Abstract
Patients with hemorrhagic stroke have high rates of morbidity and mortality, and drugs for prevention are very limited. Mendelian randomization (MR) analysis can increase the success rate of drug development by providing genetic evidence. Previous MR analyses only analyzed the role of individual drug target genes in hemorrhagic stroke; therefore, we used MR analysis to systematically explore the druggable genes for hemorrhagic stroke. We sequentially performed summary-data-based MR analysis and two-sample MR analysis to assess the associations of all genes within the database with intracranial aneurysm, intracerebral hemorrhage, and their subtypes. Validated genes were further analyzed by colocalization. Only genes that were positive in all three analyses and were druggable were considered desirable genes. We also explored the mediators of genes affecting hemorrhagic stroke incidence. Finally, the associations of druggable genes with other cardiovascular diseases were analyzed to assess potential side effects. We identified 56 genes that significantly affected hemorrhagic stroke incidence. Moreover, TNFSF12, SLC22A4, SPARC, KL, RELT, and ADORA3 were found to be druggable. The inhibition of TNFSF12, SLC22A4, and SPARC can reduce the risk of intracranial aneurysm, subarachnoid hemorrhage, and intracerebral hemorrhage. Gene-induced hypertension may be a potential mechanism by which these genes cause hemorrhagic stroke. We also found that blocking these genes may cause side effects, such as ischemic stroke and its subtypes. Our study revealed that six druggable genes were associated with hemorrhagic stroke, and the inhibition of TNFSF12, SLC22A4, and SPARC had preventive effects against hemorrhagic strokes.
Graphical Abstract
Avoid common mistakes on your manuscript.
Introduction
Hemorrhagic strokes mainly include intracerebral hemorrhage (ICH) and subarachnoid hemorrhage (SAH), which account for approximately 13% of all strokes [1]. ICH is a tissue injury caused by the rupture of microvessels deep in brain tissue and the accumulation of blood in the brain parenchyma. Secondary brain damage from later blood clot enlargement and perihematomal cerebral edema can lead to neurological deterioration and an increased likelihood of poor functional outcomes and mortality [2]. Eighty percent of SAH cases result from the rupture of intracranial aneurysms (IAs). One-third of the patients die without admission to the hospital, accompanied by complications such as rebleeding, delayed cerebral ischemia, and hydrocephalus, which seriously affects their quality of life [3, 4]. Considering the high morbidity and mortality of hemorrhagic strokes, prevention is very important. The current treatments are mainly surgery. With the use of effective drugs for preventive therapy, the incidence of hemorrhagic stroke can be reduced while avoiding damage caused by surgery; therefore, the exploration of preventive drug regimens is key.
The majority of research on drugs for ICH prophylaxis has concentrated on lipid-lowering therapies. However, the preventive effects of these agents have been inconsistent, and multiple meta-analyses have shown no evidence of the effect of lipid-lowering drugs on ICH [5, 6]. Observational studies on SAH suggest that antiplatelet agents may confer protective effects by decreasing the incidence of IA rupture [7, 8]. Moreover, different types of antihypertensive drugs have different effects on IA, with angiotensin-converting enzyme inhibitors and beta-blockers reducing the risk of IA, while calcium channel blockers have opposite effects; however, studies have shown that antihypertensive drugs are not associated with IA [9, 10]. In observational research, these findings are always constrained by the risk of confounding bias.
The rapid development of genomics has increased the prospects for the development of novel drugs. The development of drugs with genetically supported targets has a significantly higher success rate and a much lower rate of attrition [11]. Mendelian randomization (MR) is an epidemiological approach based on genome-wide association studies (GWASs), and it has been increasingly used for the discovery and validation of drug targets [12]. Single-nucleotide polymorphisms (SNPs) within a certain range near the target, such as expression quantitative trait loci (eQTLs) and protein quantitative trait loci (pQTLs), that have a significant effect on the corresponding product are used as instrumental variables (IVs) in drug target MR analysis research and are analyzed with disease GWAS datasets to determine whether the targets have an effect on the disease at the genetic level [13]. Drug target MR analysis has been performed for individual drugs for hemorrhagic stroke treatment, but no study has comprehensively analyzed potentially druggable genes associated with hemorrhagic stroke.
Accordingly, in this study, we used the eQTLs identified in blood and performed MR analysis with GWAS datasets of SAH and ICH to systematically identify effective druggable genes.
Methods
Study Design
We performed our analysis using the following procedures: (1) preliminary screening of the targets for hemorrhagic stroke using summary-data-based Mendelian randomization (SMR), (2) further two-sample MR analysis to obtain causal associations between SMR-screened targets and hemorrhagic strokes to reduce the spectrum of positive results, (3) colocalization analysis of positive targets identified by MR analysis to verify the robustness of the results, (4) search of the Drug–Gene Interaction Database for druggable genes, (5) causal analysis of cardiovascular risk factors with hemorrhagic strokes and assessment of the effects of validated druggable targets on risk factors to explore possible mediators of targets’ effects on diseases, and (6) assessment of the effects of druggable targets on other cardiovascular diseases (CVDs) and related metabolic disorders to understand possible side effects. Our MR analysis met three major assumptions: (1) IVs were significantly linked with exposures; (2) IVs were not associated with confounders that mix exposure-outcome correlations; and (3) IVs did not act directly on outcomes but rather influenced them through exposures [14]. Further details are provided below, and the process of this study is shown in Fig. 1.
Genetic Datasets
The eQTL datasets used in this study were obtained from the Genotype-Tissue Expression project (GTEx) version 8 and the eQTLGen Consortium. The GTEx project examined 15,201 RNA sequencing samples that were taken from 49 tissues of 838 postmortem donors (with 85% European individuals) and obtained the associated eQTL datasets [15]. The eQTLGen Consortium analyzed blood gene expression in 31,684 European individuals from 37 consortiums to identify eQTL information [16]. The cis-eQTL is located within the region of the regulated gene itself, such as the promoter and coding region of the gene, and is close to the regulated gene (within 1 Mb on either side of the coding gene). Furthermore, most drugs function through the circulation. Therefore, we used cis-eQTLs from blood samples from two consortia as our study’s IVs. For each detection, a false discovery rate of < 0.05 was considered significant [15, 16].
IA and ICH Datasets
The SAH data were based on a GWAS meta-analysis of intracranial aneurysms from the International Stroke Genetics Consortium. Data for individuals of European ancestry were obtained from individual genetic analysis of 23 cohorts with 7495 patients and 71,934 control individuals. Patients susceptible to IA, such as Marfan syndrome, autosomal dominant polycystic kidney disease, and Ehlers‒Danlos disease, were excluded; ultimately, a total of 4,471,083 SNPs passed the quality control. When all IA patients were categorized by imaging, 69% of the patients had SAH due to ruptured IAs (5140 patients), 28% had unruptured IAs (UIAs) (2070 patients), and 3.8% had an uncertain rupture status [17]. In our study, overall IA data, UIA data, and SAH data were analyzed in a stratified manner.
The data for ICH were based on a GWAS meta-analysis of 3026 individuals (1545 patients and 1481 control individuals) of European ancestry, with the diagnostic criteria being new onset of acute neurological deficit and the presence of intracerebral parenchymal hemorrhage on imaging; in addition, patients with trauma, tumors, vascular malformations, or secondary hemorrhage were excluded [18]. According to our stratified analysis, ICHs were grouped into lobar ICHs (684 patients; lobar ICHs mostly occurring in the cerebral cortex or cortical-subcortical junction) and deep ICHs (1161 patients; deep ICHs mostly occurring in the deeper parts of the hemisphere, cerebellum, and brainstem).
