The FDA Drops an Anvil on 23andMe – Now What? (slightly revised Wed. night)

On Friday, November 22, the FDA sent a nastygram to 23andMe, the only remaining substantial pillar of the “direct to consumer” (DTC) genomic testing industry. The letter, available here, is well worth reading in its entirety, but, in brief, it told 23andMe, in no uncertain terms, to stop marketing its “Saliva Collection Kit and Personal Genome Service” (“the PGS”) on the ground that it was an unapproved and uncleared “device” under the Federal Food, Drug, and Cosmetics Act.  It noted that the PGS, because of its lack of approval or clearance, was both adulterated and misbranded under the Act; the distribution of adulterated or misbranded devices is a federal crime.

23andMe’s initial response, click here, through an unnamed press spokesperson on its blog, was, at least on its face, conciliatory.

We have received the warning letter from the Food and Drug Administration. We recognize that we have not met the FDA’s expectations regarding timeline and communication regarding our submission. Our relationship with the FDA is extremely important to us and we are committed to fully engaging with them to address their concerns.

What’s going on – and what does it mean in the long run?

There has long been controversy over whether and how the FDA should, or can, regulate genetic tests, whether delivered, as 23andMe does, direct-to-consumer or through a medical professional.  These issues simmer below the surface of the cease and desist letter, but it is not clear that, at anything other than the deepest level, they are immediately relevant.

It appears that, for whatever reason, 23andMe chose to ignore the FDA – and the FDA chose not to take it anymore. The FDA letter says

Since July of 2009, we have been diligently working to help you comply with regulatory requirements regarding safety and effectiveness and obtain marketing authorization for your PGS device. . . . As part of our interactions with you, including more than 14 face-to-face and teleconference meetings, hundreds of email exchanges, and dozens of written communications, we provided you with specific feedback on study protocols and clinical and analytical validation requirements, discussed potential classifications and regulatory pathways (including reasonable submission timelines), provided statistical advice, and discussed potential risk mitigation strategies. . . .

However, even after these many interactions with 23andMe, we still do not have any assurance that the firm has analytically or clinically validated the PGS for its intended uses, which have expanded from the uses that the firm identified in its submissions. In your letter dated January 9, 2013, you stated that the firm is “completing the additional analytical and clinical validations for the tests that have been submitted” and is “planning extensive labeling studies that will take several months to complete.” Thus, months after you submitted your 510(k)s and more than 5 years after you began marketing, you still had not completed some of the studies and had not even started other studies necessary to support a marketing submission for the PGS. It is now eleven months later, and you have yet to provide FDA with any new information about these tests. You have not worked with us toward de novo classification, did not provide the additional information we requested necessary to complete review of your 510(k)s, and FDA has not received any communication from 23andMe since May. Instead, we have become aware that you have initiated new marketing campaigns, including television commercials that, together with an increasing list of indications, show that you plan to expand the PGS’s uses and consumer base without obtaining marketing authorization from FDA. [emphasis added]

Note particularly the sentence I bolded.

This sounds as though 23andMe started down the FDA path but, six months ago, not only stopped communicating with the agency but started new and bigger marketing efforts. That sounds as though 23andMe did not just ignore the FDA, but, while walking briskly past it, quickly turned and spat in its face.

Now, I have no inside information on this dispute.  I have the (scathing) FDA letter and the (timid) 23andMe initial response.  Undoubtedly, something was going on other than lost messages at the 23andMe side.  I don’t know whether that something was a plan to fight the FDA’s jurisdiction, on statutory or (novel) constitutional grounds, or a hope that the FDA would back down, or something else entirely. But if you ignore the FDA for 6 months, it is not surprising that the FDA is going to be unhappy with you, wholly apart from the future of FDA regulation of genetic tests.

So this may just be a slap down of an insubordinate, or royally disorganized, regulated firm. Does it say anything more about the future of FDA regulation of genetic tests?

Well, it does say that the FDA thinks it can and should regulate, to some extent, those tests. The letter lays out its reasons: “As discussed above, FDA is concerned about the public health consequences of inaccurate results from the PGS device; the main purpose of compliance with FDA’s regulatory requirements is to ensure that the tests work.”  The letter gives examples, from BRCA 1 and 2 to pharmacogenomic tests where the accuracy, and the patient’s actions on, the 23andMe tests have potentially serious health implications.

