Breathomics: may it become an affordable, new tool for early diagnosis of non-small cell lung cancer? An exploratory study on a cohort of 60 patients.

Abstract

Analysis of breath, specifically the patterns of volatile organic compounds (VOCs), has shown the potential to distinguish between patients with lung cancer (LC) and healthy individuals (HC). However, the current technology relies on complex, expensive, and low throughput analytical platforms, which provide an offline response, making it unsuitable for mass screening. A new portable device has been developed to enable fast and on-site LC diagnosis, and its reliability is being tested. Breath samples were collected from patients with histologically proven non-small cell lung cancer (NSCLC) and healthy controls using Tedlar bags and a Nafion filter attached to a one-way mouthpiece. These samples were then analyzed using an automated micro portable gas chromatography device that was developed in-house. The device consisted of a thermal desorption tube, thermal injector, separation column, photoionization detector, as well as other accessories such as pumps, valves and a helium cartridge. The resulting chromatograms were analyzed using both chemometrics and machine learning techniques. 30 NSCLC patients and 30 HC entered the study. After a training set (20 NSCLC and 20 HC), and a testing set (10 NSCLC and 10 HC), an overall specificity of 83.3%, a sensitivity of 86.7%, and an accuracy of 85.0% to identify NSCLC patients were found based on three VOCs. These results are a significant step towards creating a low-cost, user-friendly, and accessible tool for rapid on-site lung cancer screening.