作者
Qing-Shuang Yang,Jia-Qi Xu,Ruiling Liu,Jiao Zheng,Mei-Hua Bao,Zhao-Hui Zhong,Qian Feng,Yu-Bin Ding
摘要
Environmental toxicant exposure has emerged as a potential contributor to diabetes, yet systematic investigations integrating multiple chemicals and biological mechanisms remain limited. This study employed a hypothesis-generating, exposome-toxicogenomic framework to examine the associations between diverse environmental toxicants and diabetes risk and to explore underlying biological pathways. Data from 2689 NHANES 2013-2016 participants (366 with diabetes and 2323 without) were analyzed. Forty-six toxicants across seven chemical classes were evaluated using exposure-wide association studies, deletion/substitution/addition modeling, restricted cubic splines, Bayesian kernel machine regression, and quantile-based g-computation. Integrative bioinformatics analyses, including Comparative Toxicogenomics Database annotations, transcriptomic profiling, pathway enrichment, protein-protein interaction networks, and machine learning, were conducted to explore potential biological pathways and support biological plausibility. Five toxicants-glycidamide, ethylene oxide, antimony, uranium, and NAC-3HPM-were consistently associated with higher odds of diabetes (OR range: 1.22-1.34). Mixture analyses revealed cumulative risk amplification (qgcomp OR 1.39, 95 % CI 1.21-1.60), with ethylene oxide showing the highest posterior inclusion probability (>0.5). Stronger associations were observed among obese individuals. Bioinformatics analyses identified 87 overlapping toxicant-diabetes-related genes enriched in pathways related to oxidative stress, apoptosis, AGE-RAGE signaling, and atherosclerosis. Machine learning across 113 models (optimal: Elastic Net; training AUC 0.956, test AUC 0.867) highlighted 14 key genes, of which five (MAPK8, SIRT1, PIK3R1, KRAS, MAPK1) overlapped as hub genes in protein-protein interaction networks. These findings suggest that background-level exposure to environmental toxicants is associated with increased diabetes risk, potentially involving biologically relevant pathways related to mitochondrial function, insulin signaling, and inflammatory processes, with obesity acting as a potential susceptibility factor.