Continuous Longitudinal Tumor Measurement-Based Phase II Endpoints for Predicting Overall Survival (OS) Using the RECIST 1.1 Data Warehouse

• Activity Number: 177
• Type: Contributed
• Date/Time: Monday, August 5, 2013 : 10:30 AM to 12:20 PM
• Sponsor: Biopharmaceutical Section
• Author(s): Ming-Wen An*+ and Sumithra Mandrekar and Daniel J. Sargent and Xinxin Dong and Axel Grothey and Jan Bogaerts
• Companies: Vassar College and Mayo Clinic and Mayo Clinic and University of Pittsburgh and Mayo Clinic and EORTC
• Keywords: tumor response ; endpoint ; continuous ; c-index ; RECIST

Abstract

Tumor response, a common Phase II clinical trial endpoint, in solid tumors is assessed via the Response Evaluation Criteria in Solid Tumors (RECIST), which categorizes continuous tumor measurements (TM). The high failure rate of Phase III trials (e.g. 50-60%), a major obstacle in drug development, suggests inappropriate choice of Phase II endpoint. We previously reported that alternate cutpoints and categorical metrics to RECIST are no better in predicting OS. We now seek to identify simple, clinically relevant continuous metrics based on longitudinal TM. Data from 13 breast, lung, and colon trials from the RECIST 1.1 data warehouse were landmarked at 12-weeks (w). Metrics included slope from 0-6w, slope from 6-12w, and indicator of any increase (=5% vs. < 5%) in tumor from a prior assessment. We fit separate Cox models by tumor type, stratified by trial and number of baseline target lesions (< 3 vs. >3) and adjusted for baseline tumor size. No metric was better than RECIST in terms of concordance-index despite statistically significant hazard ratio. Ongoing work is needed to address statistical challenges including missing data, multicollinearity, and using early timepoints.