Identification of Possible Positive Targets via SMR Analysis
We first used the SMR method as a preliminary analytical tool, which integrates and analyzes GWAS data with eQTL data to identify causal associations between gene expression levels and complex traits [19]. We used SMR software, version 1.03, developed by Yang Lab of Westlake University, to analyze all eQTL data within eQTLGen and GTEx and the heterogeneity in dependent instrument (HEIDI) test to detect the presence of linkage disequilibrium (LD). The threshold for filtering SNPs was the standard of 5 × 10−8 and MAF > 0.01. The results were considered positive at P values < 0.05. Detailed descriptions of the software and its data parameters were obtained from https://yanglab.westlake.edu.cn/software/smr/#Overview.
Further Screening Using Two-Sample MR Analysis
If there are two different causal variants with a high degree of LD, the SMR may not have the statistical power to identify pleiotropy from linkage, leaving the result unstable [19]. We further performed two-sample MR analysis on the positive results obtained from SMR analysis. The R package Two Sample MR V.0.5.7 was used to perform the analysis. First, cis-eQTLs of positive genes were extracted as IVs, with the criterion of SNPs with P values < 5 × 10−8 and within ± 100 kb from the transcriptional start site of each gene [15, 16]. The window for LD was set to 10,000 kb with r2 < 0.001 (based on the 1000 Genomes European Reference Panel) to assure independence [20]. We also removed SNPs that could not be matched or had palindromic sequences when harmonizing the gene eQTL data with the outcome GWAS data.
The Wald ratio method was selected as the primary analysis method when only a single SNP was available, and the primary analysis method was the inverse-variance weighting (IVW) method if there were multiple SNPs. The results are expressed as the disease odds ratio (OR) per 1-SD change in gene expression. MR‒Egger regression, the weighted median, and the weighted mode were used as supplementary methods for sensitivity analysis [21]. When the number of SNPs exceeded 2, the presence of horizontal pleiotropy was determined by whether the MR‒Egger intercept was close to 0 [22]. The use of Cochran’s Q statistic to assess the presence of heterogeneity can also help assess horizontal pleiotropy, with P values < 0.05 indicating the presence of heterogeneity. In addition, leave-one-out analysis was performed by removing each SNP alone to assess the robustness of the MR results. The results were corrected by multiple comparisons using false discovery rate (FDR) correction to reduce the false-positive rate, and FDR-corrected P values < 0.05 were considered to indicate positive results.
Colocalization Analysis
Colocalization analysis is used to determine whether two phenotypes share the same causal variant within a region, which can provide strong evidence of overlapping genetic mechanisms between the two phenotypes [23]. There are five hypotheses: within the region analyzed via colocalization, H0: there is no SNP associated with exposure or outcome; H1/H2: there are SNPs associated with either exposure or outcome; H3: there are SNPs associated with both exposure and outcome but as two distinct SNPs; and H4: there are SNPs associated with both exposure and outcome and as the same SNP. We performed colocalization analysis between genes that were significant in both MR analyses and hemorrhagic strokes using the R package COLOC V.5.2.3 and default prior probabilities of P1 = 1 × 10−4, P2 = 1 × 10−4, and P12 = 1 × 10−5. P1, P2, and P12 represent the probabilities that a SNP is significantly associated with eQTLs, hemorrhagic strokes, or both, respectively. The posterior probability (PP) and likelihood of the five hypotheses were calculated using the Bayesian test, and the results were visualized using LocusCompareR [23, 24]. A range of genes ± 500 kb was taken as the region for colocalization analysis. We considered a significant colocalization analysis result of PPH4 > 0.75 or PPH3 + PPH4 > 0.8; genes strongly colocalized with hemorrhagic stroke were considered potential targets.
Assessing the Druggability of Genes
The Drug–Gene Interaction Database (DGIdb V.4.2.0) summarizing drug–gene interactions from existing studies and several drug–gene databases provides information on 42 potential drug categories and at least 49 interaction types, with a total of 5776 druggable genes identified [25]. We searched for positive targets obtained after multiple analyses from the DGIdb. A target was considered druggable if it was classified as part of the druggable genome in the drug information provided.
Evaluation of Potential Mediators
We further explored the potential mechanisms through which the identified drug targets influence hemorrhagic stroke incidence. Hypertension, smoking, and alcohol consumption are the leading causes of worldwide disease burden and obviously contribute to the occurrence of CVD [26]. Diabetes mellitus and improper glucose metabolism are the main causes of CVD, and CVD, which is the leading cause of death in patients with diabetes mellitus, is the primary preventive complication. Obesity serves as an important precursor to CVD complications [27]. Furthermore, lipids play an important role in CVD by affecting inflammation, leukocytes, and cardiovascular function, and lipid-lowering drugs are often used as critical therapeutic agents for a variety of diseases [28]. Therefore, we performed two-sample MR analyses of the following cardiovascular influences on patients with hemorrhagic stroke: hypertension (GWAS data published by the UK Biobank consortium, which were analyzed by the Neale laboratory among 360,000 people of white British ancestry) [29]; blood pressure (systolic blood pressure, diastolic blood pressure)[30]; smoking and alcohol consumption (drinks per week, smoking initiation, cigarettes per day, smoking cessation, age of initiation) [31]; lipid type (high-density lipoprotein, low-density lipoprotein, very low-density lipoprotein, total cholesterol, total triglycerides, apolipoprotein A1, and apolipoprotein B) [32]; obesity-related parameters (body mass index (BMI), body fat percentage, waist circumference, hip circumference, waist-to-hip ratio) [29, 33, 34]; and glucose traits (fasting glucose, fasting insulin, 2-h glucose, and glycated hemoglobin levels) [35]. The IVs were selected with the condition of P < 5 × 10−8, and SNPs with r2 > 0.001 and within a 10,000-kb window were removed. The final results after FDR correction were considered positive at P < 0.05. Because hypertension is a well-defined risk factor for SAH and ICH, we performed multivariable MR (MVMR) analysis of hypertension, systolic blood pressure, and diastolic blood pressure to determine the predominant risk factor [36]. The inclusion criteria for SNPs in the MVMR analysis were the same as those for univariate analysis. MR mediation analysis by the coefficient product method explored whether the other positive exposures had a direct effect on hemorrhagic strokes rather than affecting blood pressure completely. Finally, we used the risk factors as outcomes to detect the effect of the targets. If there is a significant effect of both the risk factor and risk factor on hemorrhagic stroke incidence, we can consider that these risk factors may be potential mediators.
Exploration of Potential Side Effects of Validated Targets
Considering that the danger of ischemic strokes and their contributors have similar or opposite effects on hemorrhagic strokes, we assessed the effects of the validated targets on ischemic strokes using two-sample MR analysis. The currently largest available dataset comes from a GWAS meta-analysis of stroke and stroke subtypes by the MEGASTROKE consortium, which included 446,696 European individuals and has been used in most MR analyses studying this disease [37]. We obtained data on any history of ischemic stroke (AIS), large artery stroke (LAS), small vessel stroke (SVS), or cardioembolic stroke (CES) for analysis. Data on lacunar stroke incidence were obtained from another meta-analysis that pooled previous GWASs and genes from newly recruited patients [38]. GWAS data for transient ischemic attack were obtained from Release 9 of the FinnGen consortium [39]. In addition, to investigate other potential benefits and harms of druggable genes, we also performed MR analysis of other CVDs and metabolic-related diseases, including white matter hyperintensities [40], atrial fibrillation (AF) [41], heart failure [42], chronic kidney disease [43], type-2 diabetes [44], and thoracic aortic aneurysm [45]. Some data were obtained from the UK Biobank consortium, e.g., data on venous thromboembolism (VTE), coronary atherosclerosis, major coronary heart disease event, nonrheumatic valve diseases, pulmonary embolism, peripheral artery disease [29], while other data, e.g., data on abdominal aortic aneurysm, dissection of the aorta, dissection of cerebral arteries, nonpyogenic thrombosis of the intracranial venous system, arterial embolism, and thrombosis, were obtained from the FinnGen consortium [39]. For the results of two-sample MR analysis, we considered P values < 0.05 to indicate statistical significance after FDR correction, and the parameters of analysis were the same as those listed above.
