The causal relationship between extensive perivascular space burden and ischemic stroke and its subtypes and transient ischemic attack: A Mendelian randomization study

doi:10.21203/rs.3.rs-4498156/v1

Download PDF

Research Article

The causal relationship between extensive perivascular space burden and ischemic stroke and its subtypes and transient ischemic attack: A Mendelian randomization study

https://doi.org/10.21203/rs.3.rs-4498156/v1

This work is licensed under a CC BY 4.0 License

Version 1

posted

You are reading this latest preprint version

Background

Clinical studies suggest a strong link between extensive perivascular space (EPVS) and ischemic stroke (IS), including its subtypes, and transient ischemic attack (TIA), but it's uncertain if the relationship is genetically causal.

Methods

We utilize summary data from large-scale Genome-wide Association Studies (GWAS) to investigate the association between EPVS in different locations and IS, its subtypes, and TIA through Mendelian randomization (MR) analysis. Various MR methods are employed to assess the causal relationship between EPVS and IS, its subtypes, and TIA. We apply multivariable MR to mitigate potential confounding factors and conduct sensitivity analyses to enhance result robustness. Subsequently, meta-analysis is utilized to integrate causal relationships between EPVS in different locations and IS from various sources. Additionally, reverse MR is employed to observe the impact of various IS types on EPVS. Finally, linkage disequilibrium score regression is conducted to assess genetic correlations between exposures and outcomes.

Results

EPVS burden in both the white matter (OR, 1.12; 95% CI, 1.01–1.25; P = 0.04) and the basal ganglia (OR, 1.57; 95% CI, 1.30–1.89; P < 0.01) are significant risk factors for IS. EPVS burden in the basal ganglia is also a risk for IS (small-vessel) (OR, 4.56; 95% CI, 2.57–8.27; P = 5.95E-07). Additionally, there appears to be a potential increase in extensive basal ganglia perivascular space burden following IS and TIA.

Conclusion

Extensive white matter perivascular space burden and extensive basal ganglia perivascular space burden may serve as important indicators for predicting IS.

extensive perivascular space burden

ischemic stroke

causal inference

Mendelian randomization

Stroke is a prevalent cerebrovascular disease marked by neurological deficits due to acute focal injury in the central nervous system. It is classified into hemorrhagic and ischemic types based on pathophysiology. Ischemic stroke (IS), caused by arterial blockage and decreased cerebral blood flow, is the most common form. Recently, IS incidence has been on the rise, making it a top cause of global mortality and disability, and imposing a significant socio-economic burden.¹

Perivascular spaces (PVS) are small gaps around small arteries and veins within the brain parenchyma, enclosed by the pia mater. They are thought to be essential for lymphatic drainage, waste removal, and tissue homeostasis.² Increasing evidence suggests that extensive perivascular space burden (EPVS) indicates impaired lymphatic drainage function, which is a characteristic of brain disorders, including small vessel disease,³ Parkinson's disease,⁴ cognitive impairment,⁵ and multiple sclerosis.⁶ Recent research indicates that after ischemia, cerebrospinal fluid surrounding brain tissue enters the brain via PVS within minutes, causing cerebral edema.⁷ This highlights the potential impact of EPVS in IS. Approximately 98.8% of acute IS patients reportedly exhibit observable EPVS on scans within the first 7 days post-stroke.⁸ A prospective study involving high-risk individuals for IS or transient ischemic attack (TIA) (n = 2002 subjects) found that a high burden of PVS in the basal ganglia is linked to recurrent stroke and IS.⁹ Although traditional epidemiological studies are well-designed, prospective, and involve large population sizes, the conclusions drawn from these studies may be influenced by confounding factors or reverse causality, thus precluding the establishment of definitive causal relationships.

Mendelian randomization (MR) analysis provides an alternative method for causal inference. It utilizes genetic variations as instrumental variables (IVs), which are closely linked to the exposure, unaffected by confounding factors, and not influenced by pathways other than the exposure. This effectively reduces reverse causal bias and residual confounding bias,¹⁰ resulting in stronger evidence for causal inference. In this study, we utilize MR analysis to explore the causal relationship between EPVS in different locations and IS, its subtypes and TIA, thus offering valuable insights for clinical diagnosis and treatment.

