A panel of four protein tumor markers for effective and affordable lung cancer early detection by artificial intelligence.

Abstract

8028 Background: Lung cancer is the most common and deadly malignancy worldwide. While low-dose computed tomography (LDCT) reduces mortality in high-risk populations, its high false-positive rate and the required specialized infrastructure and radiologists limit its application. This study assesses LungCanSeek, a novel blood-based protein test for lung cancer early detection. Methods: This study enrolled 1,814 participants (1,095 lung cancer, 719 non-cancer) from three independent cohorts. Blood samples were analyzed for four protein tumor markers (PTMs) using Roche cobas. Artificial intelligence (AI) algorithms were developed for lung cancer detection and subtype classification (lung adenocarcinoma (LUAD), lung squamous cell carcinoma (LUSC), and small cell lung cancer (SCLC)). A two-step lung cancer screening approach was modeled, using LungCanSeek for initial screening, followed by LDCT for LungCanSeek's positive cases. Results: LungCanSeek showed 83.5% sensitivity, 90.3% specificity, and 86.2% accuracy overall. Sensitivities of LUAD, LUSC, and SCLC were 83.3%, 81.4%, and 91.9%. Sensitivity increased with clinical stage in non-small cell lung cancer (NSCLC): 59.5% (I), 69.8% (II), 86.5% (III), and 91.3% (IV). Sensitivities of limited- and extensive-stage SCLC were 91.3% and 93.0%. The subtype classification accuracy was 77.4%. Compared with the other blood-based lung cancer early detection tests like OncImmune’s EarlyCDT-Lung (41.0% sensitivity, 91.0% specificity) and DELFI’s FirstLook-Lung (84.1% sensitivity, 50.9% specificity), LungCanSeek's performance was superior. LDCT had 93.1% sensitivity and 76.5% specificity in NLST study. A screening was modeled for 9 million high-risk adults, based on the number of 15 million eligible individuals in the USA in 2024 at a 60% rate, with a 1.2% lung cancer incidence. While LungCanSeek reduced false positives by 2.4-fold to 862,524 compared to 2,089,620 with LDCT, the two-step approach further lowered false positives by 10.3-fold to just 202,693. Additionally, LDCT’s total cost was $2,493 million, exceeding LungCanSeek’s $720 million by 3.5-fold and two-step’s $978.5 million by 2.5-fold. Conclusions: LungCanSeek is a non-invasive, easy to perform, cost-effective (reagent cost $15) and robust test for lung cancer early detection. It also provides accurate subtype prediction that may guide patients’ clinical management and monitor subtype switching during treatment. The two-step approach not only effectively reduces LDCT’s high false positives but also yields substantial economic benefits, making it a cost-effective strategy for population-wide lung cancer screening.