κ-index as a potential prognostic biomarker of disease activity in multiple sclerosis

Background: The κ-index, a quantitative marker of intrathecal κ free light chain synthesis, shows high diagnostic accuracy in multiple sclerosis (MS), but its prognostic value remains insufficiently defined. Biomarkers capable of predicting NEDA-3 loss are needed to support individualized risk stratification. Objectives: To evaluate whether the κ-index predicts NEDA-3 loss in MS, to characterize its association with baseline clinical, radiological and CSF features and to explore the complementary prognostic contribution of serum neurofilament light chain (sNfL) z-scores. Methods: A longitudinal observational cohort of 242 MS patients was analyzed with standardized CSF and serum sampling and a median follow-up of 28 months. Cox regression models tested associations between baseline variables and time to NEDA-3 loss. κ-index and sNfL z-scores were studied as continuous and dichotomized predictors, including cohort-specific thresholds identified via maximally selected rank statistics. Flexible Cox models with cubic B-splines assessed potential non-linear associations, and a bivariate spline-based model evaluated the joint contribution of κ-index and sNfL. Results: Higher κ-index values correlated with larger T2 lesion burden and CSF inflammatory markers. κ-index >6.4 independently predicted NEDA-3 loss, whereas as a continuous variable it showed a non-linear risk profile, with increasing hazard at low and very high values and a plateau at intermediate values. sNfL z-scores were linearly associated with earlier NEDA-3 loss. In a bivariate spline-based model, κ-index and sNfL provided complementary prognostic information, although discrimination remained modest (~55%). Correlation analyses across DMT classes showed no differential association between κ-index and treatment response in anti-CD20–treated patients. Conclusions: The κ-index is a biologically meaningful marker of intrathecal immune activation associated with subsequent loss of disease control. Its non-linear prognostic profile and partial complementarity with sNfL support multi-biomarker approaches and flexible survival modeling. These findings provide a methodological rationale for prospective validation and developing individualized prognostic tools in MS.