Abstract P343: Associations of Adipose Tissue Distribution With Hypertension and Diabetes in African Ancestry Men: A Novel Approach Using Compositional Data Analysis

Abstract

Background: Adipose tissue (AT) distribution, which differs by race/ethnicity, can differentially impact risk for cardiometabolic disease. Traditional analyses using AT volumes tend to place all ATs in a regression model, estimating an increase in one AT volume while “holding other volumes constant”; however, the compositional nature of the data suggests the other ATs cannot remain constant. Compositional data analysis (CODA) methods eliminate this concern by comparing logratios of components. CODA was designed for analyses of data like those of body composition , but has yet to be used in this field. We demonstrate the use of CODA for the comparison of visceral AT (VAT) and subcutaneous AT of the abdomen (ASAT) and thigh (TSAT) with prevalence of cardiometabolic diseases. Methods: This cross-sectional analysis used 615 African Caribbean men from Tobago aged 50-91 years (mean age 63.6 years, mean BMI 27.7 kg/m 2 ). AT volumes were obtained from CT scans of the abdomen (VAT and ASAT) and of the mid-thigh (TSAT). An additive log2 ratio transformation was applied to the ATs to generate the VAT, ASAT, and TSAT component effects relative to the other tissues captured in the respective CT scan. Ordinal logistic regression analyses were performed for increasing hypertension or glucose impairment categories (defined in Table) after adjustment for age, disease status, and lifestyle factors. Results: After full adjustment (Table), a 2-fold higher ASAT volume (relative to other abdominal tissues) was associated with higher odds, and a 2-fold higher TSAT volume (relative to other thigh tissues) was associated with lower odds, of being in a higher hypertension category. After BMI adjustment, no AT was associated with glucose impairment category. VAT was not a significant predictor for either outcome in African Ancestry men. Conclusion: CODA methods, which avoid problematic assumptions made in traditional analyses, found significant and opposing associations of TSAT and ASAT with hypertension categories in African Ancestry men.