NEW YORK (GenomeWeb) – Two recent actions from the US Food and Drug Administration on pharmacogenetics have the genetics community scrambling to figure out what the agency's communications mean for test manufacturers, labs, and doctors.
The FDA last week granted consumer genomics firm 23andMe authorization to sell tests for 33 variants in eight pharmacogenes (CYP2C19, CYP2C9, CYP3A5, UGT1A1, DPYD, TPMT, SLCO1B1, and CYP2D6) directly to consumers, with the caveat that results not be used to change or stop treatments without talking to a doctor and confirming it using a clinical test.
A few hours later, the agency issued a safety alert cautioning doctors and patients that most genetic tests claiming to predict a patient's response to specific drugs haven't been reviewed by the FDA. Without FDA's clearance or approval, these tests may lack clinical evidence supporting their use, and the agency told doctors they could end up harming their patients by using such tests to make treatment decisions.
The messages in these back-to-back actions puzzled industry observers, and left them wondering why FDA would allow 23andMe to sell its tests to consumers online, only to render them medically useless. Experts also wondered why the agency stipulated 23andMe's test reports need clinical confirmation but did not specify which tests are appropriate to use for confirmation. The FDA did not respond to questions for this article ahead of press time.
"They've really put us in a tough spot," said one industry insider who didn't have permission from his company to speak on the record and asked to remain anonymous. "It's just not clear what's expected of us. It's not clear what FDA is trying to accomplish. What is clear is that they're doing what's good for 23andMe, but I'm not sure that's good for patients."
Bottom line: FDA does not provide a rationale for why a clinician should repeat testing.
Deciphering FDA
23andMe has become one of the biggest players in the direct-to-consumer (DTC) genetic testing space since its launch a decade earlier, boasting more than 5 million customers. It is the only company to have garnered FDA DTC market authorization for its tests for carrier screening, genetic health risks — including three BRCA1/2 variants — and now 33 pharmacogenetic variants involved in the metabolism of more than 50 medications.
However, in recent years, 23andMe has faced increasing competition from a number of new entrants who have neither taken their tests through the FDA nor have had the agency take regulatory action against them, as it did against 23andMe in 2013. Some of these competing firms are operating CLIA-certified labs, allowing consumers' own doctors and doctors from third-party physician networks to order the tests. Other firms don't operate labs at all but reanalyze raw genetic data generated by companies like 23andMe to provide new genetic insights.
23andMe thinks FDA inaction against these firms has created an unfair playing field. After FDA published the safety alert last week, 23andMe CEO Anne Wojcicki tweeted: "We are excited about our new authorization but also concerned that the FDA is not requiring all direct-to-consumer genetic testing services to meet the high bar for analytical validity, accuracy, or user comprehension, which 23andMe does."
In the safety alert, the agency maintained that it is "looking into" companies selling unapproved tests and "will take compliance actions when appropriate." While the FDA suggested throughout the alert that its approval or clearance of tests provides assurance of clinically valid claims, it stopped short of directly recommending that companies seek its review. Instead, the agency told genetic test manufacturers to assure that their test claims and reports are in line with information in FDA-approved drug labels about genotype/drug associations.
Industry players struggled to decipher these statements. Some acknowledged that the FDA has reason to be concerned that some companies are putting the public health at risk by making inappropriate claims. But they wondered why the agency wouldn't just send cease-and-desist letters to those companies, instead of diffusely suggesting that it might take regulatory action against some firms. Industry players also wondered what would trigger FDA actions, for example marketing PGx tests directly to consumers without FDA approval or clearance, or making claims that are outside of FDA-approved drug labels.
Also, the genetics community is unclear about how the FDA defines DTC testing. When physicians in a third-party network order tests, instead of the patient's own doctor, does that fall into FDA's definition of DTC testing? To what extent can the FDA regulate this, since state laws govern who has authority to order a test?
