Using Machine learning to identify metabolomic signatures of ulcerative colitis with the classification according to disease extent

Using Machine learning to identify metabolomic signatures of ulcerative colitis with the classification according to disease extent | Research Square window.SnipcartSettings = { analytics: { enabled: false } }; (function() { var accessVector = localStorage.getItem('access_vector') || ''; window.dataLayer = window.dataLayer || []; if (accessVector) { window.dataLayer.push({ user: { profile: { profileInfo: { snid: accessVector } } } }); } })(); (function(w,d,s,l,i){w[l]=w[l]||[];w[l].push({'gtm.start':new Date().getTime(),event:'gtm.js'});var f=d.getElementsByTagName(s)[0],j=d.createElement(s),dl=l!='dataLayer'?'&l='+l:'';j.async=true;j.src='https://www.googletagmanager.com/gtm.js?id='+i+dl;f.parentNode.insertBefore(j,f);})(window,document,'script','dataLayer','GTM-K279D39R'); Browse Preprints In Review Journals COVID-19 Preprints AJE Video Bytes Research Tools Research Promotion AJE Professional Editing AJE Rubriq About Preprint Platform In Review Editorial Policies Our Team Advisory Board Help Center Sign In Submit a Preprint Cite Share Download PDF Research Article Using Machine learning to identify metabolomic signatures of ulcerative colitis with the classification according to disease extent Changchang Ge, Yi Lu, Zhaofeng Shen, Yizhou Lu, Xiaojuan Liu, and 4 more This is a preprint; it has not been peer reviewed by a journal. https://doi.org/ 10.21203/rs.3.rs-4212589/v1 This work is licensed under a CC BY 4.0 License Status: Posted Version 1 posted You are reading this latest preprint version Abstract Background : Ulcerative colitis (UC) is a metabolic-related chronic intestinal inflammatory disease. Disease extent is an intrinsic aspect of UC and influences the treatment modality. Identifying serum metabolic profiling of different disease extent may help to elucidate the underlying molecular mechanisms of UC. Methods : UC patients with different disease extent and healthy controls were included in this study. Liquid chromatography-tandem mass spectrometry (LC-MS/MS) method detected the metabolites, and the orthogonal partial least squares-discriminant analysis (OPLS-DA) were performed to identify metabolites. Lasso regression and support vector machine-recursive feature elimination (SVM-RFE) were used to screen metabolites. Receiver operating characteristic curves (ROC) were plotted to represent the diagnostic value of metabolites. Results : 93 differential metabolites between the UC and HCs, and57, 66, 44 in E1 and E2, E1 and E3, E2 and E3 groups were confirmed by OPLS-DA in the positive and negative ion mode. KEGG pathway analysis show that Arachidonic acid metabolism, Glycerophospholipid metabolism and Pentose phosphate pathway were the main disturbed metabolic pathway related to UC. Glycerophospholipid metabolism, Sphingolipid metabolism, Pyrimidine metabolism and Linoleic acid metabolism are involved UC site phenotype. Lasso and SVM-RFE screened differential metabolites relating to disease extent, and 8 metabolites were confirmed. The area under the ROC curve of all metabolites predicting the disease extent is greater than 0.7. Conclusions : There are significant differences in metabolomics among UC patients with different disease extent. Tridecanoic acid may be a potential biomarker for identifying patients with extensive colitis. Ulcerative Colitis metabolomics machine learning disease extent Figures Figure 1 Figure 2 Figure 3 Figure 4 Figure 5 1.Introduction Ulcerative Colitis (UC) is a chronic non-specific intestinal inflammatory disease, a subtype of inflammatory bowel disease(IBD), characterized by recurrent diarrhea, bloody stools, and abdominal pain. Traditionally, UC was regrated as a disease of western nations. However, epidemiological studies suggest that incidence are rising in the developing Eastern Countries, which is closely to industrialization and the westernization of lifestyle and dietary habits.[1, 2]The cause of UC is unclear, but it is generally believed that environmental factors and intestinal microbial components can stimulate the immune response of genetically susceptible individuals, and interferes with energy metabolism in immune cells of gastrointestinal tract and intestinal microbial metabolism[3, 4]. Metabolomics is a high-throughput tool for quantitatively analyzing the small molecule metabolites in biospecimens such as blood, tissue, urine, or saliva, which gives us a chance to study the metabolic alterations of UC[5].Di'Narzo et al. analyzed the serum metabolites of UC and Crohn’s disease (CD) patients, and found that several metabolites can be used to distinguish IBD patients from healthy people, and associate with disease activity.[6] Huang et al. found that patients with IBD had metabolic pathway disorder before diagnosis. The changes of unsaturated fatty acid biosynthesis and fatty acid biosynthesis related metabolic pathways were associated with UC, and most of the metabolites were negatively related to the risk of UC[7].Metabolomics provide information on metabolite-concentration changes and have great potential in discovering possible biomarkers and related metabolic pathways, as well as exploring the complex pathophysiological mechanisms of diseases. A gene association analysis study in The Lancet pointed that there were significant differences in the genetics between UC, colon CD and ileal CD, and think that disease extent is an intrinsic aspect of a patient’s disease, in part genetically determined,[8].Atreya summarized from the aspects of clinical behavior, epidemiology, genetics, and intestinal microbial groups, demonstrated that colon CD showed overlap in the disease behavior of UC[9]. In addition, a large number of studies have shown that CD patients with different lesion sites have different responses to biological agents. Meta analysis results showed that in RCT study, all biological agents had better effect on colon type CD.[10, 11]This makes us pay attention to the important significance of lesion location in the clinical and mechanism research of IBD. Compared with CD, the disease extent of ulcerative colitis has been studied less. This is because the lesions of the UC are limited to the colon and are diffused, and the boundaries between different parts are not obvious. The disease extent of UC is also not static, but rather extends or shrinks with disease activity. However, in previous retrospective study, we found that there were differences in glucose, lipid and protein metabolism related indicators between distal colitis and extensive colitis under the condition of the same level of disease activity. In recent years, advanced machine learning algorithms have been widely used to screen medical biomarkers, and can effectively identify disease related metabolite markers. Binary logistics stepwise regression is the most commonly used regression method, but it is well known that the model obtained by it is not optimal. The least absolute shrinkage and selection operator (LASSO) algorithm is a machine learning method, which constructs a penalty function to obtain a more refined model, when fewer predictors are desired, LASSO regression is the preferred option to reduce model dimensionality[12]. Compared with classical statistical methods, LASSO regression has shown better performance for prediction model selection and better identification of predictors.[13–15] In addition, Support vector machine (SVM) is another machine learning algorithm, which adopts the principle of structural risk minimization, it can iteratively filter out a subset of features with the highest accuracy from a large amount of data, and thus was used to identify potential biomarkers of disease.[16] Therefore, we compared the serum metabolic profiles of healthy and UC patients using non-targeted metabolomics method to reveal the changes in the metabolic profiles of UC. More importantly, this study is the first to combine metabolomics techniques with machine learning algorithms to search for differential metabolites and metabolic pathways relating to the disease extent of ulcerative colitis, in order to provide evidence for the mechanisms study of ulcerative colitis. 2.Materials and Methods 2.1 Study populations A total of 219 individuals, between 18 and 75 years of age, were recruited. The participants comprised 189 UC patients and 30 healthy controls. UC patients were recruited from Jiangsu Provincial Hospital of Chinese medicine affiliated to the Nanjing University of Chinese Medicine and healthy controls were enrolled through advertisements. All UC patients were diagnosed according to the combination of UC clinical, endoscopic, and histopathological criteria. Montreal classification was used to assess the disease extent, patients with a total Mayo score < 3 or partial Mayo score < 2, with Mayo endoscopic subscore of 0–1 and inactive histological disease were used to judge whether the disease is active. Exclusion criteria were: pregnancy, history of colorectal surgery, active infections, and current malignancy. The healthy controls were selected, matched to the UC patients based on age and gender, and had no current or history of severe physical disease. 2.2 Chemicals, Reagents, and Equipment Methanol, acetonitrile, ammonium acetate (NH4Ac), and ethanoic acid (AcOH) were all liquid chromatograph-mass spectrometer-grade (LC-MS) and were purchased from CNW Technologies, SIGMA-ALDRICH and Fisher Chemical. All the experiments were conducted on an ultra-high performance liquid chromatography (UHPLC) system (Vanquish, Thermo Fisher Scientific) with a UPLC BEH Amide column (1.7 µm,2.1 mm × 100 mm) coupled to Orbitrap Exploris 120 mass spectrometer (Orbitrap MS, Thermo). The ACQUITY UPLC BEH Amide column (1.7 µm,2.1 mm × 100 mm) was purchased from Waters Corporation (Milford, MA, United States). 2.3 Serum Sample Collection and Metabolites Extraction Metabolomic analysis was conducted in serum samples, which were collected from all the participants. The serum was separated from the peripheral blood samples in EDTA tubes by centrifuging at 3,000 rpm for 10 min at 4°Candwas immediately stored at − 80°C until future metabolomics analysis to minimize the metabolic degradation process. 50µL of sample was transferred to an EP tube. After the addition of 200µL of extract solution (acetonitrile: methanol = 1: 1, containing isotopically-labelled internal standard mixture), the samples were vortexed for 30 s, sonicated for 10 min in ice-water bath, and incubated for 1 h at -40 ℃ to precipitate proteins. Then the sample was centrifuged at 12000 rpm(RCF = 13800(×g),R = 8.6cm) for 15 min at 4 ℃. The resulting supernatant was transferred to a fresh glass vial for analysis. The quality control (QC) sample was prepared by mixing an equal aliquot of the supernatants from all of the samples. 2.4 LC-MS/MS Analysis and Data preprocessing LC-MS/MS analyses were performed using an UHPLC system with a UPLC BEH Amide column (2.1 mm × 100 mm, 1.7µm) coupled to Orbitrap Exploris 120 mass spectrometer .The mobile phase consisted of 25 mmol/L ammonium acetate and 25 mmo spray Voltage as 3.8 kV (positive) or -3.4 kV(negative), respectively.l/L ammonium acetate in water(pH = 9.75)(A) and acetonitrile (B). The auto-sampler temperature was 4 ℃, and the injection volume was 2µL. The Orbitrap Exploris 120 mass spectrometer was used for its ability to acquire MS/MS spectra on information-dependent acquisition (IDA) mode in the control of the acquisition software (Xcalibur, Thermo). In this mode, the acquisition software continuously evaluates the full scan MS spectrum. The ESI source conditions were set as following: sheath gas flow rate as 50 Arb, Aux gas flow rate as 15 Arb, capillary temperature 320 ℃, full MS resolution as 60000, MS/MS resolution as 15000 collision energy as 10/30/60 in NCE mode, spray Voltage as 3.8 kV (positive) or -3.4 kV(negative), respectively. In this study, X peaks were detected and X metabolites were left after relative standard deviation de-noising. Then, the missing values were filled up by half of the minimum value. Also, total ion current normalization method was employed in this data analysis. The final dataset containing the information of peak number, sample name and normalized peak area was imported to SIMCA16.0.2 software package (Sartorius Stedim Data Analytics AB, Umea, Sweden) for multivariate analysis. Data was scaled and logarithmic transformed to minimize the impact of both noise and high variance of the variables. Then an in-house MS2 database (Biotree DB) was applied in metabolite annotation. The cutoff for annotation was set at 0.3. 