A two sample mendelian randomization analysis investigates causal effects between gut microbiome and immune related Vasculitis

Abstract

Observational data suggest a link between gut microbiota and immune-related vasculitis, but causality remains unclear. A bidirectional mendelian randomization study was conducted using public genome-wide data. The inverse-variance-weighted (IVW) method identified associations and addressed heterogeneity.Families Clostridiaceae 1 and Actinomycetaceae correlated positively with granulomatosis with polyangiitis risk, while classes Lentisphaeria and Melainabacteria, and families Lachnospiraceae and Streptococcaceae showed negative associations. Behçet's disease was positively associated with the risk of family Streptococcaceae abundance. And other several gut microbiota constituents were identified as potential risk factors for immune-related vasculitis. Furthermore, combining positive association results from the IVW analysis revealed numerous shared gut microbiota constituents associated with immune-related vasculitis. MR analysis demonstrated a causal association between the gut microbiota and immune-related vasculitis, offering valuable insights for subsequent mechanistic and clinical investigations into microbiota-mediated immune-related vasculitis.

Introduction

Vasculitis is an autoimmune disorder characterized primarily by predominant vascular inflammation and destruction. Our study selectively encompasses specific vasculitis types from the 2012 Chapel Hill Nomenclature¹. We focused solely on studies with accessible genome-wide association study (GWAS) summary data related to giant cell arteritis (GCA), behçet's disease (BD), kawasaki disease (KD), and granulomatosis with polyangiitis (GPA). GCA is a systemic vasculitis distinguished by granulomatous inflammation affecting large and medium-sized arteries, which predominantly occurs in individuals aged 50 years or older, with an incidence rate of approximately 10 cases per 100,000 people². The manifestation of GCA involves granulomatous inflammation, particularly in the temporal artery, resulting in symptoms such as headache, jaw claudication, scalp discomfort, and increased C-reactive protein levels and erythrocyte sedimentation rate³. BD is a rare autoimmune disorder that causes inflammation and lesions in various body systems. It involves oral and genital ulcers, eye inflammation, and vascular issues and affects multiple organs⁴. BD is characterized by persistent mucosal ulceration and systemic vasculitis and has demographic and clinical heterogeneity. Mortality is elevated due to complications in the pulmonary artery, large vessels, nerves, and gastrointestinal tract⁵. KD manifests as an acute febrile illness and systemic vasculitis that primarily affects young children typically ranging from 6 months to 4 years of age⁶. Initially perceived as self-limiting⁷, KD is now acknowledged as a systemic vasculitis with a propensity to cause the development of coronary artery lesions, affecting up to 25% of untreated patients⁸. The emergence of coronary artery lesions associated with KD is a major contributor to pediatric heart disease in developed nations. GPA, formerly known as wegner's granulomatosis, represents a rare form of necrotizing vasculitis affecting small- to medium-sized vessels. This condition predominantly affects the upper and lower respiratory tracts, as well as the renal system⁹. The formation of necrotizing granulomas in regions around the head, neck, and kidneys can result in diverse and numerous clinical manifestations.

Gut dysbiosis, which influences mucosal immune homeostasis and gut barrier integrity, may contribute to autoimmune disorders. The established link between the microbiome and autoimmune diseases, such as systemic lupus erythematosus and rheumatoid arthritis, offers a promising therapeutic target¹⁰. Vasculitis patients exhibit dysbiosis compared to those of individuals with healthy controls. Crucially, environmental triggers, including changes in the gut microbiota, play a significant role in vasculitis onset, with an imbalance in the intestinal microbiome occurring with an increased abundance of pathogenic bacteria and a decreased abundance of beneficial bacteria¹¹. Nevertheless, investigations of the microbiome in blood and the aorta have revealed varying abundances of Actinobacteria, Proteobacteria, Bifidobacterium, and Parasutterella and minimal Bacteroidetes, Rhodococcus, Cytophagaceae, and Granulicatella among GCA patients^12–15. Several cross-sectional studies have indicated alterations in gut bacterial abundance in patients with BD compared to healthy controls. These findings revealed an enrichment of lactic acid-producing bacteria, sulfate-reducing bacteria, and certain opportunistic pathogens in the gut microbiota of BD patients; conversely, there was a deficiency in butyric acid-producing bacteria and methanogenic bacteria^16–24. KD arises from the interplay of genetic and environmental susceptibility factors alongside infectious triggers. Several researchers have proposed that KD induces microbial dysbiosis, diminishing the production of short-chain fatty acids (SCFAs) by the intestinal microbiota. This alteration potentially leads to an imbalance between helper T cells 17 and regulatory T cells, contributing to the pathogenesis of KD^25,26. Dekkema et al. conducted a comprehensive review of nasal microbiome studies in patients with ANCA-associated vasculitis (AAV), encompassing those with GPA. They found that nasal microbial dysbiosis is prevalent in active AAV patients, and immunosuppressive treatment thus has the potential to ameliorate this disturbance²⁷. Notably, investigations on GPA and the gut microbiota are lacking.

The inherent limitations within the designs of observational studies pose challenges in establishing definitive causal relationships. Constraints such as potential influence from confounding variables and biases, limited sample sizes, and variations in ethnic demographics hinder the ability to conclusively establish causality. Consequently, the causative nature of the relationship between the gut microbiota and immune-related vasculitis remains unclear. Furthermore, the directionality of this relationship—whether accidental, bidirectional, or unidirectional—remains ambiguous. Mendelian randomization (MR) is a contemporary epidemiological approach employed when the execution of randomized controlled trials is impractical. The primary data source for MR is derived from the Human Genome Project, and genetic variants are utilized as instrumental variables (IVs) to mitigate the limitations inherent in observational studies. By extracting single-nucleotide polymorphisms (SNPs) from GWASs, MR establishes a link between exposure and outcome while employing analytical techniques to mitigate confounding factors^28–30. MR investigations, akin to randomized controlled trials, yield results less susceptible to reverse causation and residual bias³¹. An increase in the number of GWASs related to the gut microbiota and disease^32,33 has led to the widespread availability of large-scale summary statistics, facilitating two-sample MR analysis with significantly enhanced statistical power.

