub-populations with improved clinical benefit within the experimental treatment group. These included 4 trials with planned biomarker analyses. Remaining trials were enriched for studies investigating anti-angiogenic therapies (83% vs. 23%, p = 0.001), including bevacizumab, aflibercept, regorafenib, and ramucirumab. Conclusions: Effect size holds potential as a measure to indicate the presence of subpopulations of patients benefiting in clinical trials. Those trials with a significant effect size despite an indeterminate HR should be examined closely for such populations. In CRC, effect size indicates the potential of a subgroup of patients who benefit significantly from anti-angiogenic agents. Further investigations are needed to validate effect size as a tool to delineate improved clinical outcomes from heterogeneous populations.