Accumulated evidence supports motor evoked potential (MEP) status and neuroimaging measures of corticospinal tract (CST) integrity as useful biomarkers for predicting upper-limb motor recovery. Our aim was to determine which biomarkers among electrophysiological and MRI variables can accurately explain the motor impairment in subacute/chronic stroke patients.

Forty patients (>1 month) were assessed by the Fugl-Meyer score, grip force and the Jebsen Taylor test. A motor composite score was computed with principal component analysis for each patient. In both hemispheres were collected resting motor threshold (rMT), MEP amplitude and latency and fractional anisotropy of the CST (FA-CST). During a motor fMRI paradigm, laterality index was computed within the primary motor cortex and cortico-cerebellar functional connectivity. Lesion volume, age and time since stroke were also collected. Analysis was done with ratio values in patients with MEP in the affected hemisphere (MEP+). First, each variable was tested in simple regression with linear or nonlinear effects. Only significant variables (P<0.1) were included in the multiple regression analysis.

In the 34 MEP+ patients (median age: 63 years, median time: 2.3 months), the multiple regression model only retained the rMT and the time since stroke as predictors. Time had non-linear effects. This model explained 38% of the motor composite score's variance.

In MEP+ patients, the resting motor threshold is the explanatory factor of the upper limb impairment. In contrast to MEP amplitude and latency, the rMT, which is a complex measure of both corticospinal excitability and influenced by the corticocortical interactions, was retained. No brain imaging measure was retained by the model. Other studies are required to determine the unexplained variance of the motor impairment.