S150 A pilot study exploring the utility of breath analysis in the diagnosis of adult pulmonary tuberculosis

Abstract

Introduction Tuberculosis (TB) is one of the top ten leading causes of mortality worldwide. Early diagnosis and treatment of active TB is central to the global TB control strategy. Simple, non-invasive point-of-care tests are needed to support this strategy. There has been recent interest in the use of breath analysis for TB diagnosis. This technique detects gaseous chemicals (volatile organic compounds (VOCs)) that are emitted from infected cells and released in exhaled breath. Instruments for measuring VOCs have become increasingly portable and accurate, offering scope for development as rapid diagnostics. Objectives To determine whether significant differences are detectable in the expired VOCs between well characterised subjects, defined as: i) treatment naive pulmonary TB (n=13); ii) pulmonary TB after completing 2 months’ treatment (n=12); iii) asymptomatic recent TB contacts (IGRA positive (n=19), IGRA negative (n=30)); and iv) IGRA negative healthy controls (n=10). Methods A single centre pilot study was performed at Glenfield Hospital, Leicester (UK). All active TB cases were microbiologically confirmed. All participants were over 16 years of age and HIV-seronegative. Latent TB infection (LTBI) was defined using QuantiFERON Gold-in tube. Participants refrained from eating and drinking for one hour prior to breath capture. End-tidal breath samples were collected into the G.A.S. BreathSpecTM (Germany) through a standard mouthpiece. The instrument only requires a single breath and provides a reliable and repeatable output. Chemical detection of breath chemicals was undertaken by GC-IMS (Gas Chromatography – Ion Mobility Spectrometry). Between-group differences were optimised using machine learning methods to extract the most discriminating features and comparisons presented as receiver operating characteristic (ROC) plots. Results (Table 1) There were significant differences in the VOC signature of the treatment naive pulmonary TB group, compared with the control group (IGRA-negative healthy controls and TB contacts) and with pulmonary TB subjects after 2 months’ therapy. The latter difference may reflect a treatment response effect and/or a drug effect of the TB medication. Although statistically significant, differences between TB contacts with and without LTBI were not sufficient to be of diagnostic value. Conclusions Breath analysis offers a potential approach to the development of non-invasive TB diagnostics and possibly TB treatment response or adherence monitoring.