Bioinformatics analysis reveals major hub genes involved with extracellular matrix and inflammatory and endocrine pathways associated with intracranial aneurysm tissue.

Background Intracranial aneurysm (IA) pathogenesis involves complex interplay between genetic predisposition and focal extracellular matrix (ECM) membrane degradation and inflammatory processes. We aimed to identify key differentially expressed genes (DEGs) that serve as hub genes (major genes with large networks) associated with IAs.

We conducted a comprehensive search of available Gene Expression Omnibus (GEO) databases for IA tissue from database inception to January 2024. This resulted in five GEO datasets, of which four were included as the discovery set, consisting of tissue from 28 IAs and 34 controls. DEGs were identified and used for enrichment analysis in evaluating Gene Ontology (GO) and Kyoto Encyclopedia of Genes and Genomes database pathways. A protein-protein interaction (PPI) DEG network was constructed to pinpoint interactions with other DEGs. The fifth GEO dataset was used to validate hub gene expressions.

We identified 1864 DEGs: 963 downregulated, 901 upregulated. Three gene clusters were linked to critical biological processes; notably, inflammatory response (GO:006954, false discovery rate [FDR] = 7.12 × 10−25), muscle contraction (GO:0006936, FDR = 1.1 × 10−3), and endocrine-related phosphatidylcholine sterol O-acyltransferase activator activity (GO:0060228, padj = 3.2 × 10−2) pathways. Eleven hub genes were identified, of which eight (COL1A, CXCR4, IL10, CXCL8, ESR1, APOE, RN1, and IGF1) were validated.

To our knowledge, this study represents the largest bioinformatics analysis to date on IAs, resulting in identification of 11 hub genes involved in ECM and immunologic pathways. These findings are consistent with existing literature; however, the potential involvement of endocrine-related processes, such as estrogen receptor signaling and cholesterol metabolism, is particularly intriguing and has not been previously well studied in this context.

Bioinformatics, intracranial aneurysms, gene expression, intracranial aneurysm, pathogenesis, ribonucleic acid sequencing

Intracranial aneurysm (IA) rupture remains a major cause of morbidity and mortality. Genetic predisposition and familial history remain well-established risk factors associated with IA development. Current understanding suggests that the pathogenesis of IAs involves a complex interplay between inherited genetic factors 1 and local processes, such as vascular extracellular matrix (ECM) remodeling and recruitment of inflammatory cells.2,3 There is a strong association between genetic predisposition and IA development. 4 Up to 20% of patients with IAs report a family history of aneurysms or subarachnoid hemorrhage, 5 and a twin-based study estimated the heritability of aneurysmal subarachnoid hemorrhage to be close to 40%. 5 However, our current understanding is limited, because common variants in genes only account for up to one-third of familial IAs. 6 These observations suggest the presence of additional genetic variants contributing to the risk of developing aneurysms that remain undiscovered. 1 It is becoming increasingly evident that complex diseases, such as IAs, are not attributed to single gene mutations with large effect sizes. Instead, they are influenced by multiple common genetic variants with modest effect sizes or a dysfunction in a network of genes that ultimately manifest the disease phenotype. The major challenge in evaluating the pathophysiology of IAs is the scarcity of available aneurysmal tissue. Access to IA tissue is primarily through surgical means, resulting in small and limited samples for genomic study. Furthermore, the adoption of endovascular treatment for IAs has further reduced the availability of tissue. Resultantly, many centers lack sufficient samples for large-scale ribonucleic acid (RNA) sequencing studies. In light of these limitations, leveraging available data and employing advanced bioinformatics tools is a viable solution. There is a shift in focus from studying individual genes to understanding comprehensive networks of genes, allowing us to understand the interplay between hundreds to thousands of genes involved in IAs. The purpose of this study was to identify and compile existing gene expression microarray data related to IAs in order to identify major genetic networks and critical pathways involved in aneurysmal development. Additionally, we sought to elucidate the relationship between local and systemic genomic variations in IAs through evaluation of gene expression changes in peripheral arterial blood compared to aneurysmal tissue. This comprehensive approach was undertaken to allow us to further our understanding between focal changes and inherited genetic mutations.

Ethics The STrengthening the Reporting of OBservational studies in Epidemiology (STROBE) reporting guidelines were followed (https://www.strobe-statement.org/). Data that support the study findings are available from the corresponding author on reasonable request and can be obtained from the Gene Expression Omnibus (GEO) database from the National Center for Biotechnology Information (NCBI) (https://www.ncbi.nlm.nih.gov/gds/), a public functional genomic data repository. Differentially expressed genes (DEGs) in tissue samples Data collection and processing Expression profile data associated with IAs were obtained from the GEO database. A query using the terms (“intracranial aneurysm”[MeSH Terms] OR “cerebral aneurysm”[All Fields]) AND (“gse”[Filter] AND “Homo sapiens”[Organism] AND “Expression profiling by array”[Filter]) retrieved 13 datasets of Homo sapiens from database inception to January 2024. The eligibility criteria included (a) original experimental dataset, (b) IA tissue and control arterial tissue sample data (obtained from superficial temporal arteries [STAs] in patients with IAs), and (c) messenger RNA (mRNA) expression profile. Studies were then excluded based on the following reasons: non-human studies, non-aneurysmal tissue, <6 samples in total, or providing microRNA expression profiles. No reporting biases were recorded in each study according to the GEO database. After identification of all eligible GEO datasets, one dataset was randomly selected as the validation set and not used for initial discovery analyses. All processing programs were executed using R language (R version 4.2.2; https://www.r-project.org/). Each included raw dataset was downloaded from the GEO database using GEOquery. Expression values for each sample were quantile