Biomarker discovery for lung adenocarcinoma diagnosis using liquid chromatography-mass spectrometry-based enhanced pseudotargeted metabolomics.

Abstract

Lung adenocarcinoma, the most prevalent subtype of non-small cell lung cancer, is often diagnosed at advanced stages due to the lack of effective early screening methods, leading to poor patient outcomes. In this study, an enhanced pseudotargeted metabolomics approach was developed using liquid chromatography-mass spectrometry, combining untargeted-level coverage with targeted quantitative accuracy while enabling simplified clinical implementation. Serum samples from early-stage lung adenocarcinoma (LUAD) patients and healthy controls were analyzed using this method to identify potential biomarkers and establish a diagnostic model for early LUAD detection. A total of 329 serum samples were divided into discovery, internal validation, and external validation cohorts. Through non-parametric tests and machine learning algorithms, 113 differential metabolites were identified. Glycerophosphocholine and glutamine were validated as potential biomarkers for early LUAD diagnosis; the diagnostic model based on these biomarkers demonstrated good discriminative power, with AUCs of 0.972 and 0.867 in the internal and external validations, respectively. Additionally, comparative analysis between stage I and stage II patients revealed significant metabolic changes including elevated levels of choline, and sphingosine, and decreased levels of 3-dehydroteasterone and PC 31:0. These findings provided new insights into the metabolic alterations associated with LUAD progression and highlighted the potential of pseudotargeted metabolomics in discovering the metabolite biomarkers for early diagnosis of LUAD.