Germline genetics of prostate cancer, PSA, and BPH using the expanded PLCO GWAS
The goal of this project is to conduct a meta-analysis of genome-wide association studies for PSA and PSA velocity and discover new genetic variants that may be used to understand the genetic underpinnings of PSA and to create risk prediction models in the future.
For this meta-analysis, genotyped data from each study will be imputed to the TopMed (preferred), 1KG Phase III (1KGP3), or Haplotype Reference Consortium (HRC) reference panel. Each study will test the association with PSA and PSA velocity assuming an additive genetic model. Analyses will be performed stratified by ancestry and adjusted for age, principal components and other covariates as available and appropriate for the study. Association statistics and related imputation quality measures will be shared, so that the meta-analyses may be conducted at participating institutions and compared.
BPH
The goal of this project is to conduct a meta-analysis of genome-wide association studies for BPH and discover new genetic variants that may be used to understand the genetic underpinnings of BPH. Results may also be used to develop polygenetic risk scores to be evaluated in relation to PSA and risk prediction models for prostate cancer.
For this meta-analysis, genotyped data from each study will be imputed to the TopMed (preferred), 1KG Phase III (1KGP3), or Haplotype Reference Consortium (HRC) reference panel. Each study will test the association with BPH and BPH velocity assuming an additive genetic model. Analyses will be performed stratified by ancestry and adjusted for age, principal components and other covariates as available and appropriate. Association statistics and related imputation quality measures will be shared, so that the meta-analyses may be conducted at participating institutions and compared.
1. Among men with a baseline PSA measurement and genotyping data, what is the risk of developing prostate cancer and ow does this vary with time (continuous, spline, categorical) when considering the following factors:
-baseline age
-family history of PCa
-genetic ancestry
-baseline PSA value
-baseline DRE
-PSA PRS
-PCA PRS
-BPH PRS
What is the shape of the risk for aggressive disease an prostate cancer death?
Possible exploratory analyses that in addition to the above factors include:
-PSA velocity or multiple PSA values (e.g., T0 and T3)
-mosaic Y loss (or PRS for y loss)
-telomere length PRS
Can we use PSA PRS, PCa PRS, BPH PRS to predict these risks? Can risk stratification optimize PSA screening?
Construction of prostate cancer PRS:
Option 1: Construct PRS with known prostate cancer hits.
Option 2: Construct PRS using SBayesR. Use Practical data (without PLCO) to generate risk estimates, then use UK Biobank as test set to construct PRS. Apply PRS to PLCO.
2. Lifetime risk of prostate cancer and prostate cancer death for men aged 55-60 years
For men with a middle-age PSA value, how can we risk stratify into low/moderate/high lifetime risk of prostate cancer using the following risk factors:
-family history of PCa
-genetic ancestry
-middle-age PSA value
-middle-age DRE
-PSA PRS
-PCA PRS
-BPH PRS
What are the years of life gained?
3. Risk of a positive biopsy among men with intermediate PSA levels
Restricting to men who underwent a biopsy, what is the risk of a positive biopsy when combining the following risk factors:
-current age
-family history of PCa
-genetic ancestry
-current PSA value
-DRE outcome
-PSA PRS
-PCa PRS
Amongst men who underwent prostate biopsy due to a positive PSA screen (i.e., not solely DRE), is PSA PRS and/or PCa PRS predictive of an outcome (PCa, aggressive PCa)? What if we restrict to positive PSAs tests less than 10/8/6 ng/mL? Can we stratify men based on risk in order to make better recommendations on when to biopsy based on age, race, family history, PSA level, DRE, PSA PRS, BPH PRS, PCa PRS, and BMI? Can we improve upon existing nomograms for prostate biopsy?
4. Localized disease. Among men with clinically localized disease, can we improve existing risk stratification models, such as D’Amico score, by including PCa PRS, BPH PRS, PSA PRS, BMI, family history, and smoking?
5. PSA GWAS meta-analysis
Taking the baseline PSA measurement for each man, perform a GWAS looking for germline susceptibility loci associated with PSA level adjusting for:
-age at PSA measurement
-significant PCAs
-BPH, BMI, smoking?
Limit to men without prostate cancer.
Potential studies: PLCO, Kaiser, MVP, PRACTICAL, SELECT
6. BPH GWAS meta-analysis
Two possible approaches for BPH: 1) self-reported BPH, 2) self-reported surgery for BPH
Perform a GWAS looking for germline susceptibility loci associated with BPH adjusting for:
-age at baseline
-significant PCAs
-other covariates(?): BMI, smoking
Potential studies: PLCO, Kaiser, UK Biobank, SELECT, PCPT, Harvard cohorts, ACS
Sonja Berndt
Stephen Chanock
Danielle Karyadi
Mitchell Machiela
Rebecca Landy
Hormuzd Katki
Paul Albert