normalized and log2 transformed for quality control and annotated with Bioconductor open-source software, v3.16 (https://bioconductor.org/news/bioc_3_16_release/). Common Entrez IDs were used to substitute and standardize gene probes with the corresponding microarray platforms. The Linear Model for Microarray package (https://bioconductor.org/packages/release/bioc/html/limma.html) was used to identify DEGs using Fisher's method to analyze differences between IAs and control groups. To account for potential batch effects across datasets, we applied the “removeBatchEffect” function from the limma package (https://rdrr.io/bioc/limma/man/removeBatchEffect.html) to the log2-transformed expression data, using dataset origin as the batch variable. A false discovery rate (FDR) of 0.05 was used to correct for multiple testing, and an adjusted p value (padj 1.5 were considered to be statistically significant. This cutoff is more stringent than the commonly used |log2 FC| > 1 and pad < 0.05, aiming to reduce false positives and focus on biologically meaningful changes. Meta-analysis of DEGs MetaVolcanoR, an open-source gene expression meta-analysis visualization tool (https://bioconductor.org/packages/release/bioc/html/MetaVolcanoR.html), was used to conduct meta-analysis across the included datasets and produce a volcano plot. A random-effects model was used to identify DEGs among all genes present in the aneurysm and control groups. Pathway enrichment analysis The R package clusterProfiler package was used to explore the Gene Ontology (GO) annotation (i.e. function) of DEGs, 7 with a cut-off criteria of padj < 0.05. The GO annotation contained three sub-ontologies: (a) biological processes (BPs), (b) cellular component (CC), and (c) molecular function (MF).7,8 Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway analysis was also used to explore key pathways among the genes of interest. 9 Construction of protein-protein interaction (PPI) network and identification of hub genes The Search Tool for the Retrieval of Interacting Genes (STRING) tool was used to construct a predictive PPI network. STRING is an online tool for building network models (https://string-db.org). 10 The PPI network was exported to and visualized using Cytoscape (v3.10.1). 11 Cytoscape is an open-source software for integrating and visualizing biomolecular interaction networks. The Molecular COmplex DEtection (MCODE) application was used to analyze and identify densely connected regions or clusters. 12 The MCODE score reflects the likelihood that the identified gene cluster is functionally relevant, where a higher score indicates more tightly connected clusters (0 = no connection). For identification of hub genes, or genes that play a statistically significant role in interactions with other DEGs, the cytoHubba Cytoscape application was used with six algorithms: three local-based methods, maximum neighborhood component (MNC), maximal clique centrality (MCC), and degree method (Deg), and three global-based methods, closeness (Clo), radiality (Rad), and stress (Str). 13 Each algorithm assigns a rank to every gene in the network based on its centrality in the network topology. For integrated ranking, we applied a consensus scoring approach: genes were ranked individually by each method, and an average rank score was computed across all algorithms. Genes with the lowest mean rank were considered hub genes. To facilitate integration and rank aggregation in R, we used dplyr functions such as full_join() to merge and align ranks across methods by gene symbol, followed by calculation of mean scores. 14 This integrated approach provided a robust prioritization of genes central to the network. Validation of hub genes using microarray GEO GSE54083 RNA sequencing (RNA-seq) data for IAs are limited due to the small number of tissue samples obtained in microsurgical clipping of aneurysms and the small amount of tissue obtained during the procedures. Therefore, we used GEO dataset GSE54083 for exploring and validating the significance of hub genes. The GSE54083 microarray dataset was acquired from the GEO database. The gene expression matrices were normalized by R package “limma” and log-transformed. The differential expression of the identified hub genes between IA tissue and control tissue was compared and visualized by the use of column graphs.

Identification of DEGs in aneurysmal tissue A total of 32 GEO databases were identified; 6 datasets did not contain aneurysmal tissue, 4 datasets were non-human tissue, 1 dataset had no control subjects, 12 datasets had non-RNA assay samples, and 3 datasets had less than 6 samples in total. One database (GSE46337) included control tissue from the middle meningeal artery, whereas six databases used the STA as the control. None of the datasets included normal intracranial vessels for comparison. To minimize bias from using different control tissues, only datasets with STA controls were selected for analysis. The exclusion of these datasets resulted in a total of five GEO datasets included in the final analysis of the aneurysmal tissue (GSE158558, GSE26969, GSE54083, GSE66238, GSE75436) (Figure 1). Four datasets were used as the discovery dataset, and the fifth, GSE54083, was randomly selected for use as the validation dataset and not included in the discovery dataset. The discovery dataset consisted of tissues obtained from 28 IAs and 34 STA controls (Table 1). Figure 2 illustrates a volcano plot that categorizes the upregulated and downregulated genes for each GSE dataset with padj at a threshold of p < 0.05. After multivariate analysis, GSE26969 had only one significant gene. Table 1. | GEO dataset | Sample source | Aneurysm samples (N) | Control tissue samples (N) | Array platform | |---|---|---|---|---| | Discovery datasets | |||| | GSE75436 | Beijing Tian Tan Hospital, China | 15 | 15 (STA) | GPL570 | | GSE26969 | Shandong University, China | 3 | 3 (STA) | GPL570 | | GSE66238 | Oregon State University, USA | 6 | 12 (STA) | GPL17303 | | GSE158558 | Fujian Medical University, China | 4 | 4 (STA) | GPL20301 | | Validation dataset | |||| | GSE54083 | National Institute of Genetics, Japan | 13 | 10 (STA) | GPL4133 | *The discovery datasets include tissue samples from 28 cerebral aneurysms and 34 control vessels. These data were published on the GEO database website (https://www.ncbi.nlm.nih.gov/gds/) between 2003 and 2020. GSE: gene set enrichment; STA: superficial temporal artery. A total of 1096 DEGs met the threshold of log2FC > 1.5 and p < 0.05. Among these genes, 627 genes were downregulated, and 469 genes were upregulated. The