2.1 Study design and ethical statement

To explore the relationship between EPVS in various locations and IS, its subtypes, and TIA, we conduct MR analysis. Figure 1 outlines the study design. In the forward MR, we consider extensive white matter, hippocampal, and basal ganglia perivascular space burdens as exposures, investigating their causal links with IS, its subtypes and TIA individually. Additionally, we also utilize multivariable MR (MVMR) to adjust for confounding factors, followed by conducting a meta-analysis to evaluate the overall effect of EPVS on IS, its subtypes and TIA from different sources. The MR analysis in this study meets three core assumptions: 1) significant correlation between IVs and exposures,¹¹ 2) no correlation between IVs and confounding factors affecting the relationship between exposures and outcomes,¹¹ and 3) IVs solely influencing the outcomes through exposures.¹²

Ethical approval is not necessary for this study as it utilizes publicly available data that has already been approved by the relevant institutional ethics committees. Moreover, it is reported according to the Strengthening the Reporting of Observational Studies in Epidemiology Using Mendelian Randomization guidelines (STROBE-MR) (Supplementary Table S1).¹⁰

2.2 Data sources

The summary data of EPVS in various sites are sourced from a cohort study involving 18 populations, encompassing over 8 million SNPs (minor allele frequency ≥ 1%) from more than 40,095 participants (mean age 66.3 ± 8.6 years, 51.7% female). To address variations in PVS quantification methods, image acquisition, and participant characteristics, we categorized PVS burden using thresholds closest to the upper quartile of the PVS distribution. Ultimately, 9,607 out of 39,822, 9,189 out of 40,000, and 9,339 out of 40,095 participants exhibit EPVS in white matter, hippocampus, and basal ganglia, respectively.¹³

The data of IS, its subtypes and TIA are sourced from multiple publicly available Genome-wide Association Studies (GWAS) datasets. Specifically, IS data are derived from the MEGA consortium (Ncase = 34,217, sample size = 440,328),¹⁴ a meta-analysis involving the UK Biobank (Ncase = 11,929, sample size = 484,121),¹⁵ and the FinnGen database (Ncase = 10,551, sample size = 212,774). Based on the TOAST classification of IS, partial GWAS information for its subtypes, including large-artery atherosclerosis (Ncase = 4,373, sample size = 150,765), cardioembolism (Ncase = 7,193, sample size = 211,763), and small-vessel (Ncase = 5,386, sample size = 198,048), is obtained from the MEGA consortium.¹⁴ Lacunar stroke (Ncase = 6,030, sample size = 225,419) is sourced from cases recruited from acute stroke hospitalization and outpatient services in Europe, the United States, South America, and Australia. This study involves a meta-analysis of MRI-diagnosed lacunar stroke patients' data and existing GWAS datasets. The patients are from hospitals in the UK, part of the UK DNA Cavernous Stroke Study and collaborators from the International Stroke Genetics Consortium.¹⁶ GWAS data of TIA are obtained from the FinnGen database and the UK Biobank,¹⁷ both of which conducted GWAS studies on large populations to identify risk loci for the disease.

2.3 Selection of IVs

During the forward MR analysis, EPVS found in various regions are considered as the exposures, while IS, its subtypes, and TIA are examined as outcomes. To meet assumption 1, this study identifies single nucleotide polymorphisms (SNPs) across the entire genome that show significant associations with EPVS at various locations (P < 1×10^− 5) and have no linkage disequilibrium (LD) (r²=0.01, kb = 5000), ensuring the independence of the selected IVs. To address potential confounding factors, we utilize MVMR analysis to control for common confounders of IS and TIA, including obesity, hypertension, diabetes, and alcohol, thereby satisfying assumption 2. To fulfill assumption 3, this study further excludes SNPs significantly associated with IS, its subtypes and TIA across the entire genome (P < 1×10^− 5). To ensure the strength of the selected IVs, we calculate the statistical strength using the F value. Specifically, F = R²/ (1 - R²) * (N - K − 1)/K, where N represents the sample size of the exposure, K is the number of SNPs, R² is the proportion of variance explained by SNPs in the exposure dataset, and R² = 2× (1-MAF) (MAF) ×(β/SD)², β denotes the effect size of the allele.¹⁸ IVs with F < 10 will be excluded. Furthermore, SNPs that are inconsistent with the exposure and outcome alleles, as well as palindromic SNPs with moderate allele frequencies, are excluded. The SNPs subjected to the rigorous screening process are utilized for the final causal analysis.

2.4 MR analysis

This study employs inverse variance weighted (IVW) as the primary method for MR analysis. When the selected SNPs are all effective IVs, the IVW method can provide the most accurate estimates of causal association effects.¹⁹ Additionally, Bayesian weighted, weighted median (WM), weighted mode, and simple mode are used as supplementary analyses. Bayesian weighted Mendelian randomization explicitly accounts for uncertainty related to weak effects from polygenic traits and can identify outliers, addressing instrumental variable assumption violations due to pleiotropy.²⁰ The WM method provides effective causal estimates when over half of the SNPs are valid IVs.²¹ Weighted mode is reliable when most individual instruments' causal effect estimates come from valid instruments, even if some IVs are considered invalid.²² Furthermore, the simple mode can serve as an unweighted empirical density function for estimating causality.²³ To enhance the robustness of results, we require consistent directions of β values across all methods while ensuring significance in IVW and Bayesian weighted results. Moreover, we use false discovery rate (FDR) correction for P-values. Significant causal relationships between EPVS and outcomes are indicated when P < 0.05 and P_FDR < 0.05, and potential causal relationships are suggested when P < 0.05 and P_FDR > 0.05.

