SAN ANTONIO – Myriad Genetics this week presented the complete validation data for an 86 SNP breast cancer risk score that it believes can help inform the care of the large proportion of women who lack mutations in well-known cancer-linked genes.
At the San Antonio Breast Cancer Symposium yesterday, Myriad researchers presented a study involving more than 1,600 women of European ancestry with and without breast cancer. They showed that the multi-SNP riskScore combined with the Tyrer-Cuzick model (a risk estimate factoring in various patient characteristics and their family history of breast cancer) was a superior predictor of a woman's five-year and lifetime risk for breast cancer compared to using Tyrer-Cuzick alone.
"This combined score has significantly greater discriminatory accuracy than Tyrer-Cuzick," Elisha Hughes, senior manager of research biostatistics at Myriad and a lead investigator in the validation study, said at SABCS. "With a general population sample, average risks from the combined score will match pretty identically to the average risks in Tyrer-Cuzick, [indicating] the calibration is the same."
The riskScore test is an adjunct to myRisk, a next-generation sequencing panel that gauges 28 genes associated with elevated risk for eight hereditary cancers. Specifically, mutations in 11 breast cancer-linked genes in myRisk ― BRCA1, BRCA2, TP53, PTEN, STK11, CDH1, PALB2, CHEK2, ATM, NBN, BARD1 ― confer high and moderate risk but show up in only about 10 percent of women.
In an effort to develop a measure with broader utility, Myriad recently launched riskScore, which the company claims can be informative for the majority of women of European descent who don't have pathogenic mutations in these genes. The women who test negative for the myRisk breast cancer risk genes "are often left with no answer at all," Myriad's Chief Medical Officer Johnathan Lancaster said in an interview at the conference. "They may know they have a family history of breast cancer, and yet, they don't have a genetic explanation."
When women unaffected by breast cancer are negative for the well-known high-risk genes, such as BRCA1/2, TP53, and others, clinicians have the difficult task of figuring out how to advise them in terms of increased surveillance with MRI, chemoprevention with medication, or bilateral mastectomy. "In [mutation] negative patients, it's a very difficult discussion, especially the pros and cons of bilateral mastectomy," said Banu Arun, co-director of the clinical cancer genetics program at MD Anderson Cancer Center, in an interview. "In the clinic, we all want models or scores that can help guide us through those discussions."
Myriad further evaluated the combined algorithm in a cohort of nearly 6,500 women of European descent without breast cancer and a negative result on myRisk, and reported that 38 percent had a lifetime risk of greater than 20 percent and 7 percent had a lifetime risk of greater than 35 percent.
In the general population, around 12 percent of women will get breast cancer in their lifetime. However, when a woman's lifetime risk for breast cancer increases to 20 percent or more, based on Tyrer-Cuzick for example, the American Cancer Society, the National Comprehensive Cancer Network (NCCN), and the American Congress of Obstetricians and Gynecologists recommend that she should be offered annual MRI screening and mammography. Those with greater than 35 percent lifetime risk could also be candidates for more aggressive interventions, Lancaster said.
At SABCS, a number of experts noted there is data suggesting that Tyrer-Cuzick can overestimate breast cancer risk in certain instances, and questioned how this might impact the ability of Myriad's combined risk score to identify those at reduced risk. There is a need in the clinic to not only identify high-risk patients but also those at reduced risk, Arun said, so those below the 20 percent risk threshold can avoid MRIs and other interventions.
RiskScore reports out a breast cancer risk estimate as a continuous variable. In the validation study, risk estimates in the 6,500-patient cohort ranged from a low of .88 percent to a high of 66 percent. Lancaster noted that the validation study provides high confidence in the accuracy of riskScore, but it's not yet clear what clinicians will do with this information.
At SABCS, expert opinions varied as to whether this and other polygenic scores were ready to be implemented in the clinic. "I do think in the future we will have more assays that are looking at this," said Julie Nangia, an assistant professor at Baylor College of Medicine, noting that she has ordered Myriad's riskScore on a few patients. However, Nangia added that there needs to be more experience with the test before doctors fully understand its clinical utility.
