Gestational diabetes mellitus (GDM) manifests as diabetes during pregnancy in women with previously normal glucose metabolism (McIntyre et al., Reference McIntyre, Catalano, Zhang, Desoye, Mathiesen and Damm2019). As a significant public health concern, GDM incidence varies globally, estimated at approximately 14.0% worldwide (Wang et al., Reference Wang, Li, Chivese, Werfalli, Sun, Yuen, Hoegfeldt, Elise Powe, Immanuel, Karuranga, Divakar, Levitt, Li, Simmons and Yang2022) and 14.8% in China (Gao et al., Reference Gao, Sun, Lu, Liu and Yuan2019), as reported by the International Diabetes Federation (IDF). GDM jeopardizes maternal health and adversely impacts fetal development. It predisposes mothers to perinatal complications such as gestational hypertension and preeclampsia, along with adverse pregnancy outcomes such as macrosomia, cesarean delivery and preterm birth. Moreover, this condition heightens the risk of future type 2 diabetes and cardiovascular diseases for mothers. Additionally, fetuses face increased risks of neonatal complications, including hyperglycemia, hyperbilirubinemia, and respiratory distress syndrome, and in the long term, childhood obesity, metabolic syndrome, and cardiovascular diseases (Kondracki et al., Reference Kondracki, Valente, Ibrahimou and Bursac2022; Lee et al., Reference Lee, Ching, Ramachandran, Yee, Hoo, Chia, Wan Sulaiman, Suppiah, Mohamed and Veettil2018; Lenoir-Wijnkoop et al., Reference Lenoir-Wijnkoop, van der Beek, Garssen, Nuijten and Uauy2015; McIntyre et al., Reference McIntyre, Catalano, Zhang, Desoye, Mathiesen and Damm2019).
The alanine aminotransferase (ALT) level, a critical liver damage biomarker, is positively correlated with diabetes risk when it is persistently elevated, as demonstrated by prior research. Some articles also suggests that a substantial correlation between the two variables is lacking. Nonetheless, the epidemiological association between ALT and GDM has been subject to scrutiny (Hua et al., Reference Hua, Qi, Kizer, Williams-Nguyen, Strickler, Thyagarajan, Daviglus, Talavera, Schneiderman, Cotler, Cai, Kaplan and Isasi2021). Traditional risk factor identification for GDM, primarily based on observational studies, is often limited by confounding factors. Mendelian randomization (MR) analysis, a novel methodological approach, overcomes these limitations by controlling for confounders, thereby elucidating the causal relationships between variables. This study employed two-sample Mendelian randomization, leveraging large-scale genomewide association study (GWAS) data and utilizing genetic markers as instrumental variables. This approach aims to delineate the causal relationship between ALT levels and GDM incidence, laying the groundwork for enhanced prediction and intervention strategies in GDM management.
Materials and Methods
Study Design and Data Sources
This study implemented two-sample MR to examine the causal link between ALT exposure and GDM outcomes. This method, utilizing distinct and independent GWAS datasets, surpasses single-sample MR in efficacy and power. ALT levels were the exposure variable, while gestational diabetes status was the outcome of interest. Instrumental variables (IVs) for the analysis were single nucleotide polymorphisms (SNPs), chosen based on three fundamental two-sample MR assumptions: (1) a strong association of all selected IVs with ALT exposure (p < 5×10^-8); (2) the independence of all selected IVs from confounders affecting ALT as well as GDM; (3) influence of all selected IVs on gestational diabetes exclusively via ALT, without alternative pathways.
There were 437,438 discovery samples and 315,572 replication samples in the ALT GWAS from the UK Biobank, Million Veteran Program, Rotterdam Study and Lifeline Study (Pazoki et al., Reference Pazoki, Vujkovic, Elliott, Evangelou, Gill, Ghanbari, van der Most, Pinto, Wielscher, Farlik, Zuber, de Knegt, Snieder, Uitterlinden, Cohort Study, Lynch, Jiang, Said, Kaplan and Million Veteran Program2021). All participants, of European descent, provided informed consent. The GWAS summary data for gestational diabetes were sourced from a Finnish database that included 6033 gestational diabetes patients and 110,330 controls, all of European descent.
