Use of Mendelian Randomization to Unveil Metabolic Markers in Helicobacter pylori Infection Authors: Tung Hoang, Jeongseon Kim Published: 2026 ABSTRACT BackgroundEnzymes encoded by Helicobacter pylori are involved in various metabolic processes. Using a Mendelian randomization (MR) framework, we investigated the metabolites associated with six antibodies against H. pylori and explored the potential underlying biological pathways.MethodsA meta-analysis of 65 genome-wide association studies was conducted to assess the genetic predisposition to approximately 3,550 metabolites. Summary-level data for > 10 million genetic variants associated with H. pylori antibodies were extracted from the UK Biobank. MR analysis was performed using inverse-variance weighting, weighted median, and Egger regression to ensure robust findings. Significant metabolites were further analyzed using enrichment analysis to identify the relevant biological pathways.ResultsA total of 100 metabolites were positively associated with H. pylori antibodies (cytotoxin-associated gene A [CagA], 16; catalase, 25; GroEL, 16; outer membrane protein [OMP], 22; urease [UREA], 15; and vacuolating cytotoxin [VacA], 6). These metabolites were linked to pathways involving the metabolism of alanine, aspartate, glutamate, glycine, serine, threonine, purine, and tryptophan in four antibodies. Additionally, 67 metabolites were negatively associated with H. pylori antibodies (CagA, 11; catalase, 6; GroEL, 13; OMP, 11; UREA, 11; and VacA, 15). These metabolites were primarily involved in pyrimidine metabolism in the three antibodies.ConclusionOur study identified numerous metabolites linked to H. pylori antibody levels, indicating that metabolic alterations are associated with infection. These changes were particularly enriched in pathways involved in amino acid, nucleotide, and coenzyme metabolism and biosynthesis. These findings highlight the systemic metabolic impact of H. pylori infection and offer insights into the biological mechanisms underlying host-pathogen interactions. INTRODUCTION Helicobacter pylori is a spiral-shaped, gram-negative, microaerophilic bacterium that primarily colonizes the stomach and duodenum, contributing to various gastrointestinal diseases, such as gastritis, peptic ulcers, and neoplastic conditions.1 It possesses virulence factors that facilitate adhesion, translocation, inflammation, and infection in host epithelial cells, and genetic heterogeneity and polymorphisms influence its pathogenicity.23 Since 1994, the World Health Organization’s International Agency for Research on Cancer has classified H. pylori as a class I bacterial carcinogen, although only about 20% of infected individuals develop disease symptoms.4 The current prevalence of H. pylori infection is estimated to be 43.9% in adults and 35.1% in children and adolescents,5 although the rates in children may vary depending on the diagnostic method, with 28.6% detected through serology and 35.9% through urea breath tests or stool antigen tests.6 Several proteins have been reported to contribute to H. pylori infection, including cytotoxin-associated gene A (CagA), catalase, GroEL, outer membrane protein (OMP), urease (UREA), and vacuolating cytotoxin (VacA).478 Identifying circulating metabolic profiles is promising for detecting high-risk populations susceptible to diseases caused by H. pylori infection. Previous studies have demonstrated the impact of H. pylori on human metabolism, highlighting significant metabolites and disrupted metabolic pathways associated with infection, while suggesting a link between H. pylori and gastric cancer.9101112 However, limitations of observational study designs, including confounding factors and reverse causation, pose challenges in establishing causal relationships. Thus, further in-depth identification of metabolomic markers and evaluation of the etiology of H. pylori infection are crucial for informing high-risk populations of H. pylori-related conditions. As genome-wide association studies (GWAS) continue to expand, the application of the Mendelian randomization (MR) framework has been shown to