Comprehensive Mendelian randomization analysis of low-density lipoprotein cholesterol and multiple cancers Authors: Hengchang Liang, Chunling Tang, Yue Sun, Mingwei Wang, Tong Tong, Qinquan Gao, Hui Xie, Tao Tan Published: 2024 ABSTRACT AbstractPurposeThe aim of this study was to investigate the causal relationship between low-density lipoprotein cholesterol (LDL-C) and five cancers (breast, cervical, thyroid, prostate and colorectal) using the Mendelian Randomization (MR) method, with a view to revealing the potential role of LDL-C in the development of these cancers.MethodsWe used gene variant data and disease data from the Genome-Wide Association Study (GWAS) database to assess the causal relationship between LDL-C and each cancer by Mendelian randomisation analysis methods such as inverse variance weighting and MR-Egger. Specifically, we selected Proprotein convertase subtilisin/kexin type 9 (PCSK9) and 3-hydroxy-3-methylglutaryl-CoA reductase (HMGCR), genes associated with LDL-C levels, as instrumental variables, extracted the corresponding single nucleotide polymorphism (SNP) data and analysed the associations of these SNPs with five cancers.In addition, sensitivity analyses and heterogeneity tests were performed to ensure the reliability of the results.ResultsThe analyses showed that when using HMGCR gene, LDL-C were significantly and positively associated with breast (OR:1.200, 95% CI:1.082–1.329, p = 0.001), prostate (OR:1.198, 95% CI:1.050–1.366, p = 0.007), and thyroid cancers (OR:8.291, 95% CI:3.189- 21.555, p = 0.00001) were significantly positively correlated, whereas they were significantly negatively correlated with colorectal cancer (OR:0.641, 95% CI:0.442–0.928, p = 0.019); the results for cervical cancer were not significant (p = 0.050). When using the PCSK9 gene, LDL-C levels were significantly and positively associated with breast (OR:1.107, 95%:CI 1.031–1.187, p = 0.005) and prostate (OR:1.219, 95%:CI 1.101–1.349, p = 0.0001) cancers, but not with cervical (p = 0.294), thyroid cancer (p = 0.759) and colorectal cancer ( p = 0.572).ConclusionAnalyses using both the HMGCR and PCSK9 genes have shown that LDL-C may be a potential risk factor for breast and prostate cancer, while analyses of the HMGCR gene have also suggested that LDL-C may increase the risk of thyroid cancer and decrease the risk of colorectal cancer.Supplementary InformationThe online version contains supplementary material available at 10.1007/s12672-024-01684-9.Keywords: Mendelian randomization, LDL cholesterol, HMGCR, PSCK9, Causality, Cancer Introduction Low-density lipoprotein cholesterol (LDL-C) is an essential component of blood lipids and has long been considered a critical risk factor for cardiovascular diseases [1]. Recent studies have started to explore the potential role of LDL-C in cancer development and progression. As the primary carrier of cholesterol, LDL-C may influence cell membrane structure and function, promoting the proliferation and migration of cancer cells [2].Despite initial investigations into the relationship between LDL-C and cancer, results have been inconsistent, necessitating further research to clarify these causal relationships. We selected breast cancer, cervical cancer, thyroid cancer, prostate cancer, and colorectal cancer for this study based on several considerations. First, these cancers have high incidence or mortality rates globally, posing significant public health challenges [3]. Second, while previous studies have extensively explored the risk factors and etiologies of these cancers, the causal relationship between LDL-C and these cancers remains controversial, highlighting the need for further investigation [4, 5]. Additionally, the availability of large sample data from genome-wide association studies (GWAS) databases for these cancers enhances the reliability and generalizability of our research findings. Previous studies have mainly used epidemiological observational studies and randomised controlled trials (RCTs) to investigate the relationship between LDL cholesterol and cancer. For example, some observational studies have reported that higher LDL cholesterol levels are associated with a higher risk of breast and prostate cancer [6]. However, there are also articles that suggest the opposite [7]. However, these studies are susceptible to confounding factors such as dietary habits, lifestyle and other metabolic risk factors. In addition, reverse causality (e.g., changes in LDL cholesterol levels as a result of cancer) may also affect the results of the studies. Mendelian randomisation (MR) method uses genetic variants as instrumental variables (IV) and simulates randomised controlled trials, which effectively reduces the effect of confounding factors.Since genes are determined at birth, the genetic variants used in the study do not change due to the onset of the disease, thus avoiding reverse causality [8–10]. 