CHICAGO (GenomeWeb) – Researchers used data from approximately 100,000 women assessed on Myriad Genetics' myRisk Hereditary Cancer test to evaluate the magnitude of breast and ovarian cancer risk associated with mutations in 25 genes on the next-generation sequencing panel.
Although the study involved the largest cohort to date, the uncertainty around the clinical significance of less-studied, moderate-risk genes on this panel remains and needs further research, experts reviewing the data said at the American Society of Clinical Oncology's annual meeting held here this week.
In two analyses presented at the conference, researchers utilized Myriad's database containing the NGS results from 95,561 women who received testing between September 2013 and September 2015. From test requisition forms that physicians filled out, researchers gathered information on patients' personal and family cancer history, age, and ethnicity.
Researchers primarily used multivariable logistic regression models to estimate the relative risk between pathogenic mutations detected in patients and breast and ovarian cancer risk. Although, in the ovarian cancer portion of the study, researchers also performed a secondary case-control analysis involving data from women without cancer who were matched to those with cancer in terms of age, race, and family history.
In the overall population, there was at least one pathogenic mutation in 7 percent of patients, while 33 percent received a variant of unknown significance. The analysis included 26,000 women with breast cancer and 5,000 with ovarian cancer. The majority of women were of European descent, and researchers excluded women who had received prior BRCA testing.
In the breast cancer subset, 10 percent had at least one pathogenic variant. Of these variants, 52 percent were in BRCA1/2, 40 percent were in other breast cancer genes, and nearly 6 percent were in Lynch syndrome genes. Eight genes were associated with conferring a two- to six-fold increase in breast cancer risk.
Among the ovarian cancer patients, 14 percent had a pathogenic mutation — around 64 percent of these mutations were in BRCA1/2; 9 percent in Lynch syndrome genes; and 27 percent in other genes. Approximately 15 percent of detected mutations in ovarian cancer patients were in genes such as ATM that are traditionally thought of as conferring moderate risk for breast cancer. Eleven genes were associated with between a two- and 40-fold increase in ovarian cancer risk.
Allison Kurian from Stanford University School of Medicine noted during an ASCO presentation 'that this is the first time that mutations in the ATM gene, which show up in approximately .5 percent and 1 percent of breast cancer patients, have been found to increase ovarian cancer risk. "That's interesting because we are finding a lot of these [mutations] in breast cancer patients," she said. "So, it really becomes important [for] their ovarian cancer risk and what to do about it." Kurian noted though that this finding needs to be confirmed in further studies.
Some limitations of the study noted by Kurian and others at the meeting were that the data derived from clinicians' reports could not be verified, and the cohort was largely European. "The rate of variants of uncertain significance was high and the clinical implications of some of these findings are not clear," Linda Mileshkin from Peter MacCallum Cancer Centre, Melbourne, said at the meeting, reviewing the data on ovarian cancer risk genes.
Experts at the meeting that reviewed the breast and ovarian data lauded the size of the cohort. However, Nadine Tung of Beth Israel Deaconess Medical Center noted that researchers used a "less conventional" logistic regression study design to explore the extent to which the genes conferred breast cancer risk. She cautioned that the lower odds ratios of the significantly associated breast cancer genes needed to be confirmed with a more conventional case control study design.
The study may be enriched for breast cancer families, Mileshkin noted as a reason for the identification of mutations in traditionally breast cancer genes in ovarian cancer patients. Some of the significant risk associations found for genes, such as STK11, "may be speculative and unreliable" due to the small number of patients with them, she said.
Importantly, in the ovarian cancer subset, few women had one or more first-degree relatives with the disease. This, Mileshkin said, highlights the need for doctors to refer all newly diagnosed women with high-grade ovarian cancer for genetic testing, regardless of family history.
BRCA1 and BRCA2, and Lynch syndrome genes are by now well known to oncologists and have been studied extensively. But because NGS testing detects mutations in more common genes that confer moderate cancer risks, and yield higher rates of VUS, it is becoming challenging for doctors to figure out what, if anything, to do based on the results. "Several people in the literature have described that perhaps the development of these panels has outstripped our ability to understand what [the results] mean," Mileshkin said.
For genes such as ATM, for example, the field currently doesn't have "well qualified" information on lifetime risks, she said, which creates uncertainty for doctors trying to determine which women to screen more frequently and which to offer risk-reducing surgical interventions. The latest National Comprehensive Cancer Network guidelines list genes it has determined to have sufficient evidence for additional screening and surgery, and some of the newer genes on NGS panels fall in the insufficient evidence category.
"The problem is management recommendations rely very heavily on accurate risk estimates, and therein lies the rub," observed Tung.
The broader availability of NGS testing may "result in more anxiety and inappropriate recommendations if it's not used with appropriate genetic counseling," Mileshkin added. She noted that in the future large, prospective case-control or family studies are needed to better understand that penetrance and prevalence of lower risk genes, as is international data sharing to facilitate reclassification of VUS.
Large studies like the one conducted using Myriad's database are "very valuable," Tung said. "But I think we're left with the impression that the clinical validity associated with a lot of these moderate-risk mutations is just not yet robust," she said. "The risk estimates are quite variable, and that translates to the fact that the clinical actionability, let alone the clinical utility for many of these moderate risk mutations, is just not certain."