Results
MR Analysis and Colocalization Analysis
Since most of the drugs act through blood circulation, we first performed SMR analysis using whole-blood eQTLs to conduct an initial screening of the targets, with P-SMR < 0.05 as the criterion for positivity. The number of positive results detected in eQTLGen were IA (822), SAH (802), UIA (656), ICH (761), lobar ICH (655), and deep ICH (784), while in GTEx, it was IA (266), SAH (275), UIA (210), ICH (262), lobar ICH (219), and deep ICH (252) (Tables S3–S14).
Subsequently, we performed two-sample MR analysis of SMR-positive targets to increase the robustness of the results, with Pfdr < 0.05 indicating positivity, and each MR analysis method required that beta values have the same direction; otherwise, data were excluded. Finally, we identified a total of 432 positive results, including 124 IA genes (eQTLGen:16; GTEx:108), 123 SAH genes (eQTLGen:7; GTEx:116), 78 UIA genes (eQTLGen:0; GTEx:78), 91 ICH genes (eQTLGen:6; GTEx:85), 9 lobar ICH genes (eQTLGen:8; GTEx:1), and 7 deep ICH genes (eQTLGen:5; GTEx:2) (Tables S15–S20). All the results were tested by the MR‒Egger intercept test to exclude the presence of horizontal pleiotropy.
The 432 positive targets were subjected to colocalization analysis to clarify the likelihood of sharing causal variants. A total of 56 targets met the criteria for PPH4 > 0.75 or PPH3 + PPH4 > 0.80 (Table S21). Finally, we searched for these 56 targets in the DGIdb to obtain information on the targets, and the final results were obtained for the drug targets categorized as part of the druggable genome. After multiple screenings, we identified six druggable genes associated with hemorrhagic stroke (Fig. 2). Among those associated with IA were KL (P-SMR = 6.37 × 10−4; OR = 0.67; 95% CI, 0.55–0.82; Pfdr = 4.12 × 10−3; H3 + H4 = 0.99), RELT (P-SMR = 1.66 × 10−3; OR = 0.44; 95% CI, 0.28–0.69; Pfdr = 8.55 × 10−3; H4 = 0.86), and TNFSF12 (P-SMR = 7.87 × 10−3; OR = 1.30; 95% CI, 1.13–1.49; Pfdr = 8.33 × 10−3; H4 = 0.88); the targets of SAH were SLC22A4 (P-SMR = 2.83 × 10−3; OR = 1.40; 95% CI, 1.16–1.69; Pfdr = 0.035; H3 + H4 = 0.91) and TNFSF12 (P-SMR = 9.33 × 10−3; OR = 1.34; 95% CI, 1.14–1.57; Pfdr = 0.027; H4 = 0.84); and for ICH, the targets were SPARC (P-SMR = 5.85 × 10−4; OR = 2.95; 95% CI, 1.65–5.26; Pfdr = 0.014; H4 = 0.86) and ADORA3 (P-SMR = 1.61 × 10−3; OR = 0.57; 95% CI, 0.40–0.80; Pfdr = 0.045; H4 = 0.77); no significant identifiable druggable gene was associated with UIA, lobar ICH, or deep ICH (Table 1). We can conclude that TNFSF12, SLC22A4, and SPARC may induce IA, SAH, and ICH and that blocking these targets may have a protective effect against hemorrhagic stroke (Fig. 3).
Mediators of Validated Druggable Genes Affecting Hemorrhagic Strokes
In analyses exploring possible mechanisms of druggable gene-induced hemorrhagic strokes, MR analysis revealed that FDR-corrected hypertension, systolic blood pressure, diastolic blood pressure, BMI, body fat percentage, and smoking initiation were associated with IA and SAH, whereas none of the risk factors were significantly associated with ICH (Table 2). Then, we performed MVMR analysis of hypertension, systolic blood pressure, and diastolic blood pressure in patients with IA and SAH to clarify which factor was predominant. The multivariate IVW results showed that hypertension (OR = 27.842; 95% CI, 1.042–743.623; P = 4.72 × 10−2) and systolic blood pressure (OR = 2.423; 95% CI, 1.361–4.312; P = 2.63 × 10−3) still had significant effects on IA, and SAH was significantly correlated with only systolic blood pressure (OR = 2.933; 95% CI, 1.519–5.664; P = 1.35 × 10−3). Although diastolic blood pressure no longer had harmful effects after adjustment (IA: OR = 0.938; P = 0.795; SAH: OR = 0.832; P = 0.515), it was excluded.
For IA, when hypertension was used as a mediator, the effects of BMI (133.5%), body fat percentage (67.4%), and smoking initiation (10.6%) were mediated in part through hypertension; however, when systolic blood pressure was used as a mediator, there was no evidence that the effects of BMI, body fat percentage, or smoking initiation were mediated through systolic blood pressure. We can assume that systolic blood pressure, hypertension, body fat percentage, and smoking initiation have direct effects on IA. Moreover, for SAH patients, there was no evidence that hypertension, BMI, body fat percentage, or smoking initiation had an effect on systolic blood pressure; therefore, we believe that systolic blood pressure, hypertension, BMI, body fat percentage, and smoking initiation can directly influence SAH incidence.
Our two-sample MR analysis of druggable genes and exposures using the same parametric criteria showed that TNFSF12 was significantly associated with hypertension (OR = 1.006; 95% CI, 1.002–1.011; P = 6.90 × 10−3), which remained significant after FDR correction (IA: Pfdr = 1.38 × 10−2; SAH: Pfdr = 2.01 × 10−2). SLC22A4 was also causally associated with hypertension (OR = 1.01; 95% CI, 1.005–1.014; P = 3.46 × 10−5) and with SAH (Pfdr = 1.38 × 10−4) (Table 3).
Effects of Validated Druggable Genes on Other Diseases
To explore the potential side effects of these targets, we analyzed whether druggable genes influence ischemic stroke and found that TNFSF12, SLC22A4, and SPARC influence the occurrence of certain ischemic stroke subtypes. Moreover, TNFSF12 was shown to reduce the risk of CES (OR = 0.82; 95% CI, 0.73–0.94; Pfdr = 0.012), and SLC22A4 was shown to reduce the risk of SVS (OR = 0.81; 95% CI, 0.69–0.96; Pfdr = 0.036). Conversely, SPARC potentially increased the risk of AIS (OR = 1.14; 95% CI, 1.00–1.12; P = 0.042) and LAS (OR = 1.43; 95% CI, 1.06–1.91; P = 0.018), but these results were no longer significant after FDR correction. The various MR methods produced consistent effect estimates and no evidence of horizontal pleiotropy or heterogeneity.
Additionally, we analyzed other cardiovascular diseases (Table S24). Although none of the SNPs were included when TNFSF12 or AF was analyzed in two-sample MR, complementary analysis with SMR revealed that TNFSF12 was protective against AF (beta = − 0.16; P = 9.80 × 10−8), and colocalization analysis also yielded a positive result (H3 + H4 = 0.99). Furthermore, TNFSF12 weakly increased the risk of VTE (OR = 1.002; 95% CI, 1.001–1.004; Pfdr = 0.015). SLC22A4 acted as a risk factor for AF (OR = 1.08; 95% CI, 1.002–1.16; Pfdr = 0.043) and increased the risk of dissection of the aorta (OR = 1.58; 95% CI, 1.12–2.23; Pfdr = 0.029), but MR‒Egger (OR = 0.78) showed a different effect. Additionally, SPARC increased the risk of arterial embolism (OR = 1.64; 95% CI, 1.11–2.42; Pfdr = 0.038). No significant effects of druggable genes on other diseases were found.