One regulatory expert further pointed out that while the FDA stipulated 23andMe's test results need confirmation, the agency didn't explain why this is necessary. "There is some incoherence between FDA's clearance of 23andme's PGx test and what the agency is telling physicians," said the expert, who asked to remain anonymous to protect clients' interests.
The FDA says it has reviewed analytical validity data showing that 23andMe's tests can accurately detect the variants they claim to. But it appears to take the position in the safety alert that only an FDA-cleared or -approved test can accurately detect these variants. "Well, if that's true, then why would a physician need to repeat 23andMe's test?" the expert posited. "Repeating a test makes sense if you're not sure about the analytical validity of the result, but if FDA did not believe 23andMe's analytical validation was sufficient, why did it clear the test? Bottom line: FDA does not provide a rationale for why a clinician should repeat testing."
Without clarity from the FDA, doctors and genetic counselors are left to figure out what to tell patients about why they need to be tested again at additional expense. Part of 23andMe's marketing message to customers is that its tests are the "first and only" to receive the FDA's nod for DTC sales. So when patients learn their PGx variants through the company and bring this information to their physician, it might be difficult to explain to patients why they need confirmation with a test that likely hasn't been FDA approved, and which could result in out-of-pocket costs.
Genetic testing firms have had a difficult time convincing insurers to recognize the value of pharmacogenetic testing. Today, insurers reimburse for very few PGx indications, even when testing is supported by guidelines and drug labeling, as in the case of CYP2C19 testing and the antiplatelet drug clopidogrel. "The out-of-pocket cost of testing starts at a few hundred dollars and many individuals will not pay for confirmatory testing, leaving the genetic counselor and treating healthcare professional in a lurch on what to do," said Scott Weissman, a genetic counselor and founder of Chicago Genetic Consultants.
Importantly, the FDA hasn't explained what it considers acceptable for confirmatory testing. The agency doesn't require labs to use an FDA-cleared or -approved test for confirmatory testing, likely because it has not cleared pharmacogenetic assays for all of the variants 23andMe is authorized to test. And while the agency has issued guidance stating its expectation to be able to review tests claiming to predict response to specific drugs, it could not make this point too explicitly in the safety alert, given its history with regulating lab-developed tests (LDTs).
It's difficult to keep up with what pharmacogenetic tests are appropriate to use and order. That's a big reason we formed CPIC.
Follow the drug label?
Industry players are interpreting FDA's communications as yet another attempt to regulate LDTs. Not only PGx tests, but most of the 74,000 commercially available genetic tests are not reviewed by the FDA, but are overseen by the Centers for Medicare & Medicaid Services under federal lab regulations, called the Clinical Laboratory Improvement Amendments.
The FDA has a long history of trying to regulate LDTs, but each time this has been thwarted by the lab industry and pathologist groups, who maintain the agency lacks the statutory authority to do so. The lab industry and pathologists believe FDA's regulations are ill-suited for laboratory processes and the agency's oversight would hinder patient access to the latest tests. Moreover, clinicians don't necessarily look for FDA's nod when deciding to use a test.
For example, St. Jude Children's Research Hospital uses reference labs for its pharmacogenetic testing, some of which may have FDA clearance or approval, while others run LDTs that don't. "Whether or not a test has FDA approval is not a critical factor in whether to deploy a test or not," said Mary Relling, head of the department of pharmaceutical sciences at St. Jude. "There are many areas of testing in clinical medicine that thrive and that don't get FDA approval. There are many, many laboratory-developed tests, especially in the cancer world, that we couldn't survive without."
Relling is also co-chair of the Clinical Pharmacogenetics Implementation Consortium (CPIC), a group of experts in drug/gene interactions from academia, healthcare systems, and industry that since 2009 has been creating evidence-based clinical practice guidelines. The group has issued 20 PGx guidelines that address 35 drugs to date and recently received a $5 million grant from the National Institutes of Health to continue its work.
"It's difficult to keep up with what pharmacogenetic tests are appropriate to use and order," Relling acknowledged. "That's a big reason we formed CPIC." CPIC's current list of members includes at least one FDA employee as "an observer," and Relling said the agency is "very aware of our work."
