Large scale evaluation of a multiple tumor marker panel for lung cancer diagnosis and histological subtyping

Early detection and accurate differential diagnosis of lung cancer (LC) remain significant clinical challenges. This study aimed to develop and validate a multi-marker panel (5LCTM) to optimize diagnostic performance, particularly for ultra-early stage LC and complex differential scenarios.

A retrospective cohort of 2423 participants (1761 LC and 662 benign controls) was analyzed. Eleven serum tumor markers (TMs) were evaluated, and a five-marker panel (5LCTM: CEA, NSE, CYFRA 21 - 1, ProGRP, and SCC-Ag) was integrated. Diagnostic models were constructed using binary logistic regression (Enter method). Diagnostic performance was assessed using receiver operating characteristic (ROC) curves, sensitivity, specificity, positive predictive value (PPV)and negative predictive value (NPV).

The 5LCTM panel significantly outperformed individual markers, achieving an AUC of 0.822 for overall LC detection. Notably, the panel exhibited promising performance in Stage 0 LC, with an AUC of 0.843, a sensitivity of 0.83, and a high NPV of 0.97. For histological subtypes, the 5LCTM model maintained high accuracy for SCLC (AUC = 0.921), SCC (AUC = 0.904), and AD (AUC = 0.801). In differential diagnosis, the panel achieved an NPV of 1.00 across nearly all BLT and BLD subgroups, including pulmonary hamartoma and organizing pneumonia. Gender-stratified analysis confirmed the model's robustness, with stable AUCs in both males (0.827) and females (0.810).

The 5LCTM panel is a robust, non-invasive tool that provides high diagnostic accuracy across all pathological stages and histological types. Its superior performance in Stage 0 LC and its exceptional NPV make it a potential screening tool for refining the triage of patients with indeterminate pulmonary nodules.