Diagnostic accuracy and age-dependent prognostic confounding of kappa free light chain parameters in multiple...

The kappa free light chain (κ-FLC) index is widely recognized as an alternative to oligoclonal bands for demonstrating intrathecal immunoglobulin synthesis in multiple sclerosis (MS). However, its prognostic value remains controversial. We investigated whether physiological age-related changes confound κ-FLC parameters and their reported prognostic associations.

In this retrospective study across two tertiary centers, we included patients with MS and controls with non-immunological neurological disease. Paired cerebrospinal fluid (CSF) and serum samples were analyzed by nephelometry in a single laboratory. Diagnostic accuracy was evaluated using receiver operating characteristic (ROC) curves. Prognostic endpoints included time to confirmed Expanded Disability Status Scale (EDSS) score of 3.0 and of 6.0, as well as conversion to progressive MS. Biomarker cut-offs were derived using maximally selected rank statistics, and associations were assessed via Cox proportional hazards regression.

We evaluated 61 patients with MS and 18 non-immunological neurological diseases over a median follow-up of 8.1 years. The κ-FLC index discriminated MS from controls with an area under the curve (AUC) of 0.93 (cut-off: 5.7), performing comparably to CSF κ-FLC (AUC: 0.94) and Q_κ-FLC (AUC: 0.97). Serum κ-FLC positively correlated with age (ρ = 0.48, p < 0.001), whereas CSF κ-FLC showed no age dependence (ρ = -0.01, p = 0.93), resulting in an age-dependent decline in the κ-FLC index (ρ = -0.26, p = 0.049). Serum albumin did not change with age (ρ = 0.12, p = 0.36), while the serum κ-FLC-to-albumin ratio increased (ρ = 0.52, p < 0.001). The IgG index was unaffected by age. Although unadjusted cut-offs appeared predictive on standard log-rank testing, no κ-FLC metric retained prognostic significance after correction for threshold optimization or multivariable adjustment for age at onset. Age at onset was the sole independent predictor across all disability endpoints (hazard ratios: 1.12-1.15).

Age selectively elevates the denominator of the κ-FLC index while leaving its intrathecally driven numerator unaffected. Consequently, the index carries an age signal that is not corrected by Q_albumin. This physiological dissociation likely explains previous discrepancies in the prognostic literature and underscores the necessity of adjusting for age in future κ-FLC outcome studies.