In 243 older adults (≥60 years) hospitalized for heart failure and enrolled in the REHAB-HF trial, pre-intervention blood protein levels predicted who would benefit most from a 12-week multidomain physical rehabilitation program. Using LASSO and ensemble tree-based modelling across 92 cardiovascular and inflammatory proteins, two proteins — GP6 and ST2 — stratified Short Physical Performance Battery (SPPB) improvement: participants in the optimal protein profile group gained 2.1 SPPB points versus a loss of 0.8 points in the least-favorable group. For 6-minute walk distance, ALCAM alone stratified response from a negligible 10-metre gain to a clinically meaningful 85-metre improvement. Importantly, none of these proteins mediated functional change, suggesting they predict rather than cause rehabilitation response.
This is a preprint that has not yet been peer-reviewed, so findings should be interpreted cautiously. The work is nonetheless meaningful: it moves precision medicine into rehabilitation science, an area historically reliant on one-size-fits-all protocols. ST2 (a known cardiac stress and inflammation marker) and ALCAM (linked to immune cell trafficking) as response stratifiers opens mechanistic hypotheses worth pursuing. Limitations include the observational proteomic sub-study design within a randomized trial, a 70% data-availability rate introducing possible selection bias, and the lack of an external validation cohort. If replicated, pre-discharge protein profiling could help clinicians prioritize or intensify rehabilitation resources for heart failure patients most likely to recover physical function — a high-value target for healthy aging.