Background/Aim: Early stroke recovery biomarkers may improve prognostication of long-term mobility post-stroke. Our aim was to examine early neuroimaging and neurophysiological biomarkers of long-term mobility recovery post-stroke.
Methods: MEDLINE/EMBASE were systematically searched. Eligible cohort studies reported the prognostic capacity of early (≤14days post-stroke) neuroimaging or neurophysiological biomarkers for mobility outcomes (impairment/activity) at a future time-point (≥14days but ≤24months post-stroke). Statistical analysis methods were categorised as associative, discriminative or validated predictive.
Results: Fourteen studies (biomarker measurement median 6.9(1.8-14) days) were included. Sixty-seven biomarker analyses were completed: 52 neuroimaging (most common lesion location n=22) and 15 neurophysiological (most common motor evoked potential status n=8). Most analyses related to 3-months post-stroke follow-up (n=31). Forty-one impairment (most common Fugl Meyer Lower Limb n=20) and 26 activity (most common Functional Ambulation Category n=17) measures were used. Analyses were largely associative (n=65, 97%, all high risk of bias due to confounding factors and low risk of bias for outcome measurement). Eighteen (27.6%) neuroimaging (most common lesion location n=7) and six (9.2%) neurophysiological (most common tibial nerve somatosensory evoked potential amplitude parameters n=4) biomarkers were significantly associated (p≤0.05) with their respective mobility outcome. One internally validated neural network model using magnetic resonance imaging data (low risk of bias) well-discriminated (AUC 0.83) independent walking ability at 6-months post-stroke.
Conclusion: Few promising early neuroimaging/neurophysiological biomarkers of mobility recovery post-stroke were identified. This may suggest that biomarkers measured early are not prognostic of long-term mobility outcomes, or that we are yet to identify the right biomarkers.