Selecting Instrument Variables
MR analysis required strict adherence to three principles: relevance, independence and exclusion restriction. Consequently, all IVs selected for further analysis underwent stringent screening. SNPs strongly associated with ALT exposure (p < 5 × 10^-8) were chosen. To ensure significance and mitigate weak IV bias, F values less than 10 were excluded. The F value was calculated as F = R^2 × (N-2)/(1-R^2), where R^2 is the variance in ALT explained by each IV. R2 = 2 × EAF × (1-EAF) × β^2, where beta indicates the allelic effect and EAF the effect allele frequency. To eliminate biases from high linkage disequilibrium among SNPs, a clumping process (r 2 < .001, physical distance = 10,000 kb) was used to ensure IV independence. Additionally, palindromic SNPs with intermediate allele frequencies were excluded to align effect alleles between the ALT and gestational diabetes datasets.
Statistical Analysis
The genetic association between ALT levels and gestational diabetes incidence was investigated using five methods: MR-Egger regression, the weighted median, the inverse-variance weighted (IVW) method, the simple mode, and the weighted mode. IVW, assuming the validity of all analyzed SNPs, was anticipated to provide the most accurate estimates and thus was the primary method in this study (Chen et al., Reference Chen, Zhang, Wu, Lei, Lei and Zhao2024). The results were statistically significant when the p value of IVW was less than .05 and IVW and MR-Egger were in the same direction. Several tests, including the Cochrane Q test and funnel plot symmetry assessment, were used to validate the results. The MR-Egger intercept test and MR-PRESSO global test were used to detect multicollinearity, with MR-PRESSO also identifying and excluding outliers to provide adjusted estimates. A leave-one-out sensitivity analysis was used to assess the impact of individual SNPs on the overall association. Statistical analyses were conducted with R software (version 4.3.2) using the TwoSampleMR package; p < .05 indicated statistical significance.
Results
Selection of Instrumental Variables
Screening identified 252 SNPs strongly linked to ALT (p < 5 × 10^-8; F value > 10) and independently related to ALT (r 2 < .001, physical distance ≤ 10,000 kb), initially serving as potential instrumental variables, with the lowest F value being 27.35. Postharmonization analysis of the ALT and gestational diabetes datasets was performed. For subsequent MR analysis, 238 SNPs were retained, including nine palindromic SNPs, namely, rs12609548, rs133015, rs13395911, rs1778793, rs4711750, rs4782568, rs7041363, rs7672435, and rs9788910. Consequently, 229 SNPs were finalized as instrumental variables.
Mendelian Randomization Analysis
Genetic links between ALT levels and gestational diabetes incidence were explored using the random-effects IVW method. A significant difference in the odds ratio (OR) was detected between people with gestational diabetes and those without gestational diabetes (p = .011, 95% CI = 2.868 [1.275-6.451]) (Table 1; Figure 1). The weighted median method corroborated a genetic causal relationship between ALT levels and gestational diabetes incidence (Figure 2). Table 1 details the five methodologies employed in our MR analysis, along with their respective outcomes.
Note: MR, Mendelian randomization; ALT, alanine aminotransferase; GDM, gestational diabetes mellitus; IVW, inverse-variance weighted.
The heterogeneity tests revealed significant variability in the impacts of genetic instrumental variables. The MR Egger method yielded a heterogeneity Q statistic of 302.6416 (degrees of freedom [df] = 227, p = .00058), highlighting notable heterogeneity among the genetic tools. The Q statistic of the IVW method was 304.5560 (df = 228, p = .00052), which further confirmed the heterogeneity. Additionally, funnel plots exhibited SNP symmetry (Figure 3).
Egger intercept and MR-PRESSO analyses indicated no pleiotropy (p = .23205), with no outliers identified in the MR-PRESSO during the analysis. The leave-one-out test confirmed that the MR analysis results were unaffected by individual SNPs, confirming the stability and robustness of the findings (Figure 4).
Discussion
This study leveraged large-scale GWAS data to examine the causal relationship between ALT levels and GDM incidence. We identified a notable association between SNPs affecting ALT levels and those affecting GDM prevalence, suggesting that prenatal interventions targeting liver disease affecting ALT can reduce the prevalence of GDM.
