Department of Gastroenterology, Baoqing Middle Road, Hongqi Street, Daxiang District, Shaoyang, Hunan Province, 422000, China, Email: pq1232020@163.com

1.1. Background: The present Mendelian randomization (MR) study aimed to investigate the potential causal relationship between metabolites and inflammatory bowel diseases (IBD), including Crohn’s disease (CD) and ulcerative colitis (UC).

1.2. Methods: A comprehensive two-sample MR analysis of data from the FinnGen database determined the causal association between 1400 metabolite traits and IBD. Additionally, sensitivity analyses were used to verify the results’ robustness and horizontal pleiotropy.

1.3. Results: Twenty-nine metabolites demonstrated protective effects on CD and UC, respectively. Thirty and thirty-four metabolites showed risk effects on CD and UC, respectively. In addition, CD and UC significantly affected one metabolite, respectively.

1.4. Conclusions: Our study highlighted the relationship between metabolites and IBD while emphasizing their significance in the pathogenesis of CD and UC.

Inflammatory bowel diseases, Crohn’s disease, Ulcerative colitis, Metabolites, MR analysis

Inflammatory bowel disease (IBD) is a group of nonspecific chronic inflammatory diseases of the gastrointestinal tract with unknown etiology. It is classified as a type of digestive disorder and is characterized by severe diarrhea, electrolyte loss, bleeding, and abdominal pain [1]. The incidence of IBD is relatively high. IBD includes Crohn’s disease (CD) and ulcerative colitis (UC). The underlying physiological process involves the production of a large number of cytokines (inflammatory factors), immune complexes, or metabolites from intestinal microbiota, which activate neutrophils and macrophages, leading to the generation of excessive reactive oxygen species and causing cellular oxidative damage, resulting in chronic inflammatory injury to the intestines. Numerous studies have confirmed the association between IBD and factors such as increased intestinal permeability, genetics, ischemia, biochemical abnormalities, infectious and parasitic pathogens, food allergies, and adverse drug reactions [2, 3]. The most prominent clinical manifestations of IBD include vomiting, diarrhea, changes in appetite, weight loss, anorexia, ascites, and peripheral edema.

With in-depth research on gut microbiota, researchers have gradually realized that its metabolites play an important role as mediators of the interaction between gut microbiota and host in the development of IBD. Metabolites of gut microbiota can be detected in various tissues, including feces, serum, urine, liver, cerebrospinal fluid, and intestinal tissue [4]. A study in the United States conducted fecal non-targeted metabolomics and metagenomic sequencing on 121 IBD patients and 34 healthy controls. The results showed that the metabolic profile of IBD patients underwent extensive changes, with over 2700 differential metabolites, including various fatty acids, bile acids, amino acids, sheath lipids, etc. The changes in metabolites were associated with a correlation to IBD inflammation, indicating that metabolites of gut microbiota play a role in maintaining intestinal homeostasis. Based on current research results, there is a wide variety of metabolites in the gut microbiota, which may be involved in various complex pathological and physiological processes of the host. Clarifying the gut microbiota host metabolic signal is crucial for revealing the pathogenesis of IBD. Mendelian randomization (MR) is a statistical analysis technique that uses genetic variation as an instrumental variable (IV) to detect and quantify causal relationships. This method has gained popularity recently due to its ability to overcome potential confounding and reverse causal effects in observational research. The approach relies on three hypotheses, namely: (1) instrumental variables are strongly correlated with exposure factors, (2) instrumental variables are not correlated with confounding factors, and (3) instrumental variables are only correlated with exposure and outcomes. MR uses genotype instrumental variables to infer the relationship between exposure factors and outcomes. Due to its ability to avoid residual confounding factors, MR produces more reliable association results than observational studies or randomized controlled trials. More research is needed to utilize the MR method to assess the association between metabolites and IBD. This study aims to investigate the potential causal relationship between metabolite and IBD using a two-sample MR approach to establish a theoretical foundation for understanding the association between metabolite and the pathogenesis of IBD.

4.1. Study design We used two-sample MR analyses to investigate the causal relationship between 1400 metabolites and IBD, which includes UC and CD. The MR analysis relies on three core assumptions: association, independence, and exclusion. These are as follows: 1) instrumental variables must have a strong correlation with exposure factors, 2) instrumental variables should not be associated with any exposure outcome related to confounding factors, and 3) instrumental variables can only be influenced by exposure factors and outcome variables.