3-hydroxy-3-methylglutaryl-CoA reductase (HMGCR) and Proprotein convertase subtilisin/kexin type 9 (PCSK9) are common targets for lowering low-density lipoprotein cholesterol (LDL-C) [11]. In addition, studies have shown that HMGCR and PSCK9 are strongly associated with cancer [12, 13].Therefore, the present study will evaluate the causal relationship between LDL-C (via HMGCR and PSCK9) and multiple cancers, aiming to provide new insights and scientific evidence for the potential application of LDL-C in cancer prevention and treatment. Methods MR study design As shown in Fig. 1, we designed an MR study to systematically assess the causal relationship between LDL-C and five common cancers (breast, cervical, thyroid, prostate, and colorectal).LDL-C was used as exposure, whereas multiple cancers were used as outcomes, and HMGCR and PSCK9 were used as instrumental variables to select SNPs with strong correlations with LDL-C from GWAS statistics according to rigorous screening criteria. Research overview and design of Mendelian randomization analyses.In order to verify the existence of causal correlations in Mendelian randomisation experiments, the following conditions need to be met (1) the instrumental variable is correlated with the exposure factor (2) the instrumental variable is not correlated with the confounders(3) the instrumental variable is not directly correlated with the outcomes Data source The GWAS ID for LDL-C is ieu-a-300, which is derived from the IEU OpenGWAS project, which encompasses large-scale genetic data related to LDL-C. The GWAS ID for LDL-C is ieu-a-300. The dataset contains 173,082 samples containing 2,437,752 Single Nucleotide Polymorphisms (SNPs).The samples are from multi-ethnic populations including both males and females.The data was collected and provided by the GLGC (Global Lipids Genetics Consortium), and the lead author of the study is Willer CJ. Cervical Cancer (ieu-b-4876), Prostate cancer (ieu-b-85) and Breast Cancer (ieu-a-1126) are from the IEU OpenGWAS project, which provides large-scale GWAS datasets.Cervical cancer contains 199,086 samples including 563 cervical cancer cases and 198,523 controls. The samples are predominantly from European women and contain 8,506,261 SNPs.Prostate cancer Contains 79,148 prostate cancer cases and 61,106 controls, mainly from European men, with 20,346,368 SNPs.Breast cancer data were obtained from the GWAS dataset provided by the Breast Cancer Association Consortium (BCAC). This dataset contains 228,951 samples, including 122,977 breast cancer cases and 105,974 controls. There are also 10,680,257 SNPs.The samples are mainly from European women. The data for colorectal cancer (ebi-a-GCST90013866) and thyroid cancer (ebi-a-GCST90013867) were obtained from the European Bioinformatics Institute's (EBI) GWAS catalogue. Colorectal cancer contains 407,746 samples and 11,037,981 SNPs.Thyroid cancer sample size and data quality were assured, containing a total of 407,746 samples and 11,035,981 SNPs. SNP data for LDL cholesterol and five cancers can be found on the IEU Open GWAS project database [14] (https://gwas.mrcieu.ac.uk/), which are publicly available summary statistics. The protocol and data collection for the original GWAS were approved by the relevant ethics committees and written informed consent was obtained from each participant prior to data collection. Therefore, no additional ethical approval was required for the use of data in this study. Selection of instrumental variables To substitute for HMGCR, seven SNPs that were LDL-C-related at the genomic significance level (P < 5.0 × 10-8) and within a ± 100 kb window of the region of the gene encoding HMGCR were obtained. Similarly, 12 LDL-C-related SNPs within ± 100 kb of PCSK9 were used to substitute for PCSK9. To maximise the benefits of the tool, a modest chain imbalance (r2 = 0.30) was allowed between the relevant SNPs to ensure sufficient independence and diversity between markers while retaining a degree of correlation for more efficient statistical analysis and interpretation of results. Data analysis We analysed the data for LDL-C and five cancers (breast, cervical, thyroid, prostate, and colorectal) using five methods (MR-Egger, weighted median, inverse variance weighted [IVW], weighted mode, and simple mode), with IVW being the decisive method of analysis, and the other methods used as references [24]. We also performed a series of sensitivity analyses because they can defend against violations of certain assumptions in MR studies. First, we performed MR-Egger regression analysis, which detects and corrects for horizontal pleiotropy and provides more robust causal effect estimates. Second, we also performed a leave-one-out sensitivity