In this study, we explored the causal connection between the gut microbiota and various immune-related vasculitis conditions through an extensive two-sample MR analysis involving four distinct vasculitis types (GCA, BD, KD, and GPA). Utilizing a bidirectional MR approach, we sought to investigate the potential causal impact of the gut microbiota on the risk of immune-related vasculitis and simultaneously assess whether genetic predisposition to immune-related vasculitis risk causally influences the composition of the gut microbiota. These analyses aimed to elucidate the role of the gut microbiota in the development of immune-related vasculitis, contributing to the eventual formulation of novel treatment strategies.

Methods

Study design

We conducted a two-sample MR analysis to explore the potential causal link between immune-related vasculitis and the gut microbiome. The study's workflow is depicted in Fig. 1. To ensure the validity of the IVs, we adhered to the three foundational assumptions of the MR design: (I) the genetic variation used as an IV must exhibit a significant association with the exposure(s); (ii) the genetic variation must be independent of confounding factors; and (iii) the variation must solely relate to the outcome(s) through the exposure(s)³⁴. Initial emphasis was placed on establishing causation, treating the gut microbiome as the exposure and immune-related vasculitis as the outcome. Subsequently, we analyzed the reverse causal direction, considering immune-related vasculitis as the exposure and the gut microbiome as the outcome (refer to Fig. 1 for detailed insights).

Figure 1.

Design of the two-sample MR study to assess the impact of the genetically inferred gut microbiome on specific immune-related vasculitis subtypes. BD Behcet's disease, GCA giant cell arteritis, KD Kawasaki disease, GPA granulomatosis with polyangiitis, N number of cases, SNPs single-nucleotide polymorphisms, MR: Mendelian randomization.

GWAS data sources

The international consortium MiBioGen conducted a comprehensive genome-wide meta-analysis of gut microbiota-related GWAS data involving 18,340 participants from 24 cohorts representing diverse ethnicities³³. The resulting dataset includes 211 GWAS summary statistics for bacterial taxa, covering 9 phyla, 16 classes, 20 orders, 35 families (including 3 with unknown classifications), and 131 genera (with 12 having unknown taxonomies). The detailed taxonomic categorizations are presented in Supplementary Table 1. In addition, MR analyses incorporated the remaining bacterial taxa from five alternative phyla and their subcategories, enhancing potential evidence for causality. After excluding 15 families and genera with unknown taxonomic classifications, a total of 196 taxa across diverse hierarchical levels were selected as the focal exposure of interest in our investigation. Additionally, GCA and BD summary statistics were derived from FinnGen Release 9³⁵ using the phenocodes "M13_GIANTCELL" and "M13_BEHCET". The GCA dataset included 366,529 samples, comprising 996 cases and 365,533 controls, while the BD dataset included 365,618 samples, comprising 85 cases and 365,533 controls. Summary data for KD were acquired from the IEU OpenGWAS database available at https://gwas.mrcieu.ac.uk/. The KD dataset included 6,190 samples, comprising 119 cases and 6,071 controls and encompassing a total of 152,542 genotyped SNPs³⁶. Summary data for GPA were acquired from the publicly available GWAS catalog (https://www.ebi.ac.uk/gwas/). The GPA dataset included 456,348 samples, comprising 135 cases and 456,213 controls³⁷.

IV selection

To identify potential associations between exposure and outcome, we employed stringent criteria for selecting SNPs as IVs. When considering the gut microbiome as the exposure factor, a p value < 1 × 10^–5 was considered to indicate statistical significance. Additionally, the linkage disequilibrium threshold was established at r² < 0.01, and the search distance for linkage disequilibrium r² values was limited to 500 kb. To assess the potential causal impact of immune-related vasculitis on the bacterial genera, we conducted a reverse MR analysis. In the case of immune-related vasculitis being the exposure, the significance level for IVs was set at a p value < 1 × 10^–4. The linkage disequilibrium threshold and clumping window were defined as r² < 0.01 and 250 kb, respectively. To assess potential weak IV bias, the F-statistic of the IVs was computed, with a threshold F-statistic < 10 indicating weak IV bias³⁸. IVs failing to meet this criterion (F-statistic < 10) were excluded to ensure robustness. The dataset was further refined by removing ambiguous and palindromic SNPs.

MR analyses

MR analyses were performed employing various methods, including inverse variance-weighted (IVW), weighted median, MR–Egger, and maximum likelihood approaches, to discern associations between the gut microbiome and three subtypes of immune-related vasculitis. The IVW approach, with an assumption of the validity of all SNPs as variables, served as the primary method. The weighted median approach provides consistent estimates under the assumption that more than half of the weights originate from valid SNPs³⁹. MR–Egger analysis, which is capable of calibrating for pleiotropy, facilitates causal inference even in the presence of pleiotropic genetic variants⁴⁰. The maximum likelihood-based approach was employed to ensure appropriate confidence interval (CI) estimation in the presence of weak IVs. Interpretative guidelines for these methods can be found elsewhere⁴¹. In sensitivity analyses, heterogeneity was assessed to gauge the compatibility of instrumental variables. Cochran’s Q statistics, implemented through the IVW and MR–Egger methods, were used to test for heterogeneity, and consideration of its effect was warranted if it was present among IVs (p < 0.05)⁴². The identification of horizontal pleiotropy, signaling that IVs are associated with outcomes through mechanisms other than causal effects and potentially leading to false-positive results (p < 0.05)⁴³, was considered crucial. Direct association testing between selected IVs and outcomes involved horizontal pleiotropy assessment using MR pleiotropy residual sum and outlier (MR-PRESSO). Leave-one-out analysis was also conducted to determine whether a single SNP disproportionately influenced the causal effect of exposure on outcomes. This approach involved iteratively omitting each SNP from IVs during IVW testing and assessing potential outliers using the TwoSampleMR package (version 0.5.7)⁴⁴. False discovery rate (FDR) correction was applied via the q value procedure, with the threshold set as a q value < 0.1⁴⁵. Taxa associations between the gut microbiota and immune-related vasculitis were considered suggestive if p < 0.05 but q ≥ 0.1. Specific FDR thresholds were established for various taxonomic levels: phyla (9), classes (16), orders (20), families (32), and genera (119). All analyses were conducted using R software (version 4.3.1). Venn diagram analysis was carried out using SUMO online software (https://angiogenesis.dkfz.de/oncoexpress/software/sumo/).

