OR21-1 Adiposity and Diabetes Genetic Risk Modulates Causal Effects of BMI and Type 2 Diabetes on NAFLD: A Wax-and-Wane Pattern

• Published in: Journal of the Endocrine Society 2019-04, Vol.3 (Supplement_1)
• Main Authors: Wang, Ningjian, Wan, Heng, Wang, Yuying, Chen, Chi, Chen, Yi, Xia, Fangzhen, Han, Bing, Li, Qin, Zhang, Wen, Jensen, Michael, Lu, Yingli
• Format: Article
• Language: English
• Subjects: Cardiovascular Endocrinology
• ISSN: 2472-1972; 2472-1972
• DOI: 10.1210/js.2019-OR21-1

Background: Strong associations of obesity and diabetes with nonalcoholic fatty liver disease (NAFLD) have been discovered, but their causal relationships and interplay of genetic background was not fully determined. We aimed to assess the causal association of type 2 diabetes and BMI with NAFLD by Mendelian randomization (MR) analysis, and whether these associations were modulated by different levels of type 2 diabetes and body mass index (BMI) genetic risk scores (GRS). Methods: 8,944 participants were enrolled from a large cohort study. Subjects with a history of excessive consumption (male ≥20 g/d, female ≥10 g/d) of pure alcohol, viral hepatitis (including hepatitis B and hepatitis C virus), using medications associated with secondary NAFLD were excluded. We calculated the weighted diabetes and BMI GRS based on 18 and 14 variants that identified and validated in East Asians. Liver steatosis was determined by ultrasound. Results: MR analysis is regarded as independent of confounders affecting the exposure-outcome relationship and guiding definite causal direction. Using MR analysis, in total subjects the causal odds ratio (OR MR ) of genetically determined BMI for NAFLD were 1.434 (95% CI 1.234, 1.665), and for genetically determined diabetes, the association was insignificant (OR 1.098, 95%CI 0.890, 1.354). Then, we found DM_GRS and BMI_GRS significantly interacted ( P for interaction 0.018). In stratification analyses, in the lowest BMI_GRS quartile, OR MR of diabetes for NAFLD were 1.581 (95%CI 1.071, 2.233) and the association was insignificant in three higher BMI_GRS quartiles. Conversely, in subjects within DM_GRS from quartile one to three, the OR MR s of BMI for NAFLD were 1.769 (95%CI 1.203, 2.601), 1.325 (95%CI 1.025, 1,711) and 1.822 (95%CI 1.025, 2.933) but the association was insignificant in highest DM_GRS. Conclusions: The causal associations of type 2 diabetes and BMI with NAFLD were significantly modulated by the diabetes and adiposity genetic risk showing a wax-and-wane pattern. Individuals with obesity and diabetic patients with low adiposity genetic risk may need screening for NAFLD.