analysis, systematically removing one SNP at a time to determine whether a single SNP drives the overall causal estimate. The heterogeneity test was performed using the analysis results of the MR Egger and IVW methods. Cochran's Q was used as a measurement indicator.If Q_pval was less than the significance level (usually 0.05), heterogeneity was considered to exist [25, 28–30]. After MR analysis, we set the significance threshold to pval < 0.05, at which point the results are generally considered statistically significant and the null hypothesis (that there is an association between exposure and outcome) can be rejected. For odds ratios (ORs), when OR > 1, it means that exposure increases the likelihood of the outcome (positive correlation). Conversely, exposure and outcome are negatively correlated. In particular, when OR = 1, it means that there is no association between exposure and outcome. All analyses were conducted with packages named TwoSampleMR in R software (version 4.3.1; R Development Core Team). Results HMGCR analysis results For the HMGCR, the results showed that elevated LDL-C levels were significantly associated with the risk of thyroid, prostate, breast, and colorectal cancer. Specific results are shown in Fig. 2: Mendelian randomised estimates of the association between HMGCR and five cancers (breast, cervical, prostate, thyroid, colorectal).CI, confidence interval. The observed two-sided P < 0.05 was considered statistically significant.OR = 1: indicates that there is no association between exposure and the outcome.OR > 1: indicates that exposure increases the likelihood of the outcome (positive association).OR < 1: indicates that exposure decreases the likelihood of the outcome (negative association) For thyroid cancer, in 407,746 samples (including 3,001 thyroid cancer cases and 287,288 controls), the IVW method showed that elevated LDL-C levels were associated with a significantly increased risk of thyroid cancer, with an OR of 8.29 (95% CI: 3.19–21.55), and a p-value of less than 0.001. This suggests that for every unit increase in LDL-C levels, there is an 8.29-fold increase in thyroid cancer risk. Prostate cancer: OR 1.20 (95% CI: 1.05–1.37), p-value 0.007, indicating that elevated LDL-C levels are associated with a significantly increased risk of prostate cancer. Breast Cancer: OR 1.20 (95% CI: 1.08–1.33), p-value 0.00052, indicating that elevated LDL-C levels are associated with a significantly increased risk of breast cancer. Colorectal Cancer: In 407,746 samples (including 58,131 colorectal cancer cases and 349,615 controls), the IVW method showed that elevated LDL-C levels were associated with a significant reduction in colorectal cancer risk with an OR of 0.64 (95% CI: 0.44–0.93) and a p-value of 0.018. This means that for every unit increase in LDL-C levels, the colorectal rectal cancer risk was reduced by 36%. For cervical cancer, HMGCR analysis showed no significant association between LDL-C levels and their risk, with an OR of 1.00 (95% CI: 1.00–1.01) and a P value of 0.050. PSCK9 analysis results For PSCK9, the results showed that elevated LDL-C levels were significantly associated with the risk of breast and prostate cancer. The specific results are shown in Fig. 3: Mendelian randomised estimates of the association between PCSK9 and five cancers (breast, cervical, prostate, thyroid, colorectal).CI, confidence interval. The observed two-sided P < 0.05 was considered statistically significant.OR = 1: indicates that there is no association between exposure and the outcome.OR > 1: indicates that exposure increases the likelihood of the outcome (positive association).OR < 1: indicates that exposure decreases the likelihood of the outcome (negative association) For breast cancer, in 228,951 samples (including 122,977 breast cancer cases and 105,974 controls), the inverse variance weighted (IVW) method showed that elevated LDL-C levels were associated with a significant increase in the risk of breast cancer, with an OR of 1.20 (95% CI: 1.08–1.33) and a p-value of 0.00052.This suggests that for every one-unit increase in LDL-C level, there is a 20% increase in breast cancer risk. In the analysis of prostate cancer the results showed that out of 140,254 samples (including 79,148 prostate cancer cases and 61,106 controls), the IVW method showed that elevated LDL-C levels were associated with a significantly increased risk of prostate cancer, with an OR of 1.20 (95% CI: 1.05–1.37) and a p-value of 0.007. this implies that for every one unit increase of LDL-C level, there was a 20% increase in prostate cancer risk. In addition, no abnormal results were found in the sensitivity and heterogeneity analyses. Details of the forest plots for the heterogeneity analysis, multiple validity tests, scatterplots, funnel plots, and