Exploring the causal effect of genetically predicted blood and cerebrospinal fluid metabolites and metabolic pathways on intracranial aneurysm

Abstract

Objective

Previous studies have identified significant alterations in metabolites in patients with intracranial aneurysm. This study aimed to evaluate the causal relationships of metabolites in both blood and cerebrospinal fluid with intracranial aneurysm at the genetic level using bidirectional Mendelian randomization analysis.

Methods

Genetic instrumental variables for 1400 blood metabolites were obtained from a genome-wide association study analysis involving 8299 individuals, while 338 cerebrospinal fluid metabolites were sourced from another genome-wide association study involving 291 individuals. Outcome data for intracranial aneurysm were retrieved from the International Stroke Genetics Consortium, and validation data were retrieved from the FinnGen study. The primary analysis employed the inverse-variance weighted method, supported by sensitivity analysis to address pleiotropy and enhance robustness. Replication analysis and meta-analysis were performed to enhance the robustness of the findings. Colocalization analysis was used to assess the potential shared genetic architecture between metabolites and intracranial aneurysm, and metabolic pathway analysis was performed using MetaboAnalyst 6.0.

Results

Following false discovery rate correction, four and five genetically predicted blood metabolites showed inversely causal associations with intracranial aneurysms and subarachnoid hemorrhage, respectively, in the discovery group, while no cerebrospinal fluid metabolites showed significant causal associations with three intracranial aneurysm phenotypes. After performing a meta-analysis incorporating the results from the replication data, we identified 1-arachidonoyl-GPC (20:4n6) (odds ratio: 0.90) and 1-palmitoyl-2-arachidonoyl-GPC (16:0/20:4n6) (odds ratio: 0.92) as potential protective factors against intracranial aneurysms as well as (N(1) + N(8))-acetylspermidine (odds ratio: 0.83) as a potential protective factor against subarachnoid hemorrhage.

Conclusions

Our findings establish a potential causal link between blood/cerebrospinal fluid metabolites and intracranial aneurysm. Furthermore, the identified metabolites and pathways provide valuable insights into the pathogenic mechanisms and progression of intracranial aneurysm.

Keywords

Metabolites genetics causality intracranial aneurysm Mendelian randomization

Introduction

Intracranial aneurysm (IA) is a severe, life-threatening cerebrovascular disorder characterized by the abnormal bulging or ballooning of cerebral arteries. An epidemiological study has estimated the prevalence of unruptured intracranial aneurysm (uIA) in the general population at approximately 3%–5%, with a relatively low rupture incidence.¹ Ruptured IA, which is the main cause of subarachnoid hemorrhage (SAH), presents a significant clinical burden, with a case-fatality rate approaching 50%. Despite technological advancements in microsurgical clipping and endovascular treatment, the outcome following rupture remains poor.² The established risk factors for IA formation include genetic predisposition, hypertension, and smoking. Another study has shown that chronic inflammation and vascular wall degeneration play crucial roles in IA formation and progression.³ However, the molecular mechanism, specific metabolite alterations, and metabolic pathways driving the formation and development of IA are not fully understood. This gap limits early detection, risk stratification, and the development of targeted treatments.

In recent years, the role of metabolic alterations in IA development and progression has garnered increasing attention. Metabolites, which are small molecules produced as intermediates or end products of cellular metabolic processes, are recognized as key contributors to the pathogenesis of cardio- and cerebrovascular disorders.⁴ Metabolomics, the comprehensive analysis of metabolites in biological fluids such as blood and cerebrospinal fluid (CSF), has emerged as a promising approach for identifying disease biomarkers and elucidating underlying biochemical mechanisms.⁵ Descriptive studies have shown significant alterations in blood metabolic profiles among IA patients. For example, a recent study identified that 19 metabolites were significantly altered between healthy individuals and uIA patients.⁶ Another study that performed untargeted metabolomics analysis reported a correlation between aneurysm development and local amino acid metabolism.⁷ However, due to the challenges involved in obtaining CSF samples from patients with uIA, studies on CSF metabolites in this population are rare. Notably, Kamińska et al. demonstrated that the interleukin 6 (IL-6) quotient is significantly elevated in uIA patients compared with that in controls, with the CSF IL-6 levels significantly surpassing the serum levels.⁸ Another study has reported significantly increased levels of both pro- and anti-inflammatory cytokines in the CSF of uIA patients compared with those in controls.⁹ Given that CSF metabolomics can provide a direct assessment of the microenvironment surrounding the aneurysm, both descriptive and causal studies of blood and CSF metabolites in IA patients are essential.

Mendelian randomization (MR) has emerged as a robust tool to address the limitations of observational research by using single nucleotide polymorphisms (SNPs) as instrumental variables (IVs) to infer causal relationships between exposures and outcomes.¹⁰ Since genetic variants are randomly assigned at conception and remain stable throughout life, MR is less susceptible to confounding and reverse causation than traditional observational studies.¹¹ Although randomized controlled trials (RCTs) can yield reliable results, examining the association between metabolites and IA remains challenging due to the high cost, ethical constraints, and difficulty in obtaining adequate CSF samples.¹² In this study, we employed a bidirectional MR approach using large-scale genome-wide association study (GWAS) data to assess causal relationships of blood and CSF metabolites with IA. Additionally, we conducted a metabolic pathway analysis to identify potential pathways and elucidate the biological functions of specific metabolites.

Methods

Study design

The flow diagram of this study is illustrated in Figure 1, which utilizes SNPs as genetic instruments. Three fundamental assumptions are critical for a valid MR study:¹⁰ (a) the relevance assumption: IVs must be strongly associated with the exposure; (b) the independence assumption: IVs must be independent of any confounders affecting the exposure–outcome relationship; and (c) the exclusion restriction assumption: IVs must influence the outcome solely through the exposure.

Figure 1.

Overall design of the bidirectional two-sample MR study. MR: Mendelian randomization.

