Causal Relationships between 1400 Serum metabolite Traits, 179 Plasma Lipids, and Hemorrhoids: A Mendelian Randomization Study and Meta-analysis

Research Paper |

| Full text views: times Abstract Download

Lu Hu^1,+, Danyang Wang^2,+, Chang You¹, Ziyan Zhao³, Wanlin Zheng⁴ , Luyao Wang², Yutong Li¹, Liangzhe Dai⁵, Hongkai Yu^2,*

1 School of Stomatology, Nanjing Medical University, Nanjing, China.
2 The Second School of Clinical Medicine, Nanjing Medical University, Nanjing, China.
3 School of Rehabilitation Medicine, Nanjing Medical University, Nanjing, China.
4 School of Public Health, Nanjing Medical University, Nanjing, China.
5 The First School of Clinical Medicine, Nanjing Medical University, Nanjing, China.
✝ These authors contributed equally to this work.
* Corresponding author: Hongkai Yu, E-mail: yuhongkai@stu.njmu.edu.cn

Citation:

Hu, L., Wang, D., You, C., Zhao, Z., Zheng, W., Wang, L., Li, Y., Dai, L., & Yu, H. (2025). Causal Relationships between 1400 Serum Metabolite Traits,179 Plasma Lipids, and Hemorrhoids: A Mendelian Randomization Study and Meta-analysis. Life Conflux, 1(4), e202. https://doi.org/10.71321/8s70we60

Copy

Abstract

Background: Hemorrhoids is a common anorectal disorder that significantly impacts patients' quality of life over the long term and imposes a substantial economic burden. However, the potential link between HD and serum metabolites and lipids has been scarcely studied, and its pathogenesis remains unclear.

Methods: The causal relationships between serum metabolite traits, plasma lipids, and HD were evaluated using Mendelian Randomization. Inverse Variance Weighted was the most important analysis approach. Sensitivity analysis was used to assess the robustness of the results. Enrichment analysis was applied to the metabolites obtained from the results of MR analysis and meta-analysis.

Results: The risk factors included Citramalate levels (OR = 1.487, 95%CI = 1.172-1.888, P = 0.001), P-cresol sulfate levels (OR = 1.476, 95%CI = 1.111-1.961, P = 0.007), and Triacylglycerol (53:3) levels (OR = 1.262, 95%CI = 1.049-1.518, P = 0.014). The protective factors included Imidazole lactate levels (OR = 0.865, 95%CI = 0.756-0.989, P = 0.034), 1-(1-enyl-stearoyl)-GPE (p-18:0) levels (OR = 0.722, 95%CI = 0.552-0.944, P = 0.017), and Phosphatidylethanolamine (18:1_18:1) levels (OR = 0.803, 95%CI = 0.666-0.969, P = 0.022). Enrichment analysis showed that risk factors were enriched in Alanine Metabolism (Holm P = 0.0039), Cysteine Metabolism (Holm P = 0.0146), Urea Cycle (Holm P = 0.0181), and Ammonia Recycling (Holm P = 0.0245). Protective factors were enriched in the Urea Cycle (Holm P = 0.00754), Ammonia Recycling (Holm P = 0.0102), Aspartate Metabolism (Holm P = 0.0147), Glutamate Metabolism (Holm P = 0.038), Arginine and Proline Metabolism (Holm P = 0.0479), and Malate-Aspartate Shuttle (Holm P = 0.0495).

Conclusion: Previous studies often focused on the causal relationships between a few potential risk factors and HD in a single population. Our study is the first to address this gap by integrating a broad range of 1400 metabolites and 179 lipids with HD across diverse populations, and identifying specific metabolic pathways involved in HD development.

Keywords: Serum Metabolites; Plasma Lipids; Hemorrhoids; Mendelian Randomization; Meta-analysis.

Introduction

Hemorrhoids (HD) is one of the most common types of anorectal disorders. Epidemiological studies indicate that over 50% of individuals experience at least one episode of HD before the age of 50 [1]. Furthermore, studies have suggested that HD may be associated with an increased risk of colorectal cancer [2]. The pathogenesis of HD is complex and involves multiple factors, including dietary habits, lifestyle, local inflammation, and abnormalities in local anatomical structures [3, 4]. HD is prevalent worldwide and has a long-term and significant negative impact on patients' quality of daily life. However, the scientific community has yet to fully understand the specific pathophysiological processes of HD and its association with risk factors.

Numerous studies have indicated significant correlations between serum metabolites and HD triggers, such as constipation, diarrhea, and varicose veins. For example, C4 and fibroblast growth factor 19 (FGF19) in serum are considered biomarkers for bile acid diarrhoea [5], highlighting the importance of serum metabolites in the pathogenesis of diarrhea, which itself may trigger HD. Notably, many studies have discovered complex interactions between serum metabolites and the gut microbiota [6-8]. Evidence from Mendelian Randomization studies supports a causal correlation between the gut microbiota and the occurrence of HD [9]. Although existing evidence suggests a potential link between serum metabolites and HD, the direct causal relationship between the two requires further research to be clarified.

Lipids are important components of cell membranes and are involved in many physiological functions, including energy storage, signal transduction, and immune responses [10-12]. Studies indicate that plasma lipids may be associated with various inflammatory diseases [13], and HD, as an inflammatory anorectal disease, may be related to vascular inflammation and tissue damage. For instance, oxidized low-density lipoprotein (ox-LDL) can activate endothelial cells and smooth muscle cells, promoting the release of inflammatory factors [14, 15], which may exacerbate the local inflammatory state in HD patients. Additionally, dyslipidemia is related to an imbalance in the gut microbiota [16], which has been proven to be associated with various intestinal diseases, including HD [9, 17]. Therefore, regulating plasma lipid levels and improving dyslipidemia may have positive significance in the prevention and treatment of HD. However, further in-depth research is needed to clarify the specific mechanisms between dyslipidemia and HD.

