Xiaoyin Huang,1,* Qianni Chen,2,* Qingling Su,1,* Jiamin Gong,1 Liqin Wu,2 Liangguang Xiang,3 Wanxin Li,1 Jun Chen,1 Hongwei Zhao,4 Wuqing Huang,1 Shanshan Du,5 Weimin Ye1,5,6 

1Department of Epidemiology and Health Statistics, Fujian Medical University, Fuzhou, People’s Republic of China; 2Department of Ultrasonography, Fuqing City Hospital Affiliated to Fujian Medical University, Fuqing, People’s Republic of China; 3Department of General Surgery, Fuqing City Hospital Affiliated to Fujian Medical University, Fuqing, People’s Republic of China; 4Division of Occupational and Environmental Health, University of Utah, Salt Lake City, UT, USA; 5Institute of Population Medicine, Fujian Medical University, Fuzhou, People’s Republic of China; 6Department of Epidemiology and Biostatistics, Karolinska Institutet, Stockholm, Sweden

*These authors contributed equally to this work

Correspondence: Shanshan Du, Institute of Population Medicine, Fujian Medical University, Fuzhou, People’s Republic of China, Tel +86 591 2286 9594, Email dushanshan1007@163.com Weimin Ye, Department of Epidemiology and Health Statistics, Fujian Medical University, Fuzhou, People’s Republic of China, Tel +86 591 2286 2023, Fax +86 591 2286 2510, Email ywm@fjmu.edu.cn

Purpose: We aimed to identify the association between obesity and nonalcoholic fatty liver disease (NAFLD) and to quantify the mediating effects of insulin resistance (IR) and chronic inflammation through observational studies and Mendelian randomization (MR).
Patients and Methods: In the current study, three IR-related indicators and three indicators of inflammation were included. The individual and combined mediated effects of IR and inflammation in the association between obesity and NAFLD were investigated in two cross-sectional studies, the Fuqing Cohort from China and the National Health and Nutrition Examination Survey (NHANES). Total, direct, and indirect effects were estimated through direct counterfactual imputation estimation, and the proportion of mediating effects was calculated. We applied a two-step MR to determine the causal mediating role of IR and chronic inflammation in the pathway between obesity and NAFLD by using single nucleotide polymorphisms as instrumental variables to predict obesity, IR, and inflammation genetically.
Results: In the Fuqing Cohort, all obese phenotypes were associated with an elevated NAFLD risk. Moreover, indicators of IR such as homeostatic model assessment of insulin resistance (HOMA-IR) and indicators of inflammation such as C-reactive protein (CRP) were significantly and positively associated with NAFLD risk. Individuals with obesity had significantly higher levels of IR and inflammation indicators compared to non-obese individuals. The indirect proportions of insulin and HOMA-IR accounted for 50.97– 66.72% in the associations between obese phenotypes and NAFLD risk, while the proportions of inflammation indicators were < 14%. Similar results were observed in the NHANES analysis. In the MR analysis, the indirect effects of HOMA-IR and CRP were statistically significant with a greater mediated proportion explained by HOMA-IR than CRP.
Conclusion: Through two population-based studies and MR, we found the causal mediation roles of IR and inflammation in the association between obesity and NAFLD, in which HOMA-IR and CRP showed stable, significant mediation effects. Furthermore, HOMA-IR showed a higher mediation effect than CRP. We emphasize the vital role of HOMA-IR in NAFLD monitoring.

Keywords: NAFLD, insulin resistance, inflammation, mediation analysis, Mendelian randomization