4.2. Genome-wide association study (GWAS) data sources We obtained data comprising approximately 19 million single nucleotide polymorphisms (SNPs) associated with IBD from data freeze 10 of the FinnGen study. This dataset included 2,033 and 5,931 patients and 409,940 and 405,386 healthy controls for CD and UC, respectively (https:// storage. Googleapis.com/finngen-public-data-r10/summary_stats/). Notably, all the cases included in this study were of European origin. A research study [5] was conducted on the Canadian Longitudinal Study on Aging (CLSA) cohort, which included 8,299 European individuals. The study involved a series of large GWASs that examined 1,091 metabolites and 309 metabolite ratios publicly available from the GWAS Catalog (accession numbers from GCST90199621 to GCST90201020) (Supplementary File 1). Of the 1,091 plasma metabolites tested, 850 belonged to eight super pathways: lipid, amino acid, xenobiotics, nucleotide, cofactor and vitamins, carbohydrate, peptide, and energy. The remaining 241 metabolites were either unknown or only partially characterized.

4.3. Selection of instrumental variables (IVs) To test the first hypothesis of MR, we selected SNPs strongly associated with each immune trait and have a P-value less than 1 × 10-5. We also ensured that their linkage disequilibrium (LD) r2 threshold is less than 0.001 within a distance of 10000 kb[6-8]. To test the second hypothesis of MR, we utilized the Phenoscanner database (http://www.phenoscanner. medschl.cam.ac.uk/) to ensure no associations between these SNPs and any known confounding factors. We excluded SNPs with significant heterogeneity and identified those significantly correlated with each metabolite. We used the F-statistic formula to measure the strength of each independent variable (IV). The formula follows: F = R2 (N-2) / (1-R2). Here, R2 denotes the proportion of the variability of physical activity explained by each IV, and N is the sample size of the GWAS for the SNP-physical activity association [9]. To calculate R2, we used the formula 2 x EAF x (1-EAF) x beta^2, where EAF represents the effect allele frequency, and beta is the standard error of the genetic effect [10] (Supplementary File 2)

