Salivary Biomarkers for Non-small Cell Lung Cancer Detection

Overview

The investigators plan to recruit patients for a prospective study in patients in need of evaluation for lung lesions suspicious for cancer. Saliva samples will be collected before diagnostic evaluation including biopsy with subsequent blinded examination of the salivary markers without knowledge of the disease status. This prospective recruitment with retrospective blinded evaluation or PRoBE design satisfies the highest standards recommended by the National Cancer Institute for biomarker development. This process limits the selection bias that can confound retrospective studies. As the primary endpoint, a pre-specified multi-marker panel will be evaluated based on the combination of sensitivity and specificity. In addition, seven pre-specified individual candidate mRNA cancer markers and six internal reference or "housekeeping" genes will be evaluated. The performance of new multi-marker panels will also be assessed and compared with the prior pre-specified model based on sensitivity and specificity combinations as well as the area under the receiver operating characteristic curve.

Consecutive eligible patients presenting to the study institutions and associated clinics will be enrolled. Inclusion and exclusion criteria are detailed separately in the section on eligibility. The target enrollment population listed in the study design section provides a greater than 85% power to achieve the pre-specified goal for the sensitivity and specificity of the pre-specified model. Saliva will be collected in vials pre filled with mRNA stabilizer. Seven mRNA biomarkers (BRAF, CCNI, EGRF, FGF19, FRS2, GREB1, and LZTS1) will be measured in a central laboratory by quantitative PCR with the laboratory personnel blinded to the patient diagnosis. The primary outcome is the validation of a pre-specified multi-marker model. This pre-specified model incorporates 3 of the cancer genes and the housekeeping gene. The model will be validated based on the sum of sensitivity and specificity exceeding 1.3 with the lower limit of the 95% confidence interval exceeding 1.0. A secondary endpoint is the development of new multi-maker models with potential improved performance. These new models will be developed by logistic regression and compared with the pre-specified model based on the area under the receiver operating characteristic curve and the maximum sum of sensitivity and specificity. Individual candidate biomarkers will also be validated based on a statistically significant up-regulation in cancer patients compared with controls. Potential new housekeeping genes will be evaluated based on their equivalence in cancer and control as well as their performance in multi-marker models in comparison with the pre-specified housekeeping gene which is GADPH..