NEW YORK – New research suggests that a multi-ancestry polygenic risk score (PRS) based on 269 genetic variants is just as good at predicting prostate cancer risk as genome-wide PRS (GW-PRS) based on a much broader set of disease-related variants.
"This investigation suggests that current GW-PRS approaches may not improve the ability to predict prostate cancer compared to PRS269 developed from multi-ancestry GWASs and fine-mapping," researchers wrote in a paper published in the American Journal of Human Genetics on Monday.
For that study, an international team led by investigators at the University of Southern California and the Fred Hutchinson Cancer Research Center compared multiple GW-PRS with a PRS based on just 269 variants, focusing on their predictive performance in men of African or European ancestry. In large testing and validation cohorts, the 269-variant PRS provided as much predictive information as scores based on more than a million variants.
"[T]hese findings suggest that if PRS were to be used in a clinical setting, a panel of 269 variants rather than [more than 1 million] variants would be able to capture genetic risk of prostate cancer that can be attributed to common genetic variation," co-first and corresponding author Burcu Darst, a researcher affiliated with USC's Keck School of Medicine and the Fred Hutchinson Cancer Center, said in an email.
In particular, after training the GW-PRS models with data for the same diverse individuals used to establish the PRS269 score — including 107,247 individuals with prostate cancer and more than 127,000 without — the researchers looked at how well the scores performed on independent case-control sets spanning some 1,586 cases and 1,047 controls with African ancestry and another 8,046 cases and 191,825 controls of European ancestry.
Their results pointed to similar predictive performance for the GW-PRS and the more targeted 269-variant risk score in these cohorts, with the PRS269 score edging out genome-wide approaches in some cases — results that were backed up by validation testing on nearly 20,000 more prostate cancer cases and almost 263,600 controls of European or African ancestry from the Million Veteran Program.
"Our work shows that our PRS269 performs better or as well as the best performing genome-wide PRS approach in both European and African ancestry men," Darst noted.
Such a pattern "is not what's been observed for other traits — often, genome-wide PRS perform better than PRS of variants significantly associated with the outcome," she explained. "Although we don't fully know the reasons for this discrepancy, we suspect that contributing factors could include our careful approach of selecting the 269 PRS variants with multi-ancestry populations, and a statistical fine-mapping approach to select variants most likely to be causally associated with disease risk, along with the strong heritability of prostate cancer."
Based on such results, the authors suggested that the 269-variant PRS will be useful for distinguishing men of African or European ancestry who have a particularly pronounced prostate cancer risk, and prioritizing these patients for enhanced screening and surveillance, while sparing genetically low-risk men from undergoing excessive screening.
"Prospective clinical investigations are warranted to evaluate the clinical utility of PRSs in prostate cancer screening and early detection," they wrote.