Results of IVW method demonstrated significant protective effects of thirty metabolites on CD described as follows (Figure 1 and Supplementary File 3): Adenosine 5’−monophosphate (AMP) to EDT A ratio (OR: 0.6685, 95% CI: 0.4958−0.9013, P-value: 0.0082), 1−stearoyl−2− linoleoyl−gpc (18:0/18:2) levels (OR: 0.7249, 95% CI: 0.6024−0.8724, P-value: 0.006), Cortisone levels (OR: 0.7317, 95% CI: 0.5691−0.9407, P-value: 0.0148), 6−hydroxyindole sulfate levels (OR: 0.7330, 95% CI: 0.5486−0.9794, P-value: 0.0356), 4−methylhexanoylglutamine levels (OR: 0.7445, 95% CI: 0.5980−0.9271, P-value: 0.0084), X−25519 levels (OR: 0.7451, 95% CI: 0.5979−0.9285, P-value: 0.0088), Linoleoylcholine levels (OR: 0.7495, 95% CI: 0.5862−0.9583, P-value: 0.0215), Uridine levels (OR: 0.7500, 95% CI: 0.5957−0.9442, P-value: 0.0143), 5− methylthioadenosine (MTA) to phosphate ratio (OR: 0.7520, 95% CI: 0.6105−0.9262, P-value: 0.0073), Maleate levels (OR: 0.7561, 95% CI: 0.5990−0.9544, P-value: 0.0186), X−17354 levels (OR: 0.7609, 95% CI: 0.6147−0.9418, P-value: 0.0120), X−24243 levels (OR: 0.7769, 95% CI: 0.6392−0.9443, P-value: 0.0112), X−22834 levels (OR: 0.7809, 95% CI: 0.6390−0.9543, P-value: 0.0157), Carnitine levels (OR: 0.7911, 95% CI: 0.6842−0.9147, P-value: 0.0016), N−lactoyl tyrosine levels (OR: 0.7915, 95% CI: 0.6471−0.9681, P-value: 0.0229), Phenylpyruvate levels (OR: 0.7946, 95% CI: 0.6480−0.9744, P-value: 0.0272), Creatinine levels (OR: 0.7957, 95% CI: 0.6702−0.9447, P-value: 0.0091), Histidine to glutamine ratio (OR: 0.7974, 95% CI: 0.6423−0.9900, P-value: 0.0403), Adenosine 5’−monophosphate (AMP) to threonine ratio (OR: 0.7994, 95% CI: 0.6816−0.9374, P-value: 0.0059), Linoleoyl ethanolamide levels (OR: 0.8010, 95% CI: 0.6890−0.9311, P-value: 0.0039), X−13866 levels (OR: 0.8023, 95% CI: 0.6442−0.9993, P-value: 0.0492), Sphingadienine levels (OR: 0.8128, 95% CI: 0.6721−0.9829, P-value: 0.0326), N−oleoyltaurine levels (OR: 0.8176, 95% CI: 0.6810−0.9816, P-value: 0.0308),X−26111 levels (OR: 0.8201, 95% CI: 0.6865−0.9797, P-value: 0.0288), Phosphoethanolamine levels (OR: 0.8392, 95% CI: 0.7119−0.9893, P-value: 0.0368), Oleoyl ethanolamide levels (OR: 0.8414, 95% CI: 0.7204−0.9827, P-value: 0.0293), Betaine levels (OR: 0.8416, 95% CI: 0.7216−0.9816, P-value: 0.0280), Metabolonic lactone sulfate levels (OR: 0.8882, 95% CI: 0.8118−0.9718, P-value: 0.0098), Glucose to sucrose ratio (OR: 0.8939, 95% CI: 0.8001−0.9987, P-value: 0.0475), Oleoyllinoleoyl-glycerol (18:1 to 18:2) to linoleoyl-arachidonoyl-glycerol (18:2 to 20:4) ratio (OR: 0.8987, 95% CI: 0.8120−0.9946, P-value: 0.0389). In addition, significant risk effects of thirty metabolites on CD were observed as follows: Pipecolate levels (OR: 1.0524, 95% CI: 1.0021−1.1053, P-value: 0.0411), 1−arachidonoyl−gpc (20:4n6) levels (OR: 1.1318, 95% CI: 1.0152−1.2617, P-value: 0.0256), X−18345 levels (OR: 1.1518, 95% CI: 1.0038−1.3216, P-value: 0.0440), 2−hydroxyoctanoate levels (OR: 1.1534, 95% CI: 1.0017−1.3280, P-value: 0.0474), 3−methoxycatechol sulfate (1) levels (OR: 1.1845, 95% CI: 1.0085−1.3913, P-value: 0.0391), Adenosine 5’−diphosphate (ADP) to 2’−deoxyuridine ratio (OR: 1.1881, 95% CI: 1.0142−1.3919, P-value: 0.0328), Trans−2−hexenoylglycine levels (OR: 1.1945, 95% CI: 1.0012−1.4253, P-value: 0.0485), Suberate (C8−DC) levels (OR: 1.1966, 95% CI: 1.0039−1.4264, P-value: 0.0452), Kynurenate levels (OR: 1.1994, 95% CI: 1.0394−1.3842, P-value: 0.0128), Uridine to 2’−deoxyuridine ratio (OR: 1.2128, 95% CI: 1.0075−1.4598, P-value: 0.0414), S−adenosylhomocysteine (SAH) levels (OR: 1.2168, 95% CI: 1.0433−1.4193, P-value: 0.0124), Adipoylcarnitine (C6−DC) levels (OR: 1.2225, 95% CI: 1.0127−1.4758, P-value: 0.0365), Dopamine 4−sulfate levels (OR: 1.2258, 95% CI: 1.0467−1.4354, P-value: 0.0115), Oxalate (ethanedioate) levels (OR: 1.2268, 95% CI: 1.0031−1.5004, P-value: 0.0466), Succinate to acetoacetate ratio (OR: 1.2279, 95% CI: 1.0027−1.5037, P-value: 0.0470), Homostachydrine levels (OR: 1.2328, 95% CI: 1.0062−1.5105, P-value: 0.0434), X−24757 levels (OR: 1.2591, 95% CI: 1.0187−1.5563, P-value: 0.0330), Isovalerate (i5:0) levels (OR: 1.2676, 95% CI: 1.0164−1.5809, P-value: 0.0354), X−11847 levels (OR: 1.2841, 95% CI: 1.0403−1.5851, P-value: 0.0199), Imidazole propionate levels (OR: 1.2883, 95% CI: 1.0526−1.5768, P-value: 0.0140), X−12216 levels (OR: 1.3035, 95% CI: 1.0660−1.5939, P-value: 0.0098), X−24728 levels (OR: 1.3182, 95% CI: 1.1032−1.5751, P-value: 0.0024), Benzoate to oleoyl−linoleoyl−glycerol (18:1 to 18:2) [2] ratio (OR: 1.3193, 95% CI: 1.0514−1.6555, P-value: 0.0167), N−carbamoylalanine levels (OR: 1.3221, 95% CI: 1.0929−1.5993, P-value: 0.0040), Threonine to alpha−ketobutyrate ratio (OR: 1.3671, 95% CI: 1.0498−1.7803, P-value: 0.0203), X−18887 levels (OR: 1.3841, 95% CI: 1.0479−1.8282, P-value: 0.0221), 2−hydroxypalmitate levels (OR: 1.3890, 95% CI: 1.0322−1.8690, P-value: 0.0300), Heptenedioate (C7:1−DC) levels (OR: 1.4124, 95% CI: 1.0520−1.8963, P-value: 0.0216), 4−oxo−retinoic acid levels (OR: 1.4719, 95% CI: 1.1194−1.9354, P-value: 0.0056), Serine to pyruvate ratio (OR: 1.5566, 95% CI: 1.1799−2.0535, P-value: 0.0017). The reliability and validity of the causal relationships identified have been further supported by the findings from three different methods, namely the weighted median, MR Egger, and simple mode (Supplementary Figure 1 and Supplementary File 3), along with a leave-one-out sensitivity analysis (Supplementary File 4). The intercept of MR-Egger was analyzed to ensure the absence of horizontal pleiotropy (Supplementary File 5). The forest plots are shown in Supplementary File 6. The stability of the results was also indicated by scatter plots (Supplementary File 7) and funnel plots (Supplementary File 8).

