[Genetic Causation Analysis of Hyperandrogenemia Testing Indicators and Preeclampsia]

Chanyu Lin,Jingbo Chen,Xiaomiao Zhao

DOI: https://doi.org/10.12182/20240560106

2024-05-20

Abstract:Objective: Some epidemiological studies have shown that pregnant women who develop preeclampsia (PE) have elevated levels of testosterone in their maternal plasma compared to women with normal blood pressure during pregnancy, revealing a potential association between hyperandrogenism in women and PE. To explore the causal relationship between hyperandrogenism and PE, this study selected total testosterone (TT), bioavailable testosterone (BIOT), and sex hormone binding globulin (SHBG) as exposure factors and PE and chronic hypertension with superimposed PE as disease outcomes. Two-sample Mendelian randomization (MR) analyses were used to genetically dissect the causal relationships between the three exposure factors (TT, BIOT, and SHBG) and the outcomes of PE and chronic hypertension with superimposed PE. Methods: Two independent genome-wide association study (GWAS) databases were used for the two-sample MR analysis. In the GWAS data of female participants from the UK Biobank cohort, single nucleotide polymorphisms (SNPs) associated with TT, BIOT, and SHBG were analyzed, involving 230454, 188507, and 188908 samples, respectively. GWAS data on PE and chronic hypertension with superimposed PE from the Finnish database were used to calculate SNP, involving 3556 PE cases and 114735 controls, as well as 38 cases of chronic hypertension with superimposed PE and 114735 controls. To meet the assumptions of instrumental relevance and independence in MR analysis, SNPs associated with exposure were identified at the genome-wide level (P<5.0×10-8), and those in linkage disequilibrium interference were excluded based on clustering thresholds of R 2<0.001 and an allele distance greater than 10000 kb. Known confounding factors, including previous PE, chronic kidney disease, chronic hypertension, diabetes, systemic lupus erythematosus, or antiphospholipid syndrome, were also identified and the relevant SNPs were removed. Finally, we extracted the outcome data based on the exposure-related SNPs in the outcome GWAS, integrating exposure and outcome data, and removing palindromic sequences. Five genetic causal analysis methods, including inverse variance-weighted method (IVW), MR-Egger regression, weighted median method, simple mode method, and weighted mode method, were used to infer causal relationships. In the IVW, it was assumed that the selected SNPs satisfied the three assumptions and provided the most ideal estimate of the effect. IVW was consequently used as the primary analysis method in this study. Considering the potential heterogeneity among the instrumental variables, random-effects IVW was used for MR analysis. The results were interpreted using odds ratios (OR) and the corresponding 95% confidence interval (CI) to explain the impact of exposure factors on PE and chronic hypertension with superimposed PE. If the CI did not include 1 and had a P value less than 0.05, the difference was considered statistically significant. Sensitivity analysis was conducted to assess heterogeneity and pleiotropy. Heterogeneity was examined using Cochran's Q test, and pleiotropy was assessed using MR-Egger intercept analysis. Additionally, leave-one-out analysis was conducted to examine whether individual SNPs were driving the causal associations. To further validate the findings, MR analyses were performed using the same methods and outcome variables, but with different exposure factors, including waist-to-hip ratio adjusted for BMI (WHRadjBMI) and 25-hydroxyvitamin D levels, with MR results for WHRadjBMI and PE serving as the positive controls and MR results for 25-hydroxyvitamin D levels and PE as the negative controls. Results: According to the criteria for selecting genetic instrumental variables, 186, 127, and 262 SNPs were identified as genetic instrumental variables significantly associated with testosterone indicators TT, BIOT, and SHBG. MR analysis did not find a causal relationship between the TT, BIOT, and SHBG levels and the risk of developing PE and chronic hypertension with superimposed PE. The IVW method predicted that genetically predicted TT (OR [95% CI]=1.018 [0.897-1.156], P=0.78), BIOT (OR [95% CI]=1.11 [0.874-1.408], P=0.392), and SHBG (OR [95% CI]=0.855 [0.659-1.109], P=0.239) were not associated with PE. Similarly, genetically predicted TT (OR [95% CI]=1.222 [0.548-2.722], P=0.624), BIOT (OR [95% CI]=1.066 [0.242-4.695], P=0.933), and SHBG (OR [95% CI]=0.529 [0.119-2.343], P=0.402) were not significantly associated with chronic hypertension with superimposed PE. Additionally, MR analysis using the MR-Egger method, weighted median method, simple mode method, and weighted mode method yielded consistent results, indicating no significant causal relationship between elevated testosterone levels and PE or chronic hypertension with superimposed PE. Heterogeneity was observed for SHBG in the analysis with PE (Cochran's Q test, P=0.01), and pleiotropy was detected for BIOT in the analysis with PE (MR-Egger intercept analysis, P=0.014), suggesting that the instrumental variables did not affect PE through BIOT. Other instrumental variables did not show significant heterogeneity or pleiotropy. Leave-one-out analysis confirmed that the results of the MR analysis were not driven by individual instrumental variables. Consistent with previous MR studies, the results of the control MR analyses using WHRadjBMI and 25-hydroxyvitamin D levels supported the accuracy of the MR analysis approach and the methods used in this study. Conclusion: The MR analysis results suggest that current genetic evidence does not support a causal relationship between TT, BIOT, and SHBG levels and the development of PE and chronic hypertension with superimposed PE. This study suggests that elevated testosterone may be a risk factor for PE but not a direct cause.