Jeanette McCarthy, MPH, PhD
This month, the US Food and Drug Administration (FDA) sent genetic testing company Inova Genomics a warning letter because they were selling a suite of pharmacogenomic (PGx) testing products, marketed under the name MediMap, without obtaining prior FDA authorization to do so.
The letter was reminiscent of the 2010s, when the FDA began sending similar warning letters to, what was at the time, a new breed of companies who were marketing health-related genetic tests directly to consumers (DTC). Among the companies receiving these letters, only 23andMe went on to comply with the FDA requirements and ultimately gained FDA authorization for several health-related tests, including PGx tests.
Generally speaking, the FDA has not stepped in to regulate genetic tests unless they pose a significant potential public health threat in that they are 1. health-related and 2. marketed directly to consumers.
The recent authorization of 23andMe’s genetic tests by the FDA, along with this new warning letter to Inova Genomics afford a glimpse into their regulatory strategy for PGx and have left many in the molecular diagnostics community unsettled.
How are genetic tests currently regulated?
We come to depend on oversight by regulatory agencies to ensure that health-related products, including genetic tests, are safe. Paramount to that safety is the idea that the laboratory assay is sufficiently accurate, i.e. has good sensitivity and specificity for detecting the sequence variant it is designed to detect. This is referred to as analytical validity. An equally important aspect of safety is the concept of clinical validity, the idea that a test can accurately predict the clinical outcome it is said to predict.
It may come as a surprise to most health care providers that the genetic diagnostic tests offered by most commercial (and non-commercial) laboratories are only loosely regulated.
Two federal agencies oversee genetic diagnostic testing: The Centers for Medicare and Medicaid Services (CMS) and the FDA. CMS regulates the laboratories that run genetic diagnostic tests. These labs need to comply with Clinical Laboratory Improvement Act (CLIA) regulation, which ensures that the lab personnel, procedures and equipment are compliant with a high level of standards.
The FDA regulates medical devices, such as test kits, but most genetic tests on the market today are not sold as kits. Instead, they are sold as services by individual laboratories and are referred to as laboratory developed tests (LDTs). While the FDA has regulatory authority over LDTs, it has historically opted to not regulate them, exercising its enforcement discretion.
What did we learn from FDA authorization of 23andMe’s genetic tests
The FDA authorization of 23andMe’s DTC health-related tests covered three main aspects of the test: user comprehension; analytical validity; and clinical validity. User comprehension is an important part of DTC testing because it’s presumed that consumers, who are ordering and receiving results, will have a low genomic literacy. Analytical validity is the most basic requirement of a genetic test, so it’s not surprising that 23andMe boasts >99% analytical accuracy. Clinical validity is a somewhat subjective concept comprised of two elements. The first is a body of literature supporting an association between the genetic variant and clinical outcome (disease/drug response). The second is a high positive or negative predictive value of the test, indicating that the results can predict accurately who will and won’t have the disease/drug response. FDA clearance of 23andMe tests appears to have focused on the first element, as several of their cleared tests are backed by a large body of evidence and yet still have poor predictive value.
What the FDA didn’t account for in its approval of the 23andMe tests was that most of the tests were incomplete. A recent presentation at the American College of Medical Genetics conference revealed that the 23andMe Breast and Ovarian Cancer Risk Test misses nearly 90% of BRCA mutation carriers. In our own analysis, we identified important PGx variants missing from the 23andMe PGx tests. The FDA addresses these shortcomings by saying that the tests should not be used as ‘diagnostic’ tests or not used to make medical decisions.
This begs the question why the FDA went to the trouble to clear them if they weren’t to be used to make health decisions. Instead, consumers who see that the test is FDA cleared may decide that the tests are safe, when in fact they have major shortcomings which may make them more of a public health threat than if they had not been cleared.
What does the Inova Genomics warning letter from the FDA say
The FDA warning letter to Inova Genomics cited two main reasons why they thought the Inova tests posed significant public health concerns. First, and probably the main reason the tests caught the FDA’s attention, is that they were sold DTC, albeit with a company-provided physician intermediary who could order the test. The concern is that the results were being delivered directly to patients and that uneducated patients may misinterpret them. This assumes that having the customer’s personal physician involved in the testing process would mitigate the potential risk, but if that physician is not educated about PGx testing, it probably makes little difference.
The second public health concern cited by the FDA is a lack of established clinical validity for the PGx tests. As stated in the FDA warning letter, “For example, the relationship between CYP2C19 genotype and drug response to escitalopram and sertraline is not established and this relationship is not described in the FDA-approved labeling for these drugs”. I’ve previously written about the inclusion of genes on PGx panels that were not clinically valid, but the two examples cited in the warning letter happen to be among the most clinically valid.
