Functional Tumor Volume by Subtype Improves Prediction of Pathological Complete Response in Breast Cancer

Motivation: In the breast cancer I-SPY2 clinical trial, improvement of predictive capability for pathological complete response (pCR) would lead to improved drug candidate identification and treatment length assignment. Goal(s): Improve functional tumor volume (FTV) predictive performance by optimizing parameter thresholds by receptor subtype. Approach: This retrospective study used DCE-MRI to calculate FTV at different enhancement curve parameter thresholds. Resulting FTV metrics were used in prediction models for pCR. Optimal thresholds were selected to maximize receiver operating characteristic area under the curve. Results: Each factor category (subtype, enhancement curve thresholds, and measurement timepoint) were important for improving pCR predictive performance. The maximum AUC was 0.77. Impact: In the breast cancer I-SPY2 clinical trial, better pathological complete response (pCR) prediction would lead to improved treatment redirection and treatment sparing, improving patient outcomes. In this study, we optimized parameters for MRI functional tumor volume calculation to improve predictions.