Background: The diagnostic value of six tumor markers was investigated and the appropriate combinations of those tumor markers to discriminate small cell lung cancer was explored.
Methods: Patients suspected with lung cancer (1938) were retrospectively analyzed. Candidate tumor markers from carcinoembryonic antigen (CEA), squamous cell carcinoma-related antigen (SCC), cytokeratin 19 fragment 21-1 (CYFRA 21-1), neuron-specific enolase (NSE), tissue polypeptide antigen (TPA), and progastrin releasing peptide (ProGRP) were selected to construct a logistic regression model. The receiver operating characteristic curve was used for evaluating the diagnostic value of the tumor markers and the predictive model.
Results: ProGRP had the highest positive rate (72.3%) in diagnosed small cell lung cancer, followed by neuron-specific enolase (68.3%), CYFRA21-1 (50.5%), carcinoembryonic antigen (45.5%), tissue polypeptide antigen (30.7%), and squamous cell carcinoma-related antigen (5.9%). The predictive model for small cell lung cancer discrimination was established, which yielded the highest area under the curve (0.888; 95% confidence interval: 0.846-0.929), with a sensitivity of 71.3%, a specificity of 95.0%, a positive predictive value of 49.0%, and a negative predictive value of 98.0%.
Conclusions: Combining tumor markers can improve the efficacy for small cell lung cancer discrimination. A predictive model has been established in small cell lung cancer differential diagnosis with preferable efficacy.
Keywords: Logistic regression model; Small cell lung cancer; Tumor marker.