The sources of GWAS data analyzed in the present study are summarized in Table 1. Human blood metabolites, including 1091 metabolites and 309 metabolite ratios, were obtained from 8299 individuals in the Canadian Longitudinal Study on Aging (CLSA) cohort. We sourced the GWAS summary statistics from the GWAS catalog: GCST90199621 - GCST90201020. Among the 1091 plasma metabolites, 850 had known identities, spanning eight super pathways: lipid, amino acid, xenobiotics, nucleotide, cofactor and vitamins, carbohydrate, peptide, and energy. The remaining 241 were classified as unknown or “partially” characterized molecules.¹³ GWAS data of CSF metabolites were derived from a subset of participants of the Wisconsin Alzheimer’s Disease Research Center (WADRC) and Wisconsin Registry for Alzheimer’s Prevention (WRAP) studies, including 291 unrelated European-ancestry individuals; finally, 338 CSF metabolites were identified. Among these, 39 were categorized as unknown molecules.¹⁴ Detailed information on blood and CSF metabolites is presented in Supplementary Table 1 and Supplementary Table 2.

Table 1.

Characteristics of GWAS data.

Variable	First author (year)	GWAS ID	Sample size	Ancestry	Article DOI/URL
Exposure
blood metabolites¹³	Chen (2023)	GCST90199621-GCST90201020	8299 cases	Canadian	doi:10.1038/s41588-022-01270-1
CSF metabolites¹⁴	Panyard (2021)	GCST90025999-GCST90026336	291 cases	European	doi.org/10.1038/s42003-020-01583-z
Outcome
uIA¹⁵	Bakker (2020)	NA	2070 cases/71952 controls	European	https://doi.org/10.6084/m9.figshare.11303372
SAH¹⁵	Bakker (2020)	NA	5140 cases/71952 controls	European	https://doi.org/10.6084/m9.figshare.11303372
IAs¹⁵	Bakker (2020)	NA	7495 cases/71934 controls	European	https://doi.org/10.6084/m9.figshare.11303372
Replication
uIA	FinnGen	finngen_R11_I9_ANEURYSM	3014 cases/412105 controls	European	https://storage.googleapis.com/finngen-public-data-r11/summary_stats/finngen_R11_I9_ANEURYSM.gz
SAH	FinnGen	finngen_R11_I9_SAH	3814 cases/408928 controls	European	https://storage.googleapis.com/finngen-public-data-r11/summary_stats/finngen_R11_I9_SAH.gz
IAs	FinnGen	finngen_R11_I9_SAHANEUR	6236 cases/408928 controls	European	https://storage.googleapis.com/finngen-public-data-r11/summary_stats/finngen_R11_I9_SAHANEUR.gz

IAs: intracranial aneurysms; uIA: unruptured intracranial aneurysm; SAH: subarachnoid hemorrhage; CSF: cerebrospinal fluid.

The discovery data were derived from a publicly available GWAS summary database created by the International Stroke Genetics Consortium (ISGC) Intracranial Aneurysm Working Group, which includes 10,754 cases and 306,882 controls of both European and East Asian ancestry. For this study, we used the GWAS summary data specific to individuals of European ancestry and analyzed three IA phenotypes as the outcomes, the IAs group consisting of 7495 ruptured and unruptured IA patients and 71934 controls; SAH group consisting of 5140 ruptured IA patients and 71952 controls; and uIA group, only comprising 2070 uIA patients and 71952 controls.¹⁵ The replication data were derived from the FinnGen study, a large public-private partnership research project aimed at collecting samples and health data from 500,000 Finnish participants to provide novel insights into disease genetics.¹⁶ The IAs group contained 6236 IA patients and 408928 controls, the SAH group contained 3814 SAH patients and 408928 controls, and the uIA group contained 3014 uIA patients and 412105 controls. There was no overlap between the exposure and outcome data. The characteristics of the study populations are presented in Supplementary Table 1. All GWAS databases were publicly available and had been approved by the corresponding ethical review board in the original GWAS.

IVs selection

To satisfy the first assumption, IV-associated blood and CSF metabolites were identified using strict screening conditions. All SNPs that had strong predictive metabolite traits at a threshold of p < 5 × 10⁻⁸ were selected; since the majority of blood and CSF metabolites either lacked or had fewer than three linked SNPs (as shown in Supplementary Table 1 and Supplementary Table 2), a more relaxed significance threshold of p < 1 × 10⁻⁵ was employed to obtain SNPs associated with blood and CSF metabolites. Linkage disequilibrium (r² < 0.001, within 10,000 kb) was then applied to assess the independence of the candidate IVs, excluding dependent genetic IVs.¹⁷ This criterion has been applied in previous studies.^18,19

Subsequently, we extracted exposure-associated SNPs from the outcome and discarded SNPs associated with the outcome (p < 5 × 10⁻⁵) to satisfy the third assumption.²⁰ We calculated the R² and F-statistics for each SNP and removed those with F-statistics <10.²¹ Then, we harmonized SNPs for exposure and outcome, and palindromic effects and allelic inconsistent SNPs were removed. To minimize potential violations of the second assumption, we queried the GWAS catalog (https://www.ebi.ac.uk/gwas/) to exclude pleiotropic SNPs that were directly related to potential confounding factors (smoking and hypertension).

MR analysis and sensitivity analysis

All analyses were performed using the TwoSampleMR (version 0.6.8) and MRPRESSO package (1.0) in R Software 4.3.2 (https://www.R-project.org). The inverse-variance weighted (IVW) method was employed (p < 0.05) as our primary approach, integrating the β values and the standard errors (SEs). The IVW method was chosen because it provides the most precise estimate of causal effects when all genetic variants are valid instruments.²² To ensure the robustness of the findings, several additional methods were used. MR–Egger regression offered an accurate estimate of the causal effect, particularly in the presence of pleiotropy.²³ The weighted median remained consistent even if 50% of the SNPs were invalid.²⁴ The weighted mode, which relaxed the IV assumptions, exhibited less bias and fewer type-I errors in comparison with other methods.²⁵

