Skip to main content
NCBI home page
Medicine logoLink to Medicine
. 2025 Dec 12;104(50):e45824. doi: 10.1097/MD.0000000000045824

Causal effect of gut microbiota and its metabolites on cerebral small vessel disease: A multivariable Mendelian randomization analysis

Hao Xue a, Jing Zang a, Fang Yang a, Lei Zhang a, Xiaoli Yang a, Miaomiao Zhang a, Xiuping Zhang a, Lin Wu a,*
PMCID: PMC12708115  PMID: 41398819

Abstract

Emerging evidence suggests a strong relationship between cerebral small vessel disease (CSVD), gut microbiota (GM), and their secreted metabolites. The major subtypes of CSVD include lacunar stroke, cerebral microbleeds, perivascular spaces (PVSs), and white matter hyperintensity (WMH) volume. However, their causal relationship remains unknown. A 2-sample Mendelian randomization (MR) analysis was undertaken on summarized data from genome-wide association studies. Univariable MR and multivariable MR were used to find the possible causal link between CSVD, GM, and their metabolites. The causal association was determined through the inverse variance weighted, weighted median, maximum likelihood, MR-Egger regression, and MR-PRESSO approaches. Multiple sensitivity tests were employed to evaluate the presence of horizontal pleiotropy and heterogeneity. A reverse MR analysis determined the potential for reverse causality. Finally, enrichment analyses were carried out to elucidate relevant biological functions. Overall, 1505 host single-nucleotide polymorphisms related to 119 GM features and 1873 related to 81 GM-derived metabolites (GMDMs) were identified as exposure variables. Two intestinal bacteria were linked with raised likelihood of WMH volume, 1 with cerebral microbleeds, 2 with lacunar stroke, and 7 with PVSs. However, 3 bacterial species were correlated with decreased WMH volume risk, 2 to lacunar stroke, and 8 to PVSs. Moreover, the possible causal link between 28 GMDMs and CSVD was observed. Enrichment analysis revealed the crucial involvement of multiple key regulatory pathways. This study provided new insights into gut biomarkers and specialized prevention and therapy for CSVD by demonstrating evidence of the causal link between GM, GMDMs, and CSVD.

Keywords: causal association, cerebral small vessel disease, gut microbiota, gut microbiota-derived metabolites, Mendelian randomization

1. Introduction

Cerebral small vessel disease (CSVD) is among the most prevalent age-associated neurological disorders observed by neurologists in medical settings and epidemiologists in population studies, more common than clinical stroke, impacting nearly 80% of individuals over the age of 60.[1,2] The term CSVD includes all conditions marked by transient or prolonged impairment of the brain’s small blood vessels (i.e., small arteries, small veins, and capillaries) resulting in ischemia- or hemorrhage-related persistent cerebral hypoperfusion and dysfunction of subcortical and cortical regions dependent on the affected vessels.[3] With the aging population, the prevalence of CSVD continues to rise annually, posing a major risk to public health.[4] However, the precise mechanisms underlying CSVD and effective treatment options for its prevention and management remain poorly defined.

The gut microbiota (GM) represent a diverse microbial ecosystem within the intestine. Current evidence demonstrated that GM and their secreted metabolites have a crucial role in cerebrovascular diseases via the microbiota–gut–brain axis (MGBA).[5] For example, notoginsenoside R1 alleviates cerebral ischemia/reperfusion injury by inhibiting the TLR4/MyD88/NF-kappaB signaling pathway through MGBA.[6] Multistrain probiotics with fructooligosaccharides improve middle cerebral artery occlusion-driven neurological deficits by revamping MGBA.[7] However, current studies predominantly focus on cerebral macrovascular disorders, with relatively few studies exploring the association between GM and CSVD. A previous study demonstrated that GM from arteriosclerotic CSVD patients stimulates elevated IL-17A secretion in neutrophils by triggering the modulator of reprogramming, isoform gamma-t (RORγt).[8] A previous study demonstrated a strong association between GM and inflammation, the integrity of the blood–brain barrier, and various imaging parameters in CSVD patients.[9] Another previous study identified microbiota diversity and structure alterations among individuals with CSVD, which were also linked to neuroimaging indicators and cognitive performance.[10] However, research exploring the correlations between GM and multiple neuroimaging forms of CSVD remains limited. Moreover, most findings are derived from observational studies, complicating the determination of the temporal correlation between exposure and predicted outcome. Moreover, the intricate nature of the intestine and cerebral vasculature presents challenges in assessing and controlling confounding variables in GM–CSVD-related investigations. Furthermore, Mendelian randomization (MR) analysis incorporates genome-wide association studies (GWAS) data, utilizing genetic variants as instrumental variables (IVs) to determine causal links.[11]

The MR method incorporates 3 assumptions: IVs must show strong links with the exposure; IVs must be independent of confounding variables influencing both the exposure and the outcome; IVs must affect the outcome exclusively through the exposure. Due to its strengths in reducing confounding and minimizing reverse causality, MR analysis observed causal correlations between GM, their secreted metabolites, and complex human illnesses, that is, neurodegenerative disorders[12] and autoimmune hypothyroidism.[13] The present study used comprehensive GWAS datasets of GM, their secreted metabolites, and CSVD phenotypes to systematically examine the potential causal link between GM, secreted metabolites, and CSVD phenotypes.

2. Materials and methods

2.1. Data sources

The GM data analyzed as exposure variables were sourced from a GWAS involving 18,340 participants of varying ethnicities.[14] This study integrated 16S rRNA gene sequencing data with genotyping information to conduct a GWAS, investigating the possible link between host genetic variants and GM composition. Sequencing data was classified at the genus and higher taxonomic levels via the SILVA reference database.[15]

Further, data on GM-derived metabolites (GMDMs) were retrieved from the largest metabolomics GWAS to date,[16] which included individuals of European descent from the TwinsUK and KORA cohorts (N = 7824). This GWAS examined correlations between single-nucleotide polymorphism (SNPs) and 486 human metabolites across both cohorts. After that, the HMDB database[17] was manually searched to identify human GM-secreted metabolites.

