O084 A preoperative breath test to predict post-oesophagectomy pulmonary infection

Abstract

Introduction Pulmonary infection (PI) is the most common complication post-oesophagectomy. Identifying patients at risk of PI could facilitate pre-emptive preventative measures. Analysis of exhaled volatile organic compounds (VOCs) is a novel method of diagnosing disease non-invasively. Given the integral contribution of the oro-respiratory system in generating breath VOCs, this study applied breathomics to develop a novel predictive model for PI prior to oesophagectomy. Methods Breath samples were collected from patients with oesophageal adenocarcinoma on the morning of their surgery using a quality-controlled methodologically optimised workflow. Breath was analysed using two gold standard analytical platforms: one (GC-MS) and two-dimensional (GCxGC-MS) thermal-desorption gas-chromatography time-of-flight mass-spectrometry, the latter being unrivalled in its chromatographic resolution. Raw spectral data was pre-processed and analysed using a tile-based Fisher ratio method to generate predictive models. Ethical approval REC: 17/WA/0161. Results Breath samples were analysed from 23 patients undergoing oesophagectomy and 12 (52%) developed PI. Principal component analysis revealed significantly distinct volatolomes discriminating PI risk status (R2Xcum 0.908, Q2cum 0.806, CV ANOVA p<0.001). A 5 VOC model had an area under the curve of 0.806 (95%CI 0.647-0.947) and 0.859 (95%CI 0.601-1) for predicting PI using GC-MS and GCxGC-MS respectively. Predictive VOCs were tentatively identified to be branched chain in structure suggesting a likely microbial origin. Conclusion This is the first predictive PI model using state-of-the-art breathomics. Identification of branched chain VOCs offers unprecedented insights into the disease biology of post-oesophagectomy PI and highlights an important role of the preoperative oro-respiratory microbiome as a risk factor for PI.