The Role of AHRR Methylation in Predicting Smoking Associated Lung Pathology in the National Lung Screening Trial Cohort
Recently, we have developed a methylation sensitive droplet digital PCR (MSddPCR) assay that precisely measures DNA methylation status at cg05575921 that can objectively determine smoking intensity. Using a similar assay, Bojesen and colleagues demonstrated cg05575921 methylation could predict which patients with a self-reported 30 pack year would benefit from LDCT screening. If their findings generalizes to American smokers, it is conceivable that DNA methylation assessments could eliminate the need for 25% of all LDCT screening, saving hundreds of millions of dollars annually while eliminating unnecessary medical radiation exposure.
In this proposal, we will test whether cg05575921 methylation can identify those smokers who are unlikely to benefit from LDCT screening using DNA and clinical data from the approximately ~4856 of these subjects enrolled in the LDCT arm of the NLST. In addition, we will explore the correlation between MSddPCR and CT derived radiomic features associated with lung cancer and chronic obstructive pulmonary disease (COPD) to develop a powerful post-CT prediction of lung cancer risk following lung nodule detection with LDCT. We hypothesize that DNA methylation at cg05575921 will identify a population of screening eligible smokers who are unlikely to benefit from CT screening and to increase the diagnostic power of the CT for those for which screening is recommended. Our proposal will have impact because it could reduce health care costs, reduce incidental radiation exposure to high risk individuals and produce better prediction of cancer in those receiving LDCT. It is highly feasible because methylation assay exist and both the DNA, clinical and image data are available. It is innovative because epigenetic tests are new to patient care. The research team is well poised to conduct the research as it includes established experts in clinical epigenetics, pulmonology, lung cancer, CT analysis and statistics. As a direct result of this research, we will determine the relationship of smoking intensity to the risk of lung cancer and COPD progression.
We hypothesize that DNA methylation at cg05575921 will identify a population of screening eligible smokers who are unlikely to benefit from CT screening and that we can develop a methylation and radiomic (objective measures) machine learning method to increase the diagnostic power of the CT for those undergoing LDCT screening. To test this hypothesis, using state-of-the-art MSddPCR and radiomic biomarkers, we will determine DNA methylation and CT features in those subjects in the LDCT arm who contributed to the ACRIN repository, then analyze the resulting data to determine likelihood of having lung cancer. Our proposal will have impact because it could reduce health care costs by hundreds of millions of dollars annually, result in more immediate intervention for those with lung cancer and reduce incidental radiation exposure to high risk individuals. It is highly feasible because the MSddPCR assays exist and the DNA, LDCT and clinical data are available. It is innovative because epigenetic assessments are new to patient care. The research team is well poised to conduct the research as it includes established experts in clinical epigenetics, pulmonology, lung cancer, CT analysis and statistics. As a direct result of this research, we will determine the relationship of smoking intensity to the risk of lung cancer and COPD progression. Specifically, we will:
Aim 1. Determine cg05575921 methylation in all available subjects with both LDCT and DNA (n≈ 4856), then determine its relationship to lung cancer.
Hypothesis: cg05575921 methylation will predict risk for lung cancer in these 30 pack year smokers.
Aim 2. Correlate cg05575921 methylation with radiological features associated with COPD and emphysema, including factors related to spatial heterogeneity and progression of disease
Hypothesis: cg05575921 methylation is associated with COPD risk, severity and the rate of COPD progression using longitudinal data from the NLST.
Aim 3. Construct a machine learning classification method to predict lung cancer risk based on a combination of features from LDCT and cg05575921 methylation, for comparison to existing multi-parameter risk models.
Hypothesis: classification performance will be enhanced beyond that achievable through current multi-parameter risk models reliant on subjective determination of radiological features and patient reported smoking history, through the use of a non-subjective, quantitative feature based machine learning approach.
Robert Philibert, University of Iowa, Psychiatry
Eric Hoffman, University of Iowa, Radiology
Jeffrey Long, University of Iowa, Psychiatry
-
Using Cg05575921 methylation to predict lung cancer risk: a potentially bias-free precision epigenetics approach.
Philibert R, Dawes K, Moody J, Hoffman R, Sieren J, Long J
Epigenetics. 2022 Dec; Volume 17 (Issue 13): Pages 2096-2108 PUBMED