Results

SNP selection

Following our screening criteria, 27,548 SNPs were identified as IVs through a large-scale GWAS. A comprehensive set of 196 taxa spanning five biological classifications (phylum, class, order, family, and genus) was chosen as the exposure conditions. In the context of immune-related vasculitis as the exposure, 10 SNPs for BD, 89 SNPs for GCA, 35 SNPs for KD, and 77 SNPs for GPA were selected as IVs. All IVs exhibited F statistics well above 10 (refer to Supplementary Table 2), indicating an absence of weak instrument bias. In investigating the causal relationship between the gut microbiota and immune-related vasculitis, the primary interpretation relied on the IVW results, complemented by findings from four additional tests: detailed results from the IVW, weighted median, MR–Egger, and maximum likelihood approaches, along with outcomes regarding heterogeneity and pleiotropy, are available in Supplementary Tables 3–10. In the case of heterogeneity or pleiotropy (p < 0.05), any IV displaying such characteristics was excluded (see Supplementary Tables 11 and 12).

Positive causal effects of the gut microbiota on immune-related vasculitis, determined after FDR correction

After FDR correction, the results of IVW analyses demonstrated that the abundance of the genus Ruminococcaceae NK4A214 group (OR = 0.734, 95% CI = 0.614–0.877, p = 0.001, q = 0.081) was negatively associated with the risk of GCA. Additionally, the abundances of the families Clostridiaceae 1 (OR = 1.699, 95% CI = 1.314–2.196, p = 5.33E-05, q = 0.001) and Actinomycetaceae (OR = 2.333, 95% CI = 1.899–2.867, p = 7.68E-16, q = 1.23E-14) were positively associated with the risk of GPA. Moreover, the abundances of the class Lentisphaeria (OR = 0.649, 95% CI = 0.545–0.773, p = 1.28E-06, q = 4.08E-06), class Melainabacteria (OR = 0.675, 95% CI = 0.567–0.804, p = 1.00E-05, q = 2.67E-05), class Negativicutes (OR = 0.696, 95% CI = 0.51–0.951, p = 0.023, q = 0.04), family Lachnospiraceae (OR = 0.653, 95% CI = 0.515–0.828, p = 4.30E-04, q = 0.003), family Porphyromonadaceae (OR = 0.503, 95% CI = 0.309–0.818, p = 5.650E-03, q = 0.03), family Ruminococcaceae (OR = 0.442, 95% CI = 0.233–0.838, p = 0.012, q = 0.06), and family Streptococcaceae (OR = 0.4, 95% CI = 0.23–0.695, p = 0.001, q = 0.007) were negatively associated with the risk of GPA (Table 1). In reverse MR analysis, BD was found to be positively associated with the risk of the family Streptococcaceae (OR = 1.021, 95% CI = 0.996–1.022, p = 0.002, q = 0.065) (Table 2).

Table 1.

Positive MR results of the causal relationship between gut microbiota and GPA and GCA risk after FDR correction.

Taxa	Gut microbiota (exposure)	Outcome	Methods	SNPs (n)	OR (95% CI)	P-value	q-value	Test of heterogeneity		Test of pleiotropy
								Cochran’s Q	P-value	Egger intercept	SE	P-value
Genus	Ruminococcaceae NK4A214 group	GCA	MR Egger	105	0.648(0.363–1.155)	0.144	1.000	96.834	0.652	0.008	0.018	0.659
			Weighted median	105	0.726(0.56–0.941)	0.016	0.927
			IVW	105	0.734(0.614–0.877)	0.001	0.081	97.031	0.673
			Simple mode	105	0.784(0.39–1.579)	0.497	1.000
			Weighted mode	105	0.537(0.286–1.01)	0.057	1.000
Class	Negativicutes	GPA	MR Egger	310	0.439(0.16–1.205)	0.111	0.296	288.058	0.787	0.028	0.03	0.348
			Weighted median	310	1.13(0.717–1.781)	0.599	0.599
			IVW	310	0.696(0.51–0.951)	0.023	0.040	288.942	0.788
			Simple mode	310	3.006(0.602–15.005)	0.181	0.321
			Weighted mode	310	2.904(0.603–13.98)	0.185	0.657
Class	Melainabacteria	GPA	MR Egger	327	0.854(0.423–1.725)	0.66	0.094	243.851	1	-0.022	0.03	0.499
			Weighted median	327	0.644(0.506–0.819)	3.35E-04	8.94E-04
			IVW	327	0.675(0.567–0.804)	1.00E-05	2.67E-05	244.31	1
			Simple mode	327	1.467(0.599–3.594)	0.403	0.500
			Weighted mode	327	1.435(0.608–3.389)	0.411	1.000
Class	Lentisphaeria	GPA	MR Egger	300	1.465(0.625–3.436)	0.38	0.608	331.138	0.091	-0.086	0.05	0.056
			Weighted median	300	0.586(0.459–0.747)	1.68E-05	5.38E-05
			IVW	300	0.649(0.545–0.773)	1.28E-06	4.08E-06	335.213	0.073
			Simple mode	300	0.164(0.053–0.512)	2.03E-03	0.008
			Weighted mode	300	0.172(0.065–0.451)	4.06E-04	0.002
Family	Streptococcaceae	GPA	MR Egger	89	0.139(0.036–0.542)	5.53E-03	0.059	101.063	0.144	0.081	0.05	0.1
			Weighted median	89	0.428(0.197–0.926)	0.031	0.125
			IVW	89	0.4(0.23–0.695)	0.001	0.007	104.275	0.114
			Simple mode	89	0.377(0.05–2.845)	0.347	0.925
			Weighted mode	89	0.336(0.063–1.799)	0.206	0.732
Family	Porphyromonadaceae	GPA	MR Egger	112	0.56(0.146–2.151)	0.4	0.854	107.26	0.556	-0.007	0.04	0.867
			Weighted median	112	0.613(0.291–1.293)	0.199	0.454
			IVW	112	0.503(0.309–0.818)	5.65E-03	0.030	107.288	0.582
			Simple mode	112	0.579(0.089–3.763)	0.568	1.000
			Weighted mode	112	0.695(0.099–4.902)	0.716	1.000
Family	Lachnospiraceae	GPA	MR Egger	533	1.076(0.474–2.446)	0.861	0.888	287.359	1	-0.029	0.02	0.213
			Weighted median	533	0.515(0.372–0.715)	6.98E-05	4.46E-04
			IVW	533	0.653(0.515–0.828)	4.30E-04	0.003	288.917	1
			Simple mode	533	0.427(0.117–1.554)	0.197	0.631
			Weighted mode	533	0.441(0.123–1.59)	0.212	0.677
Family	Clostridiaceae 1	GPA	MR Egger	344	0.869(0.373–2.029)	0.746	0.880	320.495	0.792	0.045	0.03	0.105
			Weighted median	344	1.807(1.243–2.626)	0.002	0.010
			IVW	344	1.699(1.314–2.196)	5.33E-05	0.001	323.139	0.773
			Simple mode	344	1.472(0.399–5.44)	0.562	1.000
			Weighted mode	344	1.472(0.358–6.053)	0.592	0.997
Family	Actinomycetaceae	GPA	MR Egger	332	2.262(0.992–5.157)	0.053	0.243	369.839	0.064	0.003	0.03	0.939
			Weighted median	332	2.867(2.125–3.87)	5.71E-12	1.83E-10
			IVW	332	2.333(1.899–2.867)	7.68E-16	1.23E-14	369.846	0.069
			Simple mode	332	11.45(3.591–36.505)	4.77E-05	0.002
			Weighted mode	332	10.909(3.77–31.567)	1.41E-05	4.52E-04
Family	Ruminococcaceae	GPA	MR Egger	75	0.645(0.105–3.949)	0.637	0.927	84.008	0.178	-0.025	0.057	0.663
			Weighted median	75	0.599(0.232–1.547)	0.29	0.580
			IVW	75	0.442(0.233–0.838)	0.012	0.060	84.229	0.195
			Simple mode	75	1.043(0.101–10.747)	0.972	0.972
			Weighted mode	75	0.969(0.149–6.301)	0.974	1.000

