Abstract Background and aims Biomarkers reflecting neuroplasticity, inflammation, and vascular integrity may explain variability in post-stroke recovery. This exploratory substudy of the PRACTISE trial (NCT05355831) examined longitudinal biomarker changes during stroke rehabilitation and their associations with upper-extremity (UE) motor recovery. Methods A total of 24 patients with subacute ischemic stroke (21 completers) received patient-tailored transcranial direct current stimulation (TDCS) or sham during four weeks of UE rehabilitation. UE motor function (FMA-UE) and clinical outcomes - including cognition (MoCA), depression (BDI-II), and quality of life (EQ-5D-5L) - were assessed at baseline, end-of-treatment, and at 12-weeks along with the biomarkers plasma Cathepsin-B, Cathepsin-S, E-selectin, and high-sensitivity C-reactive protein (hsCRP). Baseline MRI was rated for small-vessel-disease (SVD) burden using STRIVE criteria. Longitudinal biomarker changes were evaluated using mixed-effects regression. Results Despite substantial interindividual variability, increases in hsCRP were associated with less FMA-UE improvement from baseline to end-of-treatment (β = –0.75 ± 0.26, p = 0.01). The remaining biomarkers were not significantly related to FMA-UE improvement. Four participants who later experienced major adverse cardiovascular events showed marked elevations in hsCRP and Cathepsin-S at baseline. Sensitivity analyses revealed no significant associations with changes in cognition, depression, or quality-of-life, although a trend-level positive association was observed between E-selectin and MoCA-score. Conclusions Inflammatory activation—particularly elevated hsCRP and Cathepsin-S—was linked to poorer motor recovery and characterized individuals who later experienced major adverse cardiovascular events. These findings support further investigation of inflammatory biomarkers as indicators of both recovery potential and vascular risk during subacute stroke rehabilitation. Conflict of interest All authors have nothing to disclose.
Kolmos et al. (2026) studied this question.