CONFERENCE PROCEEDING
Predicting absolute risk of lung cancer risk using clinical risk factors and polygenic risk scores
 
More details
Hide details
1
Allelica Inc, New York, United States
 
 
Publication date: 2024-10-17
 
 
Tob. Prev. Cessation 2024;10(Supplement 1):A34
 
KEYWORDS
ABSTRACT
Lung cancer is the third most common cancer and the biggest cancer killer globally. In 2021, 1% of all deaths were due to lung cancer, and because an estimated 80% of lung cancer deaths are due to preventable risk factors, there is enormous potential for risk models that identify high-risk individuals to be implemented to drive down the unnecessary burden of this disease. Several risk factors affect an individual’s risk of developing lung cancer, including tobacco use, pollution, radiation, family history, and genetics. Whilst tobacco use has the single most significant effect on an individual’s chance of getting lung cancer, even within current and previous users, there exists a significant variation in risk. Therefore, models that can identify smokers who are at even greater risk of lung cancer due to combinations of additional risk factors are needed to help national health systems target interventions and surveillance on those who are most likely to get a disease. Here, we present a risk model for lung cancer risk prediction that incorporates age, sex, BMI, smoking history, alcohol consumption, history of lung disease, family history, and genetic risk through a polygenic risk score (PRS) for lung cancer. We develop a new PRS for lung cancer with an Odds Ratio per Standard Deviation change in PRS value of 1.33 (95%CI 1.3-1.37) following adjustment for all relevant risk factors. Adding PRS to an integrated lung cancer risk prediction model that includes clinical risk factors yields a better calibrated and more accurate overall model than a clinical model without PRS. The new PRS-integrated model can be used to significantly stratify a 5-year and lifetime risk of lung cancer in both former and current smokers. For example, current smokers with high PRS values have 3-4 times the risk of those with the lowest genetic risk values. Our model can also classify 5-year risk in former smokers and stratify these individuals above and below a 2% 5-year high-risk threshold. Our results pave the way for absolute risk prediction models that incorporate genetic risk to be used to identify individuals at risk of lung cancer and target interventions toward them.
CONFLICTS OF INTEREST
All authors are employees of Allelica Inc., a for-profit software company.
FUNDING
Funding is not provided.
eISSN:2459-3087
Journals System - logo
Scroll to top