Adding neuroimaging and neurophysiological biomarkers to clinical assessment models significantly improves prognostic accuracy for upper limb motor recovery at three to six months following a stroke, according to a systematic review and meta-analysis published in Frontiers in Neurology.
Clinical scales are routinely used to predict post-stroke motor recovery, yet these tools remain insensitive to underlying neural substrates. While neurophysiological and neuroimaging modalities capture corticospinal tract integrity, their incremental predictive contribution beyond standard clinical evaluations had not previously been systematically quantified.
To address this gap, researchers systematically searched PubMed, Embase, and Web of Science from database inception through January 31, 2026. The systematic review identified twenty-three prospective cohort studies comprising a total of 1,341 participants, with each included study evaluating at least two distinct assessment modalities. Methodological quality and risk of bias across the cohorts were appraised using the Quality in Prognosis Studies tool.
The quantitative synthesis applied random-effects meta-analytic models to calculate the pooled incremental predictive value, measured as the change in the area under the receiver operating characteristic curve between multimodal combinations and clinical-only models. To assess the stability of the pooled estimates, the authors carried out sensitivity analyses using alternative between-study variance estimators alongside the Hartung-Knapp adjustment, and explored potential heterogeneity through meta-regression.
Across the pooled literature, multimodal assessment yielded a statistically significant improvement in prognostic accuracy over clinical models alone, achieving an overall pooled incremental discrimination gain of 0.112 in area under the curve (95% CI 0.061 to 0.163, p < 0.001). Among specific diagnostic modalities, motor evoked potential status demonstrated the largest pooled incremental predictive gain, registering an increase of 0.150 (95% CI 0.023 to 0.277; I² = 71.0%).
Individual imaging parameters showed varied predictive contributions across the reviewed evidence. A single prospective study examining ipsilesional corticospinal tract fractional anisotropy reported an incremental gain of 0.220 (95% CI 0.110 to 0.330), while task-based functional magnetic resonance imaging showed an incremental improvement of 0.190 in one cohort (95% CI 0.088 to 0.292). Conversely, pooled analyses of other diffusion tensor imaging metrics did not achieve statistical significance, yielding an incremental gain of 0.074 (95% CI −0.006 to 0.155), and available data for electroencephalography, resting-state functional magnetic resonance imaging, and kinematic parameters could not be quantitatively pooled.
Assessment timing emerged as a key factor influencing prognostic performance. Meta-regression identified stroke phase as a statistically significant moderator of incremental predictive value (beta = −0.127, 95% CI −0.247 to −0.007, p = 0.039), demonstrating that incorporating biomarkers provided greater incremental predictive improvements when assessments took place during the subacute phase rather than the acute phase.
Despite these favorable predictive findings, the study authors emphasized important methodological constraints that currently limit routine clinical adoption. Because the overall evidence base relies on a small number of cohorts and the statistical prediction interval crosses zero, the investigators concluded that the findings remain statistically promising but not yet sufficiently robust to direct bedside care, underscoring the need for further validation in larger prospective cohorts before motor evoked potential-guided stratification can be integrated into routine clinical practice.
Subject of Research: Incremental predictive value of biomarkers for post-stroke upper limb motor recovery
Article Title: Incremental predictive value of multimodal over unimodal assessment for upper limb motor recovery after stroke: a systematic review and meta-analysis
Article References: Chang-Qing, Y., Yun-Lei, W., Yu-Dong, C., Lin, M., Xiao-Xia, D., & Xiao-Li, W. (2026). Incremental predictive value of multimodal over unimodal assessment for upper limb motor recovery after stroke: a systematic review and meta-analysis. Frontiers in Neurology, 17, Article 1845257. https://doi.org/10.3389/fneur.2026.1845257
Image Credits: AI Generated
DOI: 10.3389/fneur.2026.1845257
Keywords: stroke recovery, upper limb motor recovery, multimodal assessment, motor evoked potentials, corticospinal tract, fractional anisotropy, diffusion tensor imaging, task-based fMRI, resting-state fMRI, electroencephalography, kinematic parameters, area under the curve, random-effects meta-analysis, prospective cohort studies, QUIPS tool, subacute stroke, acute stroke, prognostic modeling
Tags: acute strokearea under the curvecorticospinal tractdiffusion tensor imagingelectroencephalographyfractional anisotropykinematic parametersmotor evoked potentialsmultimodal assessmentprognostic modelingprospective cohort studiesQUIPS toolrandom-effects meta-analysisresting-state fMRIstroke recoverysubacute stroketask-based fMRIupper limb motor recovery