overcome the shortcomings of observational study designs, such as confounding, reverse causation, and measurement error.13 Despite evidence linking H. pylori to metabolic alterations,14 no study has comprehensively assessed the metabolic profiles associated with H. pylori infection using an MR approach. Therefore, in this study, we used MR to systematically identify significant metabolites related to H. pylori antibodies and further explore the underlying biological pathways. METHODS Study design In this study, we investigated the genetic association between metabolic markers and H. pylori infection by using a two-sample MR framework with publicly available summary-level data. Genetic instruments for metabolite levels and estimates of associations between genetic variants and H. pylori antibody levels were obtained from large-scale GWAS. To ensure the validity of our findings, we adhered to three fundamental MR assumptions: 1) the relevance assumption, where the selected single nucleotide polymorphisms (SNPs) are strongly associated with the metabolic markers; 2) the independence assumption, which ensures that the SNPs are not linked to confounding factors; and 3) the exclusion restriction assumption, which requires that the SNPs influence H. pylori infection only through their effect on the metabolic markers.15 To specifically address the relevance assumption, we calculated F-statistics to confirm the strength of the selected instrumental variables (IVs) using the square of the ratio between the SNP and exposure effect estimate (β) and its standard error (SE).1617 Data sources Significant genetic variants for H. pylori antibodies Summary-level data for H. pylori infection were obtained from the GWAS Catalog website (https://www.ebi.ac.uk/gwas/), based on a previous genome-wide human leukocyte antigen association study conducted using UK Biobank data of more than 10 million SNPs.18 In particular, we extracted summary statistics for six anti-H. pylori seropositivity antibodies, including CagA (985 samples), catalase (1,558 samples), GroEL (2,716 samples), OMP (2,640 samples), UREA (2,251 samples), and VacA (1,571 samples).18 Significant genetic variants for metabolites To select the genetic instruments of the MR framework, we utilized summary statistics from the previous 65 GWAS of metabolomic traits (Table 1).1819202122232425262728293031323334353637383940414243444546474849505152535455565758596061626364656667686970717273747576777879808182 Overall, 353,753 significant SNPs linked to metabolites were identified in 65 previous GWAS. Of these, 74,145 SNPs were excluded due to missing beta coefficients and/or allele information, whereas 78,450 SNPs were removed due to missing SE values. Of the remaining 201,158 SNPs, the effect estimates of duplicated SNPs were pooled using random-effect models, ultimately yielding summary statistics for 24,960 SNPs. After excluding weak instruments with F-statistics ≤ 10 and then harmonizing with the summary statistics of H. pylori antibodies, more than 21,000 SNPs remained for MR analyses (Fig. 1). Table 1Description of genome-wide association studies for each phenotypePhenotypeData sourceSample sizeReferenceMetabolites249 metabolitesUK Biobank500,000 83 ProteinDIabetes REsearCh on patient straTification (DIRECT)3,029 19 2,319 metabolite featuresFINRISK02, Framingham8,738 20 LipidomeOld Order Amish650 21 355 lipid speciesKORA2,718 22 181 lipidsPROMIS + INTERVAL19,476 23 652 metabolitesAfrican American Study of Kidney Disease and Hypertension619 24 161 metabolitesHan Chinese cohort1,551 25 1,487 metabolitesGCKD cohort4,864 26 338 metabolitesWADRC and WRAP cohorts689 27 640 metabolitesHispanic Community Health Study/Study of Latinos5,435 28 174 metabolitesFenland cohort (EPIC-Norfolk) + INTERVAL30,138 29 476 metabolitesTwinsUK and SHIP-22,419 30 1,172 metabolitesGCKD cohort1,627 31 4 fatty acidsCHARGE consortium11,494 32 22 fatty acids and 15 fatty acid ratiosFramingham Heart Study Offspring Cohort2,374 33 102 lipid metabolitesAtherosclerosis Risk in Communities (ARIC)3,424 34 9 amino acidsMETabolic Syndrome In Men (METSIM)11,136 35 1,303 metabolitesQatari