Although players in the burgeoning consumer genomics industry have agreed to standardize certain parts of their genetic risk screening offerings, they have yet to agree on clinical validation issues critical to protecting the public health, according to a top US government health official.
During a December conference on consumer genomics hosted by the Centers for Disease Control and Prevention and the National Institutes of Health, representatives from three leading firms — 23andMe, Decode Genetics, and Navigenics — presented voluntary industry standards they had developed with the help of the Personalized Medicine Coalition.
Following the meeting, the PMC published the standards in a consensus document.
"The industry standards issued in the PMC document were just a good start but they are really not where they need to be," Muin Khoury, director of the Office of Public Health Genomics at the CDC, told Pharmacogenomics Reporter this week. "There needs to be some more work."
In the consensus document, the companies agree to use DNA chips with more than 99 percent accuracy covering between 600,000 and 1 million SNP markers; to characterize individual risk in terms of both absolute lifetime risk and risk relative to the general population; to use SNPs replicated in at least two well-powered studies; and to be transparent about their methods and criteria.
However, the companies still diverge on several fronts, such as how they define different disease states; the type and number of SNPs they use to calculate genetic risk for a particular disease; and which published studies they consider when deciding which SNPs to use.
"The three gene-scan companies have reached consensus on some science standards (such as standardizing the calculation of individual risk), and have identified points of disagreement that will require the federal government to set standards (for example, the point at which a SNP is considered to be adequately validated)," the PMC explains in its most recent newsletter to members.
"The gene-scan companies … recognize that subjective standards (such as what types of information they should give customers) must also be defined," the PMC notes in the newsletter.
Khoury remained skeptical that the companies have made much progress in aligning the so-called "subjective" aspects of their businesses since the December meeting.
Amy Miller, PMC's public policy director, acknowledged to Pharmacogenomics Reporter that the work to try to bridge the areas of disagreement is at a standstill since "a lot of the subjective standards are specific to the business models of the individual companies."
According to the consensus document, using the statistical methods of the three companies customers can expect an overestimate of risk for some high-risk patients using Decode's approach and an underestimate of risk using Navigenics and 23andMe's approaches. The estimate of risk using the three companies' models agree for over 99 percent of individuals, the companies note in the paper.
The main source of variation in quantifying lifetime risk between the three companies comes from the differences in the epidemiological literature, noted Decode's Chief Scientific Officer Jeffrey Gulcher. "Since there is no agreement in the literature regarding lifetime risk assessments for many disease states, instead of mandating that everyone use the same estimates, we agreed to be more transparent in how we quantify and calculate lifetime risk," Gulcher told Pharmacogenomics Reporter.
Khoury acknowledged that the PMC's efforts in getting the companies to voluntarily agree to baseline standards are "miraculous." However, the companies will need to be "nudged" by government regulation before they agree on clinical validation issues, he noted.
"I don't want to put a damper on what the companies are doing, but left to their own devices, probably they will not achieve that consensus," Khoury said. "At the end of the day, unless there is an enforcement of regulation … they're probably not going to do it."
[ pagebreak ]
The US Food and Drug Administration has not indicated any intent to regulate the burgeoning consumer genomics industry. The Centers of Medicaid and Medicare Services, which regulates most laboratory-developed tests, has said it is working with these firms to ensure that they are in line with CLIA standards.
"I think when you have competing interests, at the end there may be some agreement or disagreement based on the business model imperative," Khoury said. "But the only thing we at the CDC and NIH care about is that people don't get hurt and that they are given the right information.
"So, what [they] call 'subjective,' I think there is more objectivity to it," he said.
According to Khoury, there are scientific standards for classifying whether a SNP is validated, and consumer genomics firms should align their validation processes.
"For example … if there is no meta-analysis of a genetic association done and the SNPs don't meet … the Venice criteria [for rating the credibility of gene associations] then I wouldn't use the SNPs," Khoury explained. "To me that's not subjective, that's objective [criteria]."
Decode's Gulcher noted that when the main SNPs for a disease are not available on the chip a company is using, then surrogate markers have to be used. "The companies generally do a good job using the appropriate surrogate markers," Gulcher said.
Following the December meeting, the workshop participants wrote a paper issuing five recommendations for the industry. The recommendations include: developing and implementing industry-wide scientific standards; developing and implementing a multidisciplinary research agenda; synthesizing the information disseminated to providers and consumers; linking scientific research on validity and utility to evidence-based recommendations for the use of personal genomics tests; and considering the value of "personal utility" and develop metrics for evaluation.
The paper is currently undergoing peer review at Genetics & Medicine, and, if accepted, will likely be published this summer. NIH and CDC also plan to hold another meeting on consumer genomics in the fall to discuss the standardization of clinically validated SNPs.
"We at the government [will] hold another meeting, probably next November or December, to have a more in-depth discussion on which markers to use in genomic profiles, and more robust standards for analysis and reclassification of risk," Khoury said.
He added that the CDC's Evaluation of Genomic Applications in Practice and Prevention Working Group might review the clinical utility and validity of the clinical methods used by consumer genomics firms.
"Whatever the industry has right now doesn't cover the whole front. … Now, [consumer genomics companies] can do whatever they want," Khoury said. "But what we [at the NIH and CDC] are doing is working together with [these companies … and working in consensus with government, you'll see how flexible they can become."
23andMe and Navigenics did not answer questions for this article prior to deadline.