Breath Volatile Organic Compounds as Non‐Invasive Tools for the Detection and Monitoring of Inflammatory Bowel Disease

ABSTRACT Background and Aims Inflammatory bowel disease (IBD) diagnosis and disease evaluation require endoscopy, histology, or radiology, which are often invasive or expensive. A non‐invasive diagnostic method is urgently needed. The present study examined the utility of volatile organic compounds (VOCs) to distinguish IBD from normal controls and monitor disease progression. Methods Breath samples were collected from 98 IBD patients, 77 non‐IBD controls, and 8 non‐IBD enteritis. VOCs were extracted using the solid phase microextraction (SPME) technique and detected using high‐resolution gas chromatography–mass spectrometry (GC–MS). Four machine learning models were used in our study. Feature VOCs were those differentially presented in all these four models for diagnosis when compared patients with IBD and non‐IBD controls. The performance of the VOC panels in the best selected model was carried out in a training set and confirmed in a validation set and testing set. Results A total of seven VOCs were selected for the diagnosis of IBD. The Random Forest model showed distinct differentiation between IBD patients and the non‐IBD controls in the training set (AUC: 1.000 [1.000, 1.000]), validation set (AUC: 0.978 [0.978, 0.978]) and testing set (AUC: 0.966 [0.906, 1.000]). Among them, four VOCs were found to be related to IBD activity. Conclusions Breath VOCs appeared to be useful tools for the diagnosis and activity monitoring of IBD patients.