Sensitivity analysis was essential for evaluating horizontal pleiotropy and heterogeneity, both of which significantly affect MR results. In this investigation, we employed four methods to identify and address heterogeneity and pleiotropy: Cochran’s Q test, MR–Pleiotropy RESidual Sum and Outlier test (PRESSO), radial MR, and MR–Egger regression intercept test. Heterogeneity was considered when the p-value in Cochran’s Q test was <0.05.²⁶ MR–PRESSO and radial MR were used to detect and remove SNP outliers, providing an outlier-corrected estimate.²⁷ MR–Egger regression was applied to check for horizontal pleiotropy, with p < 0.05 indicating its presence and verifying consistency with results obtained using other methods.²⁸ The MR–Steiger test was conducted to assess whether the SNPs were more strongly associated with the exposure than the outcome, thereby improving the robustness of the IVs.²⁹ Finally, a leave-one-out (LOO) analysis was used, systematically removing SNPs one at a time to evaluate whether the results were influenced by any single SNP.³⁰

False discovery rate (FDR) correction was applied to account for multiple comparisons, with a p-value threshold of <0.05 defining strong evidence. Data with p-values <0.05 but FDR-adjusted p-values >0.05 were considered indicative of potential causal relationships.³¹

Replication analysis and meta-analysis

To enhance the robustness of the findings, we replicated the MR analysis with another GWAS dataset for three IA phenotypes. We finally identified the metabolites with causal effects on IA from the results of the meta-analysis of two MR analyses. The meta-analysis was conducted based on a random-effects IVW model.³²

Bayesian colocalization analysis

To investigate the potential shared genetic architecture between metabolites and IAs, we performed a colocalization analysis to identify loci with evidence of shared causal variants. The analysis was conducted using the Coloc package in R Studio, a Bayesian framework that assesses the probability of a single variant being responsible for associations in both traits. We evaluated five hypotheses for each locus as follows: (a) H0: no association with either trait; (b) H1: an association with Trait A only; (c) H2: an association with Trait B only; (d) H3: an association with both traits but due to distinct causal variants; and (e) H4: associations with both traits due to a shared causal variant. The Bayesian posterior probability (PP) for each hypothesis was computed. A PP >0.8 for H4 was set as the threshold indicating strong evidence for colocalization.³³

Metabolic pathway analysis

To explore the potential metabolite groups or pathways associated with the biological processes of IA, metabolic pathway enrichment analysis was conducted using MetaboAnalyst 6.0. The Kyoto Encyclopedia of Genes and Genomes (KEGG) database was employed for this analysis, and the level of significance for pathway analysis was set at 0.05.³⁴ Furthermore, this study analyzed only the metabolites that passed the recommended association threshold (p < 0.05) determined using the IVW method; additionally, the metabolites were divided into blood and CSF groups for separate analysis.

Results

Causal effects of blood metabolites on IA

We identified 13,476; 13,429; and 13,406 independent SNPs associated with IAs, SAH, and uIA, respectively, via a meticulous quality control process. Specific details regarding all SNPs utilized as genetic IVs are provided in Supplementary Table 1. The F-statistics computed for all IVs were >10, ranging from 19.5 to 5309.7, signifying the absence of weak IVs chosen from the exposure. All independent SNPs were checked in the GWAS catalog; 54, 52, and 54 SNPs were directly related to smoking and hypertension in the three groups (Supplementary Table 1).

After removing these SNPs, IVW analysis was conducted for the initial determination of 73 blood metabolites potentially exhibiting causal effects on IAs. The Cochran’s Q test indicated the existence of heterogeneity when 1,2-dilinoleoyl-GPC (18:2/18:2), arachidonate (20:4n6), and the glycerol-to-sulfate ratio were used as exposures. No outlier was detected when we performed the MR–PRESSO test. MR–Egger intercept test yielded compelling evidence, indicating the absence of horizontal pleiotropy; Steiger’s test showed that there was no reverse causal effect of IVs on the outcome. LOO analysis supported the notion that individual SNPs do not introduce bias in MR calculations (Supplementary Figure 1). Finally, following FDR correction, 1-arachidonoyl-GPC (20:4n6) levels (odds ratio (OR): 0.88, 95% confidence interval (CI): 0.81–0.93, p = 2.07 × 10⁻⁵, FDR p:0.01), 1-linoleoyl-2-arachidonoyl-GPC (18:2/20:4n6) (OR: 0.79, 95% CI: 0.71–0.88, p = 2.84 × 10⁻⁵, FDR p:0.01), 1-palmitoyl-2-arachidonoyl-gpc (16:0/20:4n6) (OR: 0.88, 95% CI: 0.82–0.94, p =7.93 × 10⁻⁵, FDR p:0.03) and (N(1) +N(8))-acetylspermidine (OR: 0.78, 95% CI: 0.69–0.88, p = 2.84 × 10⁻⁵, FDR p:0.01), showed a significant causal relationship with IAs (Figures 2 and 4); 66 blood metabolites showed potential causal associations with IAs.

Figure 2.

Forest plot of the associations between blood metabolites and IAs. IAs: intracranial aneurysms.

IVW analysis was conducted for the initial identification of 58 blood metabolites potentially exhibiting causal effects on SAH. The Cochran’s Q test (p = 0.003) indicated the existence of heterogeneity when 3-methylglutarylcarnitine was used as the exposure. The MR–Egger intercept test (p = 0.04) indicated horizontal pleiotropy when retinol (vitamin A) and linoleoyl-arachidonoyl-glycerol (18:2 to 20:4) were used as the exposures. Steiger’s test showed that there was no reverse causal effect of IVs on the outcome. LOO analysis supported the notion that individual SNPs did not introduce bias in MR calculations (Supplementary Figure 1). Finally, following FDR correction, 1-linoleoyl-2-arachidonoyl-GPC (18:2/20:4n6) (OR: 0.74, 95% CI: 0.64–0.85, p = 4.6 × 10⁻⁵, FDR p:0.02), cholesterol-to-linoleoyl-arachidonoyl-glycerol ratio (18:2 to 20:4) (OR: 1.32, 95% CI: 1.16–1.5, p = 1.57 × 10⁻⁵, FDR p:0.01), benzoate-to-linoleoyl-arachidonoyl-glycerol ratio (18:2 to 20:4) (OR: 1.59, 95% CI: 1.27–2.0, p = 1.18 × 10⁻⁵, FDR p:0.01), X-11478 (OR: 1.32, 95% CI: 1.17–1.5, p = 6.21 × 10⁻⁵, FDR p:0.02), and (N(1) + N(8))-acetylspermidine (OR: 0.75, 95% CI: 0.65–0.85, p = 2.45 × 10⁻⁵, FDR p:0.01) showed a significant causal relationship with SAH; 51 blood metabolites showed suggestive causal associations with SAH (Figures 3 and 4).

