Inflammatory Bowel Disease Is Associated With Prediagnostic Perturbances in Metabolic Pathways

Circulating metabolites integrate environmental, genetic, and microbiome data and can serve as illustrative proximal markers of complex diseases. Accordingly, metabolomics profiling has emerged as a novel approach for building mechanistic bridges that link diverse biological pathways to disease phenotype. Alterations in metabolomic profiles have been described in patients with established inflammatory bowel disease (IBD).1Aldars-García L. et al.Pharmaceuticals (Basel). 2021; 14: 1190Crossref PubMed Scopus (8) Google Scholar However, metabolomic studies to date have used postdiagnosis samples, where it is impossible to discriminate between signatures associated with IBD etiology and those resulting from IBD. Therefore, we aimed to examine the association between prediagnosis metabolites and metabolic pathways and risk of IBD. Our discovery cohort included 104 women with incident IBD (49 with Crohn’s disease [CD] and 55 with ulcerative colitis [UC]) with plasma samples collected at least 4 years before diagnosis and 208 matched control individuals nested within the prospective cohort studies of the Nurses’ Health Study (NHS) and NHSII.2Colditz G.A. et al.J Womens Health. 1997; 6: 49-62Crossref PubMed Google Scholar Our replication cohort included 25 CD and 25 UC patients with serum samples collected at least 4 years before diagnosis matched to 25 control individuals within the Proteomic Evaluation and Discovery in an IBD Cohort of Tri-service Subjects (PREDICTS) study.3Porter C.K. et al.Contemp Clin Trials Commun. 2019; 14100345Crossref PubMed Scopus (12) Google Scholar We evaluated metabolite-specific associations across 462 annotated metabolites using conditional logistic regression and conducted functional analysis of unannotated metabolites (n = 5623) using Mummichog.4Karnovsky A. Li S. Methods Mol Biol. 2020; 2104: 387-400Crossref PubMed Scopus (20) Google Scholar We adjusted for multiple comparisons using a permutation-based approach that accounts for the correlation structure of metabolites and false discovery rate for functional analysis.5Westfall P.H. Young S.S. Resampling-based multiple testing: examples and methods for p-value adjustment. Wiley, Hoboken, NJ1993Google Scholar FDR adjusted P values of <.15 were considered statistically significant, given the hypothesis-generating nature of this study. Further information on the study population, metabolomic assays, and statistical analysis is provided in the Supplementary Materials. Baseline characteristics of our discovery and replication cohorts are described in Figure 1A. In our discovery cohort, among annotated metabolites, we observed that higher prediagnostic levels of cortisone were associated with lower odds of developing CD (odds ratio [OR], 0.42 per standard deviation change of natural log–transformed metabolite; 95% confidence interval [CI], 0.25–0.70; Pcorrected = .09). In contrast, higher prediagnostic levels of phenylacetylglutamine and methionine sulfoxide were associated with a higher odds of CD with ORs (95% CIs) of 1.96 (1.29–3.00) and 2.13 (1.32–3.43), respectively (Pcorrected = .13 and .14, respectively). Although these associations were not statistically significant in PREDICTS, the magnitude and direction of the associations were similar for cortisone (OR, 0.68; 95% CI, 0.30–1.42; P = .31) and phenylacetylglutamine (OR, 2.30; 95% CI, 1.00–6.77; P = .08) but not for methionine sulfoxide (OR, 0.74; 95% CI, 0.27–1.89; P = .54). In contrast, we did not observe any significant metabolite-specific association with UC in our discovery cohort. Similarly, pathway analysis based on annotated metabolites did not reveal significant associations with the odds of developing CD or UC (all Pcorrected > .15). In our analysis of unannotated metabolites, we found several metabolic pathways to be significantly perturbated before clinical diagnosis of CD (Figure 1B). Notably, alterations in metabolic pathways involved in the biosynthesis of primary bile acids, the metabolism of amino acids and nitrogen, and steroid hormone biosynthesis were associated with CD in our discovery and validation cohorts (Figure 1B). We observed that most bile acids and steroid hormone biosynthesis metabolites were significantly associated with a lower odds of developing CD, whereas most metabolites involved in d-glutamine and d-glutamate, alanine, aspartate and glutamate, and nitrogen metabolism were associated with a higher odds of developing CD (Supplementary Table 1). For UC, both biosynthesis of unsaturated fatty acid and fatty acid biosynthesis pathways were significantly altered in prediagnosis plasma samples of patients with UC compared to control individuals (Pcorrected = 4.93 × 10–11 and .004, respectively (Figure 1B), a finding that was also replicated in PREDICTS (P values = .009 and .04, respectively) (Figure 1B). Most metabolites in these pathways were inversely associated with odds of developing UC (Supplementary Table 1). We report that the prediagnosis phases of CD and UC are associated with distinct and significant perturbances in metabolic pathways as measured by circulating metabolites. The majority of these metabolic pathways were also perturbed in an external cohort. To our knowledge, our study represents the first investigation of prediagnosis metabolomic changes in IBD. Consistent with our findings, studies using sera from newly diagnosed pediatric IBD patients and adults with established IBD have found alterations in amino acid metabolites and bile acid pathways in CD and in lipid and sphingolipid metabolism in UC.6Kolho K.