Lesion-related white matter disconnection in relapsing-remitting multiple sclerosis: clinical associations and exploratory analysis of conversion to secondary progressive multiple sclerosis

To assess whether lesion-related white matter tract disconnection is associated with clinical impairment in relapsing-remitting multiple sclerosis (RRMS) and to explore whether baseline/index disconnection patterns differ between patients who later convert to secondary progressive MS (SPMS) and those who remain RRMS. This retrospective cohort study included 95 RRMS patients, including 31 future SPMS converters and 64 non-converters. Follow-up MRI was available in 74 patients. Baseline/index and follow-up 3D T1-weighted and fluid-attenuated inversion recovery (FLAIR) images were processed using the online volBrain DeepLesionBrain pipeline to extract disconnection probabilities for 64 white matter tracts. Baseline/index disconnection was compared between groups using Mann-Whitney U tests, logistic regression, and receiver operating characteristic analysis, reduced-model sensitivity analyses, and scanner-adjusted sensitivity models. Spearman correlations assessed associations with Expanded Disability Status Scale (EDSS), Symbol Digit Modalities Test (SDMT), Timed 25-Foot Walk Test (T25-FWT), 9-Hole Peg Test (9-HPT), functional system scores, relapse burden, and lesion-derived MRI measures. False discovery rate (FDR) correction was applied. Converters had higher EDSS, greater relapse burden, slower T25-FWT, and worse 9-HPT performance than non-converters. Several baseline/index disconnection measures showed nominal group differences, but none survived FDR correction. Reduced clinical and lesion-adjusted models, the full adjusted model, and scanner-adjusted models identified no nominally significant or FDR-significant tract-level association with conversion. Binary disconnected/not-disconnected sensitivity analyses also identified no FDR-significant conversion-related tract. In contrast, 90 clinical tract-disconnection correlations remained FDR-significant. Longitudinal disconnection changes did not significantly differ between groups after correction. Lesion-related tract disconnection is strongly associated with clinical impairment in RRMS but showed only exploratory associations with SPMS conversion.