RESUMO
Prospective epidemiological evidence was lacking on the association of phthalates (PAEs) exposure with incident type 2 diabetes mellitus (T2DM) risk. In present nested case-control study, we identified 1006 T2DM cases and matched 1006 controls based on Dongfeng-Tongji cohort study, and 6 PAEs were detected in baseline serum. The conditional logistic regression model, Bayesian kernel machine regression (BKMR) model and Quantile-based g-computation were applied to evaluate the associations of determined PAEs, either as individuals or as a mixture, with incident T2DM risk. Subgroup analysis was conducted to identify the potential sensitive population of PAEs effects on T2DM. After multiple adjustment, no statistically significant association was observed between single or mixture of PAEs and incident T2DM risk in the whole population. However, serum levels of Di-n-butyl phthalate (DnBP) [OR= 2.06; 95% CI: (1.11-3.96)], Σdibutyl phthalate (ΣDBP) [OR= 1.96; 95% CI: (1.06-3.76)], and Σlow-molecular- weight phthalate (ΣLMW) [OR= 2.27; 95% CI: (1.17-4.57)] were significantly associated with T2DM in current drinker group. Moreover, significant potential interactions were observed among Di-iso-butyl phthalate (DiBP), DnBP, Butyl-benzyl phthalate (BBP), ΣDBP, and ΣLMW with drinking status on T2DM risk (P for interaction = 0.036, 0.005, 0.049. 0.010, and 0.005). We did not find significant associations between serum PAEs levels and T2DM in the whole population. However, current alcohol drinkers expose to higher levels of DnBP, ΣDBP, and ΣLMW had higher risk of T2DM.