NEW YORK – Myriad Genetics unveiled data at the American Society of Clinical Oncology's virtual annual meeting demonstrating that its polygenic score for assessing breast cancer risk can provide accurate estimates for women regardless of their ancestry.
The company launched riskScore three years ago initially as a test for estimating the five-year and lifetime risk of breast cancer for women who had never had the disease and who do not have a mutation in breast cancer-associated genes detected by its next-generation sequencing myRisk Hereditary Cancer test. However, the availability of the around 86-SNP polygenic risk score to date has been restricted to women who self-identified as having European and Ashkenazi Jewish ancestry.
Now, having recalibrated riskScore to provide more accurate breast cancer risk estimates for women in the US, regardless of their genetic ancestry, Myriad is planning to launch this version of the test later this year for women who qualify for myRisk, which gauges mutations in multiple genes conferring high or moderate risk for breast cancer. In 2022, the company will offer riskScore as a standalone, direct-to-consumer test for women who aren't eligible for the myRisk test based on their personal and family history of breast cancer.
Polygenic risk scores rely on the combinatorial power of many SNPs associated with disease risk, but these SNPs have largely been identified in genome-wide association studies done in patients of European ancestry. As such, these scores tend to overestimate disease risk and are less accurate in discerning between high- and low-risk groups in those of non-European ancestry.
For example, studies have shown that Black women have a similar incidence of breast cancer compared to white women in the US. But Myriad's 86-SNP riskScore developed for women of European ancestry overestimates the breast cancer risk in Black women by nearly twofold, said Holly Pederson, who was involved in the effort to recalibrate riskScore and directs medical breast services at the Cleveland Clinic.
Myriad wanted to address this limitation within its test and has been refining riskScore in Hispanic, African American, and other racial groups for several years. Pederson presented the culmination of those efforts at ASCO's annual meeting and unveiled a new 93-SNP riskScore, re-engineered for all ancestries using data from more than 275,000 women.
The new iteration of riskScore will test women not only for 93 breast cancer-associated SNPs but also for 56 ancestry-associated genes, in order to calculate an ancestry-specific result that corresponds to their chances of developing breast cancer in the next five years and over their lifetime. This will preclude women from having to self-report their ancestry, which can be inaccurate, especially for non-European women. "What I found during my years of seeing patients is that many patients weren't entirely sure of their ancestry, and this will no longer be a barrier for care," said Nicole Lambert, president of Myriad Genetic Laboratories.
Weighted by genetic ancestry
The 93-SNP riskScore is weighted according to 56 SNPs associated with ancestral lineage from Africa, East Asia, and Europe, the three places that account for most of the genetic diversity in the US. "There are multiple sub-clusters within each of those [continental] clusters, so using three ancestries is a simplification of the full diversity of human populations," Pederson acknowledged during her presentation at the meeting. "However, these three ancestries together should reasonably represent most of US human genetic diversity."
Data from more than 189,000 women were used to develop the score, and it was validated in data from more than 89,000 women. In these cohorts, 23 percent of women had breast cancer and around 30 percent had a first-degree relative with the disease. Roughly 10 percent of women in these cohorts self-reported as Black or African, around the same proportion self-reported as Hispanic, and around 2 percent self-reported as Asian.
To develop the score, researchers led by Myriad CSO Jerry Lanchbury and Elisha Hughes, the company's director of research biostatistics, first developed polygenic risk scores specific to people of African, Asian, and European descent using data from its own hereditary cancer testing customers with self-reported race, as well as from large consortia and genome-wide association studies. For each of the patients in the development cohort, researchers determined their "fractional ancestry" from the three continents using the 56 SNPs, which then allowed for the ancestry-adjusted calculation of their risk for developing breast cancer based on the 93 SNPs.
"The different alleles found for each SNP in an individual woman are interpreted not only as a function of her ancestral composition, but also on the frequency of that allele's presence in one of the three continental ancestries because they are each different," Pederson said. "An individual woman's polygenic risk score therefore depends not only on her genotype, but also on her ancestral derivation and the frequency of an allele in a given ancestry."
In the validation cohort, researchers wanted to see how well the re-engineered riskScore distinguished between women at high and low risk of developing breast cancer across ancestries and how the new score compared to the 86-SNP test for women of European descent. The study showed that the 93-SNP test was generally an improvement over the 86-SNP test in terms of breast cancer risk predictions for women of all ancestries. In their abstract, the authors noted that the Asian cohort was too small to demonstrate that either score was superior.
Furthermore, the validation study showed that the women with the recalibrated riskScore placed in the highest risk category — the top 1 percent — in fact had a two to threefold greater chance of developing breast cancer compared to average-risk women. For women of all self-reported ancestries, except Black women, if the test placed them in the top decile in terms of risk, they were twice as likely to develop breast cancer compared to average-risk women.
Self-reported African or Black women who were deemed by riskScore to be in the top decile in terms of risk had a 44 percent greater chance of developing breast cancer risk. Pederson said during her presentation that the re-engineered riskScore's ability to assess self-reported Black women's breast cancer risk was "significantly improved" compared to the earlier test but still "suboptimal." She added that the new score's risk discrimination in Black women will likely become more precise with additional data.
"We have known for some time that genomically based breast cancer risk stratification was biased towards SNPs from women with European ancestry and did not perform as well in women of other ancestries," said Corey Speers, assistant professor of radiation oncology at the University of Michigan Rogel Cancer Center. "This study represents an important step to 'level the field' for women of disparate ancestries and more accurately estimate breast cancer risk in these women," added Speers, who researches the biology of aggressive breast cancers and wasn't involved in the riskScore study.
