MRI-visible perivascular space location is associated with Alzheimer's disease independently of amyloid burden

Abstract

Perivascular spaces that are visible on magnetic resonance imaging (MRI) are a neuroimaging marker of cerebral small vessel disease. Their location may relate to the type of underlying small vessel pathology: those in the white matter centrum semi-ovale have been associated with cerebral amyloid angiopathy, while those in the basal ganglia have been associated with deep perforating artery arteriolosclerosis. As cerebral amyloid angiopathy is an almost invariable pathological finding in Alzheimer's disease, we hypothesized that MRI-visible perivascular spaces in the centrum semi-ovale would be associated with a clinical diagnosis of Alzheimer's disease, whereas those in the basal ganglia would be associated with subcortical vascular cognitive impairment. We also hypothesized that MRI-visible perivascular spaces in the centrum semi-ovale would be associated with brain amyloid burden, as detected by amyloid positron emission tomography using 11C-Pittsburgh B compound (PiB-PET). Two hundred and twenty-six patients (Alzheimer's disease n = 110; subcortical vascular cognitive impairment n = 116) with standardized MRI and PiB-PET imaging were included. MRI-visible perivascular spaces were rated using a validated 4-point visual rating scale, and then categorized by severity ('none/mild', 'moderate' or 'frequent/severe'). Univariable and multivariable regression analyses were performed. Those with Alzheimer's disease-related cognitive impairment were younger, more likely to have a positive PiB-PET scan and carry at least one apolipoprotein E ɛ4 allele; those with subcortical vascular cognitive impairment were more likely to have hypertension, diabetes mellitus, hyperlipidaemia, prior stroke, lacunes, deep microbleeds, and carry the apolipoprotein E ɛ3 allele. In adjusted analyses, the severity of MRI-visible perivascular spaces in the centrum semi-ovale was independently associated with clinically diagnosed Alzheimer's disease (frequent/severe grade odds ratio 6.26, 95% confidence interval 1.66-23.58; P = 0.017, compared with none/mild grade), whereas the severity of MRI-visible perivascular spaces in the basal ganglia was associated with clinically diagnosed subcortical vascular cognitive impairment and negatively predicted Alzheimer's disease (frequent/severe grade odds ratio 0.03, 95% confidence interval 0.00-0.44; P = 0.009, compared with none/mild grade). MRI-visible perivascular space severity in either location did not predict PiB-PET. These findings provide further evidence that the anatomical distribution of MRI-visible perivascular spaces may reflect the underlying cerebral small vessel disease. Using MRI-visible perivascular space location and severity together with other imaging markers may improve the diagnostic value of neuroimaging in memory clinic populations, in particular in differentiating between clinically diagnosed Alzheimer's and subcortical vascular cognitive impairment.