Still, industry players objected to the recommendation in FDA's safety alert for labs to make sure their PGx test claims align with information in drug labeling. By placing the emphasis on drug labeling, they felt the FDA is brushing aside the longstanding work of expert groups like CPIC to advance guidelines and standardize how variants are interpreted across labs.
"One of the things that we need to follow up on with the FDA is their statement that they only recognize information that is in drug labels," said Don Rule, founder of Translational Software, a company that has developed a decision support platform to help physicians make sense of patients' PGx test results and make decisions in line with the latest evidence and guidelines. "Drug labels are not generally updated after a drug reaches the market, so restricting an information provider to FDA-sourced content will set the industry back 20 years in some cases," he said.
According to Alberto Gutierrez, former director of FDA's Office of In Vitro Diagnostics and Radiological Health, the agency makes every effort to keep drug labels up to date as PGx data become available. "From the safety alert, clearly the FDA is concerned with tests that make claims that are not supported by well vetted clinical data," Gutierrez said. "It is true that sometimes drug labels do not change as quickly as the community would like, but in general the barrier to change has more to do with the quality and quantity of the data supporting the new claims."
Meanwhile, guidelines from groups like CPIC, which consistently update recommendations based on the latest evidence, don't always align with FDA-approved drug labels. But based on FDA's safety alert, test developers must figure out how to navigate these differences in their test claims. For example, CPIC has published dosing recommendations based on patients' CYP3A5 genotypes for tacrolimus, a drug to help reduce the risk of transplant rejection. However, the label for tacrolimus doesn't mention the drug's association with CYP3A5 genotypes.
Also, for CYP2C19 extensive metabolizers, CPIC recommends doctors consider giving patients a treatment other than the antidepressant citalopram because ultrarapid metabolizers are at higher risk of failing to respond to therapy from low exposure to this drug. However, the drug label only mentions the maximum dose that should be given to CYP2C19 poor metabolizers, but doesn't mention the CYP2C19 ultrarapid metabolizer group.
A downer on antidepressant PGx
Experts in the field of pharmacogenetics took particular issue with the FDA's discussion in its safety alert of genetic tests that claim to predict how well a person might respond to antidepressants. The agency said it is aware that companies are claiming their tests can help physicians identify which antidepressants will "have increased effectiveness or side effects" for their patients, but added that "the relationship between DNA variations and the effectiveness of antidepressant medication has never been established."
"That's not correct," said Relling. "I think there is some data indicating that DNA variations in some genes impact the effectiveness and certainly impact the side effects of some antidepressant medications … In this statement, the FDA has been overly negative about the potential of some pharmacogenetic testing to actually help patients who are getting antidepressants."
Certainly there are psychiatrists who feel there is insufficient evidence supporting the use of PGx testing for conditions such as depression, and a task force of the American Psychiatric Association said as much earlier this year. But many doctors find PGx tests do help patients with major depressive disorder get to the right drug, especially those who have failed numerous treatments based on trial-and-error prescribing.
Larry Lesko, who directed the Office of Clinical Pharmacology within FDA's drug division for 16 years until his retirement in 2011, felt the agency may be "underestimating the ability of healthcare providers to understand and use genetic tests." Lesko pointed to a recent review of available PGx tests for guiding depression treatments and noted that while the clinical utility data on these tests is not perfect, there is enough evidence showing that patients who receive PGx-informed depression drugs have higher response rates and are at lower risk of not tolerating the drug compared to those receiving treatment as usual.
There is little evidence about the performance ... of LDTs in pharmacotherapy as offered by a large number of companies.
Impact on pharmacogenetics
Ten years ago, Lesko and other leaders at the FDA championed the idea that pharmacogenetics was critical to advancing precision medicine and started updating drug labels with this information. The agency has published a table of drugs with PGx information in labeling and states at the top of the webpage that "pharmacogenomics can play an important role in identifying responders and non-responders to medications, avoiding adverse events, and optimizing drug dose."