How do we determine clinical validity and actionability of pharmacogenomic tests?
Most areas of medicine rely on evidence-based evaluation of a body of literature and consensus expert opinion to judge clinical validity of a test. Professional society guidelines often serve this role. In the field of PGx in the US, experts in the field rely heavily on the PharmGKB and Clinical Pharmacogenomics Implementation Consortium (CPIC) as unbiased, authoritative sources of evidence supporting the clinical validity and actionability of PGx tests. Gene-drug pairs with the highest level of evidence become the subject of guidelines from CPIC.
Both CYP2C19 associations with response to sertraline and escitalopram, cited in the FDA warning letter, are covered by CPIC guidelines and supported by the highest level of evidence of clinical validity and actionability.
The FDA drug label for pharmacogenomics
The FDA drug label is not an authoritative source of information on the clinical validity of a PGx test. There are many shortcomings of the FDA drug label with regard to PGx information. Foremost is the apparent lack of standards for selecting which pharmacogenes are included in the drug label, the strength of recommendation of testing for each, and the evidence of clinical validity. A study that reviewed FDA drug labels with PGx information found that only 36% provided convincing clinical validity evidence.
If you review the FDA drug labels for the 35 gene-drug pairs with the highest clinical validity based on both PharmGKB and CPIC (and also the subject of CPIC guidelines), you’ll find that eight of the most clinically valid pharmacogenes were not mentioned in their respective drug labels, including CYP2C19 in the labels for sertraline and escitalopram.
If you look at the opposite end of the spectrum, those drugs with PGx info in the FDA label but with little support of clinical validity from PharmGKB and CPIC, you’ll find that there are 33 gene-drug pairs that fit the criteria of being in the FDA drug label and having a CPIC grade of C or D, where expert consensus indicates no prescribing action recommended.
The most extreme example is the drug label for the Hungtington’s disease drug tetrabenazine and the gene CYP2D6.
According to PharmGKB there is no literature support found for the CYP2D6 and tetrabenazine response. According to CPIC, this gene-drug pair is a level C [1]. Therefore, by the standards set forth by CPIC and PharmGKB this PGx information is not clinically valid or actionable, and yet it is in the FDA label.
The point is simple: a mention in the FDA drug label alone does not imply a clinically valid PGx test and conversely, absence of PGx information in the drug label does not mean that a clinically valid PGx test does not exist.
How do we make pharmacogenomic testing safer?
The safety of PGx testing hinges on the quality of the tests and knowledgeable users of those tests and not simply approval by the FDA.
High quality PGx tests are backed by rigorous scientific evidence. In singling out some PGx associations as ‘not established’, despite evidence to the contrary from CPIC and PharmGKB, the FDA seems to be ignoring the work of these respected professional organizations which the field as a whole relies upon for the truth.
The very tests that the FDA warning letter called out as lacking clinical validity are on the PGx panel tests offered by many reputable testing labs. This makes me wonder whether those labs will be subject to FDA scrutiny next, and if not, why would the FDA knowingly allow those tests to be sold if they think they’re not clinically valid.
The notion that a PGx test is safer, or doesn’t require FDA approval simply because a health care provider is involved in the ordering and delivery of results overlooks the fact that the average health care provider knows little more than his patient about PGx testing.
This is not a call for the FDA to begin regulating all PGx tests. On the contrary. These early steps by the FDA to regulate PGx testing have left many leaders in the field questioning their aims and methods. Instead, this is a call to PGx test providers to be even more transparent about the clinical validity of their tests. It’s also a call to educate health care providers and consumers alike about how to recognize the value and limitations of different PGx tests.
[1] CPIC level C: there are published studies at varying levels of evidence, some with mechanistic rationale, but no prescribing actions are recommended because (a) dosing based on genetics makes no convincing difference or (b) alternatives are unclear, possibly less effective, more toxic, or otherwise impractical or (c) few published studies or mostly weak evidence and clinical actions are unclear.
For additional pharmacogenomics information and resources, visit www.precisionmedicineadvisors.com
To learn how to evaluate pharmacogenomic tests, identify patients who might be candidates for testing and implement testing into your clinical practice, enroll in our short online course, Pharmacogenomics in Practice.
Dr. McCarthy is the founder of Precision Medicine Advisors, which specializes in communicating precision medicine to lay professional audiences, providing scientifically sound, unbiased information to promote the responsible use of genomics in medicine.
Contact info: jeanette@pmedadvisors.com
