Prediction of long-term outcome in Multiple Sclerosis: A systematic review

Abstract

Background: Disease course is highly variable in Multiple Sclerosis (MS) and the prediction of individual long-term prognosis is challenging. Objective(s): To identify demographic, socio-economic, clinical, paraclinical and therapeutic factors associated with long-term disability in patients with MS. Method(s): We searched PubMed, Embase and SCOPUS with no predefined search period and until December, 31 st 2024. We included studies that examined predictors of long-term disability in patients with MS regardless of the disease phenotype. Result(s): A total of 241 studies were included. Thirty factors were identified. At baseline, higher age (Hazard ratio (HR)= 0.036-2.3), longer disease duration (HR=1.01-1.8), smoking (HR= 1.63-2.12), vascular comorbidities (HR=1.51), ethnicity (HR=1.63), high disability (HR = 1.19-1.73), disease activity (HR = 1.6-1.9), motor (HR= 0.54-2.47), cerebellar (HR=1.71-3.096), sphincter symptoms (HR=1.3-3.3) and cognitive dysfunction (HR = 3.07-3.68) were strongly associated with a poor long-term prognosis. Relapses in the previous year increased the hazard for a subsequent disability worsening. Decreased retinal layers thickness volumes were independently associated with a three to six-fold risk of EDSS progression. Blood and cerebrospinal fluid (CSF) neurofilaments measurements at baseline were associated with an increase in the hazard of disability. Imaging prognostic biomarkers were mainly initial lesion load (HR = 1.51-3.51) and the presence of at least one spinal cord lesion (HR= 1.9-4.08). Other putative prognostic factors at baseline such as gender, delayed treatment, or whole brain atrophy showed a weak or inconsistent effect on prognosis. Newly studied factors including slowly expanding lesions on MRI and CSF Kappa index are promising biomarkers in predicting long-term outcome. Conclusion(s): The most robust drivers of long-term disability in MS were related to the degree of initial brain damage and its limited repair capacity. Prevention of modifiable risk factors should be a priority. A combination of these factors is needed to build a composite prediction model. Copyright © 2025