2.5 Sensitivity analysis

We use IVW and MR Egger regression to detect heterogeneity and calculate Cochran’s Q statistic to quantify its magnitude. P < 0.05 indicates significant heterogeneity, warranting the use of a random-effects model for causal inference.²⁴ MR-Egger intercept test is utilized to analyze horizontal pleiotropy, estimating directional inference by calculating the intercept and resulting in a directional P-value. P > 0.05 suggests the absence of horizontal pleiotropy, demonstrating the robustness of the MR analysis results.²⁵ The MR-PRESSO Global test identifies outliers, whose presence is confirmed by P < 0.05, requiring their exclusion for subsequent analysis.²⁶ Besides, leave-one-out analysis assesses individual SNPs' influence on the MR results. After removing outlier SNPs, P < 0.05 in the MR Egger regression renders the MR results unreliable. In forward MR analysis, we utilize MR Steiger to ensure directional accuracy. This method assumes that the genetic variants should explain more variance during exposure than outcome, meeting the legitimate requirements of MR investigation and aiding in identifying potential bidirectional effects.²⁷ Finally, reverse MR analysis is used to observe bidirectional effects between EPVS and IS, its subtypes, and TIA, with SNP selection criteria consistent with forward MR.

2.6 MVMR analysis

MVMR analysis can evaluate direct causal effects between exposure and outcome.²⁸ Thus, to adjust for potential confounders (Obesity, hypertension, type 2 diabetes, and ongoing alcohol addiction), we perform MVMR analysis following univariable MR (UVMR) analysis to examine the independent impact of EPVS in different locations on IS, its subtypes, and TIA. We utilize Multivariable IVW, Multivariable Egger, and Multivariable Median methods, with Multivariable IVW serving as the primary method. Also, to assess result stability, Cochran’s Q statistic and I² detect result heterogeneity, while the Egger-intercept test identifies horizontal pleiotropy.

2.7 Meta- analysis

To mitigate biases stemming from various sources of GWAS data on IS, following MR analysis, we conduct meta-analysis to examine the overall impact of EPVS in different locations on IS and its subtypes, as well as TIA. Additionally, we employ I2 to assess the heterogeneity of the findings.

2.8 Linkage disequilibrium score regression and directionality tests

We employ linkage disequilibrium score regression (LDSC) analysis to summarize GWAS data and estimate heritability and genetic correlations based on single-nucleotide variants. The LD reference panel from the 1000 Genomes Project is used to compute LD scores. Finally, we utilize the LDSC tool to further evaluate the genetic associations between EPVS at different locations and IS and its subtypes.²⁹

All MR-related analyses are conducted in R (version 4.3.0) using the "TwoSampleMR",²³ "MR-PRESSO",²⁶ and "Mendelian Randomization" R packages.

3.1 Results of causality between EPVS and outcomes by UVMR

After rigorous IV selection, we establish a variable number of IVs for EPVS in different locations based on various outcomes. Detailed information on all IVs can be found in Supplementary Table S2. The F-values for all SNPs were greater than 10, indicating that the current results are not biased by weak IVs. Genetically predicted extensive white matter perivascular space burden is significantly associated with a higher risk of IS (IEU database) (OR, 1.24; 95%CI, 1.09–1.42; P = 1.09E-03; P_FDR = 9.82E-03). Considering extensive hippocampal perivascular space burden as the exposure reveals a potential causal link with IS (large-artery atherosclerosis) (OR, 1.95; 95%CI, 1.02–3.74; P = 4.50E-02; P_FDR=3.77E-01). Moreover, extensive basal ganglia perivascular space burden exhibits significant causal associations with IS and its subtypes. Specifically, it notably increases the risk of IS (IEU database) (OR, 1.53; 95%CI, 1.20–1.94; P = 5.37E-04; P_FDR = 1.91E-02), IS (FinnGen) (OR, 1.63; 95%CI, 1.17–2.29; P = 4.23E-03; P_FDR = 3.80E-02), and IS (small vessel) (OR, 2.66; 95%CI, 1.49–4.75; P = 9.46E-04; P_FDR = 1.91E-02) (Supplementary Table S3, Fig. 2 and Fig. 3).

3.2 Results of sensitivity analysis for UVMR

To ensure the reliability of our findings, we conduct several sensitivity analyses. When extensive hippocampal perivascular space burden is considered as the exposure and IS (small-vessel) as the outcome, Cochran’s Q test reveals heterogeneity, prompting the use of the random-effects model of IVW despite the absence of outliers. For IS (MEGA STROKE) as the outcome, MR-PRESSO identifies three outliers. After their removal, no heterogeneity or horizontal pleiotropy is detected. Similarly, when extensive white matter perivascular space burden is the exposure and IS (MEGA STROKE) or IS (IEU database) is the outcome, we find two outliers in each case. After excluding them, the results remain robust. Additionally, with extensive basal ganglia perivascular space burden as the exposure and IS (MEGA STROKE) or Lacunar stroke as the outcome, two outliers are detected in each analysis. Removing these outliers enhances the robustness of all sensitivity analyses (Supplementary Tables S4 and S5). Furthermore, leave-one-out analysis confirms that individual SNPs do not drive the current results (Fig. 4).