GCA giant cell arteritis, GPA granulomatosis with polyangiitis, FDR False discovery rate, MR mendelian randomization, IVW inverse variance weighted, SNPs single nucleotide polymorphisms, OR odds ratio, CI confidence intervals, SE standard error.

IVW significant values are in bold.

Table 2.

Positive MR results of the causal relationship between gut microbiota and BD risk after FDR correction.

Exposure	Taxa	Gut microbiota (outcome)	Methods	SNPs (n)	OR (95% CI)	P-value	Test of heterogeneity		Test of pleiotropy
							Cochran’s Q	P-value	Egger intercept	SE	P-value
BD	Family	Streptococcaceae	MR Egger	10	1.024(0.933–1.094)	0.569	5.046	0.753	− 0.002	0.026	0.935
			Weighted median	10	1.022(0.992–1.028)	0.015
			IVW	10	1.021(0.996–1.022)	0.002	5.053	0.830
			Simple mode	10	1.029(0.984–1.042)	0.085
			Weighted mode	10	1.025(0.983–1.04)	0.112

BD Behcet's disease, FDR False discovery rate, MR mendelian randomization, IVW inverse variance weighted, SNPs single nucleotide polymorphisms OR odds ratio, CI confidence intervals; SE: standard error.

IVW significant values are in bold.

Potential causal effects of the gut microbiota on immune-related vasculitis

BD

Forward MR analysis

The results of IVW analyses demonstrated that the abundances of the class Melainabacteria (odds ratio (OR) = 1.851, 95% CI = 0.838–2.037, p = 0.007), class Gammaproteobacteria (OR = 1.949, 95% CI = 0.695–2.571, p = 0.046), family Rhodospirillaceae (OR = 1.744, 95% CI = 0.764–2.123, p = 0.033), genus Ruminococcaceae UCG011 (OR = 1.533, 95% CI = 0.837–1.733, p = 0.021), and genus Odoribacter (OR = 2.215, 95% CI = 0.764–2.613, p = 0.011) were potentially positively associated with the risk of BD (Supplementary Table 3, Fig. 2).

Figure 2.

Positive or potentially positive associations determined from forward MR analysis of the relationship between the gut microbiota and immune-related vasculitis. BD Behcet's disease, GCA giant cell arteritis, KD Kawasaki disease, GPA granulomatosis with polyangiitis, MR Mendelian randomization, SNPs single-nucleotide polymorphisms, OR odds ratio, CI confidence interval.

Reverse MR analysis

When considering BD as an exposure and the gut microbiota as an outcome, the results of IVW analyses demonstrated that BD was potentially positively associated with the risk of the class Bacilli (OR = 1.018, 95% CI = 0.995–1.021, p = 0.008), order Lactobacillales (OR = 1.018, 95% CI = 0.995–1.021, p = 0.007), genus Holdemanella (OR = 1.022, 95% CI = 0.99–1.029, p = 0.028), and genus Streptococcus (OR = 1.018, 95% CI = 0.995–1.021, p = 0.007) (Supplementary Table 4, Fig. 3). Moreover, BD was potentially negatively associated with the risk of the phylum Verrucomicrobia (OR = 0.984, 95% CI = 0.978–1.008, p = 0.043), the class Verrucomicrobiae (OR = 0.984, 95% CI = 0.978–1.009, p = 0.045), the order Verrucomicrobiales (OR = 0.984, 95% CI = 0.978–1.009, p = 0.045), the family Verrucomicrobiaceae (OR = 0.984, 95% CI = 0.978–1.009, p = 0.045), the genus Akkermansia (OR = 0.984, 95% CI = 0.978–1.009, p = 0.042), the genus Ruminiclostridium 6 (OR = 0.984, 95% CI = 0.979–1.008, p = 0.029), and the genus Coprococcus 3 (OR = 0.985, 95% CI = 0.98–1.007, p = 0.028) (Supplementary Table 4, Fig. 3).

Figure 3.

Positive or potentially positive associations determined from reverse MR analysis of the relationship between the gut microbiota and immune-related vasculitis. BD: Behcet's disease, GCA giant cell arteritis, KD Kawasaki disease, GPA granulomatosis with polyangiitis, MR Mendelian randomization, SNPs single-nucleotide polymorphisms; OR: odds ratio; CI: confidence interval.