Discussion
We performed a large-scale MR analysis between the GWAS data of hemorrhagic strokes and the target gene expression data (eQTL) to provide genetic support. Multiple screenings by combined MR analysis and colocalization analysis identified 56 genes associated with hemorrhagic stroke, six of which were druggable targets. Our study provides strong evidence that TNFSF12, SLC22A4, and SPARC may be potential risk factors for hemorrhagic stroke and that their induction of hypertension may be involved in potential pathogenesis. Inhibiting these targets offers potential to aid in preventing the development of hemorrhagic strokes. Furthermore, numerous potential side effects have been identified, such as blockingTNFSF12 possibly increasing the occurrence of both AF and CES while simultaneously lowering the risk of VTE; inhibiting SLC22A4 seemingly reducing the occurrence of AF but increasing the incidence of SVS; and inhibiting SPARC also having the potential to protect against arterial embolism and thrombosis, with no other unfavorable effects identified.
TNF superfamily member 12 (TNFSF12), also known as TNF-related weak inducer of apoptosis, is a member of the TNF superfamily that binds to its cognate receptor Fn14 and plays a role in regulating cell proliferation, differentiation, migration, apoptosis, inflammation, and angiogenesis [46]. Our study demonstrated that TNFSF12 can increase the risk of IA and SAH. MR analysis of drug targets for ischemic stroke revealed that TNFSF12 could promote SAH and ICH via the use of pQTLs to analyze blood proteins, and we obtained similar results using different IVs [47]. However, earlier investigations on the relationship of TNFSF12 with hemorrhagic stroke have been inadequate, with only one retrospective study suggesting that TNFSF12 increases hemorrhagic transformation in ischemic stroke patients; moreover, there are no data demonstrating the molecular mechanisms involved [48]. In an animal experiment, after treating mice with TNFSF12, it was shown that TNFSF12/Fn14 induced the upregulation of endothelin-converting enzyme-1 to produce more endothelin-1, and ultimately, the experimental group showed increased blood pressure and a dose‒response relationship [49]. Our study also proved that TNFSF12 can cause the effect of hypertension, while hypertension is a clear risk factor for IA, so we can assume that hypertension is an important mediator. TNFSF12/Fn14 act mainly through activation of the NF-κB pathway, which has been identified as a major factor regulating the induction of inflammatory genes in IA lesions. NF-κB activation recruits macrophages and produces proinflammatory cytokines and matrix metalloproteinases (MMPs) to break down the extracellular matrix, inducing apoptosis of vascular smooth muscle cells and degeneration of the internal elastic lamina, leading to loss of vascular wall integrity and aneurysm formation [50]. Thus, NF-κB may be an important pathway involved in this process. On the other hand, the MMP influences the development of IA through proteolytic disruption of the extracellular matrix of the vessel wall. Aoki et al. reported high expression of MMP-2 and MMP-9 within the wall of the IA [51], whereas Nuki et al. used gene knockout to determine that the incidence of IA decreased significantly in mice with MMP-9 knockout but not significantly in mice with MMP-2 knockout [52]. Another animal study demonstrated that TNFSF12/Fn14/NF-κB could increase MMP-9 activity but not MMP-2 activity [53]. Thus, MMP-9 could be a potential mediator. Finally, Fn14 is upregulated by a variety of cytokines in vessel wall damage [54], and TNFSF12/Fn14 induces the release of HMGB1 in macrophages via NF-κB, which can further activate NF-kB [55]. This vicious cycle may induce rupture of the IA, causing SAH. Similarly, the NF-κB signaling pathway mediates interactions between endothelial cells, platelets, and inflammatory responses, which disrupts coagulation-fibrinolysis homeostasis and induces deep vein thrombosis, and we hypothesize that NF-κB is also a mediator of TNFSFS12-induced VTE [56]. We also demonstrated that TNFSF12 reduces the risk of CES, whereas previous studies suggested that TNFSF12 can cause ischemic strokes [46]. The possible reason for these inconsistent results is the mediation of AF. Previous studies have demonstrated strong evidence that TNFSF12 is a potential drug target for AF and that genetic susceptibility of AF was an important etiological factor causing CES, and our study emphasizes the protective effect of TNFSF12 on AF [57]. Another reason could be the involvement of TNFSF12/CD163. Intracerebroventricular injection of soluble Fn14-Fc decoy receptor or anti-TNFSF12 antibody immediately after mouse cerebral infarction significantly reduced ischemic cerebral infarction volume and microglia activation and extent of apoptotic cell death in animal experiments [58], and the decoy receptor CD163 may block TNFSF12/Fn14 from acting and thus protect against CES. Currently, some drugs targeting TNFSF12 are undergoing preclinical experiments, such as ASP1531, L-534–0366, MRT-101, and TSCP-2. However, their primary investigational indications are mainly rheumatoid arthritis, gliomas, and lung cancers, and drugs with stroke as an indication still need to be further studied [25].
Solute carrier family 22 member 4 (SLC22A4), as well as organic cation transporter novel 1 (OTCN1), is an organic cation transporter protein. In a previous GWAS on IA, SLC22A4 was reported as a potential pathogenic gene at six unique loci [17]. Our MR analysis demonstrated that SLC22A4 can promote the developmental rupture of IAs, causing SAH. SLC22A4 is the protein that transports carnitine, choline, and levocarnitine and can generate trimethylamine (TMA) through the action of the gut microbiota. TMA is subsequently oxidized to trimethylamine-N-oxide (TMAO), and metabolomics studies have demonstrated that TMAO induces atherosclerosis and that elevated levels of TMAO can be used as a biomarker of high risk for CVD [59]. A meta-analysis that included 8 studies involving 11,750 individuals and 6176 patients with hypertension demonstrated that TMAO increased susceptibility to hypertension, with a 12% increased risk of hypertension in those with high TMAO concentrations [60]. Whereas hypertension, as an independent risk factor for SAH, may mediate the ability of SLC22A4 to cause SAH, a significant association between SLC22A4 and hypertension was also demonstrated in our study. Choline and its metabolites, such as betaine, can also increase the risk of developing CVD [59]. A nested case‒control study (n = 16,113) investigating the association between TMAO and stroke demonstrated a significant J-shaped association between TMAO and nonischemic stroke, but no significant associations were found for carnitine, choline, or betaine [61]. In an MR analysis of modifiable risk factors for IA, only positive correlations between IA and choline and betaine were demonstrated, but no association with TMAO was found [62]. We suggest that TMAO and its related metabolites are potential mediators of SLC22A4-related SAH, but further exploration is needed to determine which metabolites play a primary role in this process. While exploring the safety of the medication, we found that SLC22A4 can increase the risk of AF; the results of a study in which 2 prospective nested case‒control studies were conducted showed that higher choline and betaine levels were associated with an increased risk of AF [63], which supports the conclusion in our study and emphasizes the role that carnitine, choline, etc. play in SLC22A4 affecting the disease. Finally, inhibition of SLC22A4 may lead to SVS. SLC22A4, the main transporter protein of ergothioneine (ET), is responsible for uptake and reduces renal excretion, thus promoting the accumulation of ET in the body. SLC22A4 is a stable antioxidant and does not autocatalytically oxidize and generate oxygen radicals, thus protecting against chronic inflammation and reducing the risk of CVD [64]. Considering its antioxidative stress effect, SLC22A4 may reduce the incidence of SVS through this effect. Currently approved ustekinumab targeting TNFSF12 is mainly used in the treatment of inflammatory diseases such as psoriasis, Crohn’s disease, and ulcerative colitis. Whether its anti-inflammatory effect can be used for subarachnoid hemorrhage as an indication needs further experimentation [25]. In addition, drugs targeting OCTN have been used in cardiovascular and cerebrovascular diseases, and meldonium, as a carnitine analog, can inhibit the production of carnitine synthesis, which has achieved good results in the protection of ischemic stroke, and its effect on hemorrhagic stroke can be investigated in the future [65].
