The Barcelona baseline risk score to predict long-term prognosis after a first demyelinating event: a prospective observational study

Abstract

Disease progression; MRI; Multiple sclerosis

Progresión de la enfermedad; Imagen por resonancia magnética; Esclerosis múltiple

Progressió de la malaltia; Imatge per ressonància magnètica; Esclerosi múltiple

Background In multiple sclerosis (MS), predicting at symptom onset who will develop early and severe disability is an unmet need with significant therapeutic implications. Here we propose the Barcelona-Baseline Risk Score (BRS) model to predict long-term disease outcomes in a flexible and generalisable manner. Methods Using prospectively acquired data from the Barcelona first-attack cohort, we created the Barcelona-BRS model as a set of six Weibull survival models of time to an Expanded Disability Status Scale score of 3.0, built with flexible combinations of predictors, including sex, age at first attack, and number and topography of T2 lesions, among others, adaptable to data availability. Data-driven risk groups were identified and compared in terms of long-term clinical and MRI outcomes, including relapse-associated worsening (RAW), progression independent of relapse activity (PIRA), conversion to secondary progressive MS (SPMS), lesional and brain volumetric data, and patient-reported/administered clinical scores, through Kaplan–Meier and mixed-effects models. Finally, we externally validated our model in a completely unseen cohort. Findings We included 1074 patients (737 [69%] female, mean age: 31.7 years) with a first demyelinating attack. Over a median follow-up of 11.9 years, 375 (35%), 298 (28%), and 94 (8.8%) developed RAW, PIRA, and SPMS, respectively. Weibull models included age at first attack, number of brain T2 lesions, and disability at first visit as main predictors. Four data-driven groups of increasing risk of unfavourable outcomes were created: Light-Green-BRS (N = 258), Dark-Green-BRS (N = 319), Orange-BRS (N = 321), and Red-BRS (N = 176), which, over time, behaved significantly differently across disability, quality of life, and MRI measures, being the Red-BRS the group with worst outcomes (p < 0.01). The results in the external validation cohort (N = 139, 100 female [72%], 34 years) mirrored those of the original one. Interpretation The robustness, flexibility, and generalisability of the Barcelona-BRS model support its consideration as a ready-to-use tool for clinical practice.