Introduction Irritable bowel syndrome (IBS) is considered to be a functional disorder of gastrointestinal tract, the cause of which remains inadequately identified. Since there are no diagnostic markers, symptoms based criteria are currently used to confirm the diagnosis. Due to similar clinical presentation, it is yet a challenging task to distinguish it from inflammatory bowel disease (IBD) in both primary care and specialist settings. Volatile compounds are chemical which may be emitted from faeces and reflect the pathophysiological environment of the bowel. Understanding changes in their pattern could provide diagnostic information about various bowel disorders.

Methods We aimed to analyse the faecal volatile pattern of patients with diarrhoea predominant IBS and compare it with IBD.

We studied 90 individuals, 20 from diarrhoea predominant IBS, 50 from active IBD (CD=25, UC 25), and 20 from healthy individuals. Diagnosis of IBS was confirmed using symptoms based criteria (Manning, Rome III) while all individual with IBD had histological diagnosis. Fresh samples were aliquoted in 10 ml vials and faecal volatile were extracted using solid phase micro-extraction fibre injected into the head space and were analysed by gas chromatography/mass spectrometery. Volatile compounds were identified using NIST library and manual visualisation using fragmentation pattern.

Results From 400 VOCs identified, 25 were selected using univariate analysis at 0.007 significant level. These were used in forward stepwise entry analysis to construct a predictive model which showed significant value in separating IBS from active IBD and healthy controls. On cross-validation, the IBS: IBD model remains significantly stable: 90% of patients were correctly diagnosed. This model has sensitivity of 95% and NPV 90%.

Conclusion These interim results show that VOCs may be used to distinguish IBS from IBD and either from healthy controls. Further data will provide more information to identify this group more clearly from IBD and allow us to further validate this model.