Figure 3.

Forest plot of the associations between blood metabolites and SAH. SAH: subarachnoid haemorrhage.

Figure 4.

Scatter plot of the association between blood metabolites and IA. (a) four blood metabolites and IAs. (b) Five blood metabolites and SAH.

Subsequently, IVW analysis was conducted for the initial determination of 34 blood metabolites potentially exhibiting causal effects on uIA. One outlier (rs272857) was detected using the MR–PRESSO test when N, N, N-trimethyl-5-aminovalerate was exposed; the Cochran’s Q test and MR–Egger intercept test yielded compelling evidence indicating the absence of heterogeneity and horizontal pleiotropy. The Steiger test suggested that the correlation between all SNPs and the exposure is greater than that with the outcome. LOO analysis supported that individual SNPs did not introduce bias in MR calculations (Supplementary Figure 1). Finally, following the FDR correction, 33 blood metabolites showed suggestive causal associations with uIA (Figure 5). Details of the analysis data for all three groups are presented in Supplementary Table 1.

Figure 5.

Forest plot of the associations between blood metabolites and uIA. uIA: unruptured intracranial aneurysm.

Causal effects of CSF metabolites on IA

We identified 6992, 6977, and 6850 independent SNPs associated with IAs, SAH, and uIA via a meticulous quality control process, respectively. Specific details regarding all SNPs utilized as genetic IVs can be accessed in Supplementary Table 2. The F-statistics computed for all IVs were >10, ranging from 19.3 to 904.8, signifying the absence of weak IVs chosen from the exposure in three groups. All independent SNPs were checked in the GWAS catalog; 11, 11, and 11 SNPs were directly related to smoking and hypertension in three groups, respectively (Supplementary Table 2).

Subsequently, IVW analysis was conducted for the initial determination of 10, 9, and 14 CSF metabolites potentially exhibiting causal effects on IAs, SAH, and uIA, respectively. No outlier was detected using the MR–PRESSO test. The Cochran’s Q test and MR–Egger intercept test yielded compelling evidence indicating the absence of heterogeneity and horizontal pleiotropy. The Steiger test suggested that the correlation between all SNPs and exposure was greater than that with the outcome; LOO analysis supported the notion that individual SNPs did not introduce bias in MR calculations (Supplementary Figure 2). Finally, following FDR correction, 10, 9, and 14 CSF metabolites indicated causal associations with IAs, SAH, and uIA, respectively. The ORs, 95% CIs, and p-values for the IVW analysis are presented in Figure 6. Details of the analysis data for all three groups are presented in Supplementary Table 2.

Figure 6.

Forest plot of the associations between CSF metabolites and IA. (a) IAs group. (b) SAH group. (c) uIA group. CSF: cerebrospinal fluid; IA: intracranial aneurysm; uIA: unruptured intracranial aneurysm.

Reverse causation between significant blood metabolites and IA

The causal effects of IAs and SAH on the identified significant blood metabolites were further assessed using MR analysis. In the IVW model, with IAs as the exposure, no causal effect was detected for the four blood metabolites at an IV significance threshold of 1 × 10⁻⁵. Similarly, with SAH as the exposure, no causal effect was observed for the five blood metabolites at an IV significance threshold of 1 × 10⁻⁵. Detailed analysis are shown in Supplementary Table 3.

Causal effects of significant blood metabolites on IAs and SAH in replicate MR analysis

Following MR analysis and sensitivity analysis, genetically predicted 1-arachidonoyl-GPC (20:4n6) levels (OR: 0.92, 95% CI: 0.87–0.98, p = 8.67 × 10⁻³), 1-linoleoyl-2-arachidonoyl-GPC (18:2/20:4n6) (OR: 0.93, 95% CI: 0.867–0.997, p = 0.04), and 1-palmitoyl-2-arachidonoyl-gpc (16:0/20:4n6) (OR: 0.95, 95% CI: 0.906–0.996, p = 0.032) showed a causal relationship with IAs; genetically predicted (N(1) + N(8))-acetylspermidine (OR: 0.92, 95% CI: 0.85–0.99, p = 0.029) showed a causal relationship with SAH. Details of the analysis data are presented in Supplementary Table 4.

Results of meta-analysis

Subsequently, we conducted a meta-analysis to combine the MR results obtained using the two different databases. The summarized results of the meta-analyses in Figure 7 reveal that an increase in 1-arachidonoyl-GPC (20:4n6) (OR = 0.9, 95% CI: 0.84–0.95, p < 0.01) and 1-palmitoyl-2-arachidonoyl-gpc (16:0/20:4n6) (OR = 1.92, 95% CI: 0.85–0.92, p = 0.02) levels led to a low risk of IAs without any heterogeneity observed in the random-effects model. Notably, although causal relationships between 1-linoleoyl-2-arachidonoyl-GPC (18:2/20:4n6) and IAs as well as between (N(1) + N(8))-acetylspermidine and SAH, were observed in both MR and replicate MR analyses, the p-value in the meta-analysis exceeded 0.05. The results of meta-analyses of all significant blood metabolites are presented in Figure 7.

Figure 7.

Forest plots for meta-analysis of MR estimates in discovery and replication databases. (a) IAs group. (b) SAH group. MR: Mendelian randomization; IAs: intracranial aneurysms; SAH: subarachnoid haemorrhage.

Colocalization

Notably, 1-arachidonoyl-GPC (20:4n6) (PP.H4 = 0.82) and 1-palmitoyl-2-arachidonoyl-GPC (16:0/20:4n6) (PP.H4 = 0.80) exhibited strong colocalization evidence with IAs at SNP rs102274 in discovery date. In contrast, 1-linoleoyl-2-arachidonoyl-GPC (18:2/20:4n6) (PP.H4 = 0.72) and (N(1) + N(8))-acetylspermidine (PP.H4 = 0.33), displayed moderate but inconclusive evidence of colocalization with IAs and SAH (Figure 8 and Supplementary Table 5).

