Corticospinal Tract Microstructure Predicts Distal Arm Motor Improvements in Chronic Stroke.

Only 37 people from one site, and this is a reuse of data collected for a trial designed to answer a different question, so the prediction model was not the trial's pre-specified comparison. Five of the 42 enrolled participants were left out because their clinical or imaging data were incomplete. The paper does not say how they differed from the 37 who were analysed. Everyone could already move their wrist and fingers a little and had mild-to-moderate impairment. The authors themselves say the results should not be generalised to people in the first months after stroke or to those with severe arm impairment. The repeated assessments are a confound the authors admit to: the active monitoring group, which had no scheduled training hours, still improved significantly, and about 200 reaching movements per assessment session probably explain that. The contribution of the ASAP training cannot be separated from it. Only time taken was measured, so the study says nothing about how well people moved, only how fast they completed the tasks. The finding contradicts earlier studies, and the authors' explanation of an inverted U-shaped relationship between tract damage and improvement is a hypothesis drawn partly from combining this study with another, not something tested here. The authors state plainly that the corticospinal tract measure should not be used to predict for an individual survivor. They also note the final model explains only about 38% of the variance in distal arm function, with several predictors they could not measure.

The text supplied contains no funding or conflict of interest declaration. The publication is indexed as supported by an NIH extramural research grant; no commercial sponsor is identified.

Do not use corticospinal tract imaging to choose which chronic stroke patients to treat. The link held across 37 people as a group, it ran opposite to what earlier studies found, and the authors themselves say the measure should not be used to make predictions for an individual survivor.