Evaluation for causal effects of socioeconomic traits on risk of female genital prolapse (FGP): a multivariable Mendelian randomization analysis.

BACKGROUND

Although observational studies have established some socioeconomic traits to be independent risk factors for pelvic organ prolapse (POP), they can not infer causality since they are easily biased by confounding factors and reverse causality. Moreover, it remains ambiguous which one or several of socioeconomic traits play predominant roles in the associations with POP risk. Mendelian randomization (MR) overcomes these biases and can even determine one or several socioeconomic traits predominantly accounting for the associations.

OBJECTIVE

We conducted a multivariable Mendelian randomization (MVMR) analysis to disentangle whether one or more of five categories of socioeconomic traits, "age at which full-time education completed (abbreviated as "EA")", "job involving heavy manual or physical work ("heavy work")", "average total household income before tax (income)", "Townsend deprivation index at recruitment (TDI)", and "leisure/social activities" exerted independent and predominant effects on POP risk.

METHODS

We first screened single-nucleotide polymorphisms (SNPs) as proxies for five individual socioeconomic traits and female genital prolapse (FGP, approximate surrogate for POP due to no GWASs for POP) to conduct Univariable Mendelian randomization (UVMR) analyses to estimate causal associations of five socioeconomic traits with FGP risk using IVW method as major analysis. Additionally, we conducted heterogeneity, pleiotropy, and sensitivity analysis to assess the robustness of our results. Then, we harvested a combination of SNPs as an integrated proxy for the five socioeconomic traits to perform a MVMR analysis based on IVW MVMR model.

RESULTS

UVMR analyses based on IVW method identified causal effect of EA (OR 0.759, 95%CI 0.629-0.916, p = 0.004), but denied that of the other five traits on FGP risk (all p > 0.05). Heterogeneity analyses, pleiotropy analyses, "leave-one-out" sensitivity analyses and MR-PRESSO adjustments did not detect heterogeneity, pleiotropic effects, or result fluctuation by outlying SNPs in the effect estimates of six socioeconomic traits on FGP risk (all p > 0.05). Further, MVMR analyses determined a predominant role of EA playing in the associations of socioeconomic traits with FGP risk based on both MVMR Model 1 (OR 0.842, 95%CI 0.744-0.953, p = 0.006) and Model 2 (OR 0.857, 95%CI 0.759-0.967, p = 0.012).

CONCLUSION

Our UVMR and MVMR analyses provided genetic evidence that one socioeconomic trait, lower educational attainment, is associated with risk of female genital prolapse, and even independently and predominantly accounts for the associations of socioeconomic traits with risk of female genital prolapse.