2.5 Statistical Analysis In this study, OPLS-DA was applied to visualize group separation and screen for effective differential metabolites. To check the robustness and predictive ability of the OPLS-DA model, a 200 times permutation was conducted for validation of the classification performance of the OPLS-DA model with the fit metrics values of R 2 Y (p < 0.05) and Q2 (p 1.5 and p < 0.05(student t test) were considered as significantly changed metabolites, and Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway analysis were used to obtain the main disordered metabolic pathways. And then, differential metabolites were further screened by the LASSO algorithm from the “glmnet” package in the R software (version 4.2.1). At the same time, the SVM-RFE algorithm in the “mlbench” and “caret” package were also used. All distinctive features are fed to the SVM-RFE classifier in the form of 10-flod cross validation (CV) and the methods were set to “svmRadial”. The differential metabolites screened by the two machine learning algorithms are intersected to further narrow the range of metabolites. Receiver operating characteristic (ROC) curves were plotted using the “pROC” package by R, and the area under the curve (AUC) was used to represent the diagnostic value of metabolites. Figure 1 shows the overall process of the study. 3.Results 3.1 Clinical characteristics of study population A total of 189 patients with UC (152 active UC and 37 inactive UC) and 30 HCs were included in this study, and the active UC were divided into ulcerative proctitis (E1), left-sided colitis (E2) and extensive colitis (E3). HCs had a median age of 38.6 years, with 17 males and 13 females. Clinical characteristics of patients with UC are listed in Table 1 . We can see that there’s no difference in the baseline data of patients in each group, such as gender, age, severity, treatment, BMI and so on. Table 1 Clinical characteristics of patients with Ulcerative Colitis E1(N = 31) E2(N = 44) E3(N = 77) Inactive UC (N = 37) P- value * Male, n(%) 12(38.7%) 26(59.1%) 44(57.1%) 21(56.8%) 0.286 Age, y (median and range) 42(32–55) 41.5(34.25-52) 44(29-49.5) 44(36–51) 0.781 Disease activity, n(%) Remission 37 Mild 23(74.2%) 20(45.5%) 44(57.1%) 0.113 Moderate 8(25.8%) 21(47.7%) 27(35.1%) Severe 0 3(6.8%) 6(7.8%) Treatment, n (%) 5-Aminosalicylic Acid 29(93.5%) 43(97.7%) 76(98.7%) 33(89.2%) 0.095 Corticosteroid 1(3.2%) 3(6.8%) 4(5.2%) 0 0.455 Immunosuppressant 0(0.0%) 1(2.3%) 2(2.6%) 0 0.630 Biological agents 1(3.2%) 2(4.5%) 2(2.6%) 0 0.645 BMI, kg/m 2 (mean ± SD) 22.48(3.01) 22.24(2.69) 21.33(2.95) 22.07(3.67) 0.359 Smoker, n(%) 4(12.9%) 7(15.9%) 6(7.8%) 9(24.3%) 0.112 *We used Kruskal-Wallis tests and ANOVA for numerical data, and χ2 test for categorical data. 3.2 Metabolic differences between UCs and HCs Figure 2 A shows the visual and intuitive results of the supervised OPLS-DA analysis on metabolic profiles in the positive and negative ion mode, and the OPLS-DA model displayed obvious discrimination between the UC and HC groups. The permutation plot test of OPLS-DA model for the UC and HC groups was reliable and robust due to avoiding overfitting (Fig. 2 B). The specific VIP score and p-value of each abovementioned metabolite is presented in Supplementary table 1 . Additionally, for each group of comparison, we calculate the corresponding ratio of the quantitative value of the differential metabolite, and take log2 fold change. The metabolites with VIP > 1.5 and p < 0.05 were considered as significantly changed metabolites. Finally, 64 differential metabolites were screened in the positive ion mode, 29 differential metabolites were screened in the negative ion mode, and the cluster analysis heat map of 93 metabolites was shown on the right. Figure 2 shows the results of the cluster analysis of the differential metabolites between the UC and HC groups. KEGG pathway analysis show that Arachidonic acid metabolism, Glycerophospholipid metabolism and Pentose phosphate pathway were the main disturbed metabolic pathway related to UC. (Fig. 2 D) 3. 3 Metabolic differences between UC patients with different disease extent Next, we focused on analyzing the differences in serum metabolic profiles of UC patients with different disease extent. Figure 3 shows OPLS-DA score scatter plot and permutation plot test of OPLS-DA under positive and negative ion mode, respectively. There’re 44, 53 and 32 differential metabolites obtained from OPLS-DA model under positive ion mode, and13,13 and 12 differential metabolites under negative ion mode in E1 and E2, E1 and E3, E2 and E3 groups. The specific VIP score and p-value of each abovementioned metabolite presented in Supplementary table 1 . In order to further explore the differences of serum metabolic profiles in patients with UC at different disease extent, we performed KEGG enrichment analysis. (Supplementary table 2 ). In Table 2, we listed all metabolic pathways with P < 0.05. Notably, the metabolites are mainly involved Glycerophospholipid metabolism, Sphingolipid metabolism, Pyrimidine metabolism and Linoleic acid metabolism. The KEGG Enrichment bubble were showed in Fig. 4 A. Table2 Significantly changed pathways in patients with UC at different lesion sites Groups Pathway name Total Hits Raw p FDR E1 vs E2 Glycerophospholipid metabolism 36 3 0.001672 0.14046 Sphingolipid metabolism 21 2 0.008937 0.37535 Pyrimidine metabolism 39 2 0.029406 0.7356 Linoleic acid metabolism 5 1 0.035028 0.7356 E1 vs E3 Sphingolipid metabolism 21 2 0.002539 0.21329 Linoleic acid metabolism 5 1 0.01923 0.80767 alpha-Linolenic acid metabolism 13 1 0.049357 1 E2 vs E3 Glycerophospholipid metabolism 36 5 1.03E-07 8.68E-06 Sphingolipid metabolism 21 2 0.003526 0.14809 Linoleic acid metabolism 5 1 0.022406 0.62738 3.4 Identification of key metabolites by machine learning In order to obtain the key metabolites related to the disease extent of ulcerative colitis, we used two machine learning algorithms, lasso and SVM-RFE, to further screen 57, 66 and 44 differential metabolites obtained from OPLS-DA model. In the above three groups of patients, LASSO regression screened 17, 10 and 9 differential metabolites respectively, and the SVM-REF algorithm generated 57, 3 and 9 outputs respectively. The screening results of the two machine learning algorithms were intersected to obtain 17, 2 and 3 differential metabolites respectively (Fig. 4 B and C). We can see that the E1 group and E2 group compared to screen out 17 different metabolites. Due to the large number of metabolites, we selected the three most significant metabolites according to the results of SVM-REF, which are PS(22:2(13Z,16Z)/22:0), Allocholic acid and Cholic acid. Compared with group E1 and group E3, two main differential metabolites were screened, which were Tridecanoic acid and PC(20:0/P-18:1(11Z)). Compared with group E2 and group E3, three main metabolites were screened, which were Pipericine, Pelargonic acid and Tridecanoic acid. Then, we drew the ROC curves of the above differential metabolites respectively to further clarify their potential value in predicting the location of UC lesions. Figure 5 shows the inter-group comparison box chart of the above differential metabolites, and the area under the ROC curve of all metabolites predicting the disease extent is greater than 0.7. It is noteworthy that the serum content of Tridecanoic acid in E3 group is significantly lower than that in E1 and E2 groups, and its area under the ROC curve can reach 0.84, which may be a potential biomarker for identifying patients with extensive colitis. Discussion Ulcerative colitis is a metabolic related inflammatory disease. It is widely believed that genetic, environmental, dietary, psychological and other factors interact to cause intestinal microbial metabolism disorder, and participate in the complex mechanism of IBD.[17] Previous studies have also confirmed that UC patients have obvious metabolic disorders compared with healthy people.[18] Our study found that arachidonic acid metabolism, glycerophospholipid metabolism and Pentose phosphate pathway were disordered in UC patients, and arachidonic acid metabolism is the most obvious metabolic pathways, which was consistent with the previous research. Previous studies have observed an increase in the arachidonic acid composition of phospholipids in the colon mucosa of patients with active ulcerative colitis[19]. In addition, Kun Liu et al. applied metabolomics techniques to identify biomarkers for efficacy evaluation of Mesalazine in the treatment of ulcerative colitis and found that IBD dysregulation was related to the biosynthesis of primary bile acid, the metabolism of arachidonic acid, the metabolism of sphingolipid, fatty acid elongation, and glycerophospholipid metabolism[20]. Studies have found that sulfasalazine, 5-aminosalicylic acid and corticosteroids can control intestinal inflammation by regulating arachidonic acid metabolism, the mechanism may be related to the synthesis and release of prostaglandins and leukotrienes[21]. An animal experimental study also found that the traditional Chinese medicine prescription - Huanglian Jiedu Decoction can alleviate UC of mice by regulating arachidonic acid metabolism and glycerophospholipid metabolism[22]. Therefore, arachidonic acid metabolism plays an important role in the study of UC, and the results of this study also support this view. Clinically, the disease activity of patients greatly affects treatment decisions and is closely associated with adverse outcomes in UC, therefore, a large number of clinical studies focus on disease activity and neglect the value of disease extent in the study of UC. Recent mechanistic studies of IBD, particularly genotypic studies targeting patients with CD at different sites, have made us realize that the disease extent is also an intrinsic aspect of UC and has important clinical significance. In this study, we explored the metabolic profiling and potential biological correlates of UC with difference disease extent through the approach of combining metabolomics and machine learning. The OPLS-DA model preliminarily screened the different metabolites between groups, and pathway analysis shows that Glycerophospholipid metabolism, Sphingolipid metabolism, Pyrimidine metabolism and Linoleic acid metabolism are involved in the formation of ulcerative colitis site phenotype. At present, the most studied metabolism pathway of UC is Fatty acid metabolism (FAM), especially short-chain Fatty acid metabolism, which have been proved to be closely related to the pathogenesis of UC. Patients with UC have alterations in FAM genes that impair this metabolic pathway[23], which increases colonic permeability and promotes intestinal immune inflammation[24, 25]. It has been suggested that short chain fatty acid (SCFA) is an important energy source for colonic mucosal cells, especially in the distal colon[26]. Our previous retrospective study also found that the levels of low-density lipoprotein cholesterol, high-density lipoprotein cholesterol, apolipoprotein A1, apolipoprotein E and total cholesterol in patients with distal colitis were significantly higher than those with extensive colitis. However, no other studies