Mendelian Randomization (MR), as an innovative method for causal inference, has been widely applied in epidemiological and genetic research in recent years. MR utilizes single nucleotide polymorphisms (SNPs) as naturally occurring instrumental variables (IVs). By analyzing the association between genetic variants and diseases, MR infers the potential causal relationship between exposure factors and outcomes. Since genetic variations are determined before an individual's birth and are not directly influenced by environmental factors, MR analysis can simulate the conditions of randomized controlled trials, effectively reducing confounding factors and reverse causation, thereby providing more reliable causal inference evidence.

Previous MR studies have identified obesity and plasma low-density lipoprotein (LDL) levels as risk factors for HD [18, 19], but these studies often focused on only a few potential risk factors. For the first time, this study conducted a comprehensive scan of the causal relationship between HD and metabolites and lipids in blood, using 1,400 serum metabolites and 179 plasma lipids as exposure factors. Moreover, previous MR studies often focused on a single population, raising concerns about the generalizability of their conclusions to other populations. To address this issue, we obtained GWAS data from individuals of African American or Afro-Caribbean, South Asian, and East Asian descent, and performed a meta-analysis of MR results across these populations to obtain robust conclusions. We also conducted external validation using GWAS data on HD from individuals of European ancestry.

Materials and methods

Study design
The three assumptions underlying the causal interpretation of MR analysis and the design of this study are displayed in Figure 1. The IVs used in the analysis need to satisfy three assumptions: (1) IVs must be strongly associated with the exposure variable; (2) IVs must follow the principle of random assignment and not be influenced by confounding factors; (3) IVs should not be directly related to the outcome but should influence the outcome only through the potential causal relationship between the exposure and the outcome [20].

Figure 1. Flowchart of study design. (A) Three assumptions of MR analysis. (B) Study workflow.

GWAS data for serum metabolites and HD
The data used in this study were all obtained from publicly available GWAS datasets. GWAS summary statistics for 1,400 serum metabolite traits were acquired from the genome-wide association study by Chen et al. (SNPs = 1,540,000,000 approximately). The study cohort included 8,299 individuals of European ancestry [21]. GWAS summary statistics for plasma lipids were obtained from the genome-wide association analysis by Ottensmann et al. [22] (SNPs = 11,318,730). The study sample comprised 7,147 Finnish individuals. HD data were retrieved from the IEU Open GWAS [23], sourced from the UK Biobank. The three HD datasets used in this study have the following GWAS IDs: ukb-e-455_AFR (SNPs = 15,526,533), ukb-e-455_CSA (SNPs = 981,032), and ukb-e-455_EAS (SNPs = 8,191,759). The ukb-e-455_AFR dataset includes samples from 6,458 African American or Afro-Caribbean individuals. The ukb-e-455_CSA dataset includes samples from 8,640 South Asian individuals. The ukb-e-455_EAS dataset includes samples from 2,658 East Asian individuals. Table S1 presents the baseline characteristics of the GWAS summary statistics used in preliminary analysis. In external validation, we selected ebi-a-GCST90086078 as the GWAS dataset for HD. The ebi-a-GCST90086078 data were derived from 56,637 individuals of European ancestry.

Selection of instrumental variables
The selected IVs must meet three criteria: First, SNPs associated with serum metabolite traits and plasma metabolites must have a genome-wide significance threshold of 1×10⁻⁵. Second, to ensure the independence of selected SNPs, the linkage disequilibrium (LD) threshold is set to r² ≤ 0.001 with a physical distance greater than 10,000 kb to eliminate LD. Third, the strength of the IVs is evaluated using the F-statistic, and SNPs with an F-statistic less than 10 are excluded.

Mendelian randomization analysis
MR analysis was performed using the "TwoSampleMR" package (version 0.6.4) in R (version 4.3.1). Statistical significance was defined as a P < 0.05. The causal relationship between exposure and HD was estimated using three methods: Inverse Variance Weighted (IVW), MR-Egger, and Weighted Median. IVW was the primary analysis method, as it employs weighted linear regression to assess the association between IVs and the outcome. When the genetic variants satisfy the three assumptions of instrumental variables and are not affected by pleiotropy, IVW provides a consistent estimate of the causal effect between exposure and outcome. MR-Egger accounts for potential heterogeneity in IVs and can detect and correct for bias, thus being used to evaluate horizontal pleiotropy [24]. When the results of these three methods were inconsistent, the IVW results were used as the primary evaluation, while the "forest plot" packages were employed for graphical representations. We followed the STROBE-MR (Strengthening the reporting of observational studies in epidemiology using mendelian randomization) guidelines for MR results reporting [25].

Sensitivity analysis
Sensitivity analyses included heterogeneity testing, horizontal pleiotropy testing, and leave-one-out testing. Heterogeneity was assessed using MR-Egger and IVW methods, quantified by Cochran's Q statistic. A P > 0.05 indicated that heterogeneity was not statistically significant. Horizontal pleiotropy was evaluated using the MR-Egger regression intercept and MR-PRESSO. For the MR-Egger regression intercept, a P > 0.05 suggested that horizontal pleiotropy could be ignored [26]. For MR-PRESSO, a P > 0.05 after the Global Test indicated that horizontal pleiotropy could be ignored. The stability of the results was assessed using the leave-one-out method; if the remaining results showed no significant changes after removing a single SNP, the results were considered stable. MR-PRESSO was performed using the MR-PRESSO package (version 1.0) in R (version 4.3.1), while the "forest plot" packages were employed for graphical representations. Funnel plots were used to assess whether there were biases in the results.

Enrichment analysis
Enrichment analysis for metabolites was conducted using MetaboAnalyst 6.0 [27]. Metabolites involved in serum metabolite traits with causal relationships to HD were classified into risk factor groups and protective factor groups according to the following rules: First, for serum metabolite levels, metabolites were classified as either risk factors or protective factors based on the sign of the beta value. For serum metabolite ratios, if the beta value was positive, the numerator component was classified as a risk factor, and the denominator component was classified as a protective factor; if the beta value was negative, the classification was reversed. Enrichment analysis was conducted separately for the two groups. The data were sourced from The Small Molecule Pathway Database (SMPDB). A significance threshold of P < 0.05 was set for this study.