GCA

Forward MR analysis

The results of IVW analyses demonstrated that the genera Lachnospiraceae UCG004 (OR = 1.214, 95% CI = 1.015–1.453, p = 0.034), Ruminococcus 2 (OR = 1.199, 95% CI = 1.003–1.434, p = 0.047), Ruminococcus gnavus group (OR = 1.142, 95% CI = 1.018–1.28, p = 0.023), Holdemania (OR = 1.225, 95% CI = 1.044–1.437, p = 0.013), Flavonifractor (OR = 1.225, 95% CI = 1.048–1.432, p = 0.011), Dialister (OR = 1.2, 95% CI = 1.029–1.4, p = 0.02), Bilophila (OR = 1.226, 95% CI = 1.043–1.44, p = 0.013), and Eubacterium nodatum group (OR = 1.148, 95% CI = 1.048–1.257, p = 0.003) were potentially positively associated with the risk of GCA (Supplementary Table 5, Fig. 2). Moreover, the phylum Verrucomicrobia (OR = 0.796, 95% CI = 0.673–0.943, p = 0.008), family Porphyromonadaceae (OR = 0.833, 95% CI = 0.696–0.998, p = 0.048), genus Veillonella (OR = 0.872, 95% CI = 0.768–0.991, p = 0.036), genus Erysipelatoclostridium (OR = 0.835, 95% CI = 0.72–0.968, p = 0.017), and genus Adlercreutzia (OR = 0.858, 95% CI = 0.745–0.988, p = 0.034) were potentially negatively associated with the risk of GCA (Supplementary Table 5, Fig. 2).

Reverse MR analysis

When GCA was considered an exposure and the gut microbiota was considered an outcome, the results of IVW analyses demonstrated that GCA was potentially positively associated with the risk of the genus Erysipelotrichaceae UCG003 (OR = 1.079, 95% CI = 0.987–1.083, p = 0.001) (Supplementary Table 6, Fig. 3). Moreover, GCA was potentially negatively associated with the risk of the family Victivallaceae (OR = 0.974, 95% CI = 0.964–1.014, p = 0.044), genus Alistipes (OR = 0.985, 95% CI = 0.982–1.005, p = 0.011), and genus Ruminococcaceae UCG010 (OR = 0.986, 95% CI = 0.98–1.008, p = 0.046) (Supplementary Table 6, Fig. 3).

KD

Forward MR analysis

The results of IVW analyses demonstrated that the phylum Lentisphaerae (OR = 2.8, 95% CI = 0.675–3.623, p = 0.016), genus Lachnospira (OR = 5.506, 95% CI = 0.522–8.428, p = 0.016), and genus Victivallis (OR = 2.368, 95% CI = 0.785–2.692, p = 0.006) were potentially positively associated with the risk of KD (Supplementary Table 7, Fig. 2). Moreover, the family Prevotellaceae (OR = 0.094, 95% CI = 0.053–2.416, p = 0.015), genus Ruminiclostridium 9 (OR = 0.217, 95% CI = 0.131–2.023, p = 0.029), genus Lactobacillus (OR = 0.317, 95% CI = 0.205–1.798, p = 0.038), and genus Bifidobacterium (OR = 0.196, 95% CI = 0.098–2.473, p = 0.048) were potentially negatively associated with the risk of KD (Supplementary Table 7, Fig. 2).

Reverse MR analysis

When KD was considered an exposure and the gut microbiota was considered an outcome, the results of IVW analyses demonstrated that KD was potentially positively associated with the risk of the genus Eubacterium oxidoreducens group (OR = 1.013, 95% CI = 0.993–1.018, p = 0.041) (Supplementary Table 8, Fig. 3). Moreover, KD was potential negatively associated with the risk of the order Bacillales (OR = 0.979, 95% CI = 0.976–1.005, p = 0.005), order Actinomycetales (OR = 0.988, 95% CI = 0.985–1.004, p = 0.01), family Actinomycetaceae (OR = 0.988, 95% CI = 0.985–1.004, p = 0.01), family Ruminococcaceae (OR = 0.994, 95% CI = 0.991–1.003, p = 0.044), family Defluviitaleaceae (OR = 0.987, 95% CI = 0.985–1.004, p = 0.006), family Rikenellaceae (OR = 0.994, 95% CI = 0.991–1.004, p = 0.048), genus Alistipes (OR = 0.993, 95% CI = 0.99–1.003, p = 0.02), genus Eubacterium eligens group (OR = 0.991, 95% CI = 0.989–1.003, p = 0.017), genus Ruminococcus torques group (OR = 0.993, 95% CI = 0.991–1.003, p = 0.018), genus Ruminococcus 2 (OR = 0.993, 95% CI = 0.991–1.004, p = 0.048), genus Streptococcus (OR = 0.993, 95% CI = 0.991–1.004, p = 0.044), genus Actinomyces (OR = 0.988, 95% CI = 0.985–1.004, p = 0.014), genus Defluviitaleaceae UCG011 (OR = 0.987, 95% CI = 0.985–1.003, p = 0.005), genus Butyricimonas (OR = 0.991, 95% CI = 0.988–1.005, p = 0.037), and genus Romboutsia (OR = 0.992, 95% CI = 0.99–1.004, p = 0.03) (Supplementary Table 8, Fig. 3).

GPA

Forward MR analysis

The results of IVW analyses demonstrated that the genera Turicibacter (OR = 1.465, 95% CI = 1.001–2.143, p = 0.049), Paraprevotella (OR = 1.513, 95% CI = 1.049–2.182, p = 0.027), Parabacteroides (OR = 1.899, 95% CI = 1.17–3.082, p = 0.009), Christensenellaceae R.7 group (OR = 1.688, 95% CI = 1.058–2.693, p = 0.029), Butyricimonas (OR = 1.58, 95% CI = 1.077–2.316, p = 0.019), and Anaerotruncus (OR = 1.684, 95% CI = 1.019–2.784, p = 0.042) were potentially positively associated with the risk of GPA (Supplementary Table 9, Fig. 2). Moreover, the order Gastranaerophilales (OR = 0.679, 95% CI = 0.477–0.966, p = 0.031), family Family XIII (OR = 0.602, 95% CI = 0.369–0.982, p = 0.042), genus Lachnospira (OR = 0.544, 95% CI = 0.342–0.865, p = 0.01), genus Blautia (OR = 0.616, 95% CI = 0.381–0.997, p = 0.048), and genus Bacteroides (OR = 0.583, 95% CI = 0.349–0.975, p = 0.04) were potentially negatively associated with the risk of GPA (Supplementary Table 9 ,Fig. 2).

Reverse MR analysis.

