Tumor twins in breast cancer: toward hybrid biological-digital models for adaptive precision oncology

Precision oncology in breast cancer has improved patient stratification but remains limited by static biomarkers that fail to capture tumor heterogeneity and therapy-induced evolution. Tumor twins have emerged as patient-specific models designed to mirror tumor behavior and predict treatment response. Two complementary approaches have matured: biological tumor twins, including patient-derived organoids and tumor-on-chip systems, and digital tumor twins integrating clinical, imaging, and multi-omics data into computational models. However, these paradigms have largely evolved independently, and current computational tools often function merely as one-way enabling technologies rather than dynamically coupled systems. In this Review, we synthesize recent advances in both fields and introduce the conceptual framework of genuine hybrid tumor twins. These bidirectionally coupled, closed-loop systems leverage real-time non-destructive biological sensing to continuously calibrate computational simulations, which in turn enable predictive digital models to actively guide experimental and clinical interventions. We address critical biological and operational bottlenecks, including culture-induced experimental bias, validation, data interoperability, and clinical implementation. Finally, we outline a practical three-phase translational roadmap spanning technical platform harmonization, retrospective uncertainty calibration, and adaptive prospective clinical trials. By bridging continuous biological monitoring with dynamic computational forecasting, hybrid tumor twins provide a realistic blueprint to advance adaptive, real-time precision oncology in breast cancer.