-L. et al.J Crohns Colitis. 2017; 11: 321-334Crossref PubMed Google Scholar,7Scoville E.A. et al.Metabolomics. 2018; 14: 17Crossref PubMed Scopus (79) Google Scholar Our findings that a number of metabolic pathways are altered in the prediagnosis stage of CD and UC are also biologically plausible. In CD, steroid hormone biosynthesis pathways, which include cortisone, androgens, and estrogens, were consistently altered in NHS and PREDICTS, with most having an inverse relationship with disease risk. These results are in line with prior epidemiologic studies showing that exogenous (oral contraceptives) and endogenous (testosterone) hormones are associated with risk of CD.8Khalili H. et al.Gut. 2013; 62: 1153-1159Crossref PubMed Scopus (137) Google Scholar Bile acid metabolites may reflect alterations in diet or microbiome with secondary bile acids exerting anti-inflammatory effects in intestinal epithelial and lamina propria cells.9Thomas J.P. et al.Front Immunol. 2022; 13829525Google Scholar Amino acids, in particular glutamine, are implicated in tight junction regulation and barrier function, which has been shown to be altered in patients with preclinical CD.10Turpin W. et al.Gastroenterology. 2020; 159: 2092-2100.e5Abstract Full Text Full Text PDF PubMed Scopus (85) Google Scholar Finally, consistent with our findings, previous epidemiologic studies have shown an inverse association between higher intake of dietary fatty acid and risk of UC, likely through the anti-inflammatory effect of fatty acids on mucosal immunity.11Ananthakrishnan A.N. et al.Gut.. 2014; 63: 776-784Google Scholar The strengths of our study include the availability of blood samples before diagnosis of IBD in our discovery and validation cohort and the complementary nature of the cohorts, NHS and PREDICTS, with the former including older women and the latter including primarily younger men. Primary limitations of our study include the small sample size, which may have limited our ability to find more modest associations and provide more precise estimates. In addition, the heterogeneity between the 2 cohorts with regard to age and sex may have also influenced our ability to detect consistent associations between the cohorts. In summary, using 2 independent cohorts, we identified significant biologically plausible perturbation in metabolic pathways involved in the biosynthesis of primary bile acids, the metabolism of amino acids and nitrogen, and steroid hormone biosynthesis in CD and fatty acid biosynthesis in UC more than 4 years before diagnosis. Our findings highlight specific metabolic pathways perturbed years before IBD diagnosis that may be important in the development of clinical disease. Further characterization of these metabolic pathways and their associated metabolites may uncover novel biomarkers, provide insights into disease pathogenesis, and point to possible strategies for IBD prevention. Note: To access the supplementary material accompanying this article, visit the online version of Gastroenterology at www.gastrojournal.org, and at https://doi.org/10.1053/j.gastro.2022.09.007. Nurses’ Health Study/Nurses’ Health Study II group authors: Ashwin Ananthakrishnan,1,2 Raji Balasubramanian,3 Kristin E Burke,1,2 Prasanna Challa,1,2 Clary Clish,4 Amit Joshi,1,2 Emily Walsh Lopes,1,2 and James Richter1,2; 1Division of Gastroenterology, Massachusetts General Hospital, and Harvard Medical School, Boston, Massachusetts; 2Clinical and Translation Epidemiology Unit, Massachusetts General Hospital and Harvard Medical School, Boston, Massachusetts; 3University of Massachusetts Amherst School of Public Health, Amherst, Massachusetts; and 4Broad Institute of MIT and Harvard, Cambridge, Massachusetts. PREDICTS group authors: Jean-Frederic Colombel,1 Francesca Petralia,2 Manasi Agrawal,1 Joana Torres,1,3 Renee M. Laird,4 Mark S. Riddle,4,5 Joseph A. Murray,6 and Thierry Dervieux7; 1Henry D. Janowitz Division of Gastroenterology, Icahn School of Medicine at Mount Sinai, New York City, New York; 2Department of Genetics and Genomic Sciences, Icahn School of Medicine at Mount Sinai, New York, New York, New York; 3Hospital Beatriz Ângelo, Loures, Lisbon, Portugal; 4Enteric Diseases Department, Naval Medical Research Center, Silver Spring, Maryland; 5VA Sierra Nevada Health Care System, Reno, Nevada; 6Division of Gastroenterology and Hepatology, Mayo Clinic, Rochester, Minnesota; and 7Prometheus Laboratories, San Diego, California. Ryan C. Ungaro, MD, MS, (Conceptualization: Equal; Funding acquisition: Supporting; Investigation: Equal; Methodology: Equal; Writing – original draft: Equal). Xinwei Hua, PhD (Conceptualization: Equal; Formal analysis: Lead; Investigation: Equal; Methodology: Equal; Writing – original draft: Equal). Lauren M. Petrick, PhD (Formal analysis: Supporting; Investigation: Supporting; Methodology: Supporting; Writing – review & editing: Supporting). Andrew T. Chan, MD, MS (Funding acquisition: Supporting; Methodology: Supporting; Writing – review & editing: Supporting). Chad K. Porter, PhD (Investigation: Supporting; Methodology: Supporting; Resources: Supporting; Writing – review & editing: Supporting). Hamed Khalili, MD, MS (Conceptualization: Lead; Funding acquisition: Supporting; Investigation: Lead; Methodology: Lead; Writing – review & editing: Lead). Download .xlsx (.01 MB) Help with xlsx files Supplementary Table 1