Optimizing Biomarker Models for Biologically Heterogeneous Cancers: A Nested Model Approach for Lung Cancer.

Area of Science:

• Oncology
• Biomarker Discovery
• Cancer Subtypes

Background:

• Cancer subtypes, particularly in lung cancer, exhibit heterogeneous biology, complicating biomarker development.
• Traditional models struggle to integrate diverse histologic subtypes due to their distinct biological characteristics.
• Early lung cancer detection is challenged by this heterogeneity.

Purpose of the Study:

• To explore a nested biomarker model to address cancer subtype heterogeneity.
• To improve the accuracy of early lung cancer detection through advanced modeling.
• To enhance multicancer early detection strategies.

Main Methods:

• Analysis of blood biomarkers in 337 patients across two clinical sites.
• Development of a nested biomarker model accounting for histologic subtype heterogeneity.
• Comparison of the nested model against traditional logistic regression and the Mayo Clinic model.

Main Results:

• The nested model demonstrated comparable overall performance to existing models (AUC 77.6 training, 77.3 testing).
• The nested subtype versus benign model showed superior performance, especially for small cell lung cancer prediction.
• The patient cohort included diverse malignant and benign nodules, reflecting real-world lung cancer heterogeneity.

Conclusions:

• Nested biomarker models show potential for improving early cancer detection in biologically diverse cancers.
• Addressing cancer heterogeneity is crucial for effective biomarker development.
• Further validation in larger cohorts is necessary to confirm the predictive benefit of this approach.