When GPA was considered an exposure and the gut microbiota was considered an outcome, the results of IVW analyses demonstrated that GPA was potentially positively associated with the risk of the genera Holdemanella (OR = 1.01, 95% CI = 0.996–1.012, p = 0.02) and Barnesiella (OR = 1.006, 95% CI = 0.997–1.008, p = 0.022) (Supplementary Table 10, Fig. 3). Moreover, GPA was potentially negatively associated with the risk of the genera Eubacterium oxidoreducens group (OR = 0.99, 95% CI = 0.987–1.004, p = 0.023), Ruminococcaceae UCG009 (OR = 0.992, 95% CI = 0.989–1.004, p = 0.037), Lachnospiraceae UCG004 (OR = 0.993, 95% CI = 0.992–1.002, p = 0.008), Roseburia (OR = 0.995, 95% CI = 0.993–1.002, p = 0.037), and Veillonella (OR = 0.991, 95% CI = 0.99–1.003, p = 0.008) (Supplementary Table 10, Fig. 3).

Potential causal interactions between the gut microbiota and immune-related vasculitis

Taken together, the positive association results from IVW analysis (p < 0.05) revealed numerous shared gut microbiota constituents associated with immune-related vasculitis. Figure 4 illustrated the intricate intersections among immune-related vasculitis subtypes using Venn diagrams. Specifically, the Venn diagrams highlighted the overlapping gut microbiota components that were common across different vasculitis subtypes, suggesting potential shared pathways in the disease mechanism. These intersections underscored the complexity of the gut microbiota's role in vasculitis pathogenesis. Consequently, an interaction diagram depicting the interplay of the gut microbiota in immune-related vasculitis was constructed (Fig. 5) to provide a clearer visual representation of these associations and their potential implications.

Figure 4.

Venn diagram of the interaction patterns of the gut microbiota in immune-related vasculitis. BD Behcet's disease, GCA giant cell arteritis; KD: Kawasaki disease, GPA granulomatosis with polyangiitis, O outcome, E exposure.

Figure 5.

Interaction diagram of the gut microbiota in immune-related vasculitis. BD Behcet's disease, GCA giant cell arteritis, KD Kawasaki disease, GPA granulomatosis with polyangiitis.

The gut microbiota positively interacts with at least three types of immune-related vasculitis

The gut microbiota, particularly Ruminococcaceae/Ruminococcus, exhibits prominent interactions with immune-related vasculitis. Notably, the abundance of the family Ruminococcaceae demonstrated a negative association with the risk of GPA, and KD exhibited a negative association with the risk of the family Ruminococcaceae. Additionally, the abundances of specific genera within Ruminococcaceae, such as Ruminococcaceae UCG011, Ruminococcus gnavus group and Ruminococcus 2, exhibited positive associations with the risk of BD and GCA. Conversely, the abundance of Ruminococcaceae NK4A214 group exhibited a negative association with the risk of GCA. Furthermore, KD, GPA and GCA were negatively associated with the risk of the genera Ruminococcus 2, Ruminococcus torques group, Ruminococcaceae UCG009, and Ruminococcaceae UCG010.

The second set of gut microbiota constituents interacting with immune-related vasculitis included Eubacterium, Lachnospira/Lachnospiraceae, and Holdemanella/Holdemania. Specifically, the abundance of the genus Eubacterium nodatum group exhibited a positive association with the risk of GCA, while KD demonstrated a positive association with the risk of the genus Eubacterium oxidoreducens group. Conversely, GPA was negatively associated with the risk of the genus Eubacterium oxidoreducens group. The abundances of the family Lachnospiraceae and genus Lachnospira are negatively associated with the risk of GPA. The abundances of the genera Lachnospira and Lachnospiraceae UCG004 demonstrated positive associations with the risk of KD and GCA, respectively. Conversely, GPA was negatively associated with the risk of the genus Lachnospiraceae UCG004. BD and GPA both showed positive associations with the risk of the genus Holdemanella. Additionally, the abundance of the genus Holdemania was positively associated with the risk of GCA.

The interaction of the gut microbiota with two types of immune-related vasculitis

BD was negatively associated with the risk of the phylum Verrucomicrobia, family Verrucomicrobiaceae, class Verrucomicrobiae, and order Verrucomicrobiales. Concurrently, the abundance of the phylum Verrucomicrobia demonstrated a negative association with the risk of GCA.

GCA and GPA

GPA was negatively associated with the risk of the genus Veillonella, and the abundance of the genus Veillonella was negatively associated with the risk of GCA. Simultaneously, the abundance of the family Porphyromonadaceae showed a negative association with the risk of GCA and GPA.

BD and GPA

The abundance of the class Melainabacteria demonstrated a positive association with the risk of BD; conversely, it exhibited a negative association with the risk of GPA. Similarly, the abundance of the family Streptococcaceae exhibited a negative association with the risk of GPA; conversely, BD showed a positive association with the risk of the family Streptococcaceae.

GCA and KD

GCA and KD were both negatively associated with the risk of the genus Alistipes, and GCA was also negatively associated with the risk of the family Victivallaceae. Conversely, the abundance of the genus Victivallis was positively associated with the risk of KD.

KD and GPA

Both KD and GPA were negatively associated with the risk of the genus Eubacterium eligens group, with KD also exhibiting negative associations with the risk of the order Actinomycetales, family Actinomycetaceae, genus Actinomyces and genus Butyricimonas. In contrast, the abundances of the family Actinomycetaceae and genus Butyricimonas were positively associated with the risk of GPA.

BD and KD

BD was positively associated with the risk of the order Lactobacillales and the genus Streptococcus. Conversely, KD was negatively associated with the risk of the genus Streptococcus. BD is also negatively associated with the risk of the genus Ruminiclostridium 6. The abundances of the genera Lactobacillus and Ruminiclostridium 9 were negatively associated with the risk of KD.

Discussion

This MR study represents the first examination of the potential causal association between the gut microbiota and immune-related vasculitis, which we expect will serve as a foundation for future longitudinal investigations of alterations in the microbiome preceding the onset of immune-related vasculitis. Unexpectedly, genetic predispositions to colonization with the class Melainabacteria, class Lentisphaeria, and family Actinomycetaceae demonstrated causal links with GPA. Additionally, we identified specific gut microbiota constituents that could serve as potential risk factors for immune-related vasculitis. These findings will support public health interventions aimed at mitigating the risk of immune-related vasculitis.