top (most significant p-value) 10 upregulated and downregulated genes are highlighted in Table 2. Table 2. | Ensembl gene symbol | Chromosome location | Log2-FC [95% CI] | p-value | Function | |---|---|---|---|---| | Upregulated | |||| | ADPIOQ-AS1 | 3q27.3 | 9.17 [7.20; 11.1] | 9.20 × 10−20 | Fat metabolism and insulin sensitivity | | PANTR1 | 2q12.1 | 5.47 [4.23; 6.70] | 4.80 × 10−18 | Upstream gene regulation, associated with squamous cell carcinoma and high-grade glioma | | PGM5P4 | 2q14.1 | 4.89 [3.86; 5.93] | 1.61 × 10−20 | Inhibits lung cancer progression | | SORBS2 | 4q35.1 | 3.74 [3.18; 4.29] | 6.10 × 10−40 | ABL kinases and actin cytoskeleton | | MYH11 | 16p13.11 | 3.90 [3.11; 4.69] | 4.15 × 10−22 | Smooth muscle myosin | | Downregulated | |||| | COL11A1 | 1p21.1 | −5.67 [−7.18; −4.16] | 2.17 × 10−13 | Encodes alpha chains of type XI collagen | | DSP | 6p24.3 | −3.25 [−3.83; −2.67] | 6.50 × 10−28 | Cell-cell adhesion | | CCL4L2 | 17q12 | −4.55 [−5.94; −3.16] | 1.42 × 10−10 | Cytokine gene in inflammatory and immunoregulatory processes | | PTN | 7q33 | −2.94 [−3.58; −2.29] | 5.81 × 10−19 | Cell migration, angiogenesis and tumorigenesis | | CCL3 | 17q12 | −4.16 [−5.45; −2.88] | 1.93 × 10−10 | Cytokine storm inflammatory response | CI: confidence interval; Log2FC: log-transformed fold change values representing differential expression. GO enrichment and KEGG pathway analyses of DEGs in aneurysmal tissue The 1864 DEGs identified from the discovery dataset were submitted for enrichment analyses (Figure 3). We identified three main categories of GO enrichment: BP, CC, and MF. In the BP category, key pathways included regulation of actin filament-based processes (128 genes, padj = 1.23 × 10−7), regulation of cell-substrate adhesion (80 genes, p = 1.35 × 10−5), and response to transforming growth factor beta (85 genes, p = 7.89 × 10−6). Enriched CC involved focal adhesion (135 genes, p = 1.67 × 10−7), the actin cytoskeleton (144 genes, p = 4.01 × 10−6), and collagen-containing ECM (111 genes, p = 2.95 × 10−3). At the MF level, ECM structural constituents (25 genes, p = 1.79 × 10−7), immune receptor activity (18 genes, p = 1.33 × 10−4), and cell-cell adhesion mediator activity (10 genes, p = 4.25 × 10−4) were significantly enriched (Table 3). Table 3. | Pathway category | Total genes | Adjusted p-value | |---|---|---| | GO enrichment biological process (BP) | || | GO:0032970 Regulation of actin filament-based process | 128 | 1.23 × 10−7 | | GO: 0010810 Regulation of cell-substrate adhesion | 80 | 1.35 × 10−5 | | GO:0071559 Response to transforming growth factor beta | 85 | 7.89 × 10−6 | | GO enrichment cellular component (CC) | || | GO:0005925 Focal adhesion | 135 | 1.67 × 10−7 | | GO:0015629 Actin cytoskeleton | 144 | 4.01 × 10−6 | | GO:0062023 Collagen-containing ECM | 111 | 2.95 × 10−3 | | GO enrichment molecular function (MF) | || | GO:0006201 ECM structural constituent | 25 | 1.79 × 10−7 | | GO:0140375 Immune receptor activity | 18 | 1.33 × 10−4 | | GO:0098632 Cell-cell adhesion mediator activity | 10 | 4.25 × 10−4 | | KEGG pathway | || | HSA04820 Cytoskeleton in muscle cells | 74 | 1.52 × 10−3 | | HSA04921 Oxytocin signaling pathway | 54 | 8.72 × 10−6 | | HSA04371 Apelin signaling pathway | 44 | 7.81 × 10−4 | | HSA04066 HIF-1 signaling pathway | 35 | 4.17 × 10−2 | ECM: extracellular matrix; GEO: Gene Expression Omnibus; GO: Gene Ontology; HSA: human serum albumin; KEGG: Kyoto Encyclopedia of Genes and Genomes. KEGG pathway analysis identified enrichment in cytoskeleton regulation in muscle cells (74 genes, p = 1.52 × 10−3) and oxytocin signaling (54 genes, p = 8.72 × 10−6), apelin signaling (44 genes, p = 7.81 × 10−4), and the HIF-1 signaling pathway (35 genes, p = 4.17 × 10−2) in vascular cells. These results suggest involvement of cytoskeletal dynamics, ECM remodeling, immune signaling, and vascular response pathways in aneurysmal tissue. Identification of key chromosomal genes and construction of PPI network The DEGs were entered into the STRING database for evaluation of PPIs. In the PPI network, each node represents a single protein conduct from a gene, and each edge represents a PPI. Our network analysis identified 882 nodes and 4619 edges, with an average clustering coefficient of 0.373 (PPI enrichment p-value <1.0 × 10−16), indicating a biologically connected group of proteins. The network was sent to Cytoscape for identification of statistically significant modules of the PPI network using MCODE (Figure 4). A threshold of four for MCODE was set, resulting in three clusters reaching significance. Cluster 1 had an MCODE score of 28, with 36 nodes and 493 edges. GO analysis showed that the proteins in Cluster 1 were involved with inflammatory response (GO:006954, padj = 7.12 × 10−25). KEGG pathway analysis showed Cluster 1 proteins to be associated with vital protein interaction with cytokine and cytokine receptor (hsa04061, padj = 1.27 × 10−14). Cluster 2 had 31 nodes and 84 edges (MCODE score of 5.6). GO analysis showed Cluster 2 proteins were associated with muscle contraction (GO:0006936, padj = 1.1 × 10−3), and KEGG pathway analysis showed Cluster 2 to be associated with vascular smooth muscle contraction (hsa04270, padj = 2.25 × 10−5). Cluster 3 had six nodes and 11 edges (MCODE score of 4.4). GO analysis demonstrated Cluster 3 proteins to be associated with phosphatidylcholine-sterol O-acyltransferase activator activity (GO:0060228, padj = 1.1 × 10−2) and the KEGG pathway to be associated with cholesterol metabolism (hsa04979, padj = 3.2 × 10−2). Identification and analysis of hub genes CytoHubba with consensus scoring for six algorithms (MCC, MNC, Deg, Clo, Rad, and Str) identified the 11 hub genes (Table 4). Of the 11 identified hub genes, two were associated with ECM structure and remodeling: COL1A1 and FN1. Four genes were related to inflammatory and immune responses: CXCL8, IL10, CD4, and CD8A. Two genes, ESR1 and APOE, were linked to endocrine signaling and lipid metabolism. Three genes—CXCR4, TNFRSF family member (presumed TN), and APOE (also involved here)—were associated with cell signaling, migration, and vascular repair processes. Table 4. | Gene symbol | Full name | Chromosome | Function | Diseases or alterations associated with gene | |---|---|---|---|---| | COL1A1 | Collagen type I alpha 1 chain | 17q21.33 | Encodes pro-alpha1 chains of type I collagen | Osteogenesis imperfecta types I–IV, EDS type VIIA, classical EDS, Caffey disease, idiopathic osteoporosis | | CXCR4 | C-X-C motif chemokine receptor 4 | 2q22.1 | G-protein-coupled receptor for stromal-derived factor-1 (SDF-1/CXCL12) | WHIM