OC-107 Gas chromatography sensor platform for diagnosing inflammatory bowel disease and irritable bowel syndrome using faecal samples. (22nd June 2015)

• Record Type: Journal Article
• Authors: Aggio, R; White, P; de Lacy Costello, B; Ratcliffe, N; Probert, C
• Is Part Of: Gut. Volume 64(2015)Supplement 1
• Page Start: A53
• Page End: A54
• Publication Date: 2015-06-22
• Subjects: Gastroenterology -- Periodicals; 616.33
• DOI: 10.1136/gutjnl-2015-309861.107
• Languages: English
• ISSNs: 0017-5749

Abstract

Abstract : Introduction: Inflammatory bowel disease (IBD) is a chronic gastrointestinal disease caused by an aberrant immune response in the gut. Irritable bowel syndrome (IBS) is a common disease that may be confused with IBD. Faecal calprotectin concentration is the current recommended biomarker to rule out IBD, but there are no tests to positively diagnose IBS. Consequently, expensive and invasive methods involving colonoscopy and biopsy are still used by many clinicians to rule out other pathologies. Here, we propose a platform for accurately diagnose IBD and IBS based on faecal samples. Method: The platform is composed of a gas chromatography coupled to a metal oxide gas sensor (OdoReader) and a computer algorithm. The OdoReader separates the volatile organic compounds (VOCs) present in biological samples, while the computer algorithm identifies resistance patterns associated with specific medical conditions and builds mathematical models to classify unknown samples. This platform was applied to faecal samples collected from 98 patients attending the gastroenterology clinic: 45 patients with active IBD; 24 patients with diarrhoea-predominant IBS; and 29 healthy donors (Control). The following comparisons were performed: IBD vs. IBS; IBD vs. Control; IBS vs. Control; IBD vs. non-IBD (IBS + Control); IBS vs. non-IBS (IBD + Control); Control vs. non-Control (IBD + IBS). The mathematical models developed were validated through rigorous validation schemes namely 10-Fold Cross Validation, Double Cross Validation and their Monte Carlo variations. Results: Conclusion: This is the first description of an investigation for the positive diagnosis of IBS. The platform presented here classified samples with accuracies from 85% to 94% using rigorous validation schemes. These schemes provide an estimate of out-of-sample predictive accuracy for similar populations. Therefore, the results reported here indicate a successful differentiation of biological samples from patients with IBD, IBS and healthy donors. These results also indicate a better performance than calprotectin alone for diagnosing IBD, which has been reported as 85% before the application of validation schemes. This platform could be used at point of care for non-invasive diagnosis of IBS and IBD. Disclosure of interest: None Declared.