Results

Causal effects of serum metabolite traits on HD
Results of African American or Afro-Caribbean population
After rigorous and comprehensive sensitivity analysis and excluding all results that failed the sensitivity analysis, 23 serum metabolite traits were identified as having a reliable causal relationship with HD, with 13 serum metabolite levels and ratios being risk factors for HD and 10 being protective factors (Figure S1-4). Among the risk factors, the top three with the most significant causal relationship with HD were: Cortolone glucuronide (1) levels (OR = 1.544, 95%CI = 1.124-2.123, P = 0.007), 3-methyl-2-oxobutyrate levels (OR = 2.079, 95%CI = 1.189-3.635, P = 0.010), and Glycocholate levels (OR = 1.405, 95%CI = 1.063-1.857, P = 0.017). Among the protective factors, the top three with the most significant causal relationship with HD were: 1-(1-enyl-palmitoyl)-2-oleoyl-GPE (p-16:0/18:1) levels (OR = 0.566, 95%CI = 0.413-0.776, P = 0.0004), X-11299 levels (OR = 0.439, 95%CI = 0.257-0.747, P = 0.002), and Aspartate to phosphate ratio (OR = 0.422, 95%CI = 0.226-0.791, P = 0.007). All MR results and sensitivity analysis outcomes are listed in Table S2.

Results of South Asian population
After conducting rigorous and comprehensive sensitivity analysis and excluding results that did not pass the sensitivity test, 32 results were considered reliable. 17 serum metabolite traits were identified as risk factors for HD (Figure S1, 5-7), with the three most significant causal effects being: Gamma-glutamyltyrosine levels (OR = 1.748, 95%CI = 1.258-2.427, P = 0.001), Adenosine 5'-monophosphate (AMP) to phosphate ratio (OR = 1.910, 95%CI = 1.267-2.879, P = 0.002), and Guaiacol sulfate levels (OR = 1.851, 95%CI = 1.245-2.750, P = 0.002). 15 serum metabolite traits were identified as protective factors for HD, with the top three significant causal effects being: 1-(1-enyl-stearoyl)-GPE (p-18:0) levels (OR = 0.623, 95%CI = 0.447-0.867, P = 0.005), Sphingomyelin (d18:2/24:1, d18:1/24:2) levels (OR = 0.570, 95%CI = 0.385-0.845, P = 0.005), and Glycerol to mannitol to sorbitol ratio (OR = 0.704, 95%CI = 0.541-0.914, P = 0.009). All MR results and their sensitivity analysis results are shown in Table S3.

Results of East Asian population
34 serum metabolite traits with causal relationship with HD was identified by MR analysis and sensitivity analysis, including 14 risk factors and 20 protective factors (Figure S1, 8-10). The top three risk factors with the most significant causal relationship with HD are: X-25271 levels (OR = 4.134, 95%CI = 1.798-9.507, P = 0.001), 5-hydroxyindole sulfate levels (OR = 2.610, 95%CI = 1.469-4.638, P = 0.001), and Spermidine to pyruvate ratio (OR = 2.405, 95%CI = 1.347-4.294, P = 0.003). The top three protective factors with the most significant causal relationship with HD are: Xanthurenate levels (OR = 0.409, 95%CI = 0.220-0.759, P = 0.005), Homostachydrine levels (OR = 0.318, 95%CI = 0.140-0.723, P = 0.006), and X-21796 levels (OR = 0.402, 95%CI = 0.203-0.799, P = 0.009). All MR results that passed the sensitivity analysis and their sensitivity analysis results are shown in Table S4.

Causal effects of plasma lipids on HD
Results of African American or Afro-Caribbean population
After MR analysis and comprehensive sensitivity analysis, 4 lipids were identified to have a causal relationship with HD, 2 of which are risk factors and 2 are protective factors for HD (Figure S11-14). Phosphatidylcholine (18:2_20:4) levels (OR = 1.588, 95%CI = 1.097-2.299, P = 0.014) and Phosphatidylcholine (O-16:1_16:0) levels (OR = 1.676, 95%CI = 1.109-2.533, P = 0.014) were identified as increasing the risk of HD. Phosphatidylcholine (O-16:0_20:3) levels (OR = 0.462, 95%CI = 0.293-0.727, P = 0.001) and Phosphatidylethanolamine (O-18:1_18:2) levels (OR = 0.569, 95%CI = 0.384-0.842, P = 0.005) were identified as protective factors. Details of the results of MR analysis and sensitivity analysis were shown in Table S5.

Results of South Asian population
After a rigorous and comprehensive sensitivity analysis, 10 plasma lipids were considered to have a reliable causal relationship with HD (Figure S11-14). Among them, 8 plasma lipids were identified as risk factors for HD: Triacylglycerol (53:4) levels (OR = 1.247, 95%CI = 1.021-1.523, P = 0.030), Triacylglycerol (52:3) levels (OR = 1.332, 95%CI = 1.062-1.672, P = 0.013), Triacylglycerol (56:7) levels (OR = 1.340, 95%CI = 1.038-1.730, P = 0.024), Triacylglycerol (56:4) levels (OR = 1.373, 95%CI = 1.019-1.850, P = 0.037), Triacylglycerol (56:8) levels (OR = 1.389, 95%CI = 1.095-1.761, P = 0.007), Triacylglycerol (53:2) levels (OR = 1.399, 95%CI = 1.045-1.874, P = 0.024), Triacylglycerol (53:3) levels (OR = 1.444, 95%CI = 1.141-1.828, P = 0.002), Ceramide (d42:1) levels (OR = 1.462, 95%CI = 1.102-1.939, P = 0.008). Additionally, 2 plasma lipids were identified as protective factors against HD: Sphingomyelin (d36:2) levels (OR = 0.730, 95%CI = 0.540-0.986, P = 0.040), Sterol ester (27:1/18:1) levels (OR = 0.772, 95%CI = 0.609-0.979, P = 0.033). All MR results and their sensitivity analysis results are shown in Table S6.