cohort996 36 72 lipids and lipoproteinsMETabolic Syndrome In Men (METSIM) (non-diabetes)8,372 37 25 peptidesAtherosclerosis Risk in Communities (ARIC)3,424 38 5 fatty acidsNutrition and Health of Aging Population in China (NHAPC), Multi-Ethnic Study of Atherosclerosis (MESA), Atherosclerosis Risk in Communities study (ARIC), Coronary Artery Risk Development in Young Adults study (CARDIA), Cardiovascular Health Study (CHS), Genetics of Lipid Lowering Drugs and Diet Network (GOLDN), Nurses’ Health Study (NHS), the Health Professionals Follow-Up Study (HPFS), and Invecchiare in Chianti Study (InCHIANTI)15,541 39 644 metabolitesThree clinics1,960 40 3 branched-chain amino acidsFenland cohort, KORA, TwinsUK16,596 41 70 amino acidsAtherosclerosis Risk in Communities (ARIC)3,424 42 4 caffeine metabolites and paraxanthine/caffeine ratioProspective Study of the Vasculature in Uppsala Seniors (PIVUS), Study of Health in Pomerania TREND (SHIP-TREND), Swiss Kidney Project on Genes in Hypertension (SKIPOGH), TwinGene, TwinsUK and Uppsala Longitudinal Study of Adult Men (ULSAM)9,876 43 217 metabolites and exomeFramingham Heart Study (FHS), Atherosclerosis Risk in Communities (ARIC) Study3,604 44 5,961 metabolic featuresUppsala Longitudinal Study of Adult Men (ULSAM), Prospective Investigation of the Vasculature in Uppsala Seniors (PIVUS), TwinGene3,738 45 16 carnitine-related metabolitesAtherosclerosis Risk in Communities (ARIC) study, TwinsUK cohort1,456 46 488 (Metabolon) and 160 (Biocrates) metabolitesTwinsUK cohort1,001 47 123 lipid and metaboliteGenome Center of University of Tartu Cohort (EGCUT), Erasmus Rucphen Family Study (ERF), Finnish Twin Cohort, a subsample of FINRISK 1997 (FR97), Genetic Predisposition of Coronary Heart Disease in Patients Verified with Coronary Angiogram (COROGENE), Genetics of METabolic Syndrome, Helsinki Birth Cohort Study (HBCS), Cooperative Health Research in the Region of Augsburg (KORA), Northern Finland Birth Cohort 1966 (NFBC 1966), FINRISK subsample of incident cardiovascular cases and controls (PredictCVD), EGCUT sub-cohort (PROTE), The Cardiovascular Risk in Young Finns Study (YFS). Leiden Longevity Study (LLS) and Netherlands Twin Register (NTR)24,925 48 Choline, trimethylamine N-oxide, betaine, homocysteine, methionine, dimethylglycine, sarcosine, glycine, citrulline, arginine, ornithine , ureaThe Cleveland Clinic GeneBank study4,280 49 9 fatty acidsNutrition and Health of Aging Population in China (NHAPC), Multi-Ethnic Study of Atherosclerosis (MESA), Atherosclerosis Risk in Communities Study (ARIC), the Coronary Artery Risk Development in Young Adults Study (CARDIA), the Cardiovascular Health Study (CHS), the Invecchiare in Chianti Study (InCHIANTI)21,349 50 63 metabolitesCATHGEN cohort3,512 51 26 amino acids, 36 acylcarnitines and 34 metabolite ratiosCentral European cohort2,107 52 55 metabolites + 53 metabolitesSHIP-0 cohort, KORA F4 cohort5,552 53 129 metabolitesTwinsUK, KORA, EGCUT, LLS, QIMR, ERF and NTR8,660 54 7 phospholipid fatty acidCoronary Artery Risk Development in Young Adults (CARDIA), Cardiovascular Health Study (CHS), Genetics of Lipid-Lowering Drugs and Diet Network (GOLDN), Health Professionals Follow-Up Study (HPFS), Multi-Ethnic Study of Atherosclerosis (MESA), Nurses’ Health Study (NHS), and Women's Genome Health Study (WGHS)8,013 55 HistidineAtherosclerosis Risk in Communities (ARIC) study, The Framingham Heart Study (FHS) Offspring Cohort, Cardiovascular Health Study (CHS), Rotterdam Study (RS), Jackson Heart Study (JHS)2,623 56 42 unique metabolitesErasmus Rucphen Family (ERF) study2,118 57 14 fatty acidsFramingham Heart Study (FHS) Offspring2,633 58 3 fatty acidsAtherosclerosis Risk in Communities (ARIC) study, Coronary Artery Risk Development in Young Adults (CARDIA) study, CHS, the Health Professionals Follow-up Study (HPFS), Multi-Ethnic Study of Atherosclerosis (MESA), Nurses’ Health Study (NHS), Women’s Genome Health Study (WGHS)10,129 59 5 n6 fatty acidsARIC, CARDIA, CHS, InCHIANTI, MESA8,631 60 529 metabolitesTwinsUK cohort, KORA7,824 61 308 metabolitesARIC (Atherosclerosis Risk in Communities)1,260 62 11 