(UroToday.com) In a plenary session at the Society of Urologic Oncology Annual Meeting focused on early prostate cancer diagnosis and treatment, Dr. Burcu Darst discussed the role of polygenic risk scores (PRS) to improve prostate-specific antigen (PSA)-based prostate cancer screening.
Dr. Darst began by highlighting that prostate cancer represents one of the largest areas of health disparities in the United States. While African ancestry has been long recognized as a risk factor for prostate cancer and prostate cancer-related mortality, the majority of research into underlying genetic causes of prostate cancer has focused on men with European ancestry. Thus, polygenic models developed to date have poor generalizability to more diverse populations.
In an effort to overcome this, Dr. Darst and colleagues undertook an effort to combine genome-wide association study (GWAS) data from diverse populations with the goal of identifying both novel variants and also stronger markers of risk in known regions. This work encompasses more than 110,000 cases with 86 novel risk variants along with hundreds of previously recognized variants. Together, these form a polygenic risk score encompassing 269 variants. Across ethnic groups, this PRS categorization was strongly associated with prostate cancer diagnosis.
These findings were subsequently validated in independent populations of European and African ancestry men. Even more notably, the magnitude of association with PRS category and prostate cancer diagnosis was stronger among younger men. As a result, PRS is likely most informative for predicting prostate cancer risk in men 55 years of age or younger.
While PRS was associated with prostate cancer risk, it did not distinguish between aggressive and non-aggressive phenotypes. However, the preponderance of cases (51% among European ancestry men and 45% among African ancestry men) of aggressive prostate cancer were among men in the upper two deciles of PRS risk categories.
The authors then combined this PRS with known rare variants such as HOXB13, BRCA2, ATM, and CHEK2. Among men of European ancestry, there was a linear association between PRS category and prostate cancer risk, with an increased risk seen for carriers of these rare variants, across each category of PRS.
A similar observation was then replicated among African ancestry men, both overall and among those with aggressive disease. Therefore, they suggest that considering both the presence of rare variants as well as PRS in conjunction can improve predictive models. Using a large, multi-ethnic cohort, the authors found that adding PRS to models including PSA, PSA, and free PSA, and the 4K panel, improved risk prediction of overall prostate cancer, high-grade prostate cancer, and prostate cancer-specific mortality.
Interestingly, the authors found that the association between PRS and prostate cancer risk is relatively stable over time. However, the predictive ability of PSA changes dramatically over time with a substantially stronger association immediately prior to diagnosis than 2-, 5-, or 10-years prior. This overall effect was comparable whether overall prostate cancer risk was considered or it was limited to either aggressive or non-aggressive prostate cancer subgroups.
In simulation studies, the use of PRS to risk stratify men for the initiation of PSA screening (to begin when the absolute 10-year risk of prostate cancer exceeded 4%) was found to reduce overdiagnosis, increase quality-adjusted life years, decrease mortality, and have lower overall costs than an age-based screening approach. This approach was also superior to an MRI-first, age-based screening approach.
Presented by: Burcu Darst, PhD, University of Southern California, Los Angeles, California