Fitness seascapes are necessary for realistic modeling of the evolutionary response to drug therapy

Abstract

Predicting the evolution of drug resistance in infectious diseases may enable us to make rational drug choices to avoid resistance and exploit known molecular mechanisms of resistance. Fitness landscapes are commonly used in computational studies to model the genotype-fitness mapping. Canonical fitness landscapes do not intrinsically model varying selection pressure. However, a challenge to predicting the emergence of drug resistance is that disease agents in a patient will never experience a constant environment - the selection pressure imposed by a drug will vary as a result of the drug pharmacokinetic profile, dosing schedule, and spatial concentration gradient. Furthermore, because of evolutionary costs of resistance and different levels of genotype-specific resistance, different drug concentrations may be optimal for the fitness of different genotypes. Explicit modeling of these trade-offs and the resulting rank-order changes in fitness may be important to accurately predict the evolution of a disease population within a patient or other heterogeneous environments. Fitness seascapes extend the fitness landscape model by allowing the mapping to vary according to an arbitrary environmental parameter (e.g., drug concentration), allowing us to model the evolution of resistance with realistic pharmacological considerations. Here, we explore the importance and utility of fitness seascapes in predicting the emergence of resistance. First, we show how modeling genotype-specific dose response curves is necessary to accurately predict evolutionary outcomes in changing environments. Then, using an empirical fitness seascape measured in engineered E. coli, we performed computational experiments observing the impact of the rate of change in drug concentration and simulated patient nonadherence on the probability of evolutionary escape. We found that a greater rate of change in drug concentration resulted in a lower rate of resistance, or a lower rate of evolutionary escape. In simulated patients, higher rates of drug regimen nonadherence were associated with greater rates of resistance. Our work integrates an empirical fitness seascape into an evolutionary model with realistic pharmacological considerations. Future work may leverage this platform to optimize dosing regimens or design adaptive therapies to avoid resistance.

Diverse mutant selection windows shape spatial heterogeneity in evolving populations

“…Dose response curves differ in their IC 50 and drug-free growth rates. Empirical fitness seascapes bearing a similar structure and demonstrating fitness tradeoffs have been reported by others 7, 19, 25 .…”

“…Others have shown computationally and experimentally that drug diffusion gradients and differential drug penetration, which permit heterogeneous MSWs, facilitate antibiotic resistance 11, . Considering drug pharmacokinetics in time, Nande and Hill, among others, have shown that drug absorption rate and patient nonadherence can impact the evolution of resistance 12, 25 . Our results suggest that this is likely a function of both the time spent in the MSW and, in the case of nonadherence, the drug concentration dropping back down into a MSW.…”

“…We next sought to investigate MSWs in physiologically-relevant temporal and spatial models of drug kinetics. First, to illustrate MSWs in time, we simulated a patient undergoing a daily drug dosing regimen with a constant rate of nonadherence, using x (mm) methods described by King et al 25 . Briefly, we model the serum drug concentration using a 1-compartment pharmacokinetic model and convolve this model with an impulse train with each impulse representing a correctly administered drug dose (see Methods).…”

Promoting extinction or minimizing growth? The impact of treatment on trait trajectories in evolving populations

“…Accounting for their temporal variability (e.g. under the effect of treatment), then sometimes referring to them as fitness seascapes, has important consequences for the understanding of adaptation, such as resistance evolution (Lässig et al, 2017; King et al, 2022). For example, Hemez et al (2020) found in a simulation study that the drug mode of action (bacteriostatic vs. bactericidal) was changing the shape of the fitness landscape.…”

Evolutionary druggability: leveraging low-dimensional fitness landscapes towards new metrics for antimicrobial applications

“…Future work may seek to elaborate on these and other studies that highlight the role of environments in crafting plasticity and adaptive evolution [53][54][55]. Such efforts may include studies of "fitness seascapes" and evolution in fluctuating environments, as has been examined by others [56][57][58]. Like many others, we believe that we are only at the beginning of a larger, holistic effort to supplement existing metrics and concepts developed within clinical medicine, all towards more effective therapies that can improve outcomes at the bedside.…”