locus-metabolite associationsCohorte Lausannoise (CoLaus)1,436 63 TasteSensomics3 metabolitesAtherosclerosis Risk in Communities (ARIC) Study1,260 64 217 metabolitesThe Framingham Heart Study (FHS) Offspring cohort2,076 65 163 metabolitesKORA (Kooperative Gesundheitsforschung in der Region Augsburg)1,757 66 14 metabolitesRelationship between Insulin Sensitivity and Cardiovascular disease risk study (RISC), Botnia study, European network on Functional Genomics of type 2 diabetes (EUGENE2 consortium), Stanford Insulin Suppression Test (IST) Cohort,1,345 67 Meta-Analyses of Glucose and Insulin-related traits Consortium (MAGIC),DIAGRAM (Diabetes Genetics Replication and Meta-analysis Consortium)KORA (Cooperative Health Research in the Region of Augsburg) and UKtwins studiesInCHIANTI (Invecchiare in Chianti, aging in the Chianti area) study4 fatty acidsAtherosclerosis Risk in Communities (ARIC) Study, Cardiovascular Health Study (CHS), Coronary Artery Risk Development in Young Adults (CARDIA), Invecchiare in Chianti (InCHIANTI) Study, Multi-Ethnic Study of Atherosclerosis (MESA)8,961 68 2,264 metabolitesCancer of the Prostate in Sweden (CAPS)891 69 355 metabolitesKORA1,768 70 130 metabolitesCardiovascular Risk in Young Finns Study (YFS)6,608 71 Northern Finland Birth Cohort 1966 (NFBC66)115 phospholipid + 33 sphingolipidsErasmus Rucphen Family (ERF), MICROS, Northern Swedish Population Health Survey (NSPHS), Orkney Complex Disease Study (ORCADES), ROAS (CROATIA_Vis) study4,043 72 117 metabolitesNorthern Finland Birth Cohort 1966 (NFBC19668,330 73 The Cardiovascular Risk in Young Finns Study (YF)The Helsinki Birth Cohort Study (HBCS)Health2000 GenMets Study (GenMets)The Dietary, Lifestyle, and Genetic determinants of Obesity and Metabolic Syndrome study (DILGOM)117 metabolitesNorthern Finland Birth Cohort 1966 (NFBC1966)8,330 74 The Cardiovascular Risk in Young Finns Study (YF)Helsinki Birth Cohort Study (HBCS)The Dietary, Lifestyle and Genetic determinants of Obesity and Metabolic Syndrome (DILGOM)Health 2000 GenMets Study (H2000)4 metabolitesMolTWIN cohort, MolOBB cohort211 75 250 metabolitesKORA F4, TwinsUK2,820 76 4 n-3 polyunsaturated fatty acids (PUFAs)The Atherosclerosis Risk in Communities (ARIC)8,866 77 Cardiovascular Health Study (CHS)Coronary Artery Risk Development in Young Adults (CARDIA)InCHIANTIMulti-Ethnic Study of Atherosclerosis (MESA)59 metabolitesSHIP862 78 163 metabolites featureKORA, TwinsUK2,231 79 33 sphingolipidsERF, MICROS, NSPHS, ORCADES, Vis4,400 80 6 omega-3 and omega-6 fatty acidsInCHIANTI, Genetics of Lipid Lowering Drugs and Diet Network (GOLDN) Study2,286 81 363 metabolitesKORA F3 500K284 82 Description of genome-wide association studies for each phenotype Flowchart for identification of instrumental variables. SNP = single nucleotide polymorphism, GWAS = genome-wide association study, CagA = cytotoxin-associated gene A, OMP = outer membrane protein, UREA = urease, VacA = vacuolating cytotoxin. Statistical analysis Fig. 2 summarizes the different approaches used for MR analysis. In the primary analysis, we applied an inverse-variance weighted (IVW) meta-analysis with multiplicative random effects, assuming balanced pleiotropy across instruments.84 This method utilized SNP-specific Wald estimates derived by dividing the SNP-outcome association by the SNP-exposure association. These estimates were then combined using inverse-variance weighting, thus providing an efficient causal estimate under the assumption that all the instruments are valid. Conceptual framework of different approaches for MR analysis. SNP = single nucleotide polymorphism, IVW = inverse-variance weighted, WM = weighted median, MR = Mendelian randomization. To assess the robustness of our findings and account for potential violations of MR assumptions, we performed sensitivity analyses using the weighted median method85 and the MR-Egger regression.86 The weighted median method provides an estimate of whether at least 50% of the weight comes from valid instruments and offers robustness against horizontal pleiotropy. MR-Egger regression, which includes an intercept term, allows for the