Results of the IVW method demonstrated significant protective effects of twenty-nine metabolites on CD described as follows (Figure 2 and Supplementary File 3): 1−stearoyl−2−linoleoyl−gpc (18:0/18:2) levels (OR: 0.7926, 95% CI: 0.7029−0.8937, P-value: 0.0001), X−12680 levels (OR: 0.8039, 95% CI: 0.6991−0.9244, P-value: 0.0022), Salicyluric glucuronide levels (OR: 0.8182, 95% CI: 0.7250−0.9233, P-value: 0.0011), X−19438 levels (OR: 0.8297, 95% CI: 0.7530−0.9142, P-value: 0.0001), Glucose to maltose ratio (OR: 0.8485, 95% CI: 0.7371−0.9767, P-value: 0.0221), Adenosine 5’−diphosphate (ADP) to glycerate ratio (OR: 0.8596, 95% CI: 0.7678−0.9625, P-value: 0.0087), X−15461 levels (OR: 0.8609, 95% CI: 0.7757−0.9554, P-value: 0.0048), Phenylalanine to phosphate ratio (OR: 0.8637, 95% CI: 0.7752−0.9621, P-value: 0.0078), X−17351 levels (OR: 0.8678, 95% CI: 0.7666−0.9824, P-value: 0.0250), Phosphate to phosphoethanolamine ratio (OR: 0.8687, 95% CI: 0.7569−0.9971, P-value: 0.0453), Caffeine to linoleate (18:2n6) ratio (OR: 0.8732, 95% CI: 0.7834−0.9732, P-value: 0.0142), Dimethyl sulfone levels (OR: 0.8765, 95% CI: 0.7831−0.9811, P-value: 0.0219), N−acetylthreonine levels (OR: 0.8799, 95% CI: 0.7841−0.9873, P-value: 0.0295), Tyramine O−sulfate levels (OR: 0.8895, 95% CI: 0.8007−0.9883, P-value: 0.0293), 3−methoxycatechol sulfate (2) levels (OR: 0.8902, 95% CI: 0.7962−0.9954, P-value: 0.0412),