GCA, characterized by granulomatous inflammation in large and medium-sized vessels, primarily affects elderly patients³. No specific investigation has been dedicated to exploring the gut microbiome in GCA patients. However, studies focusing on the microbiome in the blood and aorta have highlighted distinctions in GCA patients. These distinctions include a decrease in Actinobacteria abundance and an increase in Proteobacteria abundance and minimal Bacteroidetes abundance compared to those in individuals with noninflammatory thoracic aortic aneurysms. Moreover, in contrast to patients with non-GCA temporal arteritis, GCA patients had variable proportions of Proteobacteria, Bifidobacterium, Parasutterella, and Granulicatella. Notably, there was variation in the abundance of Rhodococcus, an unidentified member of the family Cytophagaceae, in blood samples. Additionally, GCA patients exhibit dissimilarities in microbiome composition between the temporal artery and thoracic artery^12–15. Our study first revealed a negative association between the abundance of the genus Ruminococcaceae NK4A214 group and the risk of GCA. Ruminococcaceae NK4A214 group is linked to fiber degradation⁴⁶. Additionally, research has indicated its role in advancing type 2 diabetes in a Mexican cohort⁴⁷. Moreover, the abundance of Ruminococcaceae NK4A214 group effectively differentiated between chronic kidney disease patients and healthy controls⁴⁸. Moreover, our study identified a complex interaction network between immune-related vasculitis and Ruminococcaceae spp., suggesting the potential relevance of this genus to the onset and progression of immune-related vasculitis. Additionally, we identified other gut microbiota constituents, including the genera Veillonella, Ruminococcus 2, Flavonifractor, and Ruminococcus gnavus, as potential risk factors for GCA.

BD is an uncommon form of vasculitis, its etiology remains elusive, and it involves multiple organs. The disease is characterized by mucocutaneous symptoms such as oral and genital aphthosis and aseptic folliculitis. Additionally, patients may experience ocular complications such as uveitis, vascular issues leading to thrombosis, and further manifestations in the articular, gastrointestinal, and neurological systems⁴. Research on the role of the gut microbiome in BD patients surpasses that for other vasculitis types. 16S rRNA sequencing revealed that BD patients exhibit gut microbiota dysbiosis, which impacts intestinal immune function and influences BD progression. These associations were initially highlighted in mouse models. Ongoing clinical trials are exploring microbial therapies for BD, though the efficacy of dietary interventions remains unclear. Cross-sectional analyses revealed distinct gut bacterial profiles in BD patients, which were characterized by elevated abundances of lactic acid and sulfate-producing bacteria but reduced abundances of lactic butyric acid producers and methanogens^16–24. Notably, research has shown specific microbial shifts, such as increases in Tenericutes abundance and decreases in Deferribacteres and Verrucomicrobia abundance, in BD mice. Additionally, treatments with butyrate or Eubacterium rectale, a butyrate-producing bacterium, mitigated BD symptoms, suggesting therapeutic potential⁴⁹. According to reverse MR analysis, our study revealed that BD was positively associated with the risk of the family Streptococcaceae. Initially, the role of Streptococcus spp. in BD was investigated, given their prevalence in the oral bacterial community and association with oral diseases. Elevated abundances of distinct streptococcal serotypes have been observed in the oral mucosa of BD patients compared to those of healthy individuals^50,51. Further substantiating the pathogenic potential of Streptococcus in BD, mouse studies revealed that introducing Streptococcus sanguinis from BD patients led to the manifestation of BD-related symptoms⁵². Therefore, the aforementioned preliminary findings align with our results. Our study delineated a multifaceted interaction network between Streptococcus spp. and immune-related vasculitis, encompassing BD, KD, and GPA, indicating the potential involvement of these bacteria in the initiation and progression of these conditions. Furthermore, we identified additional gut microbiota constituents, such as the classes Melainabacteria and Gammaproteobacteria, familyRhodospirillaceae, and genera Odoribacter and Ruminococcaceae UCG011, that may serve as potential risk factors for BD.

KD is severe systemic vasculitis frequently associated with coronary artery aneurysms⁵³. The susceptibility of the intestinal microbiome to environmental factors is believed to influence KD development^54,55. Associations with KD have been identified for Bacteroidetes and Dorea⁵⁶, and Fusobacteria, Shigella, and Streptococcus have also been suggested as potential influencing factors⁵⁷. While the Ruminococcus abundance increases during the nonacute stages of KD, Streptococcus becomes more enriched in the acute phases⁵⁸. A comparison of microbial sequences from throat, rectum, and blood samples highlighted similarities between the blood and gut microbiota⁵⁹. In a murine model of exposure to a Lactobacillus cell wall extract, both bacteria and fungi were demonstrated to influence KD progression and severity⁶⁰. Antibiotic administration has also been associated with KD development⁶¹. After treatment with immunoglobulin/antibiotics, a decrease in the abundances of harmful bacteria and an increase in the abundances of beneficial bacteria have been observed⁵⁷. This microbial shift could enhance intestinal permeability, allowing intestinal pathogens to induce irregular immune reactions. Our study revealed that the abundance of the genus Lactobacillus was potentially negatively associated with the risk of KD. Moreover, KD was potentially negatively associated with the risk of the family Ruminococcaceae, genus Ruminococcus torques group, genus Ruminococcus 2, and genus Streptococcus. These findings diverge from those of the aforementioned observational studies, which we primarily attribute to the limited sample size of KD patients in our MR analysis. Consequently, the conclusions drawn regarding the causal association between KD and the gut microbiota should be interpreted as indicative rather than definitive, warranting further investigation in subsequent research.

