Computational Modeling of Skeletal Muscle Adaptation to Changes in Physical Activity

The scientific community has long sought to understand the relationship between muscle adaptation and physical activity through numerous experimentations from the tissue level to the subcellular level. Mechanical overuse results in increases in muscle size (hypertrophy) while disuse results in decreases in muscle size (atrophy). One hope is that by expanding our understanding at the cellular level we will be able to improve muscle hypertrophy or mitigate atrophy for patients in populations such as sports medicine, recreational conditioning, and bed rest patients. Computational modeling provides the opportunity to compile the vast number of individual experimental observations into a comprehensive framework with the ability to predict observed outcomes. There is a need for novel computational modeling frameworks that can predict human skeletal muscle adaptations to various states of activity and can treat each muscle cell type independently. This study seeks to advance our theoretical understanding of this relationship by compiling experimental data on muscle adaptation, postulating an activity-based differential equation that reflects macro insights on the cellular scale, integrating the differential equation into a computational framework, and validating empirically-derived parameters that can accurately predict muscle adaptation to various physical activities. Additionally, a novel relationship between fiber recruitment and exercise intensity was constructed by analyzing protein accretion rates by exercise intensity. These parameters were then used to simulate the effect of eight weeks of bed-rest and eight weeks of RE on muscle fiber area.

By analyzing atrophy and hypertrophy simulations, we confirmed that fiber adaptation to physical activity could be accurately represented by three parameters: protein synthesis per nuclei per day (βs), the rate of protein degradation per fiber CSA (βd), and the number of myonuclei per fiber CSA. Additionally, simulations of muscle adaptation across 29 muscle groups revealed that muscle architecture can be an accurate predictor of muscle atrophy and hypertrophy. Percentages of fiber types I, IIA, and IIB were significantly correlated with normalized hypertrophy (R2 = 0.66, p < 0.05 for all). The strongest predictor of muscle hypertrophy included fiber type distributions as well as initial CSA (R2 = 0.90, p < 0.05). Initial fiber CSA of all fiber types were significantly correlated with muscle atrophy (Type I: R2 = 0.28, p = 0.003; Type IIA: R2 = 0.33, p = 0.001; Type IIB: R2 = 0.33, p = 0.001). Simulations also revealed that muscle hypertrophy was significantly correlated to muscle atrophy (R2 = 0.9318, p < 0.001), indicating that muscle groups were more prone to either atrophy or hypertrophy based on their architecture. Simulations of detraining and subsequent retraining support the theory of muscle memory, myonuclei retained during periods of disuse enable muscle to quickly adapt to increases in physical activity, by accurately predicting muscle atrophy upon a period of detraining and subsequent hypertrophy upon retraining.