Retinol (Vitamin A) to linoleoyl−arachidonoyl−glycerol (18:2 to 20:4) [1] ratio (OR: 0.8925, 95% CI: 0.8101−0.9832, P-value: 0.0212), Nonanoylcarnitine (C9) levels (OR: 0.8927, 95% CI: 0.8283−0.9621, P-value: 0.0030), X−12731 levels (OR: 0.8938, 95% CI: 0.8189−0.9756, P-value: 0.0119), 3−ureidopropionate levels (OR: 0.8967, 95% CI: 0.8083−0.9946, P-value: 0.0392), Mannose to glycerol ratio (OR: 0.8976, 95% CI: 0.8078−0.9974, P-value: 0.0445), 1−oleoyl−2−linoleoyl−GPE (18:1/18:2) levels (OR: 0.9019, 95% CI: 0.8509−0.9559, P-value: 0.0005), Cortolone glucuronide (1) levels (OR: 0.9064, 95% CI: 0.8231−0.9982, P-value: 0.0458), Vanillic acid glycine levels (OR: 0.9126, 95% CI: 0.8493−0.9806, P-value: 0.0126), 1−stearoyl−2−linoleoyl−GPE (18:0/18:2) levels (OR: 0.9197, 95% CI: 0.8530−0.9915, P-value: 0.0291), 21−hydroxypregnenolone disulfate levels (OR: 0.9260, 95% CI: 0.8578−0.9996, P-value: 0.0487), 2’−o−methylcytidine levels (OR: 0.9306, 95% CI: 0.8839−0.9797, P-value: 0.0062), 3−aminoisobutyrate levels (OR: 0.9378, 95% CI: 0.8832−0.9957, P-value: 0.0356), Glyco− beta−muricholate levels (OR: 0.9409, 95% CI: 0.8936−0.9906, P-value: 0.0205), Decadienedioic acid (C10:2−DC) levels (OR: 0.9454, 95% CI: 0.8938−0.9999, P-value: 0.0496). In addition, significant risk effects of thirty-four metabolites on UC were observed as follows: 1−arachidonoyl− GPE (20:4n6) levels (OR: 1.0788, 95% CI: 1.0023−1.1611, P-value: 0.0433), X−24949 levels (OR: 1.0883, 95% CI: 1.0101−1.1725, P-value: 0.0262), Isovalerylglycine levels (OR: 1.0997, 95% CI: 1.0043−1.2041, P-value: 0.0402), 1−oleoylglycerol (18:1) levels (OR: 1.1022, 95% CI: 1.0023−1.2121, P-value: 0.0447), Gentisate levels (OR: 1.1038, 95% CI: 1.0019−1.2160, P-value: 0.0457), 1−(1−enyl−palmitoyl)−2−oleoyl−GPE (p−16:0/18:1) levels (OR: 1.1062, 95% CI: 1.0114−1.2098, P-value: 0.0272), 3−(3−hydroxyphenyl)propionate sulfate levels (OR: 1.1064, 95% CI: 1.0016−1.2221, P-value: 0.0465), Dihomo−linoleoylcarnitine (C20:2) levels (OR: 1.1081, 95% CI: 1.0012−1.2264, P-value: 0.0473), Adenosine 5’−diphosphate (ADP) to choline ratio (OR: 1.1090, 95% CI: 1.0079−1.2201, P-value: 0.0338), Adenosine 5’−diphosphate (ADP) to valine ratio (OR: 1.1099, 95% CI: 1.0135−1.2153, P-value: 0.0244), Arachidate (20:0) levels (OR: 1.1112, 95% CI: 1.0143−1.2174, P-value: 0.0235), Phosphate to threonine ratio (OR: 1.1135, 95% CI: 1.0159−1.2205, P-value: 0.0216), X−24728 levels (OR: 1.1182, 95% CI: 1.0201−1.2258, P-value: 0.0171), Etiocholanolone glucuronide levels (OR: 1.1186, 95% CI: 1.0317−1.2129, P-value: 0.0066), 1−arachidonoyl− gpc (20:4n6) levels (OR: 1.1238, 95% CI: 1.0501−1.2026, P-value: 0.0007), S−methylcysteine sulfoxide levels (OR: 1.1244, 95% CI: 1.0088−1.2534, P-value: 0.0342), Chiro−inositol levels (OR: 1.1266, 95% CI: 1.0102−1.2564, P-value: 0.0322), Adenosine 5’−diphosphate (ADP) to N−acetylglucosamine to N−acetylgalactosamine ratio (OR: 1.1274, 95% CI: 1.0156−1.2516, P-value: 0.0244), Oxalate (ethanedioate) levels (OR: 1.1281, 95% CI: 1.0146−1.2542, P-value: 0.0259), Arachidonate (20:4n6) to oleate to vaccenate (18:1) ratio (OR: 1.1286, 95% CI: 1.0351−1.2306, P-value: 0.0061), 1−stearoyl−2−docosahexaenoyl−gpc (18:0/22:6) levels (OR: 1.1356, 95% CI: 1.0417−1.2379, P-value: 0.0039), Uridine to 2’−deoxyuridine ratio (OR: 1.1402, 95% CI: 1.0328−1.2588, P-value: 0.0093), Arachidonoylcholine levels (OR: 1.1422, 95% CI: 1.0111−1.2902, P-value: 0.0325), Eicosenoate (20:1) levels (OR: 1.1451, 95% CI: 1.0070−1.3022, P-value: 0.0388), X−25343 levels (OR: 1.1499, 95% CI: 1.0181−1.2988, P-value: 0.0245), 3−methyladipate levels (OR: 1.1504, 95% CI: 1.0005−1.3228, P-value: 0.0491), Homovanillate (hva) levels (OR: 1.1531, 95% CI: 1.0437−1.2739, P-value: 0.0051), 1− (1−enyl−palmitoyl)−2−arachidonoyl−GPE (p−16:0/20:4) levels (OR: 1.1570, 95% CI: 1.0115−1.3235, P-value: 0.0334), X−24494 levels (OR: 1.1713, 95% CI: 1.0617−1.2922, P-value: 0.0016), Glutamate to alanine ratio (OR: 1.1727, 95% CI: 1.0279−1.3380, P-value: 0.0178), 1−(1− enyl−stearoyl)−2−arachidonoyl−GPE (p−18:0/20:4) levels (OR: 1.1823, 95% CI: 1.0700−1.3063, P-value: 0.0010), Cholesterol to cortisol ratio (OR: 1.1854, 95% CI: 1.0462−1.3431, P-value: 0.0076), 4−oxo−retinoic acid levels (OR: 1.2084, 95% CI: 1.0333−1.4132, P-value: 0.0178), Tetradecadienoate (14:2) levels (OR: 1.2804, 95% CI: 1.0853−1.5106, P-value: 0.0034). The causal relationships identified have been validated through three different methods: the weighted median, MR Egger, and simple mode (Supplementary Figure 2 and Supplementary File 3), along with a leave-one-out sensitivity analysis (Supplementary File 9). The absence of horizontal pleiotropy has been ensured by analyzing the intercept of MR-Egger (Supplementary File 5). The forest plots are shown in Supplementary File 10. The stability of the results has been indicated by scatter plots (Supplementary File 11) and funnel plots (Supplementary File 12). Therefore, the reliability and validity of the identified causal relationships have been further supported.