The outcome datasets were obtained from the latest GWAS corresponding to all CSVD phenotypes. Specifically, these included lacunar stroke (248,929 controls and 6030 cases),[18] white matter hyperintensity (WMH) volume (N = 22,138),[19] cerebral microbleeds (CMBs) (N = 25,862),[20] and perivascular spaces (PVSs), comprising WM PVSs (N = 38,598), basal ganglia PVSs (N = 38,903), and hippocampal PVSs (N = 38,871).[21] The selected GWAS datasets were predominantly composed of individuals of European ancestry. All datasets were available publicly, demonstrating minimal sample interaction, and were conducted as per ethical approvals from the original studies.

2.2. Selection of IVs

Multiple quality control procedures were performed to select appropriate IVs. The IVs were chosen from SNPs linked to GM and their secreted metabolites at the level of P < 1 × 10−5[22] The PLINK clumping method was used to select lead SNPs, applying r² < 0.001 as linkage disequilibrium within a window of 10,000 kb. Thereafter, the IVs were evaluated via the F-statistic, ensuring values exceeded the cutoff of 10 to reduce the risk of weak instrument bias.[23] The formula utilized was R²(n − k − 1)/k(1 − R²), where n, k, and R² depict the sample size, the IV number, and the percentage of variance defined by the IVs, respectively.[24]

2.3. Statistical analyses

A 2-sample MR analysis was used to observe the possible causal correlation between GM, GMDMs, and CSVD. The inverse variance weighted (IVW) method integrates Wald ratios for individual IVs using an intercept of 0. Without horizontal pleiotropy was utilized for analysis to provide unbiased estimates.[25] Causal effect estimates are shown as odds ratio (ORs) with 95% confidence intervals (CIs) for binary outcomes.[11] Multiple hypotheses were tested using false discovery rates (FDRs), with statistical significance defined as FDR-adjusted P-values < .05. These correlations yielding P-values < .05 but not exceeding the FDR-adjusted threshold were interpreted as indicative of a potential relationship. Further sensitivity analyses were undertaken to verify the consistency of the findings and included maximum likelihood estimations,[26] a weighted median approach,[27] MR-Egger regression,[28] and MR-PRESSO[29] methodologies. A leave-one-out analysis was utilized to examine the effects of all SNPs on the causal estimates, while Cochran Q statistic examined heterogeneity. Moreover, MR-Egger and MR-PRESSO analyses were utilized to examine possible horizontal pleiotropy between IVs and the outcome.[30]

The MVMR analysis incorporated extensive GM and GMDMs identified from the univariable analysis to determine independent microbial taxa and metabolites. A reverse MR analysis was carried out to evaluate a possible reverse causality between CSVD and both GM and GMDMs. The study used the same datasets and analytical parameters as applied in the forward MR, with the roles of exposure and outcome variables reversed.

To further investigate the biological functions of GM and GMDMs in CSVD, GO and KEGG analyses were performed as per the lead SNPs associated with all identified GM and GMDMs. The lead SNPs corresponding to the causal microbial taxa and metabolites related to CSVD were annotated to their nearest genes. The schematic diagram of the study design is illustrated in Figure 1. All analyses were undertaken via “TwoSampleMR,” “MRPRESSO,” and “MVMR” in R software. Enrichment analyses were conducted through the web-based platform “Metascape.”[31]

Figure 1.

Figure 1.

Open in a new tab

Study flowchart. The entire workflow of MR analysis. CSVD = cerebral small vessel disease; GWAS = genome-wide association studies; MR = Mendelian randomization; SNP = single-nucleotide polymorphism.

3. Results

Approximately 1505 leading SNPs correlated with 119 bacteria were selected for the GM analysis, whereas 1873 leading SNPs linked to 81 traits were identified for GMDMs. All IVs demonstrated F-statistic values > 10, ensuring sufficient instrument strength and indicating a low possibility of bias arising from weak instruments in the analysis. Comprehensive details of the IVs used are provided in Tables S1 to S4, Supplemental Digital Content, https://links.lww.com/MD/Q702.

3.1. GM and CSVD correlations

The IVW analysis indicated that elevated genetically predicted abundance of Turicibacter (OR: 1.117; 95% CI: 1.027–1.215; P = .010) and Lachnospiraceae UCG-008 (OR: 1.078; 95% CI: 1.009–1.153; P = .026) were linked with elevated WMH volume risk. However, higher genetically predicted abundances of Fusicatenibacter (OR: 0.907; 95% CI: 0.832–0.989; P = .027), Eisenbergiella (OR: 0.909; 95% CI: 0.842–0.983; P = .016), and Alistipes (OR: 0.889; 95% CI: 0.804–0.984; P = .023) were linked to a reduced likelihood of WMH volume. The IVW estimations showing higher genetically predicted Bifidobacterium (OR: 1.305; 95% CI: 1.030–1.653; P = .027) was linked with a raised CMB risk. The IVW predictions of higher Streptococcus (OR: 1.304; 95% CI: 1.069–1.590; P = .009) and Lachnospiraceae UCG010 (OR: 1.296; 95% CI: 1.038–1.618; P = .022) were linked to increased chances of lacunar stroke, while those of Lachnospiraceae UCG001 (OR: 0.794; 95% CI: 0.675–0.935; P = .006), and Fusicatenibacter (OR: 0.817; 95% CI: 0.675–0.989; P = .038) were associated with a lower risk of lacunar stroke (Fig. 2, Table 1).

Figure 2.

Figure 2.

Open in a new tab

Associations of genetically predicted gut microbiota with risk of CSVD using the IVW method. The colored block represents the point estimation of OR and the horizontal black line represents the 95% CI of OR. BG PVSs = basal ganglia perivascular spaces; CI = confidence interval; CMBs = cerebral microbleeds; CSVD = cerebral small vessel disease; hippocampal PVSs = hippocampal perivascular spaces; IVW = inverse variance weighted; OR = odds ratio; WMH volume = white matter hyperintensity volume; WM PVSs = white matter perivascular spaces.

Table 1.

MR analyses of gut microbiota on CSVD by IVW method.

