Synergy of biological and mathematical models to optimize checkpoint immunotherapy
Checkpoint blockade immunotherapy (CBI) provided long-term progression-free survival to one-third of metastatic melanoma patients and is now approved to treat more than 20 oncologic indications. However, most patients treated with CBI eventually become resistant. Since immunosuppressive reactions within the tumor environment are a significant cause of resistance to therapy, there is a pressing need to optimize CBI. One key problem in optimizing CBI is that the number of variables to consider is substantial and continuously growing. These variables include immune circuitries promoting or opposing tumor rejection, their interplay, the drugs to administer, their dosage, and at what time during cancer immunotherapy. A significant concern for the years to come is that such variables will become so numerous that identifying optimal CBI regimens in preclinical and clinical studies would exceed available money, patients, and time.
Here, we propose to tackle this issue using math-aided immunological experimentation. The Marangoni (immunology) and Lowengrub (mathematics) labs teamed up to generate a mathematical model of immune reactions within the tumor that is deeply rooted in biology, is fitted to experimental data, and correctly recapitulates the outcome of immunotherapies, including PD-1 and CTLA-4 blockade. The model generated more than 1100 sets of immune parameters representing individual immune responses to tumor and checkpoint immunotherapy (eMice). The eMice variably responded to simulated PD-1 blockade: some showed complete tumor control, and others no control at all, mimicking the variable response to immunotherapy observed in mice and patients. We are now analyzing the parameters enriched in responding eMice, which are candidate biomarkers to predict a good response to CBI in patients. We are also using our model to find the combination of CBI type, dose, and administration time predicted to be the most effective. We will validate all these predictions using biological experiments.