Unraveling the link between sterol ester and colorectal cancer: a two-sample mendelian randomization study (2024)

Background

Several studies reported the sterol ester (SE), a subclass of subtype of cholesterol ester (CE), is associated with the incidence of colorectal cancer (CRC). Nevertheless, the causal relationship of SE on CRC remains unknown.

Methods

A two-sample Mendelian randomization study was performed with the summary statistics of SE (27:1/14:0) which is from the largest available genome-wide association study meta-analysis(n = 377277) conducted by FinnGen consortium. The summary data were obtained from UK Biobank repository (377673 cases and 372016 controls). And we used a relative relaxed filter (p < 5 × 10^− 6 and LD r² < 0.01) of instrumental variables to explore the causal effect and complete the sensitive analysis with the threshold p < 5 × 10^− 8 and LD r² < 0.01, MR Egger intercept, MR-PRESSO, and leave-one-out method, which all support the causal assessment. Inverse variance weighted, MR-Egger, weighted median, simple mode, and weighted model, were used to examine the causal association between SE (27:1/14:0) and CRC. Cochran’s Q statistics were used to quantify the heterogeneity of instrumental variables.

Results

The IVW results showed that SE (27:1/14:0) (OR = 1.004; 95% CI 1.002, 1.005; p < 0.001) have genetic causal relationship with CRC. The results of weighted median, weighted mode, and simple mode are all consistent with IVW model. However, the result from the MR-Egger method (OR = 1.005; 95% CI 1.004, 1.009; p = 0.052) didn’t demonstrate a significant result. There was no heterogeneity, horizontal pleiotropy or outliers, and results were normally distributed. The results of MR analysis were not driven by a single SNP. And results from two filter threshold is consistent.

Conclusion

Altogether, genetically predicted sterol ester (27:1/14:0) plays a causal association role in the incidence of CRC. This finding will provide a new screening and diagnosis indicator of CRC in the future.

Colorectal cancer (CRC) is a highly prevalent malignant tumor of the digestive tract, ranking third in tumor incidence and becoming the leading cause of cancer-related death in men and the second leading cause in women recently [1]. In 2020, the annual number of new colorectal cancer cases reached 1.9million, and presently, the incidence rate is even higher, significantly contributing to the global cancer burden [2]. Typically, significant delays occurring from the onset of symptoms to the diagnosis of early-onset colorectal cancer cause the asymptomatic at the beginning [3]. Therefore, it is crucial to prevent the occurrence of colorectal cancer through targeted health interventions aiming at addressing its etiology.

Colorectal cancer (CRC) is a multifactorial disease influenced by various risk factors, such as excess ethanol intake in alcoholic drinks [4], smoking [5], overweight [6], bacterial species [7] in gut including Fusobacterium nucleatum, enterotoxigenic Bacteroides fragilis, and pks + E. coli etc. Furthermore, certain lipid metabolism pathways are linked to an elevated risk of CRC [8]. And The disordered levels of lipids in colorectal cancer tissue and patient serum may be correlated with initiation of colorectal cancer [9]. Additionally, lipids such as phosphatidylcholine, phosphatidylethanolamine, and sphingomyelin may serve as diagnostic markers for various stages of colorectal cancer [10]. Nevertheless, the association between lipids and CRC warrants further investigation and validation in a larger cohort.

Sterol ester (SE) [11], a subtype of cholesteryl ester (CE), are a group of lipids implicated in the onset and progression of CRC. Liu et al. examined the expression of enzymes associated with cholesteryl ester metabolism and found a strong correlation between the expression of enzymes involved in cholesteryl ester synthesis (ACAT1 and ACAT2) and the occurrence of CRC [12]. However, Sadek et al. [13]. discovered that free-form plant sterol esters have been shown to inhibit colon cancer via suppressing inflammation and inducing apoptosis. SE (27:1/14:0), also known as CE 14:0 [14], has been shown to be associated with insulin resistance [15]. However, its relationship with CRC has not been reported. Given the contentious association between SE (27:1/14:0) and CRC, additional investigations are warranted to elucidate this relationship.

Mendelian randomization (MR) serves as an analytical method for making causal inferences within the realm of epidemiological etiology [16]. To unveil the causal relationship between the CRC and SE (27:1/14:0), we conducted a two-sample Mendelian randomization utilizing methods including “MR-Egger,” “Weighted median (WM)”, “Inverse variance weighted,” “Simple mode,” and “Weighted mode.” Genome-wide association studies (GWAS) have pinpointed numerous variants associated with diseases and traits located within noncoding regions of the genome [17]. By incorporating instrumental variables as genetic predictors, the causal relationship between genes and diseases remains unaffected by common confounding factors such as environmental influences, socioeconomic variables, and individual behaviors. Thus, this study is to investigate the causal relationship between the CRC and SE (27:1/14:0) and provide some clinical helpful clinical insights for the diagnosis of CRC patients.

