Abstract
BACKGROUND: Currently, there is a lack of satisfactory biomarkers for lung cancer screening and differential diagnosis. Metabolomics can detect the differences of metabolites in different pathological states, which may be helpful in the identification of lung cancer. METHODS: Patients with pulmonary nodules were divided into two groups: benign pulmonary nodules (BPN) and primary lung cancer (PLC). Healthy population group(HPG) was enrolled. Half of PLC and HPG were selected as the testing subset and the validating subset, respectively. Two groups with different pulmonary lesions were compared with HPG, respectively, and compared with each other. Univariate and multivariate data statistical analysis method was used to select the metabolites. Their discriminating ability was verified by the ROC curve in the validating subset. RESULTS: The testing subset was comprised of BPN(n = 32), PLC(n = 80), and HPG(n = 48). The significant metabolites selected by the comparison of PLC and HPG were 1-salicylate glucuronide, s-nitrosoglutathione, and dihydrocaffeic acid 3-O-glucuronide. Their AUC of ROC was all greater than 0.95(0.957-0.995) in the validating subset, which indicated they had a strong ability to differentiate PLC from HPG. From the comparison of the two pulmonary nodule groups with each other, three metabolites with good sensitivity and specificity were selected, whose discriminating ability of PLC from BPN was satisfactory. CONCLUSION: Through the research of plasma metabolomics, some significant metabolites were detected among PLC, BPN, and HPG, which have the potential to be the promising biomarkers for lung cancer screening and differential diagnosis.