NEW YORK – Diagnostics firms Myriad Genetics and Allelica say their polygenic risk scores (PRS) are becoming more effective in assessing breast cancer risk in patients with non-European ancestry, though the firms continue to seek out large, diverse datasets for further study.
Both firms have conducted research focused on using PRS to predict breast cancer risk in patients with African, Asian, Hispanic, or mixed ancestry — ancestry groups that had been overlooked in previous PRS. The PRS tests are now available for patients with any ancestry as these firms validate their tests in datasets that have greater diversity. However, some ancestries remain underrepresented in those datasets, a gap the companies hope to address.
Myriad's PRS test, called RiskScore, was originally available only to European and Ashkenazi Jewish ancestry patients due to limitations in the large genome-wide association studies used to validate the test, said Shelly Cummings, VP of oncology medical affairs and head of oncology genetic counseling at Myriad. The SNPs and clinical data originally used to validate the RiskScore test were largely from Northern European individuals, she added. In 2021, the firm launched its updated RiskScore test for patients of all ancestries.
"We decided to dig deeper into different ancestries to look at these SNPs," Cummings said. "We studied [the European SNPs] to see if they perform the same way in multiple ancestries and found there are some subtle differences."
Allelica's ancestry-specific PRSs are a more recent addition to this space and have only been available to order since September. However, the company is pushing ahead to validate its tests in different ancestries and is partnering with institutions to access more diverse data, said Allelica CSO George Busby.
"What we've tried to do with these multi-ancestry PRSs is to ensure that we've developed a test that can capture genetic risk across the vast majority of individuals," Busby said. "There's still work to do, but it's an advance [in the field] to be able to say we can accurately assess risk across these different groups even if there's sort of marginal differences in that [assessment]."
Reclassifying risk in diverse patients
Studies exploring Myriad and Allelica's multiple ancestry PRS tests at the American Society of Clinical Oncology's annual meeting this month suggest that they can refine estimates of patients' risk. In practice, PRS, which assesses sometimes hundreds of SNPs associated with disease risk, can be combined with clinical factors, such as family history of cancer or age of menopause, to predict an individual's five-year and lifetime risk of developing breast cancer. For those who are considered high risk, their doctors can do more frequent or enhanced screening for cancer.
Olufunmilayo Olopade, director of the Center for Clinical Cancer Genetics and Global Health at the University of Chicago Medical Center, detailed results of a study that evaluated Myriad's multiple ancestry PRS in more than 203,000 patients with a representative proportion of patients with Black/African ancestry, East Asian ancestry, South Asian ancestry, and Hispanic/Latino ancestry.
The study showed that the PRS test adjusted for ancestry could change which patients were identified as having a higher or lower risk of breast cancer development compared to the Tyrer-Cuzick model alone, a measure of breast cancer risk based on family history of cancer and other clinical factors.
The multiple ancestry PRS (MA-PRS) in addition to the Tyrer-Cuzick model reclassified breast cancer risk for 1 in 6 women in the study, or 15.7 percent. About a quarter of women who were classified as high risk, or having a greater than 20 percent lifetime risk of breast cancer, by Tyrer-Cuzick were reclassified as low risk when the PRS was added. The multiple ancestry PRS also identified about 10 percent of women at high risk of developing cancer who had been classified as low risk by Tyrer-Cuzick.
"The MA-PRS was well calibrated and performed well in predicting risk of breast cancer in each ancestral population," Olopade said. "For women of all ancestries, the MA-PRS has significantly improved risk prediction beyond clinical factors. Incorporation of MA-PRS into traditional risk assessment models advances the field and may lead to more accurate breast cancer risk prediction on diverse populations."
Similarly, data from a validation study of Allelica's PRS test in multiple ancestries, also presented at ASCO, showed that the addition of its PRS to Tyrer-Cuzick could also led to breast cancer risk reclassification.
Daniel Morganstern, a medical oncologist at Hartford Healthcare Cancer Institute, reported at ASCO that Allelica's ancestry-specific PRSs were well calibrated across African, East Asian, South Asian, European, and mixed-ancestry patients. The study used a dataset of more than 175,000 women of diverse ancestry in the UK Biobank and Women's Health Initiative cohorts.
The combination of Allelica's PRS with the Tyrer-Cuzick model led to about 13 percent of African ancestry patients, 19 percent of Asian ancestry patients, and 14 percent of mixed-ancestry patients who were low risk by Tyrer-Cuzick alone to have their risk level upgraded when the PRS was added.
Meanwhile, among patients classified as high risk by Tyrer-Cuzick, 25 percent of African ancestry, 48 percent of Asian ancestry, and 36 percent of mixed-ancestry patients had their risk downgraded when PRS was added. Morganstern said that including PRS with Tyrer-Cuzick clinical factors led to an average improvement of 8 percent in the accuracy of risk classification overall.
"We have built and validated three ancestry-specific polygenic risk scores with performance at least as good and often better than previously published polygenic risk scores," Morganstern said. "These results suggest that genetic information from polygenic risk scores has an important role to play in breast cancer risk prediction in the future."
Remaining gaps
While both studies showed PRS could inform breast cancer risk predictions in three large ancestry groups, some gaps remain. The datasets used for both Myriad and Allelica's research had low representation of Indigenous American individuals, and Olopade noted that there were few Pacific Islander patients included the Myriad study.
While the three continental-level ancestries — European, African, and Asian — can account for most people, Allelica's Busby acknowledges that using these three categories to validate the test means it may be less accurate for Indigenous groups, for example.
"Everyone who is given a test has to fall into one of either one of those three groups or is a mixture of those three groups," he explained. "It turns out that actually that's a relatively good approximation at least at a high level of most ancestry across individuals. But we're very aware that [Indigenous Americans] are a group of people that are potentially not well captured by our test."
He noted that the mixed-ancestry PRS test can still assess risk in this group because it combines risk markers from across the three continental ancestry groups. "We just have to be transparent and communicate the fact that that performance varies, but there still is utility even if the performance is not exactly the same in all groups," he added.
To further validate its PRS tests, Allelica has built a platform in which its customers around the world can share data on the test's usage to further improve its performance. Busby added that customers from South America, North America, Europe, and Asia are currently using that platform. The company is also working with Mount Sinai in New York to evaluate the multi-ancestry PRSs in its BioMe BioBank, which includes about 30,000 individuals.
Allelica has also released a virtual educational program to teach clinicians about how to use PRS in practice for cancer risk assessment. The series also includes a lecture on representation gaps in genetic datasets and how that affects genetic cancer risk prediction.
Myriad is also continuing to dig deeper into ancestry subgroups in longitudinal studies, Cummings said. The company is partnering with health organizations and medical societies to help improve access to genetic testing for patients from diverse backgrounds.
"This is a gap across all of medicine, not having [data from] diverse populations that are getting genetic testing," Cummings said. "We will continue to do studies in this area, like the research presented by Dr. Olopade, to look at subpopulations that we didn't have opportunities to look at before."
Myriad is also working to raise awareness among patients and educate clinicians about PRS for diverse ancestries. Cummings said her group at the company focuses on clinician education about these tests and how to use them in practice.
"[Patient awareness of PRS] stems from enhancing education at the clinician level," she said. "Many of these tests that are available now weren't in the training that clinicians had, so it's hard to stay up to date. My medical team and others are trying to educate clinicians on how few individuals are actually getting genetic testing and precision medicine, and that [these tools] should be used across all groups, all ancestries, all populations."