MRI predictors of cognition in subcortical ischemic vascular disease and Alzheimer's disease

Abstract

Background: Causes of cognitive impairment in subcortical ischemic vascular disease (SIVD) are less well understood than in AD, but have been thought to result from direct effects of subcortical lacunes and white matter lesions, perhaps related to disruption of important cortical-subcortical pathways.

Objective: To examine the relation between cognitive abilities and quantitative MRI measures of subcortical cerebrovascular disease and cortical and hippocampal atrophy.

Methods: Subjects were 157 participants in a multicenter study of SIVD and AD who included cognitively normal, cognitively impaired, and demented individuals with and without subcortical lacunar infarcts. Dependent variables were neuropsychological tests of global cognitive function, memory, language, and executive function. Independent variables were quantitative MRI measures of volume of lacunar infarcts in specific subcortical structures, volume of white matter lesion (WML), volume of cortical gray matter (cGM), and total hippocampal volume (HV). Multiple regression analyses were used to identify MRI predictors of cognition.

Results: Subcortical lacunes were not related to cognitive measures independent of effects of other MRI variables. WML was independently related to selected, timed measures. HV and cGM were strong and independent predictors of cognitive variables, with effects that did not differ in subjects with and without subcortical lacunes.

Conclusions: Results suggest that cognitive impairment associated with subcortical ischemic vascular disease is primarily a result of associated hippocampal and cortical changes.

Figure

Percent of total variance (R²) of neuropsychological test variables explained by sequential regression models. Demographic variables (D) alone were included in the first model. Volume of thalamic lacunes was added in model 2, volume of white matter lesion (WML) was added in model 3, volume of cortical gray matter (cGM) in model 4, and hippocampal volume (HV) in model 5. For each neuropsychological test, each bar shows the amount of variance accounted for by all of the effects in that specific model. Differences between adjacent bars correspond to the incremental explanatory power associated with the addition of the last variable. In general, results show slight improvement in prediction associated with thalamic lacunes, modest effects associated with WML, and robust improvement in prediction when cGM and HV were added. (Neuropsychological tests: MDRS Tot = Mattis Dementia Rating Scale total score; MDRS I-P = Initiation-Perseveration subscale of the Mattis Dementia Rating Scale; BNT = Boston Naming Test; MAS LR = List recall trial of the World List Learning Test of the Memory Assessment Scales; FAS = FAS letter fluency total score; Animals = animal category fluency total score.) □ = demographic (D); = D + Lac; = D + Lac + WML; = D + Lac + WML + cGM; ■ = D + Lac + WML + cGM + HV.