Retinal nerve fiber layer and ganglion cell layer thinning as biomarkers of cognitive decline in multiple sclerosis: an 18-month prospective cohort study.

Background

Retinal neurodegeneration measured by optical coherence tomography (OCT) has emerged as a candidate biomarker of central nervous system (CNS) damage in multiple sclerosis (MS). However, its relationship with cognitive dysfunction - a clinically relevant and underappreciated dimension of disease burden - remains poorly characterized.

Objective

To evaluate cross-sectional and longitudinal associations between retinal nerve fiber layer (RNFL) and ganglion cell layer (GCL) thickness and cognitive performance in MS patients without a history of optic neuritis.

Methods

A prospective 18-month cohort study was conducted at a tertiary MS center in Buenos Aires, Argentina. OCT was performed using a spectral domain OCT (Spectralis, Heidelberg Engineering), and cognitive assessment used the Brief International Cognitive Assessment for MS (BICAMS) together with the Brief Repeatable Battery of Neuropsychological Tests (BRB-N) at baseline and 18 months. Multiple linear regression, linear mixed-effects models, and Benjamini-Hochberg false discovery rate (FDR) correction were applied.

Results

Sixty-one patients completed follow-up (mean age 41.2 ± 13.4 years; 72.1% female; mean EDSS 3.1 ± 1.1). GCL thickness was a significant independent predictor of an age/education-adjusted PASAT-3 z-score after adjustment for EDSS (β = 0.041 SD/µm, p = 0.005, R² = 0.21). This association survived FDR correction (p_FDR = 0.007). Effect sizes for RNFL group differences were medium-to-large (Cohen's d = 0.62 for PASAT-3; d = 0.64 for 9HPT). Longitudinally, patients with normal RNFL showed significant improvement in phonemic verbal fluency over 18 months (Δ = +4.0 points, p = 0.009), whereas those with abnormal RNFL showed no change (Δ = +0.3, p = 0.81); however, the RNFL group × time interaction in a mixed-effects model did not reach statistical significance (p = 0.073), indicating this represents a trend rather than a confirmed differential trajectory.

Conclusions

GCL thinning independently predicts impaired information processing speed in MS, after accounting for physical disability. A trend toward better-preserved phonemic verbal fluency trajectories in patients with normal baseline RNFL was observed but did not reach significance in a group × time interaction model, warranting confirmation in larger cohorts. These findings support GCL thickness as a candidate biomarker of CNS neurodegeneration in MS.