Some industry observers pointed to the strong language in FDA's recent safety alert as conveying the opposite sentiment, or perhaps a signal that the agency wants to shut down all PGx testing going on in the field that it hadn't approved or cleared. Since the majority of PGx tests haven't been reviewed by the FDA and already face a difficult reimbursement climate, this could place additional pressure on an already beleaguered segment of the industry.
Lesko, who is now emeritus professor at the University of Florida's College of Pharmacy, said he believes the agency still wants to advance PGx testing to improve pharmacotherapy and empower patients. While he acknowledged that FDA's authorization for 23andMe's test is confusing due to the caveats restricting its use in clinical care, he felt the warnings about unapproved LDTs, and the unproven claims some labs are making, are warranted. "There is little evidence about the performance, especially comparative performance, of LDTs in pharmacotherapy as offered by a large number of companies," he said.
However, Lesko also recognized that FDA's safety alert seems to suggest that "if the PGx test leads to a decision about a dose adjustment for a medicine that is not approved and covered in the label, this could be potentially problematic for the prescriber and the manufacturer."
Despite their confusion over the agency's communications, some genetic testing companies reacted quickly to incorporate the FDA's concerns into their services. For example, Color Genomics, which recently launched a PGx test, now includes a footnote on its website with FDA's message to consumers about not changing medications based on a genetic test report without talking to their own doctor. "We emphasize that the results of our PGx test should always be considered by a healthcare provider in the context of other, non-genetic factors that may impact a patient's medications," said a Color spokesperson, adding that the company has taken a conservative approach in its PGx service.
The company is currently reporting variants in CYP2C19 and CYP2D6 involved in metabolizing a handful of drugs, and will roll out additional gene/drug reports in the next six months. Color states on its website that it has decided which genes to report on after reviewing CPIC guidelines and FDA labeling, and is only reporting information related to medications that have FDA-approved labels describing how these genes influence their use.
Color's tests can be ordered by the patient's own doctor or a physician from a third-party network, but unlike 23andMe's service, comes with complimentary access to pharmacists and board-certified genetic counselors to help consumers and their doctors better understand the results. The company's willingness to provide these supporting services has helped it gain a good reputation among healthcare providers and genetic counselors.
Others are going a similar route. "We're doing this responsibly by providing the companion services in cases where it makes sense for the individuals and physicians to have them," said Mirza Cifric, CEO of Veritas Genetics, a company that offers a whole-genome sequencing test and provides consumers information about their potential health risks and ability to metabolize more than 170 drugs.
Veritas last week expanded its offerings to include a premium service for those who want more in-depth analysis of certain disease risks and a diagnostic service for those with a personal or family history of a genetic condition. These tests are intended to be actionable and backed by the latest evidence and professional guidelines, according to Cifric, who added that Veritas has already had conversations with the FDA about taking its services through the agency.
Similar to Color, Veritas also offers free genetic counseling to customers with actionable results and PharmD support to physicians to help interpret PGx reports. "Anything short of that is really not providing a complete service to the individuals and doctors," Cifric said.
While FDA's authorizations for 23andMe's BRCA and PGx tests come with special controls or requirements to inform customers about what the tests can and cannot do, 23andMe is not required to offer genetic counseling. When FDA authorized 23andMe's BRCA test for three variants, doctors and genetic counselors felt that since there are thousands of cancer-linked variants in BRCA1 and BRCA2, the company should provide genetic counseling to ensure customers understand the test's limitations.
"People aren't educated enough to understand how incomplete 23andMe's test is," Cifric said, adding that he believes companies like Veritas are the next generation of consumer genomics firms that have learned from the mistakes of first-generation firms like 23andMe.
Recalling the kerfuffle over 23andMe's BRCA test, he predicted there may be a similar backlash against the PGx test from the medical community. "That hurts us all unfortunately," Cifric said. "It creates a negative perception of the whole genetics industry when there are better choices, like our service."