Monday, February 23, 2015

Why FDA's Go-Ahead for 23andMe is Significant

On February 19, the FDA (here) and 23andMe (here and here) announced that the FDA has cleared a Bloom syndrome carrier screening genetic test developed by 23andMe - and has cleared it for direct-to-consumer marketing, fitting the 23andMe business model.

The announcement was quickly picked up by Forbes (here), Genetic Engineering News (here), and by subscription sources like Genomeweb and Gray Sheet (here, here).

Although the FDA's press release does not link to the 10-page clearance letter, another source does, here.



What Happened?

The FDA has three classifications for devices: Class I, Class II, and Class III.  Class I devices are defined by category in federal regulations and do not require FDA clearance or approval (example: tongue depressers, exercise equipment.)  Class III devices are "high risk" and generally require a PMA and clinical trials.  Combination diagnostics (like a KRAS kit assay) have generally been in this category.

In between fall the Class II devices.  These are based on the 1976 FDA device law which has a special category for already-marketed pre-1976 devices and later devises with the same intended use.  This is the 510(k) pathway, based on a seemingly-simple law which has developed an enormously complex body of policy, practice, and guidance.

In recent years, the FDA has been able to create "de novo" 510(k) clearances, essentially deeming a device can be treated as if it had a predicate device, if a product-specific body of rules are followed, captured in guidance documents.

So:  The FDA clearance of the 23andMe Bloom test is a Class II, "de novo 510(k)" clearance.  The FDA is in the process of creating a new regulatory classification to hold it (which will be placed at 21 CFR 866.5940).  (For more on test categories, see here).  The FDA will probably publish a guidance document for this new category.  The letter to 23andMe (which the FDA knows will get a lot of public scrutiny) is so long because it contains information that would normally appear in a category guidance document.  The 23andMe letter is a sort of Trojan Horse through which the FDA can release quite a bit of information about what is really a new policy approach.

Why is the regulatory classification interesting?

FDA's Relationship with "LDTs that are DTC." First, in the current FDA LDT guidance, the FDA has a footnote separating out DTC genetic tests from all other lab developed tests, and banning the marketing of DTC genetic tests unless they have been FDA reviewed.   (Guidance, page 5, footnote 4, here.)

FDA Innovates!  Second, the FDA views its efforts to provide DTC pregnancy, cholesterol, and HIV tests as a hallmark of its ability to carry out innovative regulatory approaches (see FDA comments in its own press release on the 23andMe Bloom clearance).   So: At the same time critics are saying that the FDA's regulation of LDTs may cripple innovation, the FDA leadership can point to this genetics DTC clearance as an example of the agency's high capacity for rapid innovation.

Broad versus Narrow FDA Test Categories.   Last year, the FDA approved a physician-order, cystic fibrosis sequencing test from Illumina as a 510(k) device, K132750 (here).  That CF sequencing tests falls into its own very product specific 510(k) category at 21 CFR 866.5900, which is for "Cystic fibrosis transmembrane conductance regulator (CFTR) gene mutation detection system." (Here; see also table at bottom of this webpage).

In contrast, the 23andMe Bloom genetic test review also created a new, open-ended category of tests called carrier screening tests - in FDA speak, an "Autosomal recessive carrier screening gene mutation diagnostic system."   (As of now, all of this information is found only within the 10-page FDA clearance letter to 23andMe, here.)

The use of "open-ended" genetic test regulatory policies for broad categories of tests (like, all tests that are "carrier genetic screening") could be very important.  The FDA is talking about regulating LDTs, and possibly, farming out LDT review to third parties such as CAP or New York State.  Such third parties couldn't create new Federal regulatory 510(k) categories for each new gene implemented as an LDT.  But CAP or New York State might be able to apply clear-enough 510(k) rules under newer, broader test categories which could have names like "carrier screening test."

No Physician Prescription Required.  Many IVD diagnostic tests carry labeling, "For prescription use only."  There's no such restriction written into the 23andMe clearance.

NEJM and DTC Genetics; AMP and DTC Genetics.  In May 2014, the New England Journal ran an article quite critical of 23andMe (here), including its direct-to-consumer aspect.  In creating an FDA-endorsed DTC genetic test, the FDA has made an end-run around the NEJM authors. Interesting: the Association for Molecular Pathologist in February 2015 issued a new position statement that is more favorable to certain DTC genetic tests than the AMP's prior policy (here).

