Abstract P049: Cognitive Function and Brain Structure and Risk for Incident Diabetes: The Atherosclerosis Risk in Communities Study

Abstract

Introduction: Diabetes is an established risk factor for accelerated brain aging; however recent epidemiological studies suggest individuals with lower levels of cognitive functioning may be at increased risk for developing diabetes. Hypothesis: We hypothesized that lower baseline levels of cognitive function and MRI measures of brain pathology are associated with increased risk of diabetes. Methods: We used data on 11,503 adult men and women with cognitive function assessment (ARIC visit 2, 1990-1992) and 1,383 individuals who completed brain MRI (ARIC visit 3, 1993-1995) and were followed through 2012. Cognitive function was calculated as the average of z-scores from three unique neuropsychological tests. Board-certified radiologists adjudicated severity of sulcal size, ventricular size, and white matter hyperintensity (WMH) from brain MRI to a 10-point visual matching scale. Incident diabetes was ascertained at ARIC visits 3, 4 (1996-1998), and 5 (2010-2012) [self-report of diabetes medication use, self-report of a physician diagnosis, fasting glucose ≥ 126 mg/dL (≥ 7.0 mmol/L) or by non-fasting glucose ≥ 200 mg/dL (≥ 11.1 mmol/L)] and at annual telephone follow-up calls by self-report of physician diagnosis or diabetes medication use. Cox proportional hazards regression analysis was used to estimate hazard ratios (HR) for incident diabetes according to 1 standard deviation (SD) increment of cognitive function and severity of each brain MRI measure. Minimally adjusted models included age, sex, race, and field center. More fully adjusted models included other potentially relevant characteristics and diabetes risk factors. Results: Mean follow-up for the cognitive function analysis was 16.7 years and 14.2 years for the MRI analyses. Summary statistics and hazard estimates are presented in the Table (below). Conclusion: Measures of brain atrophy were associated with an increased incidence of diabetes in more fully adjusted models, but cognitive function and other measures of brain structure were not.