Correlations between four common measures of cognition in patients with secondary progressive multiple sclerosis

Background: Cognitive impairment is common in patients with multiple sclerosis (MS), especially in those with secondary progressive MS (SPMS), in whom neurodegeneration is the predominant feature. The oral Symbol Digit Modalities Test (SDMT), Paced Auditory Serial Addition Test (PASAT) and Brief Visuospatial Memory Test Revised (BVMT‐R) were used to track changes in cognition in the EXPAND study. It is critical to evaluate whether these tests are redundant or complementary in the evaluation of progressive disease. Objectives/Aims: Evaluate the relationship between commonly used cognitive tests in patients with SPMS and identify whether these cognitive tests provide unique insights on disease progression in patients with MS. Methods: EXPAND was a 36 month randomized, placebo‐controlled trial of siponimod 2 mg/day in patients with SPMS. Crosssectional, pairwise Pearson correlations (r) between SDMT, PASAT, and both the immediate learning and delayed recall indices of the BVMT‐R (BVMTR‐TL; BVMTR‐DR) were calculated by treatment group and combined. Correlations were examined for Baseline and for the change in these scores from the first postbaseline measurement to Month 24. Correlations were considered to be strong (>0.6), intermediate (0.4‐0.6) or weak (< 0.4); p< 0.05 indicates that the correlation was statistically different to zero. Results: Data were analysed for 1644 patients (siponimod, 1098; placebo, 546). In both treatment groups at Baseline, strong correlations (0.89) were found between BMVT‐R‐TL and BMVTR‐ DR. Intermediate correlations (range, 0.45‐0.59) were observed between all other tests (all p< 0.0001). The strong correlations between BMVT‐R‐TL and ‐DR persisted when measuring change from Baseline to Month 24 (siponimod, 0.68; placebo, 0.72; both p< 0.0001). The strength of correlations in change from Baseline to Month 24 were weak between all other tests in both treatment groups (range, 0.05‐0.17; p>0.05 in many cases). Conclusions: Cognitive outcomes measured were correlated at Baseline, confirming overlap in variance for BVMT‐R indices and intermediate shared variance for other measures at a single time point. The weak correlations of change to Month 24 between SDMT, PASAT and BVMT‐R may suggest that each test tracks different aspects of cognitive decline and/or has different test characteristics when applied repeatedly in patients with SPMS.