Study on optimizing serum-based lung cancer diagnosis using a multi-biomarker approach.

Abstract

8032 Background: Lung cancer remains the leading cause of cancer-related mortality, largely due to the challenges in early diagnosis. Elevated levels of serum carcinoembryonic antigen (CEA), serum amyloid A (SAA), and osteopontin (OPN) have been reported in various cancers. However, the diagnostic efficiency of these biomarkers as a combined panel for lung cancer detection is not well understood. This study evaluates the diagnostic value of combining these three biomarkers for lung cancer screening. Methods: Serum samples from 1,429 lung cancer patients and 1,000 healthy donors were analyzed for CEA, SAA, and OPN levels to assess their diagnostic accuracy. The data were divided into three independent cohorts: a development set, a training set, and a validation set. The diagnostic performance of each individual biomarker and the combined multi-biomarker panel was evaluated. Results: In the discovery set, the most influential variables in the support vector machine model were OPN, SAA, CEA, in that order, while the least influential were NSE, SCC, CYFRA. The optimized multi-biomarker panel comprising SAA, CEA, and OPN, with a cut-off value of 0.61, demonstrated sensitivity, specificity, positive predictive value (PPV), and negative predictive value (NPV) of 83.59%, 90.10%, 93.16%, and 77.29%, respectively, in the training set. Validation confirmed robust performance with sensitivity of 81.21%, specificity of 86.86%, PPV of 90.87%, and NPV of 74.15%. The multi-biomarker panel outperformed individual biomarkers across all stages of lung cancer, achieving AUC values of 0.9320 in the training set and 0.9230 in the validation set. Conclusions: The combined multi-biomarker panel of CEA, SAA, and OPN significantly improves diagnostic performance compared to single biomarkers. This panel represents a promising non-invasive tool for early and accurate lung cancer diagnosis.