Abstract
OBJECTIVE: Establishing biomarkers to predict multiple sclerosis (MS) disability accrual has been challenging using a single biomarker approach, likely due to the complex interplay of neuroinflammation and neurodegeneration. Here, we aimed to investigate the prognostic value of single and multimodal biomarker combinations to predict four-year disability progression in patients with MS. METHODS: In total, 111 MS patients were followed up for four years to track disability accumulation based on the Expanded Disability Status Scale (EDSS). Three clinically relevant modalities (MRI, OCT and blood serum) served as sources of potential predictors for disease worsening. Two key measures from each modality were determined and related to subsequent disability progression: lesion volume (LV), gray matter volume (GMV), retinal nerve fiber layer, ganglion cell-inner plexiform layer, serum neurofilament light chain (sNfL) and serum glial fibrillary acidic protein. First, receiver operator characteristic (ROC) analyses were performed to identify the discriminative power of individual biomarkers and their combinations. Second, we applied structural equation modeling (SEM) to the single biomarkers in order to determine their causal inter-relationships. RESULTS: Baseline GMV on its own allowed identification of subsequent EDSS progression based on ROC analysis. All other individual baseline biomarkers were unable to discriminate between progressive and non-progressive patients on their own. When comparing all possible biomarker combinations, the tripartite combination of MRI, OCT and blood biomarkers achieved the highest discriminative accuracy. Finally, predictive causal modeling identified that LV mediates significant parts of the effect of GMV and sNfL on disability progression. CONCLUSION: Multimodal biomarkers, i.e. different major surrogates for pathology derived from MRI, OCT and blood, inform about different parts of the disease pathology leading to clinical progression.