leave-one-out sensitivity analyses can be found in the additional file. Discussion Interpretation of the findings In this study, we assessed the causal association between LDL-C and five common cancers (breast, cervical, thyroid, prostate and colorectal) using MR analysis. By analysing PSCK9 and HMGCR, we found a significant association between elevated LDL-C levels and the risk of certain cancers [15, 31]. For the HMGCR analyses, the results showed that elevated LDL-C levels were significantly associated with the risk of thyroid, prostate, breast and colorectal cancer. Particularly for thyroid cancer, the risk increased 8.29-fold for each unit increase in LDL-C levels. This may be due to the fact that LDL-C affects the metabolism and proliferation of thyroid cells through the HMGCR pathway [16]. Similarly, LDL-C was significantly associated with an increased risk of prostate and breast cancer, further supporting the potential role of LDL-C in the development and progression of these cancers [17, 18]. In addition, elevated LDL-C levels were associated with a significantly lower risk of colorectal cancer, a result that warrants further exploration and may involve different biological mechanisms [19, 32]. For the PSCK9 analysis, the results showed that elevated LDL-C levels were significantly associated with the risk of breast and prostate cancer [20]. Specifically, each unit increase in LDL-C levels was associated with a 20 per cent increase in the risk of breast and prostate cancer, respectively. This is consistent with previous findings suggesting that PSCK9 is not only tightly associated with LDL-C but also highly expressed in various tumour-derived cell lines [21]. However, PSCK9 analyses showed no significant association between LDL cholesterol levels and the risk of cervical, thyroid, and colorectal cancers, but some studies have illustrated a positive association between PSCK9 inhibitors and cervical cancers [22]. Moreover, some studies suggest that PSCK9 promotes colorectal cancer and is a therapeutic target [23].This may indicate that LDL cholesterol has a different mechanism of action in these cancers, and further studies are needed. We note that only the HMGCR results suggest a significant association between LDL-C and thyroid and colorectal cancers. This may be related to the way LDL-C is regulated by HMGCR, a key enzyme in the cholesterol synthesis pathway, which directly affects the amount of cholesterol (including LDL-C) synthesised in the body over a wide range of effects. In contrast, PCSK9, whose primary role is to regulate plasma LDL-C levels by degrading the low-density lipoprotein receptor (LDL-R), has an impact that is more limited to the recycling of LDL-C than to its synthesis process. Thus, HMGCR may influence a broader network of cellular metabolism, including the cholesterol requirement of thyroid and colorectal cancer cells. Research significance Therapeutic strategies to lower LDL-C levels, such as the use of statins, have been widely used in the prevention and treatment of cardiovascular disease [26]. The results of our study suggest that these drugs may also have potential benefits in reducing the risk of certain cancers. Specifically, for breast and prostate cancers, high levels of LDL-C may be an independent risk factor, and therefore, by lowering LDL-C levels, it may help reduce the incidence of these cancers. The findings suggest that the mechanism of action of LDL-C in different cancers may differ. For example, PSCK9 analysis showed no significant association between LDL-C levels and the risk of cervical, thyroid and colorectal cancers, which may indicate that LDL-C affects some cancers but not others through specific pathways. This provides clues for further mechanistic studies, and exploring the specific pathways and molecular mechanisms of LDL-C's action in different cancers will help to unravel the etiological and developmental processes of cancer. In the public health arena, the results of this study suggest the need to reassess strategies for LDL-C management, especially in high-risk populations. For example, in areas with a high prevalence of breast and prostate cancers, active management of LDL-C levels may help to reduce the incidence of these cancers. In addition, for thyroid cancer, the findings suggest that special attention needs to be paid to the management of LDL-C levels, as significantly elevated LDL-C levels are associated with an increased risk of thyroid cancer. The findings of this study reveal the potential risk of LDL-C in specific cancers, but cancer risk factors are multilayered, encompassing dietary, nutritional, and micronutrient factors. The impact of diet and nutrition cannot be ignored when examining risk factors for prostate cancer. Relevant studies have shown that specific dietary and nutritional