Skip to Main Content

An official website of the United States government

PLCO: Effects of high-dimensional genetic variation (GWAS) on PSA trajectories and their relationship with prostate cancer incidence

Principal Investigator
Name
Grace Hong
Degrees
Ph.D.
Institution
NCI
Position Title
ORISE fellow
Email
About this CDAS Project
Study
PLCO (Learn more about this study)
Project ID
PLCO-787
Initial CDAS Request Approval
May 18, 2021
Title
PLCO: Effects of high-dimensional genetic variation (GWAS) on PSA trajectories and their relationship with prostate cancer incidence
Summary
We are planning a number of projects related to exploring high-dimensional analytic methods for examining the association between genome-wide mutations and the incidence of prostate cancer. We plan on using statistical methodology that we have developed to identify mutational patterns that have the strongest association with prostate cancer incidence. We also plan to examine whether germline mutational signatures modify the association between PSA trajectories and prostate-cancer risk. Such a higher-order interaction would help to identify whether PSA may be more useful for risk prediction for certain genetic signatures.
Aims

1. Genetic effects on the risk of prostate cancer
1.1 We will develop a high-dimensional predictor of cancer risk incorporating hidden SNPs that are jointly important but have weak marginal associations with prostate cancer risk. This predictor will be compared with established polygenic risk scores (PRS) published in the literature.
1.2 We will develop a high-dimensional risk predictor (incorporating joint effects) to distinguish histologic types of prostate cancer

2. Identifying genomic patterns that show enhanced association between PSA trajectories and prostate cancer incidence
2.1 We will develop a PRS that identifies enhanced differences between the PSA trajectories between cases and controls. This will be done using mixed model methodology as well as quantile regression models for longitudinal data
2.2 We will develop high-dimensional modeling approach that incorporates hidden SNPS that are jointly important in identifying enhanced differences between the trajectories between cases and controls.
2.3 We will compare the approaches developed in 2.1 and 2.2 in identifying genomic patterns in the PLCO-Prostate Cancer dataset
2.4 The sub-aims in 2 will be studied for etiology by focusing on limiting the longitudinal PSA data to 6 months before cancer diagnosis and to early detection by extending the longitudinal models to include a change-point in the trajectory.

Collaborators

Paul Albert, Danping Liu, Mitchell Machiela, Sonja Berndt, Justin Han