More definitive guidance
Cleveland Clinic, where Pederson works, hasn't yet incorporated polygenic risk scores into standard disease risk estimation workflows. The academic medical center is participating in a prospective study, called GENRE-2, using a 300-SNP breast cancer polygenic risk score developed by Fergus Couch at the Mayo Clinic. In that study, researchers are tracking if this score helps patients make decisions about breast cancer prevention, such as whether to take endocrine therapy.
Outside of the research setting, however, the lack of validation in non-European populations has been a big reason holding up adoption of polygenic risk scores for breast cancer and other diseases. "Clinically, the polygenic risk score is really in its infancy," Pederson said. "Previous to this, really due to concerns over applicability in non-European populations and interpretation and communication of the results, we have not utilized polygenic risk scores at Cleveland Clinic."
Even though Myriad has been offering the 86-SNP riskScore for European women as part of myRisk at no additional cost, Cleveland Clinic has been opting out of that information, according to Pederson. This study, she believes, may very well change that, since to the best of her knowledge Myriad's test is the only breast cancer polygenic risk score that has been calibrated to be informative for all ancestries.
Speers noted as a positive that the training and validation cohorts in the study presented at ASCO included tens of thousands of women and were well balanced in terms of the factors that are most likely to influence breast cancer risk. He is eagerly awaiting peer-reviewed publication of the data, upon which he expects that riskScore will represent "an important step forward for providing equitable and accurate test results for women of all ancestral backgrounds."
With the increasing use of multi-gene tests, like myRisk, which look for pathogenic variants in moderate-risk genes alongside well-known high-risk genes like BRCA1/2, more patients are receiving results where the management implications aren't well established. This can be particularly difficult when women's personal or family history of cancer doesn't offer straightforward clues as to their future cancer risks.
Myriad and others developing polygenic risk scores are betting that these tests will provide risk information when large NGS panels turn up negative or even refine risk estimates when considered alongside mutations in moderate-penetrance genes and relieve uncertainties around patient management. "If patients have a genetic mutation in CHEK2, which is a moderate-risk gene, we tell them they have an estimated lifetime [breast cancer] risk of about 30 percent," Pederson said. "But when you look at the risk stratification that can be achieved by a polygenic risk score, patients may have a risk as low as 6.6 percent over the course of her life or a 70 percent risk, which is similar to a patient with a BRCA1/2 [high-risk] mutation."
Women she treats overwhelmingly want to know this information, Pederson said.
Although she believes that Myriad's new riskScore is "sufficiently validated and calibrated" in all ancestries, she would like to see the test factor in patients' clinical features that also increase their chances of developing breast cancer. At her own practice, patients' decisions about having preventive mastectomies or oophorectomies to mitigate their cancer risk aren't just based on genetic testing but also on a variety of other clinical factors, as well as patients' own priorities for their health and family planning.
The genetic test result is "just one piece of information," she said. "While it is useful in and of itself, it'll be even more useful for a woman to get an estimate in combination with those other [clinical] factors. … It just allows for more precise estimates and better conversations."
Myriad's 86-SNP score for European women incorporates the Tyrer-Cuzick risk model, which evaluates breast cancer risk based on features like age, body mass index, age of first period, and family history of cancer. Myriad is working on integrating clinical risk features into the recalibrated riskScore, Pederson said, adding that this work will likely be presented at a medical meeting by year-end.
Access to all
Myriad is planning to launch the recalibrated riskScore for all ancestries later this year but in the near-term will maintain it as a physician-ordered test offered alongside myRisk. Next year, however, the company wants to launch riskScore as a standalone test through a DTC model for the estimated 93 million women who don't qualify for testing for high- or moderate-penetrance breast cancer risk genes based on stringent personal and family cancer history criteria, as well as the National Comprehensive Cancer Network's guidelines. "This will allow us to provide a precise risk estimate to all women: myRisk for those who qualify, standalone riskScore for those who don’t," said Lambert.
Myriad's DTC plans for riskScore also raise questions about how the company will navigate the regulatory landscape. The US Food and Drug Administration has been clear about its intent to regulate labs marketing genetic tests for assessing disease risk directly to consumers.
23andMe is the only company that currently sells FDA-authorized genetic tests for gauging disease risk, including for cancer, which people can order online without any physician involvement. Other companies offering testing in CLIA-certified labs have found ways around FDA oversight by using third-party physician networks to review and approve customers' online orders. However, this is a controversial model because often the physicians approving test orders don't have much interaction with the patients.
Myriad demurred on its specific regulatory plans, saying that it is still ironing out the specific DTC model it will employ when it launches riskScore as a standalone test next year. "We are currently assessing the regulatory requirements, talking with stakeholders, and creating the specific launch plans," Lambert said.
Pederson backed efforts to broaden access to cancer risk testing, recognizing that using current testing guidelines, largely based on personal and family history of cancer, the healthcare system has identified only a minority of patients with mutations in high-risk genes. At the same time, the rapid introduction of broad NGS panels has made it difficult for physicians lacking genetic expertise to accurately interpret test reports.
As such, a broad marketing strategy for polygenic risk scores must include a robust education plan for patients and providers, Pederson said, including genetic counseling support and resources to help primary care providers interpret test reports and relay nuanced risk information to patients.
Lambert assured that Myriad currently makes genetic counselors available to any doctor or patient ordering germline genetic testing and that these resources would also be available in the consumer-facing service. "We are in the process of evaluating what other services would be desirable as we prepare for the launch of the consumer version in 2022," she said.
Ultimately, given the popularity of DTC genetic testing, "something real like this, if it is priced right and marketed correctly, would really provide women with information that they really want," Pederson said.