Causal Association between Circulating Lipids and Cervical Cancer in European and East Asian Populations: A Bidirectional and Multivariate Mendelian Randomization Study

Abstract

Background: Numerous epidemiological and genetic studies have probed the potential link between circulating lipid levels and the risk of various cancers, with a specific focus on cervical cancer (CC). While several observational studies have delved into the impact of lipids on CC, their conclusions are often compromised by confounding variables, reverse causality, and inconsistent findings. Mendelian-randomization, an approach that uses genetic variation as an instrumental variable to explore causality, effectively addresses these challenges, offering more reliable insights. In this context, we aimed to assess the influence of lipid profiles on cervical cancer among individuals of European and East Asian descent using Mendelian randomization. A nuanced understanding of the relationship between lipid profiles and cervical cancer across these populations could pave the way for innovative predictive models and preventive measures against cervical cancer.

Method: This study employed two-sample bidirectional Mendelian randomization and multivariate Mendelian randomization (MVMR) methods for the analysis. Lipid profile data from the European population were sourced from the UK Biobank and GWAS databases, encompassing HDL cholesterol (n=403,943), LDL cholesterol (n=440,546), triglycerides (n=343,992), and total cholesterol (n=115,082). Additionally, cervical cancer data were amalgamated from the MRC-IEU and GWAS resources (n=462,933). Data from the East Asian population were derived from the East Asian GWAS study, incorporating summary statistics for HDL cholesterol (n=2,342), LDL cholesterol (n=72,866), triglycerides (n=111,667), total cholesterol (n=128,305), and cervical cancer (n=462,933).

The primary analyses were confirmed using inverse variance weighting as the principal method, with the addition of weighted median, MR-Egger, and simple median estimates. The MR-Egger intercept test was employed to evaluate horizontal pleiotropy, while heterogeneity was measured using Cochran's Q. High-density lipoprotein cholesterol (HDL-C), low-density lipoprotein cholesterol (LDL-C), and triglycerides (TG) were included as core lipid predictors in the MVMR model. All analyses adhered to the international Mendelian randomization reporting standards for quality control.

Result:  The causal inference analysis, based on Mendelian randomization, demonstrated a positive association between genetically predicted high-density lipoprotein cholesterol (HDL-C) and the risk of cervical cancer in populations of European descent (IVW β=0.20, 95% CI=1.04-1.44, p=0.017). Conversely, triglycerides (TG) exhibited a negative causal association (β=-0.26, 95% CI=0.65-0.92, p=0.003). No significant causal effect was observed for low-density lipoprotein cholesterol (LDLC) or total cholesterol (TC) (p>0.3). The reverse MR analysis did not indicate a reverse causal effect of cervical cancer on lipid metabolism. Sensitivity analyses confirmed the absence of pleiotropy (Pleiotropy p>0.05) and non-significant heterogeneity (Q_ p>0.05) in the forward MR instrumental variables. A random-effects model was employed to correct for heterogeneity in the reverse analyses. Only TG demonstrated a significant negative causal relationship with cervical cancer in populations of Asian descent (IVW β =-0.59, 95% CI=0.32-0.96, p=0.034). This association remained robust after multivariate MR analysis (MV-IVW β=-0.70, 95% CI=0.31-0.80, p=0.004). Conversely, reverse MR showed a weak positive impact of cervical cancer on LDL-C (β=0.013, 95% CI=1.003-1.023, p=0.011). All analyses were further confirmed by MR-Egger regression and the leave-one-out method, revealing no potential confounding bias. 

Conclusion: Our research uncovered a potential causal link between HDL-C and TG levels and the incidence of cervical cancer. Furthermore, we observed significant heterogeneity in these relationships within European and Asian populations

Keywords: circulating lipids, cervical cancer, Mendelian randomization, causal inference, genetic epidemiology, instrumental variables