PTU-093 Metabolomic profiling in inflammatory bowel disease. (22nd June 2015)

• Record Type: Journal Article
• Authors: Hildebrand, DR; Trivedi, D; Xu, Y; Rattray, N; Ross, NP; Correa, ES; Satsangi, J; Goodacre, R; Watson, AJ
• Is Part Of: Gut. Volume 64(2015)Supplement 1
• Page Start: A102
• Page End: A102
• Publication Date: 2015-06-22
• Subjects: Gastroenterology -- Periodicals; 616.33
• DOI: 10.1136/gutjnl-2015-309861.208
• Languages: English
• ISSNs: 0017-5749

Abstract

Abstract : Introduction: Inflammatory bowel disease is classified into 2 main types; Crohn's disease (CD) and ulcerative colitis (UC). Differentiation can prove clinically challenging. Metabolomic profiling is an emerging science that is utilised for biomarker discovery. We aimed to use metabolomic profiling to differentiate between patients with CD, UC and healthy controls (HC). Method: Serum and urine samples were collected from 40 UC patients, 43 CD patients, 1 patient with indeterminate colitis and 62 HCs. Gas chromatography (GC) and reverse phase ultra-high performance liquid chromatograph (RP-UHPLC) were employed as separation techniques of choice coupled with time-of-flight or orbitrap mass spectrometry (MS) serving as detector, respectively. Chemometric and statistical analysis was carried out using partial least square models (cross-validated via bootstrapping). Classification was visualised with confusion matrices and classification tables. Metabolite data was subjected to univariate analysis (Kruskal-Wallis test with p-values corrected for groups). Results: GC-MS and LC-MS analysis of urine showed that samples can be classified using their metabolic profiles for borderline differentiation between UC and CD but may be prone to false positives. LCMS data showed a slightly better classification for HC but weaker for UC and CD. The serum samples showed acceptable classification of UC and CD but poor classification of HC, with the GC-MS being the more favourable platform. Univariate analysis was performed to identify significant metabolite variables between the three groups (p < 0.05). Combined analyses revealed 343 significant analytes (serum GC-MS 263, serum LC-MS 16, urine GC-MS 15 and urine LC-MS 49), of which 47 robust analytes were significant after correcting for groups. Conclusion: This study highlights the potential for metabolomic profiling to differentiate between HC, UC and CD. Of those metabolites identified hippuric acid (p = 0.02), ethylbenzene (0.049), and octanoeic acid (p = 0.001) have previously been identified as potential biomarkers in IBD. Disclosure of interest: None Declared.