Abstract 2813: Serum biomarker panel detects lung cancer in never smokers

Introduction: Lung cancer is the leading cause of cancer mortality worldwide with more than 1 million deaths each year. Although most lung cancers are attributable to smoking tobacco, it is estimated that as many as 25% of all lung cancer cases are in subjects who have never smoked. A number of clinical, epidemiological, and molecular variations suggest that lung cancers that arise in smokers and non-smokers are significantly different. Biomarker panels may have considerable value when combined with imaging protocols in detecting and diagnosing lung cancer. We previously employed a novel mass spectrometry-based approach to identify serum biomarkers which we have previously shown to detect non-small cell lung cancer (NSCLC) in a smoking population representing all 4 stages of disease. In this study we extend these findings to a cohort of lung cancer subjects who have never smoked. Methodology: Initially 9 biomarkers were assayed in serum collected from smoking subjects with NSCLC and appropriate controls. More than 600 specimens collected from 4 independent sites were employed in the study. Samples were randomly divided into a training set (NSCLC n=128, controls n = 191) and a testing set (NSCLC n=141, controls n=175) and used to develop a regression-based algorithm for lung cancer detection. Subsequently, an independent validation study was undertaken in cohort of lung cancer subjects who had never smoked (interview questionnaire). All stages of cancer (stage I n = 8, stage II n = 4, stage III n = 17, stage IV n = 11) and all major histological cell types (adenocarcinoma n = 21, squamous n = 7, bronchioloalveolar = 8, others n=4) were included. Controls were matched on age/gender (n=40). Results: A global 6-marker regression model identified smoking associated cancer cases with good performance (Training AUC=0.877; Testing AUC=0.868). All stages of cancer were distinguished as well as all of the major histological cell types. Fitting of the model to data from the never smoker cohort revealed that the algorithm again discriminated the malignant cases with strong performance AUC=0.906 (sensitivity = 83% at specificity = 83%). Conclusion: Lung cancer biomarkers identified initially through proteomic analysis have shown robust performance in a study cohort of lung cancer subjects who have never smoked. The findings from these studies suggest that these biomarkers may provide suitable performance across all lung cancer populations to design tests for a variety