P12 Asthma breathomics – a systematic review of exhaled volatile organic compounds associated with diagnosis and disease characteristics

Introduction Breathomics is the study of the metabolome – those metabolites associated with a biological system – via the sampling of exhaled breath. As a potential non-invasive indicator of disease processes, breathomics has been applied to a wide range of diseases including asthma. We aimed to assess the evidence for the use of breathomics in the identification of disease and disease characteristics in adults with asthma. Design A systematic review and qualitative analysis of the published literature on exhaled volatile organic compounds in adult asthma. Methods We conducted online databases searches - including PubMed, Embase and OVID medline - in November 2018. We included studies of adult asthma (physician diagnosed or diagnosed according to recognised guidelines), collecting exhaled breath volatiles by any method and presenting primary data. Results Twenty studies were identified; methodologically heterogenous they exhibited a variable risk of bias. Meta-analysis was deemed inappropriate and a qualitative, narrative analysis presented. Assessment using the CASP diagnostic checklist (Critical Appraisal Skills Programme, 2017) revealed studies to be of largely good quality, however, scores were reduced due to the hypothesis-generating stage of the research; none were studies of diagnostic test accuracy. Those studies comparing healthy controls and participants with asthma reported moderate or greater accuracy in the discrimination of samples, or significant differences in compound levels. Asthma phenotypes were differentiated with similarly high levels of accuracy in all but one study. Nine studies named those compounds which they had identified as significant; seventy six compounds were reported in total, of which nine were reported in two papers, and two (acetone and isoprene) featured in three. Conclusion Results are encouraging but there was little concordance between studies in respect of the compounds upon which discriminatory models were based, and models based on such large data-sets are at risk of over-fitting. Validation using independent prospective cohorts and larger participant numbers is required; success would constitute an important step towards non-invasive disease monitoring and the development of personalised medicine in asthma.