Subsequently, we conducted a reverse MR, using UC and CD as outcomes and metabolites with positive results found in the above analysis as exposures. We discovered that CD onset could decrease the level of X−24728 (Figure 3 and Supplementary File 3) (OR: 0.9498, 95% CI: 0.9096− 0.9918, P-value: 0.0195). We found that UC onset could decrease the level of Tetradecadienoate (14:2) (Figure 4 and Supplementary File 3) (OR: 0.9660, 95% CI: 0.9393−0.9934, P-value: 0.0155). Three different methods have been utilized to validate the causal relationships that have been identified. These methods are the weighted median, MR Egger, simple mode (Figure 3 and Figure 4), and sensitivity analysis (Supplementary File 13). The intercept of MR-Egger has been analyzed to ensure the absence of horizontal pleiotropy (Supplementary File 5). The forest plots are shown in Supplementary File 14. The stability of the results is indicated by scatter plots (Supplementary File 15) and funnel plots (Supplementary File 16). Therefore, the reliability and validity of the identified causal relationships have been further supported.

Our study evaluated the causal relationship between 1400 metabolites and the incidence of IBD by using the MR analysis of extensive publicly available genetic data. Results demonstrated significant causal effects of 60 and 63 metabolites on CD and UC, respectively. In addition, reverse MR results revealed that the onsets of UC and CD are each associated with one metabolite. Moreover, sensitivity analysis confirmed the relationship between identified metabolites and IBD. Amino acids (AA) have high nutritional value and are essential for intestinal growth and maintaining mucosal integrity and barrier function. AA reserves vary in different tissues and ecological niches [12]. Excessive dietary protein may produce potentially harmful bacterial metabolites in the intestine, affecting the repair of epithelial cells. Some of these bacterial metabolites can inhibit the respiration and proliferation of colonic epithelial cells, affecting barrier function [13]. Research [14] has shown that the metabolism of AA has a profound impact on cell function. Immune cells are dynamic when responding to infections and changes in the tissue environment, indicating that they heavily rely on metabolic states. Scoville et al. [15] have demonstrated that some AA and tricarboxylic acid cycle-related metabolites have undergone significant changes in CD patients. In clinical trials, Benjamin et al. [16] investigated the role of glutamine in treating active Crohn’s disease, and the results showed that glutamine and whey protein have practical effects in improving intestinal mucosal permeability and mucosal structure. A study by Singh et al. [17] adding arginine (Arg) to the diet resulted in better weight loss, shorter colon length, and less histological damage. The high Arg diet also increased intestinal microbiota diversity in mice, suggesting a protective effect on colitis models. A study [18] on colitis in mice treated with 5% acetic acid found that adding glycine (Gly) to the diet did not affect survival rate or colon length-to-weight ratio. However, Gly supplementation significantly reduced the expression of interleukin-1B and I-10 in colitis mice. Li et al.’s study[19] found that injecting glutamic acid increases cell proliferation and antioxidants and reduces inflammation in the colon mucosa. Tryptophan also reduces intestinal inflammation [20].