Figure 8.

Colocalization analysis assessing potential shared genetic architecture between blood metabolites and IAs. (a) 1-arachidonoyl-GPC (20:4n6). (b) 1-palmitoyl-2-arachidonoyl-gpc (16:0/20:4n6). IAs: Intracranial aneurysms.

Metabolic pathways

As shown in Supplementary Table 6 and Figure 9, five metabolic pathways associated with blood metabolites were identified that participated in the pathogenesis of IAs. The most significant pathway was arginine biosynthesis (p = 0.0032), which involves two metabolites, N-acetylornithine and urea. Eight metabolic pathways associated with CSF metabolites were identified that participated in the pathogenesis of IAs. The most significant pathway was glyoxylate and dicarboxylate metabolism (p = 7.42 × 10⁻⁵), which involved three metabolites, glycine, isocitric acid, and glutamine.

Figure 9.

Pathway analysis for metabolites using MetaboAnalyst 6.0. (a) Blood metabolites pathway analysis. (b) CSF metabolites pathway analysis. CSF: cerebrospinal fluid.

Discussion

The present findings advance our understanding of the metabolites and metabolic processes involved in IA pathogenesis. Specifically, 48 blood metabolites, 22 blood metabolite ratios, and 10 CSF metabolites were identified to have potential causal associations with IAs. Forty-three blood metabolites, 13 blood metabolite ratios, and 9 CSF metabolites were identified to have potential causal associations with SAH. Furthermore, 23 blood metabolites, 10 blood metabolite ratios, and 14 CSF metabolites demonstrated potential causal links with uIA. After FDR correction and meta-analysis, our results revealed that three blood metabolites, 1-arachidonoyl-GPC (20:4n6), 1-linoleoyl-2-arachidonoyl-GPC (18:2/20:4n6), and 1-palmitoyl-2-arachidonoyl-gpc (16:0/20:4n6) served as protective factors against IAs; (N(1) + N(8))-acetylspermidine served as a protective factor against SAH; and no blood metabolites exhibited a significant causal association with uIA. In addition, no CSF metabolites exhibited a significant causal association with three IA phenotypes.

The results from the two MR analyses suggested a causal relationship between (N(1) + N(8))-acetylspermidine and SAH, and our KEGG pathway analysis of metabolites in both blood and CSF suggested that the arginine synthesis pathway played a role in IA pathogenesis. It is well-established that arginine is converted into ornithine by the action of arginine decarboxylase, with ornithine acting as the direct precursor for spermidine synthesis. (N(1) + N(8))-acetylspermidine, an acetylated derivative of spermidine, demonstrates high blood stability and functions as spermidine’s storage and transport form.¹³ As a vital polyamine, spermidine plays a central role in various physiological processes, including cell growth, proliferation, gene expression regulation, protein synthesis, and autophagy.³⁵ Previous studies have suggested a protective role of spermidine against vascular disorders. For example, high dietary intake of spermidine has been inversely associated with cardiovascular disease and hypertension in a population-based prospective cohort.³⁶ Spermidine may also reverse age-related arterial stiffness by increasing nitric oxide bioavailability, reducing oxidative damage to endothelial cells, and enhancing autophagy in aging mouse models.³⁷ In addition, a single-cell study on IA first clearly identified that vascular endothelial cells are the core target cells of DNA methylation regulation during IA pathogenesis. Among them, DNA methylation–related genes (MRGs) such as DNMT1, DNMT3A, and DNMT3B have established causal associations with the occurrence and progression of aneurysms.³⁸ Another study has confirmed that exogenous supplementation of spermine does not alter the protein expression levels of the aforementioned DNMTs; however, it can selectively activate DNMT3A and exert a significant regulatory effect on the activity of DNMT3B.³⁹ Considering these in combination with the series of findings from our study, we hypothesize that spermidine may participate in the pathological process of IA and exert a protective effect via a multi-dimensional synergistic pathway involving metabolic regulation, alleviation of vascular oxidative stress, and epigenetic modification. However, this hypothesis warrants further verification and refinement through additional basic experimental and clinical research.

Notably, the random-effects model meta-analysis yielded a non-significant p-value (0.08), likely attributable to the high observed heterogeneity $I^{2} = 85%$ . Two key factors may underpin this heterogeneity. Th first involves clinical phenotyping discrepancies—the ISGC study defined SAH as aneurysm-related, while the FinnGen cohort used the broader International Classification of Diseases, 10th Revision (ICD-10) code I60 (potentially including nonaneurysmal SAH) and thus introduced diagnostic misclassification. The second factor is the marked population heterogeneity, with the multi-country European ISGC cohort and the exclusive Finnish FinnGen cohort representing distinct genetic and environmental backgrounds. Unfortunately, reliance on summary-level data precluded stratified analyses to statistically adjust for these variations. Nevertheless, the OR remained below 1, implying a potential protective role of this metabolite in aneurysm rupture. Inclusion of additional studies in future meta-analyses may mitigate heterogeneity and enhance statistical power.

It is well-established that disorders of lipid metabolism, characterized by lipid accumulation and peroxidation in the vascular wall, contribute to the formation and rupture of IA.⁴⁰ In this study, we identified two blood metabolites, 1-arachidonoyl-GPC (20:4n6) and 1-palmitoyl-2-arachidonoyl-GPC (16:0/20:4n6), belonging to the lysophosphatidylcholine (LPC) and phosphatidylcholine (PC) classes, respectively that exhibit protective associations with IA. Under physiological conditions, the metabolic balance between PC and LPC is dynamically maintained through the Lands’ cycle.⁴¹ LPC is traditionally recognized as a proinflammatory mediator implicated in endothelial dysfunction and atherosclerotic plaque progression, which appears to contradict our findings. However, emerging evidence suggests that LPC also exerts beneficial effects under specific pathological conditions, such as pulmonary arterial hypertension.⁴² Furthermore, Huang et al. have reported that multiple lipid species, including LPC and PC, characterized by a total acyl chain carbon number ≥54 and/or total double bond number ≥4, are significantly decreased in the plasma of patients with aortic dissection.⁴³ In addition, the specific biological activities of PC, such as involvement in lipoprotein assembly and secretion, modulation of inflammation, apoptosis, and intracellular signaling, are often influenced by the fatty acid composition at the sn-1 and sn-2 positions of its side chains.⁴⁴ More unsaturated acyl chains at the sn-2 position can enhance cell membrane fluidity, inhibit the maturation and aggregation of intracellular lipid droplets, and reduce aberrant lipid deposition in vascular endothelial cells.⁴⁵ This may represent an important mechanism underlying its protective effects. Our enrichment analysis further established an association between the biosynthesis pathway of unsaturated fatty acids and IA. Taken together, these findings highlight the remarkable context-dependent complexity in the physiological functions of PC and LPC with distinct side-chain structures. They also indicate that the lipid metabolic disturbance in the IA wall is not simply a recapitulation of the lipid abnormalities observed in atherosclerosis. Interestingly, our colocalization analysis provided further support for a potential shared causal variant, rs102274 (mapped gene: TMEM258 queried by GWAS catalog), between the two protective metabolites and IA, indicating the potential genetic overlap and susceptibility between blood metabolites and IA.