Secreted protein acidic and rich in cysteine (SPARC), which is also known as osteonectin/BM-40, is a matricellular glycoprotein that plays a key role in the regulation of cell–matrix interactions, tissue repair, and tissue calcification [66]. Our study identified SPARC as a risk factor for ICH and confirmed that SPARC has the potential to increase the risk of arterial embolism. There are more than ten drugs known to target SPARC, and the marketed drugs include tizanidine hydrochloride, baclofen, and NAB-PACLITAXEL [25]. Tizanidine hydrochloride and baclofen are mainly used in multiple sclerosis, but also for the relief of poststroke myotonia. Tizanidine hydrochloride and baclofen are mainly used in multiple sclerosis, but also for the relief of myasthenia gravis after stroke, and stroke remains an investigational indication for these drugs because inflammation caused by multiple sclerosis impairs the normal physiological functioning of vascular endothelial cells, which is strongly associated with hemorrhagic stroke [67]. PARC is closely associated with inflammatory responses and oxidative stress in tissues. The conversion of anti-inflammatory macrophages to proinflammatory macrophages can predispose patients to intravascular plaque formation and further recruit monocytes [68], and animal experiments have demonstrated that SPARC deficiency delays macrophage infiltration and proinflammatory cytokine expression [69]. SPARC also increased the apoptosis of vascular smooth muscle through the mitochondrial pathway while promoting the production of MMP, thus causing rupture of the internal elastic lamina of the vascular wall. Tan et al. also demonstrated that SPARC can upregulate the expression of NOX4 through TGF-β1, which contributes to the induction of an inflammatory phenotype transformation and apoptosis in smooth muscle cells, thereby worsening vascular wall damage [66, 70]. SPARC-mediated inflammation and vessel wall damage may be responsible for the development of ICH and arterial embolism and thrombosis. Another study collected cerebrospinal fluid specimens from 27 patients with cerebral amyloid angiopathy and reported that platelet-derived growth factor (PDGF)-BB was positively correlated with the titer of anti-Aβ antibody; moreover, PDGF-BB might be involved in smooth muscle cell repair and thus have a vasoprotective effect [71]. Raines et al. reported that SPARC could interact with PDGF-BB and inhibit its binding to the receptor to exert a vasoprotective effect and that its expression was more frequent in injured vessels than in control vessels [72], which might mediate the occurrence of ICH. Finally, SPARC is an adipocyte-secreted proinflammatory adipokine, and chronic inflammation in adipose tissue caused by its elevation is an important marker of obesity. SPARC also causes insulin resistance and is positively correlated with BMI, waist circumference, fasting blood glucose, and fasting insulin; therefore, SPARC may be a key participant in obesity and type-2 diabetes mellitus [73]. All of these factors are important in the pathogenesis of CVD. Our study concluded that these factors are not mediators of the effect of SPARC on ICH incidence.
In addition to the three druggable genes that may contribute to hemorrhagic stroke, we also identified three protective targets. Klotho (KL) is an antiaging gene expressed mainly in the kidney that stimulates the production of various antioxidant enzymes by increasing the activation of FoxO and signaling via the Nrf2 pathway and inhibits inflammatory pathways such as the TGF-β and NF-κB pathways [74], which may play a protective role against IA. A lack of KL leads to hyperphosphatemia and hypercalcemia. MR analysis also revealed a positive correlation between blood calcium levels and IA [75]. Wang et al. showed that KL gene deficiency leads to hyperphosphatemia by inducing KL in mice, after which hyperphosphate-induced or NF-κB-mediated MMP-2/9 can cause abdominal aortic aneurysm [76]. Considering that abdominal aortic aneurysm, an extracranial subtype of IA, has some similarities with the pathogenesis of IA, we hypothesized that KL deficiency may influence IA, which is consistent with our findings. The RELT TNF receptor (RELT) is a member of the TNFR superfamily that is expressed mainly in blood lymphoid tissues. This protein has proinflammatory effects and activates NF-κB, as well as apoptosis. There is a lack of correlation between RELT and cardiovascular disease, and the mechanism by which it protects against ICH is unknown. The adenosine A3 receptor (ADORA3) is an adenosine receptor that is broadly distributed in immunological and inflammatory cells and participates in the regulation of the central nervous, cardiovascular, peripheral, and immune systems as an important regulator of inflammatory responses. In addition to its anti-inflammatory effects, adenosine has been associated with vascular wall repair and protection, but it acts mainly through ADORA1/2, and it is unclear whether ADORA3 is involved. ADORA3 has been found to alleviate dysfunction caused by cerebral ischemia and brain injury; however, its effect on ICH has not been explored [77]. HU-010 targeting ADORA3 is undergoing a phase III clinical trial in the treatment of stroke as an indication [25].
Stroke is the second leading cause of death and the third leading cause of disability, and the global burden of stroke has increased dramatically, not only because of population growth and aging but also because of the dramatic increase in exposure to important risk factors, so prevention is always key [78]. Drugs that prevent hemorrhagic stroke, if available, may offer benefits for long-term protection and perioperative management of high-risk patients. Since the success rate of drugs development with gene-supported targets is twice as high as that of drugs without gene-supported targets, the primary objective of this study is to provide genetic evidence for drug development in hemorrhagic stroke in order to maximize drug efficacy, reduce drug toxicity and side effects, and reduce healthcare costs and to achieve precision medication and precision medicine. Our study provides the first comprehensive exploration of targets involved in SAH and ICH using eQTL data to provide genetic support for hemorrhagic stroke drug development and is the first to report that TNFSF12, SLC22A4, and SPARC have preventive effects on IA, SAH, and ICH. Despite the lack of an association between such targets and hemorrhagic strokes in previous studies, our study expands the horizon for future research. Most of the data used for the study came from the latest and largest GWAS. We performed two MR analyses and colocalization analyses to more clearly locate the targets, and only those targets with positive results after three screenings were included. The results were also corrected for FDR and tested for heterogeneity and horizontal pleiotropy to ensure the robustness of the results and to reduce the false-positive rate. Furthermore, the low repeatability of the eQTLGen data with the GTEx data, about 14%, and the presence of a 15% non-European population in GTEx may led to different results in different databases for these targets. We explored the two major databases to reduce the omission of valid targets while setting strict inclusion criteria [15, 16]. Meanwhile, eQTLGen has a larger sample size resulting in stronger evidence for the results relative to GTEx. We limited the targets to druggable genes, which can reduce the attrition rate associated with the development of drugs and improve the therapeutic efficacy for hemorrhagic stroke treatment. Our study also re-emphasizes the harmful role of hypertension. Considering the safety of the drug, we analyzed the targets associated with other CVD and showed that inhibiting targets that are harmful to hemorrhagic stroke patients may increase the risk of certain ischemic stroke subtypes and that attention should be given to the detection and the prevention of related side effects if these drugs are used in the clinic.
However, there are several limitations in our MR analysis. One of the key limitations is that MR analysis only provides evidence at the gene level, and the results usually indicate the effect of a relatively high drug dose on the disease over a short period of time, and therefore, the effect of the drug on the organism may be different in practical applications. At the same time, in order to ensure that the IVs were strongly correlated with exposure, we set the P threshold at 5 × 10−8 and the LD window was also very strict in order to ensure the independence of the IVs. As a result, many targets with only one SNP could only be subjected to the Wald ratio method and could not be tested for horizontal pleiotropy and heterogeneity, but the strict screening criteria resulted in high test power F and smaller variance R2 for the SNPs we analyzed (Table S25), indicating that our results are still robust and confident [79]. Furthermore, the data in this study were limited to European-ancestry individuals. Cross-ethnic genome-wide association studies have shown that many of the loci identified in European populations can be reproduced in non-European populations, suggesting broad cross-ethnic genetic similarities [80]. But the genetic unknowns of minority groups prevent the sharing of valid genetic information, and we cannot rashly deduce the generalizability of the findings to other groups and races. There is a need to expand cross-racial research to reduce health inequalities. And the reproducibility of the study makes the results more credible and more appropriate for generalization. We included only a small number of risk factors for CVD in our analysis, without in-depth studies on mediating mechanisms, which inevitably missed some possible potential associations. Finally, the exploration of adverse effects was limited to a small number of CVDs and was not validated for systemic diseases and symptoms. Because the effects of drugs on their targets are complex, many off-target effects cannot be explored by MR analysis, so further exploration through subsequent basic and clinical trials conducted in a comprehensive cross-ethnic manner is needed.
