Modeling Measurement Error in a Biomarker on the Pathway from Smoking to Lung Cancer

• Activity Number: 72
• Type: Topic Contributed
• Date/Time: Monday, August 9, 2004 : 8:30 AM to 10:20 AM
• Sponsor: Section on Statistics in Epidemiology
• Author(s): Sally W. Thurston*+
• Companies: University of Rochester
• Keywords: Bayesian ; biomarker ; DNA adducts ; lung cancer ; measurement error ; smoking

Abstract

Molecular biologists have identified specific cellular changes, called biomarkers, which enable them to better characterize the pathway from chemical exposure to initiation of some cancers. In lung cancer, one such biomarker is DNA adducts in lung tissue. Carcinogens derived from cigarette smoke can bind to DNA to form such adducts, and this process is believed to initiate smoking-induced lung cancer. The goal of this work is to incorporate knowledge of such underlying biological mechanisms into a useful statistical framework to improve cancer risk estimates. The model uses measurements of adducts in lung cells and in blood cells; the latter are needed because lung adducts cannot be measured in controls. Adduct measurements in each type of cell are known to vary within individuals. By introducing a latent variable for true lung DNA adducts, I allow for measurement error in both types of observed adduct measurements, but assume greater measurement error in blood adducts. Gibbs sampling is used to obtain the posterior distributions of model parameters. Predicted and observed case status agree for approximately 75% of the sample.