A1131
Title: Bayesian meta-analysis of penetrance for cancer risk with an extension to include studies with ascertainment bias
Authors: Thanthirige Lakshika M Ruberu - University of Texas at Dallas (United States)
Danielle Braun - Harvard TH Chan School of Public Health (United States)
Giovanni Parmigiani - Dana-Farber Cancer Institute (United States)
Swati Biswas - University of Texas at Dallas (United States) [presenting]
Abstract: Multi-gene panel testing allows many cancer susceptibility genes to be tested quickly at a lower cost, making such testing accessible to a broader population. Thus, more patients carrying pathogenic germline mutations in various cancer-susceptibility genes are being identified. This creates both opportunity and urgency to provide appropriate risk-reducing guidance, which depends on accurate age-specific cancer risk (penetrance) estimates for each gene. A meta-analysis approach is proposed using a Bayesian hierarchical random-effects model to estimate penetrance by integrating studies that report various risk measures (e.g., penetrance, relative risk, odds ratio), while accounting for uncertainty. Using a Markov chain Monte Carlo algorithm, posterior distributions are derived to estimate penetrance and credible intervals. The method is assessed through simulations involving risk estimates for two moderate-risk breast cancer genes, ATM and PALB2, and superior performance is demonstrated over existing methods in terms of coverage probability and mean squared error. The model is also extended to account for ascertainment bias by incorporating a bias term with appropriate priors. Simulations show that adjusting for this bias leads to more accurate and precise penetrance estimates than ignoring it or discarding biased studies. Finally, the method is applied to estimate breast cancer penetrance in carriers of pathogenic variants in ATM and PALB2.