Uric acid is a powerful antioxidant that eliminates over half of the free radicals in the circulatory system [21]. Yun et al. [22] the intestine is essential for uric acid distribution and clearance in rats. In a mouse colitis model, brewing yeast increased uric acid levels in the intestine, which worsened colitis by increasing intestinal mucosal permeability. Guo et al. [23] used a mouse model of hyperuricemia to find that it can damage intestinal barrier function, increase intestinal permeability, and elevate serum TNF-α and IL-6. LV et al. [24] uric acid damages the intestinal barrier through a molecular mechanism involving the transport protein TSPO, activated by reactive oxygen species. This triggers the NLRP3 inflammasome, reduces the expression of tight junction proteins, and leads to intestinal epithelial dysfunction. The serum uric acid/creatinine ratio is a biomarker that reflects endogenous uric acid production in the body. It can accurately measure uric acid metabolism and eliminate the effects of different renal functions and nutritional states. A study conducted by Zhu et al. [25] the uric acid/creatinine ratio is associated with disease activity in CD patients but not in UC patients. The study also revealed that induction therapy significantly decreased the uric acid/creatinine ratio in CD patients with high uric acid/creatinine ratio and anti-brewing yeast antibody (ASCA) positivity. This suggests that uric acid may be a new indicator of CD disease activity, and elevated uric acid levels may increase the risk of CD. Additionally, studies have shown that uric acid is also a risk factor for UC. Tian et al. [26] discovered that serum uric acid levels in UC patients were significantly higher than those in healthy individuals and identified serum uric acid as an independent risk factor for UC. The above literature suggests that uric acid may be a risk factor for IBD, and its elevated levels have an impact on intestinal inflammation. We want to acknowledge a few limitations to our study. First, the study only included populations of European ancestry. This means we cannot establish genetic differences between ethnic groups, countries, and regions. Therefore, the results cannot be generalized. Second, the lack of comprehensive clinical data could have helped the feasibility of conducting subgroup analyses and impeded the determination of specific causal relationships.

Our study found causal relationships between metabolites and IBD, emphasizing their complex interactions. These insights could lead to early interventions and treatments for IBD.

The data supporting this study’s findings are openly available in the Finland database at https://www.finngen.fi/en/access_results. Further inquiries can be directed to the corresponding author

1. Liu Y, Wang X, Hu CA: Therapeutic Potential of Amino Acids in Inflammatory Bowel Disease. Nutrients 2017; 9(9).

2. Imhann F, Van der Velde KJ, Barbieri R, Alberts R, Voskuil MD, Vich Vila A, Collij V, Spekhorst LM, Van der Sloot KWJ, Peters V et al: The 1000IBD project: multi-omics data of 1000 inflammatory bowel disease patients; data release 1. BMC gastroenterology 2019;19(1):5.

3. Kaplan GG, Ng SC: Globalisation of inflammatory bowel disease: perspectives from the evolution of inflammatory bowel disease in the UK and China. The lancet Gastroenterology & hepatology 2016;1(4):307-316.

4. Lavelle A, Sokol H: Gut microbiota-derived metabolites as key actors in inflammatory bowel disease. Nature reviews Gastroenterology & hepatology 2020;17(4):223-237.

5. Chen Y, Lu T, Pettersson-Kymmer U, Stewart ID, Butler-Laporte G, Nakanishi T, Cerani A, Liang KYH, Yoshiji S, Willett JDS et al: Genomic atlas of the plasma metabolome prioritizes metabolites implicated in human diseases. Nature genetics 2023;55(1):44-53.

6. Auton A, Brooks LD, Durbin RM, Garrison EP, Kang HM, Korbel JO, Marchini JL, McCarthy S, McVean GA, Abecasis GR: A global reference for human genetic variation. Nature 2015; 526(7571):68-74.

7. Wang C, Zhu D, Zhang D, Zuo X, Yao L, Liu T, Ge X, He C, Zhou Y, Shen Z: Causal role of immune cells in schizophrenia: Mendelian randomization (MR) study. BMC psychiatry 2023; 23(1):590.

8. Noyce AJ, Kia DA, Hemani G, Nicolas A, Price TR, De PabloFernandez E, Haycock PC, Lewis PA, Foltynie T, Davey Smith G et al: Estimating the causal influence of body mass index on risk of Parkinson disease: A Mendelian randomisation study. PLoS medicine 2017; 14(6):e1002314.

9. Burgess S, Thompson SG: Avoiding bias from weak instruments in Mendelian randomization studies. International journal of epidemiology 2011; 40(3):755-764.

10. Papadimitriou N, Dimou N, Tsilidis KK, Banbury B, Martin RM, Lewis SJ, Kazmi N, Robinson TM, Albanes D, Aleksandrova K et al: Physical activity and risks of breast and colorectal cancer: a Mendelian randomisation analysis. Nature communications 2020; 11(1):597.

