ABSTRACT
BACKGROUND: Predicting the long-term disability outcomes of multiple sclerosis (MS) cases is challenging. OBJECTIVE: We prospectively analysed our previous MS cohort with initial cerebrospinal fluid (CSF) proteomics data to reveal disability markers after 8.2±2.2 years of follow-up. METHODS: Patients with regular follow-up visits were assigned into two groups: those with an age-related MS severity (ARMSS) score ≥5 (unfavourable course group, N = 27) and ARMSS score <5 (favourable course group, N = 67). A machine learning-based algorithm was applied to reveal candidate poor prognosis-associated initial CSF proteins, which were measured in an independent MS cohort (verification group, N = 40) by ELISA. Additionally, the correlation of initial clinical and radiological parameters with long-term disability was analysed. RESULTS: CSF alpha-2-macroglobulin (P = 0.0015), apo-A1 (P = 0.0016), and haptoglobin (P = 0.0003) protein levels, as well as cerebral lesion load (>9 lesions) on magnetic resonance imaging, gait disturbance (P = 0.04), and bladder/bowel symptoms (P = 0.01) were significantly higher in the unfavourable course group than in the favourable course group. Optic nerve involvement evident on initial magnetic resonance imaging (P = 0.002) and optic neuritis (P = 0.01) were more frequent in the favourable course group. CONCLUSION: The herein identified initial CSF protein levels, in addition to the clinical and radiological parameters at disease onset, have predictive value for long-term disability in MS cases.