Exposure Outcome OR (95% CI) P P FDR
Alistipes WMH volume 0.889 (0.804–0.984) .023 .058
Eisenbergiella WMH volume 0.909 (0.842–0.983) .016 .077
Fusicatenibacter WMH volume 0.907 (0.832–0.989) .027 .055
Lachnospiraceae UCG008 WMH volume 1.078 (1.009–1.153) .026 .057
Turicibacter WMH volume 1.117 (1.027–1.215) .010 .076
Bifidobacterium CMBs 1.305 (1.030–1.653) .027 .053
Streptococcus Lacunar stroke 1.304 (1.069–1.590) .009 .119
Lachnospiraceae UCG010 Lacunar stroke 1.296 (1.038–1.618) .022 .058
Lachnospiraceae UCG001 Lacunar stroke 0.794 (0.675–0.935) .006 .318
Fusicatenibacter Lacunar stroke 0.817 (0.675–0.989) .038 .053
Lactococcus BG PVSs 0.980 (0.961–0.999) .042 .049
Lachnospiraceae UCG008 BG PVSs 0.972 (0.945–0.999) .046 .051
Dorea BG PVSs 0.961 (0.925–0.998) .038 .052
Bilophila BG PVSs 0.967 (0.939–0.995) .023 .055
Bacteroides BG PVSs 0.949 (0.908–0.992) .020 .066
Ruminococcaceae UCG005 Hippocampal PVSs 1.032 (1.001–1.064) .040 .049
Holdemania Hippocampal PVSs 1.031 (1.008–1.055) .009 .095
Haemophilus Hippocampal PVSs 0.974 (0.951–0.999) .039 .049
Clostridium sensu stricto 1 Hippocampal PVSs 0.956 (0.922–0.992) .018 .073
Victivallis WM PVSs 1.021 (1.003–1.040) .020 .062
Ruminococcaceae UCG004 WM PVSs 1.037 (1.005–1.070) .021 .059
Ruminococcaceae NK4A214 group WM PVSs 1.039 (1.006–1.072) .020 .059
Eubacterium fissicatena group WM PVSs 1.025 (1.001–1.050) .038 .050
Clostridium sensu stricto 1 WM PVSs 0.957 (0.922–0.992) .017 .075
Anaerotruncus WM PVSs 1.041 (1.005–1.079) .026 .055
Open in a new tab

BG PVSs = basal ganglia perivascular spaces, CI = confidence interval, CMBs = cerebral microbleeds, CSVD = cerebral small vessel disease, Hippocampal PVSs = hippocampal perivascular spaces, IVW = inverse variance weighted, MR = Mendelian randomization, OR = odds ratio, WM PVSs = white matter perivascular spaces, WMH volume = white matter hyperintensity volume.

A possible causal link between GM taxa and PVSs was further assessed using the same analytical approach. The IVW results demonstrated that higher genetically predicted levels of Lactococcus (OR: 0.980; 95% CI: 0.961–0.999; P = .042), Lachnospiraceae UCG-008 (OR: 0.972; 95% CI: 0.945–0.999; P = .046), Dorea (OR: 0.961; 95% CI: 0.925–0.998; P = .038), Bilophila (OR: 0.967; 95% CI: 0.939–0.995; P = .023), and Bacteroides (OR: 0.949; 95% CI: 0.908–0.992; P = .020) were substantially correlated to a reduced risk of basal ganglia (BG) PVSs. The IVW estimations depicting an elevated genetically predicted Ruminococcaceae UCG005 (OR: 1.032; 95% CI: 1.001–1.064; P = .040) and Holdemania (OR: 1.031; 95% CI: 1.008–1.055; P = .009) were correlated with a greater likelihood of hippocampal PVSs, while Haemophilus (OR: 0.974; 95% CI: 0.951–0.999; P = .039), and Clostridium sensu stricto 1 (OR: 0.956; 95% CI: 0.922–0.992; P = .018) were linked to a lower chance of hippocampal PVSs. The IVW estimations depicting elevated predictions of Victivallis (OR: 1.021; 95% CI: 1.003–1.040; P = .020), Ruminococcaceae UCG004 (OR: 1.037; 95% CI: 1.005–1.070; P = .021), Ruminococcaceae NK4A214 group (OR: 1.039; 95% CI: 1.006–1.072; P = .020), Eubacterium fissicatena group (OR: 1.025; 95% CI: 1.001–1.050; P = .038), and Anaerotruncus (OR: 1.041; 95% CI: 1.005–1.079; P = .026) were linked with an elevated risk of WM PVSs, while a higher Clostridium sensu stricto 1 (OR: 0.957; 95% CI: 0.922–0.992; P = .017) level was linked with lower WM PVS likelihood (Fig. 2, Table 1).

The Cochran Q statistic revealed no substantial heterogeneity, while the leave-one-out sensitivity analysis confirmed the absence of SNP influence on the causal relationship (Figures S1–S6, Supplemental Digital Content, https://links.lww.com/MD/Q703). Moreover, the MR-Egger intercept and MR-PRESSO results suggested no considerable horizontal pleiotropy (Table S6, Supplemental Digital Content, https://links.lww.com/MD/Q702). After FDR correction, Lactococcus (PFDR = .049) was negatively associated with BG PVSs, Ruminococcaceae UCG005 (PFDR = .049) was positively associated with hippocampal PVSs, and Haemophilus (PFDR = .049) was negatively associated with hippocampal PVSs (Table 1, Table S5, Supplemental Digital Content, https://links.lww.com/MD/Q702).

An MVMR analysis evaluated the effects of intestinal microbes on CSVD risk. This analysis demonstrated no independent associations between GM and CSVD (Table 2, Table S7, Supplemental Digital Content, https://links.lww.com/MD/Q702). Moreover, no causal link was found between CSVD and GM, as shown by the reverse MR analysis (Tables S8 and S9, Supplemental Digital Content, https://links.lww.com/MD/Q702).

Table 2.

Multivariable MR analyses results of gut microbiota and CSVD.