have linked the above pathways to the disease extent of UC. The relationship between metabolic pathways and the disease extent of UC requires further research, which will help to refine the study of UC site phenotypes, facilitate drug development and guide clinical drug use. Due to the large number of metabolites involved in the metabolic pathway, our research further screened the differential metabolites through machine learning algorithm, and finally obtained 8 new biomarkers related to the lesion site. After analysis, it was found that the serum content of Tridecanoic acid in E3 group is significantly lower than that in E1 and E2 groups, and its area under the ROC curve can reach 0.84, which may be a potential biomarker for identifying patients with extensive colitis. Tridecanoic acid, a saturated fatty acid of medium-chain length, has been little studied. Some studies have found that tridecanoic acid can inhibit E. coli persistent cell formation and repressed biofilm activity[27]. Another study found that the content of tridecanoic acid in lung cancer patients was significantly lower than that in the healthy control group. The accuracy rate of identifying lung cancer patients with tridecanoic acid combined with other four metabolites can reach 82.9%[28]. This study confirmed that the serum metabolomics of UC patients with different disease extent were different under the same disease activity. Although we do not have a good explanation for this discrepancy, we think it can be considered in two ways. First, this difference may be caused by the different range of intestinal inflammation. This suggests that the above metabolites can be helpful to predict the range of intestinal lesions in UC and predict the disease progression. Second, these differences may be related to the intrinsic molecular structure of the colon and the molecular mechanism of intestinal inflammation in UC. In clinical practice, many drugs for treating UC are selective to the target, and biological agents are representative drugs. If we can dig deeply into the molecular differences of UC patients in different parts, it will be helpful for clinical precision medicine. Of course, there are some defects in this study. Our study is the first one to study the difference of metabolites in different disease extent, and the results lack evidence support. Additionally, this study only used serum samples for metabolomics detection, subsequent multi-sample and multi-omics studies are required for subsequent cross-validation. Therefore, we plan to collect intestinal mucosa samples from patients with UC at different sites to further validate our findings. Abbreviations UC：Ulcerative Colitis IBD：Inflammatory bowel disease CD ：Crohn’s disease VIP：Vasoactive intestinal peptide QC：Quality Control CV：Coefficient of variation OPLS-DA：Orthogonal Partial Least Squares-Discriminant Analysis LC-MS/MS: Liquid chromatography-tandem mass spectrometry ROC: Receiver operating characteristic curves LASSO: Least absolute shrinkage and selection operator SVM: Support vector machine LC-MS: Liquid chromatograph-mass spectrometer-grade AUC: Area under the curve FAM: Fatty acid metabolism SCFA: short chain fatty acid Declarations Ethics approval and consent to participate The study was approved by the independent Ethics Committee of Jiangsu Provincial Hospital of traditional Chinese medicine affiliated to the Nanjing University of Chinese Medicine (ethics number:2019NL-097-08); all participants provided informed consent. Consent for publication Not applicable. Availability of data and materials The data supporting the conclusions of this article is included within the article. Competing interests The authors report no conflict of interest. Funding This work was supported by a grant from the National Key Research and Development Program of China [grant number 2017YFC1700104] and Jiangsu Province Traditional Chinese Medicine Science and Technology Development Plan [grant number ZT202103]. Authors' contributions G.C.C. and L.Y. drafted the manuscript. S.Z.F. analyzed and interpreted all data. L.Y.Z., L.X.J., Z.M.Y. and L.Y.J. collected all data. Z.L. and S.H. conceived the study. All authors have approved the final draft of the manuscript. Acknowledgements We wish to thank the technical support provided by Shanghai Biotree Biomedical Technology Company. References Ng SC, Shi HY, Hamidi N, Underwood FE, Tang W, Benchimol EI, et al. Worldwide incidence and prevalence of inflammatory bowel disease in the 21st century: a systematic review of population-based studies. The Lancet. 2017;390:2769–78. Zhao M, Gönczi L, Lakatos PL, Burisch J. The Burden of Inflammatory Bowel Disease in Europe in 2020. Journal of Crohn’s and Colitis. 2021;15:1573–87. Chang JT. Pathophysiology of Inflammatory Bowel Diseases. N Engl J Med. 2020;383:2652–64. Adolph TE, Meyer M, Schwärzler J, Mayr L, Grabherr F, Tilg H. The metabolic nature of inflammatory bowel diseases. Nat Rev Gastroenterol Hepatol. 2022;:1–15. Cheng J, Lan W, Zheng G, Gao X. Metabolomics: A High-Throughput Platform for Metabolite Profile Exploration. Methods Mol Biol. 2018;1754:265–92. Di’Narzo AF, Houten SM, Kosoy R, Huang R, Vaz FM, Hou R, et al. Integrative Analysis of the Inflammatory Bowel Disease Serum Metabolome Improves Our Understanding of Genetic Etiology and Points to Novel Putative Therapeutic Targets. Gastroenterology. 2022;162:828-843.e11. Hua X, Ungaro RC, Petrick LM, Chan AT, Porter CK, Khalili H, et al. Inflammatory Bowel Disease Is Associated With Prediagnostic Perturbances in Metabolic Pathways. Gastroenterology. 2022;0. Cleynen I, Boucher G, Jostins L, Schumm LP, Zeissig S, Ahmad T, et al. Inherited determinants of Crohn’s disease and ulcerative colitis phenotypes: a genetic association study. The Lancet. 2016;387:156–67. Atreya R, Siegmund B. Location is important: differentiation between ileal and colonic Crohn’s disease. Nat Rev Gastroenterol Hepatol. 2021;18:544–58. Reinisch W, Colombel J-F, D’Haens G, Sandborn WJ, Rutgeerts P, Geboes K, et al. Characterisation of Mucosal Healing with Adalimumab Treatment in Patients with Moderately to Severely Active Crohn’s Disease: Results from the EXTEND Trial. Journal of Crohn’s & Colitis. 2017;11:425–34. Wu Y, Zhang L, Cao J, Wang H, Ye C, Zhuoma D, et al. Efficacy of infliximab treatment on the mucosal healing of different intestinal segments in patients with ileocolonic Crohn’s disease. Therap Adv Gastroenterol. 2020;13:1756284820976923. Wang H, Lengerich BJ, Aragam B, Xing EP. Precision Lasso: accounting for correlations and linear dependencies in high-dimensional genomic data. Bioinformatics. 2019;35:1181–7. Kim SM, Kim Y, Jeong K, Jeong H, Kim J. Logistic LASSO regression for the diagnosis of breast cancer using clinical demographic data and the BI-RADS lexicon for ultrasonography. Ultrasonography. 2018;37:36–42. Göbl CS, Bozkurt L, Tura A, Pacini G, Kautzky-Willer A, Mittlböck M. Application of Penalized Regression Techniques in Modelling Insulin Sensitivity by Correlated Metabolic Parameters. PLoS One. 2015;10:e0141524. Oh E, Yoo TK, Park E-C. Diabetic retinopathy risk prediction for fundus examination using sparse learning: a cross-sectional study. BMC Med Inform Decis Mak. 2013;13:106. Huang X, Zeng J, Zhou L, Hu C, Yin P, Lin X. A New Strategy for Analyzing Time-Series Data Using Dynamic Networks: Identifying Prospective Biomarkers of Hepatocellular Carcinoma. Sci Rep. 2016;6:1–11. Ananthakrishnan AN. Epidemiology and risk factors for IBD. Nat Rev Gastroenterol Hepatol. 2015;12:205–17. Gallagher K, Catesson A, Griffin JL, Holmes E, Williams HRT. Metabolomic Analysis in Inflammatory Bowel Disease: A Systematic Review. J Crohns Colitis. 2021;15:813–26. Nishida T, Miwa H, Shigematsu A, Yamamoto M, Iida M, Fujishima M. Increased arachidonic acid composition of phospholipids in colonic mucosa from patients with active ulcerative colitis. Gut. 1987;28:1002–7. Liu K, Jia B, Zhou L, Xing L, Wu L, Li Y, et al. Ultraperformance Liquid Chromatography Coupled with Quadrupole Time-of-Flight Mass Spectrometry-Based Metabolomics and Lipidomics Identify Biomarkers for Efficacy Evaluation of Mesalazine in a Dextran Sulfate Sodium-Induced Ulcerative Colitis Mouse Model. J Proteome Res. 2021;20:1371–81. Peppercorn MA. Sulfasalazine. Ann Intern Med. 1984;101:377–86. Yuan Z, Yang L, Zhang X, Ji P, Hua Y, Wei Y. Mechanism of Huang-lian-Jie-du decoction and its effective fraction in alleviating acute ulcerative colitis in mice: Regulating arachidonic acid metabolism and glycerophospholipid metabolism. J Ethnopharmacol. 2020;259:112872. Heimerl S, Moehle C, Zahn A, Boettcher A, Stremmel W, Langmann T, et al. Alterations in intestinal fatty acid metabolism in inflammatory bowel disease. Biochim Biophys Acta. 2006;1762:341–50. Tan M, Ye J, Zhou Z, Ke X, Yu X, Huang K. Fatty Acid Metabolism in Immune Cells: A Bioinformatics Analysis of Genes Involved in Ulcerative Colitis. DNA Cell Biol. 2020;39:1573–82. E DH, M H, P E, M P, Y G, P R. In vivo butyrate metabolism and colonic permeability in extensive ulcerative colitis. Gastroenterology. 1998;115. Bharucha SB. Colonic metabolism of short-chain fatty acids. Am J Gastroenterol. 1992;87:1885. Jin X, Zhou J, Richey G, Wang M, Hong SMC, Hong SH. Undecanoic Acid, Lauric Acid, and N-Tridecanoic Acid Inhibit Escherichia coli Persistence and Biofilm Formation. J Microbiol Biotechnol. 2021;31:130–6. Qi S-A, Wu Q, Chen Z, Zhang W, Zhou Y, Mao K, et al. High-resolution metabolomic biomarkers for lung cancer diagnosis and prognosis. Sci Rep. 2021;11:11805. Additional Declarations No competing interests reported. Supplementary Files Supplementarytable1DifferentiallyExpressedMetabolites.xlsx Supplementarytable2KEGGEnrichmentdata.xlsx Cite Share Download PDF Status: Posted Version 1 posted You are reading this latest preprint version Research Square lets you share your work early, gain feedback from the community, and start making changes to your manuscript prior to peer review in a journal. As a division of Research Square Company, we’re committed to making research communication faster, fairer, and more useful. We do this by developing innovative software and high quality services for the global research community. Our growing team is made up of researchers and industry professionals working together to solve the most critical problems facing scientific publishing. Also discoverable on Platform About Our Team In Review Editorial Policies Advisory Board Help Center Resources Author Services Accessibility API Access RSS feed Manage Cookie Preferences © Research Square 2026 | ISSN 2693-5015 (online) Privacy Policy Terms of Service Do Not Sell My Personal Information {"props":{"pageProps":{"initialData":{"identity":"rs-4212589","acceptedTermsAndConditions":true,"allowDirectSubmit":true,"archivedVersions":[],"articleType":"Research Article","associatedPublications":[],"authors":[{"id":287732914,"identity":"2161cea3-0893-4cc3-a614-f5fc136cfb7b","order_by":0,"name":"Changchang Ge","email":"","orcid":"","institution":"Jiangsu Provincial Hospital of Chinese medicine, Affiliated hospital of Nanjing University of Chinese Medicine","correspondingAuthor":false,"prefix":"","firstName":"Changchang","middleName":"","lastName":"Ge","suffix":""},{"id":287732915,"identity":"fc90922e-779f-4c5c-af23-9307d151f547","order_by":1,"name":"Yi Lu","email":"","orcid":"","institution":"Jiangsu Provincial Hospital of Chinese medicine, Affiliated hospital of Nanjing University of Chinese Medicine","correspondingAuthor":false,"prefix":"","firstName":"Yi","middleName":"","lastName":"Lu","suffix":""},{"id":287732916,"identity":"9cf61b3b-e571-42a8-ac11-5e34efbab757","order_by":2,"name":"Zhaofeng Shen","email":"","orcid":"","institution":"Jiangsu Provincial Hospital of Chinese medicine, Affiliated hospital of Nanjing University of Chinese Medicine","correspondingAuthor":false,"prefix":"","firstName":"Zhaofeng","middleName":"","lastName":"Shen","suffix":""},{"id":287732917,"identity":"875bc116-2f45-480d-bf11-11148a0672a9","order_by":3,"name":"Yizhou