Abstract B027: Modeling the emergence of therapy resistance with fitness seascapes

Abstract

Abstract One approach to the study of therapy resistance is through an evolutionary lens, where novel sub-types emerge through a random process and are selected for by an increased fitness. Computational models are used to understand the parameter spaces that facilitate resistance, optimize therapy regimens, and predict the evolutionary trajectory. Fitness landscapes are structures that map genotype to fitness and are commonly used in computational models of evolution. Most fitness landscape models assume the environment is constant. However, this obscures evolutionary trade-offs that depend on environmental change, i.e., where a resistant mutant trades a lower growth rate in the absence of drug in exchange for drug resistance. Furthermore, a disease agent population in a patient will never experience a constant drug concentration; the drug concentration will vary according to pharmacokinetic effects and spatial relationships to blood vessels. Here, we explored fitness seascapes as an extension to fitness landscapes. Fitness seascapes model a variable genotype-phenotype mapping. In this work, we used fitness seascapes that map genotypes to drug dose-response curves. In contrast to fitness landscapes, the fitness rank-order of genotypes changes as a function of drug concentration, representing evolutionary tradeoffs. Using a novel computational platform, we showed that modeling evolutionary tradeoffs with fitness seascapes dramatically impacted the evolutionary trajectory of a population subject to changing drug concentrations. We also showed how mutant selection windows, a commonly studied structure in evolutionary medicine, are embedded in fitness seascapes. Using an empirical fitness seascape measured in genetically engineered S. cerevisiae, we then explored the utility of fitness seascapes in two in silico experiments. In the first experiment, we subjected an evolving population to varying rates of change of drug concentration, finding that populations experiencing a lower rate of change were more likely to become resistant, while those experiencing a higher rate of change were more likely to go extinct. This aligns with similar previously published in vitro experiments. Next, we simulated patients undergoing therapy with varying levels of drug regimen nonadherence using a pharmacokinetic model. We found that a higher rate of drug regimen nonadherence was associated with a higher rate of resistance, while a higher rate of adherence was associated with a higher rate of elimination of the disease agent. Future work may reveal patterns of nonadherence that promote resistance. Here, we have demonstrated the utility of fitness seascapes in studying resistance. Our results highlight the importance of evolutionary trade-offs when modeling the emergence of resistance under varying drug concentrations. These results have implications for diseases across kingdoms, including viral and bacterial infections and cancer. In the future, fitness seascapes may be used to optimize drug dosing regimens or understand differences in metabolism in silico. Citation Format: Eshan S. King, Emily Dolson, Jacob G. Scott. Modeling the emergence of therapy resistance with fitness seascapes [abstract]. In: Proceedings of the AACR Special Conference on the Evolutionary Dynamics in Carcinogenesis and Response to Therapy; 2022 Mar 14-17. Philadelphia (PA): AACR; Cancer Res 2022;82(10 Suppl):Abstract nr B027.