detection and correction of directional pleiotropy but is less statistically efficient than IVW.848586 All MR analyses were performed using R statistical software (R Foundation for Statistical Computing, Vienna, Austria).87 To further investigate the mechanisms linking metabolites and H. pylori antibodies, we performed enrichment analyses of H. pylori antibody-associated metabolic pathways. Using the list of significant metabolites, these analyses were conducted using the web-based tool MetaboAnalyst (version 6.0).88 Overrepresentation analysis was performed against 80 curated KEGG human metabolic pathways (updated December 2023).88 The analysis employed a hypergeometric test to evaluate whether specific pathways were enriched with input metabolites more than would be expected by chance, based on the background of all known human metabolites.88 Ethics statement This study utilized findings from published articles, thereby exempting them from ethical approval. RESULTS Genetically predicted metabolites and H. pylori antibody levels Table 2 presents the results of the two-sample MR analysis evaluating the relationships between metabolites and H. pylori antibody levels. In total, the associations between approximately 3,550 metabolites and six H. pylori antibodies were examined (Supplementary Tables 1, 2, 3, 4, 5, 6). Using the IVW method, we identified 135 metabolites linked to CagA (135 metabolites), catalase (105 metabolites), GroEL (107 metabolites), OMP (123 metabolites), UREA (105 metabolites), and VacA (115 metabolites). Table 2Number of significant metabolites associated with H. pylori antibody levels in according to different modelsH. pylori antibodiesNo. metabolic traits analyzedInverse-variance weightedWeighted medianMR-EggerTotalCagA3,5581351065416 increased and 11 decreased metabolitesCatalase3,562105945825 increased and 6 decreased metabolitesGroEL3,558107945416 increased and 13 decreased metabolitesOMP3,5611231027022 increased and 11 decreased metabolitesUREA3,559105855315 increased and 11 decreased metabolitesVacA3,56011587506 increased and 15 decreased metabolitesMR = Mendelian randomization, CagA = cytotoxin-associated gene A, OMP = outer membrane protein, UREA = urease, VacA = vacuolating cytotoxin. Number of significant metabolites associated with H. pylori antibody levels in according to different models MR = Mendelian randomization, CagA = cytotoxin-associated gene A, OMP = outer membrane protein, UREA = urease, VacA = vacuolating cytotoxin. Sensitivity analyses conducted using the weighted median and MR-Egger methods identified 106 and 54 metabolites associated with CagA levels, 94 and 58 metabolites linked to catalase levels, 94 and 54 metabolites associated with GroEL levels, 102 and 70 metabolites associated with OMP levels, 85 and 53 metabolites associated with UREA levels, and 87 and 90 metabolites related to VacA levels, respectively. A total of 16 metabolites were positively associated and 11 were negatively associated with CagA antibody levels across all three methods. Similarly, 25 metabolites showed a positive association and 6 a negative association with catalase antibody levels, 16 metabolites were positively linked, and 13 were negatively linked to GroEL antibody levels. Additionally, 22 metabolites exhibited a positive association and 11 exhibited a negative association with OMP antibody levels. Similarly, 15 metabolites were positively associated and 11 were negatively associated with UREA antibody levels, whereas 6 metabolites showed a positive association and 15 showed a negative association with VacA antibody levels. KEGG pathways linking metabolites and H. pylori antibodies Several biological pathways were identified through enrichment analysis using significant metabolites (Tables 3 and 4). In Particularly, metabolites positively associated with H. pylori antibodies were linked to several significant KEGG pathways, including arginine biosynthesis for CagA (P = 0.036), OMP (P = 0.045), and UREA (P = 0.036) antibody levels; fructose and mannose metabolism (P = 0.013); galactose metabolism (P = 0.018); and amino sugar and nucleotide sugar