Abstract Lung cancer is the world's leading cause of cancer deaths. Early detection through low-cost metabolite screening approaches could reduce lung cancer mortality. Here we report the results of a large-scale, quantitative metabolomics study aimed at identifying plasma biomarkers for lung cancer detection in a Chinese population. A cohort of 410 patients, including 137 healthy controls, 189 biopsy-confirmed individuals with stage I/II lung cancer, and 84 individuals with biopsy-confirmed stage III/IV cancer, were studied. Plasma samples were collected and analyzed using an in-house-developed, targeted liquid chromatography-mass spectrometry (LC-MS) metabolomics method that detects and quantifies 138 metabolites. Logistic regression was used to develop an optimal biomarker panel that could distinguish all-stage lung cancer patients from healthy controls. The resulting six-metabolite panel achieved an area under the curve (AUC) of 95.0%. A second biomarker panel was developed to distinguish early-stage lung cancer from healthy controls and reached an AUC of 94.3%. All models were developed on an initial training set and then fully validated on a separate holdout set. Our proposed biomarker models show significant improvements over previously published models for metabolite-based lung cancer diagnosis and detection. These metabolite biomarker panels are intended for the development of a low-cost, minimally invasive blood test for lung cancer screening in China.
Chen et al. (Tue,) studied this question.