Abstract
BACKGROUND: Gestational diabetes mellitus (GDM) is a multifaceted and complex condition. Genetic factors, maternal exposure to bisphenol A (BPA), and thyroid-stimulating hormone (TSH) levels have been associated with GDM. However, existing findings are inconsistent, and evidence regarding their interactions remains limited. This study aimed to identify single-nucleotide variants (SNVs) associated with GDM and to examine whether the genetic influence on GDM would be modulated by maternal BPA and TSH levels during pregnancy. METHODS: This case-control study was nested within a prospective cohort of 2,884 pregnant women in South China from July 2016 to December 2020. Significant SNVs between cases and controls were identified by whole-exome sequencing and validated by Sequenom MassARRAY. Functional and pathway enrichment analyses were applied to explore potential biological pathways. The relationship between GDM and maternal SNVs' genotype, BPA, and TSH was evaluated by logistic regression models and marginal effect analyses. RESULTS: We identified 308 missense variants among 1,770 SNVs linked to GDM. After validation, the allele frequencies of PPARGC1A rs8192678 C > T (p = 0.005, FDR = 0.077) and GCK rs2971672 A > C (p = 0.007, FDR = 0.077) showed significant differences between cases and controls. In an exploratory analysis using logistical regression, the odds ratio (OR) for GDM was 0.417 (95% CI: 0.225-0.774) among women with the TT genotype of PPARGC1A rs8192678 and 0.470 (95% CI: 0.262-0.846) among those with the CC genotype of GCK rs2971672 compared to the wild type. Sub-population analysis revealed that urinary BPA levels were linked to an increased risk of GDM, with an OR of 2.295 (95% CI: 1.361-3.867). The protective effect ofPPARGC1A rs8192678 in GDM was confirmed and was non-linearly modified by sqrt-BPA levels. Additionally, this effect was modified by sqrt-TSH in a dose-dependent manner. The protective association was strongest at moderate BPA exposure levels (e.g., at sqrt-BPA = 2 and 3, the dy/dx for CT + TT vs. CC was -0.20 and -0.194, respectively; p < 0.01). At the highest level of BPA or TSH, the protective genetic effect was attenuated and became statistically non-significant. CONCLUSION: The study highlights the associations between GDM and the missense variant of PPARGC1A rs8192678, further revealing that the genetic effect is modified slightly by urinary BPA and serum TSH levels. The modification displayed a quasi-U-shaped distribution in relation to BPA and decreased as TSH levels increased.