factors may be associated with the development of prostate cancer [33]. In addition, micronutrients such as selenium and zinc may also have an important impact on cancer susceptibility [34]. In prostate cancer diagnosis, advanced diagnostic methods of biomarker analysis combined with cholesterol level monitoring demonstrate high potential to help improve diagnostic accuracy [35]. Limitations of the study Although this study provides new insights, there are some limitations. Firstly, MR analysis relies on the validity of the selected genetic variants as IVs. If there is horizontal pleiotropy in these instrumental variables, it may affect the accuracy of causal inference. Second, the GWAS data used in this study were mainly from European populations [27], and the results may not be applicable to other ethnic populations. Therefore, future studies should consider populations of different ethnicities and regions to verify the generalisability of these findings.In addition, some cancers have various subtypes (e.g., hormone receptor positive or negative, HER2 status), which we did not specifically explore. Future research directions Future studies should further explore the specific mechanisms of LDL-C's role in different cancers, especially for thyroid, prostate and breast cancers. In addition, investigating whether there are racial or geographic differences in the relationship between LDL-C levels and cancer risk, as well as the underlying biological basis, will help to develop personalised prevention and treatment strategies and we want to explore the potential relationship between LDL-C and a variety of cancers in the context of more data. At the same time, we also need to consider the impact of specific subtypes of cancer. Conclusion Analyses using both the HMGCR and PCSK9 genes have shown that LDL-C may be a significant potential risk factor for breast and prostate cancer. The elevated levels of LDL-C were consistently associated with an increased risk of these cancers, indicating a possible mechanistic link through lipid metabolism pathways influencing cancer cell proliferation and survival. Furthermore, analyses of the HMGCR gene alone have suggested that LDL-C may also increase the risk of thyroid cancer and decrease the risk of colorectal cancer. Specifically, elevated LDL-C levels were strongly associated with a markedly higher risk of thyroid cancer, potentially due to HMGCR-mediated cholesterol synthesis affecting thyroid cell function and growth. Conversely, the inverse relationship between LDL-C and colorectal cancer risk points to complex and possibly tissue-specific metabolic and inflammatory processes influenced by cholesterol levels. These findings highlight the multifaceted role of LDL-C in cancer development and underscore the importance of considering genetic pathways when assessing cancer risks associated with lipid levels. Supplementary Information Supplementary Material 1. Author contributions Hengchang Liang contributed mainly to data collection, code construction, and graphics, and manuscript writing. Chunling Tang contributed to several revisions. Yue Sun contributed to the design and improvement of the concept, as well as the revision of the manuscript. Mingwei Wang contributed to the design and improvement of the concept. Tong Tong contributed to the design and improvement of the concept. Qinquan Gao contributed to the improvement of the study. Hui Xie contributed to the conception of the study, data collection, code construction, and manuscript revision. Tao Tan gave the final touches to the article. Funding This work is supported by Macao Polytechnic University Grant (RP/FCA-15/2022). This work is supported by Macao Polytechnic University Grant (RP/FCA-10/2023). Data availability The data described in this Data Note can be freely and openly accessed on IEU OpenGwas Project under [https://gwas.mrcieu.ac.uk/]. Please see references [14] for details and links to the data. Declarations Ethics approval and consent to participate The protocol and data collection for the original GWAS were approved by the relevant ethics committees and written informed consent was obtained from each participant prior to data collection. Therefore, no additional ethical approval was required for the use of data in this study. Competing interests The authors declare no competing interests. Footnotes Publisher's Note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. Contributor Information Hui Xie, Email: h.xie@xnu.edu.cn. Tao Tan, Email: taotan@mpu.edu.mo. References Associated Data Supplementary Materials Supplementary Material 1. Data Availability Statement The data described in this Data Note can be freely and openly accessed on IEU OpenGwas Project under [https://gwas.mrcieu.ac.uk/]. Please see references [14] for details and links to the data.