Additionally, although previous descriptive studies have reported associations between local CSF metabolites and the presence of uIA, our MR analysis did not support a causal relationship after FDR correction. This discrepancy does not invalidate past observations but may be explained by several methodological and biological considerations. First, the statistical power of the current CSF GWAS was limited by its modest sample size (n = 291),¹⁴ particularly when compared with large-scale blood metabolites cohorts (n = 8299). However, it might be the only publicly available GWAS study on this topic to date. Second, the metabolites assessed in the CSF GWAS (covering 338 compounds primarily from carbohydrate, lipid, amino acid, and organic acid) may not have captured key inflammatory mediators previously implicated in uIA pathogenesis, such as members of the interleukin family.⁴⁶ Third, descriptive studies can identify correlations but cannot establish direction; it remains plausible that aneurysm formation alters the local CSF metabolic environment, rather than metabolites driving aneurysm development. Despite the practical challenges of obtaining CSF samples, future large-scale, targeted CSF metabolomic studies remain warranted in this domain.

Our study has certain limitations. First, when selecting SNPs for metabolites, a less stringent threshold (p < 1 × 10⁻⁵) was used instead of the conventional threshold (p < 5 × 10⁻⁸) to ensure a sufficient number of SNPs for analysis. Second, the available demographic data were not intact, which may have introduced bias and heterogeneity. Third, due to the limitations in current metabolomics detection techniques, many metabolites remain unidentified, potentially impacting the robustness of the conclusions. Furthermore, the GWAS data used in this study were derived from individuals of European ancestry across multiple countries. The genetic architecture underlying metabolite levels as well as their causal relationships with IA may differ in nonEuropean populations due to variations in allele frequencies, linkage disequilibrium patterns, and gene–environment interactions. This highlights a crucial direction for future research; large-scale metabolomic GWAS studies in diverse ancestral populations (e.g. African and Asian) are urgently needed to determine which aspects of metabolite-associated IA risk are shared across humanity and which are population-specific.

Conclusion

Our study highlighted the potential causal relationship between specific blood metabolites and IA and identified the potential metabolic pathways involved in IA pathogenesis. Future GWAS with larger sample sizes are needed to verify the causal link between CSF metabolites and IA. The metabolites identified in this preliminary investigation may serve as potential noninvasive indicators of IA instability; however, this requires further validation through in vivo experiments and clinical studies.

Supplemental Material

sj-pdf-1-imr-10.1177_03000605261442755 - Supplemental material for Exploring the causal effect of genetically predicted blood and cerebrospinal fluid metabolites and metabolic pathways on intracranial aneurysm

Supplemental material, sj-pdf-1-imr-10.1177_03000605261442755 for Exploring the causal effect of genetically predicted blood and cerebrospinal fluid metabolites and metabolic pathways on intracranial aneurysm by Lu Ding, Wei Cao, Zhou Zhou, Zhaojun Mei, Bo Chen, Xinyu Lu and Wei Chen in Journal of International Medical Research

Supplemental Material

sj-pdf-2-imr-10.1177_03000605261442755 - Supplemental material for Exploring the causal effect of genetically predicted blood and cerebrospinal fluid metabolites and metabolic pathways on intracranial aneurysm

Supplemental material, sj-pdf-2-imr-10.1177_03000605261442755 for Exploring the causal effect of genetically predicted blood and cerebrospinal fluid metabolites and metabolic pathways on intracranial aneurysm by Lu Ding, Wei Cao, Zhou Zhou, Zhaojun Mei, Bo Chen, Xinyu Lu and Wei Chen in Journal of International Medical Research

Supplemental Material

sj-xls-3-imr-10.1177_03000605261442755 - Supplemental material for Exploring the causal effect of genetically predicted blood and cerebrospinal fluid metabolites and metabolic pathways on intracranial aneurysm

Supplemental material, sj-xls-3-imr-10.1177_03000605261442755 for Exploring the causal effect of genetically predicted blood and cerebrospinal fluid metabolites and metabolic pathways on intracranial aneurysm by Lu Ding, Wei Cao, Zhou Zhou, Zhaojun Mei, Bo Chen, Xinyu Lu and Wei Chen in Journal of International Medical Research

Supplemental Material

sj-xls-4-imr-10.1177_03000605261442755 - Supplemental material for Exploring the causal effect of genetically predicted blood and cerebrospinal fluid metabolites and metabolic pathways on intracranial aneurysm

Supplemental material, sj-xls-4-imr-10.1177_03000605261442755 for Exploring the causal effect of genetically predicted blood and cerebrospinal fluid metabolites and metabolic pathways on intracranial aneurysm by Lu Ding, Wei Cao, Zhou Zhou, Zhaojun Mei, Bo Chen, Xinyu Lu and Wei Chen in Journal of International Medical Research

Supplemental Material

sj-xls-5-imr-10.1177_03000605261442755 - Supplemental material for Exploring the causal effect of genetically predicted blood and cerebrospinal fluid metabolites and metabolic pathways on intracranial aneurysm