syndrome, cancer metastasis, HIV entry co-receptor | | IL10 | Interleukin 10 | 1q32.1 | Anti-inflammatory cytokine; regulates immune response | Inflammatory bowel disease, Crohn's disease, ulcerative colitis, autoimmune disorders | | CXCL8 | C-X-C motif chemokine ligand 8 (IL-8) | 4q13.3 | Chemotactic factor for neutrophils; pro-inflammatory cytokine | Elevated in infections, cancers, inflammatory diseases (e.g. RA, COPD, psoriasis) | | ESR1 | Estrogen receptor 1 | 6q25.1 | Nuclear hormone receptor for estrogen; regulates gene expression | Breast cancer, endometriosis, osteoporosis, cardiovascular disease | | CD4 | CD4 molecule | 12p13.31 | Co-receptor in T-helper cells; binds MHC class II | HIV entry receptor, immunodeficiency, autoimmune diseases | | CD8A | CD8a molecule | 2p11.2 | Co-receptor for MHC class I on cytotoxic T-cells | Immunodeficiency, viral infections, cancer immune responses | | APOE | Apolipoprotein E | 19q13.32 | Lipid transport and injury repair in the CNS | Alzheimer's disease (especially ε4 allele), cardiovascular disease, type III hyperlipoproteinemia | | FN1 | Fibronectin 1 | 2q35 | Glycoprotein involved in cell adhesion, growth, migration, differentiation | Glomerulopathy with fibronectin deposits, cancer, wound healing defects | | TNF | Tumor necrosis factor | 6p21.33 | Pro-inflammatory cytokine; regulates immune cells | Rheumatoid arthritis, psoriasis, ankylosing spondylitis, Crohn's disease, septic shock | | IGF1 | Insulin-like growth factor 1 | 12q23.2 | Mediates effects of growth hormone; cell proliferation and survival | Laron syndrome (deficiency), cancer, acromegaly (elevated levels), growth retardation | COPD: chronic obstructive pulmonary disease; EDS: Ehlers-Danlos syndrome; HIV: human immunodeficiency virus; RA: rheumatoid arthritis; WHIM: warts, hypogammaglobulinemia, infections, and myelokathexis. Validation of expression of the 11 hub genes The validation dataset (GSE54083) was used to validate differential expression of the 11 hub genes. Our analysis found eight genes to reach statistical significance (COL1A1, CXCR4, IL10, CXCL8, ESR1, APOE, FN1, and IGF1). One hub gene was downregulated (COL1A1), and seven hub genes were upregulated (CXCR4, IL10, CXCL8, ESR1, APOE, FN1, and IGF1). Figure 5 shows the expression levels between aneurysmal and control tissues.

The main purpose of our study was to identify genes associated with major networks (hub genes) and key regulatory pathways for understanding underlying IA mechanisms. Using a combined GEO dataset approach, we identified all available datasets to date, resulting in a discovery set consisting of four datasets with 28 IA tissues. We identified 963 downregulated and 901 upregulated genes. We constructed a PPI network analysis for these genes to identify hub genes and found 11 genes as potential key players in the development of IAs. Of these relevant genes, eight were validated to be differentially expressed in a separate validation dataset: one hub gene was downregulated, and seven hub genes were upregulated. The identification of novel genes and gene networks is critical to understanding the predisposition for developing IAs. Although genome-wide association studies have previously identified genetic loci associated with IAs,1,15–18 our current understanding of the genomic variances in IAs only explain a small subset of the familial disease. Recently, the International Consortium on Aneurysm (ICAN) study group identified 11 new risk loci and demonstrated a polygenic architecture. However, this still only explains half of disease heritability. 16 There is currently no genetic risk prediction test to identify individuals at risk for IAs for clinical triaging. Unlike Mendelian diseases where a single gene results in a large effect, there is increasing understanding that IAs and other complex diseases involve a large array of genetic variations with a small effect. 19 The limited availability of IA tissue and decreasing availability of IA tissue for research due to advancement of endovascular therapy have prompted increasing effort into developing and using bioinformatics analysis for large-scale evaluation of gene and protein networks. Prior studies using GEO databases have been successful in identifying novel genes but have been limited due to their small number of subjects. Using two GEO datasets, Zhao et al. recently identified SLC2A12 to be a potential gene of interest in the pathogenesis of IAs. 20 In a subsequent study, that group also identified the cytokine-cytokine receptor interaction signaling pathway to also likely be involved in IA pathogenesis. 21 Similarly, Wu et al. utilized three GEO datasets and identified 79 DEGs and found two genes of interest, KIAA0226L and UPP1, to be differentially expressed. 22 Consistent with our study, Zhong et al. identified eight hub genes across four datasets, including COL1A1 and POSTN. 23 The variability in identified differential genes across these microarray analyses likely stems from variations across methods and platforms but substantiates the hypothesis that the pathophysiology of aneurysms involves alterations in a large subset of genes. To address these challenges, we performed a comprehensive screen for all available GEO datasets and incorporated all available data existing for human IA tissue in this study. Furthermore, we selected one dataset as a validation dataset to confirm that the differential expressions found in our hub genes were validated. Our pathway enrichment analysis showed that collagen-containing ECM remodeling, immune signaling, and endocrine pathways play a major role in the development of IAs. This is consistent with our current understanding of the pathophysiology of IA. The pathophysiology of aneurysm development includes ECM defect and degradation, hemodynamic stress at arterial junctions, and acute and chronic inflammation resulting in aneurysmal dilatation and eventual rupture. 24 The DEGs identified across the datasets clustered into three main functional categories: ECM structural components, inflammatory and immune-related genes, and signaling pathways linked to cell migration and vascular remodeling. Of the 11 hub genes identified, COL1A and fibronectin (FN1) are directly involved in ECM structure and remodeling. COL1A1 gene has been associated with aneurysms in the context of the Ehlers-Danlos syndrome. 25 Mutations in the COL1A1 gene, such as the p.Arg312Cys variant, have also been associated with vascular fragility. 