Integrating Clinical, Genetic, and MRI Biomarkers to Predict Disability Progression in Multiple Sclerosis
Massimo Filippi1, Ferdinando Clarelli4, Paolo Preziosa2, Melissa Sorosina4, Elisabetta Mascia4, Antonino Giordano3, Rosetta Pedotti7, Catarina Raposo7, Jens Wuerfel7, Stefano Magon7, Martina Rubin2, Elisabetta Pagani5, Paola Valsasina5, Paola Maria V. Rancoita8, Federica Esposito6, Maria Rocca2
1Neuroimaging Research Unit, Division of Neuroscience, Neurology Unit, Neurorehabilitation Unit, and Neurophysiology Service, 2Neuroimaging Research Unit, Division of Neuroscience, and Neurology Unit, 3Laboratory of Human Genetics of Neurological Disorders, IRCCS San Raffaele Scientific Institute, and Vita-salute San Raffaele University, 4Laboratory of Human Genetics of Neurological Disorders, 5Neuroimaging Research Unit, Division of Neuroscience, 6Neurology Unit, and Laboratory of Human Genetics of Neurological Disorders, IRCCS San Raffaele Scientific Institute, 7F. Hoffmann-La Roche, and employees of Roche, 8University Centre for Statistics in the Biomedical Sciences (CUSSB), Vita-Salute San Raffaele University
Objective:
To evaluate the value of clinical, genetic and brain MRI features in predicting time to Expanded Disability Status Scale (EDSS) score ≥6.0 in multiple sclerosis (MS).
Background:
Reliable predictors of disability progression are crucial to identify high-risk patients and enable personalized treatment. While clinical, MRI, and genetic features may individually predict prognosis, their combined predictive value in real-world cohorts is unclear.
Design/Methods:
We retrospectively collected demographic, clinical, genetic and 3T brain MRI data from 440 MS patients. Four demographic/clinical factors (sex, age, disease duration, exposure to moderate-efficacy [ME] and high-efficacy [HE]-disease-modifying therapies [DMTs]) and 10 MRI variables (white matter [WM] lesion volume [LV], global/regional volumes, microstructural damage) were pre-specified. Polygenic risk score (PRS) for MS severity (International MS Genetics Consortium) and Leukocyte Telomere Length (UK Biobank) were computed using clumping+thresholding across seven p-value thresholds (<5*10-6, <5*10-5, <5*10-4, <5*10-3, <0.05, <0.1, <0.2), generating 49 integrated models. Predictors of time to EDSS≥6.0 were identified using lasso-penalized Cox models. Performance was evaluated with the c-index.
Results:
The cohort (61% females; median age=40.43 years [IQR=32.15;47.92] years; median disease duration=8.38 years [IQR=2.54;15.85]; median EDSS=2.0 [IQR=1.5;4.0]; 78% relapsing-remitting, 22% progressive) was followed for a median of 7.79 years [IQR=4.47;12.30]. Ninety-three patients (21.1%) reached EDSS≥6.0. The best-performing model (c-index=0.795) identified older age (β=0.320), lower exposure to ME-DMTs (β=-0.639) and HE-DMTs (β=-0.356), higher MS severity p<0.2-PRS (β=0.073), higher <5*10-5-PRS Leukocyte Telomere Length (β=0.114), higher brain WM LV (β=0.140), lower normalized brain volume (β=-0.361), lower normalized gray matter volume (β=-0.061), lower normal-appearing gray matter fractional anisotropy (β=-0.075), and higher mean diffusivity in normal-appearing WM (β=0.119) and WM lesions (β=0.067), as predictors of faster progression.
Conclusions:
Combining demographic, clinical, genetic, and MRI features provides robust prognostic models of MS disability progression. Multimodal integration may inform individualized monitoring and treatment strategies to delay disability. Validation in independent cohorts will support validity of the top model(s).
Disclaimer: Abstracts were not reviewed by Neurology® and do not reflect the views of Neurology® editors or staff.