A four-protein marker panel combined with artificial intelligence demonstrates high accuracy in a multicenter study.
A new blood-based test that combines four protein markers with artificial intelligence has demonstrated potential for early lung cancer detection in a multicenter retrospective study published in Translational Lung Cancer Research.
The LungCanSeek test, developed by SeekIn Inc, was evaluated in a study of 1,814 participants, including 1,095 lung cancer patients, led by Dr Mao Mao and a multidisciplinary team from Chinese hospitals. The AI-driven algorithm combines protein marker levels with participant age and gender to detect lung cancer and its subtypes.
The test measures levels of four protein tumor markers—CEA, CYFRA 21-1, ProGRP, and SCCA—and can be performed using standard clinical laboratory equipment at a cost of $15 per test.
Two-Step Screening Model
Researchers modeled a two-step screening approach to address the high false-positive rate and cost associated with low-dose computed tomography (LDCT). In this model, high-risk individuals are first screened with the blood test, and only those testing positive are referred for LDCT imaging.
This approach maintained high detection rates of true cancer cases while reducing unnecessary follow-up procedures and associated costs. The strategy could enable population-wide lung cancer screening programs in settings where access to expensive imaging is limited.
Key highlights:
- High Accuracy: LungCanSeek demonstrated an overall sensitivity of 83.5% and specificity of 90.3% in detecting lung cancer.
- Subtype Identification: The test successfully classified lung cancer subtypes—lung adenocarcinoma (LUAD), lung squamous cell carcinoma (LUSC), and small cell lung cancer (SCLC)—with 77.4% accuracy.
- Early Stage Detection: Sensitivity reached 67.2% for early-stage cases, critical for improving survival rates.
- Cost-Effectiveness: The two-step screening approach (LungCanSeek followed by low-dose CT for positives) reduced false positives by over tenfold and screening costs by 2.5 times compared to low-dose CT screening.
- Simple Implementation: The test leverages four widely available protein markers (CEA, CYFRA 21-1, ProGRP, SCCA) and standard lab equipment, with an estimated reagent cost of $15 per test.
“LungCanSeek’s performance surpasses many existing blood-based and imaging-only screening approaches. Its affordability and ease of use make it especially valuable for expanding early lung cancer detection in resource-constrained settings,” says Mao, corresponding author and founder and CEO at SeekIn Inc, in a release.
The authors note the test’s potential for implementation in national screening strategies and its ability to reduce healthcare system burdens while boosting screening coverage.
ID 296458859 © Skypixel | Dreamstime.com