Disability profiles in progressive multiple sclerosis reflect pathology distribution, independent of clinical phenotype

Abstract Multiple sclerosis (MS) is a clinically heterogeneous disease affecting both neurological and cognitive functions. In this study, we defined clinical profiles based on combined cognitive and motor assessments in progressive MS patients, examined their overlap with classical progressive phenotypes, and linked each profile to MRI measures of network disconnection and grey matter atrophy. We analysed baseline data from 580 participants [277 primary progressive (PP)MS, 303 secondary progressive (SP)MS] in the phase 3b CONSONANCE (NCT03523858) study. Clinical assessments included: Expanded Disability Status Scale (EDSS), 9-Hole Peg Test (9HPT), timed 25-Foot Walk Test (25FWT), Symbol Digit Modalities Test (SDMT) and Brief Visuospatial Memory Test Revised (BVMT-R). MRI acquisition comprised Fluid Attenuated Inversion Recovery for lesion segmentation, and 3D-T1 weighted images for disconnectome reconstruction and volumetry. Independent component analysis (ICA) was applied to regional disconnection and volumes to respectively identify patterns (ICs) of disconnection and grey matter atrophy. Latent profile analysis (LPA) identified disability profiles from multimodal clinical data. To identify the variables most strongly associated with each LPA-defined profile we trained one-versus-all eXtreme Gradient Boosting (XGBoost) classifiers, and computed SHapley Additive exPlanations (SHAP) values. Based on prevailing clinical manifestation, LPA delineated three profiles: motor disability (n=138, 23.8%), cognitive disability (n=181, 31.2%), and global disability (n=261, 45.0%). Profile prevalence did not differ by clinical phenotype (χ², p=0.675). Compared with the other profiles, motor disability one was associated with relatively preserved connectivity and tissue volume in the limbic and frontal networks, and preserved volumes in supplementary motor cortex and postcentral gyrus. The cognitive disability profile had relatively lower connectivity and volumes in the cerebellar vermis, temporal pole, posterior cingulate gyrus, and parietal operculum, lower default-mode-network connectivity and prefrontal–visual-integration-network volumes. The global disability profile had lower fronto-parietal-network connectivity and lower grey matter, superior frontal gyrus, pallidum, cuneus and cerebellum volumes. Our findings reveal that progressive MS encompasses at least three clinically and biologically distinct subtypes transcending the traditional PPMS/SPMS divide. The cognitive disability profile is driven by focal network disconnection and targeted grey matter loss in critical cognitive hubs, whereas the global disability profile reflects widespread grey matter atrophy across motor and associative regions. Identifying these profiles could be useful in clinical trials by matching outcome measures to underlying patterns of pathology and their clinical manifestations.