Results of East Asian population
After MR analysis and sensitivity analysis, 7 plasma lipids were considered to have a reliable causal relationship with HD, and all were identified as protective factors for HD (Figure S11-14): Sterol ester (27:1/20:3) levels (OR = 0.541, 95%CI = 0.331-0.885, P = 0.014), Ceramide (d42:1) levels (OR = 0.610, 95%CI = 0.387-0.960, P = 0.033), Phosphatidylcholine (16:0_16:1) levels (OR = 0.468, 95%CI = 0.243-0.901, P = 0.023), Phosphatidylcholine (18:0_18:3) levels (OR = 0.400, 95%CI = 0.164-0.977, P = 0.044), Phosphatidylcholine (O-16:0_16:0) levels (OR = 0.352, 95%CI = 0.160-0.770, P = 0.009), Phosphatidylethanolamine (18:1_18:1) levels (OR = 0.540, 95%CI = 0.343-0.850, P = 0.008), Triacylglycerol (54:7) levels (OR = 0.600, 95%CI = 0.365-0.987, P = 0.044). Details of the results of MR analysis and sensitivity analysis were shown in Table S7.

Meta-analysis of the results of MR
A meta-analysis will be conducted on the MR analysis results of serum metabolite traits and plasma lipids whose results of MR analysis were positive in at least one population, to evaluate the overall effect of serum metabolite traits and plasma lipids on HD across samples from different populations.

Meta-analysis of the causal relationships between serum metabolite traits and HD
Figure 2 shows the meta-analysis results of the effects of serum metabolites on HD. The heterogeneity test results indicate that a fixed-effect model should be chosen (P values are all greater than 0.05). The factors that increase the risk of HD are: Citramalate levels (OR = 1.487, 95%CI = 1.172-1.888, P = 0.001), P-cresol sulfate levels (OR = 1.476, 95%CI = 1.111-1.961, P = 0.007), Guaiacol sulfate levels (OR = 1.481, 95%CI = 1.113-1.971, P = 0.007), Gamma-glutamyltyrosine levels (OR = 1.472, 95%CI = 1.127-1.923, P = 0.005), Glycocholate levels (OR = 1.202, 95%CI = 1.018-1.420, P = 0.030), X-12839 levels (OR = 1.205, 95%CI = 1.007-1.442, P = 0.042), X-21258 levels (OR = 1.287, 95%CI = 1.007-1.645, P = 0.044), Adenosine 5'-monophosphate (AMP) to phosphate ratio (OR = 1.665, 95%CI = 1.219-2.275, P = 0.001), Pyruvate to 3-methyl-2-oxobutyrate ratio (OR = 1.341, 95%CI = 1.094-1.643, P = 0.005) (Figure 2A). The factors that reduce the risk of HD are: Imidazole lactate levels (OR = 0.865, 95%CI = 0.756-0.989, P = 0.034), 1-(1-enyl-stearoyl)-GPE (p-18:0) levels (OR = 0.722, 95%CI = 0.552-0.944, P = 0.017), Lanthionine levels (OR = 0.767, 95%CI = 0.597-0.983, P = 0.037), Sphingomyelin (d18:2/24:1, d18:1/24:2) levels (OR = 0.735, 95%CI = 0.548-0.985, P = 0.040), Dihomo-linolenoylcarnitine (C20:3n3 or 6) levels (OR = 0.827, 95%CI = 0.705-0.971, P = 0.020), X-17325 levels (OR = 0.803, 95%CI = 0.656-0.984, P = 0.034), X-17351 levels (OR = 0.786, 95%CI = 0.622-0.992, P = 0.043), X-24306 levels (OR = 0.765, 95%CI = 0.609-0.960, P = 0.021), Alpha-ketoglutarate to kynurenine ratio (OR = 0.817, 95%CI = 0.686-0.973, P = 0.024), Aspartate to phosphate ratio (OR = 0.665, 95%CI = 0.460-0.962, P = 0.030), Salicylate to oxalate (ethanedioate) ratio (OR = 0.705, 95%CI = 0.537-0.925, P = 0.012) (Figure 2B).

Figure 2. Meta-analysis results of serum metabolite traits. (A) The meta-analysis identified 9 serum metabolite traits as risk factors for HD. (B) The meta-analysis identified 11 serum metabolite traits as protective factors for HD.

Meta-analysis of the causal relationships between plasma lipids and HD
Figure 3 shows the meta-analysis results of the effects of plasma lipids on HD. The heterogeneity test results indicate that a fixed-effect model should be chosen (P values are all greater than 0.05). Triacylglycerol (53:3) levels have been identified as a risk factor for HD (OR = 1.262, 95%CI = 1.049-1.518, P = 0.014) (Figure 3A). Phosphatidylethanolamine (18:1_18:1) levels are a protective factor against HD (OR = 0.803, 95%CI = 0.666-0.969, P = 0.022) (Figure 3B).

Figure 3. Meta-analysis results of plasma lipids. (A) The meta-analysis identified Triacylglycerol (53:3) levels as risk factors for HD. (B) The meta-analysis identified Phosphatidylethanolamine (18:1_18:1) levels as protective factors for HD.

Enrichment analysis of serum metabolites
According to the rules proposed in the Methods section, the serum metabolite characteristics with positive meta-analysis results were classified into risk factor groups and protective factor groups. Serum metabolites that could not be retrieved in HMDB, PubChem, and KEGG were excluded (refer to Table S8 for metabolite information in different databases). Metabolite enrichment analysis showed that the enriched metabolic pathways for the risk factor group were: Alanine Metabolism (Holm P = 0.0039), Cysteine Metabolism (Holm P = 0.0146), Urea Cycle (Holm P = 0.0181), Ammonia Recycling (Holm P = 0.0245) (Figure 4A, Table S9). The enriched metabolic pathways for the protective factor group included: Urea Cycle (Holm P = 0.00754), Ammonia Recycling (Holm P = 0.0102), Aspartate Metabolism (Holm P = 0.0147), Glutamate Metabolism (Holm P = 0.038), Arginine and Proline Metabolism (Holm P = 0.0479), Malate-Aspartate Shuttle (Holm P = 0.0495) (Figure 4B, Table S10).

Figure 4. Serum metabolites enrichment analysis results. (A) Enrichment analysis results of serum metabolites in the risk factor group. (B) Enrichment analysis results of serum metabolites in the protective factor group.