Lu","email":"","orcid":"","institution":"Jiangsu Provincial Hospital of Chinese medicine, Affiliated hospital of Nanjing University of Chinese Medicine","correspondingAuthor":false,"prefix":"","firstName":"Yizhou","middleName":"","lastName":"Lu","suffix":""},{"id":287732918,"identity":"152c28ee-f214-4bc9-9ce3-e4300d4d36ce","order_by":4,"name":"Xiaojuan Liu","email":"","orcid":"","institution":"Jiangsu Provincial Hospital of Chinese medicine, Affiliated hospital of Nanjing University of Chinese Medicine","correspondingAuthor":false,"prefix":"","firstName":"Xiaojuan","middleName":"","lastName":"Liu","suffix":""},{"id":287732919,"identity":"ce42b7e0-93b4-415a-babe-3f6aa2d78c12","order_by":5,"name":"Mengyuan Zhang","email":"","orcid":"","institution":"Jiangsu Provincial Hospital of Chinese medicine, Affiliated hospital of Nanjing University of Chinese Medicine","correspondingAuthor":false,"prefix":"","firstName":"Mengyuan","middleName":"","lastName":"Zhang","suffix":""},{"id":287732920,"identity":"6f33223a-9164-4a9d-b069-81937d1207b0","order_by":6,"name":"Yijing Liu","email":"","orcid":"","institution":"Jiangsu Provincial Hospital of Chinese medicine, Affiliated hospital of Nanjing University of Chinese Medicine","correspondingAuthor":false,"prefix":"","firstName":"Yijing","middleName":"","lastName":"Liu","suffix":""},{"id":287732921,"identity":"ea8dcbaf-ad14-4c56-a00d-76bba5e18eb0","order_by":7,"name":"Hong Shen","email":"","orcid":"","institution":"Jiangsu Provincial Hospital of Chinese medicine, Affiliated hospital of Nanjing University of Chinese Medicine","correspondingAuthor":false,"prefix":"","firstName":"Hong","middleName":"","lastName":"Shen","suffix":""},{"id":287732922,"identity":"051e84af-85d2-442b-9fe8-b9d6b2b10e4d","order_by":8,"name":"Lei Zhu","email":"data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAAZAAAAAyAQMAAABI0h/eAAAABlBMVEX///8AAABVwtN+AAAACXBIWXMAAA7EAAAOxAGVKw4bAAAA+UlEQVRIiWNgGAWjYDACCQbGAwwG/3jYmJkPADlgYEBICwNQywEZfva2BFK0MBywkew5A1eJX4v87B6DAx8K7vAY3Mj59sCyzS6agb15mwRDzR2cWhjnnDE4OMPgGVBL7nYDybbk3AaeY2USDMee4dTCLJFjcJjHgBmkZZuEZNuB3AaJHDMJxobDOLWwgbT8AWvJeQbRIv8GvxYekBYGoEVA77NBbeHBr0VCIq3gYI9BGg8wkM0kJM4l57bxpBVbJBzDrUV+RvLGBz/+2NgDo/KZtESZXW4/++GNNz7U4NaCFhYg34FYCcRpAIb4B2JVjoJRMApGwYgCAKfOUsQD9COJAAAAAElFTkSuQmCC","orcid":"","institution":"Jiangsu Provincial Hospital of Chinese medicine, Affiliated hospital of Nanjing University of Chinese Medicine","correspondingAuthor":true,"prefix":"","firstName":"Lei","middleName":"","lastName":"Zhu","suffix":""}],"badges":[],"createdAt":"2024-04-03 12:05:53","currentVersionCode":1,"declarations":"","doi":"10.21203/rs.3.rs-4212589/v1","doiUrl":"https://doi.org/10.21203/rs.3.rs-4212589/v1","draftVersion":[],"editorialEvents":[],"editorialNote":"","failedWorkflow":false,"files":[{"id":54373422,"identity":"a7be2773-3ae8-4d9e-a3d5-5efa23c8b8fe","added_by":"auto","created_at":"2024-04-09 13:28:02","extension":"png","order_by":1,"title":"Figure 1","display":"","copyAsset":false,"role":"figure","size":55027,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eThe overall process of the study.\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eWe recruited a total of 219 volunteers, including 189 UC patients and 30 healthy controls. Firstly, we conducted metabolomics testing on 219 serum samples. Secondly, construct an OPLS-DA model to screen for differential metabolites and perform KEGG enrichment analysis. Then, LASSO and SVM-RFE algorithms were used to further screen for differential metabolism. Finally, potential biomarkers were obtained and ROC curves were plotted.\u003c/p\u003e","description":"","filename":"Slide1.png","url":"https://assets-eu.researchsquare.com/files/rs-4212589/v1/1d4008f9087abefd25154c3e.png"},{"id":54373425,"identity":"5c5eb0a9-f06f-44d7-8afa-4687b4ebbf3f","added_by":"auto","created_at":"2024-04-09 13:28:03","extension":"png","order_by":2,"title":"Figure 2","display":"","copyAsset":false,"role":"figure","size":238086,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eDifference of serum metabolites between UC patients and healthy people.\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003e(A)OPLS-DA score scatter plot between the UC and HC groups; (B) Permutation plot test of OPLS-DA model between the UC and HC groups; (C) Heatmap of hierarchical clustering analysis for the HC and UC group; (D) KEGG Enrichment bubble for the UC and HC groups.\u003c/p\u003e","description":"","filename":"Slide2.png","url":"https://assets-eu.researchsquare.com/files/rs-4212589/v1/c6c290400df0fd3561624190.png"},{"id":54373426,"identity":"217a794e-dfe3-4c52-a1d3-b60284990ddc","added_by":"auto","created_at":"2024-04-09 13:28:03","extension":"png","order_by":3,"title":"Figure 3","display":"","copyAsset":false,"role":"figure","size":188237,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eDifference of serum metabolites between E1, E2 and E3.\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003e(A-C) OPLS-DA score scatter plot and permutation plot test between E1 and E2, E1 and E3, E2 and E3 under positive ion mode; (D-F) OPLS-DA score scatter plot and permutation plot test between E1 and E2, E1 and E3, E2 and E3 under negative ion mode.\u003c/p\u003e","description":"","filename":"Slide3.png","url":"https://assets-eu.researchsquare.com/files/rs-4212589/v1/1d416f943c84939081070382.png"},{"id":54373825,"identity":"6a387e7a-0443-41eb-9d7a-32388f02904d","added_by":"auto","created_at":"2024-04-09 13:36:03","extension":"png","order_by":4,"title":"Figure 4","display":"","copyAsset":false,"role":"figure","size":133150,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eScreening of differential metabolites related to disease extent.\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003e(A) KEGG Enrichment bubble for the group E1 and E2, E1 and E3, E2 and E3; (B) LASSO regression algorithm for the group E1 and E2, E1 and E3, E2 and E3;(C) Intersection differential metabolites between LASSO regression algorithm and SVM-RFE algorithm.\u003c/p\u003e","description":"","filename":"Slide4.png","url":"https://assets-eu.researchsquare.com/files/rs-4212589/v1/6aa080896c6e7e3d553ecf64.png"},{"id":54373423,"identity":"99b4c746-c244-4598-bd84-b0804423e7d7","added_by":"auto","created_at":"2024-04-09 13:28:02","extension":"png","order_by":5,"title":"Figure 5","display":"","copyAsset":false,"role":"figure","size":90724,"visible":true,"origin":"","legend":"\u003cp\u003eInter-group comparison box chart and the area under the ROC curve of differential metabolites predicting the disease extent.\u003c/p\u003e","description":"","filename":"Slide5.png","url":"https://assets-eu.researchsquare.com/files/rs-4212589/v1/60a6e20c04a794323153f605.png"},{"id":58670192,"identity":"5ea6265c-055e-4d11-a5a0-acc62d1919fe","added_by":"auto","created_at":"2024-06-19 14:29:39","extension":"pdf","order_by":0,"title":"","display":"","copyAsset":false,"role":"manuscript-pdf","size":1336912,"visible":true,"origin":"","legend":"","description":"","filename":"manuscript.pdf","url":"https://assets-eu.researchsquare.com/files/rs-4212589/v1/485f0152-304b-4c08-902e-5b3f643070f6.pdf"},{"id":54373827,"identity":"ddd2dc4b-f26e-4f41-8b5f-53d77a1b31ca","added_by":"auto","created_at":"2024-04-09 13:36:03","extension":"xlsx","order_by":1,"title":"","display":"","copyAsset":false,"role":"supplement","size":688632,"visible":true,"origin":"","legend":"","description":"","filename":"Supplementarytable1DifferentiallyExpressedMetabolites.xlsx","url":"https://assets-eu.researchsquare.com/files/rs-4212589/v1/5217052f7c0e8649f0d467e7.xlsx"},{"id":54373828,"identity":"277887c7-2dad-4533-b70e-d4c89b65788c","added_by":"auto","created_at":"2024-04-09 13:36:03","extension":"xlsx","order_by":2,"title":"","display":"","copyAsset":false,"role":"supplement","size":19107,"visible":true,"origin":"","legend":"","description":"","filename":"Supplementarytable2KEGGEnrichmentdata.xlsx","url":"https://assets-eu.researchsquare.com/files/rs-4212589/v1/e4c971209bc81ab5a0621527.xlsx"}],"financialInterests":"No competing interests reported.","formattedTitle":"Using Machine learning to identify metabolomic signatures of ulcerative colitis with the classification according to disease extent","fulltext":[{"header":"1.Introduction","content":"\u003cp\u003eUlcerative Colitis (UC) is a chronic non-specific intestinal inflammatory disease, a subtype of inflammatory bowel disease(IBD), characterized by recurrent diarrhea, bloody stools, and abdominal pain. Traditionally, UC was regrated as a disease of western nations. However, epidemiological studies suggest that incidence are rising in the developing Eastern Countries, which is closely to industrialization and the westernization of lifestyle and dietary habits.[1, 2]The cause of UC is unclear, but it is generally believed that environmental factors and intestinal microbial components can stimulate the immune response of genetically susceptible individuals, and interferes with energy metabolism in immune cells of gastrointestinal tract and intestinal microbial metabolism[3, 4].\u003c/p\u003e \u003cp\u003eMetabolomics is a high-throughput tool for quantitatively analyzing the small molecule metabolites in biospecimens such as blood, tissue, urine, or saliva, which gives us a chance to study the metabolic alterations of UC[5].Di'Narzo et al. analyzed the serum metabolites of UC and Crohn\u0026rsquo;s disease (CD) patients, and found that several metabolites can be used to distinguish IBD patients from healthy people, and associate with disease activity.[6] Huang et al. found that patients with IBD had metabolic pathway disorder before diagnosis. The changes of unsaturated fatty acid biosynthesis and fatty acid biosynthesis related metabolic pathways were associated with UC, and most of the metabolites were negatively related to the risk of UC[7].Metabolomics provide information on metabolite-concentration changes and have great potential in discovering possible biomarkers and related metabolic pathways, as well as exploring the complex pathophysiological mechanisms of diseases.\u003c/p\u003e \u003cp\u003eA gene association analysis study in \u003cem\u003eThe Lancet\u003c/em\u003e pointed that there were significant differences in the genetics between UC, colon CD and ileal CD, and think that disease extent is an intrinsic aspect of a patient\u0026rsquo;s disease, in part genetically determined,[8].Atreya summarized from the aspects of clinical behavior, epidemiology, genetics, and intestinal microbial groups, demonstrated that colon CD showed overlap in the disease behavior of UC[9]. In addition, a large number of studies have shown that CD patients with different lesion sites have different responses to biological agents. Meta analysis results showed that in RCT study, all biological agents had better effect on colon type CD.[10, 11]This makes us pay attention to the important significance of lesion location in the clinical and mechanism research of IBD. Compared with CD, the disease extent of ulcerative colitis has been studied less. This is because the lesions of the UC are limited to the colon and are diffused, and the boundaries between different parts are not obvious. The disease extent of UC is also not static, but rather extends or shrinks with disease activity. However, in previous retrospective study, we found that there were differences in glucose, lipid and protein metabolism related indicators between distal colitis and extensive colitis under the condition of the same level of disease activity.