Exposure Outcome OR (95% CI) P
Alistipes WMH volume 1.027 (0.964–1.093) .408
Eisenbergiella WMH volume 0.980 (0.935–1.028) .416
Fusicatenibacter WMH volume 0.989 (0.927–1.056) .740
Lachnospiraceae UCG008 WMH volume 0.991 (0.951–1.032) .654
Turicibacter WMH volume 1.007 (0.971–1.044) .707
Streptococcus Lacunar stroke 1.084 (0.991–1.186) .079
Lachnospiraceae UCG010 Lacunar stroke 0.981 (0.903–1.065) .642
Lachnospiraceae UCG001 Lacunar stroke 0.949 (0.865–1.040) .262
Fusicatenibacter Lacunar stroke 1.001 (0.891–1.124) .991
Lactococcus BG PVSs 1.005 (0.995–1.014) .360
Lachnospiraceae UCG008 BG PVSs 1.006 (0.993–1.020) .371
Dorea BG PVSs 0.993 (0.973–1.012) .461
Bilophila BG PVSs 0.998 (0.981–1.014) .776
Bacteroides BG PVSs 0.998 (0.978–1.019) .843
Ruminococcaceae UCG005 Hippocampal PVSs 0.998 (0.983–1.012) .759
Holdemania Hippocampal PVSs 1.004 (0.992–1.016) .495
Haemophilus Hippocampal PVSs 1.002 (0.992–1.012) .740
Clostridium sensustricto 1 Hippocampal PVSs 0.988 (0.976–1.001) .064
Victivallis WM PVSs 1.007 (0.998–1.017) .123
Ruminococcaceae UCG004 WM PVSs 1.009 (0.996–1.023) .159
Ruminococcaceae NK4A214 group WM PVSs 1.000 (0.985–1.015) .997
Eubacterium fissicatena group WM PVSs 0.996 (0.985–1.007) .487
Clostridium sensu stricto 1 WM PVSs 0.992 (0.983–1.001) .067
Anaerotruncus WM PVSs 1.013 (0.998–1.028) .081
Open in a new tab

Multivariable MR (MVMR) analyses were conducted by using the inverse variance weighted (IVW) method.

BG PVSs = basal ganglia perivascular spaces, CI = confidence interval, CSVD = cerebral small vessel disease, Hippocampal PVSs = Hippocampal perivascular spaces, MR = Mendelian randomization, OR = odds ratio, WM PVSs = white matter perivascular spaces, WMH volume = white matter hyperintensity volume.

3.2. Correlation between GMDMs and CSVD

The high concentrations of glycodeoxycholate (OR: 1.142; 95% CI: 1.005–1.297; P = .041), isoleucine (OR: 1.066; 95% CI: 1.001–1.136; P = .047), trans-4-hydroxyproline (OR: 1.529; 95% CI: 1.111–2.104; P = .009), and tyrosine (OR: 1.599; 95% CI: 1.025–2.495; P = .039) were strongly linked to elevated WMH volume risk, while high level of phenylacetate (OR: 0.754; 95% CI: 0.596–0.954; P = .019) protected against WMH volume. Increased abundances of cysteine (OR: 2.691; 95% CI: 1.014–7.146; P = .047) and Ketoleucine (OR: 3.455; 95% CI: 1.042–11.456; P = .043) were remarkably related to an elevated risk of CMBs, while a higher level of indolepropionate (OR: 0.526; 95% CI: 0.292–0.950; P = .033) protected against CMBs. Furthermore, 3-indoxyl sulfate (OR: 0.533; 95% CI: 0.286–0.994; P = .048), adrenate (OR: 0.541; 95% CI: 0.293–0.999; P = .049), allantoin (OR: 0.672; 95% CI: 0.456–0.990; P = .044), and glycine (OR: 0.610; 95% CI: 0.389–0.957; P = .032) protected against lacunar stroke, while aspartate (OR: 3.004; 95% CI: 1.195–7.548; P = .019), phenylalanine (OR: 21.631; 95% CI: 1.376–340.006; P = .029), taurodeoxycholate (OR: 1.425; 95% CI: 1.088–1.866; P = .010), and trans-4-hydroxyproline (OR: 2.198; 95% CI: 1.080–4.471; P = .030) were showed to elevate the possibility of lacunar stroke (Table 3).

Table 3.

MR analyses of gut microbiota-derived metabolites on CSVD by IVW method.

Exposure Outcome OR (95% CI) P P FDR
Glycodeoxycholate WMH volume 1.142 (1.005–1.297) .041 .050
Isoleucine WMH volume 1.066 (1.001–1.136) .047 .049
Phenylacetate WMH volume 0.754 (0.596–0.954) .019 .071
Trans-4-hydroxyproline WMH volume 1.529 (1.111–2.104) .009 .081
Tyrosine WMH volume 1.599 (1.025–2.495) .039 .050
Cysteine CMBs 2.691 (1.014–7.146) .047 .050
Indolepropionate CMBs 0.526 (0.292–0.950) .033 .050
Ketoleucine CMBs 3.455 (1.042–11.456) .043 .049
3-Indoxyl sulfate Lacunar stroke 0.533 (0.286–0.994) .048 .049
Adrenate Lacunar stroke 0.541 (0.293–0.999) .049 .049
Allantoin Lacunar stroke 0.672 (0.456–0.990) .044 .050
Aspartate Lacunar stroke 3.004 (1.195–7.548) .019 .068
Glycine Lacunar stroke 0.610 (0.389–0.957) .032 .051
Phenylalanine Lacunar stroke 21.631 (1.376–340.006) .029 .051
Taurodeoxycholate Lacunar stroke 1.425 (1.088–1.866) .010 .067
Trans-4-hydroxyproline Lacunar stroke 2.198 (1.080–4.471) .030 .049
Arabinose BG PVSs 0.928 (0.875–0.985) .014 .080
Choline BG PVSs 0.757 (0.591–0.970) .028 .053
Glycodeoxycholate BG PVSs 0.950 (0.916–0.986) .007 .175
Arginine Hippocampal PVSs 0.872 (0.769–0.989) .034 .048
Dimethylarginine Hippocampal PVSs 0.844 (0.722–0.987) .033 .049
Indolepropionate Hippocampal PVSs 1.094 (1.018–1.176) .015 .078
Isoleucine Hippocampal PVSs 1.026 (1.003–1.050) .024 .056
Kynurenine Hippocampal PVSs 1.132 (1.013–1.265) .029 .049
Methionine Hippocampal PVSs 1.292 (1.004–1.663) .047 .051
Indoleacetate WM PVSs 0.906 (0.828–0.991) .032 .050
Taurochenodeoxycholate WM PVSs 0.941 (0.899–0.984) .008 .141
Trans-4-hydroxyproline WM PVSs 0.867 (0.764–0.985) .028 .051
Open in a new tab

BG PVSs = basal ganglia perivascular spaces, CI = confidence interval, CMBs = cerebral microbleeds, CSVD = cerebral small vessel disease, Hippocampal PVSs = Hippocampal perivascular spaces, IVW = inverse variance weighted; WMH volume, white matter hyperintensity volume, MR = Mendelian randomization, OR = odds ratio, WM PVSs = white matter perivascular spaces.