26 FN1 encodes a high-molecular-weight glycoprotein that is a critical component of the ECM and plays a major role in cell adhesion and maintenance of vascular integrity.27,28 Dysregulation of FN1 has been linked with inflammation, endothelial dysfunction, and weakening of the arterial wall, which are all hallmarks of aneurysm formation and progression. 29 It has also been associated with the promotion of thoracic aortic aneurysm in the rodent model and found to have elevated expression in plasma of patients with IA.30–33 Its identification as a hub gene supports its relevance in the mechanisms behind IA development. Four hub genes (CXCL8, IL-10, CD4, and CD8A) reflected immune and inflammatory responses, including cytokine signaling and T-cell activation. CXCL8 is a member of the CXC chemokine family responsible for mediating the inflammatory response by encoding interleukin-8 (IL-8). 32 CXCL8 mRNA expression is also noted to be statistically elevated in human abdominal aortic aneurysms (AAAs), and the activation of the CXCL8–CXCR1–2 pathway has been identified to be a distinctive feature of AAA and a potential target for AAA stabilization. 34 Inhibition of chemokines, such as CXCR2, has significantly reduced the size of AAA and attenuated inflammation and phenotypic changes in the vasculature. 35 Taken together, similar to AAA pathogenesis, this suggest CXCL8 may be a potential target for IAs. Interleukin 10 (IL-10) is also a key anti-inflammatory cytokine that plays a role in immune responses. IL-10 is thought to be associated with macrophage activation and vascular wall inflammation. Three hub genes, ESR1, APOE and IGF1, are linked to hormonal signaling and lipid metabolism, suggesting a possible endocrine contribution. Estrogen receptor 1 (ESR1) encodes the alpha subtype of the estrogen receptor. Estrogen signaling through ESR has been associated with endothelial cell function, alteration in vascular tone, and ECM integrity. 36 ESR1 regulates vascular endothelial growth factor A, a key factor in regulation of vascular tissue. 37 ESR1 and female hormones may play a role in sex differences in aneurysm rupture risk, as women are disproportionately affected by this disease. 38 The identification of ESR1 as a hub gene in IA tissue supports the hypothesis that hormonal regulation may play a role in IA pathophysiology and highlights the need for further research into sex-specific mechanisms. APOE has been associated with worse neurologic outcome after ruptured IA and cerebral amyloid angiopathy.39–42 Insulin-like growth factor 1 (IGF1) is a hormone involved in cell growth, tissue maintenance, and maintaining vascular integrity of vascular smooth muscle cells and endothelial cells. IGF1 has been proposed as a potential mechanism contributing to pathogenesis of intracerebral hemorrhages in aging.43,44 Our identification of hub genes and associated pathways aligns with the growing understanding of the complex molecular interactions involved in IA pathogenesis. Recent genome-wide association studies have identified novel risk loci and, through drug-target enrichment analysis, suggested potential links between IA and antiepileptic as well as sex hormone drugs, highlighting the possible involvement of endocrine pathways. 16 Although genome-wide association studies provide important insights into genetic variants associated with disease risk, network-based analysis enables deeper exploration of how these genes interact within biological pathways, offering a more integrated view of IA mechanisms. The identification of hub genes involved in IA pathogenesis has clinically relevant implications. Our findings align with the existing understanding that aneurysmal development is a multifactorial disease, combining an underlying genetic predisposition with a focal gene-environment interaction. These hub genes, which represent major regulators in complex networks and pathways, may have the potential to serve as biomarkers to improve rupture risk stratification beyond our current conventional imaging. Currently, there is no biomarker or panel of biomarker used clinically for screening of IA, but further exploration of multiple gene or specific expression may allow for surveillance, allowing for more timely intervention and personalized surveillance.45,46 Understanding biological pathways linked to these hub genes may reveal not only diagnostic but also novel therapeutic targets, including modulation of inflammatory responses or ECM remodeling. These targets could complement or enhance current endovascular treatments, potentially through direct intra-arterial therapies. Notably, genes involved in endocrine-signaling pathways, such as ESR1 and IGF1, present promising targets for further exploration of IA pathogenesis. Ultimately, integrating molecular biomarkers derived from these hub genes with clinical and radiographic data may advance our diagnostic and treatment approaches for IA. Future development of a panel of biomarkers for screening, incorporating these hub genes and their associated pathway, could be a valuable next step in changing the care for patients with IA. Additionally, future studies will focus on validating the identified hub genes in patient cohorts and integrating these findings with clinical and imaging data. This will involve developing computational models that combine gene expression profiles with hemodynamic and clinical variables. Such approaches aim to translate our bioinformatic results into clinically actionable tools to improve patient care.

The major limitations of this study are the heterogeneity that we cannot account for across the several studies and the meta-analysis study design. We acknowledge that meta-analysis of several microarrays would dampen the effects of differential genes with small variations and render them nonsignificant. Our study uses STA tissue as the control for IA tissue, consistent with prior transcriptomic studies. To reduce heterogeneity, we excluded one dataset that used MMA samples as controls. Although the STA is commonly used in genomic analyses due to its accessibility, it is an extracranial vessel and differs from intracranial arteries in embryologic origin, structural features, and exposure to hemodynamic forces. These differences may influence baseline gene expression and potentially confound the interpretation of DEGs observed in IA tissue. As such, some transcriptomic changes may reflect vascular bed-specific characteristics rather than IA-specific pathology. Furthermore, the gene expression changes associated with IAs and a causal relationship cannot be established. Despite this, we were able to identify 1864 significant genes, of which 11 were clearly identified to have major connections with other DEGs.

