A mechanistic digital twin model for epigenetic therapy optimization in triple-negative breast cancer

Abstract

Epigenetic therapies offer a promising approach to cancer treatment by modulating chromatin states that govern tumor cell identity, plasticity and therapeutic response. However, predicting and optimizing the effects of such interventions remains challenging. Here, we developed a digital twin framework that integrates mechanistic models of chromatin regulation, in vitro cell-state and treatment response data, and pharmacokinetics to simulate tumor progression and therapeutic response. We applied this framework to triple-negative breast cancer (TNBC), an aggressive disease in which chromatin dysregulation contributes to tumor progression, and investigated combination treatment with an EZH2 inhibitor promiting chromatin opening and an AKT inhibitor, which together enhance expression of GATA3 and BMF. Parameterized and validated using in vitro treatment response data, the model enables in silico clinical trials of alternative combination regimens and treatment schedules. These simulations identify regimens that achieve comparable therapeutic effects to reference schedules while substantially reducing cumulative drug exposure. We further demonstrated the digitan twin's ability of identifying personalized therapeutic strategies by incorporating patient-specific treatment-response data. Our work establishes a mechanistic digital twin framework for predicting tumor responses to chromatin-modifying therapies and provides a quantitative approach for optimizing treatment combinations and schedules across diverse cancer contexts.