Utility of accessible neuroimaging markers of cerebrovascular disease in identifying cerebral amyloidosis across the Alzheimer’s disease spectrum

Abstract

AbstractBackgroundAlzheimer’s disease (AD) is defined by the presence of cognitive impairment and cerebral amyloidosis and is associated with several cerebrovascular disease (CVD) risk factors including age, apolipoprotein‐E ε4 (APOE‐ε4) genotype, hypertension, diabetes and hyperlipidemia. Despite this association, the clinical utility of individual CVD biomarkers in screening for AD pathology is understudied. Therefore, we sought to examine the relationship between cerebral amyloidosis and magnetic resonance imaging (MRI)‐based CVD measures, including cerebral microbleeds, ischemic infarction, and white matter hyperintensities (WMH). We hypothesized that CVD biomarker burden would be positively associated with amyloidosis. Given recent studies have shown plasma phospho‐tau181 (P‐tau181) is a promising early biomarker of amyloidosis, we further hypothesized that combining P‐tau181 with CVD biomarkers would predict amyloid status in cognitively unimpaired and impaired persons.MethodParticipant data were extracted from the Alzheimer’s Disease Neuroimaging Initiative database (N = 1,430). Volumetric data (WMH volume [WMHv], medial temporal lobe volume [MTLv]) were normalized and log‐transformed. Adjusted odds ratio (OR) was calculated using logistic regression with dichotomized amyloid‐PET status as the outcome. Area under the curve (AUC) was compared by the likelihood ratio test.ResultAfter adjustment for age, sex, APOE‐ε4 status, MTLv, cognition (MoCA), and other CVD biomarkers of interest, we found that WMHv alone was significantly associated with amyloidosis in both cognitively impaired (OR = 1.21) and unimpaired (OR = 1.25) persons. We then developed an accessible diagnostic model incorporating WMHv, significant covariates and plasma P‐tau181. Addition of P‐tau181 to the model improved diagnostic accuracy in cognitively impaired persons (AUC = 0.85 to 0.87, p < 0.001), and did not attenuate the relationship between WMHv and amyloidosis.ConclusionWe newly identify WMH, but not other CVD lesions, as predictive of amyloidosis in both preclinical and symptomatic AD. The current study adds to the literature regarding which CVD biomarkers are most predictive of AD pathology. Measurement of WMH alongside P‐tau181 and other biomarkers may produce an accessible screening algorithm for AD. Study generalizability is limited by ascertainment bias and lack of diversity. Future research will compare the performance of WMH to standard AD biomarkers in a more diverse, community‐based sample.