Immunosignature Technology Detects Stage I Lung Cancer From A Drop Of Serum

Early detection of lung cancer could have a tremendous impact as newly diagnosed patients suffer an overall 5-year survival rate of 16.8%, largely due to late detection. The use of computed tomography (CT) for earlier lung cancer detection in a high risk population was shown to reduce lung cancer mortality by ∼20% in a 2011 NLST study report. The US Preventative Services Task Force has recommended low dose CT be used for screening high risk populations despite a high false positive rate (23.5%) and 2-10 mSv of radiation exposure per scan. Our laboratories have developed a new concept for disease detection based on immunosignatures. The immunosignature (IS) technology uses arrays of hundreds of thousands of unique peptides designed to broadly survey an individual9s antibody binding repertoire from a drop of blood. Arrayed library peptides that bind differentially between case and control samples are selected as input into classification algorithms. The technology holds promise for detecting the presence of virtually any disease that generates a significant B-cell response. For cancer, the technology has accurately distinguished patient samples representing several different cancers from each other as well as from non-cancer controls. To determine the feasibility of using IS technology to detect early stage lung cancer among high risk and/or symptomatic individuals, a cohort of samples representing 44 pathology-confirmed stage I lung cancer cases and 44 age, gender, and smoking-history/status matched controls were assayed on high density peptide arrays. All control samples were from individuals with benign lung diseases, such as chronic respiratory illnesses and/or benign lung nodules (55% of control samples), that could be interpreted as suspicious of cancer and require differential diagnosis. Peptides with statistically significant difference in binding signal between cases and controls were identified by Bonferroni adjusted t-test and standardized log odds ratios. A support vector machine classification algorithm was trained using the 100 most significant features. Feature selection and classifier model performance were evaluated by applying 100 iterations of a four-fold cross-validation routine. The cross-validated performance metrics of the 100-feature immunosignature classifier were sensitivity of 91.9% (95% confidence interval (CI) 78-3%-97.5), specificity of 94.3% (95%CI 81.3%-98.6%) and an overall accuracy of 93.1% (95% CI 85.7%-97.5%). These results need verification in larger cohorts and validation in blinded studies; however, they are deemed promising for the early detection of lung cancer. Citation Format: Theodore M. Tarasow, Michael W. Rowe, Mojgan Haddad, Kathryn Sykes. Immunosignature technology detects stage I lung cancer from a drop of serum. [abstract]. In: Proceedings of the 106th Annual Meeting of the American Association for Cancer Research; 2015 Apr 18-22; Philadelphia, PA. Philadelphia (PA): AACR; Cancer Res 2015;75(15 Suppl):Abstract nr 2010. doi:10.1158/1538-7445.AM2015-2010