\u003c/p\u003e \u003cp\u003eIn recent years, advanced machine learning algorithms have been widely used to screen medical biomarkers, and can effectively identify disease related metabolite markers. Binary logistics stepwise regression is the most commonly used regression method, but it is well known that the model obtained by it is not optimal. The least absolute shrinkage and selection operator (LASSO) algorithm is a machine learning method, which constructs a penalty function to obtain a more refined model, when fewer predictors are desired, LASSO regression is the preferred option to reduce model dimensionality[12]. Compared with classical statistical methods, LASSO regression has shown better performance for prediction model selection and better identification of predictors.[13\u0026ndash;15] In addition, Support vector machine (SVM) is another machine learning algorithm, which adopts the principle of structural risk minimization, it can iteratively filter out a subset of features with the highest accuracy from a large amount of data, and thus was used to identify potential biomarkers of disease.[16]\u003c/p\u003e \u003cp\u003eTherefore, we compared the serum metabolic profiles of healthy and UC patients using non-targeted metabolomics method to reveal the changes in the metabolic profiles of UC. More importantly, this study is the first to combine metabolomics techniques with machine learning algorithms to search for differential metabolites and metabolic pathways relating to the disease extent of ulcerative colitis, in order to provide evidence for the mechanisms study of ulcerative colitis.\u003c/p\u003e"},{"header":"2.Materials and Methods","content":"\u003cdiv id=\"Sec3\" class=\"Section2\"\u003e \u003ch2\u003e2.1 Study populations\u003c/h2\u003e \u003cp\u003eA total of 219 individuals, between 18 and 75 years of age, were recruited. The participants comprised 189 UC patients and 30 healthy controls. UC patients were recruited from Jiangsu Provincial Hospital of Chinese medicine affiliated to the Nanjing University of Chinese Medicine and healthy controls were enrolled through advertisements.\u003c/p\u003e \u003cp\u003eAll UC patients were diagnosed according to the combination of UC clinical, endoscopic, and histopathological criteria. Montreal classification was used to assess the disease extent, patients with a total Mayo score\u0026thinsp;\u0026lt;\u0026thinsp;3 or partial Mayo score\u0026thinsp;\u0026lt;\u0026thinsp;2, with Mayo endoscopic subscore of 0\u0026ndash;1 and inactive histological disease were used to judge whether the disease is active. Exclusion criteria were: pregnancy, history of colorectal surgery, active infections, and current malignancy. The healthy controls were selected, matched to the UC patients based on age and gender, and had no current or history of severe physical disease.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec4\" class=\"Section2\"\u003e \u003ch2\u003e2.2 Chemicals, Reagents, and Equipment\u003c/h2\u003e \u003cp\u003eMethanol, acetonitrile, ammonium acetate (NH4Ac), and ethanoic acid (AcOH) were all liquid chromatograph-mass spectrometer-grade (LC-MS) and were purchased from CNW Technologies, SIGMA-ALDRICH and Fisher Chemical. All the experiments were conducted on an ultra-high performance liquid chromatography (UHPLC) system (Vanquish, Thermo Fisher Scientific) with a UPLC BEH Amide column (1.7 \u0026micro;m,2.1 mm \u0026times; 100 mm) coupled to Orbitrap Exploris 120 mass spectrometer (Orbitrap MS, Thermo). The ACQUITY UPLC BEH Amide column (1.7 \u0026micro;m,2.1 mm \u0026times; 100 mm) was purchased from Waters Corporation (Milford, MA, United States).\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec5\" class=\"Section2\"\u003e \u003ch2\u003e2.3 Serum Sample Collection and Metabolites Extraction\u003c/h2\u003e \u003cp\u003eMetabolomic analysis was conducted in serum samples, which were collected from all the participants. The serum was separated from the peripheral blood samples in EDTA tubes by centrifuging at 3,000 rpm for 10 min at 4\u0026deg;Candwas immediately stored at \u0026minus;\u0026thinsp;80\u0026deg;C until future metabolomics analysis to minimize the metabolic degradation process.\u003c/p\u003e \u003cp\u003e50\u0026micro;L of sample was transferred to an EP tube. After the addition of 200\u0026micro;L of extract solution (acetonitrile: methanol\u0026thinsp;=\u0026thinsp;1: 1, containing isotopically-labelled internal standard mixture), the samples were vortexed for 30 s, sonicated for 10 min in ice-water bath, and incubated for 1 h at -40 ℃ to precipitate proteins. Then the sample was centrifuged at 12000 rpm(RCF\u0026thinsp;=\u0026thinsp;13800(\u0026times;g),R\u0026thinsp;=\u0026thinsp;8.6cm) for 15 min at 4 ℃. The resulting supernatant was transferred to a fresh glass vial for analysis. The quality control (QC) sample was prepared by mixing an equal aliquot of the supernatants from all of the samples.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec6\" class=\"Section2\"\u003e \u003ch2\u003e2.4 LC-MS/MS Analysis and Data preprocessing\u003c/h2\u003e \u003cp\u003eLC-MS/MS analyses were performed using an UHPLC system with a UPLC BEH Amide column (2.1 mm \u0026times; 100 mm, 1.7\u0026micro;m) coupled to \u003cem\u003eOrbitrap Exploris 120 mass spectrometer\u003c/em\u003e .The mobile phase consisted of 25 mmol/L ammonium acetate and 25 mmo spray Voltage as 3.8 kV (positive) or -3.4 kV(negative), respectively.l/L ammonium acetate in water(pH\u0026thinsp;=\u0026thinsp;9.75)(A) and acetonitrile (B). The auto-sampler temperature was 4 ℃, and the injection volume was 2\u0026micro;L. The \u003cem\u003eOrbitrap Exploris 120 mass spectrometer\u003c/em\u003e was used for its ability to acquire MS/MS spectra on information-dependent acquisition (IDA) mode in the control of the acquisition software (Xcalibur, Thermo). In this mode, the acquisition software continuously evaluates the full scan MS spectrum. The ESI source conditions were set as following: sheath gas flow rate as 50 Arb, Aux gas flow rate as 15 Arb, capillary temperature 320 ℃, full MS resolution as 60000, MS/MS resolution as 15000 collision energy as 10/30/60 in NCE mode, spray Voltage as 3.8 kV (positive) or -3.4 kV(negative), respectively.\u003c/p\u003e \u003cp\u003eIn this study, X peaks were detected and X metabolites were left after relative standard deviation de-noising. Then, the missing values were filled up by half of the minimum value. Also, total ion current normalization method was employed in this data analysis. The final dataset containing the information of peak number, sample name and normalized peak area was imported to SIMCA16.0.2 software package (Sartorius Stedim Data Analytics AB, Umea, Sweden) for multivariate analysis. Data was scaled and logarithmic transformed to minimize the impact of both noise and high variance of the variables. Then an in-house MS2 database (Biotree DB) was applied in metabolite annotation. The cutoff for annotation was set at 0.3.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec7\" class=\"Section2\"\u003e \u003ch2\u003e2.5 Statistical Analysis\u003c/h2\u003e \u003cp\u003eIn this study, OPLS-DA was applied to visualize group separation and screen for effective differential metabolites. To check the robustness and predictive ability of the OPLS-DA model, a 200 times permutation was conducted for validation of the classification performance of the OPLS-DA model with the fit metrics values of R\u003csup\u003e2\u003c/sup\u003eY (p\u0026thinsp;\u0026lt;\u0026thinsp;0.05) and Q2 (p\u0026thinsp;\u0026lt;\u0026thinsp;0.05). The metabolites with VIP\u0026thinsp;\u0026gt;\u0026thinsp;1.5 and p\u0026thinsp;\u0026lt;\u0026thinsp;0.05(student t test) were considered as significantly changed metabolites, and Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway analysis were used to obtain the main disordered metabolic pathways.\u003c/p\u003e \u003cp\u003eAnd then, differential metabolites were further screened by the LASSO algorithm from the \u0026ldquo;glmnet\u0026rdquo; package in the R software (version 4.2.1). At the same time, the SVM-RFE algorithm in the \u0026ldquo;mlbench\u0026rdquo; and \u0026ldquo;caret\u0026rdquo; package were also used. All distinctive features are fed to the SVM-RFE classifier in the form of 10-flod cross validation (CV) and the methods were set to \u0026ldquo;svmRadial\u0026rdquo;. The differential metabolites screened by the two machine learning algorithms are intersected to further narrow the range of metabolites. Receiver operating characteristic (ROC) curves were plotted using the \u0026ldquo;pROC\u0026rdquo; package by R, and the area under the curve (AUC) was used to represent the diagnostic value of metabolites. Figure\u0026nbsp;\u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e1\u003c/span\u003e shows the overall process of the study.\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003c/div\u003e"},{"header":"3.Results","content":"\u003cdiv id=\"Sec9\" class=\"Section2\"\u003e\n \u003ch2\u003e3.1 Clinical characteristics of study population\u003c/h2\u003e\n \u003cp\u003eA total of 189 patients with UC (152 active UC and 37 inactive UC) and 30 HCs were included in this study, and the active UC were divided into ulcerative proctitis (E1), left-sided colitis (E2) and extensive colitis (E3). HCs had a median age of 38.6 years, with 17 males and 13 females. Clinical characteristics of patients with UC are listed in Table \u003cspan class=\"InternalRef\"\u003e1\u003c/span\u003e. We can see that there\u0026rsquo;s no difference in the baseline data of patients in each group, such as gender, age, severity, treatment, BMI and so on.\u003c/p\u003e\n \u003ctable id=\"Tab1\" border=\"1\"\u003e\n \u003ccaption language=\"En\"\u003e\n \u003cdiv class=\"CaptionNumber\"\u003eTable 1\u003c/div\u003e\n \u003cdiv class=\"CaptionContent\"\u003e\n \u003cp\u003eClinical characteristics of patients with Ulcerative Colitis\u003c/p\u003e\n \u003c/div\u003e\n \u003c/caption\u003e\n \u003cthead\u003e\n \u003ctr\u003e\n \u003cth align=\"left\"\u003e\u0026nbsp;\u003c/th\u003e\n \u003cth align=\"left\"\u003e\n \u003cp\u003eE1(N\u0026thinsp;=\u0026thinsp;31)\u003c/p\u003e\n \u003c/th\u003e\n \u003cth align=\"left\"\u003e\n \u003cp\u003eE2(N\u0026thinsp;=\u0026thinsp;44)\u003c/p\u003e\n \u003c/th\u003e\n \u003cth align=\"left\"\u003e\n \u003cp\u003eE3(N\u0026thinsp;=\u0026thinsp;77)\u003c/p\u003e\n \u003c/th\u003e\n \u003cth align=\"left\"\u003e\n \u003cp\u003eInactive UC (N\u0026thinsp;=\u0026thinsp;37)\u003c/p\u003e\n \u003c/th\u003e\n \u003cth align=\"left\"\u003e\n \u003cp\u003e\u003cem\u003eP-\u003c/em\u003evalue *\u003c/p\u003e\n \u003c/th\u003e\n \u003c/tr\u003e\n \u003c/thead\u003e\n \u003ctbody\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eMale, n(%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e12(38.7%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e26(59.1%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e44(57.1%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e21(56.8%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e0.286\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eAge, y (median and range)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e42(32\u0026ndash;55)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e41.5(34.25-52)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e44(29-49.5)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e44(36\u0026ndash;51)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e0.781\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eDisease activity, n(%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eRemission\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e37\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eMild\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e23(74.2%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e20(45.5%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e44(57.1%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"left\" rowspan=\"3\"\u003e\n \u003cp\u003e0.113\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eModerate\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e8(25.8%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e21(47.7%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e27(35.1%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eSevere\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e0\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e3(6.8%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e6(7.8%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eTreatment, n (%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e5-Aminosalicylic Acid\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e29(93.5%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e43(97.7%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e76(98.7%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e33(89.2%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e0.095\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eCorticosteroid\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e1(3.2%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e3(6.8%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e4(5.2%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e0\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e0.455\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eImmunosuppressant\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e0(0.0%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e1(2.3%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e2(2.6%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e0\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e0.630\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eBiological agents\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e1(3.2%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e2(4.5%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e2(2.6%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e0\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e0.645\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eBMI, kg/m\u003csup\u003e2\u003c/sup\u003e (mean\u0026thinsp;\u0026plusmn;\u0026thinsp;SD)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e22.48(3.01)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e22.24(2.69)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e21.33(2.95)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e22.07(3.67)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e0.359\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eSmoker, n(%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e4(12.9%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e7(15.9%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e6(7.8%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e9(24.3%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e0.112\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\" colspan=\"6\"\u003e\n \u003cp\u003e*We used Kruskal-Wallis tests and ANOVA for numerical data, and \u0026chi;2 test for categorical data.\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003c/tbody\u003e\n \u003c/table\u003e\n\u003c/div\u003e\n\u003cdiv id=\"Sec10\" class=\"Section2\"\u003e\n \u003ch2\u003e3.2 Metabolic differences between UCs and HCs\u003c/h2\u003e\n \u003cp\u003eFigure \u003cspan class=\"InternalRef\"\u003e2\u003c/span\u003eA shows the visual and intuitive results of the supervised OPLS-DA analysis on metabolic profiles in the positive and negative ion mode, and the OPLS-DA model displayed obvious discrimination between the UC and HC groups. The permutation plot test of OPLS-DA model for the UC and HC groups was reliable and robust due to avoiding overfitting (Fig. \u003cspan class=\"InternalRef\"\u003e2\u003c/span\u003eB). The specific VIP score and p-value of each abovementioned metabolite is presented in Supplementary table \u003cspan class=\"InternalRef\"\u003e1\u003c/span\u003e. Additionally, for each group of comparison, we calculate the corresponding ratio of the quantitative value of the differential metabolite, and take log2 fold change.\u003c/p\u003e\n \u003cp\u003eThe metabolites with VIP\u0026thinsp;\u0026gt;\u0026thinsp;1.5 and p\u0026thinsp;\u0026lt;\u0026thinsp;0.05 were considered as significantly changed metabolites. Finally, 64 differential metabolites were screened in the positive ion mode, 29 differential metabolites were screened in the negative ion mode, and the cluster analysis heat map of 93 metabolites was shown on the right. Figure \u003cspan class=\"InternalRef\"\u003e2\u003c/span\u003e shows the results of the cluster analysis of the differential metabolites between the UC and HC groups. KEGG pathway analysis show that Arachidonic acid metabolism, Glycerophospholipid metabolism and Pentose phosphate pathway were the main disturbed metabolic pathway related to UC. (Fig. \u003cspan class=\"InternalRef\"\u003e2\u003c/span\u003eD)\u003c/p\u003e\n\u003c/div\u003e\n\u003ch3\u003e3. 3 Metabolic differences between UC patients with different disease extent\u003c/h3\u003e\n\u003cp\u003eNext, we focused on analyzing the differences in serum metabolic profiles of UC patients with different disease extent. Figure \u003cspan class=\"InternalRef\"\u003e3\u003c/span\u003e shows OPLS-DA score scatter plot and permutation plot test of OPLS-DA under positive and negative ion mode, respectively. There\u0026rsquo;re 44, 53 and 32 differential metabolites obtained from OPLS-DA model under positive ion mode, and13,13 and 12 differential metabolites under negative ion mode in E1 and E2, E1 and E3, E2 and E3 groups. The specific VIP score and p-value of each abovementioned metabolite presented in Supplementary table \u003cspan class=\"InternalRef\"\u003e1\u003c/span\u003e.\u003c/p\u003e\n\u003cp\u003eIn order to further explore the differences of serum metabolic profiles in patients with UC at different disease extent, we performed KEGG enrichment analysis. (Supplementary table \u003cspan class=\"InternalRef\"\u003e2\u003c/span\u003e). In Table 2, we listed all metabolic pathways with P\u0026thinsp;\u0026lt;\u0026thinsp;0.05. Notably, the metabolites are mainly involved Glycerophospholipid metabolism, Sphingolipid metabolism, Pyrimidine metabolism and Linoleic acid metabolism. The KEGG Enrichment bubble were showed in Fig. \u003cspan class=\"InternalRef\"\u003e4\u003c/span\u003eA.\u003c/p\u003e\n\u003ctable border=\"1\" cellspacing=\"0\" cellpadding=\"0\" width=\"89%\"\u003e\n \u003ctbody\u003e\n \u003ctr\u003e\n \u003ctd width=\"100%\" colspan=\"6\"\u003e\n \u003cp\u003e\u003cstrong\u003eTable2 Significantly changed pathways in patients with UC at different lesion sites\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd width=\"14.583333333333334%\"\u003e\n \u003cp\u003e\u003cstrong\u003eGroups\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"48.958333333333336%\"\u003e\n \u003cp\u003e\u003cstrong\u003ePathway name\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"7.291666666666667%\"\u003e\n \u003cp\u003e\u003cstrong\u003eTotal\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"6.25%\"\u003e\n \u003cp\u003e\u003cstrong\u003eHits\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"11.458333333333334%\"\u003e\n \u003cp\u003e\u003cstrong\u003eRaw p\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"11.458333333333334%\"\u003e\n \u003cp\u003e\u003cstrong\u003eFDR\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd width=\"14.583333333333334%\" rowspan=\"4\"\u003e\n \u003cp\u003eE1 vs E2\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"48.958333333333336%\"\u003e\n \u003cp\u003eGlycerophospholipid metabolism\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"7.291666666666667%\"\u003e\n \u003cp\u003e36\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"6.25%\"\u003e\n \u003cp\u003e3\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"11.458333333333334%\"\u003e\n \u003cp\u003e0.001672\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"11.458333333333334%\"\u003e\n \u003cp\u003e0.14046\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd width=\"57.31707317073171%\"\u003e\n \u003cp\u003eSphingolipid metabolism\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"8.536585365853659%\"\u003e\n \u003cp\u003e21\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"7.317073170731708%\"\u003e\n \u003cp\u003e2\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"13.414634146341463%\"\u003e\n \u003cp\u003e0.008937\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"13.414634146341463%\"\u003e\n \u003cp\u003e0.37535\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd width=\"57.31707317073171%\"\u003e\n \u003cp\u003ePyrimidine metabolism\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"8.536585365853659%\"\u003e\n \u003cp\u003e39\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"7.317073170731708%\"\u003e\n \u003cp\u003e2\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"13.414634146341463%\"\u003e\n \u003cp\u003e0.029406\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"13.414634146341463%\"\u003e\n \u003cp\u003e0.7356\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd width=\"57.31707317073171%\"\u003e\n \u003cp\u003eLinoleic acid metabolism\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"8.536585365853659%\"\u003e\n \u003cp\u003e5\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"7.317073170731708%\"\u003e\n \u003cp\u003e1\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"13.414634146341463%\"\u003e\n \u003cp\u003e0.035028\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"13.414634146341463%\"\u003e\n \u003cp\u003e0.7356\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd width=\"14.583333333333334%\" rowspan=\"3\"\u003e\n \u003cp\u003eE1 vs E3\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"48.958333333333336%\"\u003e\n \u003cp\u003eSphingolipid metabolism\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"7.291666666666667%\"\u003e\n \u003cp\u003e21\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"6.25%\"\u003e\n \u003cp\u003e2\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"11.458333333333334%\"\u003e\n \u003cp\u003e0.002539\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"11.458333333333334%\"\u003e\n \u003cp\u003e0.21329\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd width=\"57.31707317073171%\"\u003e\n \u003cp\u003eLinoleic acid metabolism\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"8.536585365853659%\"\u003e\n \u003cp\u003e5\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"7.317073170731708%\"\u003e\n \u003cp\u003e1\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"13.414634146341463%\"\u003e\n \u003cp\u003e0.01923\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"13.414634146341463%\"\u003e\n \u003cp\u003e0.80767\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd width=\"57.31707317073171%\"\u003e\n \u003cp\u003ealpha-Linolenic acid metabolism\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"8.536585365853659%\"\u003e\n \u003cp\u003e13\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"7.317073170731708%\"\u003e\n \u003cp\u003e1\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"13.414634146341463%\"\u003e\n \u003cp\u003e0.049357\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"13.414634146341463%\"\u003e\n \u003cp\u003e1\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd width=\"14.583333333333334%\" rowspan=\"3\"\u003e\n \u003cp\u003eE2 vs E3\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"48.958333333333336%\"\u003e\n \u003cp\u003eGlycerophospholipid metabolism\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"7.291666666666667%\"\u003e\n \u003cp\u003e36\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"6.25%\"\u003e\n \u003cp\u003e5\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"11.458333333333334%\"\u003e\n \u003cp\u003e1.03E-07\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"11.458333333333334%\"\u003e\n \u003cp\u003e8.68E-06\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd width=\"57.31707317073171%\"\u003e\n \u003cp\u003eSphingolipid metabolism\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"8.536585365853659%\"\u003e\n \u003cp\u003e21\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"7.317073170731708%\"\u003e\n \u003cp\u003e2\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"13.414634146341463%\"\u003e\n \u003cp\u003e0.003526\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"13.414634146341463%\"\u003e\n \u003cp\u003e0.14809\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd width=\"57.31707317073171%\"\u003e\n \u003cp\u003eLinoleic acid metabolism\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"8.536585365853659%\"\u003e\n \u003cp\u003e5\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"7.317073170731708%\"\u003e\n \u003cp\u003e1\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"13.414634146341463%\"\u003e\n \u003cp\u003e0.022406\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"13.414634146341463%\"\u003e\n \u003cp\u003e0.62738\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003c/tbody\u003e\n\u003c/table\u003e\n\u003cdiv id=\"Sec12\" class=\"Section2\"\u003e\n \u003ch2\u003e3.4 Identification of key metabolites by machine learning\u003c/h2\u003e\n \u003cp\u003eIn order to obtain the key metabolites related to the disease extent of ulcerative colitis, we used two machine learning algorithms, lasso and SVM-RFE, to further screen 57, 66 and 44 differential metabolites obtained from OPLS-DA model. In the above three groups of patients, LASSO regression screened 17, 10 and 9 differential metabolites respectively, and the SVM-REF algorithm generated 57, 3 and 9 outputs respectively. The screening results of the two machine learning algorithms were intersected to obtain 17, 2 and 3 differential metabolites respectively (Fig. \u003cspan class=\"InternalRef\"\u003e4\u003c/span\u003eB and C).