3.3 Results of Meta-analysis after UVMR

To mitigate biases arising from varied GWAS data sources on IS, we perform meta-analyses after UVMR to synthesize the findings regarding EPVS in different locations on IS from diverse sources (Fig. 5). The results indicate that when extensive white matter perivascular space burden is considered as the exposure and IS from various sources as the outcomes, no discernible causal relationship is observed, albeit with moderate heterogeneity (P = 0.07, I² = 63%). Conversely, when extensive basal ganglia perivascular space burden is considered as the exposure and IS from different sources as the outcome, a significant causal relationship is detected, with no observed heterogeneity in the current causal association (P = 0.43, I² = 0%).

3.4 Results of MVMR analysis between EPVS and outcomes

To assess the direct causal link between EPVS in different locations and IS and its subtypes, we perform MVMR analyses. In these analyses, we adjust for confounders such as obesity, hypertension, type 2 diabetes, and ongoing alcohol addiction. The results are shown in Supplementary Table S6 and Fig. 6. After accounting for confounding factors, the outcomes of UVMR change. Notably, when extensive white matter perivascular space burden is considered as the exposure and IS (IEU database) as the outcome, the causal relationship remains significant (OR, 1.23; 95%CI, 1.05–1.43; P = 9.65E-03). However, when extensive hippocampal perivascular space burden is the exposure, the causal link with IS (large artery atherosclerosis) becomes non-significant (OR, 1.55; 95%CI, 0.75–3.18; P = 2.36E-01). Conversely, when extensive basal ganglia perivascular space burden is the exposure, a causal relationship is established with IS (MEGA STROKE) (OR, 1.32; 95%CI, 1.01–1.71; P = 4.11E-02). Moreover, the causal associations between extensive basal ganglia perivascular space burden and IS (IEU database) (OR, 1.76; 95%CI, 1.37–2.25; P = 8.51E-06), IS (FinnGen) (OR, 1.68; 95%CI, 1.21–2.34; P = 1.97E-03), and IS (small-vessel) (OR, 4.56; 95%CI, 2.51–8.27; P = 5.95E-07) persist (Table 1).

Table 1

Multivariable mendelian randomization analysis results between perivascular space burden and ischemic stroke by adjusting for all confounders
Type of Perivascular space measurement / Exposure	Data source	Type of ischemic stroke	nSNP	Methods of multivariable MR	Beta	SE	P-value	OR (95%CI)
Extensive white matter perivascular space burden	MEGA STROKE	Ischemic stroke	193	Multivariable IVW	0.045	0.086	6.06E-01	1.05 (0.88–1.24)
				Multivariable Median	-0.049	0.127	6.96E-01	0.95 (0.74–1.22)
				Multivariable Egger	-0.016	0.119	8.92E-01	0.98 (0.78–1.24)
	IEU database	Ischemic stroke	193	Multivariable IVW	0.204	0.079	9.65E-03	1.23 (1.05–1.43)
				Multivariable Median	0.146	0.109	1.79E-01	1.16 (0.94–1.43)
				Multivariable Egger	0.295	0.105	5.16E-03	1.34 (1.09–1.65)
	FinnGen	Ischemic stroke	195	Multivariable IVW	0.068	0.109	5.29E-01	1.07 (0.87–1.32)
				Multivariable Median	0.241	0.157	1.25E-01	1.27 (0.94–1.73)
				Multivariable Egger	0.244	0.144	9.15E-02	1.28 (0.96–1.69)
Extensive hippocampal perivascular space burden	MEGA STROKE	Ischemic stroke (large artery atherosclerosis)	163	Multivariable IVW	0.436	0.368	2.36E-01	1.55 (0.75–3.18)
				Multivariable Median	0.953	0.476	4.53E-02	2.59 (1.02–6.60)
				Multivariable Egger	1.001	0.472	3.42E-02	2.72 (1.08–6.87)
Extensive basal ganglia perivascular space burden	MEGA STROKE	Ischemic stroke	165	Multivariable IVW	0.275	0.134	4.11E-02	1.32 (1.01–1.71)
				Multivariable Median	0.534	0.183	3.55E-03	1.71 (1.19–2.44)
				Multivariable Egger	0.156	0.178	3.79E-01	1.17 (0.83–1.66)
	IEU database	Ischemic stroke	168	Multivariable IVW	0.563	0.126	8.51E-06	1.76 (1.37–2.25)
				Multivariable Median	0.455	0.170	7.29E-03	1.58 (1.13–2.20)
				Multivariable Egger	0.650	0.173	1.76E-04	1.92 (1.36–2.69)
	FinnGen	Ischemic stroke	168	Multivariable IVW	0.520	0.168	1.97E-03	1.68 (1.21–2.34)
				Multivariable Median	0.422	0.253	9.49E-02	1.52 (0.93–2.50)
				Multivariable Egger	0.515	0.221	1.96E-02	1.67 (1.09–2.58)
	MEGA STROKE	Ischemic stroke (small-vessel)	167	Multivariable IVW	1.517	0.304	5.95E-07	4.56 (2.51–8.27)
				Multivariable Median	1.547	0.440	4.38E-04	4.70 (1.98–1.13)
				Multivariable Egger	1.202	0.404	2.94E-03	3.33 (1.51–7.35)
MR, mendelian randomization; IVW, inverse variance weighted.

