2023 North America Conference on Lung Cancer (NACLC) - Posters and Abstracts-PP01.076 Palina Woodhouse NACLC23 Abstract

PP01.076 Palina Woodhouse NACLC23 Abstract

Pdf Summary

Early detection of lung cancer is crucial for improving survival rates, however, there is currently no widely-used non-invasive blood-based biomarker for lung cancer. This is partly because the disease is highly heterogeneous, with different subtypes requiring tailored treatment approaches. In this study, researchers aimed to investigate whether a nested model approach could improve the performance of a biomarker prediction model for lung cancer.<br /><br />The study included 337 patients from Vanderbilt University Medical Center and the Veterans Affairs Hospital. Blood samples were collected and analyzed for levels of different biomarkers associated with lung cancer. The data was split into a training set and a testing set. Logistic regression models were trained for each subtype of lung cancer and for the overall dataset.<br /><br />The results showed that the nested prediction model, which incorporated information from each subtype model, performed better than both the Mayo model and a generalized linear model. The nested model had an improved area under the curve (AUC) of 0.78 in the training set and 0.77 in the testing set, compared to the AUC of 0.63 for the Mayo model.<br /><br />This study demonstrates that the nested model approach can improve the performance of biomarker prediction models for lung cancer by accounting for the histologic heterogeneity of the disease. The findings have important implications for the development of non-invasive blood-based biomarkers for early detection of lung cancer.

Keywords

lung cancer

early detection

blood-based biomarker

heterogeneous disease

tailored treatment

nested model approach

biomarker prediction model

Vanderbilt University Medical Center

Veterans Affairs Hospital

logistic regression models