11. Burgess S, Thompson SG: Interpreting findings from Mendelian randomization using the MR-Egger method. European journal of epidemiology 2017; 32(5):377-389.

12. Behringer S, Wingert V, Oria V, Schumann A, Grünert S, CieslarPobuda A, Kölker S, Lederer AK, Jacobsen DW, Staerk J et al: Targeted Metabolic Profiling of Methionine Cycle Metabolites and Redox Thiol Pools in Mammalian Plasma, Cells and Urine. Metabolites 2019; 9(10).

13. Vidal-Lletjós S, Beaumont M, Tomé D, Benamouzig R, Blachier F, Lan A: Dietary Protein and Amino Acid Supplementation in Inflammatory Bowel Disease Course: What Impact on the Colonic Mucosa? Nutrients 201; 9(3).

14. Buck MD, Sowell RT, Kaech SM, Pearce EL: Metabolic Instruction of Immunity. Cell 2017, 169(4):570-586.

15. Scoville EA, Allaman MM, Brown CT, Motley AK, Horst SN, Williams CS, Koyama T, Zhao Z, Adams DW, Beaulieu DB et al: Alterations in Lipid, Amino Acid, and Energy Metabolism Distinguish Crohn’s Disease from Ulcerative Colitis and Control Subjects by Serum Metabolomic Profiling. Metabolomics : Official journal of the Metabolomic Society 2018; 14(1):17.

16. Benjamin J, Makharia G, Ahuja V, Anand Rajan KD, Kalaivani M, Gupta SD, Joshi YK: Glutamine and whey protein improve intestinal permeability and morphology in patients with Crohn’s disease: a randomized controlled trial. Digestive diseases and sciences 2012; 57(4):1000-1012.

17. Singh K, Gobert AP, Coburn LA, Barry DP, Allaman M, Asim M, Luis PB, Schneider C, Milne GL, Boone HH et al: Dietary Arginine Regulates Severity of Experimental Colitis and Affects the Colonic Microbiome. Frontiers in cellular and infection microbiology 2019; 9:66.

18. Wu X, Zheng Y, Ma J, Yin J, Chen S: The Effects of Dietary Glycine on the Acetic Acid-Induced Mouse Model of Colitis. Mediators of inflammation 2020, 2020:5867627.

19. Li TT, Zhang JF, Fei SJ, Zhu SP, Zhu JZ, Qiao X, Liu ZB: Glutamate microinjection into the hypothalamic paraventricular nucleus attenuates ulcerative colitis in rats. Acta pharmacologica Sinica 2014; 35(2):185-194.

20. Kim CJ, Kovacs-Nolan JA, Yang C, Archbold T, Fan MZ, Mine Y: l-Tryptophan exhibits therapeutic function in a porcine model of dextran sodium sulfate (DSS)-induced colitis. The Journal of nutritional biochemistry 2010; 21(6):468-475.

21. Alvarez-Lario B, Macarrón-Vicente J: Is there anything good in uric acid? QJM : monthly journal of the Association of Physicians 2011, 104(12):1015-1024.

22. Yun Y, Yin H, Gao Z, Li Y, Gao T, Duan J, Yang R, Dong X, Zhang L, Duan W: Intestinal tract is an important organ for lowering serum uric acid in rats. PloS one 2017; 12(12):e0190194.

23. Guo Y, Li H, Liu Z, Li C, Chen Y, Jiang C, Yu Y, Tian Z: Impaired intestinal barrier function in a mouse model of hyperuricemia. Molecular medicine reports 2019; 20(4):3292-3300.

24. Lv Q, Xu D, Ma J, Wang Y, Yang X, Zhao P, Ma L, Li Z, Yang W, Liu X et al: Uric acid drives intestinal barrier dysfunction through TSPOmediated NLRP3 inflammasome activation. Inflammation research : official journal of the European Histamine Research Society [et al] 2021; 70(1):127-137.

25. Zhu F, Feng D, Zhang T, Gu L, Zhu W, Guo Z, Li Y, Lu N, Gong J, Li N: Altered uric acid metabolism in isolated colonic Crohn’s disease but not ulcerative colitis. Journal of gastroenterology and hepatology 2019; 34(1):154-161.

26. Tian S, Li J, Li R, Liu Z, Dong W: Decreased Serum Bilirubin Levels and Increased Uric Acid Levels are Associated with Ulcerative Colitis. Medical science monitor : international medical journal of experimental and clinical research 2018; 24:6298-6304.

Qin Peng. Causal Role Of Metabolites In Inflammatory Bowel Disease: Mendelian Randomization (MR) Study. International Journal of Gastroenterology and Hepatology 2024.