\u003c/p\u003e\n \u003cp\u003eWe can see that the E1 group and E2 group compared to screen out 17 different metabolites. Due to the large number of metabolites, we selected the three most significant metabolites according to the results of SVM-REF, which are PS(22:2(13Z,16Z)/22:0), Allocholic acid and Cholic acid. Compared with group E1 and group E3, two main differential metabolites were screened, which were Tridecanoic acid and PC(20:0/P-18:1(11Z)). Compared with group E2 and group E3, three main metabolites were screened, which were Pipericine, Pelargonic acid and Tridecanoic acid.\u003c/p\u003e\n \u003cp\u003eThen, we drew the ROC curves of the above differential metabolites respectively to further clarify their potential value in predicting the location of UC lesions. Figure \u003cspan class=\"InternalRef\"\u003e5\u003c/span\u003e shows the inter-group comparison box chart of the above differential metabolites, and the area under the ROC curve of all metabolites predicting the disease extent is greater than 0.7. It is noteworthy that the serum content of Tridecanoic acid in E3 group is significantly lower than that in E1 and E2 groups, and its area under the ROC curve can reach 0.84, which may be a potential biomarker for identifying patients with extensive colitis.\u003c/p\u003e\n\u003c/div\u003e"},{"header":"Discussion","content":"\u003cp\u003eUlcerative colitis is a metabolic related inflammatory disease. It is widely believed that genetic, environmental, dietary, psychological and other factors interact to cause intestinal microbial metabolism disorder, and participate in the complex mechanism of IBD.[17] Previous studies have also confirmed that UC patients have obvious metabolic disorders compared with healthy people.[18]\u003c/p\u003e\n\u003cp\u003eOur study found that arachidonic acid metabolism, glycerophospholipid metabolism and Pentose phosphate pathway were disordered in UC patients, and arachidonic acid metabolism is the most obvious metabolic pathways, which was consistent with the previous research. Previous studies have observed an increase in the arachidonic acid composition of phospholipids in the colon mucosa of patients with active ulcerative colitis[19]. In addition, Kun Liu et al. applied metabolomics techniques to identify biomarkers for efficacy evaluation of Mesalazine in the treatment of ulcerative colitis and found that IBD dysregulation was related to the biosynthesis of primary bile acid, the metabolism of arachidonic acid, the metabolism of sphingolipid, fatty acid elongation, and glycerophospholipid metabolism[20]. Studies have found that sulfasalazine, 5-aminosalicylic acid and corticosteroids can control intestinal inflammation by regulating arachidonic acid metabolism, the mechanism may be related to the synthesis and release of prostaglandins and leukotrienes[21]. An animal experimental study also found that the traditional Chinese medicine prescription - Huanglian Jiedu Decoction can alleviate UC of mice by regulating arachidonic acid metabolism and glycerophospholipid metabolism[22]. Therefore, arachidonic acid metabolism plays an important role in the study of UC, and the results of this study also support this view.\u003c/p\u003e\n\u003cp\u003eClinically, the disease activity of patients greatly affects treatment decisions and is closely associated with adverse outcomes in UC, therefore, a large number of clinical studies focus on disease activity and neglect the value of disease extent in the study of UC. Recent mechanistic studies of IBD, particularly genotypic studies targeting patients with CD at different sites, have made us realize that the disease extent is also an intrinsic aspect of UC and has important clinical significance.\u003c/p\u003e\n\u003cp\u003eIn this study, we explored the metabolic profiling and potential biological correlates of UC with difference disease extent through the approach of combining metabolomics and machine learning. The OPLS-DA model preliminarily screened the different metabolites between groups, and pathway analysis shows that Glycerophospholipid metabolism, Sphingolipid metabolism, Pyrimidine metabolism and Linoleic acid metabolism are involved in the formation of ulcerative colitis site phenotype. At present, the most studied metabolism pathway of UC is Fatty acid metabolism (FAM), especially short-chain Fatty acid metabolism, which have been proved to be closely related to the pathogenesis of UC. Patients with UC have alterations in FAM genes that impair this metabolic pathway[23], which increases colonic permeability and promotes intestinal immune inflammation[24, 25]. It has been suggested that short chain fatty acid (SCFA) is an important energy source for colonic mucosal cells, especially in the distal colon[26]. Our previous retrospective study also found that the levels of low-density lipoprotein cholesterol, high-density lipoprotein cholesterol, apolipoprotein A1, apolipoprotein E and total cholesterol in patients with distal colitis were significantly higher than those with extensive colitis. However, no other studies have linked the above pathways to the disease extent of UC. The relationship between metabolic pathways and the disease extent of UC requires further research, which will help to refine the study of UC site phenotypes, facilitate drug development and guide clinical drug use.\u003c/p\u003e\n\u003cp\u003eDue to the large number of metabolites involved in the metabolic pathway, our research further screened the differential metabolites through machine learning algorithm, and finally obtained 8 new biomarkers related to the lesion site. After analysis, it was found that the serum content of Tridecanoic acid in E3 group is significantly lower than that in E1 and E2 groups, and its area under the ROC curve can reach 0.84, which may be a potential biomarker for identifying patients with extensive colitis. Tridecanoic acid, a saturated fatty acid of medium-chain length, has been little studied. Some studies have found that tridecanoic acid can inhibit E. coli persistent cell formation and repressed biofilm activity[27]. Another study found that the content of tridecanoic acid in lung cancer patients was significantly lower than that in the healthy control group. The accuracy rate of identifying lung cancer patients with tridecanoic acid combined with other four metabolites can reach 82.9%[28].\u003c/p\u003e\n\u003cp\u003eThis study confirmed that the serum metabolomics of UC patients with different disease extent were different under the same disease activity. Although we do not have a good explanation for this discrepancy, we think it can be considered in two ways. First, this difference may be caused by the different range of intestinal inflammation. This suggests that the above metabolites can be helpful to predict the range of intestinal lesions in UC and predict the disease progression. Second, these differences may be related to the intrinsic molecular structure of the colon and the molecular mechanism of intestinal inflammation in UC. In clinical practice, many drugs for treating UC are selective to the target, and biological agents are representative drugs. If we can dig deeply into the molecular differences of UC patients in different parts, it will be helpful for clinical precision medicine.\u003c/p\u003e\n\u003cp\u003eOf course, there are some defects in this study. Our study is the first one to study the difference of metabolites in different disease extent, and the results lack evidence support. Additionally, this study only used serum samples for metabolomics detection, subsequent multi-sample and multi-omics studies are required for subsequent cross-validation. Therefore, we plan to collect intestinal mucosa samples from patients with UC at different sites to further validate our findings.\u003c/p\u003e"},{"header":"Abbreviations","content":"\u003cp\u003eUC：Ulcerative Colitis\u003c/p\u003e\n\u003cp\u003eIBD：Inflammatory bowel disease\u003c/p\u003e\n\u003cp\u003eCD\u0026nbsp;：Crohn\u0026rsquo;s disease\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eVIP：Vasoactive intestinal peptide\u003c/p\u003e\n\u003cp\u003eQC：Quality Control\u003c/p\u003e\n\u003cp\u003eCV：Coefficient of variation\u003c/p\u003e\n\u003cp\u003eOPLS-DA：Orthogonal Partial Least Squares-Discriminant Analysis\u003c/p\u003e\n\u003cp\u003eLC-MS/MS: Liquid chromatography-tandem mass spectrometry\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eROC: Receiver operating characteristic curves\u003c/p\u003e\n\u003cp\u003eLASSO: Least absolute shrinkage and selection operator\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eSVM: Support vector machine\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eLC-MS: Liquid chromatograph-mass spectrometer-grade\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eAUC: Area under the curve\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eFAM: Fatty acid metabolism\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eSCFA: short chain fatty acid\u003c/p\u003e"},{"header":"Declarations","content":"\u003cp\u003e\u003cstrong\u003eEthics approval and consent to participate\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThe study was approved by the independent Ethics Committee of Jiangsu Provincial Hospital of traditional Chinese medicine affiliated to the Nanjing University of Chinese Medicine (ethics number:2019NL-097-08); all participants provided informed consent.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eConsent for publication\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eNot applicable.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eAvailability of data and materials\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThe data supporting the conclusions of this article is included within the article.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eCompeting interests\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThe authors report no conflict of interest.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eFunding\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThis work was supported by a grant from the National Key Research and Development Program of China [grant number 2017YFC1700104] and Jiangsu Province Traditional Chinese Medicine Science and Technology Development Plan [grant number ZT202103].