Cognition in multiple sclerosis within the modern diagnostic and treatment era

Slowed information processing speed has long been considered the principal cognitive deficit in multiple sclerosis, with slowed cognitive speed presumed responsible for downstream deficits in other functions, including memory. This speed-centric model was established over three decades ago before disease-modifying therapies were available, and contrasts with our current clinical experience, but nonetheless remains dominant and unquestioned. We re-evaluated this model among current patients with multiple sclerosis diagnosed and cared for within the modern diagnostic and treatment era (2001-25). Within a case-control cohort, persons with early relapsing-remitting disease (≤5.0 years since diagnosis, n = 170) performed worse than neurologically-healthy controls (n = 45) on memory but not cognitive speed, and cognitive speed among patients remained normal and stable over the next 6 years despite subtle memory decline. Relative to four historical studies of early relapsing-remitting disease (diagnosed using older criteria), effect sizes for case-control differences in our cohort were much lower for cognitive speed, but comparable for memory. An independent clinical cohort of 1004 consecutive patients aged 18-65 years with relapse-onset multiple sclerosis completed standard-of-care cognitive screenings between 2018 and 2025. We captured data from three independent periods. During the first period (n = 642), rates of poor performance (≤7.5th percentile) did not differ from normative expectations for cognitive speed (8.4%) or attention (9.0%), but were nonetheless elevated for verbal memory (23.8%) and visuospatial memory (14.8%). Patients during the second (n = 123) and third (n = 239) periods demonstrated memory deficits despite normal cognitive speed on co-normed tasks. Objective cognitive speed among current patients was remarkably similar to healthy normative expectations [z-score: mean (standard deviation), 0.03 (1.14); median (interquartile range), 0.00 (-0.67, 0.67)], and was much better than across several historical comparisons from 20-25 years ago. Self-reported cognitive deficits within the total clinical cohort versus control respondents indicated worst disease-related difficulties in expressive language (e.g. word-finding), followed by working memory and episodic memory, with a small difference in executive/speed that was fully explained by mood in relapsing-remitting disease. Current patient-reported attention/executive deficits were lower than 35 years ago, despite comparable memory difficulties. As an exception, attention/executive and cognitive speed deficits were observed in secondary-progressive disease. The speed-centric model of cognitive dysfunction in multiple sclerosis is inaccurate within the modern diagnostic and treatment era. Memory deficits remain prevalent, and we highlight working memory maintenance as an important target for further investigation. The field requires testable models of memory dysfunction informed by contemporary cognitive neuroscience, with the goal of developing heretofore elusive treatments for memory deficits.