High levels of arabinose (OR: 0.928; 95% CI: 0.875–0.985; P = .014), choline (OR: 0.757; 95% CI: 0.591–0.970; P = .028), and glycodeoxycholate (OR: 0.950; 95% CI: 0.916–0.986; P = .007) protected against BG PVSs. Similarly, high level of indolepropionate (OR: 1.094; 95% CI: 1.018–1.176; P = .015), isoleucine (OR: 1.026; 95% CI: 1.003–1.050; P = .024), kynurenine (OR: 1.132; 95% CI: 1.013–1.265; P = .029), and methionine (OR: 1.292; 95% CI: 1.004–1.663; P = .047) were substantially correlated with greater probability of hippocampal PVSs, while higher level of arginine (OR: 0.872; 95% CI: 0.769–0.989; P = .034) and dimethylarginine (OR: 0.844; 95% CI: 0.722–0.987; P = .033) protected against hippocampal PVSs. In addition, indoleacetate (OR: 0.906; 95% CI: 0.828–0.991; P = .032), taurochenodeoxycholate (OR: 0.941; 95% CI: 0.899–0.984; P = .008), and trans-4-hydroxyproline (OR: 0.867; 95% CI: 0.764–0.985; P = .028) protected against WM PVSs (Table 3).

The Cochran Q test revealed no substantial heterogeneity, while the leave-one-out sensitivity analysis validated that the observed causal links were not influenced by any SNP (Figures S7–S12, Supplemental Digital Content, https://links.lww.com/MD/Q703). Further, the MR-Egger intercept and the MR-PRESSO global test results indicated no horizontal pleiotropy (Table S6, Supplemental Digital Content, https://links.lww.com/MD/Q702). After FDR correction, isoleucine (PFDR = .049) was positively associated with WMH volume. Ketoleucine (PFDR = .049) was positively associated with CMBs. Moreover, 3-indoxyl sulfate (PFDR = .049) and adrenate (PFDR = .049) were negatively associated with lacunar stroke, while trans-4-hydroxyproline (PFDR = .049) was positively associated with lacunar stroke. Arginine (PFDR = .048) and dimethylarginine (PFDR = .049) were negatively related to hippocampal PVSs, while kynurenine (PFDR = .049) was positively associated with hippocampal PVSs (Table 3, Table S5, Supplemental Digital Content, https://links.lww.com/MD/Q702).

MVMR analyses were conducted to evaluate the effects of microbial metabolites on susceptibility to CSVD. The findings revealed an inverse correlation between aspartate levels and lacunar stroke (OR: 0.702; 95% CI: 0.504–0.978; P = .036). Indolepropionate protected against hippocampal PVSs (OR: 0.974; 95% CI: 0.951–0.998; P = .032) (Table 4 and Table S7, Supplemental Digital Content, https://links.lww.com/MD/Q702). However, reverse MR analysis indicated no causal relationship between CSVD and GM metabolites, except for a significant link observed between lacunar stroke and glycine levels (Tables S8 and S9, Supplemental Digital Content, https://links.lww.com/MD/Q702).

Table 4.

Multivariable MR analyses results of gut microbiota-derived metabolites and CSVD.

Exposure Outcome OR (95% CI) P
Glycodeoxycholate WMH volume 0.993 (0.961–1.026) .674
Isoleucine WMH volume 1.011 (0.994–1.029) .202
Phenylacetate WMH volume 1.012 (0.942–1.087) .747
Trans-4-hydroxyproline WMH volume 0.917 (0.782–1.075) .284
Tyrosine WMH volume 1.411 (0.978–2.037) .066
Cysteine CMBs 1.020 (0.793–1.312) .879
Indolepropionate CMBs 0.707 (0.499–1.001) .050
Ketoleucine CMBs 1.228 (0.624–2.418) .552
3-Indoxyl sulfate Lacunar stroke 1.080 (0.845–1.381) .537
Adrenate Lacunar stroke 0.988 (0.775–1.259) .920
Allantoin Lacunar stroke 0.899 (0.729–1.108) .319
Aspartate Lacunar stroke 0.702 (0.504–0.978) .036
Glycine Lacunar stroke 0.838 (0.660–1.064) .146
Phenylalanine Lacunar stroke 1.682 (0.833–3.396) .147
Taurodeoxycholate Lacunar stroke 0.973 (0.883–1.072) .580
Trans-4-hydroxyproline Lacunar stroke 0.942 (0.818–1.084) .402
Arabinose BG PVSs 0.985 (0.953–1.018) .358
Choline BG PVSs 0.999 (0.838–1.190) .988
Glycodeoxycholate BG PVSs 0.994 (0.974–1.014) .570
Arginine Hippocampal PVSs 0.967 (0.916–1.022) .238
Dimethylarginine Hippocampal PVSs 0.995 (0.912–1.085) .904
Indolepropionate Hippocampal PVSs 0.974 (0.951–0.998) .032
Isoleucine Hippocampal PVSs 0.997 (0.993–1.001) .154
Kynurenine Hippocampal PVSs 0.975 (0.931–1.022) .294
Methionine Hippocampal PVSs 0.996 (0.912–1.089) .938
Indoleacetate WM PVSs 1.015 (0.981–1.050) .404
Taurochenodeoxycholate WM PVSs 0.999 (0.979–1.021) .962
Trans-4-hydroxyproline WM PVSs 1.020 (0.985–1.057) .263
Open in a new tab

Multivariable MR (MVMR) analyses were conducted by using the inverse variance weighted (IVW) method.

BG PVSs = basal ganglia perivascular spaces, CI = confidence interval, CMBs = cerebral microbleeds, CSVD = cerebral small vessel disease, Hippocampal PVSs = Hippocampal perivascular spaces, MR = Mendelian randomization, OR = odds ratio, WM PVSs = white matter perivascular spaces, WMH volume = white matter hyperintensity volume.

3.3. Enrichment analysis

The enrichment analysis of GM, metabolites, and CSVD revealed substantial involvement of several crucial regulatory pathways. GO and KEGG analyses identified 41 GO biological processes (BPs) and 3 KEGG pathways related to WMH volume, including pathways related to metal ion translocation and lymphocyte-mediated immune responses. Similarly, 6 GO BPs and 1 KEGG pathway showed substantial enrichment for CMBs, for example, female sexual differentiation. After GO and KEGG enrichment analyses, 77 GO BPs and 1 KEGG pathway were found to be remarkably enriched in lacunar stroke, involving processes such as myeloid leukocyte differentiation, regulation of MAP kinase activity, and the Fc epsilon RI signaling pathway. In the case of basal ganglia perivascular spaces (BG PVSs), 31 GO BPs and 2 KEGG pathways showed significant enrichment, including the amide biosynthetic process and metal ion transport (Fig. 3).

