A 2-year multimodal prospective cohort study to predict sustained progression independent of relapse activity in multiple sclerosis

Background Disability accrual in multiple sclerosis (MS) is heterogeneous. We assessed whether a 2-year multimodal evaluation could identify patients who were developing progression independent of relapse activity (PIRA) or relapse-associated worsening (RAW). Methods Among the 81 MS patients with a 2-year follow-up and outcomes determined through 5 years, a composite for PIRA was developed using Firth’s penalized logistic regression analysis; discrimination was calculated as the area under the curve (AUC) using repeated stratified nested cross-validation. Results In total, 21 patients (25.9%) met the PIRA criteria, 9 (11.1%) met the RAW criteria, and 1 met both criteria. A multivariate Firth regression analysis selected two variables associated with PIRA after adjustment: Modified Fatigue Impact Scale (MFIS) total score (OR 2.04, 95% CI 1.17–3.83; p = 0.011) and cerebrospinal fluid (CSF) chitinase-3-like protein 1 (CHI3L1) (OR 2.00, 95% CI 1.11–5.26; p = 0.018). Notably, the Symbol Digit Modalities Test (ΔSDMT) showed an association in the univariate analysis (OR 0.47, 95% CI 0.23–0.82; p = 0.006) but was attenuated in the multivariate analysis [odds ratio (OR) 0.64; p = 0.141]. Oligoclonal immunoglobulin G band (OCB) positivity was universal in PIRA, but the findings are inconclusive due to sparse OCB-negative observations and incomplete OCB data. Worsening gait [ΔTimed 25-Foot Walk (ΔT25FW)] and new T2 lesions showed associations in the univariate analysis that were not retained in the multivariate analysis. None of the variables had a p -value of less than 0.10 for RAW. The median out-of-sample AUC was 0.74 (95% CI 0.68–0.78). Conclusion A two-year multimodal assessment identified a biologically coherent PIRA composite driven by intrathecal neuroinflammation and fatigue. No reliable RAW predictor was found, and external validation is warranted.