A multi-layered approach to elucidate mechanisms of physical function in response to rehabilitation in heart failure with preserved ejection fraction

Heart failure with preserved ejection fraction (HFpEF) is an increasingly common cause of morbidity and mortality in older adults that is driven by cardiac and non-cardiac mechanisms. Physical rehabilitation improves frailty and functional capacity in HFpEF, though underlying mechanisms remain less clear. We quantified >5,000 circulating proteins across two randomized clinical trials of rehabilitation in HFpEF (REHAB-HF, SECRET-II), identifying proteins associated with prognostic measures of physical function (short physical performance battery, 6-minute walk distance) and protein changes after rehabilitation. Using an artificial intelligence (AI)-enabled multiplex network analysis (MENTOR-IA), we identified biologically plausible networks central to this "physical function proteome," including endothelial remodeling, mitochondrial metabolism, calcium handling, and immune modulation. Expression of prioritized proteins at the transcriptional level localized to heart, skeletal muscle, and brain tissue, with several cognate transcripts implicated in frailty via tissue-specific transcriptome-wide genetic association studies. In addition, using novel human genetic approaches, we implicated select proteins as mediating tissue-specific genetic effects on frailty. These findings motivated us to construct multi-protein signatures of physical function, which correlated with functional changes observed with rehabilitation in REHAB-HF and SECRET-II and that were associated with heart failure and multi-dimensional clinical outcomes in >26,000 individuals. These findings collectively delineate a multi-system molecular program underlying physical function impairment and rehabilitation response in HFpEF, offering insights into potential precision risk estimators and therapeutic targets for surveillance and promotion of physiologic resilience.