Post-stroke cognitive impairment afflicts up to 52% of survivors within 6 months, and 20% progress to clinical dementia within 5 years — yet the subacute phase remains largely untargeted therapeutically. The PROTECT trial is a Phase 3, Bayesian adaptive RCT enrolling up to 120 subacute stroke patients (≤6 months post-event) across multiple sites, randomizing them to 12 weeks of moderate-to-high-intensity multimodal resistance and aerobic training versus a low-intensity comparator. Primary outcome is the ADAS-Cog-13; secondary outcomes span structural and perfusion neuroimaging alongside blood biomarkers of neurodegeneration and inflammation. Power calculations used Cohen's d=0.63 from prior exercise RCTs, with two pre-planned interim analyses enabling early stopping.
The trial's strategic value lies in targeting the subacute inflammatory and vascular cascade — a mechanistically plausible but underexplored intervention window. Exercise is already known to reduce BDNF-linked neurodegeneration and systemic inflammation in chronic stroke, but whether those benefits translate during acute neurological vulnerability is genuinely unknown. The Bayesian adaptive design is methodologically sophisticated, enabling efficient sample reallocation and early stopping without inflating type I error — a meaningful improvement over classic fixed-sample stroke trials. Limitations include modest target enrollment (N=90 completers), which constrains subgroup analysis, and the absence of a true passive control. Still, if positive, this would be the first Phase 3 evidence that structured exercise during early stroke recovery can measurably alter dementia trajectory — a paradigm-shifting outcome for rehabilitation medicine.
