Patterns of white matter hyperintensities associated with cognition in middle-aged cognitively healthy individuals

Abstract

White matter hyperintensities (WMH) are commonly detected in the brain of elderly individuals and have been associated with a negative impact on multiple cognitive domains. We aim to investigate the impact of global and regional distribution of WMH on episodic memory and executive function in middle-aged cognitively unimpaired participants [N = 561 (45-75 years)] enriched for Alzheimer's disease risk factors. WMH were automatically segmented from FLAIR, T1 and FSE MR images. WMH load was calculated both globally and regionally. At each cerebral lobe, regional WMH load was measured at four equidistant layers extending from the lateral ventricles to juxtacortical areas. Cognition was measured by The Memory Binding Test (MBT) and WAIS-IV subtests. Global composite z-scores were calculated for the two cognitive domains. Association between global and regional WMH measurements were sought against cognitive measures, both in global composite scores and in individual subtests. We adjusted cognition and WMH burden for the main sociodemographic (age, sex and education) and genetic factors (APOE-ε4). Memory and executive function were significantly associated with global WMH load. Regionally, lower executive performance was mainly associated with higher deep WMH load in frontal areas and, to a lower degree, in occipital, parietal and temporal regions. Lower episodic memory performance was correlated with higher WMH burden in deep frontal and occipital areas. Our novel methodological approach of regional analysis allowed us to reveal the association between cognition and WMH in strategic brain locations. Our results suggest that, even a small WMH load can impact cognition in cognitively unimpaired middle-aged subjects.

Keywords: Cardiovascular risk; Hypertension; Magnetic resonance imaging; Preclinical Alzheimer’s disease; White matter lesions.

Fig. 1

Example of the segmentation of one participant and the Bullseyes plot representation. On the left figure, the left column, T1w MR image from one participant is presented in the three views: coronal, axial and sagittal (from up to bottom). In the central and right columns, the same views are shown with the layer and lobar segmentations overlaid, respectively. Abbreviations: FR, frontal right; FL, frontal left; TR, temporal right; TL: temporal left; PR, parietal right; PL, parietal left; OR, occipital right; OL, occipital left and BG, basal ganglia

Fig. 2

Cross-correlation matrix between pairs of the cognitive measures depicting results of Pearson’s correlations. The hot colour scale in the upper diagonal represents statistical significance (p-values), in which grey coloured regions depict non-significant correlations. Values down the main diagonal represent strength of correlation (r), which was always positive. The two cognitive domains were highly significant correlated with their respective cognitive measures. Memory measures were highly correlated among them, whereas executive function measures only showed a modest correlation among themselves but also with memory scores. Abbreviation: MBT, Memory Binding Test

Fig. 3

Regional patterns of correlation between cognition and WMH. Effect sizes of these correlation measures by Spearman’s Rho are shown in those areas that presented statistical significant association (p < 0.05). All the other regions are depicted in beige. Cognition and WMH were adjsuted by age, sex, education and number of APOE-ε4 alleles. WMH were also adjusted by TIV. Abbreviations: FR, frontal right; FL, frontal left; TR, temporal right; TL: temporal left; PR, parietal right; PL, parietal left; OR, occipital right; OL, occipital left; BG, basal ganglia; WMH, white matter hyperintensities; TIV, total intracranial volume