GPA is a rare necrotizing vasculitis primarily affecting the upper and lower respiratory tracts and the renal system and involving small to medium-sized vessels⁹. Recent research has elucidated the association between GPA and the nasal microbiome. Rhee et al. observed fluctuations in the nasal microbiota of GPA patients, particularly in the Corynebacterium-to-Staphylococcus ratio, with distinct changes preceding disease relapses⁶². Using deep sequencing, another group discerned a microbial imbalance in GPA patients at the bacterial and fungal tiers, with immunosuppressive therapy associated with a more normalized profile⁶³. Wagner et al. identified enrichment of Staphylococcus aureus in patients with active GPA, contrasting with the prevalence of Staphylococcus epidermidis in patients with inactive GPA⁶⁴. Furthermore, Lamprecht et al. observed reduced microbiome diversity in GPA patients and revealed enhanced colonization of Staphylococcus aureus, along with pathogens such as Haemophilus influenzae and rhinoviruses, emphasizing the microbial dysbiosis⁶⁵. Together, these findings underscore the nuanced interplay between GPA and the nasal microbiome, emphasizing the potential roles of specific bacterial genera. Further investigations are essential to clarify the underlying mechanisms and possible therapeutic approaches. Niccolai et al. investigated the gut microbiota in eosinophilic granulomatosis with polyangiitis patients and identified an increase in the abundance of potential pathobionts, specifically Enterobacteriaceae and Streptococcaceae, during active disease⁶⁶. Concurrently, Yu et al. noted increased Actinomyces and Streptococcus abundances and reduced abundances of SCFA-producing taxa in microscopic polyangiitis patients⁶⁷. To date, no published study has explored the association between GPA and the gut microbiota. The present MR investigation was conducted based on the most recent GWAS datasets for GPA. The GWAS dataset for GPA included 135 cases and 456,213 controls from the publicly available GWAS catalog. Consequently, our study revealed that the abundances of the families Clostridiaceae 1 and Actinomycetaceae were positively associated with the risk of GPA. Moreover, the abundances of the classes Lentisphaeria, Melainabacteria, and Negativicutes and the families Lachnospiraceae, Porphyromonadaceae, Ruminococcaceae, and Streptococcaceae were negatively associated with the risk of GPA. This study represents the first exploration of the causal association between GPA and the gut microbiota on the bases of a comprehensive GWAS sample, yielding robust and credible findings. Through this research, we identified a shared gut microbiota signature, notably involving the Streptococcaceae family, between GPA and other AAVs.

The relationship between the gut microbiota and immune-related vasculitis highlights the potential regulatory role of the gut microbiome in these diseases. While there are some common microbiota associated with various forms of vasculitis, the specific associations often vary between diseases, indicating both shared and unique underlying mechanisms influenced by the gut microbial composition. For example, the family Ruminococcaceae exhibits distinct associations across diseases, showing a negative association with GPA but a positive association with BD and GCA. This underscores the importance of a detailed understanding of microbial taxa rather than generalized interpretations. Common microbiota such as Eubacterium and Lachnospiraceae are relevant in multiple vasculitis pathways, with certain genera playing possible proinflammatory roles, as seen with the Eubacterium nodatum group in GCA. Conversely, the negative associations of Verrucomicrobia and Veillonella with GCA suggest potential protective roles or altered states in disease contexts. The bidirectional relationships within the microbial ecosystem, exemplified by Actinomycetales and Actinomycetaceae, highlight their context-specific roles in different diseases, as they are negatively associated with KD and GPA but positively associated with other vasculitis. Moreover, the positive associations of BD with Lactobacillales and Streptococcus contrast with their negative associations with KD, emphasizing the importance of considering broader microbial interactions. In summary, while there are common microbiota patterns that correlate with various vasculitis diseases, these associations can differ significantly, necessitating further exploration into their causative, functional, and therapeutic implications in vasculitis.

The current study has several distinct strengths. First, this study provides the first application of a 2-sample bidirectional MR approach, probing the causal ties between the gut microbiota and immune-related vasculitis subtypes, namely, GCA, BD, KD, and GPA. This methodology diminishes issues such as reverse causation and confounding effects, which are often prevalent in observational studies. To enhance the robustness of our findings, we integrated multiple MR framework strategies, ensuring consistency in the results both pre- and postoutlier adjustments while also minimizing variability. For comprehensive genetic insights, we utilized an expansive GWAS dataset at the summary level. The observed disparities between exposure and outcome data further validate our conclusions.

Nevertheless, several limitations were evident in this research. Initially, the limited number of SNPs employed as IVs reduced overall statistical robustness. However, given that our F-statistics consistently surpassed 10, the risk of significant instrumental bias in our conclusions remained minimal. Underscoring the necessity for larger sample sizes in subsequent MR investigations. Second, the predominant European ancestry of our participants prohibits direct generalization to diverse racial or ethnic populations, emphasizing the need for replicative studies. Third, the GWAS data included in this study only captured differences in gut microbiota between vasculitis patients and healthy controls. The GWAS data did not perform subgroup analyses based on the clinical manifestations or specific organ involvement in BD or GPA patients. Therefore, we could not investigate the impact of gut microbiota on immune-related vasculitis affecting specific organs. Fourth, while MR was valuable for understanding the impact of genetic determinants, it had limitations in assessing the influence of non-genetic factors. This constituted a limitation of such studies, as they could not account for environmental, lifestyle, and other non-genetic influences that significantly contributed to the pathogenesis of immune-related vasculitis. Fifth, this research did not address the potential effects of treatments such as proton pump inhibitors on the gut microbiota. The inability to control for these treatment effects could impact the interpretation of our results. Finally, our analysis focused solely on bacterial taxa at the genus level, not achieving more granular classifications such as species or strains. Utilizing advanced shotgun metagenomic techniques in microbiota GWASs could increase the precision of the results.

Conclusions

In summary, our results substantiate the causal roles of the class Melainabacteria, Lentisphaeria and family Actinomycetaceae in GPA. Moreover, we identified distinct gut microbiota elements that might act as potential triggers for immune-related vasculitis. This research offers fresh perspectives on the mechanisms underlying the progression of gut microbiota-associated immune-related vasculitis.

Author contributions

Yuan H, Zeng X, and Chen S conceptualized and designed the study; Chen S retrieved the data; Chen S analyzed, interpreted, and drafted the article; Nie R, Wang C, Luan H, Xu M, and Gui Y revised the article; Chen S, Nie R, and Wang C generated the graphs and tables. All the authors approved the final version for submission.

Funding

This work was supported by funding from STI2030-Major Projects (2021ZD0200600, 2021ZD0200603), the National Key Research and Development Program (2022YFC2009600) (2022YFC2009602), the “Beijing Major Epidemic Prevention and Control Key Specialty Construction Project” (2022).

Data availability

This study utilized publicly available datasets, which were obtained from the MiBioGen database (https://mibiogen.gcc.rug.nl/), the GWAS catalog (https://www.ebi.ac.uk/gwas/), the IEU OpenGWAS database (https://gwas.mrcieu.ac.uk/) and the FinnGen consortium (https://www.finngen.fi/).

Competing interests

The authors declare no competing interests.

Supplementary Information

The online version contains supplementary material available at 10.1038/s41598-024-68205-0.