Meanwhile, we conduct sensitivity analyses to affirm the stability of the results, employing Cochran's Q test and I² to detect heterogeneity. We find no significant heterogeneity in any of the results. Additionally, no instances of horizontal pleiotropy are identified (Supplementary Table S7).

3.5 Results of Meta-analysis after MVMR

Following adjustment for relevant confounding factors in the MVMR analysis, we conducted a meta-analysis to synthesize the results of EPVS in different locations on IS from various sources (Fig. 7). The findings reveal a significant causal relationship between extensive white matter perivascular space burden as the exposure and IS from different sources as the outcome, without any observed heterogeneity (P = 0.35, I² = 6%). Additionally, when considering extensive basal ganglia perivascular space burden as the exposure and IS from different sources as the outcome, the causal relationship remains significant, with no observed heterogeneity in the current causal association (P = 0.26, I² = 25%).

3.6 Reverse causal relationship between IS, TIA and EPVS

In the reverse MR analysis, IS and TIA are treated as exposures, while EPVS in different locations are considered as outcomes. Through comprehensive sensitivity analyses, outliers are removed to ensure result stability, and causal associations affected by horizontal pleiotropy are eliminated. Ultimately, we find potential causal relationships between extensive basal ganglia perivascular space burden and IS (FinnGen) (β, 0.021; 95%CI, 0.002–0.041; P = 3.46E-02, P_FDR = 1.01E-01) and TIA (FinnGen) (β, 0.016; 95%CI, 0.000-0.032; P = 4.50E-02, P_FDR =1.01E-01) (Supplementary Tables S8 and S9). Furthermore, there is no evidence of heterogeneity or horizontal pleiotropy in these results. Leave-one-out analysis confirms that the observed causal effects are not driven by individual SNPs (Supplementary Tables S10, S11 and Fig. 8).

3.7 Genetic correlations between EPVS and IS and its subtypes

As indicated in Supplementary Table S12, in addition to a comprehensive MR analysis, we also utilize LDSC regression to evaluate the genetic correlations between EPVS in different locations and IS and its subtypes. We observe genetic correlations between extensive basal ganglia perivascular space burden and IS from various sources. Specifically, extensive basal ganglia perivascular space burden shows positive genetic correlations with IS (MEGA STROKE) (Rg = 0.504, P = 1.43E-05), IS (IEU database) (Rg = 0.390, P = 1.97E-03), IS (FinnGen) (Rg = 0.616, P = 8.05E-03), and IS (small-vessel) (Rg = 0.877, P = 4.96E-03).

This study thoroughly investigates the relationship between EPVS in various locations and IS, its subtypes, and TIA using MR analysis. After adjusting for confounders and integrating data from multiple sources, we confirm that EPVS in the white matter and basal ganglia region increases the risk of IS. Moreover, EPVS in the basal ganglia region is associated with the occurrence of IS (small-vessel subtype). Reverse MR results suggest a potential increase in extensive basal ganglia perivascular space burden following the occurrence of IS and TIA. Therefore, monitoring EPVS in these regions could be valuable for preventing, monitoring, and treating IS in clinical practice.

PVSs, also known as Virchow-Robin spaces, are potential gaps between the brain's vascular system and the meningeal layers, initially described by pathologists Rudolf Virchow and Charles Robin in the 19th century.³⁰ These spaces create pathways along small arteries and veins in the subarachnoid space, facilitating the transport of cerebrospinal fluid and the exchange of substances within cells to effectively clear metabolic byproducts from the brain.³¹ IS is the most common type of stroke, characterized by high incidence, mortality, and disability rates.³² Clinical risk factors for IS are categorized into controllable and uncontrollable types. Uncontrollable risk factors include age, gender, and race, while controllable factors encompass hypertension, diabetes, hyperlipidemia, smoking, alcohol abuse, carotid atherosclerosis, heart disease, and blood disorders. However, even with effective management of known risk factors, many patients still experience worsening or recurrence of ischemia. Therefore, there remains a need to identify and control additional unknown risk factors.