Some have argued that the FDA does not have jurisdiction over these direct-to-consumer tests.  (And they include some very smart colleagues and friends of mine.)  How can a plastic tube for collecting saliva be a regulated device? Welcome to FDA law.  The federal statute defines “devices” (“medical” doesn’t appear in the law) as, among other things,

an instrument, apparatus, implement, machine, contrivance, implant, in vitro reagent, or other similar or related article, . .  which is–

(2) intended for use in the diagnosis of disease or other conditions, or in the cure, mitigation, treatment, or prevention of disease, in man or other animals

As the FDA’s letter points out, 23andMe markets its genetic tests as,

“providing “health reports on 254 diseases and conditions,” including categories such as “carrier status,” “health risks,” and “drug response,” and specifically as a “first step in prevention” that enables users to “take steps toward mitigating serious diseases” such as diabetes, coronary heart disease, and breast cancer. Most of the intended uses for PGS listed on your website, a list that has grown over time, are medical device uses under section 201(h) of the FD&C Act.”

A plastic tube for collecting spit doesn’t look much like an MRI machine or an implantable left ventricular assist device, but it is an “article” that, from its marketing information, 23andMe “intend[s] for [medical] use.”

Of course, the decision about jurisdiction (if 23andMe actually dares make it) will ultimately be one of statutory interpretation for the courts, informed by and giving appropriate deference to reasonable interpretations by the agency charged with implementing the statute – the FDA.  One never knows about courts, but I think the 23andMe jurisdictional argument is a loser. And, a reality of regulated industries, especially those regulated by the FDA, is that fighting your regulator may well, in the long run, be a very bad idea.

Ok, so FDA does intend to assert regulatory authority here, which its thinks 23andMe has been flouting (not “flaunting” – common misuse). Any other take-aways?

That’s hard to tell.  We’ve all been waiting . . . and waiting . .  . and waiting . . . for long-promised draft FDA guidance on regulating LDT’s, “laboratory developed tests.”  These include, but are not limited to, genetic tests and have been expected to include considerations of clinical validity. Rumor has it that the FDA thought they were done in mid-2012.  “After the election,” the word was.  Well, we all thought that meant the 2012 election, but that guidance is still stuck somewhere in the Administration.  FDA had promised a risk-based regulatory scheme, but we don’t know what it is.  We do know from its letter that it is worried that 23andMe’s tests have not been validated yet, neither analytically (do they accurately detect the single nucleotide polymorphisms (SNPs) they say they detect) nor  clinically (do those SNPs mean what 23andMe says they mean?  Presumably any long-term FDA strategy for dealing with LDTs, including LDT genetic tests will focus on analytical and clinical validation. (Hint – the clinical validation will generally be a lot tougher than the analytic validation. See the GAO report on terrible inconsistencies in clinical interpretation of the same SNPs by the DTC genomic testing industry, available for download here.) [Italicized portions of this paragraph revised.]

Does this letter tell us anything about the FDA’s position on the DTC part of testing? That’s not really clear. The letter mentions the fear that direct-to-consumer patients may self-manage in harmful ways, but it isn’t clear how much turns on that.  It is the case that, at least under current FDA practice, this would not be an issue if 23andMe were not a DTC company.  The FDA has not regulated so-called “laboratory developed tests” (LDTs), performed from (sort of) scratch by clinical laboratories.  23andMe has a clinical laboratory, but the saliva is not being collected by a doctor, with her own equipment, in her office, but by a tube 23andMe sends out – a “kit” in FDA parlance. FDA regulates diagnostic tests when sold as kits – to consumers, to doctors, or to clinical labs – but not when done directly by the labs based on samples collected by others and sent to them.

Finally, in the long run (which in this field is about 12 months right now), what does this say about genetic testing and the FDA?  In one sense, almost nothing. The SNP chip method that 23andMe uses was never very good at providing useful genetic information. Its advantage has been its low cost. But as sequencing gets cheaper and cheaper, SNP chips have already largely become obsolete for most genetic testing. Don’t look only for SNPs, do a whole exome (the 2 percent or so of the genome that “codes” for protein) sequence for under $1000 or a whole genome sequence for under $5000 (today – under $1000 soon).

In other, probably much more important news, on November 20, two days before its action against 23andMe, FDA just approved as a medical device a next generation sequencing platform from Illumina, two cystic fibrosis tests for that platform, and a kit that allows doctors to come up with other tests for the platform.  Here.  As Francis Collins, director of NIH (and one of the discoverers of CFTR1, the “cystic fibrosis gene,” and Margaret Hamburg, commissioner of the FDA, said in an article in last week’s New England Journal of Medicine, here, this “opens the door for the transformation of research, clinical care, and patient engagement.”

THAT is the future.  The DTC SNP chip era is ending.  23andMe’s two main competitors, Navigenics and deCODEme, already left the market.  23andMe’s SNP chips are slipping rapidly into the past.  And 23andMe knows it. Earlier this year it offered, for $999, a “pilot” of whole exome sequencing.  If the firm survives, it will surely be by offering whole exome and whole genome sequencing.  And, I hope, doing so after analytic and clinical validation that convinces the FDA that its methods are safe and effective. Because tests of unknown accuracy are no better than no tests at all.  I applaud the FDA for taking seriously its mandate to protect public health.