Discussion

In the development of HD, the dilation and congestion of hemorrhoidal veins are the main pathological features. Inflammation and damage of local vascular endothelium play a crucial role in the occurrence of HD [3]. However, it remains unclear whether local inflammation in hemorrhoid patients causes changes in blood composition or if abnormal blood components trigger local inflammation, leading to the development of HD. In this study, we combined metabolomics, lipidomics, and genomics for the first time to explore the causal relationships between serum metabolites, plasma lipids, and HD using MR analysis, and employed meta-analysis to investigate the overall effects of MR results across different population samples.

It was observed that many lipids or lipoids and their metabolites are included among the protective factors for HD. Lipoids such as ceramides can maintain the integrity of the intestinal barrier by enhancing tight junctions in intestinal epithelial cells [28], thereby preventing harmful substances and bacteria from entering the bloodstream [29]. Additionally, fatty acids, especially unsaturated fatty acids, play an important role in maintaining vascular health. Polyunsaturated fatty acids (PUFAs) such as eicosapentaenoic acid (EPA) can reduce the risk of HD by inhibiting platelet aggregation and decreasing inflammation and damage in the vascular endothelium [30].

Metabolite enrichment analysis results indicate that Alanine metabolism and Cysteine metabolism may play key roles in the pathogenesis of HD. Abnormal activation of Alanine metabolism may influence the tumor necrosis factor-alpha (TNF-α) signaling pathway, thereby modulating the inflammatory response and lipid metabolism processes [31, 32]. Additionally, enhanced Cysteine metabolism is closely associated with the inflammatory response in venous tissue [33]. This inflammatory response may lead to endothelial damage in the venous wall. Furthermore, enhanced Cysteine metabolism is associated with increased production of matrix metalloproteinases (MMPs) [34-36]. The upregulation of MMPs expression promotes collagen degradation and extracellular matrix (ECM) remodeling, leading to significant changes in the structure and function of the venous wall. These changes may ultimately result in the development of varicose veins [37, 38]. When these pathological changes occur in the veins of the anorectal region, they may trigger the formation of HD.

Metabolite enrichment analysis revealed a series of metabolic pathways with potential protective effects. Specifically, Aspartate metabolism, Glutamate metabolism, Arginine and Proline metabolism, and the Malate-Aspartate shuttle play key roles in regulating the pathophysiological processes related to HD. Aspartate metabolism reduces oxidative stress by affecting intracellular metabolic processes such as the TCA cycle and pyruvate metabolism, thereby improving endothelial function [39-41]. Regulation of Glutamate metabolism may have antithrombotic effects by modulating the expression of nitric oxide synthase 3 (NOS3) and the metabolism of Arginine, Glutamate, and Glutamine [42-45]. The metabolic product of Arginine, nitric oxide (NO), has been shown to suppress inflammation, reduce inflammatory responses, and minimize tissue damage [46]. Proline metabolism reduces inflammation-induced tissue damage by activating immune cells such as T cells and natural killer (NK) cells [47]. The Malate-Aspartate shuttle plays a crucial role in maintaining intracellular NADH balance by transporting NADH to the mitochondria, reducing cytoplasmic NADH concentration, and thus lowering oxidative stress production. Excess accumulation of NADH in the cytoplasm may provide substrates for NAD(P)H oxidase, thereby promoting oxidative stress and inflammatory responses [48, 49]. The Malate-Aspartate shuttle may alleviate inflammation in the anorectal region by inhibiting this pathway, thereby preventing the occurrence of HD. In summary, the regulation of these metabolic pathways may offer new insights into the prevention and treatment of HD.

Interestingly, urea cycle and ammonia recycling were observed to exhibit both promoting and inhibiting effects in the development of HD. Urea cycle and ammonia recycling are key pathways in the metabolism, utilization, and excretion of ammonia, and are crucial for maintaining ammonia homeostasis in the internal environment [50-53]. On the one hand, high ammonia concentrations have been shown to promote inflammatory responses [54, 55], which may enhance the release of inflammatory mediators and, in turn, contribute to the development of HD. On the other hand, moderate ammonia levels have been found to mitigate the damage of inflammatory responses to pancreatic β-cells through the regulation of glutamine metabolism [56, 57], suggesting that ammonia may have anti-inflammatory effects at certain concentrations. Therefore, urea cycle and ammonia recycling play a critical role in maintaining ammonia balance in the internal environment, and subtle changes in ammonia levels can influence the development of HD by modulating inflammatory responses. This finding not only reveals the complex roles of the urea cycle and ammonia recycling in the pathophysiology of HD but also provides new perspectives for further research, suggesting that regulating ammonia metabolism pathways may offer potential strategies for the prevention and treatment of HD.

Triacylglycerol (53:3) levels were found to be positively associated with HD risk, while Phosphatidylethanolamine (18:1_18:1) levels were negatively correlated with HD risk. Research suggests that Triacylglycerol (53:3) may increase the risk of thrombosis [58], as well as induce inflammatory responses such as leukocyte adhesion and exudation, which in turn may promote the development of varicose veins [59, 60]. When these pathological changes are localized to the anorectal region, they may lead to the formation of HD. Phosphatidylethanolamine (18:1_18:1) is an important phospholipid compound with various physiological functions in the human body, but research on this specific phospholipid remains limited. These findings provide new insights into the role of plasma lipids in the pathogenesis of HD and may offer a scientific basis for developing new therapeutic strategies.

This study reports for the first time the causal relationship between serum metabolites, plasma lipids, and the occurrence of HD. It identifies risk factors for HD and provides new insights into the mechanisms underlying HD. Through MR analysis, we identified a series of metabolites and lipids associated with the risk of HD occurrence, and explored their overall causal relationships with HD in different population samples through meta-analysis. These findings provide a new theoretical foundation for the clinical diagnosis and treatment strategies of HD.

