P132 Detection and monitoring of IBD based on faecal volatile organic compounds
S. Bosch*1, D. Wintjens2, A. Wicaksono3, J. Kuijvenhoven4, P. Stokkers5, R. van der Hulst4, E. Daulton3, M. Pierik2, J. A. Covington3, N. K. De Boer1, T. G. de Meij6
1Amsterdam UMC, Gastroenterology and Hepatology, Amsterdam, The Netherlands, 2MUMC+, Gastroenterology and Hepatology, Maastricht, The Netherlands, 3University of Warwick, School of Engineering, Coventry, UK, 4Spaarne Gasthuis, Gastroenterology and Hepatology, Hoofddorp, The Netherlands, 5OLVG West, Gastroenterology and Hepatology, Amsterdam, The Netherlands, 6Amsterdam UMC, Pediatric gastroenterology, Amsterdam, The Netherlands
The gold standard to detect and monitor inflammatory bowel disease (IBD) remains endoscopic assessment which is invasive and costly. Faecal calprotectin (FCP) is the most commonly used non-invasive biomarker to assess IBD but lacks specificity. Faecal volatile organic compounds (VOC) are molecular end-products thought to represent both metabolic processes in the human body and the interaction between microbiota and host. The aim of the current study was to evaluate the potential of faecal VOC patterns to detect IBD and to identify disease exacerbation.
Patients aged 18 years and older with an established diagnosis of IBD collected a faecal sample prior to their scheduled consult at the outpatient clinic of either the Maastricht University Medical Centre (MUMC+) or the Amsterdam University Medical Centres (Amsterdam UMC). The healthy control (HC) group consisted of patients without mucosal abnormalities observed during their scheduled colonoscopy at the Amsterdam UMC. Active disease was defined as an FCP level of ≥250 mg/g, remission was defined as FCP <100 mg/g combined with a Harvey–Bradshaw Index <4 points for Crohn’s disease (CD) or Simple Clinical Colitis Activity Index <3 points for ulcerative colitis (UC). Faecal samples were measured by means of gas chromatography-ion mobility spectrometry (G.A.S. Flavourspec). The data were split into three sets, 70% for training and validation and the remaining 30% as test set. A Wilcoxon rank-sum test was used to find the 100 most discriminatory features and Random Forest classification was used to provide statistical results.
A total of 497 faecal samples were provided by 281 IBD patients and compared with 224 samples from 224 HC. Of these, 294 were CD samples (107 active disease, 84 remission) and 203 were UC samples (83 active disease, 64 remission). Outcomes of the Random Forest classification are given in Table 1. IBD, UC and CD could be discriminated from HC with high accuracy both in active state and remission. No difference in VOC pattern was observed between UC and CD, and between active disease state and remission.
Table 1. Differences in VOC pattern between groups of inflammatory bowel disease patients and healthy controls
|AUC (95% CI)||Sensitivity||Specificity||PPV||NPV||p-value|
|IBD vs. HC||0.97 (0.94–1)||0.97||0.95||0.99||0.88||<0.0001|
|CD active vs. HC||0.98 (0.96–1)||1||0.95||0.84||1||<0.0001|
|CD remission vs. HC||0.97 (0.95–1)||1||0.93||0.74||1||<0.0001|
|CD active vs. CD remission||0.49 (0.36–0.62)||0.33||0.77||0.73||0.38||0.562|
|UC active vs. HC||0.97 (0.95–1)||1||0.91||0.88||1||<0.0001|
|UC remission vs. HC||0.97 (0.95–0.99)||0.96||0.95||0.70||1||<0.0001|
|UC active vs. UC remission||0.62 (0.43–0.81)||0.85||0.43||0.78||0.55||0.094|
|CD active vs. UC active||0.55 (0.41–0.68)||0.54||0.67||0.74||0.46||0.228|
|CD remission vs. UC remission||0.52 (0.33–0.72)||0.53||0.62||0.78||0.36||0.393|
We demonstrated that faecal VOC patterns can discriminate IBD, CD and UC from HC both during active disease state and remission, though there is no difference between UC and CD. These characteristics imply that faecal VOC patterns may hold potential as non-invasive biomarkers for IBD disease detection. Based on clinical activity, active disease could not be discriminated from remission, which hamper its potential to detect disease exacerbation.