Hank Greely

31 Responses to The FDA Drops an Anvil on 23andMe – Now What? (slightly revised Wed. night)
  1. Where do you get the $5000 price for whole genomes today?


    1. I have heard from people in the industry, as early as Sepember 2011, that soe of the companies were selling genomes (for research purposes and thus not CLIA approved) for about $5000 and as low as $3400 for package deals (10 genomes or more). In 2012, Genome Web Daily reported that Complete Genomics had shipped 600 whole genomes in the last quarter of 2011 for an average $4200 each. I do think hospitals buying clinical grade genomes from CLIA-approved labs are paying more – around $9k, I’ve heard.

      The NHGRI pubishes the costs of a human geome sequence at its sequencing centers. The last data, from April 2013, showed the cost at about $5800. Go to this page,, and click on the link below the second graph, a link labelled Sequencing Cost Table. That will download an excel spread sheet with these costs. Interestingly, that number hadn’t really fallen in a year (4 quarterly data points). I think we’ll need a new technology to get much lower. Oxford Nanopore has announced a “pilot” of a (sort of) $1000 sequencer. That may prove to be the next next gen technology – or not.

  2. The NHGRI prices are wholesale – buy a machine and run it all the time. The few genomes that have been sold cheaply are misleading. They are a small amount of excess capacity that is sold at marginal cost, well below average cost, largely so that the company can brag about low prices, when it isn’t actually selling anything.

    23andMe is a good example – they sold a fair number of exomes at $1k, probably far below cost, but they won’t accept any more customers until it’s cheaper.

    But when I asked someone else about your numbers, I was pointed to DNA DTC, which does appear to be accepting customers at $900/exome and $7k/genome, pretty close to your numbers. But it might be a mirage, too.

    Didn’t Oxford Nanopore announce $1k genomes by 2012? The current pilot has $1k operating cost, but I haven’t been able to determine how much DNA it can sequence per unit.


  3. Almost always in health care, “prices” are wildly different from “costs”. Usually it’s in the other direction – prices much much higher than costs – but sometimes the other. I think $5k isn’t a bad number in this context . . . some will be higher, some will be lower, and the actual “cost” will vary depending on company circumstances and accounting conventions. BGI has such a huge set of machines up and running; I wouldn’t be surprised if their negotiated rates are a lot lower than $5k.

    As to Oxford, yep – in early 2012, they and Ion Torrent promised the $1k genome by the end of 2012. Still waiting. But at least Oxford Nano is out with something – I haven’t heard anything from Ion Torrent on cheap next NGS since mid-2012.


  4. Pingback: minglingken

  5. I suspect (although I have no inside knowledge whatsoever) that 23andMe’s management and/or legal team may believe that if the FDA tried to stop or regulate their marketing they might have a winning defense in the First Amendment after Sorrell v.IMS Health which implied that singling out marketing speech for special treatment would violate content neutrality. That, combined with the disapproval of so-called “paternalism” of protecting people from the truth because they might misuse or misunderstand the information from the original Virginia Pharmacy decision, is the argument I could see them thinking they might make successfully. I think this is terrible argument in terms of its consequences for regulation generally. (Whatever you think of the FDA, I think it is even worse to contemplate a market for drugs and medical devices which is unregulated.) There are, of course, many other possible explanations but that seemed like a plausible one.


  6. Could be – or they could be planning to challenge whether their tests are, in fact (or in law), ‘devices’ under the Act. You’d think, though, if that were their plan they would have taken the offensive with perhaps a declaratory judgment suit, or at least a press release, rather than be non-responsive for 6 months and then letting themselves get hammered (or anviled) by FDA.

    I suspect 23andMe had some plausible reason for their actions, but I don’t know what it was.


  7. Pingback: FDA vs 23andMe: Link Round Up

  8. Isn’t it more than a coincident that 23andMe appoints a new President this summer and immediately stops communicating with the FDA.

    It’s suicidal to run 23andMe like Gilt, the online luxury goods retailer, but it appears that’s exactly the business strategy.

    Coincidence? I think not.


