Abstract
BACKGROUND: Increasing pulmonary nodule presentations in lung adenocarcinoma patients reveal diagnostic limitations of CT-based invasiveness assessment. The critical unmet need lies in developing non-invasive biomarkers differentiating invasive adenocarcinoma from premalignant lesions and benign nodules, while characterizing metabolic trajectory from health to metastatic disease. METHODS: Untargeted metabolomics analyzed plasma samples from 102 subjects stratified into four cohorts: confirmed adenocarcinoma (n = 35), benign nodules (n = 22), precursor lesions (n = 24), and healthy controls (n = 21). Multivariate analysis identified discriminative metabolites for constructing an infiltration prediction model. RESULTS: Three diagnostic groups exhibited distinct metabolic profiles. Hexaethylene glycol, tetraethylene glycol, and Met-Thr showed stage-dependent concentration gradients. Progressive malignancy correlated with elevated levels of 41 metabolites. An eight-metabolite panel achieved AUC 0.933 (0.873-0.994) in distinguishing precursors from early malignancies, sustained through internal validation (AUC 0.934, 0.905-0.966). CONCLUSIONS: Met-Thr depletion inversely correlates with malignancy progression, while eight-metabolite signatures demonstrate diagnostic potential for preoperative infiltration assessment in nodular adenocarcinoma.