Path for 23andMe was Not Easy

Here is where the actual 10-page clearance letter is especially important.  The letter describes what the FDA did from scratch - in much more detail than most clearance letters.  In part, this is probably because the corresponding "De Novo Guidance" for this test category has not appeared.  Some requirements include:
  • Labeling, Consumer Packaging, 23andMe Home Page must all include clear links to informational materials for this test.  The FDA recognizes the test may be DTC literally "over the counter," but this may include website orders, etc.  
  • Extensive accuracy development information must be undertaken. These must be undertaken on actual clinical specimens.  FDA asks for 20 tests on patients with more common clinical variants and 3 tests on patients with uncommon variants.  
    • This is a significant barrier to entry.  23andMe might assert its quality and precision exceed what is required under CLIA regulations. See also review details at Genetic Engineering News (here).
  • Precision information must be performed "on multiple instruments, multiple operators, multiple non consecutive days, with multiple reagent lots."  Try that, St. Mary's Hospital.
  • Usability studies were conducted on hundreds of individuals for usability of the saliva collection device and also the usability of the consumer-friendly test information.
    • FDA:  Test reports given to users must: a) define the condition being tested and related symptoms; b) explain the intended use and limitations of the test; c) explain the relevant ethnicities regarding the variant tested; d) explain carrier status and relevance to the user’s ethnicity; e) provide links to additional information pertaining to situations where the user is concerned about their test results or would like follow-up information as indicated in test labeling). The study shall assess participants’ ability to understand the following comprehension concepts: the test’s limitations, purpose, and results. Study participants must be untrained, naïve to the test subject of the study and be provided only the materials that will be available to them when the test is marketed. 
  • Links.  Prominent links must connect patients with certified genetic counselors.  
  • Warnings.  Extensive explicit limitations and warnings must be provided.  "The results of this test are intended to be interpreted by a board-certified clinical molecular geneticist or equivalent...your ethnicity may affect your your results are interpreted...the test does not diagnose any health condition...this test does not detect all genetic variants..."
  • Expansion Path.  Tests may be expanded using "clinical validity...supported by peer-reviewed journal articles, professional society recommendations, or both, including ACMG or ACOG guideline-recommend testing..."
Implications?
  • The FDA "shut down" 23andMe in November/December 2013, and the submission of this Bloom syndrome genetic test for clearance was not secret, it was announced by 23andMe in June 2014 in its company blog and picked up by media like Forbes (here) and Genomeweb at that time.   
  • The FDA leadership will leverage this clearance as a landmark demonstration of its ability to innovate.
  • The use of a "class" category (the class of "autosomal carrier screening tests") is important as a similar test category approach may be used in future FDA efforts to classify and control LDTs.
    • Interestingly, some other FDA IVD regulatory categories (e.g. 864.1860, immunohistochemistry) already comprise Class I, II, or III devices under one category listing.  Under IHC, the "risk class" for a new IVD IHC test depends on the indication for the IHC test within the device category type.   When the risk class depends on the "indication" this could be subject to misuse, such as if an uncommon but low risk indiciation is proposed for review, but the test would mostly be used off label in higher risk scenarios clinically.
  • Some tests might have dual indications leading to different risk classes.  For example, what if a lab claimed it was offering DTC gene testing "for carrier screening" but some patients would employ it for a medical use of the same gene sequence, like management of gene-positive status with drugs or surgery?   
    • Would some tests be disallowed for carrier screening simply because the kit, once released, "might" or "could" be misused for medical applications for certain genes within?   
    • How would the FDA apply this additional review of hypothetical future off label uses?  Is a warning "don't do that!" enough?  
    • The regulatory category (866.5940) will encompass both DTC and physician-prescription status for carrier screening genetic tests.  Will the FDA handle this case by case?   
  • 23andMe says it will not market the Bloom test until a more comprehensive panel of carrier tests can be offered together.   
  • The guidance does not seem to impact physician-ordered and CLIA-based LDT genetic carrier testing.  But it certainly stirs the waters.   
    • Imagine that the same Bloom genetic test is offered in an FDA sanctioned way as a DTC test, and by a CLIA lab LDT as a physician-ordered test.    Would the FDA view the CLIA lab LDT physician ordered test with as much discretion as it's had in the past?
  • While the FDA requires the test explanation to be consumer-friendly and tested on hundreds of consumers, details of the performance characteristics and the genomic limitations could not possibly be directly understood by most consumers.
  • The FDA seems to state in its 23andMe letter that future carrier tests would be class II yet not require additional 510(k) reviews.   How this would work is puzzling (since THIS test had voluminous safety and accuracy review) and the regulatory experts I've talked to find it mysterious at best. 
Below, a screen shot comparing the cystic fibrosis Class II category (866.5900) with the new carrier screening genetic test category (866.5940).