Conclusion
The large-scale MR analysis in the present study revealed the effects of six druggable genes on hemorrhagic stroke incidence. TNFSF12 promoted IA and SAH, SLC22A4 increased the risk of IA rupture leading to SAH, and SPARC was a risk factor for ICH, for which hypertension may be a potential mediator. In addition, certain ischemic stroke side effects that may result from the inhibition of these targets are also suggested. In conclusion, our study provides genetic evidence for drug targets, and further experiments are needed to assess the effectiveness and safety of these agents.
Data Availability
The genetic data used in this study are all previously publicly accessible GWAS datasets. The data analyzed and obtained in this study process are available in the supplementary data or in the database repositories provided in the original literature.
References
Unnithan AKA, M Das J, Mehta P (2023) Hemorrhagic stroke. In: StatPearls. StatPearls Publishing, Treasure Island. http://www.ncbi.nlm.nih.gov/books/NBK559173/
Sheth KN (2022) Spontaneous intracerebral hemorrhage. New Engl J Med 387:1589–1596. https://doi.org/10.1056/NEJMra2201449
Neifert SN, Chapman EK, Martini ML et al (2021) Aneurysmal subarachnoid hemorrhage: the last decade. Transl Stroke Res 12:428–446. https://doi.org/10.1007/s12975-020-00867-0
Lawton MT, Vates GE (2017) Subarachnoid hemorrhage. N Engl J Med 377:257–266. https://doi.org/10.1056/NEJMcp1605827
Hackam DG, Woodward M, Newby LK et al (2011) Statins and intracerebral hemorrhage: collaborative systematic review and meta-analysis. Circulation 124:2233–2242. https://doi.org/10.1161/CIRCULATIONAHA.111.055269
McKinney JS, Kostis WJ (2012) Statin therapy and the risk of intracerebral hemorrhage: a meta-analysis of 31 randomized controlled trials. Stroke 43:2149–2156. https://doi.org/10.1161/STROKEAHA.112.655894
Hudson JS, Nowicki KW, Lucke-Wold B et al (2023) Clopidogrel Is Associated with Reduced Likelihood of Aneurysmal Subarachnoid Hemorrhage: a Multi-Center Matched Retrospective Analysis. Transl Stroke Res. https://doi.org/10.1007/s12975-023-01179-9
Hasan DM, Mahaney KB, Brown RD et al (2011) Aspirin as a promising agent for decreasing incidence of cerebral aneurysm rupture. Stroke 42:3156–3162. https://doi.org/10.1161/STROKEAHA.111.619411
Jabbarli R, Darkwah Oppong M, Chihi M et al (2023) Regular medication as a risk factor for intracranial aneurysms: a comparative case-control study. Eur Stroke J 8:251–258. https://doi.org/10.1177/23969873221129080
Räisänen S, Huttunen J, Huuskonen TJ et al (2022) Use of antihypertensive medication and formation of de novo intracranial aneurysms. Eur J Neurol 29:2708–2715. https://doi.org/10.1111/ene.15430
Nelson MR, Tipney H, Painter JL et al (2015) The support of human genetic evidence for approved drug indications. Nat Genet 47:856–860. https://doi.org/10.1038/ng.3314
Gordillo-Marañón M, Zwierzyna M, Charoen P et al (2021) Validation of lipid-related therapeutic targets for coronary heart disease prevention using human genetics. Nat Commun 12:6120. https://doi.org/10.1038/s41467-021-25731-z
Schmidt AF, Finan C, Gordillo-Marañón M et al (2020) Genetic drug target validation using Mendelian randomisation. Nat Commun 11:3255. https://doi.org/10.1038/s41467-020-16969-0
Lawlor DA, Harbord RM, Sterne JAC et al (2008) Mendelian randomization: using genes as instruments for making causal inferences in epidemiology. Stat Med 27:1133–1163. https://doi.org/10.1002/sim.3034
Ardlie Kristin G, Barbeira Alvaro N, Battle Alexis et al (2020) The GTEx Consortium atlas of genetic regulatory effects across human tissues. Science (New York, NY) 369:1318–1330. https://doi.org/10.1126/science.aaz1776
Võsa U, Claringbould A, Westra H-J et al (2021) Large-scale cis- and trans-eQTL analyses identify thousands of genetic loci and polygenic scores that regulate blood gene expression. Nat Genet 53:1300–1310. https://doi.org/10.1038/s41588-021-00913-z
Bakker MK, van der Spek RAA, van Rheenen W et al (2020) Genome-wide association study of intracranial aneurysms identifies 17 risk loci and genetic overlap with clinical risk factors. Nat Genet 52:1303–1313. https://doi.org/10.1038/s41588-020-00725-7
Woo D, Falcone GJ, Devan WJ et al (2014) Meta-analysis of genome-wide association studies identifies 1q22 as a susceptibility locus for intracerebral hemorrhage. Am J Hum Genet 94:511–521. https://doi.org/10.1016/j.ajhg.2014.02.012
Zhu Z, Zhang F, Hu H et al (2016) Integration of summary data from GWAS and eQTL studies predicts complex trait gene targets. Nat Genet 48:481–487. https://doi.org/10.1038/ng.3538
Genomes Project Consortium, Abecasis GR, Auton A et al (2012) An integrated map of genetic variation from 1,092 human genomes. Nature 491:56–65. https://doi.org/10.1038/nature11632
Burgess S, Davey Smith G, Davies NM et al (2019) Guidelines for performing Mendelian randomization investigations: update for summer 2023. Wellcome Open Res 4:186. https://doi.org/10.12688/wellcomeopenres.15555.3
Bowden J, Davey Smith G, Burgess S (2015) Mendelian randomization with invalid instruments: effect estimation and bias detection through Egger regression. Int J Epidemiol 44:512–525. https://doi.org/10.1093/ije/dyv080
Giambartolomei C, Vukcevic D, Schadt EE et al (2014) Bayesian test for colocalisation between pairs of genetic association studies using summary statistics. PLoS Genet 10:e1004383. https://doi.org/10.1371/journal.pgen.1004383
Liu B, Gloudemans MJ, Rao AS et al (2019) Abundant associations with gene expression complicate GWAS follow-up. Nat Genet 51:768–769. https://doi.org/10.1038/s41588-019-0404-0
Freshour SL, Kiwala S, Cotto KC et al (2021) Integration of the Drug-Gene Interaction Database (DGIdb 4.0) with open crowdsource efforts. Nucleic Acids Res 49:D1144–D1151. https://doi.org/10.1093/nar/gkaa1084
Lim SS, Vos T, Flaxman AD et al (2012) A comparative risk assessment of burden of disease and injury attributable to 67 risk factors and risk factor clusters in 21 regions, 1990–2010: a systematic analysis for the Global Burden of Disease Study 2010. Lancet 380:2224–2260. https://doi.org/10.1016/S0140-6736(12)61766-8
Joseph JJ, Deedwania P, Acharya T et al (2022) Comprehensive Management of cardiovascular risk factors for adults with type 2 diabetes: a scientific statement from the American Heart Association. Circulation 145:e722–e759. https://doi.org/10.1161/CIR.0000000000001040
Soppert J, Lehrke M, Marx N et al (2020) Lipoproteins and lipids in cardiovascular disease: from mechanistic insights to therapeutic targeting. Adv Drug Deliv Rev 159:4–33. https://doi.org/10.1016/j.addr.2020.07.019
Liam Abbott, Sam Bryant et al (2018) UK Biobank — Neale lab. https://www.nealelab.is/uk-biobank. Accessed 1 Aug 2018