Recent research on IS and EPVS has sparked debate. Some studies suggest that EPVS serves as a risk factor for IS, with varying relationships observed between EPVS in different regions and IS.⁹ Conversely, other research poses that EPVS is merely an MRI manifestation of cerebral small vessel disease and lacks significant association with IS onset.³³ This ongoing debate highlights the controversial nature of the EPVS-IS relationship. Moreover, several studies indicate that risk factors for cerebrovascular disease, such as hypertension,³⁴ diabetes,³⁵ age,³⁶ and gender,³⁷ may also influence EPVS. Consequently, EPVS might be linked to the occurrence of IS.

Recent research suggests that atherosclerosis in the carotid artery system may disrupt pulsatile blood flow dynamics and impede interstitial fluid drainage in the brain, leading to EPVS.³⁸ In IS patients, over 50% of intracranial atherosclerotic stenosis is independently associated with more than 20 centrum semiovale EPVS, but not with basal ganglia EPVS.⁹ Another study indicates that atherosclerosis in the conduit arteries between the heart and the brain may correlate with a higher burden of EPVS, suggesting that EPVS could serve as a marker of systemic arterial aging and its impact on pulsatile hemodynamics.³⁹

Given that atherosclerosis is a key pathogenic mechanism of IS,⁴⁰ we speculate whether EPVS may contribute to IS by facilitating the occurrence and progression of atherosclerosis. Further longitudinal studies are essential to elucidate the relationship between EPVS burden and luminal narrowing progression. In our study, we investigated the genetic association between EPVS burden in various locations and the onset of IS and its subtypes, revealing a clear causal link between EPVS burden in the white matter and basal ganglia regions and IS occurrence. Moreover, our findings suggest a potential increase in EPVS burden in the basal ganglia region following IS and TIA, underscoring the need for further exploration of underlying mechanisms.

It is essential to recognize several limitations in our study. Firstly, setting the genome-wide significance threshold at 1×10 − 5 may introduce bias. Secondly, our study samples were exclusively from European ancestry populations, limiting the generalizability of our findings to other ethnic groups. Lastly, due to dataset constraints, we couldn't assess the individual-level impact of EPVS burden on IS. Therefore, prospective cohort studies are still necessary. Nevertheless, we rigorously evaluated the causal relationship between EPVS burden in various locations and IS, its subtypes, and TIA through comprehensive MR analysis, providing a foundation for further clinical diagnosis and treatment.

This MR analysis confirms that an increased EPVS burden in the white matter and basal ganglia regions significantly raises the risk of IS. Moreover, elevated EPVS burden in the basal ganglia region also contributes to the occurrence of IS (small-vessel). Therefore, clinical monitoring of EPVS burden is essential. However, future research should delve deeper into the associations and potential mechanisms involved.

Declaration of competing interest

The authors declare that they have no any competing interests.

Acknowledgment

We would like to thank the Duperron MG et.al, MEGA STRIOKE consortium, FinnGen consortium, the IEU open GWAS project, and the UK biobank for providing complete GWAS data.

Data availability statement

The original GWAS data for perivascular space burden can be found at the GWAS CatLog (https://www.ebi.ac.uk/gwas/studies/GCST90134433). The GWAS data for ischemic stroke and its subtype can be found at https://www.finngen.fi/en, https://gwas.mrcieu.ac.uk/, and http://www.megastroke.org/index.html.

Funding

This work was supported by the National Natural Science Foundation of China (82071468 and 82271507).

CRediT author contributions

Xuehong Chu: Conceptualization, Methodology, Software, Formal analysis, Writing - Original Draft. Yingjie Shen: Conceptualization, Data Curation, Software and Writing - Review & Editing. Yaolou Wang: Investigation, Data Curation, and Visualization. Xiao Dong: Software, Resources, and Visualization. Yuanyuan Liu: Conceptualization, Validation, and Resources. Chuanhui Li: Software and Writing - Review & Editing. Wenbo Zhao: Validation, and Formal analysis. Xunming Ji: Supervision, Project administration, and Funding acquisition. Miaowen Jiang: Writing - Review & Editing, Investigation, Resources. Software, and Visualization. Ming Li: Writing – review & editing, Software and Visualization, Funding acquisition, Resources, Supervision. Chuanjie Wu: Conceptualization, Writing - Review & Editing, Supervision, Project administration, and Funding acquisition. All participating authors give their consent for this work to be published.

Ethics approval

No additional ethical approvals were required for this study, as we used publicly available data that had been approved by the relevant ethical and institutional review boards.

Supporting information

Additional information can be found in the Supplemental Tables (1-12).

