Multi-centre discovery and validation study evaluating breath biomarkers for the diagnosis of lung cancer - the LuCID study.

Abstract

Background: Volatile Organic Compound (VOC) research for lung cancer detection has faced study design and analytical methodology challenges limiting translation into clinical practice. We evaluated the diagnostic value of breath biomarkers in patients under investigation for suspected lung cancer. Method(s): In a multi-centre prospective case-control study involving 1844 subjects under investigation, breath samples from subjects with a conclusive diagnosis were analysed using gas-chromatography mass-spectrometry. A staged approach was adopted: an Exploratory method for targeted analysis of 63 VOCs associated with lung cancer, followed by an Optimised method for biomarker discovery and finally, evaluation of the optimised panel in a separate validation cohort. Results were compared to the Liverpool Lung Project (LLP) risk model. Finding(s): Using breath VOCs from 677 controls and 518 cases the Exploratory method showed only 2 literature-reported compounds differed significantly between cases and controls. The Optimised method detected 102 VOCs, with ten differing between cases and controls. However, in a validation cohort the 10-VOC panel differentiated cases from controls cohort with a modest AUC: 0.54+/-0.14 for early-stage disease, 0.58+/-0.16 for advanced stage disease and 0.58+/-0.11 for all cases, which did not differ significantly from the LLP model. Combining VOCs with the LLP model did not significantly improve diagnostic performance (AUC 0.64+/-0.11). Interpretation(s): Although some potential biomarkers were identified, their diagnostic performance did not surpass an epidemiological risk model. The study highlights the importance of careful trial design to avoid false-positive findings and indicates a need for more targeted approaches to enhance signal-to-noise ratio in breath biomarker research. Copyright The copyright holder for this preprint is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. It is made available under a CC-BY 4.0 International license.