A mechanistic model of curative combination therapy explains lymphoma clinical trial results

Combinations of chemotherapies are used to treat many cancer types as they elicit higher cure rates and longer responses than single drugs. Several rationales contribute to the efficacy of combinations, including overcoming inter-patient and intra-tumor heterogeneity and improving efficacy through additive or synergistic pharmacological effects. We present a quantitative model that unifies these phenomena to simulate the clinical activity of curative combination therapies. This mechanistic simulation describes kinetics of tumor growth and death in response to treatment and outputs progression-free survival (PFS) distributions in patient populations. We applied this model to first-line combination therapy for Diffuse Large B-Cell Lymphoma, which is cured in most patients by the 5-drug combination RCHOP. This mechanistic model reproduced clinically observed PFS distributions, kinetics of tumor killing measured by circulating tumor DNA, and the adverse prognostic effect of tumor proliferation rate. The outcomes of nine phase 3 trials of new therapies combined with RCHOP were accurately predicted by the model, based on new therapies efficacies in trials in patients with relapsed or refractory disease. Finally, we used the model to explore how drug synergy and predictive biomarkers affect the chance of success of randomized trials. These findings show that curative combination therapies can be understood in quantitative and kinetic detail, and that predictive simulations can be used to aid the design of new treatment regimens and clinical trials in curative-intent settings. SIGNIFICANCEA novel model that incorporates pharmacological interactions in the presence of inter-patient and intra-tumor heterogeneity explains and predicts combination clinical trial outcomes of curative regimes used to treat Diffuse Large B-cell lymphoma. This model can be used to understand and inform optimal design of drug combinations and clinical trials.