Ding Y, DeGracia D, Geng X, et al. Perspectives on effect of spleen in ischemic stroke. Brain Circ. 2022;8:117–20. 10.4103/bc.bc_53_22.
Rasmussen MK, Mestre H, Nedergaard M. Fluid transport in the brain. Physiol Rev. 2022;102:1025–151. 10.1152/physrev.00031.2020.
Brown R, Benveniste H, Black SE, et al. Understanding the role of the perivascular space in cerebral small vessel disease. Cardiovasc Res. 2018;114:1462–73. 10.1093/cvr/cvy113.
Sundaram S, Hughes RL, Peterson E, et al. Establishing a framework for neuropathological correlates and glymphatic system functioning in Parkinson's disease. Neurosci Biobehav Rev. 2019;103:305–31520190524. 10.1016/j.neubiorev.2019.05.016.
Nation DA, Sweeney MD, Montagne A, et al. Blood-brain barrier breakdown is an early biomarker of human cognitive dysfunction. Nat Med. 2019;25(20190114):270–6. 10.1038/s41591-018-0297-y.
Cavallari M, Egorova S, Healy BC, et al. Evaluating the Association between Enlarged Perivascular Spaces and Disease Worsening in Multiple Sclerosis. J Neuroimaging. 2018;28:273–7. 10.1111/jon.12490.
Mestre H, Du T, Sweeney AM et al. Cerebrospinal fluid influx drives acute ischemic tissue swelling. Science 2020; 367 20200130. 10.1126/science.aax7171.
Liang Y, Chan YL, Deng M, et al. Enlarged perivascular spaces in the centrum semiovale are associated with poststroke depression: A 3-month prospective study. J Affect Disord. 2018;228:166–17220171122. 10.1016/j.jad.2017.11.080.
Lau KK, Li L, Lovelock CE, et al. Clinical Correlates, Ethnic Differences, and Prognostic Implications of Perivascular Spaces in Transient Ischemic Attack and Ischemic Stroke. Stroke. 2017;48:1470–7. 10.1161/strokeaha.117.016694.
Skrivankova VW, Richmond RC, Woolf BAR, et al. Strengthening the Reporting of Observational Studies in Epidemiology Using Mendelian Randomization: The STROBE-MR Statement. JAMA. 2021;326:1614–21. 10.1001/jama.2021.18236.
Smith GD, Ebrahim S. Mendelian randomization': can genetic epidemiology contribute to understanding environmental determinants of disease? Int J Epidemiol. 2003;32:1–22. 10.1093/ije/dyg070.
Emdin CA, Khera AV, Kathiresan S, Mendelian Randomization. JAMA. 2017;318:1925–6. 10.1001/jama.2017.17219.
Duperron MG, Knol MJ, Le Grand Q, et al. Genomics of perivascular space burden unravels early mechanisms of cerebral small vessel disease. Nat Med. 2023;29:950–96220230417. 10.1038/s41591-023-02268-w.
Malik R, Chauhan G, Traylor M, et al. Multiancestry genome-wide association study of 520,000 subjects identifies 32 loci associated with stroke and stroke subtypes. Nat Genet. 2018;50(20180312):524–37. 10.1038/s41588-018-0058-3.
Sakaue S, Kanai M, Tanigawa Y, et al. A cross-population atlas of genetic associations for 220 human phenotypes. Nat Genet. 2021;53(20210930):1415–24. 10.1038/s41588-021-00931-x.
Traylor M, Persyn E, Tomppo L, et al. Genetic basis of lacunar stroke: a pooled analysis of individual patient data and genome-wide association studies. Lancet Neurol. 2021;20:351–61. 10.1016/s1474-4422(21)00031-4.
Sudlow C, Gallacher J, Allen N, et al. UK biobank: an open access resource for identifying the causes of a wide range of complex diseases of middle and old age. PLoS Med. 2015;12:e1001779. 10.1371/journal.pmed.1001779.
Burgess S, Thompson SG. Bias in causal estimates from Mendelian randomization studies with weak instruments. Stat Med. 2011;30(20110322):1312–23. 10.1002/sim.4197.
Chen X, Kong J, Diao X, et al. Depression and prostate cancer risk: A Mendelian randomization study. Cancer Med. 2020;9(20201007):9160–7. 10.1002/cam4.3493.
Zhao J, Ming J, Hu X, et al. Bayesian weighted Mendelian randomization for causal inference based on summary statistics. Bioinformatics. 2020;36:1501–8. 10.1093/bioinformatics/btz749.
Bowden J, Davey Smith G, Haycock PC, et al. Consistent Estimation in Mendelian Randomization with Some Invalid Instruments Using a Weighted Median Estimator. Genet Epidemiol. 2016;40:304–14. 10.1002/gepi.21965.
Hartwig FP, Davey Smith G, Bowden J. Robust inference in summary data Mendelian randomization via the zero modal pleiotropy assumption. Int J Epidemiol. 2017;46:1985–98. 10.1093/ije/dyx102.
Hemani G, Zheng J, Elsworth B, et al. The MR-Base platform supports systematic causal inference across the human phenome. Elife. 2018;7:20180530. 10.7554/eLife.34408.
Greco MF, Minelli C, Sheehan NA, et al. Detecting pleiotropy in Mendelian randomisation studies with summary data and a continuous outcome. Stat Med. 2015;34:2926–40. 10.1002/sim.6522.