Surendran P, Feofanova EV, Lahrouchi N et al (2020) Discovery of rare variants associated with blood pressure regulation through meta-analysis of 1.3 million individuals. Nat Genet 52:1314–1332. https://doi.org/10.1038/s41588-020-00713-x
Saunders GRB, Wang X, Chen F et al (2022) Genetic diversity fuels gene discovery for tobacco and alcohol use. Nature 612:720–724. https://doi.org/10.1038/s41586-022-05477-4
Tg R, Gm L, Q W, et al (2022) Characterising metabolomic signatures of lipid-modifying therapies through drug target mendelian randomisation. PLoS biology 20(2): e3001547. https://doi.org/10.1371/journal.pbio.3001547
Sakaue S, Kanai M, Tanigawa Y et al (2021) A cross-population atlas of genetic associations for 220 human phenotypes. Nat Genet 53:1415–1424. https://doi.org/10.1038/s41588-021-00931-x
Shungin D, Winkler TW, Croteau-Chonka DC et al (2015) New genetic loci link adipose and insulin biology to body fat distribution. Nature 518:187–196. https://doi.org/10.1038/nature14132
Chen J, Spracklen CN, Marenne G et al (2021) The trans-ancestral genomic architecture of glycemic traits. Nat Genet 53:840–860. https://doi.org/10.1038/s41588-021-00852-9
Burgess S, Thompson SG (2015) Multivariable Mendelian randomization: the use of pleiotropic genetic variants to estimate causal effects. Am J Epidemiol 181:251–260. https://doi.org/10.1093/aje/kwu283
Malik R, Chauhan G, Traylor M et al (2018) Multiancestry genome-wide association study of 520,000 subjects identifies 32 loci associated with stroke and stroke subtypes. Nat Genet 50:524–537. https://doi.org/10.1038/s41588-018-0058-3
Traylor M, Persyn E, Tomppo L et al (2021) Genetic basis of lacunar stroke: a pooled analysis of individual patient data and genome-wide association studies. Lancet Neurol 20:351–361. https://doi.org/10.1016/S1474-4422(21)00031-4
Kurki MI, Karjalainen J, Palta P et al (2023) FinnGen provides genetic insights from a well-phenotyped isolated population. Nature 613:508–518. https://doi.org/10.1038/s41586-022-05473-8
Traylor M, Tozer DJ, Croall ID et al (2019) Genetic variation in PLEKHG1 is associated with white matter hyperintensities (n = 11,226). Neurology 92:e749–e757. https://doi.org/10.1212/WNL.0000000000006952
Roselli C, Chaffin MD, Weng L-C et al (2018) Multi-ethnic genome-wide association study for atrial fibrillation. Nat Genet 50:1225–1233. https://doi.org/10.1038/s41588-018-0133-9
Shah S, Henry A, Roselli C et al (2020) Genome-wide association and Mendelian randomisation analysis provide insights into the pathogenesis of heart failure. Nat Commun 11:163. https://doi.org/10.1038/s41467-019-13690-5
Wuttke M, Li Y, Li M et al (2019) A catalog of genetic loci associated with kidney function from analyses of a million individuals. Nat Genet 51:957–972. https://doi.org/10.1038/s41588-019-0407-x
Xue A, Wu Y, Zhu Z et al (2018) Genome-wide association analyses identify 143 risk variants and putative regulatory mechanisms for type 2 diabetes. Nat Commun 9:2941. https://doi.org/10.1038/s41467-018-04951-w
Roychowdhury T, Lu H, Hornsby WE et al (2021) Regulatory variants in TCF7L2 are associated with thoracic aortic aneurysm. Am J Hum Genet 108:1578–1589. https://doi.org/10.1016/j.ajhg.2021.06.016
Méndez-Barbero N, Gutiérrez-Muñoz C, Blázquez-Serra R et al (2020) Tumor Necrosis Factor-Like Weak Inducer of Apoptosis (TWEAK)/Fibroblast Growth Factor-Inducible 14 (Fn14) Axis in Cardiovascular Diseases: Progress and Challenges. Cells-basel 9:405. https://doi.org/10.3390/cells9020405
Chong M, Sjaarda J, Pigeyre M et al (2019) Novel drug targets for ischemic stroke identified through mendelian randomization analysis of the blood proteome. Circulation 140:819–830. https://doi.org/10.1161/CIRCULATIONAHA.119.040180
da Silva-Candal A, Pérez-Mato M, Rodríguez-Yáñez M et al (2020) The presence of leukoaraiosis enhances the association between sTWEAK and hemorrhagic transformation. Ann Clin Transl Neur 7:2103–2114. https://doi.org/10.1002/acn3.51171
Martínez-Miguel P, Medrano-Andrés D, Griera-Merino M et al (2017) Tweak up-regulates endothelin-1 system in mouse and human endothelial cells. Cardiovasc Res 113:207–221. https://doi.org/10.1093/cvr/cvw239
Pawlowska E, Szczepanska J, Wisniewski K et al (2018) NF-κB-mediated inflammation in the pathogenesis of intracranial aneurysm and subarachnoid hemorrhage. Does autophagy play a role? Int J Mol Sci 19(4):1245. https://doi.org/10.3390/ijms19041245
Aoki T, Kataoka H, Morimoto M et al (2007) Macrophage-derived matrix metalloproteinase-2 and -9 promote the progression of cerebral aneurysms in rats. Stroke 38:162–169. https://doi.org/10.1161/01.STR.0000252129.18605.c8
Nuki Y, Tsou T-L, Kurihara C et al (2009) Elastase-induced intracranial aneurysms in hypertensive mice. Hypertension 54:1337–1344. https://doi.org/10.1161/HYPERTENSIONAHA.109.138297
Sastre C, Fernández-Laso V, Madrigal-Matute J et al (2014) Genetic deletion or TWEAK blocking antibody administration reduce atherosclerosis and enhance plaque stability in mice. J Cell Mol Med 18:721–734. https://doi.org/10.1111/jcmm.12221
Muñoz-García B, Martín-Ventura JL, Martínez E et al (2006) Fn14 is upregulated in cytokine-stimulated vascular smooth muscle cells and is expressed in human carotid atherosclerotic plaques: modulation by atorvastatin. Stroke 37:2044–2053. https://doi.org/10.1161/01.STR.0000230648.00027.00
Moreno JA, Sastre C, Madrigal-Matute J et al (2013) HMGB1 expression and secretion are increased via TWEAK-Fn14 interaction in atherosclerotic plaques and cultured monocytes. Arterioscl Throm Vas 33:612–620. https://doi.org/10.1161/ATVBAHA.112.300874
Wang Z, Fang C, Yao M et al (2023) Research progress of NF-κB signaling pathway and thrombosis. Front Immunol 14:1257988. https://doi.org/10.3389/fimmu.2023.1257988
Wang Q, Richardson TG, Sanderson E et al (2022) A phenome-wide bidirectional Mendelian randomization analysis of atrial fibrillation. Int J Epidemiol 51:1153–1166. https://doi.org/10.1093/ije/dyac041
Yepes M, Brown SAN, Moore EG et al (2005) A soluble Fn14-Fc decoy receptor reduces infarct volume in a murine model of cerebral ischemia. Am J Pathol 166:511–520
Wang Z, Klipfell E, Bennett BJ et al (2011) Gut flora metabolism of phosphatidylcholine promotes cardiovascular disease. Nature 472:57–63. https://doi.org/10.1038/nature09922
Ge X, Zheng L, Zhuang R et al (2020) The gut microbial metabolite trimethylamine N-oxide and hypertension risk: a systematic review and dose–response meta-analysis. Adv Nutr 11:66–76. https://doi.org/10.1093/advances/nmz064
Liu D, Gu S, Zhou Z et al (2023) Associations of plasma TMAO and its precursors with stroke risk in the general population: a nested case-control study. J Intern Med 293:110–120. https://doi.org/10.1111/joim.13572
Tian D, Zhang L, Zhuang Z et al (2022) A two-sample Mendelian randomization analysis of modifiable risk factors and intracranial aneurysms. Sci Rep 12:7659. https://doi.org/10.1038/s41598-022-11720-9