Figure 3.

Figure 3.

Open in a new tab

Enrichment analysis of GM, metabolites, and CSVD. (A) Functional enrichment analysis of the network by using GO in WMH volume; (B) pathway enrichment analysis of the network by using KEGG in WMH volume; (C) functional enrichment analysis of the network by using GO in CMBs; (D) pathway enrichment analysis of the network by using KEGG in CMBs; (E) functional enrichment analysis of the network by using GO in lacunar stroke; (F) pathway enrichment analysis of the network by using KEGG in lacunar stroke; (G) functional enrichment analysis of the network by using GO in BG PVSs; (H) pathway enrichment analysis of the network by using KEGG in BG PVSs; (I) functional enrichment analysis of the network by using GO in hippocampal PVSs; (J) pathway enrichment analysis of the network by using KEGG in hippocampal PVSs; (K) functional enrichment analysis of the network by using GO in WM PVSs; (L) pathway enrichment analysis of the network by using KEGG in WM PVSs. BG PVSs = basal ganglia perivascular spaces; CMBs = cerebral microbleeds; CSVD = cerebral small vessel disease; GM = gut microbiota; GO = gene ontology; KEGG = kyoto encyclopedia of genes and genomes; hippocampal PVSs = hippocampal perivascular spaces; WM PVSs = white matter perivascular spaces; WMH volume = white matter hyperintensity volume.

Furthermore, 96 GO BPs and 18 KEGG pathways were substantially enriched in hippocampal perivascular spaces (hippocampal PVSs), highlighting pathways such as activation of innate immune responses, amide metabolism, and regulation of TRP channels by inflammatory mediators. Moreover, 30 GO BPs and 3 KEGG pathways were considerably enriched in WM PVSs, that is, cellular response to tumor necrosis factor (Fig. 3).

4. Discussion

Previous studies have demonstrated that GM possesses substantial heritability and is crucial in developing cardiovascular diseases, emphasizing the importance of conducting microbial GWAS in CSVD. This study’s MR analyses defined the causal links among gut microbial composition, GMDMs, and CSVD. Using summary statistics from the latest and most comprehensive GWAS datasets, causal links were established between 25 gut microbial genera and various CSVD subtypes. Furthermore, results indicated that elevated levels of 28 specific metabolites may act as protective factors or risk contributors for different forms of CSVD.

In relation to the GM, anaerobic commensals belonging to the Lachnospiraceae family can synthesize compounds structurally similar to carnitine. These analogs accumulate with carnitine in the brain’s white matter, interfering with its biological function.[32] Besides, a previous study has demonstrated that elevated carnitine levels correlate with reduced WMH volumes.[33] Moreover, clinical evidence indicated that L-carnitine administration is beneficial in treating lacunar stroke.[34] Collectively, these results indicate that Lachnospiraceae may enhance the likelihood WMH volume and lacunar stroke by impairing carnitine activity, aligning with observations from the current analysis. However, other studies reported that Lachnospiraceae UCG001 showed a negative association with cardiovascular risk factors after walnut intake,[35] further supporting these findings. This underscores the necessity of performing comprehensive investigations at the species level and in different CSVD forms to examine the specific mechanistic role modulated by GM communities.

Previous studies revealed that individuals diagnosed with CSVD showed a higher burden of Bifidobacterium than healthy individuals, whereas Dorea was found in reduced abundance. These observations align with the current findings. Furthermore, an increased abundance of Bifidobacterium has been positively associated with poorer clinical outcomes.[36]

Concerning the association between GMDMs and CSVD, previous clinical reports have demonstrated that elevated isoleucine levels (hyperisoleucinemia) are related to the occurrence of white matter lesions,[37,38] supporting the present observations. Previous studies have shown that higher soluble tyrosine kinase was associated with larger WMH volumes,[39] which supported the current results. Previous univariable MR studies reported that aspartate and trans-4-hydroxyproline are risk factors for lacunar stroke, while adrenate is a protective factor for lacunar stroke,[40] which was in line with the present study. L-arginine is a substrate for eNOS, whose activity and generation of endothelial NO play crucial neurovascular protective roles. Recent studies indicate that aged eNOS heterozygous mice are a unique model of spontaneous CSVD.[41] L-arginine has been shown to increase blood flow in cerebral ischemia.[42] The findings outlined above substantiate the conclusion that arginine is a protective factor against CSVD. Interestingly, univariable MR analysis identified aspartate and indolepropionate as risk factors for lacunar stroke and hippocampal PVSs, respectively. However, after adjusting for the influence of other GMDMs in multivariable MR analysis, aspartate showed a protective role against lacunar stroke, while indolepropionate showed a protective effect on hippocampal PVSs. These findings emphasize the necessity of conducting studies at the level of individual metabolites to clarify the underlying mechanisms in terms of GM-derived metabolic pathways.

Moreover, GO and KEGG enrichment analyses revealed that numerous GO BPs and KEGG pathways are critically involved in the association between GM, their metabolites, and CSVD, consistent with previous research. For example, sex-specific differences have been documented in CSVD patients, where elevated serum uric acid levels were correlated with an increased prevalence of CMBs in males, while a lower prevalence was observed in females.[43] Chronic inflammatory responses and leukocyte infiltration are also recognized as hallmark pathological CSVD features.[44]

A main strength of the present study is the novelty in establishing causal associations between GM, their metabolites, and CSVD using univariable and multivariable MR methodologies. Comprehensive sensitivity analyses showed the reliability and consistency of the findings. Next, incorporating the most recent and large-scale GWAS facilitated the analysis of extensive genetic datasets, thus enhancing the statistical power and credibility of the results relative to smaller randomized controlled trials. This study provides a theoretical basis for elucidating the mechanistic pathways through which specific microbial taxa affect CSVD and contributes to identifying novel biomarkers. However, the study also has certain limitations. Firstly, caution should be applied in interpreting the findings due to insufficient IVs meeting the strict genome-wide threshold of significance. A more reliable threshold (P < 1 × 10⁻⁵) was used to reduce this limitation to select exposure-associated SNPs. Secondly, the taxonomic resolution of the GM analysis was limited to the genus level, precluding assessment at the particular species level. Lastly, studies have revealed that gene–biota interactions vary considerably across ethnic populations. Thus, generalizability of the findings may be affected, as the underlying GWAS were primarily conducted in populations with European ancestry.

