From an editorial commenting on a recent prostate cancer screening trial using polygenic risk scoring to stratify patients:
“After decades of undergoing epidemiologic study, prostate cancer remains an enigmatic disease; the only established risk factors are older age, family history of prostate cancer, and African ancestry (apart from uncommon cases among men with variants in the genes associated with hereditary breast cancer and the Lynch syndrome). Of interest, then, is a study of a polygenic risk score as an adjunct to screening, the results of which are reported by McHugh et al. in this issue of the Journal.
Currently, screening for prostate cancer is problematic. The prostate-specific antigen (PSA) test gives rise to many false positive results. Imaging of the prostate with magnetic resonance imaging (MRI) has been interposed between actionable PSA levels and the decision to obtain multiple biopsy samples. Treatments for prostate cancer are associated with substantial morbidity (related to urinary incontinence with surgery, bowel incontinence with radiotherapy, and erectile dysfunction with both). [..]
The advent of genome-wide association studies led to the understanding that hundreds of genetic variants, each associated with a low relative risk of prostate cancer, underlie the genetic predisposition to the disease. When such variants are combined in the calculation of a polygenic risk score, which is obtained by running a genome array, the distribution of risk in the population approximates a normal distribution. Men in the highest decile of risk for prostate cancer have more than 3 times the risk of those who are at average risk, and men in the lowest decile have about one third the risk of those at average risk. A polygenic risk score does not yield a diagnosis of disease; rather it is a risk factor that may be included as part of a screening program. The genotyping panel used to calculate polygenic risk scores in the study by McHugh et al. is validated only for men of European ancestry and performs less well among men of African and other ancestries.
McHugh et al. examined the outcomes of a workup in men with a polygenic risk score in at least the 90th percentile for risk of prostate cancer. Of 6393 participants who had their polygenic risk score calculated with the use of DNA extracted from saliva, 745 had a score in at least the 90th percentile of risk in the population, and 468 agreed to undergo MRI and biopsy. A total of 187 participants received a diagnosis of prostate cancer at a median age of 64 years, of whom 103 had disease classified as intermediate or higher risk and therefore warranted further treatment according to National Comprehensive Cancer Network criteria. Of these 103 men, 74 (71.8%) had cancer that would not have been detected with the use of the screening recommendations currently used in the United Kingdom because these participants had either a PSA level lower than the action threshold or an MRI result that would not have led to a recommendation for biopsy. [..]
Thus, if a screening program started with an assessment of a polygenic risk score, a substantial number of clinically significant cases would be discovered that otherwise would have been missed. Genome arrays can be run in large research studies for less than $20, need only be obtained once in a person’s lifetime, and can also be used to calculate a polygenic risk score for a wide variety of other diseases. Should a polygenic risk score be a first step in a screening pathway for an asymptomatic man who is worried about his risk of prostate cancer?
Currently, this question is moot. Starting a prostate cancer screening program with the assessment of a polygenic risk score would require large-scale investment in the management and analysis of genome arrays and would raise a host of questions about the storage of genetic data on populations and the use of the data in calculating and potentially counseling on the risk of many other diseases. The idea that genetic testing could be a consumer-led activity, as exemplified by 23andMe, seems to have stalled in the marketplace. In addition, the sensitivity and specificity of PSA in screening for prostate cancer could be improved with the addition of other blood-based and urine biomarkers. The study by McHugh et al. suggests that, if available, a polygenic risk score for prostate cancer would be a useful component of a multimodality screening program that assesses age, family history of prostate cancer, PSA, and MRI results as triage tools before biopsy is recommended. To make this integrated program a reality, however, changes to infrastructure would be needed to make running and analyzing a regulated genome array as easy as requesting a PSA level or ordering an MRI. Clearly, we are far from that future.
A possible first step that would require less infrastructure could be to order a polygenic risk score only for men with a positive PSA result, then use the polygenic risk score to determine who should undergo an MRI, and then use all the information to determine whether biopsy is recommended. Studies such as the new TRANSFORM study in the United Kingdom may help achieve this goal and will attempt to overenroll men of African ancestry.
Critics of polygenic risk scores are justifiably concerned that giving people access to risk scores for dozens of diseases could lead to demand for cost-ineffective screening tests. A polygenic risk score should always be viewed as a measure of risk and not as a stand-alone screening test, and the potential contribution of each polygenic risk score to the relevant disease-specific screening pathway will need careful assessment. The current study is a first step on a long road to evaluating new components of any disease screening pathway.”
Full editorial, DJ Hunter, NEJM, 2025.4.9