We must acknowledge that, while the results showed consistency across diverse populations, the genetic and environmental backgrounds of these specific populations may influence the generalizability of the findings. Therefore, we conducted external validation in a European ancestry population. We selected ebi-a-GCST90086078 as the GWAS dataset for HD and conducted MR analysis using 1,400 serum metabolites and 179 plasma lipids as outcomes. The ebi-a-GCST90086078 data were derived from 56,637 individuals of European ancestry. SNPs were selected with a genome-wide significance threshold of 1×10⁻⁵. Based on the MR analysis results, in the European ancestry population, the causal relationships between HD and metabolites and lipids are illustrated in Figure 5. By intersecting the MR analysis results with those of the previous meta-analysis, we observed that Triacylglycerol (53:3) levels (GCST90277402) showed a significant positive association with HD in the results of European ancestry population, in line with the meta-analysis results of African American or Afro-Caribbean, South Asian, and East Asian populations. This indicated the generalizability of this finding. GCST90200518 also demonstrated a causal relationship with HD in both the external validation and meta-analysis; however, the direction of the effect was opposite. For detailed results, please refer to the Raw Data. Nevertheless, we observed that the results in the European ancestry population differ markedly from those in the other three populations, reminding us that the population limitations still exist.

Figure 5. Results of external validation. (A) Forest plot shows the causal associations between HD and serum metabolites in the European ancestry population. (B) Forest plot shows the causal associations between HD and plasma lipids in the European ancestry population. (C) Venn diagram illustrates the overlap between the MR analysis in the European ancestry population and the previous meta-analysis results. (D) Leave-one-out analysis shows the influence of each individual SNP on the overall causal effect between GCST90277402 and HD, used to assess horizontal pleiotropy in the MR analysis. (E) Funnel plot indicates no apparent bias among SNPs in the MR analysis between GCST90277402 and HD. (F) Scatter plot exhibits the beta estimates of the MR results between GCST90277402 and HD using different analytical methods.

Besides, since we only used summary statistics for estimation, we did not evaluate the causal association in different groups, such as different gender or age groups. Another major limitation of this study is the lack of consideration of potential confounding factors, including BMI, smoking status, diet, environmental exposures and so on. These confounders may also contribute to the development of HD and thereby compromise the accuracy and reliability of the analytical results.

However, despite these preliminary results providing valuable clues for the pathophysiological research of HD, there is still a lack of in-depth understanding regarding the precise connection between these metabolites and lipids and HD on a multi-omics level. Therefore, future research needs to adopt multi-omics approaches, such as proteomics, metabolomics, and genomics, to comprehensively explore the interactions and potential molecular mechanisms between these biomarkers and HD. Through this interdisciplinary research approach, we aim to further uncover the complex pathological processes of HD and provide a scientific basis for developing more effective prevention and treatment strategies.

Conclusion

In-depth MR analysis and meta-analysis were conducted in this study to reveal the potential causal relationship between serum metabolites, plasma lipids, and the occurrence of HD. Specific metabolites and lipid markers were found significantly associated with the risk of HD and identified metabolite-enriched pathways, providing new insights into its pathophysiology. Our findings offer new insights into the clinical diagnosis and treatment of HD and provide a reference for future multi-omics studies to uncover deeper molecular mechanisms, laying the foundation for the development of effective prevention and treatment strategies.

Abbreviations

Extracellular matrix: ECM; Eicosapentaenoic acid: EPA; Ge-nome-wide association study: GWAS; Hemorrhoids: HD; Instru-mental variable: IV; Inverse Variance Weighted: IV; Low-density lipoprotein: LDL; Matrix metalloproteinase: MMP; Mendelian randomization: MR; Natural killer: NK; Nitric oxide: NO; Nitric oxide synthase 3: NOS3; Polyunsaturated fatty acid: PUFA; Sin-gle nucleotide polymorphism: SNP; Strengthening the report-ing of observational studies in epidemiology using mendelian randomization: STROBE-MR; Tumor necrosis factor-alpha: TNF-α.

Supplementary Materials

Soure Data for Supplementary Figures and Supplementary Tables.

Declarations

Author contributions

Lu Hu and Danyang Wang made equal contributions to this manuscript. Lu Hu: Writing - original draft, Investigation, Data curation, Project administration. Danyang Wang: Formal analy-sis, Methodology, Supervision, Writing - review & editing. Chang You: Writing - review & editing. Ziyan Zhao: Writing - review & editing. Wanlin Zheng: Writing - review & editing. Luyao Wang: Writing - review & editing. Yutong Li: Visualization. Liangzhe Dai: Visualization. Hongkai Yu: Writing - original draft, Investi-gation, Data curation, Project administration. All authors read and approved the final manuscript.

Acknowledgements

Not Applicable

Funding information

Not Applicable.

Ethics approval and consent to participate

Not Applicable.

Competing Interests

The authors declare that they have no existing or potential commercial or financial relationships that could create a con-flict of interest at the time of conducting this study.

Data availability

The original contributions presented in the study are included in the article, further inquiries can be directed to the corre-sponding authors.

References

[1] Agarwal N, Singh K, Sheikh P, Mittal K, Mathai V, & Kumar A. (2017). Executive Summary - The Association of Colon & Rectal Surgeons of India (ACRSI) Practice Guidelines for the Management of Haemorrhoids-2016. Indian J Surg, 79(1), 58-61.

[2] Wu EB, Sung FC, Lin CL, Wu KL, & Chen KB. (2021). Colorectal Cancer Risk in Patients with Hemorrhoids: A 10-Year Population-Based Retrospective Cohort Study. Int J Environ Res Public Health, 18(16).

[3] Pata F, Sgró A, Ferrara F, Vigorita V, Gallo G, & Pellino G. (2021). Anatomy, Physiology and Pathophysiology of Haemorrhoids. Rev Recent Clin Trials, 16(1), 75-80.

[4] Zheng T, Ellinghaus D, Juzenas S, Cossais F, Burmeister G, Mayr G, et al. (2021). Genome-wide analysis of 944 133 individuals provides insights into the etiology of haemorrhoidal disease. Gut, 70(8), 1538-1549.

[5] Vijayvargiya P, Camilleri M, Carlson P, Lueke A, O'Neill J, Burton D, et al. (2017). Performance characteristics of serum C4 and FGF19 measurements to exclude the diagnosis of bile acid diarrhoea in IBS-diarrhoea and functional diarrhoea. Aliment Pharmacol Ther, 46(6), 581-588.