To conclude the current MR analysis explored the possible causal relationships between GM, their metabolites, and CSVD. The results may provide valuable insights into identifying novel biomarkers for developing targeted strategies to prevent and manage CSVD. However, further validation via RCTs and mechanistic experimental studies is needed to confirm these associations and clarify the underlying biological pathways.

Acknowledgments

The authors would like to thank all the reviewers who participated in the review and MJEditor (www.mjeditor.com) for its linguistic assistance during the preparation of this manuscript.

Author contributions

Conceptualization: Lin Wu.

Investigation: Hao Xue.

Formal analysis: Fang Yang, Lei Zhang.

Software: Xiaoli Yang, Miaomiao Zhang.

Visualization: Xiaoli Yang, Miaomiao Zhang.

Writing – original draft: Hao Xue.

Writing – review & editing: Jing Zang, Xiuping Zhang, Lin Wu.

Supplementary Material

medi-104-e45824-s001.xls (2.8MB, xls)
Open in a new tab
medi-104-e45824-s002.pdf (518.4KB, pdf)
Open in a new tab

Abbreviations:

BPs
biological processes
BG
basal ganglia
CI
confidence interval
CMBs
cerebral microbleeds
CSVD
cerebral small vessel disease
FDRs
false discovery rates
GM
gut microbiota
GMDMs
GM-derived metabolites
GWAS
genome-wide association studies
IVs
instrumental variables
IVW
inverse variance weighted
MGBA
microbiota–gut–brain axis
MR
Mendelian randomization
MVMR
multivariable MR
OR
odds ratio
WM PVSs
white matter perivascular spaces
PVSs
perivascular spaces
SNP
single-nucleotide polymorphism
WMH
white matter hyperintensity

The authors have no funding and conflicts of interest to disclose.

The datasets generated during and/or analyzed during the current study are available from the corresponding author on reasonable request.

Supplemental Digital Content is available for this article.

How to cite this article: Xue H, Zang J, Yang F, Zhang L, Yang X, Zhang M, Zhang X, Wu L. Causal effect of gut microbiota and its metabolites on cerebral small vessel disease: A multivariable Mendelian randomization analysis. Medicine 2025;104:50(e45824).

