Abstract
OBJECTIVE: To evaluate the sensitivity and specificity of a novel biosensor platform using the olfactory system of rats to detect volatile organic compound signatures of lung cancer present in urine samples from patients eligible for lung cancer screening. METHODS: Urine samples were collected from patients who met United States Preventive Services Task Force lung cancer screening criteria and underwent computed tomography scans, including individuals with proven lung cancer diagnosis and without lung cancer. Samples were analyzed using the biosensor platform, and biosensor responses were recorded. A dedicated machine learning algorithm assessed various behavioral parameters to estimate lung cancer risk. The system was validated against previously diagnosed cases and evaluated for sensitivity, specificity, positive predictive value, and negative predictive value. RESULTS: A total of 238 patients were included. Among them, 144 (60%) were diagnosed with lung cancer, of whom 84 (58%) were male. Non–small cell lung cancer was diagnosed in 135 of 144 patients, with adenocarcinoma identified in 110 cases. Seven patients had small cell lung cancer, and 2 had large cell lung cancer. Stage distribution was 1 (stage 0), 89 (stage I), 12 (stage II), 18 (stage III), and 22 (stage IV). The biosensor platform achieved a positive predictive value of 91%, negative predictive value of 86%, sensitivity of 91%, and specificity of 86% (area under the curve = 0.86). CONCLUSIONS: In the era of precision medicine, the novel urine biosensor platform offers a simple, noninvasive approach to assist in identifying candidates for early lung cancer screening. Its integration could improve adherence to screening protocols and aid in the diagnosis and management of nonspecific pulmonary nodules.