Diabetes is related to cognition but not plasma amyloid‐β 42/40 in an African American cohort

Background The prevalence of clinical Alzheimer’s disease (AD) and type 2 diabetes is higher in African Americans (AA) relative to non‐Hispanic White Americans. Although diabetes is a risk factor for Alzheimer’s clinical syndrome, it is unclear whether diabetes associates with amyloid‐β, a pathological hallmark of AD. We investigated whether diabetic status was associated with age‐related decline in plasma amyloid‐β (Aβ) 42/40 ratio and cognition in a predominantly dementia‐free AA cohort. Method Participants (N=314) were middle‐aged and older adults (Table 1) enrolled in the African Americans Fighting Alzheimer’s in Midlife cohort. EDTA plasma was collected at 1‐8 visits for Aβ 42/40 quantification using immunoprecipitation mass spectrometry (C 2 N Diagnostics, St. Louis, MO). 52% had >1 plasma Aβ 42/40 measurements over an average interval of 4.9 years ( SD =3). Recall (immediate and delayed) and executive functioning were assessed using Rey’s Auditory Verbal Learning Test and Trails B, respectively. Diabetic status was determined from clinician‐report, self‐report, self‐report of antidiabetic medication use or fasting glucose ≥126 mg/dL. Cognitive data from the first of 1‐6 visits and diabetic status were obtained at the same visit as baseline plasma Aβ 42/40 for 96% of participants. Linear mixed‐effects models tested diabetic status as a moderator of age‐related decline in plasma Aβ 42/40 ratio and cognition. If age x diabetic status was not significant, diabetic status was tested without the interaction. Covariates included APOE4 carrier status and sex, and education was included when cognition was the outcome. Results Diabetic status did not significantly moderate age‐related decline in plasma Aβ 42/40 (β=0.0001, p =.21) or predict average plasma Aβ 42/40 ratio (β=‐0.001, p =.22). Diabetes was also not significantly related to aging‐associated decline in cognition but was a significant predictor of worse average Trails B scores (Table 2). Results from a cognitively unimpaired subsample (n=261) revealed a similar pattern. Conclusion In a predominantly dementia‐free AA cohort, diabetes was unrelated to plasma Aβ 42/40 but was related to worse Trails B performance. Poor Trails B performance has been linked in previous studies to cerebrovascular dysfunction. Determining mechanisms, vascular and AD pathological pathways, that link diabetes to increased dementia risk in AAs warrants further investigation.