Inflammatory bowel diseases (IBD) represent a heterogeneous group of disorders including Crohn's disease (CD) and ulcerative colitis (UC). Currently clinicians use clinical, endoscopic, histopathologic, serologic and radiologic tools to diagnose and differentiate between CD and UC, but in about 10% of IBD cases a discrimination between CD and UC is not possible. Several experimental markers exist for the differential diagnosis and stratification of IBD, but their limited accuracy restricts their use in daily clinical practice. We therefore tested exhaled volatile organic compounds (VOCs; the breath metabolome) for the potential diagnosis and differentiation of IBD. We conducted a prospective study of diagnostic testing, recruiting subjects under 3 groups: 1) IBD, 2) other inflammatory gastrointestinal disease (OGD), such as diverticulitis, microscopic colitis, celiac disease, radiation enteritis and 3) healthy controls (HC). Exclusion criteria were the use of antibiotics 2 weeks prior to breath analysis, surgical history of current diverting ileostomy, colostomy or total abdominal colectomy. Subjects were recruited from the gastroenterology or medicine inpatient or outpatient units of a tertiary referral center. Exhaled breath was collected and its metabolome profile analyzed using selective ion flow tube-mass spectrometry (SIFT-MS) following a mouth rinse with water in patients who were fasting for 8 hours. The VOC ion peak data was correlated to diagnosis, medication intake, disease location and clinical disease activity scores (Lichtiger score (LS) or Harvey Bradshaw Index (HBI)). Statistical analysis was performed using analysis of covariance (ANCOVA) and Kruskal-Wallis test. Seventy-six subjects with a mean age (in years) for CD: 48.2 ± 13.5 (n = 17), UC: 43.3 ± 17.7 (n = 9), OGD: 52.5 ± 20.2 (n = 13) and healthy controls: 37.6 ± 13.3 (n = 34) were recruited. Age and gender adjusted breath VOCs were significantly higher in the IBD group and the OGD group compared to the HC group (Table 1). The area under curve (AUC) on receiver operating characteristic (ROC) graphs of most VOCs demonstrated values more than 0.7 representing a robust difference between OGD and IBD groups versus healthy controls (Table1). Discriminant analysis for CD or OGD versus HC showed that a combination of dimethyl sulfide, ethane and 2-propanol classified subjects with a misclassification rate of 14.5% (Wilks' Lambda 0.24 (P < 0.001)) (Fig. 1). When discriminating CD from UC median acetonitrile levels in the exhaled breath were significantly higher in CD patients (15.8 versus 12.6 parts per billion (ppb), P = 0.025). In addition, higher median levels of dimethyl sulfide and isoprene were noted in patients with biologics use versus no biologics use (3.9 versus 3.1 ppb, P = 0.039 and 69.8 versus 27.1 ppb, P = 0.039, respectively). None of the VOCs were linked to disease location and interestingly, none of the VOCs correlated with clinical disease activity. Analysis of the exhaled breath metabolome indicated a high diagnostic accuracy for discrimination of IBD or OGD from HC. Exhaled VOCs may be helpful in differentiating CD from UC as well as disease activity in UC patients. Ongoing studies are investigating a link between VOCs and endoscopic/radiologic disease activity as well as disease courses.
Read full abstract