IntroductionLung cancer is the leading cause of cancer related mortality owing to the advanced stage it is usually detected because the available diagnostic tests are expensive and invasive; therefore, they cannot be used for general screening.ObjectivesTo increase robustness of previous biomarker panels—based on metabolites in sweat samples—proposed by the authors, new samples were collected within different intervals (4 months and 2 years), analyzed at different times (2012 and 2014, respectively) by different analysts to discriminate between LC patients and smokers at risk factor.MethodsSweat analysis was carried out by LC–MS/MS with minimum sample preparation and the generated analytical data were then integrated to minimize variability in statistical analysis.ResultsPanels with capability to discriminate LC patients from smokers at risk factor were obtained taken into account the variability between both cohorts as a consequence of the different intervals for samples collection, the times at which the analyses were carried out and the influence of the analyst. Two panels of metabolites using the PanelomiX tool allow reducing false negatives (95 % specificity) and false positives (95 % sensitivity). The first panel (96.9 % specificity and 83.8 % sensitivity) is composed by monoglyceride MG(22:2), muconic, suberic and urocanic acids, and a tetrahexose; the second panel (81.2 % specificity and 97.3 % sensitivity) is composed by the monoglyceride MG(22:2), muconic, nonanedioic and urocanic acids, and a tetrahexose.ConclusionThe study has allowed obtaining a prediction model more robust than that obtained in the previous study from the authors. |