Papandreou C, Bulló M, Hernández-Alonso P et al (2021) Choline metabolism and risk of atrial fibrillation and heart failure in the PREDIMED study. Clin Chem 67:288–297. https://doi.org/10.1093/clinchem/hvaa224
Cheah IK, Halliwell B (2021) Ergothioneine, recent developments. Redox Biol 42:101868. https://doi.org/10.1016/j.redox.2021.101868
Zhu Y, Zhang G, Zhao J et al (2013) Efficacy and safety of mildronate for acute ischemic stroke: a randomized, double-blind, active-controlled phase II multicenter trial. Clin Drug Investig 33:755–760. https://doi.org/10.1007/s40261-013-0121-x
Jiang S, Sun H-F, Li S et al (2023) SPARC: a potential target for functional nanomaterials and drugs. Front Mol Biosci 10:1235428. https://doi.org/10.3389/fmolb.2023.1235428
Hong Y, Tang HR, Ma M et al (2019) Multiple sclerosis and stroke: a systematic review and meta-analysis. BMC Neurol 19:139. https://doi.org/10.1186/s12883-019-1366-7
Ryu S, Sidorov S, Ravussin E et al (2022) The matricellular protein SPARC induces inflammatory interferon-response in macrophages during aging. Immunity S1074–7613(22):00340–00345. https://doi.org/10.1016/j.immuni.2022.07.007
Toba H, de Castro Brás LE, Baicu CF et al (2015) Secreted protein acidic and rich in cysteine facilitates age-related cardiac inflammation and macrophage M1 polarization. Am J Physiol Cell Physiol 308:C972-982. https://doi.org/10.1152/ajpcell.00402.2014
Tan X, Li T, Zhu S et al (2020) Induction of SPARC on oxidative stress, inflammatory phenotype transformation, and apoptosis of human brain smooth muscle cells via TGF-β1-NOX4 pathway. J Mol Neurosci 70:1728–1741. https://doi.org/10.1007/s12031-020-01566-z
Sakai K, Noguchi-Shinohara M, Ikeda T et al (2021) Cerebrospinal fluid cytokines and metalloproteinases in cerebral amyloid angiopathy-related inflammation. Acta Neurol Scand 143:450–457. https://doi.org/10.1111/ane.13382
Raines EW, Lane TF, Iruela-Arispe ML et al (1992) The extracellular glycoprotein SPARC interacts with platelet-derived growth factor (PDGF)-AB and -BB and inhibits the binding of PDGF to its receptors. Proc Natl Acad Sci U S A 89:1281–1285
Ren Y, Zhao H, Yin C et al (2022) Adipokines, hepatokines and myokines: focus on their role and molecular mechanisms in adipose tissue inflammation. Front Endocrinol 13:873699. https://doi.org/10.3389/fendo.2022.873699
Prud’homme GJ, Kurt M, Wang Q (2022) Pathobiology of the Klotho antiaging protein and therapeutic considerations. Front Aging 3:931331. https://doi.org/10.3389/fragi.2022.931331
Zhang Z, Yuan S, Larsson SC et al (2023) Serum calcium, 25-hydroxyvitamin D, and parathyroid hormone levels in relation to aneurysmal subarachnoid hemorrhage. Mol Neurobiol 60:3004–3009. https://doi.org/10.1007/s12035-023-03254-6
Wang Q, Wang S, Sun Z (2021) Kidney-specific Klotho gene deletion causes aortic aneurysm via hyperphosphatemia. Hypertension 78:308–319. https://doi.org/10.1161/HYPERTENSIONAHA.121.17299
Borea PA, Gessi S, Merighi S et al (2018) Pharmacology of adenosine receptors: the state of the art. Physiol Rev 98:1591–1625. https://doi.org/10.1152/physrev.00049.2017
Valery L Feigin, Benjamin A Stark, Catherine Owens Johnson et al (2021) Global, regional, and national burden of stroke and its risk factors, 1990–2019: a systematic analysis for the Global Burden of Disease Study 2019. Lancet Neurol 20:795–820. https://doi.org/10.1016/S1474-4422(21)00252-0
Burgess S, Thompson SG, CRPCHD Genetics Collaboration (2011) Avoiding bias from weak instruments in Mendelian randomization studies. Int J Epidemiol 40:755–764. https://doi.org/10.1093/ije/dyr036
Lu H, Zhang S, Jiang Z, Zeng P (2023) Leveraging trans-ethnic genetic risk scores to improve association power for complex traits in underrepresented populations. Brief Bioinform 24:232. https://doi.org/10.1093/bib/bbad232
Acknowledgements
The authors would like to thank the International Stroke Genetics Consortium (ISGC). GWAS summary statistics data for intracranial aneurysms, intracerebral hemorrhage, and white matter hyperintensities were accessed through the ISGC Cerebrovascular Disease Knowledge Portal. The authors would like to thank the MEGASTROKE consortia for providing summary statistics data for ischemic stroke and thank all the studies that provided publicly available GWAS summary data. We gratefully acknowledge the participants and investigators of the UK Biobank and FinnGen studies and gratefully acknowledge the IEU Open GWAS Project for providing access to the GWAS summary-level data analyzed in this research. The MEGASTROKE project received funding from sources specified at megastroke.org/acknowledgements.html. The investigator list of MEGASTROKE is available at megastroke.org/authors.html. Data related to genetic diversity fuels gene discovery for tobacco and alcohol use (retrieved from the Data Repository for the University of Minnesota, https://doi.org/10.13020/przg-dp88).
Funding
This study was supported by Guangzhou Science and Technology Key Research and Development Program (No. 202206010130).
Author information
Authors and Affiliations
Contributions
Guang-Zhong Chen and Yong Yang contributed to the study conception and design, analysis of the results, and manuscript’s writing guidance and revision. Lun-Zhe Yang contributed to the study conception, data collection, code manipulation of Mendelian randomization, analysis of the results, manuscript’s writing, and creation of figures and tables. Chuan Hong, Qi-Zhe Wu, Xiong-Jie Shi, and Yi-Lin Liu provided data collection and contributed to manuscript’s revision and creation of figures and tables. All authors read and commented on previous versions of the manuscript and approved the final manuscript.
Corresponding author
Ethics declarations
Ethics Approval
This study was a reanalysis of previously published data and did not require ethical consent. Informed consent and ethical approval for the study of data were obtained from all participants according to the protocol of the original GWAS data.
Consent to Participate
Not applicable.
Competing Interests
The authors declare no competing interests.
Additional information
Publisher's Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Yong Yang, Chuan Hong, Qi-Zhe Wu, Xiong-Jie Shi, and Yi-Lin Liu are co-second authors.
Supplementary Information
Below is the link to the electronic supplementary material.
Rights and permissions
Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/.
About this article
Cite this article
Yang, LZ., Yang, Y., Hong, C. et al. Systematic Mendelian Randomization Exploring Druggable Genes for Hemorrhagic Strokes. Mol Neurobiol (2024). https://doi.org/10.1007/s12035-024-04336-9
Received:
Accepted:
Published:
DOI: https://doi.org/10.1007/s12035-024-04336-9