Supplemental material, sj-xls-5-imr-10.1177_03000605261442755 for Exploring the causal effect of genetically predicted blood and cerebrospinal fluid metabolites and metabolic pathways on intracranial aneurysm by Lu Ding, Wei Cao, Zhou Zhou, Zhaojun Mei, Bo Chen, Xinyu Lu and Wei Chen in Journal of International Medical Research

Supplemental Material

sj-xls-6-imr-10.1177_03000605261442755 - Supplemental material for Exploring the causal effect of genetically predicted blood and cerebrospinal fluid metabolites and metabolic pathways on intracranial aneurysm

Supplemental material, sj-xls-6-imr-10.1177_03000605261442755 for Exploring the causal effect of genetically predicted blood and cerebrospinal fluid metabolites and metabolic pathways on intracranial aneurysm by Lu Ding, Wei Cao, Zhou Zhou, Zhaojun Mei, Bo Chen, Xinyu Lu and Wei Chen in Journal of International Medical Research

Supplemental Material

sj-xls-7-imr-10.1177_03000605261442755 - Supplemental material for Exploring the causal effect of genetically predicted blood and cerebrospinal fluid metabolites and metabolic pathways on intracranial aneurysm

Supplemental material, sj-xls-7-imr-10.1177_03000605261442755 for Exploring the causal effect of genetically predicted blood and cerebrospinal fluid metabolites and metabolic pathways on intracranial aneurysm by Lu Ding, Wei Cao, Zhou Zhou, Zhaojun Mei, Bo Chen, Xinyu Lu and Wei Chen in Journal of International Medical Research

Supplemental Material

sj-pdf-8-imr-10.1177_03000605261442755 - Supplemental material for Exploring the causal effect of genetically predicted blood and cerebrospinal fluid metabolites and metabolic pathways on intracranial aneurysm

Supplemental material, sj-pdf-8-imr-10.1177_03000605261442755 for Exploring the causal effect of genetically predicted blood and cerebrospinal fluid metabolites and metabolic pathways on intracranial aneurysm by Lu Ding, Wei Cao, Zhou Zhou, Zhaojun Mei, Bo Chen, Xinyu Lu and Wei Chen in Journal of International Medical Research

Footnotes

Acknowledgment

The authors thank the contributions of the ISGC Intracranial Aneurysm working group as well as the participants and investigators of the FinnGen study. The authors acknowledge the use of ChatGPT (OpenAI) for language polishing and proofreading.

Author contributions

Study conceptualization and investigation: WC and LD; Methodology: WC, ZZ, and LD; Data curation, Formal analysis, and Visualization: WC, LD, ZZ, XYL, and ZJM; Writing–original draft: LD and WC; Writing–review and editing: WC, BC, and LD; Funding acquisition: XYL; Project administration, and Supervision: WC, and LD.

All authors critically reviewed the manuscript and approved the final draft.

Consent for publication

Not applicable.

Data availability statement

The analysis data are provided within the manuscript or in supplementary information files. The original GWAS data presented in the study are openly available for download by qualified researchers, as shown in .

Declaration of conflicting interests

The authors declare no commercial or financial relationships that could be construed as a potential conflict of interest.

Ethical approval and consent to participate

All GWAS databases were publicly available and had received approval from the corresponding ethical review board in the original GWAS, and all participants had provided written informed consent.

Funding

This study was supported by grants from the Zhenjiang High-level Leading Talents Cultivation Scientific Research Project (2021-169DT-19) and Wu Jieping Medical Foundation (320.6750.2024-2-18).

Supplemental material

Supplemental material for this article is available online.

ORCID iD

Wei Chen

References

Pontes

FGB

Da Silva

Baptista-Silva

, et al. Treatments for unruptured intracranial aneurysms. Cochrane Database Syst Rev 2021; 5: CD013312.

Claassen

Park

Spontaneous subarachnoid haemorrhage. Lancet 2022; 400: 846–862.

Liu

, et al. The markers and risk stratification model of intracranial aneurysm instability in a large Chinese cohort. Sci Bull (Beijing) 2023; 68: 1162–1175.

Qiu

Cai

Yao

, et al. Small molecule metabolites: discovery of biomarkers and therapeutic targets. Signal Transduct Target Ther 2023; 8: 132.

Wang

Lam

, et al. Integration of lipidomics and metabolomics for in-depth understanding of cellular mechanism and disease progression. J Genet Genomics 2020; 47: 69–83.

Sun

Zhang

, et al. Plasma metabolic signatures for intracranial aneurysm and its rupture identified by pseudotargeted metabolomics. Clin Chim Acta 2023; 538: 36–45.

Zhang

Feng

, et al. Study the local metabolic changes of aneurysms through microcatheter sampling. Sci Rep 2024; 14: 19552.

Kamińska

Dymicka-Piekarska

Chrzanowski

, et al. IL-6 Quotient (the ratio of cerebrospinal fluid IL-6 to serum IL-6) as a biomarker of an unruptured intracranial aneurysm. J Inflamm Res 2021; 14: 6103–6114.

Kamińska

Maciejczyk

Ćwiklińska

, et al. Pro-inflammatory and anti-inflammatory cytokines levels are significantly altered in cerebrospinal fluid of unruptured intracranial aneurysm (UIA) patients. J Inflamm Res 2022; 15: 6245–6261.

10.

Zuber

Grinberg

Gill

, et al. Combining evidence from Mendelian randomization and colocalization: review and comparison of approaches. Am J Hum Genet 2022; 109: 767–782.

11.

Zuccolo

Holmes

MV.

Commentary: Mendelian randomization-inspired causal inference in the absence of genetic data. Int J Epidemiol 2017; 46: 962–965.

12.

Reith

Landray

Devereaux

, et al. Randomized clinical trials–removing unnecessary obstacles. N Engl J Med 2013; 369: 1061–1065.

13.

Chen

Pettersson-Kymmer

, et al. Genomic atlas of the plasma metabolome prioritizes metabolites implicated in human diseases. Nat Genet 2023; 55: 44–53.

14.

Panyard

Kim

Darst

, et al. Cerebrospinal fluid metabolomics identifies 19 brain-related phenotype associations. Commun Biol 2021; 4: 63.

15.