Increasing evidence supports a correlation between ALT levels and GDM risk. A study of 94 GDM patients reported by An et al. (Reference An, Ma, Zhang, Lin, Xiang, Chen and Tan2022) revealed a negative correlation between early pregnancy AST/ALT levels and GDM risk. Conversely, a prospective study involving 1128 patients indicated a positive association between early pregnancy ALT/AST levels and GDM (Song et al., Reference Song, Zhang, Qiao, Duo, Xu, Peng, Zhang, Chen, Nie, Sun, Yang, Wang, Lu, Sun, Fu, Dong, Yuan and Zhao2022), identifying them as independent risk factors. Research by Erdoğan et al. (Reference Erdoğan, Ozdemir, Doğan, Sezer, Atalay, Meriç, Yilmaz and Koca2014) also suggested that the ALT concentration is a predictive marker for GDM. Nevertheless, some studies have reported no significant correlation between ALT levels and GDM risk (Kong et al., Reference Kong, Liu, Guo, Gao, Zhong, Zhou, Chen, Xiong, Yang, Hao and Yang2018; Zhao et al., Reference Zhao, Zhang, Zhang, Varkaneh, Rahmani, Clark, Ryan, Abdulazeem and Salehisahlabadi2020).
There is no evidence that ALT causes GDM based on observational studies. The cooccurrence of ALT with conditions such as intrahepatic cholestasis during pregnancy and elevated AST complicates its relationship with GDM. Genome-wide association studies are instrumental in dissecting complex diseases and identifying key genetic contributors beyond single-gene analyses. Our research, using extensive data, provides genetic insight into the causal relationship between ALT levels and GDM under both intricate and interrelated conditions.
The relationship between ALT and GDM is likely complex. Liver stress, metabolic imbalances, insulin resistance, and inflammation related to ALT have implications for GDM (Peracchi & Polverini, Reference Peracchi and Polverini2022). Insulin resistance, which is crucial in GDM development, may impair liver function and elevate ALT levels (Sakurai et al., Reference Sakurai, Kubota, Yamauchi and Kadowaki2021). GDM has been associated with metabolic disorders and chronic inflammation (Bakhshimoghaddam et al., Reference Bakhshimoghaddam, Razmi, Malihi, Mansoori and Ahangarpour2023), potentially exacerbating liver stress and ALT levels (Huang et al., Reference Huang, Guo, Xu, Tang, Wang, Jin and Wang2019). Moreover, the interplay between inflammation and insulin resistance could intensify GDM progression (Zheng et al., Reference Zheng, Ke, Feng, Yuan, Zhou, Yu, Wang and Feng2016). Fatty liver disease, a GDM risk factor, can increase ALT levels and contribute to GDM development (Ajmera et al., Reference Ajmera, Gunderson, VanWagner, Lewis, Carr and Terrault2016; Chen et al., Reference Chen, Hawa, Berinstein, Reddy, Kassab, Platt, Hsu, Steiner, Louissaint, Gunaratnam and Sharma2021). Hormonal changes during pregnancy may also impact insulin sensitivity and metabolism, affecting alanine aminotransferase (ALT) levels.
A strength of our study is that it is the first GWAS exploring the causal relationship between ALT and GDM. The two-sample MR method addresses observational study limitations such as reverse causation, confounding factors, and biases. Rigorous selection of instrumental variables ensured accurate results. Various tests for sensitivity, horizontal pleiotropy, and heterogeneity reinforced the stability and reliability of the ALT-GDM association.
However, there are limitations. The participants were exclusively of European descent, leaving the generalizability of our findings to other populations uncertain. Pleiotropy was adjusted using MR intercepts and MR-PRESSO global tests, and residual confounding factors could bias the results. Finally, reliance on genome-wide association meta-analyses limits stratified analyses by country, ethnicity, or age group, potentially restricting the applicability of the observed ALT effects to specific populations.
Conclusion
This study has established a link between ALT and GDM, enhancing our comprehension of their inherent connection and laying the groundwork for future targeted interventions. Further investigation is essential to ascertain the generalizability of these associations and their implications for clinical practice.
Author contribution
Lihua Yin: Conceptualization, methodology, software, visualization, writing — original draft, review and editing. Yifang Hu, Xiaoxia Hu: Software, Writing — Review and editing. Xiaolei Huang and Yingyuan Chen discussed and revised the manuscript. Yisheng Zhang: Project administration, funding acquisition. All the authors read and approved the final manuscript.
Financial support
This work was supported by the Science and Technology Project of Ningbo City (No. 2019C50091), the Medical and Health Science and Technology Project of Zhejiang Province (No. 2022KY297, No. 2021Y02), and the Key Technology Research and Development Project of Ningbo (No. 2023Z185).
Competing interests
The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.