Exposures: Ottensmann et al. analysis

We utilized data from Ottensmann et al.‘s genome-wide association analysis of plasma lipidome, which identified 495 genetic associations with 179 lipids [18]. Ottensmann et al. reported 11,318,730 associated genome-wide SNPs from 377,277 FinnGen participants. Initially, we applied a threshold (p < 5 × 10^− 8) to identify independent SNPs. Subsequently, we clustered the SNPs within a 5000-kb region based on linkage disequilibrium (LD) with an r² > 0.01. Due to the limited number of SNPs, we set a relatively relaxed threshold (p < 5 × 10^− 6) and clumped LD r² > 0.01 in the 5000-kb region to obtain enough IVs to complete sensitivity analysis.

Outcomes

Summary statistics of CRC were obtained from a GWAS comprising 377,673 cases and 372,016 controls of European population [19]. Our exposure and outcome population samples were non-overlapping.

We use the significant independent SE (27:1/14:0) associated SNPs to assess the causal effect between CRC and SE (27:1/14:0). All of the SNPs are available in the Summary data of CRC. The SNPs were used as IVs to perform the MR analysis (Supplementary Table 1). Ethical approval was obtained for each of the original GWAS, and detailed information can be found in the respective publications.

MR analysis

To investigate the causal relationship via the MR analysis, the IVs should satisfied the three assumptions [20]: (1) the selected IVs should be directly associated with sterol ester (27:1/14:0); (2) the selected IVs should not be associated with confounders; (3) the selected IVs must exert no effects on the CRC other than SE (27:1/14:0) (Instrument Strength Independent of Direct Effect (INSIDE) assumption).

We utilized the classic Inverse Variance Weighted model (IVW) to perform the primary MR analysis, known for providing a stable and accurate assessment of causal relationships in the absence of directional pleiotropy [21, 22]. Furthermore, MR-Egger regression, WM, Simple Mode, and Weighted Mode were employed as complementary methods to evaluate the causal effect. Under the INSIDE assumption, the MR-Egger method can estimate the horizontal average pleiotropic effect by conducting a weighted linear regression that relies on instrument strength independent of direct effects [20, 23]. The WM method plays a crucial role when the majority of genetic variants are invalid. It allows for obtaining a robust overall causal estimate, ensuring reliability even in the presence of invalid variants [24]. The Simple Mode and Weighted Mode are both mode-based methods, capable of estimating the causal effect of individual SNPs to form clusters. To elaborate, the Simple Mode selects the causal estimation from the largest cluster of SNPs, while the Weighted Mode assigns weights to each SNP [25].

The TwoSampleMR package (version 0.5.10) in R (version 4.2.3) was utilized to conduct the five MR methods.

Pleiotropy, heterogeneity, and sensitivity evaluation

The results from all five methods are required same directions, with p-values less than 0.05 indicating significance. We used the MR-Egger method to return the intercept values for testing directional pleiotropy (p-value less than 0.05 was considered indicative of directional pleiotropy) [23]. Cochran’s Q statistic (Cochran Q-derived p < 0.05) was used as a marker of heterogeneity, and it served as an assessment of potential horizontal pleiotropy in the IVW analysis. To further strengthen our results, we also employed the MR-Pleiotropy Residual Sum and Outlier (MR-PRESSO) method to assess and correct for horizontal pleiotropy [26]. This method is less biased and offers greater precision compared to IVW and MR-Egger, especially when the percentage of horizontal pleiotropic variants is below 10% [27]. We also performed Leave-one-out and single SNP analysis to determine whether a single SNP was driving the causal estimates. Additionally, another MR analysis was conducted with a stricter threshold of p < 5 × 10^− 8) and LD with r² < 0.01 for the complement of sensitivity analysis (supplementary materials).

After filtering of the SNPs from the Finngen at a genome-wide significance threshold of p < 5 × 10^− 6 and LD r² < 0.01, 14 SNPs associated with sterol ester (27:1/14:0) were identified (supplementary Table 1). We conducted random-effects IVW models, along with two-sample MR analysis, utilizing these SNPs as instruments. The results indicate a genetically predicted causal effect of SE on the risk of CRC (OR = 1.004; 95% CI 1.002, 1.005; p < 0.001) (Fig.1A and C). Though, the result from the MR-Egger method (OR = 1.005; 95% CI 1.004, 1.009; p = 0.052) didn’t demonstrate a significant result (supplementary Table 2).

The results of WM, Simple mode, and weighted mode are all consistent with IVW models (Fig.1A). There was no heterogeneity observed among the selected SNPs, as demonstrated by both the MR-Egger (Cochran’s Q = 9.951; p = 0.620) and IVW methods (Cochran’s Q = 10.089; P = 0.687). Additionally, MR-Egger analysis was conducted to assess directional pleiotropy among the selected SNPs, revealing no evidence of pleiotropy (β intercept = -8.63E-05; SE = 0.000233; p = 0.717) influencing the results. The MR-PRESSO analysis also indicated no heterogeneity, as shown by a global heterogeneity test result of p < 0.001. Furthermore, the symmetry of the funnel plot indicates that there was no pleiotropy (Fig.1B). The leave-one-out sensitivity analysis revealed that the causal association between sterol ester (27:1/14:0) and CRC is not driven by a single SNP (Fig.1D). The associations of each variant with SE and the risk of CRC are depicted in Figure S3.