Bakker

Van der Spek

RAA

Van Rheenen

, et al. Genome-wide association study of intracranial aneurysms identifies 17 risk loci and genetic overlap with clinical risk factors. Nat Genet 2020; 52: 1303–1313.

16.

Kurki

Karjalainen

Palta

; FinnGenet al. FinnGen provides genetic insights from a well-phenotyped isolated population. Nature 2023; 613: 508–518.

17.

Chang

Chow

Tellier

, et al. Second-generation PLINK: rising to the challenge of larger and richer datasets. Gigascience 2015; 4: 7.

18.

Qin

Havulinna

Liu

, et al. Combined effects of host genetics and diet on human gut microbiota and incident disease in a single population cohort. Nat Genet 2022; 54: 134–142.

19.

Rühlemann

Hermes

Bang

, et al. Genome-wide association study in 8,956 German individuals identifies influence of ABO histo-blood groups on gut microbiome. Nat Genet 2021; 53: 147–155.

20.

Meng

Tan

, et al. Causal association between common rheumatic diseases and glaucoma: a Mendelian randomization study. Front Immunol 2023; 14: 1227138.

21.

Burgess

Butterworth

Thompson

SG.

Mendelian randomization analysis with multiple genetic variants using summarized data. Genet Epidemiol 2013; 37: 658–665.

22.

Pierce

Burgess

Efficient design for Mendelian randomization studies: subsample and 2-sample instrumental variable estimators. Am J Epidemiol 2013; 178: 1177–1184.

23.

Bowden

Davey Smith

Burgess

Mendelian randomization with invalid instruments: effect estimation and bias detection through Egger regression. Int J Epidemiol 2015; 44: 512–525.

24.

Bowden

Davey Smith

Haycock

, et al. Consistent estimation in Mendelian randomization with some invalid instruments using a weighted median estimator. Genet Epidemiol 2016; 40: 304–314.

25.

Hartwig

Davey Smith

Bowden

Robust inference in summary data Mendelian randomization via the zero modal pleiotropy assumption. Int J Epidemiol 2017; 46: 1985–1998.

26.

Cohen

Chalumeau

Cohen

, et al. Cochran’s Q test was useful to assess heterogeneity in likelihood ratios in studies of diagnostic accuracy. J Clin Epidemiol 2015; 68: 299–306.

27.

Verbanck

Chen

Neale

, et al. Detection of widespread horizontal pleiotropy in causal relationships inferred from Mendelian randomization between complex traits and diseases. Nat Genet 2018; 50: 693–698.

28.

Burgess

Bowden

Fall

, et al. Sensitivity analyses for robust causal inference from Mendelian randomization analyses with multiple genetic variants. Epidemiology 2017; 28: 30–42.

29.

Hemani

Tilling

Davey Smith

Orienting the causal relationship between imprecisely measured traits using GWAS summary data. PLoS Genet 2017; 13: e1007081.

30.

Burgess

Thompson

SG.

Interpreting findings from Mendelian randomization using the MR-Egger method. Eur J Epidemiol 2017; 32: 377–389.

31.

Curtin

Schulz

Multiple correlations and Bonferroni’s correction. Biol Psychiatry 1998; 44: 775–777.

32.

Yuan

Larsson

SC.

Coffee and caffeine consumption and risk of kidney stones: a Mendelian randomization study. Am J Kidney Dis 2022; 79: 9–14.e1.

33.

Giambartolomei

Vukcevic

Schadt

, et al. Bayesian test for colocalisation between pairs of genetic association studies using summary statistics. PLoS Genet 2014; 10: e1004383.

34.

Kanehisa

Goto

Sato

, et al. KEGG for integration and interpretation of large-scale molecular data sets. Nucleic Acids Res 2012; 40: D109–D114.

35.

Madeo

Eisenberg

Pietrocola

, et al. Spermidine in health and disease. Science 2018; 359: eaan2788.

36.

Eisenberg

Abdellatif

Schroeder

, et al. Cardioprotection and lifespan extension by the natural polyamine spermidine. Nat Med 2016; 22: 1428–1438.

37.

LaRocca

Gioscia-Ryan

Hearon

, et al. The autophagy enhancer spermidine reverses arterial aging. Mech Ageing Dev 2013; 134: 314–320.

38.

Maimaiti

Turhon

Abulaiti

, et al. DNA methylation regulator-mediated modification patterns and risk of intracranial aneurysm: a multi-omics and epigenome-wide association study integrating machine learning, Mendelian randomization, eQTL and mQTL data. J Transl Med 2023; 21: 660.

39.

Fukui

Soda

Takao

, et al. Extracellular spermine activates DNA methyltransferase 3A and 3B. Int J Mol Sci 2019; 20: 1254.

40.

Pan

Shi

, et al. Intracranial aneurysms and lipid metabolism disorders: from molecular mechanisms to clinical implications. Biomolecules 2023; 13: 1652.

41.

Law

Chan

Marathe

, et al. An updated review of lysophosphatidylcholine metabolism in human diseases. Int J Mol Sci 2019; 20: 1149.

42.

Chen

Luo

, et al. Metabolomics reveals metabolite changes of patients with pulmonary arterial hypertension in China. J Cell Mol Med 2020; 24: 2484–2496.

43.

Huang

Lai

, et al. Plasma lipidomics identifies unique lipid signatures and potential biomarkers for patients with aortic dissection. Front Cardiovasc Med 2021; 8: 757022.

44.

Morita

Ikeda

Regulation of membrane phospholipid biosynthesis in mammalian cells. Biochem Pharmacol 2022; 206: 115296.

45.

Offermanns

Endothelial lipid droplets drive atherosclerosis and arterial hypertension. Trends Endocrinol Metab 2024; 35: 453–455.

46.

Turhon

Maimaiti

Gheyret

, et al. An immunogenic cell death-related regulators classification patterns and immune microenvironment infiltration characterization in intracranial aneurysm based on machine learning. Front Immunol 2022; 13: 1001320.

Supplementary Material

Please find the following supplemental material available below.

For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.

For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.

0.82 MB

0.20 MB

0.14 MB

15.82 MB

0.02 MB

0.30 MB

7.62 MB

0.21 MB

0.00 MB