In this study, we identified major hub genes identified from IA tissue that were associated with the ECM and inflammatory and endocrine pathways. Further investigation into these identified hub genes and their network-associated genes is needed for clinical translation into IA screening and therapeutic targets. Abbreviations and acronyms - AAA Abdominal aortic aneurysm - APOE Apolipoprotein E - BP Biological processes - CC Cellular component - CD4 Cluster of differentiation 4 - CD8A Cluster of differentiation 8 alpha chain - CLO Closeness - COL1A Collagen type I alpha 1 chain - CXCL8 CXC motif chemokine ligand 8 - CXCR4 CXC motif chemokine receptor 4 - Deg Degree method - DEG Differentially expressed genes - ECM Extracellular matrix - ESR1 Estrogen receptor 1 - FC Fold change - FDR False discovery rate - FN1 Fibronectin - GEO Gene Expression Omnibus - GSE Gene Set Enrichment - GO Gene Ontology (Project) - IGF1 Insulin-like growth factor 1 - IL-8 Interleukin-8 - IL-10 Interleukin-10 - KEGG Kyoto Encyclopedia of Genes and Genomes - MCC Maximal clique centrality - MCODE Molecular COmplex DEtection - MF Molecular function - MMP Matrix metalloproteinase - MNC Maximum neighborhood component - mRNA Messenger ribonucleic acid - microRNA Micro-ribonucleic acid - NBCI National Center for Biotechnology Information - p adj Adjusted p value - PPI Protein-protein interaction - PRISMA Preferred Reporting Items for Systematic reviews and Meta-Analyses - Rad Radiality - RNA Ribonucleic acid - RNA-seq RNA sequencing - STA Superficial temporal artery - Str Stress - STRING Search Tool for the Retrieval of Interacting Genes - STROBE STrengthening the Reporting of OBservational studies in Epidemiology - TNF Tumor necrosis factor Footnotes Author contributions: Conception and design: Lai. Acquisition of data: Lai. Analysis and interpretation of data: Lai, Morgan, Tutino, Siddiqui, and Levy. Drafting the manuscript: Lai and Morgan. Revising the manuscript: All authors. Final approval of the version to be submitted: All authors. Accountable for all aspects of the work: All authors. Data availability: Data that support the study findings are available from the corresponding author on reasonable request and can be obtained from the Gene Expression Omnibus (GEO) database from the National Center for Biotechnology Information (https://www.ncbi.nlm.nih.gov/gds/), a public functional genomic data repository. Declarations of conflicting interests: The authors declared the following potential conflicts of interest with respect to the research, authorship, and/or publication of this article: Lai: Research Grants: CNS Young Investigator Grant, UB-CAT, Joe Niekro Foundation Grant, Bee Foundation Grant. Morgan: None. Tutino: Financial interest/investor/stock options/ownership: Neurovascular Diagnostics, Inc. Siddiqui: Financial Interest/Investor/Stock Options/Ownership: Adona Medical, Inc., Basecamp Vascular SAS, Bend IT Technologies, Ltd, BlinkTBI, Inc, Borvo Medical, Inc., CerebrovaKP, Code Zero Medical, Inc., Cognition Medical, Collavidence, Inc., Contego Medical, Inc., CVAID Ltd, E8, Inc., Endostream Medical, Ltd, FreeOx Biotech, SL, Galaxy Therapeutics, Inc., Hyperion Surgical, Inc., Imperative Care, Inc., InspireMD, Ltd, Instylla, Inc., IRRAS AB, Launch NY, Inc., Neurolutions, Inc., Neurovascular Diagnostics, Inc., NeXtGen Biologics, Peijia Medical, PerFlow Medical, Ltd, Physician X, LLC, Piraeus Medical, Inc., Prometheus Therapeutics, Inc., Q’Apel Medical, Inc., QAS.ai, Inc., Radical Catheter Technologies, Inc., Rist Neurovascular, Inc. (Purchased 2020 by Medtronic), Sense Diagnostics, Inc., Serenity Medical, Inc., Silk Road Medical, Sim & Cure, Spinnaker Medical, Inc., StimMed, LLC, Synchron, Inc., T.G. Medical, Inc., Tulavi Therapeutics, Inc., Vastrax, LLC, Viseon, Inc., Viz.ai, Whisper Medical, Inc., Willow Medtech, Inc. Consultant/Advisory Board: Asahi Intecc Co. Ltd, Canon Medical Systems USA, Inc., CerebrovaKP, Cerenovus, Contego Medical, Inc., Cordis, Endostream Medical, Ltd, FreeOx Biotech, SL, Hyperfine Operations, Inc., Imperative Care, InspireMD, Ltd, IRRAS AB, Medtronic, MicroVention (now Terumo Neuro), Minnetronix Neuro, Inc., Peijia Medical, Perflow Medical, Ltd, Piraeus Medical, Inc., Prometheus Therapeutics, Inc., Q’Apel Medical, Inc., Serenity Medical, Inc., Shockwave Medical, Inc., StimMed, LLC, Stryker Neurovascular., Synchron Australia Pty Ltd, T.G. Medical, Inc., Vastrax, LLC, Vesalio, Viz.ai, Inc., WL Gore. National PI/Steering Committees: Cerenovus EXCELLENT and ARISE II Trial; Medtronic SWIFT PRIME, VANTAGE, EMBOLISE and SWIFT DIRECT Trials; MicroVention (now Terumo Neuro) FRED Trial & CONFIDENCE Study; MUSC POSITIVE Trial; Penumbra 3D Separator Trial, COMPASS Trial, INVEST Trial, MIVI neuroscience EVAQ Trial; Rapid Medical SUCCESS Trial; InspireMD C-GUARDIANS IDE Pivotal Trial; Patents: Patent No. US 11,464,528 B2, Date: October 11, 2022, CLOT RETRIEVAL SYSTEM FOR REMOVING OCCLUSIVE CLOT FROM A BLOOD VESSEL, Applicant and Assignee: Neuravi Limited (Galway), Role: Co-Inventor. Levy: Consulting fees: Clarion, GLG Consulting, Guidepoint Global, Medtronic, StimMed, Mosaic; Payment or honoraria for lectures, presentations, speakers bureaus, manuscript writing or educational events: Medtronic, Penumbra, MicroVention (now Terumo Neuro), Integra; Patents planned, issued, or pending: Ultrasonic Surgical Blade; Participation on a Data Safety Monitoring Board or Advisory Board: NeXtGen Biologics, Cognition