TY - JOUR T1 - Exploiting metabolomic approaches to aid in the diagnosis of lung cancer JF - European Respiratory Journal JO - Eur Respir J DO - 10.1183/13993003.congress-2016.PA2871 VL - 48 IS - suppl 60 SP - PA2871 AU - Adian Mironas AU - Simon Cameron AU - Keiron O'Shea AU - Chuan Lu AU - Paul Lewis AU - Luis Mur AU - Keir Lewis Y1 - 2016/09/01 UR - http://erj.ersjournals.com/content/48/suppl_60/PA2871.abstract N2 - INTRODUCTION: Metabolomics approaches and platforms allow high throughput biochemical screening of biological samples that can identify important biochemical pathways and allow rapid, accurate detection of early disease. We have shown Fourier Transform Infra-Red Spectroscopy (FTIR) is useful [1] and now combine it Mass Spectrometry (MS) to investigate novel biomarkers in lung cancer (LC).METHODS: Spontaneous sputum was collected from 23 patients with LC+ (17 NSCLC, 6 SCLC, 1 clinico-radiological diagnosis) and 33 healthy controls (LC-). Samples were analysed using flow infusion electrospray ion mass spectrometry. Data-mining of the derived metabolite profiles using artificial neural networks and receiver operating characteristic / area under the curve (ROC/AUC) were calculated.RESULTS: LC+ class could be detected when compared to LC- with an accuracy of 0.68, LC- versus controls with an accuracy of 0.85 and LC+ vs LC- with an accuracy of 0.79. Further datamining based on Random Forest identified metabolites with AUC values of > 0.8 which distinguished between LC+/LC- and included the metabolite Ganglioside GM1, involved in cellular membrane turnover and which has previously been linked to LC [2].CONCLUSION: Non-invasive metabolomic approaches based on sputum are feasible, can differentiate lung cancer from controls and can identify specific discriminating metabolites. These offer potential for diagnosis and identifying biological targets for intervention.REFERENCES: {1} BMC Cancer 2010; 10: 64 [2] Lung Cancer. 28 (2000) 29–36. ER -