Refining neuropsychological assessment strategies in multiple sclerosis: insights from long-term cognitive...

Cognitive evolution in people with multiple sclerosis (pwMS) may follow different trajectories. This study investigated long-term cognitive evolution in pwMS while controlling for methodological factors and psychological variables, to refine neuropsychological monitoring.

A cohort of pwMS (n = 148) underwent neuropsychological evaluations at three time points: baseline (BL), re-baseline (RBL; ~ 1 year later), and long-term follow-up (LTFU; ~ 6 years after BL). Annualized cognitive change scores were analyzed to compare changes during BL-RBL interval with those in RB-LTFU. Participants were classified over time as cognitively preserved (CP) or cognitively impaired (CI) and McNemar's test evaluated classification changes. Sensitivity analysis was conducted in a subgroup of recently diagnosed pwMS (n = 91). Reliable Change Indices (RCI) at the 95% confidence level were calculated for each neuropsychological test to identify significant improvements or declines between assessments, while accounting for practice effects. Exploratory analysis examined baseline demographic and clinical differences between pwMS who showed cognitive recovery and those with persistent CI.

After controlling for test-retest effects, cognitive improvements occurred across several domains during BL-RB, while performance remained stable in RBL-LTFU. Shifts between CP and CI were more frequent in BL-RBL. Nearly 40% of pwMS classified as CI at BL reverted to CP at RBL, and approximately 50% reverted at LTFU. CP participants remained stable, with conversion rates to CI below 1%. RCI confirmed a higher number of reliable improvements than declines in the BL-RBL interval. Cognitive status at RBL and female sex were associated with cognitive recovery at LTFU.

Cognitive trajectories in pwMS are dynamic with early improvements partly reflecting baseline measurement bias. Clinically, CI individuals at BL should undergo RBL assessment to improve long-term prognostic accuracy, whereas CP participants can be monitored avoiding unnecessary evaluations, optimizing resource allocations.