[6] Zhao R, Zhang Q, Huang T, Tian Y, Guan G, & Lin Y. (2024). Effect of the Combination of Clostridium butyricum and Mycelium of Phellinus igniarius on Intestinal Microbiota and Serum Metabolites in DSS-Induced Colitis. Nutrients, 16(1).

[7] Chen F, Dai X, Zhou CC, Li KX, Zhang YJ, Lou XY, et al. (2022). Integrated analysis of the faecal metagenome and serum metabolome reveals the role of gut microbiome-associated metabolites in the detection of colorectal cancer and adenoma. Gut, 71(7), 1315-1325.

[8] Liu R, Hong J, Xu X, Feng Q, Zhang D, Gu Y, et al. (2017). Gut microbiome and serum metabolome alterations in obesity and after weight-loss intervention. Nat Med, 23(7), 859-868.

[9] Yu M, Shang Y, Han L, & Yu X. (2024). Bowel Habits, Obesity, Intestinal Microbiota and Their Influence on Hemorrhoidal Disease: a Mendelian Randomization Study. Clin Exp Gastroenterol, 17, 157-164.

[10] Han X, & Gross RW. (2022). The foundations and development of lipidomics. J Lipid Res, 63(2), 100164.

[11] Yoon H, Shaw JL, Haigis MC, & Greka A. (2021). Lipid metabolism in sickness and in health: Emerging regulators of lipotoxicity. Mol Cell, 81(18), 3708-3730.

[12] Zeng C, Wen B, Hou G, Lei L, Mei Z, Jia X, et al. (2017). Lipidomics profiling reveals the role of glycerophospholipid metabolism in psoriasis. Gigascience, 6(10), 1-11.

[13] Nowowiejska J, Baran A, & Flisiak I. (2023). Lipid Alterations and Metabolism Disturbances in Selected Inflammatory Skin Diseases. Int J Mol Sci, 24(8).

[14] Szmitko PE, Wang CH, Weisel RD, Jeffries GA, Anderson TJ, & Verma S. (2003). Biomarkers of vascular disease linking inflammation to endothelial activation: Part II. Circulation, 108(17), 2041-2048.

[15] Wang R, Zhang Y, Xu L, Lin Y, Yang X, Bai L, et al. (2016). Protein Inhibitor of Activated STAT3 Suppresses Oxidized LDL-induced Cell Responses during Atherosclerosis in Apolipoprotein E-deficient Mice. Sci Rep, 6, 36790.

[16] Aron-Wisnewsky J, Warmbrunn MV, Nieuwdorp M, & Clément K. (2021). Metabolism and Metabolic Disorders and the Microbiome: The Intestinal Microbiota Associated With Obesity, Lipid Metabolism, and Metabolic Health-Pathophysiology and Therapeutic Strategies. Gastroenterology, 160(2), 573-599.

[17] Kattar SA, Jurjus R, Pinon A, Leger DY, Jurjus A, Boukarim C, et al. (2020). Metformin and Probiotics in the Crosstalk between Colitis-Associated Colorectal Cancer and Diabetes in Mice. Cancers (Basel), 12(7).

[18] Huang J, Gui Y, Qin H, & Xie Y. (2023). Causal association between adiposity and hemorrhoids: a Mendelian randomization study. Front Med (Lausanne), 10, 1229925.

[19] Wang Z, Yang M, Shi R, & Wang J. (2024). Association between low-density lipoprotein and hemorrhoids in east Asian populations: A multivariate Mendelian randomization study. Asian J Surg, 47(10), 4514-4515.

[20] Lawlor DA, Harbord RM, Sterne JA, Timpson N, & Davey Smith G. (2008). Mendelian randomization: using genes as instruments for making causal inferences in epidemiology. Stat Med, 27(8), 1133-1163.

[21] Chen Y, Lu T, Pettersson-Kymmer U, Stewart ID, Butler-Laporte G, Nakanishi T, et al. (2023). Genomic atlas of the plasma metabolome prioritizes metabolites implicated in human diseases. Nat Genet, 55(1), 44-53.

[22] Ottensmann L, Tabassum R, Ruotsalainen SE, Gerl MJ, Klose C, Widén E, et al. (2023). Genome-wide association analysis of plasma lipidome identifies 495 genetic associations. Nat Commun, 14(1), 6934.

[24] Bowden J, Davey Smith G, & Burgess S. (2015). Mendelian randomization with invalid instruments: effect estimation and bias detection through Egger regression. Int J Epidemiol, 44(2), 512-525.

[26] Hemani G, Bowden J, & Davey Smith G. (2018). Evaluating the potential role of pleiotropy in Mendelian randomization studies. Hum Mol Genet, 27(R2), R195-r208.

[28] Li Y, Nicholson RJ, & Summers SA. (2022). Ceramide signaling in the gut. Mol Cell Endocrinol, 544, 111554.

[29] Yu S, Sun Y, Shao X, Zhou Y, Yu Y, Kuai X, et al. (2022). Leaky Gut in IBD: Intestinal Barrier-Gut Microbiota Interaction. J Microbiol Biotechnol, 32(7), 825-834.

[30] Jain AP, Aggarwal KK, & Zhang PY. (2015). Omega-3 fatty acids and cardiovascular disease. Eur Rev Med Pharmacol Sci, 19(3), 441-445.

[31] D'Souza BN, Yadav M, Chaudhary PP, Ratley G, Lu MY, Alves DA, et al. (2024). Derivation of novel metabolic pathway score identifies alanine metabolism as a targetable influencer of TNF-alpha signaling. Heliyon, 10(13), e33502.

[32] Chen X, Xun K, Chen L, & Wang Y. (2009). TNF-alpha, a potent lipid metabolism regulator. Cell Biochem Funct, 27(7), 407-416.

[33] Tiwary SK, Kumar A, Mishra SP, Kumar P, & Khanna AK. (2020). Study of association of varicose veins and inflammation by inflammatory markers. Phlebology, 35(9), 679-685.

[34] Jacob MP. (2003). Extracellular matrix remodeling and matrix metalloproteinases in the vascular wall during aging and in pathological conditions. Biomed Pharmacother, 57(5-6), 195-202.