When clinicians first learned about the increased risks conferred by BRCA1 and BRCA2, many were reluctant to recommend their patients take preventive action, such as mastectomies and oophorectomies, recalled Lancaster, who before joining Myriad directed the Center for Women's Oncology at Moffitt Cancer Center. "At that time, a lot of docs didn't have confidence in their understanding of what having a BRCA1 mutation meant," he said. "It took a good number of years for … people to realize that it is okay to do something pretty radical based on that risk [information]."
He predicted a similar trajectory in terms of the integration of riskScore in clinical practice.
In the case of NGS panels, such as myRisk, many community oncologists already feel uncertain about how to manage patients with mutations in genes that only modestly increase a woman's chances of breast cancer or that have unclear penetrance. SNP-based risk scores could also provide more definitive risk estimates when used alongside modest risk genes, such as ATM.
Women with BRCA1 or BRCA2 gene mutations have up to a 65 percent and 45 percent lifetime risk of breast cancer, respectively, and guidelines recommend more frequent screenings and risk-reducing surgeries starting at certain ages. Comparatively, for women who have mutations in the ATM gene, which has shown to increase lifetime risk for breast cancer to between 17 percent and 52 percent, the NCCN provides specific guidelines for screening but states there is insufficient evidence to issue recommendations for risk-reducing mastectomy.
Even in the case of BRCA1/2, not all mutations are equal, Lancaster said. "We're still trying to understand within a family why one member developed breast cancer at 31 because she carries a BRCA1 mutation and a sister with the exact same mutation doesn't develop it at all or develops it in her 70s or 80s," he said.
Although Myriad is currently only marketing riskScore for women with negative myRisk results, according to Lancaster, a number of experts have expressed interest in working with the company to study whether these SNPs can improve understanding of breast cancer risk for women who have mutations in one of the 11 genes on the myRisk panel.
Ian Campbell from the Peter MacCallum Cancer Center in Australia said that his group has developed a polygenic breast cancer risk test and plans to use it on the 5 percent to 10 percent of women who have mutations in low-to-moderate risk genes that aren't by themselves actionable.
Fergus Couch, breast cancer genetics researcher at Mayo Clinic, disagreed that polygenic risk scores are ready to be implemented in the clinic. However, he felt these scores should continue to be used in the research setting since they appear to be a useful addendum to traditional risk measures, like Tryer-Cuzick, for women who don't have mutations in high-risk genes.
One example is the WISDOM trial, which is a preference-tolerant randomized trial comparing a risk-based mammography approach against annual mammography in 100,000 women ages 40 to 75. For women receiving personalized screening, researchers are using the Breast Cancer Surveillance Consortium's model to assess their non-genetic risk factors, as well as testing them for nine high-penetrance genes and with a 96-SNP polygenic test performed by Color to determine their five-year risk for developing breast cancer. Both the BCSC and the polygenic risk score (which will soon include 153 SNPs) have been calibrated for use in different ethnicities.
Myriad's riskScore currently is only for women whose doctors indicate they are of European descent. However, every patient tested on myRisk, regardless of ethnicity, will also get their lifetime disease risk estimates according to Tyrer-Cuzick.
Lancaster said Myriad is collecting data from other ethnic groups so that riskScore is more widely applicable. "We want to be able to offer this to every single patient irrespective of ethnicity," he said.
Although there is a need for polygenic risk scores in the clinic and some are using them, most doctors will likely want to see prospective data on whether these measures help with their decision making and improve patient outcomes before broadly adopting them, Arun said. It would also help if there was eventually data on how these polygenic risk scores compare with each other.
"They are all built on different cohorts," she said. "But wouldn't it be great, in the ideal world, if all of these cohorts and risk scores were compared."
Ultimately, in order to make use of these risk models and definitively understand their utility, as many as a million women need to be studied in trials like WISDOM, said Laura V 'ant Veer, one of the principal investigators in that effort. "This is my plea during this meeting, that we team up and discuss how to bring this all together," she said.