Time-dependent patterns of failure and treatment benefit from adjuvant therapy for resectable colon cancer: Lessons from the 20,800-patient (pt) ACCENT dataset

Abstract

4008 Background: To explore and inform 3 critical issues facing trialists and practicing oncologists testing adjuvant therapy (AT) for stage II and III colon cancer: 1) The nature and duration of AT benefit, 2) detailed information on survival and recurrence (rec) rates, including long-term event rates (> 10 years (yrs)), and 3) adequacy of statistical models for power and sample size calculations based on a disease-free survival (DFS) endpoint. Methods: The dataset assembled by the Adjuvant Colon Cancer Endpoints (ACCENT) Group, a collection of individual pt data from 18 trials and over 20,800 pts testing 5-FU based AT for pts with stage II and III colon cancer, was analyzed. Results: 1) Significant DFS benefit from AT were manifest in the first two years post randomization; after 2 years DFS rates in the AT and control groups were not significantly different. Significant Overall Survival (OS) benefit of AT was consistent over the 8 year follow-up period. 2) After 5 yrs, rec rates were <1.5%/yr, and after 8 yrs, < 0.5%/yr. 3) For DFS, a Weibull model provided superior model fit compared to the standard exponential model. In the range of Weibull values seen in trials in the dataset, however, the use of the Weibull model did not meaningfully impact power and sample size (<2% impact on power for the same sample size), assuming a constant effect of AT on DFS. Additionally, modeling a non-constant AT effect (as observed for DFS) had a similar negligible impact on statistical power compared to the standard exponential model, for the same sample size. Conclusions: AT provides significant benefit for DFS, primarily within the first 2 years, translating into long-term OS benefit. Long term follow-up for rec yields few events. Despite some increased complexity, existing methods remain appropriate for power and sample size determination for the DFS endpoint, pending further confirmation. No significant financial relationships to disclose.