A blood-based metabolic signature for earlier detection of lung cancer

Metabolomic profiling was applied to 1,259 sera collected within five years of a lung cancer diagnosis and 8,470 non-case sera from ever-smoker participants in the Prostate, Lung, Colorectal, and Ovarian (PLCO) Cancer Screening Trial to identify metabolic signatures for the earlier detection of lung cancer. We develop a panel of five lung cancer-associated metabolites (L5MetP: diacetylspermine, creatine riboside, arginine, hypoxanthine, and acylcarnitine C18:1) to assess 1-year risk of lung cancer. In the testing set, the L5MetP achieves an area under the receiver operating characteristic curve (AUC) of 0.79 (95% CI: 0.76-0.82) for 1-year risk of lung cancer. The L5MetP also complements existing clinical risk models that predict risk incident lung cancer as well as those that predict death from lung cancer. The L5MetP provides a means for identifying ever-smoker individuals who are lung cancer to inform on the need for lung cancer screening.