Bowden J, Davey Smith G, Burgess S. Mendelian randomization with invalid instruments: effect estimation and bias detection through Egger regression. Int J Epidemiol. 2015;44:512–25. 10.1093/ije/dyv080.
Verbanck M, Chen CY, Neale B, et al. Detection of widespread horizontal pleiotropy in causal relationships inferred from Mendelian randomization between complex traits and diseases. Nat Genet. 2018;50(20180423):693–8. 10.1038/s41588-018-0099-7.
Hemani G, Tilling K, Davey Smith G. Orienting the causal relationship between imprecisely measured traits using GWAS summary data. PLoS Genet. 2017;13:e1007081. 10.1371/journal.pgen.1007081.
Sanderson E. Multivariable Mendelian Randomization and Mediation. Cold Spring Harb Perspect Med. 2021;11:20210201. 10.1101/cshperspect.a038984.
Bulik-Sullivan BK, Loh PR, Finucane HK, et al. LD Score regression distinguishes confounding from polygenicity in genome-wide association studies. Nat Genet. 2015;47:291–5. 10.1038/ng.3211.
Jessen NA, Munk AS, Lundgaard I, et al. The Glymphatic System: A Beginner's Guide. Neurochem Res. 2015;40:2583–259920150507. 10.1007/s11064-015-1581-6.
Wardlaw JM, Benveniste H, Nedergaard M, et al. Perivascular spaces in the brain: anatomy, physiology and pathology. Nat Rev Neurol. 2020;16(20200224):137–53. 10.1038/s41582-020-0312-z.
Tuo QZ, Zhang ST, Lei P. Mechanisms of neuronal cell death in ischemic stroke and their therapeutic implications. Med Res Rev 2022; 42: 259–305. 20210506. 10.1002/med.21817.
Doubal FN, MacLullich AM, Ferguson KJ, et al. Enlarged perivascular spaces on MRI are a feature of cerebral small vessel disease. Stroke. 2010;41:450–45420100107. 10.1161/strokeaha.109.564914.
Zhu YC, Tzourio C, Soumaré A, et al. Severity of dilated Virchow-Robin spaces is associated with age, blood pressure, and MRI markers of small vessel disease: a population-based study. Stroke. 2010;41:2483–90. 10.1161/strokeaha.110.591586.
Ferguson SC, Blane A, Perros P, et al. Cognitive ability and brain structure in type 1 diabetes: relation to microangiopathy and preceding severe hypoglycemia. Diabetes. 2003;52:149–56. 10.2337/diabetes.52.1.149.
Zhu YC, Dufouil C, Mazoyer B, et al. Frequency and location of dilated Virchow-Robin spaces in elderly people: a population-based 3D MR imaging study. AJNR Am J Neuroradiol. 2011;32:709–13. 10.3174/ajnr.A2366.
Patankar TF, Mitra D, Varma A, et al. Dilatation of the Virchow-Robin space is a sensitive indicator of cerebral microvascular disease: study in elderly patients with dementia. AJNR Am J Neuroradiol. 2005;26:1512–20.
Mikami T, Tamada T, Suzuki H, et al. Influence of hemodynamics on enlarged perivascular spaces in atherosclerotic large vessel disease. Neurol Res. 2018;40:1021–102720180829. 10.1080/01616412.2018.1509827.
Gutierrez J, DiTullio M, YK KC, et al. Brain arterial dilatation modifies the association between extracranial pulsatile hemodynamics and brain perivascular spaces: the Northern Manhattan Study. Hypertens Res. 2019;42:1019–102820190401. 10.1038/s41440-019-0255-1.
Bir SC, Kelley RE. Carotid atherosclerotic disease: A systematic review of pathogenesis and management. Brain Circ 2022; 8: 127–136. 20220921. 10.4103/bc.bc_36_22.

No competing interests reported.

SupplementaryTables.xlsx
Supplemental Tables (1-12).

Download PDF

Version 1

posted

You are reading this latest preprint version

The causal relationship between extensive perivascular space burden and ischemic stroke and its subtypes and transient ischemic attack: A Mendelian randomization study

Status:

Version 1

Abstract

Background

Methods

Results

Conclusion

Figures

1 Introduction

2 Materials and Methods

2.1 Study design and ethical statement

2.2 Data sources

2.3 Selection of IVs

2.4 MR analysis

2.5 Sensitivity analysis

2.6 MVMR analysis

2.7 Meta- analysis

2.8 Linkage disequilibrium score regression and directionality tests

3 Results

3.1 Results of causality between EPVS and outcomes by UVMR

3.2 Results of sensitivity analysis for UVMR

3.3 Results of Meta-analysis after UVMR

3.4 Results of MVMR analysis between EPVS and outcomes

3.5 Results of Meta-analysis after MVMR

3.6 Reverse causal relationship between IS, TIA and EPVS

3.7 Genetic correlations between EPVS and IS and its subtypes

4 Discussion

5 Conclusion

Declarations

References

Additional Declarations

Supplementary Files

Status:

Version 1