Predicting cognitive decline in multiple sclerosis: a 5-year follow-up study
Research output: Contribution to journal › Article › Academic › peer-review
Cognitive decline is common in multiple sclerosis and strongly affects overall quality of life. Despite the identification of crosssectional MRI correlates of cognitive impairment, predictors of future cognitive decline remain unclear. The objective of this study was to identify which MRI measures of structural damage, demographic and/or clinical measures at baseline best predict cognitive decline, during a 5-year follow-up period. A total of 234 patients with clinically definite multiple sclerosis and 60 healthy control subjects were examined twice, with a 5-year interval (mean = 4.9 years, standard deviation = 0.9). An extensive neuropsychological evaluation was performed at both time points and the reliable change index was computed to evaluate cognitive decline. Both whole-brain and regional MRI (3 T) measures were assessed at baseline, including white matter lesion volume, diffusion-based white matter integrity, cortical and deep grey matter volume. Logistic regression analyses were performed to determine which baseline measures best predicted cognitive decline in the entire sample as well as in early relapsing-remitting (symptom duration 510 years), late relapsing-remitting (symptom duration 510 years) and progressive phenotypes. At baseline, patients with multiple sclerosis had a mean disease duration of 14.8 (standard deviation = 8.4) years and 96/234 patients (41%) were classified as cognitively impaired. A total of 66/234 patients (28%) demonstrated cognitive decline during follow-up, with higher frequencies in progressive compared to relapsing-remitting patients: 18/33 secondary progressive patients (55%), 10/19 primary progressive patients (53%) and 38/182 relapsing-remitting patients (21%). A prediction model that included only whole-brain MRI measures (Nagelkerke R2 = 0.22, P50.001) showed cortical grey matter volume as the only significant MRI predictor of cognitive decline, while a prediction model that assessed regional MRI measures (Nagelkerke R2 = 0.35, P50.001) indicated integrity loss of the anterior thalamic radiation, lesions in the superior longitudinal fasciculus and temporal atrophy as significant MRI predictors for cognitive decline. Disease stage specific regressions showed that cognitive decline in early relapsing-remitting multiple sclerosis was predicted by white matter integrity damage, while cognitive decline in late relapsing-remitting and progressive multiple sclerosis was predicted by cortical atrophy. These results indicate that patients with more severe structural damage at baseline, and especially cortical atrophy, are more prone to suffer from cognitive decline. New studies now need to further elucidate the underlying mechanisms leading to cortical atrophy, evaluate the value of including cortical atrophy as a possible outcome marker in clinical trials as well as study its potential use in individual patient management.