metabolism (P = 0.027) for VacA antibody levels (Table 3). In addition, metabolites negatively associated with H. pylori antibodies were linked to various significant KEGG pathways, including pantothenate and CoA biosynthesis for catalase (P = 0.013) and UREA (P = 0.026) antibody levels; β-alanine metabolism for catalase (P = 0.014) and UREA (P = 0.027) antibody levels; pyrimidine metabolism for catalase (P = 0.025), GroEL (P = 0.050), and UREA (P = 0.050) antibody levels; fructose and mannose metabolism for OMP (P = 0.039) and UREA (P = 0.003) antibody levels; and galactose metabolism for UREA levels (P = 0.035) (Table 4). Table 3KEGG pathways of metabolites positively associated with H. pylori antibodiesPathwayNo. of known metabolitesExpectaP valuebCagAArginine biosynthesis140.0360.036Alanine, aspartate and glutamate metabolism280.0730.071Porphyrin metabolism310.0810.078Tryptophan metabolism410.0940.090CatalaseAlanine, aspartate and glutamate metabolism280.0910.088Glycine, serine and threonine metabolism330.1070.103Tryptophan metabolism410.1330.126Purine metabolism700.2270.208Steroid hormone biosynthesis870.2830.253GroELFructose and mannose metabolism200.0520.068Galactose metabolism270.0700.068Glycine, serine and threonine metabolism330.0860.083Amino sugar and nucleotide sugar metabolism420.1090.105Purine metabolism700.1820.170Steroid hormone biosynthesis870.2260.208OMPArginine biosynthesis140.0460.045Alanine, aspartate and glutamate metabolism280.0910.088Glycine, serine and threonine metabolism330.1070.103Tryptophan metabolism410.1330.126Purine metabolism700.2270.208UREAArginine biosynthesis140.0360.036Alanine, aspartate and glutamate metabolism280.0730.071Tryptophan metabolism410.1070.102Purine metabolism700.1820.170VacAFructose and mannose metabolism200.0130.013Galactose metabolism270.0180.018Amino sugar and nucleotide sugar metabolism420.0270.027CagA = cytotoxin-associated gene A, OMP = outer membrane protein, UREA = urease, VacA = vacuolating cytotoxin.aExpected number of input metabolites that fall into the pathway under the null hypothesis of random chance, calculated based on the proportion of metabolites in the pathway relative to the entire known set.bProbability of observing as many input metabolites that fall into the pathway as expected by chance. KEGG pathways of metabolites positively associated with H. pylori antibodies CagA = cytotoxin-associated gene A, OMP = outer membrane protein, UREA = urease, VacA = vacuolating cytotoxin. aExpected number of input metabolites that fall into the pathway under the null hypothesis of random chance, calculated based on the proportion of metabolites in the pathway relative to the entire known set. bProbability of observing as many input metabolites that fall into the pathway as expected by chance. Table 4KEGG pathways of metabolites negatively associated with H. pylori antibodiesPathwayNo. of known metabolitesExpectaP valuebCatalasePantothenate and CoA biosynthesis200.0130.013β-alanine metabolism210.0140.014Pyrimidine metabolism390.0250.025GroELPyrimidine metabolism390.0510.050Purine metabolism700.0910.089OMPFructose and mannose metabolism200.0390.039Galactose metabolism270.0530.052Arginine and proline metabolism360.0700.069Amino sugar and nucleotide sugar metabolism420.0820.080Steroid hormone biosynthesis870.1700.160UREAFructose and mannose metabolism200.0260.026Pantothenate and CoA biosynthesis200.0260.026β-alanine metabolism210.0270.027Galactose metabolism270.0350.035Pyrimidine metabolism390.0510.050Amino sugar and nucleotide sugar metabolism420.0550.054VacALysine degradation300.0590.058Porphyrin metabolism310.0600.059Tryptophan metabolism410.0800.078OMP = outer membrane protein, UREA = urease, VacA = vacuolating cytotoxin.aExpected number of input metabolites that fall into the pathway under the null hypothesis of random chance, calculated based on the proportion of metabolites in the pathway relative to the entire known set.bProbability of observing as many input metabolites that fall into the pathway as expected by chance. KEGG pathways of metabolites negatively associated with H. pylori antibodies OMP = outer membrane