Medical; Endostream Medical, IRRAS AB; Leadership or fiduciary role in other board, society, committee or advocacy group, paid or unpaid: CNS, ABNS, UBNS; Stock or stock options (shareholder or ownership interest): NeXtGen Biologics, RAPID Medical, Claret Medical, Cognition Medical, Imperative Care, StimMed, Three Rivers Medical, Q’Apel, Dendrite; Other financial or non-financial interests: Haniva Medical Technology (Chief Medical Officer); Medtronic (National PI: Steering Committees for SWIFT Prime and SWIFT Direct trials; SHIELD trial; Site PI: STRATIS Study – Sub I); Penumbra (National PI: THUNDER trial); MicroVention (now Terumo Neuro) (Site PI: CONFIDENCE Study). Funding: The authors received no financial support for the research, authorship, and/or publication of this article. ORCID iDs: Pui Man Rosalind Lai https://orcid.org/0000-0002-8310-0474 Adnan H Siddiqui https://orcid.org/0000-0002-9519-0059 Elad I Levy https://orcid.org/0000-0002-6208-3724

- 1.Bakker MK, Ruigrok YM. Genetics of intracranial aneurysms. Stroke 2021; 52: 3004–3012. [DOI] [PubMed] [Google Scholar] - 2.Bruno G, Todor R, Lewis Iet al. et al. Vascular extracellular matrix remodeling in cerebral aneurysms. J Neurosurg 1998; 89: 431–440. [DOI] [PubMed] [Google Scholar] - 3.Hudson JS, Hoyne DS, Hasan DM. Inflammation and human cerebral aneurysms: current and future treatment prospects. Future Neurol 2013; 8: 10.2217/fnl.13.40. doi: 10.2217/fnl.13.40 [DOI] [PMC free article] [PubMed] [Google Scholar] - 4.Toader C, Eva L, Bratu BG, et al. Intracranial aneurysms and genetics: An extensive overview of genomic variations, underlying molecular dynamics, inflammatory indicators, and forward-looking insights. Brain Sci 2023; 13: 1454. doi: 10.3390/brainsci13101454 [DOI] [PMC free article] [PubMed] [Google Scholar] - 5.Brown RD, Jr., Huston J, Hornung R, et al. Screening for brain aneurysm in the familial intracranial aneurysm study: frequency and predictors of lesion detection. J Neurosurg. 2008;108:1132–1138. [DOI] [PMC free article] [PubMed] [Google Scholar] - 6.Schievink WI, Schaid DJ, Michels VVet al. et al. Familial aneurysmal subarachnoid hemorrhage: a community-based study. J Neurosurg 1995; 83: 426–429. [DOI] [PubMed] [Google Scholar] - 7.Gene Ontology C, Aleksander SA, Balhoff J, et al. The gene ontology knowledgebase in 2023. Genetics 2023; 224: iyad031. doi: 10.1093/genetics/iyad031 [DOI] [PMC free article] [PubMed] [Google Scholar] - 8.Ashburner M, Ball CA, Blake JA, et al. Gene ontology: tool for the unification of biology. The Gene Ontology Consortium. Nat Genet 2000; 25: 25–29. [DOI] [PMC free article] [PubMed] [Google Scholar] - 9.Kanehisa M, Goto S. KEGG: Kyoto Encyclopedia of Genes and Genomes. Nucleic Acids Res 2000; 28: 27–30. [DOI] [PMC free article] [PubMed] [Google Scholar] - 10.Szklarczyk D, Kirsch R, Koutrouli M, et al. The STRING database in 2023: protein-protein association networks and functional enrichment analyses for any sequenced genome of interest. Nucleic Acids Res 2023; 51: D638–D646. [DOI] [PMC free article] [PubMed] [Google Scholar] - 11.Shannon P, Markiel A, Ozier O, et al. Cytoscape: a software environment for integrated models of biomolecular interaction networks. Genome Res 2003; 13: 2498–2504. [DOI] [PMC free article] [PubMed] [Google Scholar] - 12.Bader GD, Hogue CW. An automated method for finding molecular complexes in large protein interaction networks. BMC Bioinformatics 2003; 4: 2. doi: 10.1186/1471-2105-4-2 [DOI] [PMC free article] [PubMed] [Google Scholar] - 13.Chin CH, Chen SH, Wu HH, et al. cytoHubba: identifying hub objects and sub-networks from complex interactome. BMC Syst Biol 2014; 8: S11. [DOI] [PMC free article] [PubMed] [Google Scholar] - 14.Wickham HRF, Henry L., Muller K. A grammar of data manipulation. R package version 1.1.4. - 15.Kim BJ, Hong EP, Youn DH, et al. and First Korean Stroke Genetics Association R. Genome-wide association study of the relationship between matrix metalloproteinases and intracranial aneurysms. J Clin Neurol. 2022;18:163–170. [DOI] [PMC free article] [PubMed] [Google Scholar] - 16.Bakker MK, van der Spek RAA, van Rheenen W, 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. [DOI] [PMC free article] [PubMed] [Google Scholar] - 17.Foroud T, Koller DL, Lai D, et al. Genome-wide association study of intracranial aneurysms confirms role of anril and SOX17 in disease risk. Stroke 2012; 43: 2846–2852. [DOI] [PMC free article] [PubMed] [Google Scholar] - 18.Gaastra B, Alexander S, Bakker MK, et al. A genome-wide association study of outcome after aneurysmal subarachnoid haemorrhage: discovery analysis. Transl Stroke Res 2023; 14: 681–687. [DOI] [PMC free article] [PubMed] [Google Scholar] - 19.Alg VS, Sofat R, Houlden Het al. et al. Genetic risk factors for intracranial aneurysms: a meta-analysis in more than 116,000 individuals. Neurology 2013; 80: 2154–2165. [DOI] [PMC free article] [PubMed] [Google Scholar] - 20.Zhao C, Ma Z, Shang J, et al. Bioinformatics analysis reveals potential biomarkers associated with the occurrence of intracranial aneurysms. Sci Rep 2022; 12: 13282. [DOI] [PMC free article] [PubMed] [Google Scholar] - 21.Zhu H, Tan J, Zhao Y, et al. Potential role of the chemotaxis system in formation and progression of intracranial aneurysms through weighted gene co-expression network analysis. Int J Gen Med 2022; 15: 2217–2231. [DOI] [PMC free article] [PubMed] [Google Scholar] - 22.Wu J, Chen ZJ, Liang J, et al. Identifying and validating