Lung cancer (LC) is a widespread cancer that is the cause of the highest mortality rate accounting for 25% of all cancer deaths. To date, most LC patients are diagnosed at the advanced stage owing to the lack of obvious symptoms in the early stage and the limitations of current clinical diagnostic techniques. Therefore, developing a high throughput technique for early screening is of great importance. In this work, we established an effective and rapid salivary metabolic analysis platform for early LC diagnosis and combined metabolomics and transcriptomics to reveal the metabolic fluctuations correlated to LC. Saliva samples were collected from a total of 150 volunteers including 89 patients with early LC, 11 patients with advanced LC, and 50 healthy controls. The metabolic profiling of noninvasive samples was investigated on an ultralow noise TELDI-MS platform. In addition, data normalization methods were screened and assessed to overcome the MS signal variation caused by individual difference for biomarker mining. For untargeted metabolic profiling of saliva samples, around 264 peaks could be reliably detected in each sample. After multivariate analysis, 23 metabolites were sorted out and verified to be related to the dysfunction of the amino acid and nucleotide metabolism in early LC. Notably, transcriptomic data from online TCGA repository were utilized to support findings from the salivary metabolomics experiment, including the disorder of amino acid biosynthesis and amino acid metabolism. Based on the verified differential metabolites, early LC patients could be clearly distinguished from healthy controls with a sensitivity of 97.2% and a specificity of 92%. The ultralow noise TELDI-MS platform displayed satisfactory ability to explore salivary metabolite information and discover potential biomarkers that may help develop a noninvasive screening tool for early LC.
Keywords: high-throughput; lung cancer; noninvasive diagnosis; salivary metabolomics.