protein, UREA = urease, VacA = vacuolating cytotoxin. aExpected number of input metabolites that fall into the pathway under the null hypothesis of random chance, calculated based on the proportion of metabolites in the pathway relative to the entire known set. bProbability of observing as many input metabolites that fall into the pathway as expected by chance. DISCUSSION Although evidence indicates that metabolomic signatures are associated with H. pylori infection, the exact cause and underlying mechanisms remain unknown. Here, we investigated genetically predicted metabolites and six H. pylori antibodies using MR. The results showed 100 positive and 67 negative associations between the metabolites and six H. pylori antibodies. Functional analyses further showed that these metabolites were significantly enriched in the biosynthesis and metabolism of amino acids (e.g., glycine, serine, threonine, and arginine), nucleotides (e.g., purine and pyrimidine), and coenzyme-related pathways (e.g., pantothenate and CoA). A previous study detected 26 metabolites as independent predictors of H. pylori infection, including 13 amino acids, 10 lipids, 1 carbohydrate, 1 partially characterized molecule, and 1 nucleotide.10 Among these, two metabolites were replicated in our current study, isobutyrylcarnitine (UREA) and etiocholanolone glucuronide (for VacA). First, the MR approach, compared with observational study designs, reduces bias and minimizes the risk of reverse causation. Observational studies are prone to confounding and cannot establish causality as effectively as MR. In addition, our study assessed various H. pylori antibodies rather than relying solely on IgG levels, thus providing a more comprehensive measure of bacterial infection. Moreover, our study screened a much larger set of 3,550 metabolites across six different H. pylori antibodies than the 615 metabolites analyzed in the previous untargeted approach. This broad analysis identified hundreds of significant metabolites, demonstrating the enhanced detection power of our approach. Metabolites linked to H. pylori may play a mediating role in the association between H. pylori infection and related conditions, particularly gastric cancer.10 A previous study highlighted adenine as a key mediator given its role in abnormal cell proliferation, while cancer cells require substantial amounts of adenine for ATP production and nucleic acid synthesis.1089 Given that adenine may be identified as a primary carcinogenic factor in gastric cancer, it can influence metabolic pathways in response to inflammation and promote precancerous lesions.1090 However, we were unable to confirm the genetically predicted adenine and H. pylori antibody levels owing to the lack of IVs. H. pylori has been classified as a group 1 carcinogen for the majority of non-cardia gastric cancers.91 Thus, identifying metabolic markers for H. pylori could provide potential biomarkers for gastric cancer, as H. pylori infection is known to influence metabolic changes. Other studies have investigated the association between metabolites and H. pylori infection.929394 Wu et al.92 found that H. pylori eradication was linked to changes in the pretreatment levels of three metabolites. In another study, the H. pylori eradication group had lower citrulline but higher serine and alanine concentrations than the non-eradication group.93 Additionally, Dai et al.94 identified several metabolites correlated with Helicobacter abundance, including 16 fatty acids, 24 glycerophospholipids, 15 amino acids, 10 nucleosides and nucleotides, and 9 carbohydrates. Six metabolites, including CagA, catalase, GroEL, OMP, UREA, and VacA, play distinct roles in H. pylori infection and pathogenicity. CagA, a well-known virulence factor, is injected into host cells, disrupting signaling pathways and contributing to gastric carcinogenesis.95 Catalase helps H. pylori evade host immune responses by neutralizing reactive oxygen species.96 GroEL, a heat shock protein, aids bacterial survival under stress conditions and may trigger immune responses. OMPs facilitate bacterial adhesion to gastric epithelial