In this study, a two-sample MR analysis was conducted using European GWAS data to investigate the relationship between CRC and sterol ester (27:1/14:0). Our findings indicate that sterol ester (27:1/14:0) exhibits a causal effect on CRC risk. To validate the conclusion, an MR analysis was also conducted using the 2 instrumental variables (IVs) meeting the threshold criteria (p < 5 × 10^− 8 and LD r² < 0.01). We employed the IVW method and obtained a similar result (OR = 1.007; 95% CI 1.002, 1.11; p = 0.007), reinforcing the same conclusion (supplementary materials).

SE (27:1/14:0), as a subtype of cholesterol esters, has been demonstrated to be associated with CRC [28, 29]. In a study conducted by Bershteĭn et al. [28]. on CRC tissue, they observed a significant increase in cholesterol esters among CRC patients. Additionally, Munir et al. [29]. compared isogenic primary and metastatic colon cancer cell lines and observed that, compared to SW480, CE content is significantly decreased in SW620 cells. This suggests that CE may play an important role in CRC progression. Enzymes involved in CE metabolism are strongly correlated with the pathogenesis, progression, and heterogeneity of CRC [12]. However, the precise causal relationship between SE (27:1/14:0) and CRC, which could be beneficial for CRC diagnosis, remains unclear.

There is some experimental evidence for the association between sterol ester and CRC. In a previous study, adenomatous polyposis coli mice were fed with plant sterol ester, revealing that a high intake of plant sterol ester accelerates intestinal tumorigenesis in females. Gene expression analysis from the mucosa of the small intestine illustrated up-regulated genes associated with cell cycle control and cholesterol biosynthesis in female mice fed with plant SE [30]. SE (27:1/14:0) may influence the cell cycle of epithelial cells, potentially leading to the development of CRC.

This study explores the causal relationship between SE (27:1/14:0) and CRC without via MR methods. All the methods indicate a significant result except MR-Egger method while MR-Egger analysis revealed no evidence of pleiotropy (β intercept = -8.63E-05; SE = 0.000233; p = 0.717). To validate the selection of IVs and exclude potentially relevant variants, thereby enhancing the robustness of the causal estimate, we conducted an MR-PRESSO analysis, which revealed a global heterogeneity test result of p < 0.001. Thus, the non-significant from MR-Egger may due to the limited number of samples and its low power. In addition, all the SNPs were filter at the threshold of p < 5 × 10^− 6 and LD r² < 0.01, indicating a lower likelihood of weak instrument bias. And the MR result from the threshold of p < 5 × 10^− 8 and LD r² < 0.01 is consist with the previous one which is further decrease the potential bias. What’s more, we used several approaches to assess the sensitive and pleiotropy, indicating the similar result.

Nonetheless, this study has some limitations that need to be acknowledged. Firstly, all SNP data are from European cohorts, and further validation of our findings in other ethnicities is needed. Additionally, we employed a relatively relaxed threshold (p < 5 × 10^− 6 and LD r² < 0.01), as only 2 SNPs were identified under the more stringent criteria of p < 5 × 10^− 8 and LD r² < 0.01. Furthermore, the experimental validation is warrant for the causal relationship between SE (27:1/14:0). Lastly, SE (27:1/14:0) is just a subclass of sterol esters, or even lipids in general. Further exploration is needed to better understand the broader role of lipids in CRC.

Altogether, the association between genetically predicted SE (27:1/14:0) and CRC has been established. This finding will contribute to the screening and diagnosis of CRC in the future.

SE:

sterol ester

CE:

cholesterol esters

CRC:

Colorectal cancer

LD:

linkage disequilibrium

IVW:

Variance Weighted model

WM:

Weighted Median

IVs:

instrumental variables

Authors and Affiliations

1. Department of General Surgery, The Ganzhou People’s Hospital, 18 Meiguan Avenue, Ganzhou, Jiangxi, China

   Chuanyuan Liu,Junfeng Xie,Baolong Ye,Junqiao Zhong&Xin Xu

2. Department of Ultrasound, The First Affiliated Hospital of University of South China, No.69, Chuanshan Road, Shigu District, Hengyang, Hunan, 421000, China

   Xin Xu

Contributions

CYL: Methodology, data curation, formal analysis, visualization, original draft writing. JFX, BLY, JQZ: revised the manuscript, and XX: design of the work. All authors reviewed the manuscript.

Corresponding author

Correspondence to Xin Xu.

Ethics approval and consent to participate

Not applicable.

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.

Publisher’s note

Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Open Access This article is licensed under a Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International License, which permits any non-commercial use, sharing, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if you modified the licensed material. You do not have permission under this licence to share adapted material derived from this article or parts of it. The images or other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by-nc-nd/4.0/.

Reprints and permissions