Patient-Reported Outcomes as Early Indicators of Disease Progression and Predictors of Survival via Machine Learning in Breast Cancer

Patient-reported outcomes (PROs) offer a non-invasive, low-cost way to capture patients’ experiences of symptoms, functioning, and quality of life. Yet, their potential as early indicators of tumor burden, disease progression, and survival remain unclear. In this study, we retrospectively analyzed 450,147 longitudinal PRO entries from 2,738 breast cancer patients pooled from four clinical trials, including both early- and late-stage disease, covering 15 PRO measures assessing symptoms, functioning, and quality of life. Among patients with radiographically confirmed disease progression, 90.6% experienced at least one deterioration in PROs prior to relapse detection (median PRO deterioration time 85 days vs. relapse time 1,338 days), indicating that PROs often worsen before imaging-confirmed relapse. Using Cox proportional hazards models, PRO deterioration was significantly associated with metastatic sites, tumor burden, and survival. Functional PROs were more affected by metastatic sites, whereas symptomatic PROs showed stronger associations with tumor burden and survival outcomes. Appetite loss had the strongest correlation with tumor burden, while fatigue was the most prognostic symptom for both overall survival (OS) and progression-free survival (PFS). The predictive value of PROs for OS was then evaluated using gradient boosting machine learning models. Models integrating PRO deterioration times of all subscales together with PFS achieved the best predictive performance, correctly identifying patient survival outcomes in over 95% of cases (AUC-ROC = 0.954), outperforming models using PROs (AUC-ROC = 0.838) or PFS alone (AUC-ROC = 0.896). This indicates that integrating PROs with PFS enhances the prediction of OS, providing a more powerful approach than using either measure alone. Together, these findings provide quantitative evidence that PROs can serve as early and complementary predictors of disease progression and survival, supporting their use as patient-centered biomarkers in breast cancer management. Our findings align with FDA and EMA efforts to integrate PROs into oncology endpoints, supporting more patient-centered regulatory evaluation.