Abstract:Abstract Background: Although polygenic risk score (PRS), which estimates an individual’s genetic risk to a given trait/disease, has been considered as a promising tool for precision medicine, the individual-level uncertainty in lung cancer PRS and the extent to which its effect on downstream clinical applications remains largely unexplored. Method: Lung cancer PRSs with confidence interval (CI) were constructed via two approaches for each individual - 1) a GWAS-based cross-validation and bootstrapping method (PRS-CV) based on 16 SNP loci that have been validated in Caucasian population, and 2) a Bayesian method using LDPred2 (PRS-Bayes), among 18,146 lung cancer cases and 12,894 cancer-free controls of European ancestry in the International Lung Cancer Consortium (ILCCO). Individuals were classified into different genetic risk subgroups based on the relationship between their own PRS mean/PRS CI and the population level threshold. Multivariable analyses controlling for age, gender and smoking history were conducted to discover the relative risk of lung cancer for different PRS risk subgroups. Lung cancer risk prediction models incorporating both PRS risk subgroups and non-genetic risk factors were constructed and evaluated using five-fold cross-validation. Results: Considerable variance in PRS point estimates at the individual level was observed for both methods, with an average standard deviation (s.d.) of 0.12 (95% CI = 0.09-0.15) for PRS-CV and a much larger s.d. of 0.88 (95% CI = 0.68-1.11) for PRS-Bayes. With a credible level p = 95%, PRS-CV only classified 24.3% of the low genetic risk and 15.6% of the high genetic risk population with certainty, while PRS-Bayes was unable to find any eligible individuals. Only 19% of the individuals were commonly identified as high genetic risk using the two PRS estimators. An increased relative risk of lung cancer was observed for PRS risk subgroups identified by PRS CI approach (OR = 2.92, 95% CI = 2.26-3.78, p-value = 2.5*10−16) compared to by PRS mean (OR = 2.20, 95% CI = 1.99-2.45, p-value = 1.6*10−50). Similar improvement was observed in stratified analyses by gender, lung cancer histologies and smoking history, with the largest increase observed in never smokers. Improved risk prediction performance was consistently observed in individuals identified by PRS CI approach and it achieved the best performance when incorporating age, gender and smoking packyears (AUC: 0.73, 95% CI = 0.72-0.74). Conclusion: Our study first characterized the uncertainty of lung cancer PRS for individuals and evaluated the potential impacts on subsequent risk stratification and prediction in populations with European ancestry. Lung cancer PRS estimated using different methods have modest correlations at the individual level, highlighting the importance of taking serious considerations integrating PRS in clinical interpretation and implementation in precision medicine. Citation Format: Xinan Wang, Ziwei Zhang, Yi Ding, Tony Chen, Lorelei Mucci, Christopher I. Amos, Xihong Lin, David C. Christiani. Impact of individual level uncertainty of lung cancer polygenic risk score on risk stratification and prediction [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2023; Part 1 (Regular and Invited Abstracts); 2023 Apr 14-19; Orlando, FL. Philadelphia (PA): AACR; Cancer Res 2023;83(7_Suppl):Abstract nr 1184.

Development and validation of a lung cancer polygenic risk score incorporating susceptibility variants for risk factors

Risk assessment for colorectal cancer via polygenic risk score and lifestyle exposure: a large-scale association study of East Asian and European populations

Development and Evaluation of a Polygenic Risk Score for Lung Cancer in Never-Smoking Women: A Large-Scale Prospective Chinese Cohort Study

Identification of risk loci and a polygenic risk score for lung cancer: a large-scale prospective cohort study in Chinese populations

Impact of individual level uncertainty of lung cancer polygenic risk score (PRS) on risk stratification

Abstract 1184: Impact of individual level uncertainty of lung cancer polygenic risk score on risk stratification and prediction

Polygenic Risk Score in Personalized Screening of Lung Cancer: a Prospective Cohort Study in Chinese

Novel strategy for disease risk prediction incorporating predicted gene expression and DNA methylation data: a multi‐phased study of prostate cancer

Using a Polygenic Risk Score to Improve the Risk Prediction of Non–Small Cell Lung Cancer in Taiwan

Predicting Lung Cancer in Korean Never-Smokers With Polygenic Risk Scores

Genome-wide Association Study Based Risk Prediction Model in Predicting Lung Cancer Risk in Chinese

Prediction of lung cancer risk in a Chinese population using a multifactorial genetic model

Stratifying Lung Adenocarcinoma Risk with Multi-ancestry Polygenic Risk Scores in East Asian Never-Smokers

Association of smoking and polygenic risk with the incidence of lung cancer: a prospective cohort study

Pan-cancer analysis demonstrates that integrating polygenic risk scores with modifiable risk factors improves risk prediction

Performances of Lung Cancer Risk Prediction Models for Multi-Ancestries: A Systematic Review and Meta-Analysis

Predicting Lung Adenocarcinoma Prognosis with a Novel Risk Scoring Based on Platelet-Related Gene Expression.

Genetic risk and gastric cancer: polygenic risk scores in population-based case-control study

Associations of genetic risk and smoking with incident COPD

Implications of Lifestyle Factors and Polygenic Risk Score for Absolute Risk Prediction of Colorectal Neoplasm and Risk-Adapted Screening

The risk stratification and predictive performance of a new combined polygenic risk score for hepatocellular carcinoma