Diagnostic value of CEA and CYFRA 21-1 tumor markers in primary lung cancer

Abstract

Lung cancer is sometimes difficult to differentiate from benign lung diseases expressing nodular shadow in imaging study. We assessed the diagnostic value of two commonly used tumor markers in distinguishing primary lung cancer from benign lung disease. The serum levels of carcinoembryonic antigen (CEA) and cytokeratin 19 fragments (CYFRA 21-1) were retrospectively analyzed in 655 lung cancer patients and 237 patients with benign lung disease. The standard cut-off levels of 3.2ng/mL CEA and 3.5ng/mL CYFRA 21-1 and twice these respective levels (6.4ng/mL and 7.0ng/mL) were used. CEA and CYFRA 21-1 levels were elevated in 32% and 11% of benign lung disease patients, respectively. CEA sensitivity and specificity for lung cancer diagnosis was 69% and 68% respectively, while that for CYFRA 21-1 was 43% and 89%, respectively. Thus, the combined value for the specificity of the two tumor markers was greater than either alone. Patients were grouped depending on their hospital status, and prevalence rates were determined. The prevalence rate of lung cancer in admitted patients was 51%, the prevalence rate of lung cancer in outpatients was 12%, and the prevalence rate of lung cancer identified during health check-ups was 0.1%. Positive predictive values (PPVs) were calculated using Bayes’ theorem, and varied with the serum tumor marker and prevalence rate: PPVs of CEA [prevalence rate] were 69.2% [51%], 22.7% [12%], and 0.22% [0.1%], while PPVs of CYFRA 21-1 were 80.3% [51%], 34.8% [12%], and 0.39% [0.1%]. However, PPVs for lung cancer diagnosis at a prevalence rate of 51% were 87.3% or higher when the patient exhibited positive CEA and CYFRA 21-1, or CEA or CYFRA 21-1 levels twice the standard cut-off. Our results indicate that CEA and CYFRA 21-1 are reliable serum tumor markers for the diagnosis of lung cancer in addition to CT scans when combined or used individually at twice the standard cut-off level in high prevalence rate groups. The prevalence rate should therefore be taken into account when these serum tumor markers are used as diagnostic tools for lung cancer.