References

  • [1].de Leeuw FE, de Groot JC, Achten E, et al. Prevalence of cerebral white matter lesions in elderly people: a population based magnetic resonance imaging study. The Rotterdam scan study. J Neurol Neurosurg Psychiatry. 2001;70:9–14. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • [2].Hilal S, Akoudad S, van Duijn CM, et al. Plasma amyloid-β levels, cerebral small vessel disease, and cognition: the rotterdam study. J Alzheimers Dis. 2017;60:977–87. [DOI] [PubMed] [Google Scholar]
  • [3].Rajeev V, Chai YL, Poh L, et al. Chronic cerebral hypoperfusion: a critical feature in unravelling the etiology of vascular cognitive impairment. Acta Neuropathol Commun. 2023;11:93. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • [4].Markus HS, de Leeuw FE. Cerebral small vessel disease: Recent advances and future directions. Int J Stroke. 2023;18:4–14. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • [5].Zou X, Wang L, Xiao L, Wang S, Zhang L. Gut microbes in cerebrovascular diseases: gut flora imbalance, potential impact mechanisms and promising treatment strategies. Front Immunol. 2022;13:975921. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • [6].Zhang S, Chen Q, Jin M, et al. Notoginsenoside R1 alleviates cerebral ischemia/reperfusion injury by inhibiting the TLR4/MyD88/NF-κB signaling pathway through microbiota-gut-brain axis. Phytomedicine. 2024;128:155530. [DOI] [PubMed] [Google Scholar]
  • [7].Rahman Z, Bhale NA, Dikundwar AG, Dandekar MP. Multistrain probiotics with fructooligosaccharides improve middle cerebral artery occlusion-driven neurological deficits by revamping microbiota–gut–brain axis. Probiotics Antimicrob Proteins. 2024;16:1251–69. [DOI] [PubMed] [Google Scholar]
  • [8].Cai W, Chen X, Men X, et al. Gut microbiota from patients with arteriosclerotic CSVD induces higher IL-17A production in neutrophils via activating RORγt. Sci Adv. 2021;7:eabe4827. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • [9].Shi Y, Zhao E, Li L, et al. Alteration and clinical potential in gut microbiota in patients with cerebral small vessel disease. Front Cell Infect Microbiol. 2023;13:1231541. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • [10].Huang C, Zhang W, Shen Z, et al. The association between alpha diversity of gut microbiota, neuroimaging markers and cognitive function in cerebral small vessel disease. Brain Res. 2024;1827:148757. [DOI] [PubMed] [Google Scholar]
  • [11].Greenland S. An introduction to instrumental variables for epidemiologists. Int J Epidemiol. 2000;29:722–9. Erratum in: Int J Epidemiol. 47358 (2018) [DOI] [PubMed] [Google Scholar]
  • [12].Ning J, Huang S-Y, Chen S-D, Zhang Y-R, Huang Y-Y, Yu J-T. Investigating casual associations among gut microbiota, metabolites, and neurodegenerative diseases: a Mendelian randomization study. J Alzheimers Dis. 2022;87:211–22. [DOI] [PubMed] [Google Scholar]
  • [13].Liu X, Yuan J, Liu S, et al. Investigating causal associations among gut microbiota, metabolites and autoimmune hypothyroidism: a univariable and multivariable Mendelian randomization study. Front Immunol. 2024;14:1213159. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • [14].Kurilshikov A, Medina-Gomez C, Bacigalupe R, et al. Large-scale association analyses identify host factors influencing human gut microbiome composition. Nat Genet. 2021;53:156–65. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • [15].Quast C, Pruesse E, Yilmaz P, et al. The SILVA ribosomal RNA gene database project: improved data processing and web-based tools. Nucleic Acids Res. 2013;41:D590–6. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • [16].Shin SY, Fauman EB, Petersen A-K, et al. An atlas of genetic influences on human blood metabolites. Nat Genet. 2014;46:543–50. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • [17].Wishart DS, Feunang YD, Marcu A, et al. HMDB 4.0: the human metabolome database for 2018. Nucleic Acids Res. 2018;46:D608–17. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • [18].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. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • [19].Smith SM, Douaud G, Chen W, et al. An expanded set of genome-wide association studies of brain imaging phenotypes in UK Biobank. Nat Neurosci. 2021;24:737–45. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • [20].Knol MJ, Lu D, Traylor M, et al. Association of common genetic variants with brain microbleeds: a genome-wide association study. Neurology. 2020;95:e3331–43. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • [21].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;29:950–62. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • [22].Lin D, Zhu Y, Tian Z, et al. Causal associations between gut microbiota, gut microbiota-derived metabolites, and cerebrovascular diseases: a multivariable Mendelian randomization study. Front Cell Infect Microbiol. 2023;13:1269414. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • [23].Staiger D, Stock JH. Instrumental variables regression with weak instruments. Econometrica. 1997;65:557–86. [Google Scholar]
  • [24].Palmer TM, Lawlor DA, Harbord RM, et al. Using multiple genetic variants as instrumental variables for modifiable risk factors. Stat Methods Med Res. 2012;21:223–42. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • [25].Burgess S, Dudbridge F, Thompson SG. Combining information on multiple instrumental variables in Mendelian randomization: comparison of allele score and summarized data methods. Stat Med. 2016;35:1880–906. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • [26].Thompson JR, Minelli C, Abrams KR, Tobin MD, Riley RD. Meta-analysis of genetic studies using Mendelian randomization – a multivariate approach. Stat Med. 2005;24:2241–54. [DOI] [PubMed] [Google Scholar]
  • [27].Bowden J, Davey Smith G, Haycock PC, Burgess S. Consistent estimation in Mendelian randomization with some invalid instruments using a weighted median estimator. Genet Epidemiol. 2016;40:304–14. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • [28].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. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • [29].Verbanck M, Chen CY, Neale B, Do R. Detection of widespread horizontal pleiotropy in causal relationships inferred from Mendelian randomization between complex traits and diseases. Nat Genet. 2018;50:693–8. Erratum in: Nat Genet. 501196 (2018) [DOI] [PMC free article] [PubMed] [Google Scholar]
  • [30].Hemani G, Tilling K, Davey Smith G. Orienting the causal relationship between imprecisely measured traits using GWAS summary data. PLoS Genet. 2017;13:e1007081. Erratum in: PLoS Genet13e1007149 (2017) [DOI] [PMC free article] [PubMed] [Google Scholar]
  • [31].Zhou Y, Zhou B, Pache L, et al. Metascape provides a biologist-oriented resource for the analysis of systems-level datasets. Nat Commun. 2019;10:1523. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • [32].Hulme H, Meikle LM, Strittmatter N, et al. Microbiome-derived carnitine mimics as previously unknown mediators of gut-brain axis communication. Sci Adv. 2020;6:eaax6328. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • [33].Yaqub A, Vojinovic D, Vernooij MW, et al. Plasma trimethylamine N-oxide (TMAO): associations with cognition, neuroimaging, and dementia. Alzheimers Res Ther. 2024;16:113. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • [34].Suslina ZA, Fedorova TN, Maksimova MY, Kim EK. Antioksidantnoe deĭstvie mildronata i L-karnitina pri lechenii bol’nykh s sosudistymi zabolevaniiami golovnogo mozga [Antioxidant activity of mildronate and L-carnitine in the treatment of patients with cerebrovascular diseases]. Eksp Klin Farmakol. 2003;66:32–5. (in Russian) [PubMed] [Google Scholar]
  • [35].Tindall AM, McLimans CJ, Petersen KS, Kris-Etherton PM, Lamendella R. Walnuts and vegetable oils containing oleic acid differentially affect the gut microbiota and associations with cardiovascular risk factors: follow-up of a randomized, controlled, feeding trial in adults at risk for cardiovascular disease. J Nutr. 2020;150:806–17. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • [36].He P, Jiang C, Ni J, et al. Identifying gut microbiota with high specificity for ischemic stroke with large vessel occlusion. Sci Rep. 2024;14:14086. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • [37].Poll-The BT, Wanders RJA, Ruiter JPN, et al. Spastic diplegia and periventricular white matter abnormalities in 2-methyl-3-hydroxybutyryl-CoA dehydrogenase deficiency, a defect of isoleucine metabolism: differential diagnosis with hypoxic-ischemic brain diseases. Mol Genet Metab. 2004;81:295–9. [DOI] [PubMed] [Google Scholar]
  • [38].Wang XL, Li CJ, Xing Y, Yang YH, Jia JP. Hypervalinemia and hyperleucine-isoleucinemia caused by mutations in the branched-chain-amino-acid aminotransferase gene. J Inherit Metab Dis. 2015;38:855–61. [DOI] [PubMed] [Google Scholar]
  • [39].Raman MR, Himali JJ, Conner SC, et al. Circulating vascular growth factors and magnetic resonance imaging markers of small vessel disease and atrophy in middle-aged adults. Stroke. 2018;49:2227–9. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • [40].Guo MN, Hao X-Y, Tian J, et al. Human blood metabolites and lacunar stroke: a Mendelian randomization study. Int J Stroke. 2023;18:109–16. [DOI] [PubMed] [Google Scholar]
  • [41].Katusic ZS, d’Uscio LV, He T. Emerging roles of endothelial nitric oxide in preservation of cognitive health. Stroke. 2023;54:686–96. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • [42].Peters N, Freilinger T, Opherk C, Pfefferkorn T, Dichgans M. Effects of short term atorvastatin treatment on cerebral hemodynamics in CADASIL. J Neurol Sci. 2007;260:100–5. [DOI] [PubMed] [Google Scholar]
  • [43].Jeong SM, Yoo TG, Nam Y-S, et al. Sex-dependent effects of uric acid on cerebral microbleed: a cross-sectional study in the general population. Eur J Neurol. 2017;24:1300–6. [DOI] [PubMed] [Google Scholar]
  • [44].Fu Y, Yan Y. Emerging role of immunity in cerebral small vessel disease. Front Immunol. 2018;9:67. [DOI] [PMC free article] [PubMed] [Google Scholar]

Articles from Medicine are provided here courtesy of Wolters Kluwer Health

RESOURCES