\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eAuthors\u0026apos; contributions\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eG.C.C. and L.Y. drafted the manuscript. S.Z.F. analyzed and interpreted all data. L.Y.Z., L.X.J., Z.M.Y. and L.Y.J. collected all data. Z.L. and S.H. conceived the study. All authors have approved the final draft of the manuscript.\u003cstrong\u003eAcknowledgements\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eWe wish to thank the technical support provided by Shanghai Biotree Biomedical Technology Company.\u003c/p\u003e"},{"header":"References","content":"\u003col\u003e\n\u003cli\u003eNg SC, Shi HY, Hamidi N, Underwood FE, Tang W, Benchimol EI, et al. Worldwide incidence and prevalence of inflammatory bowel disease in the 21st century: a systematic review of population-based studies. The Lancet. 2017;390:2769\u0026ndash;78.\u003c/li\u003e\n\u003cli\u003eZhao M, G\u0026ouml;nczi L, Lakatos PL, Burisch J. The Burden of Inflammatory Bowel Disease in Europe in 2020. Journal of Crohn\u0026rsquo;s and Colitis. 2021;15:1573\u0026ndash;87.\u003c/li\u003e\n\u003cli\u003eChang JT. Pathophysiology of Inflammatory Bowel Diseases. N Engl J Med. 2020;383:2652\u0026ndash;64.\u003c/li\u003e\n\u003cli\u003eAdolph TE, Meyer M, Schw\u0026auml;rzler J, Mayr L, Grabherr F, Tilg H. The metabolic nature of inflammatory bowel diseases. Nat Rev Gastroenterol Hepatol. 2022;:1\u0026ndash;15.\u003c/li\u003e\n\u003cli\u003eCheng J, Lan W, Zheng G, Gao X. Metabolomics: A High-Throughput Platform for Metabolite Profile Exploration. Methods Mol Biol. 2018;1754:265\u0026ndash;92.\u003c/li\u003e\n\u003cli\u003eDi\u0026rsquo;Narzo AF, Houten SM, Kosoy R, Huang R, Vaz FM, Hou R, et al. Integrative Analysis of the Inflammatory Bowel Disease Serum Metabolome Improves Our Understanding of Genetic Etiology and Points to Novel Putative Therapeutic Targets. Gastroenterology. 2022;162:828-843.e11.\u003c/li\u003e\n\u003cli\u003eHua X, Ungaro RC, Petrick LM, Chan AT, Porter CK, Khalili H, et al. Inflammatory Bowel Disease Is Associated With Prediagnostic Perturbances in Metabolic Pathways. Gastroenterology. 2022;0.\u003c/li\u003e\n\u003cli\u003eCleynen I, Boucher G, Jostins L, Schumm LP, Zeissig S, Ahmad T, et al. Inherited determinants of Crohn\u0026rsquo;s disease and ulcerative colitis phenotypes: a genetic association study. The Lancet. 2016;387:156\u0026ndash;67.\u003c/li\u003e\n\u003cli\u003eAtreya R, Siegmund B. Location is important: differentiation between ileal and colonic Crohn\u0026rsquo;s disease. Nat Rev Gastroenterol Hepatol. 2021;18:544\u0026ndash;58.\u003c/li\u003e\n\u003cli\u003eReinisch W, Colombel J-F, D\u0026rsquo;Haens G, Sandborn WJ, Rutgeerts P, Geboes K, et al. Characterisation of Mucosal Healing with Adalimumab Treatment in Patients with Moderately to Severely Active Crohn\u0026rsquo;s Disease: Results from the EXTEND Trial. Journal of Crohn\u0026rsquo;s \u0026amp; Colitis. 2017;11:425\u0026ndash;34.\u003c/li\u003e\n\u003cli\u003eWu Y, Zhang L, Cao J, Wang H, Ye C, Zhuoma D, et al. Efficacy of infliximab treatment on the mucosal healing of different intestinal segments in patients with ileocolonic Crohn\u0026rsquo;s disease. Therap Adv Gastroenterol. 2020;13:1756284820976923.\u003c/li\u003e\n\u003cli\u003eWang H, Lengerich BJ, Aragam B, Xing EP. Precision Lasso: accounting for correlations and linear dependencies in high-dimensional genomic data. Bioinformatics. 2019;35:1181\u0026ndash;7.\u003c/li\u003e\n\u003cli\u003eKim SM, Kim Y, Jeong K, Jeong H, Kim J. Logistic LASSO regression for the diagnosis of breast cancer using clinical demographic data and the BI-RADS lexicon for ultrasonography. Ultrasonography. 2018;37:36\u0026ndash;42.\u003c/li\u003e\n\u003cli\u003eG\u0026ouml;bl CS, Bozkurt L, Tura A, Pacini G, Kautzky-Willer A, Mittlb\u0026ouml;ck M. Application of Penalized Regression Techniques in Modelling Insulin Sensitivity by Correlated Metabolic Parameters. PLoS One. 2015;10:e0141524.\u003c/li\u003e\n\u003cli\u003eOh E, Yoo TK, Park E-C. Diabetic retinopathy risk prediction for fundus examination using sparse learning: a cross-sectional study. BMC Med Inform Decis Mak. 2013;13:106.\u003c/li\u003e\n\u003cli\u003eHuang X, Zeng J, Zhou L, Hu C, Yin P, Lin X. A New Strategy for Analyzing Time-Series Data Using Dynamic Networks: Identifying Prospective Biomarkers of Hepatocellular Carcinoma. Sci Rep. 2016;6:1\u0026ndash;11.\u003c/li\u003e\n\u003cli\u003eAnanthakrishnan AN. Epidemiology and risk factors for IBD. Nat Rev Gastroenterol Hepatol. 2015;12:205\u0026ndash;17.\u003c/li\u003e\n\u003cli\u003eGallagher K, Catesson A, Griffin JL, Holmes E, Williams HRT. Metabolomic Analysis in Inflammatory Bowel Disease: A Systematic Review. J Crohns Colitis. 2021;15:813\u0026ndash;26.\u003c/li\u003e\n\u003cli\u003eNishida T, Miwa H, Shigematsu A, Yamamoto M, Iida M, Fujishima M. Increased arachidonic acid composition of phospholipids in colonic mucosa from patients with active ulcerative colitis. Gut. 1987;28:1002\u0026ndash;7.\u003c/li\u003e\n\u003cli\u003eLiu K, Jia B, Zhou L, Xing L, Wu L, Li Y, et al. Ultraperformance Liquid Chromatography Coupled with Quadrupole Time-of-Flight Mass Spectrometry-Based Metabolomics and Lipidomics Identify Biomarkers for Efficacy Evaluation of Mesalazine in a Dextran Sulfate Sodium-Induced Ulcerative Colitis Mouse Model. J Proteome Res. 2021;20:1371\u0026ndash;81.\u003c/li\u003e\n\u003cli\u003ePeppercorn MA. Sulfasalazine. Ann Intern Med. 1984;101:377\u0026ndash;86.\u003c/li\u003e\n\u003cli\u003eYuan Z, Yang L, Zhang X, Ji P, Hua Y, Wei Y. Mechanism of Huang-lian-Jie-du decoction and its effective fraction in alleviating acute ulcerative colitis in mice: Regulating arachidonic acid metabolism and glycerophospholipid metabolism. J Ethnopharmacol. 2020;259:112872.\u003c/li\u003e\n\u003cli\u003eHeimerl S, Moehle C, Zahn A, Boettcher A, Stremmel W, Langmann T, et al. Alterations in intestinal fatty acid metabolism in inflammatory bowel disease. Biochim Biophys Acta. 2006;1762:341\u0026ndash;50.\u003c/li\u003e\n\u003cli\u003eTan M, Ye J, Zhou Z, Ke X, Yu X, Huang K. Fatty Acid Metabolism in Immune Cells: A Bioinformatics Analysis of Genes Involved in Ulcerative Colitis. DNA Cell Biol. 2020;39:1573\u0026ndash;82.\u003c/li\u003e\n\u003cli\u003eE DH, M H, P E, M P, Y G, P R. In vivo butyrate metabolism and colonic permeability in extensive ulcerative colitis. Gastroenterology. 1998;115.\u003c/li\u003e\n\u003cli\u003eBharucha SB. Colonic metabolism of short-chain fatty acids. Am J Gastroenterol. 1992;87:1885.\u003c/li\u003e\n\u003cli\u003eJin X, Zhou J, Richey G, Wang M, Hong SMC, Hong SH. Undecanoic Acid, Lauric Acid, and N-Tridecanoic Acid Inhibit Escherichia coli Persistence and Biofilm Formation. J Microbiol Biotechnol. 2021;31:130\u0026ndash;6.\u003c/li\u003e\n\u003cli\u003eQi S-A, Wu Q, Chen Z, Zhang W, Zhou Y, Mao K, et al. High-resolution metabolomic biomarkers for lung cancer diagnosis and prognosis. Sci Rep. 2021;11:11805.\u003cstrong\u003e\u003c/strong\u003e\u003c/li\u003e\n\u003c/ol\u003e"}],"fulltextSource":"","fullText":"","funders":[],"hasAdminPriorityOnWorkflow":false,"hasManuscriptDocX":true,"hasOptedInToPreprint":true,"hasPassedJournalQc":"","hasAnyPriority":false,"hideJournal":true,"highlight":"","institution":"","isAcceptedByJournal":false,"isAuthorSuppliedPdf":false,"isDeskRejected":"","isHiddenFromSearch":false,"isInQc":false,"isInWorkflow":false,"isPdf":false,"isPdfUpToDate":true,"isWithdrawnOrRetracted":false,"journal":{"display":true,"email":"[email protected]","identity":"researchsquare","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":true,"externalIdentity":"","sideBox":"","snPcode":"","submissionUrl":"/submission","title":"Research Square","twitterHandle":"researchsquare","acdcEnabled":true,"dfaEnabled":false,"editorialSystem":"","reportingPortfolio":"","inReviewEnabled":false,"inReviewRevisionsEnabled":true},"keywords":"Ulcerative Colitis, metabolomics, machine learning, disease extent","lastPublishedDoi":"10.21203/rs.3.rs-4212589/v1","lastPublishedDoiUrl":"https://doi.org/10.21203/rs.3.rs-4212589/v1","license":{"name":"CC BY 4.0","url":"https://creativecommons.org/licenses/by/4.0/"},"manuscriptAbstract":"\u003cp\u003e\u003cstrong\u003eBackground\u003c/strong\u003e: Ulcerative colitis (UC) is a metabolic-related chronic intestinal inflammatory disease. Disease extent is an intrinsic aspect of UC and influences the treatment modality. Identifying serum metabolic profiling of different disease extent may help to elucidate the underlying molecular mechanisms of UC.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eMethods\u003c/strong\u003e: UC patients with different disease extent and healthy controls were included in this study. Liquid chromatography-tandem mass spectrometry (LC-MS/MS) method detected the metabolites, and the orthogonal partial least squares-discriminant analysis (OPLS-DA) were performed to identify metabolites. Lasso regression and support vector machine-recursive feature elimination (SVM-RFE) were used to screen metabolites. Receiver operating characteristic curves (ROC) were plotted to represent the diagnostic value of metabolites.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eResults\u003c/strong\u003e: 93 differential metabolites between the UC and HCs, and57, 66, 44 in E1 and E2, E1 and E3, E2 and E3 groups were confirmed by OPLS-DA in the positive and negative ion mode. KEGG pathway analysis show that Arachidonic acid metabolism, Glycerophospholipid metabolism and Pentose phosphate pathway were the main disturbed metabolic pathway related to UC. Glycerophospholipid metabolism, Sphingolipid metabolism, Pyrimidine metabolism and Linoleic acid metabolism are involved UC site phenotype. Lasso and SVM-RFE screened differential metabolites relating to disease extent, and 8 metabolites were confirmed. The area under the ROC curve of all metabolites predicting the disease extent is greater than 0.7.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eConclusions\u003c/strong\u003e: There are significant differences in metabolomics among UC patients with different disease extent. Tridecanoic acid may be a potential biomarker for identifying patients with extensive colitis.\u003c/p\u003e","manuscriptTitle":"Using Machine learning to identify metabolomic signatures of ulcerative colitis with the classification according to disease extent","msid":"","msnumber":"","nonDraftVersions":[{"code":1,"date":"2024-04-09 13:27:58","doi":"10.21203/rs.3.rs-4212589/v1","editorialEvents":[{"type":"communityComments","content":0}],"status":"published","journal":{"display":true,"email":"[email protected]","identity":"researchsquare","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":true,"externalIdentity":"","sideBox":"","snPcode":"","submissionUrl":"/submission","title":"Research Square","twitterHandle":"researchsquare","acdcEnabled":true,"dfaEnabled":false,"editorialSystem":"","reportingPortfolio":"","inReviewEnabled":false,"inReviewRevisionsEnabled":true}}],"origin":"","ownerIdentity":"8ec98496-c534-4fe6-ba40-6b516acde761","owner":[],"postedDate":"April 9th, 2024","published":true,"recentEditorialEvents":[],"rejectedJournal":[],"revision":"","amendment":"","status":"posted","subjectAreas":[],"tags":[],"updatedAt":"2024-06-19T14:21:30+00:00","versionOfRecord":[],"versionCreatedAt":"2024-04-09 13:27:58","video":"","vorDoi":"","vorDoiUrl":"","workflowStages":[]},"version":"v1","identity":"rs-4212589","journalConfig":"researchsquare"},"__N_SSP":true},"page":"/article/[identity]/[[...version]]","query":{"redirect":"/article/rs-4212589","identity":"rs-4212589","version":["v1"]},"buildId":"qtupq5eGEP_6zYnWcrvyt","isFallback":false,"isExperimentalCompile":false,"dynamicIds":[84888],"gssp":true,"scriptLoader":[]}