Predicting Overall Survival and Progression-Free Survival Using Tumor Dynamics in Advanced Breast Cancer Patients

Prediction of survival endpoints, e.g., overall survival (OS) and progression-free survival (PFS), based on early observations, i.e., tumor size, may facilitate early decision making in oncology drug development. In this paper, using data from six randomized trials for first- or second-line advanced breast cancer (ABC) treatments with various mechanisms of action, tumor size change from baseline at different observation time points was evaluated as a predictor for survival endpoints using different modeling approaches. The aim is to establish a predictive model where tumor size change from baseline can be used as a treatment independent predictive marker for PFS and OS in first- and second-line ABC. The results showed that tumor size change at single time point (TSP) or up to certain time points as a time-varying covariate (TSTVC) were significant predictors for OS and PFS in the survival models along with other covariates identified for each line of treatment. TSP and TSTVC models performed similarly for first-line treatments; TSTVC performed significantly better for second-line treatments. Eight weeks was selected as the recommended early evaluation time of tumor size change to predict OS and PFS in both first- and second-line treatment, while better prediction can be achieved for first-line OS by using 16 weeks tumor size change. The result of this study is treatment independent and can be used to predict the outcome of the clinical trials using early readout of tumor size change for the classes of drugs that have been evaluated in this study.