Distinguished Lecture Series | Andrea Bild, PhD

Forecasting Tumor Dynamics and Drug Response: A Path to a National Clinical Trial Precision oncology has largely been built around static, single timepoint, biomarkers, even though we know that tumors are not static. They evolve, adapt, and change under the pressure of treatment. Our work began by asking if we could move beyond measuring what a tumor is today and begin forecasting where it is going next. We developed an algorithm called FORESEE to model how resistance evolves in ER+/HER2-negative breast cancer. Using longitudinal molecular and clinical data, we found that very different tumors can converge on shared resistance trajectories, some emerging within weeks of starting therapy. These early changes can forecast later treatment failure. And in experimental models, when we targeted the programs driving these trajectories, we could suppress resistance and redirect tumor evolution. To address tumor evolution in patients, our clinical trials need to be dynamic too. Existing trial infrastructure was not designed to repeatedly learn from a tumor, anticipate its next move, and adapt treatment in response.​​​​​​​ That realization led me to build ADAPT (ADvanced Analysis for Precision cancer Therapy) at ARPA-H, which is a national clinical framework designed to bring longitudinal measurement, tumor-evolution forecasting, and adaptive treatment decisions to patients at scale. Today, ADAPT is open and enrolling patients, bringing a journey that began with moving beyond static biomarkers into the clinic so that we tailor treatments to a changing tumor before resistance takes hold.