  9. No idea whether it is a coincidence or not. Similarly, I don’t know whether the departure of Ashley Gould, 23andMe’s general counsel, in July 2013 was a coincidence. She had been GC there since 2007; left to become VP legal at something called Hyperion Therapeutics. Some speculation that this was important in a David Dobbs blog for the New Yorker, quoting 23andME co-founder Linda Avey (no longer with the company)to that effect:


  10. Pingback: THE FDA CALLS A PENALTY ON 23andMe | The DNA Exchange

  11. Pingback: The FDA Orders 23andMe to Stop Marketing Medical Tests

  12. Pingback: The Berkeley Science Review » The 23andMe Shutdown

  13. Pingback: The Sky is Falling for Personal Genomics! Oh, nevermind. It’s just a cease & desist letter from the FDA to 23andMe.

  14. Pingback: Why the FDA’s anvil dropped on 23andMe | Tech Auntie

  15. Pingback: Collecting Thoughts on the FDA vs. 23andMe | Emerging Technologies Librarian

  16. Peter W. Huber Esq., author of the new book: The Cure in the Code: How 20th Century Law is Undermining 21st Century Medicine believes that 23andme has a novel and winning argument based on the First Amendment.

    Might H. Greely comment on what Peter Huber is saying in his book and herein?

    “Peter Huber, an expert in science, technology, and public policy, demonstrates why Washington’s one-size-fits-all drug policies can’t deal with diseases rooted in the complex molecular diversity of human bodies. Washington is ill-equipped to handle the torrents of data that now propel the advance of molecular medicine and is reluctant to embrace the statistical methods of the digital age that can. Obsolete economic policies, often rationalized as cost-saving measures, stifle innovation and suppress investment in the medicine that can provide the best cures at the lowest cost.

    In the 1980s, an AIDS diagnosis was a death sentence, until the FDA loosened its throttling grip and began streamlining and accelerating approval of life-saving drugs. The Cure in the Code shows patients, doctors, investors, and policy makers what we must now do to capture the full life-saving and cost-saving potential of the revolution in molecular medicine. America has to choose. At stake for America is the power to lead the world in mastering the most free, fecund, competitive, dynamic, and intelligent natural resource on the planet—the molecular code that spawns human life and controls our health.”

    For more about Huber’s First Amendment arguments see also:


  17. Pingback: 23andMe Saga Doesn't Bode Well for EMR Genetics Integration | EMR and EHR

  18. Sequencing is continuously getting cheaper and cheaper. But we ahould stop quoting these misleading (strictly marketing) figures. Even if we could get our genomes sequenced for free, it would still cost a lot to properly interpret the data and to communicate its significance back to the “sequenced” people and all concerned relatives. Add to it that this may have to be done over an over, as our genetic knowledge improves.


    1. It is true that real analysis is currently much more expensive than sequencing – but what is the marginal cost of 23andMe’s analysis of the SNPs? I assume close to zero, as their calls are, no doubt, embodied in software. In the not-so-long run, WGS calls will be embedded in software, too. The cost of creating – and updating – the software to make the calls will be very high indeed, but it will be spread over all genomes sequenced. I think the question will be how much human judgment will be needed for each genome – probably won’t need much human intervention to decide whether a 185delAG variant on BRCA1 is dangerous or 88 CAG repeats in the Huntingtin gene, but just how many variants of unknown significance will need human action for each genome will probably determine the cost.

      So, I agree with you in the short run, but not in the mid-term . . . say, 5 years or more out.

  19. Pingback: Why the World Keeps Talking About 23andMe | NewsCenterd

  20. Pingback: Why the World Keeps Talking About 23andMe | Wonderful Tips

  21. Pingback: Why the World Keeps Talking About 23andMe – Technology – The Huffington Post | Internet Technology Business

  22. Pingback: Why the World Keeps Talking About 23andMe |

  23. Pingback: Why the World Keeps Talking About 23andMe | Political Ration

  24. Pingback: Mobile Technology News, December 14, 2013 | DigitalWorkshed

  25. Definitely, as time goes by, we’ll getter better at interpreting the clinical impacts of our genomic variants. And this will progressively be more and more automatable, thus more and more cost-effective. We will also get better at conveying, explaining all what we have learned from these sequencing efforts to the concerned patients. As the information to be communicated will continue to increase, we’ll need to invent, develop new, alternative means that will diminish the consultant’s time spent in direct contact with the counselors or medical staff. Obviously personal contacts (of various kinds, including when needed by psychologists) will be necessary to ensure that the significance and implications of all these results are properly understood and digested by the patients. I have no idea what the economics of these steps entails, but at present they far outweigh the sequencing costs, and I fear this will continue for quite some time.


  26. Pingback: Why the World Keeps Talking About 23andMe - Freshwadda Brooks | Coming Soon!

  27. Pingback: Why the World Keeps Talking About 23andMe | My Blog

  28. Pingback: Today in Technology December 14, 2013 | Tech

  29. Pingback: Long-Awaited Announcement from the FDA on LDTs

Leave a Comment