The genetic adjustment of prostate-specific antigen could reduce over-diagnosis, de-escalate invasive testing, and improve the detection of aggressive disease in patients with prostate cancer.
The genetic adjustment of prostate-specific antigen (PSA) could reduce over-diagnosis, de-escalate invasive testing, and improve the detection of aggressive disease in patients with prostate cancer, according to data presented at the 2022 AACR Annual Meeting.1
“[PSA], when it comes to prostate cancer screening and detection, remains controversial. And the reason for that is that although PSA can certainly be a warning sign of prostate cancer…PSA elevation can also occur due to other factors, for example, older age, infections, or the presence of other conditions,” Linda Kachuri, PhD, MPH, from the Department of Epidemiology and Biostatistics at the University of California San Francisco, said during a presentation of the results.
“And so, because of these issues with sensitivity as well as specificity, currently, PSA screening is not recommended for systematic use at the population level. In particular, the US Preventive Task Force gave it a C grading. So, although it is a biomarker that has tremendous potential right now, it's not really being used systematically for screening.”
Therefore, the researchers aimed to characterize genetic determinants of PSA levels in men without prostate cancer, hypothesizing that test accuracy could be improved and personalized.
In the 3-part study, the researchers conducted a genome-wide association study (GWAS) of PSA levels in men without prostate cancer; developed and validated a PSA genetic score, calculating a genetically adjusted PSA; and evaluated the clinical utility of genetically adjusted PSA in the context of detection of aggressive prostate cancer.
In GWAS, the largest of its kind, 95,768 men without prostate cancer were identified using data from the UK Biobank, BioVU, PLCO, and Kaiser Permanente cohorts. In total, the study found 128 statistically significant PSA-associated variants (P <5×10-8), 82 of which were newly discovered, previously unreported variants.
“So, what's the next step? What do we do with this information? One way to utilize this is to construct a genetic score,” Kachuri explained. “A genetic score or a polygenic score, as these are commonly referred to, is really just a simple cumulative measure of someone's genetic predisposition. The way that these are calculated is that we construct a weighted sum of somebody's genotypes for the variance of interest.”
To validate their PSA genetic score, the researchers used data from 2 cancer prevention trials: the Prostate Cancer Prevention Trial (PCPT; n = 5725) and the Selenium and Vitamin E Cancer Prevention trial (SELECT; n = 22,173).
The PSA genetic score explained 7.3% (P = 7.0×10-98) of variations in baseline PSA in the PCPT trial and 8.8% (P = 7.0×10-476) in the SELECT trial. Moreover, the score was not associated with prostate cancer status, according to the abstract, in the PCPT (OR, 0.98; P = .71) or SELECT (OR, 1.04; P = .98) trials, “which confirms that it reflects benign PSA variation,” the researchers wrote.
“This is encouraging, because we're able to show that these genetic variants are robust predictors to be effective,” Kachuri said.
Next, the researchers explored the potential clinical utility of PSA genetic adjustment by examining reclassification at thresholds used for biopsy referrals in a real-world setting at Kaiser Permanente.
“The idea is that every person has a particular value for this genetic score, which captures their unique genetics. What we'd like to do is account for their information in their PSA value,” Kachuri said. “One way that you can think about this is that the observed PSA value is a little bit noisy, it can reflect variation in PSA due to cancer, but it can also capture other factors that influence PSA. And one of these factors is genetics. What we're effectively doing here is that we're using this genetic score to try to correct some of that noise. As a result of this, someone's genetically-adjusted PSA value can be either higher or lower than their measured PSA based on their genetics.”
The researchers found that by using the PSA genetic score instead of PSA, 19.6% of negative biopsies would have been avoided in men without prostate cancer, and 15.7% of patients with a Gleason score of less than 7 would have been reclassified as ineligible for biopsy.
Using the same 2 studies, the researchers evaluated genetically adjusted PSA in the context of detection of aggressive prostate cancer, defined as Gleason score 7 or more, PSA of 10 ng/mL or more, T3 or T4 stage, and/or distant or nodal metastases. They found that the genetic risk score, vs baseline PSA, had an increased association with aggressive prostate cancer and induced a higher area under the curve (AUC) in the PCPT trial (OR, 3.03; P = 3.5×10-7; AUC, 0.72 vs. 0.68, respectively) and the SELECT trial (OR, 3.37; P = 3.5×10-11; AUC, 0.78 vs. 0.74).
“Our findings are exciting because we're able to show that we can use these genetic discoveries that are coming out of Genome Wide Association Studies to potentially improve the detection of prostate cancer, and hopefully try to make PSA a more useful and accurate screening biomarker,” Kachuri concluded, adding that future studies should include PSA genetic score performance in ancestrally diverse population, as this study was mostly in men of European descent, and also the characterization of genetic adjustment of PSA in a range of clinical settings and patient populations.
Louis M. Weiner, MD, from Georgetown Lombardi Comprehensive Cancer Center, who served as the press briefing moderator, commended Kachuri, also questioning how she sees this technology being adapted into routine practice.
“I'm actually quite optimistic that this is something that can be relatively easier to translate than perhaps some other applications of genetic risk scores,” she answered. “And part of that is that genetic testing and measurement of genetic information is already happening in a lot of different contexts and different health care systems. This type of study does not require whole genome sequencing, it can be something that can be easily accomplished with genotyping, which is not as expensive and also easier to implement.
Also, the reason why I'm optimistic about the translation of this is because the complicated part is calculating the genetic risk score. But then the implementation is straightforward, because we're still using PSA, which is a biomarker that people are very familiar with, and clinicians are familiar with. This has a slightly lower adoption period.”