NEW YORK – A tool to estimate how a person's genetics influence their prostate-specific antigen (PSA) levels may refine prostate cancer screening efforts.
Scientists from the University of California at San Francisco developed a PSA-specific polygenic score (PGS-PSA) that appeared to explain nearly 10 percent of constitutive PSA variation. Their study was published last week in the journal Nature Medicine.
Measuring PSA levels is the most common tool for screening for prostate cancer. Individually varying PSA levels, however, means that this screen often leads to overdiagnosis and overtreatment.
"If we identify those genetic factors that contribute to the PSA production, and then take away that part, what remains of the PSA levels is more likely to be indicative of a tumor," said Rebecca Graff, assistant professor of epidemiology and biostatistics at UCSF and one of the study's coauthors.
Graff and her colleagues conducted genome-wide, ancestry-stratified meta-analyses of PSA levels among nearly 96,000 men without prostate cancer, using data from the UK BioBank, the Kaiser Permanente Genetic Epidemiology Research on Adult Health and Aging (GERA) cohort, the PLCO Cancer Screening Trial, the Vanderbilt University Medical Center BioVU, and the Malmö Diet and Cancer Study.
The meta-analysis revealed 128 variants associated with PSA production, 82 of which mapped to genetic regions where PSA signals had not previously been reported.
Because the datasets involved skewed heavily toward European ancestries, the majority of the PSA-associated variants — 96 — reached genome-wide significance in that population. Three variants reached genome-wide significance in individuals of East Asian ancestry, as did two among those of Hispanic/Latino ancestry and one in those of African ancestry.
The researchers validated the PGS-PSA in two separate GWAS datasets: the Prostate Cancer Prevention Trial (PCPT) and the Selenium and Vitamin E Cancer Prevention Trial (SELECT).
Here, the PGS-PSA accounted for approximately 8 percent of the variation in baseline PSA levels in the PCPT cohort and for between roughly 10 and 11 percent of variation among those in the SELECT cohort. In both cases, the PGS-PSA performed best in men of European ancestry.
Jonathan Shoag, an assistant professor of urology at Case Western Reserve University who was not involved with the study, commented that this validation test showed little to no advantage over just PSA alone.
"This is an interesting attempt to adjust for genetic differences in PSA levels to make screening better for prostate cancer," he said, while adding "I would interpret this as a largely negative study in the context of our tools today for understanding prostate cancer risk."
The researchers also assessed how their PGS-PSA might impact biopsy eligibility. Analyzing 2,363 men who had negative biopsies and 24,811 who did not have a biopsy, all from the GERA dataset, they found that inclusion of the PGS-PSA reclassified nearly 30 percent of biopsy-negative men as being below the threshold for biopsy eligibility.
Adjusting PSA values for genetics appeared to confer the most benefit in cases of aggressive prostate cancer, defined as those with Gleason scores of at least seven, among men of European ancestry. Cancers with a Gleason score of seven are often called moderately differentiated or intermediate grade, while those with scores from eight to 10 are called poorly differentiated or high grade.
Case Western's Shoag remained skeptical, however, that the current results show real benefit.
"We have better tools to understand prostate cancer risk," he said. "We have a lot of biomarkers which do far better than the genetically adjusted PSA levels, such as prostate MRI, 4K score, ExoDx, [and] IsoPSA, at predicting prostate cancer risk."
Graff acknowledged that the current work is at an early stage and needs to undergo refinement before becoming useful in a real-world setting.
"[The] GERA cohort does not exactly offer real-world data," she said, "but it is much closer to the real world than clinical trials. We had information on men who had negative biopsy outcomes, which is difficult to obtain in other studies. However, since some of the data from this cohort was included in the GWAS used to develop the PSA genetic score, it is likely that we have overestimated the number of negative biopsies that can be prevented with genetic adjustment."
Moving forward, the UCSF researchers are working to improve the predictive power of the PGS-PSA, particularly in men of non-European ancestry.
"That's the next step that we're taking before we think about how we could potentially translate the research into a real-world setting," Graff said.
"We're already working on a follow-up study," added Linda Kachuri, the paper's first author. "It's very much a matter of increasing sample size for discovery, so having a larger GWAS for PSA levels, that includes a larger number of people from those ancestry groups."
The UCSF team is currently working with the Department of Veterans Affairs' Million Veterans program on that study and hopes to submit an abstract describing their findings so far at November's American Society of Human Genetics annual conference.
Although a clinically validated PGS-PSA has clear commercial applications, Graff and Kachuri said that there are currently no plans to pursue commercialization.
"We see our approach as something that's complementary to having a genetic score for prostate cancer," Kachuri said.
A "best possible" predictive model for prostate cancer risk, she added, will need to include both estimations of genetic risk and of PSA production.
"Improving the accuracy of PSA is one piece of it," she said.