Identifying and understanding cognitive profiles in multiple sclerosis: a role for visuospatial memory functioning

Research output: Contribution to journalArticleAcademicpeer-review

Abstract

BACKGROUND: The heterogeneous nature of cognitive impairment in people with multiple sclerosis (PwMS) hampers understanding of the underlying mechanisms and developing patient-tailored interventions. We aim to identify and classify cognitive profiles in PwMS, comparing these to cognitive status (preserved versus impaired).

METHODS: We included 1213 PwMS (72% female, age 45.4 ± 10.7 years, 83% relapsing-remitting MS). Cognitive test scores were converted to Z-scores compared to healthy controls for the functions: attention, inhibition, information processing speed (IPS), verbal fluency and verbal/visuospatial memory. Concerning cognitive status, impaired cognition (CI) was defined as performing at Z ≤ - 1.5 SD on ≥ 2 functions. Cognitive profiles were constructed using latent profile analysis on all cognitive functions. Cognitive profiles or status was classified using gradient boosting decision trees, providing the importance of each feature (demographics, clinical, cognitive and psychological functioning) for the overall classification.

RESULTS: Six profiles were identified, showing variations in overall performance and specific deficits (attention, inhibition, IPS, verbal fluency, verbal memory and visuospatial memory). Across the profiles, IPS was the most impaired function (%CI most preserved profile, Profile 1 = 22.4%; %CI most impaired profile, Profile 6 = 76.6%). Cognitive impairment varied from 11.8% in Profile 1 to 95.3% in Profile 6. Of all cognitive functions, visuospatial memory was most important in classifying profiles and IPS the least (area under the curve (AUC) = 0.910). For cognitive status, IPS was the most important classifier (AUC = 0.997).

CONCLUSIONS: This study demonstrated that cognitive heterogeneity in MS reflects a continuum of cognitive severity, distinguishable by distinct cognitive profiles, primarily explained by variations in visuospatial memory functioning.

Original languageEnglish
JournalJournal of neurology
Early online date26 Feb 2024
DOIs
Publication statusE-pub ahead of print - 26 Feb 2024

Keywords

  • Classification
  • Cognition
  • Latent profile analysis
  • Multiple sclerosis
  • Phenotypes
  • Profiles

Cite this