key genes mediating intracranial aneurysm rupture using weighted correlation network analysis and exploration of personalized treatment. Ann Transl Med 2022; 10: 1057. [DOI] [PMC free article] [PubMed] [Google Scholar] - 23.Zhong F, Yao F, Xu S, et al. Identification and validation of hub genes and molecular classifications associated with chronic myeloid leukemia. Front Immunol 2023; 14: 1297886. [DOI] [PMC free article] [PubMed] [Google Scholar] - 24.Jung KH. New pathophysiological considerations on cerebral aneurysms. Neurointervention 2018; 13: 73–83. [DOI] [PMC free article] [PubMed] [Google Scholar] - 25.Takeda R, Yamaguchi T, Hayashi S, et al. Clinical and molecular features of patients with COL1-related disorders: implications for the wider spectrum and the risk of vascular complications. Am J Med Genet A 2022; 188: 2560–2575. [DOI] [PMC free article] [PubMed] [Google Scholar] - 26.Adham S, Dupuis-Girod S, Charpentier E, et al. Classical Ehlers-Danlos syndrome with a propensity to arterial events: a new report on a French family with a COL1A1 p.(Arg312Cys) variant. Clin Genet 2020; 97: 357–361. [DOI] [PubMed] [Google Scholar] - 27.Zollinger AJ, Smith ML. Fibronectin, the extracellular glue. Matrix Biol 2017; 60-61: 27–37. [DOI] [PubMed] [Google Scholar] - 28.Dalton CJ, Lemmon CA. Fibronectin: molecular structure, fibrillar structure and mechanochemical signaling. Cells 2021; 10: 2443. doi: 10.3390/cells10092443 [DOI] [PMC free article] [PubMed] [Google Scholar] - 29.Resnikoff HA, Schwarzbauer JE. Increased basal fibronectin is sufficient to promote excess endothelial cell matrix assembly causing localized barrier dysfunction. Mol Biol Cell 2024; 35: ar120. [DOI] [PMC free article] [PubMed] [Google Scholar] - 30.Wang Z, Chen YE, Chang L. Unleashing PD-1: a duel of immunity in aortic aneurysm formation. J Clin Invest 2024; 134: e182554. doi: 10.1172/JCI182554 [DOI] [PMC free article] [PubMed] [Google Scholar] - 31.Wang J, Syrett CM, Kramer MC, et al. Unusual maintenance of X chromosome inactivation predisposes female lymphocytes for increased expression from the inactive X. Proc Natl Acad Sci USA 2016; 113: E2029–E2038. [DOI] [PMC free article] [PubMed] [Google Scholar] - 32.Cambier S, Gouwy M, Proost P. The chemokines CXCL8 and CXCL12: molecular and functional properties, role in disease and efforts towards pharmacological intervention. Cell Mol Immunol 2023; 20: 217–251. [DOI] [PMC free article] [PubMed] [Google Scholar] - 33.Chen M, Cavinato C, Hansen J, et al. FN (Fibronectin)-integrin alpha5 signaling promotes thoracic aortic aneurysm in a mouse model of Marfan syndrome. Arterioscler Thromb Vasc Biol 2023; 43: e132–e150. [DOI] [PMC free article] [PubMed] [Google Scholar] - 34.Kokje VBC, Gabel G, Dalman RL, et al. CXCL8 hyper-signaling in the aortic abdominal aneurysm. Cytokine 2018; 108: 96–104. [DOI] [PubMed] [Google Scholar] - 35.Sun B, Li F, Lai S, et al. Inhibition of CXCR2 alleviates the development of abdominal aortic aneurysm in Apo E-/- mice. Acta Cir Bras 2021; 36: e360105. [DOI] [PMC free article] [PubMed] [Google Scholar] - 36.Sanchez AM, Flamini MI, Zullino S, et al. Estrogen receptor-alpha promotes endothelial cell motility through focal adhesion kinase. Mol Hum Reprod 2011; 17: 219–226. [DOI] [PubMed] [Google Scholar] - 37.Fatima LA, Campello RS, Santos RS, et al. Estrogen receptor 1 (ESR1) regulates VEGFA in adipose tissue. Sci Rep 2017; 7: 16716. [DOI] [PMC free article] [PubMed] [Google Scholar] - 38.Tada Y, Wada K, Shimada K, et al. Estrogen protects against intracranial aneurysm rupture in ovariectomized mice. Hypertension 2014; 63: 1339–1344. [DOI] [PMC free article] [PubMed] [Google Scholar] - 39.Csajbok LZ, Nylen K, Ost M, et al. Apolipoprotein E polymorphism in aneurysmal subarachnoid haemorrhage in West Sweden. Acta Neurol Scand 2016; 133: 466–474. [DOI] [PubMed] [Google Scholar] - 40.Gallek MJ, Conley YP, Sherwood PR, et al. APOE Genotype and functional outcome following aneurysmal subarachnoid hemorrhage. Biol Res Nurs 2009; 10: 205–212. [DOI] [PMC free article] [PubMed] [Google Scholar] - 41.Heinsberg LW, Turi E, Ren D, et al. Evaluation of APOE genotype and ability to perform activities of daily living following aneurysmal subarachnoid hemorrhage. Biol Res Nurs 2018; 20: 177–182. [DOI] [PMC free article] [PubMed] [Google Scholar] - 42.Rosalind Lai PM, Du R. Role of genetic polymorphisms in predicting delayed cerebral ischemia and radiographic vasospasm after aneurysmal subarachnoid hemorrhage: a meta-analysis. World Neurosurg 2015; 84: 933–41.e2. [DOI] [PubMed] [Google Scholar] - 43.Fulop GA, Ramirez-Perez FI, Kiss T, et al. IGF-1 Deficiency promotes pathological remodeling of cerebral arteries: a potential mechanism contributing to the pathogenesis of intracerebral hemorrhages in aging. J Gerontol A Biol Sci Med Sci 2019; 74: 446–454. [DOI] [PMC free article] [PubMed] [Google Scholar] - 44.Kulseng B, Myhre HO. Is insulin growth factor-1 (IGF-1) playing a role for aneurysm formation in patients with pituitary gland tumor? Int Angiol 2006; 25: 433–435. [PubMed] [Google Scholar] - 45.Hussain S, Barbarite E, Chaudhry NS, et al. Search for biomarkers of intracranial aneurysms: a systematic review. World Neurosurg 2015; 84: 1473–1483. [DOI] [PubMed] [Google Scholar] - 46.Wu Y, Zhao Z, Kang S, et al. Potential application of peripheral blood biomarkers in intracranial aneurysms. Front Neurol 2023; 14: 1273341. [DOI] [PMC free article] [PubMed] [Google Scholar]