[35] Mannello F, & Raffetto JD. (2011). Matrix metalloproteinase activity and glycosaminoglycans in chronic venous disease: the linkage among cell biology, pathology and translational research. Am J Transl Res, 3(2), 149-158.

[36] Raffetto JD, & Khalil RA. (2021). Mechanisms of Lower Extremity Vein Dysfunction in Chronic Venous Disease and Implications in Management of Varicose Veins. Vessel Plus, 5.

[37] Kucukguven A, & Khalil RA. (2013). Matrix metalloproteinases as potential targets in the venous dilation associated with varicose veins. Curr Drug Targets, 14(3), 287-324.

[38] MacColl E, & Khalil RA. (2015). Matrix Metalloproteinases as Regulators of Vein Structure and Function: Implications in Chronic Venous Disease. J Pharmacol Exp Ther, 355(3), 410-428.

[39] Bierhansl L, Conradi LC, Treps L, Dewerchin M, & Carmeliet P. (2017). Central Role of Metabolism in Endothelial Cell Function and Vascular Disease. Physiology (Bethesda), 32(2), 126-140.

[40] Sabbatinelli J, Prattichizzo F, Olivieri F, Procopio AD, Rippo MR, & Giuliani A. (2019). Where Metabolism Meets Senescence: Focus on Endothelial Cells. Front Physiol, 10, 1523.

[41] Sharp CD, Houghton J, Elrod JW, Warren A, Jackson THt, Jawahar A, et al. (2005). N-methyl-D-aspartate receptor activation in human cerebral endothelium promotes intracellular oxidant stress. Am J Physiol Heart Circ Physiol, 288(4), H1893-1899.

[42] Tao Q, Ma N, Fan L, Ge W, Zhang Z, Liu X, et al. (2024). Multi-Omics Approaches for Liver Reveal the Thromboprophylaxis Mechanism of Aspirin Eugenol Ester in Rat Thrombosis Model. Int J Mol Sci, 25(4).

[43] Chatterjee M, Ehrenberg A, Toska LM, Metz LM, Klier M, Krueger I, et al. (2020). Molecular Drivers of Platelet Activation: Unraveling Novel Targets for Anti-Thrombotic and Anti-Thrombo-Inflammatory Therapy. Int J Mol Sci, 21(21).

[44] Lee NT, Ong LK, Gyawali P, Nassir C, Mustapha M, Nandurkar HH, et al. (2021). Role of Purinergic Signalling in Endothelial Dysfunction and Thrombo-Inflammation in Ischaemic Stroke and Cerebral Small Vessel Disease. Biomolecules, 11(7).

[45] Li Y, Li R, Feng Z, Wan Q, & Wu J. (2020). Linagliptin Regulates the Mitochondrial Respiratory Reserve to Alter Platelet Activation and Arterial Thrombosis. Front Pharmacol, 11, 585612.

[46] Grisham MB, Jourd'Heuil D, & Wink DA. (1999). Nitric oxide. I. Physiological chemistry of nitric oxide and its metabolites:implications in inflammation. Am J Physiol, 276(2), G315-321.

[47] Phang JM, Liu W, & Zabirnyk O. (2010). Proline metabolism and microenvironmental stress. Annu Rev Nutr, 30, 441-463.

[48] Ying W, & Xiong ZG. (2010). Oxidative stress and NAD+ in ischemic brain injury: current advances and future perspectives. Curr Med Chem, 17(20), 2152-2158.

[49] Chen H, Wang C, Wei X, Ding X, & Ying W. (2015). Malate-Aspartate Shuttle Inhibitor Aminooxyacetate Acid Induces Apoptosis and Impairs Energy Metabolism of Both Resting Microglia and LPS-Activated Microglia. Neurochem Res, 40(6), 1311-1318.

[50] Walker V. (2009). Ammonia toxicity and its prevention in inherited defects of the urea cycle. Diabetes Obes Metab, 11(9), 823-835.

[51] Matsumoto S, Häberle J, Kido J, Mitsubuchi H, Endo F, & Nakamura K. (2019). Urea cycle disorders-update. J Hum Genet, 64(9), 833-847.

[52] Abdoun K, Stumpff F, & Martens H. (2006). Ammonia and urea transport across the rumen epithelium: a review. Anim Health Res Rev, 7(1-2), 43-59.

[53] Walker V. (2014). Ammonia metabolism and hyperammonemic disorders. Adv Clin Chem, 67, 73-150.

[54] Ferrario M, Pastorelli R, Brunelli L, Liu S, Zanella do Amaral Campos PP, Casoni D, et al. (2021). Persistent hyperammonia and altered concentrations of urea cycle metabolites in a 5-day swine experiment of sepsis. Sci Rep, 11(1), 18430.

[55] Aldridge DR, Tranah EJ, & Shawcross DL. (2015). Pathogenesis of hepatic encephalopathy: role of ammonia and systemic inflammation. J Clin Exp Hepatol, 5(Suppl 1), S7-s20.

[56] Fu A, Alvarez-Perez JC, Avizonis D, Kin T, Ficarro SB, Choi DW, et al. (2020). Glucose-dependent partitioning of arginine to the urea cycle protects β-cells from inflammation. Nat Metab, 2(5), 432-446.

[57] Rojas Á, García-Lozano MR, Gil-Gómez A, Romero-Gómez M, & Ampuero J. (2022). Glutaminolysis-ammonia-urea Cycle Axis, Non-alcoholic Fatty Liver Disease Progression and Development of Novel Therapies. J Clin Transl Hepatol, 10(2), 356-362.

[58] Vayá A, Falcó C, Simó M, Ferrando F, Mira Y, Todolí J, et al. (2007). Influence of lipids and obesity on haemorheological parameters in patients with deep vein thrombosis. Thromb Haemost, 98(3), 621-626.

[59] Saghazadeh A, Hafizi S, & Rezaei N. (2015). Inflammation in venous thromboembolism: Cause or consequence? Int Immunopharmacol, 28(1), 655-665.

[60] Poredos P, Spirkoska A, Rucigaj T, Fareed J, & Jezovnik MK. (2015). Do blood constituents in varicose veins differ from the systemic blood constituents? Eur J Vasc Endovasc Surg, 50(2), 250-256.

Figures