Assessment of a Combined Panel of Six Serum Tumor Markers for Lung Cancer

Abstract

Rationale: We have previously identified six serum tumor markers (TMs) (carcinoembryonic antigen, carbohydrate antigen 15.3, squamous cell carcinoma–associated antigen, cytokeratin-19 fragment, neuron-specific enolase, and pro–gastrin-releasing peptide) related to the presence of lung cancer (LC). Objectives: To validate their individual performance in an independent cohort, and to explore if their combined assessment (≥1 abnormal TM value) is a more accurate marker for LC presence. Methods: We determined these six TMs in 3,144 consecutive individuals referred to our institution by their primary care physician because of the clinical suspicion of LC. Measurements and Main Results: LC was excluded in 1,316 individuals and confirmed in 1,828 patients (1,563 with non–small cell LC and 265 with small cell LC). This study validated the previously reported performance of each individual TM. We also showed that their combined assessment (≥1 abnormal TM) had a better sensitivity, specificity, negative predictive value, and positive predictive value (88.5, 82, 83.7, and 87.3%, respectively) than each TM considered individually and that it increased the diagnostic performance (area under the curve) of a clinical model that included tumor size, age, and smoking status. In patients with radiographic nodules less than 3 cm, the negative predictive value of the TM panel was 71.8%, hence providing some support for a more conservative diagnostic approach. Finally we identified two TMs (neuron-specific enolase and pro–gastrin-releasing peptide) that differentiate the risk of non–small cell LC from that of small cell LC. Conclusions: The combined assessment of a panel of six serum TMs is a more accurate marker for LC presence than these same TMs considered individually. The potential of these TMs in the diagnostic and screening settings deserves further research.