cells, thereby promoting colonization and persistence.97 UREA is crucial for H. pylori survival in the acidic stomach environment because it hydrolyzes urea into ammonia, which neutralizes gastric acid. Finally, VacA induces host cell vacuolation, disrupts immune responses, and contributes to tissue damage. Together, these metabolites interact in complex ways, influencing H. pylori infection, host immune response, and disease progression. Given the distinct characteristics of H. pylori antibodies, none of the metabolites in our study was associated with all six antibody levels. This suggests that further studies are needed to explore metabolite-antibody specificity and confirm metabolic signatures for H. pylori infection across different antibody responses. Regarding the underlying biological pathways linking metabolomic markers with H. pylori infection, our study is in line with previous studies on functional impacts, including glycine, serine, and threonine metabolism, and phenylalanine, tyrosine, and tryptophan biosynthesis.10 Serine-glycine-one-carbon metabolism is a critical metabolic pathway in cancer, providing one-carbon units for nucleotide synthesis, methylation reactions, and redox balance.9899100 Its dysregulation supports tumor growth by enhancing DNA replication, epigenetic modifications, and antioxidant defense, making it a potential target for cancer therapy.9899100 In addition, tyrosine, phenylalanine, and tryptophan metabolism have been implicated in gastroesophageal cancer, with alterations in these aromatic amino acids serving as potential biomarkers.101 Despite its strengths, this study had several limitations that need to be addressed. In this study, although the findings were derived using different MR analysis methods, none of the metabolites were consistently associated with all six antibodies examined. Future studies should aim to identify more robust metabolic markers by integrating larger datasets and refining the analytical approaches. Moreover, owing to the lack of full summary statistics for all metabolites, we were unable to confirm the effect of genetically predicted H. pylori infection on metabolites, which limits the interpretation of bidirectional MR associations. Future research should focus on specific candidate metabolites to establish stronger causal relationships. Furthermore, there is a potential limitation in applying our findings to populations with higher H. pylori prevalence, such as those in East Asia,102103 as our MR analyses were based primarily on European ancestry summary statistics. In particular, global variation in H. pylori prevalence arises from methodological heterogeneity, including differences in diagnostic tests, sampling strategies, definitions of infection status, and study populations,5 which may influence SNP-H. pylori associations. In addition, population-specific factors, such as diet, hygiene, and antibiotic resistance, further complicate comparability.104 In East Asia, resistance to metronidazole (≤ 51%) and clarithromycin (≤ 37%) is notably high, potentially affecting host-pathogen interactions and metabolomic responses.105106 Another contributing factor is that genetic susceptibility to H. pylori infection differs across ancestries. For example, variants in TLR1 have been suggested to be associated with infection risk in Europeans, but these associations may not replicate in other populations due to distinct genetic architectures.107 These considerations highlight a key limitation of our study and underscore the need for future MR studies that leverage ancestry-specific GWAS datasets, particularly from high-burden regions, to validate and extend the generalizability of our findings. In summary, we screened a large number of metabolites and identified hundreds of significant associations with H. pylori antibody levels. Pathway enrichment analysis revealed that these metabolites are involved in biologically relevant pathways, such as amino acid and nucleotide metabolism and coenzyme-related pathways. These findings highlight the potential role of metabolic regulation in host response to H. pylori infection.