FDA shuts down 23andMe

November 25, 2013

Wow!

The Food and Drug Administration has ordered DNA testing company 23andMe to stop marketing its over-the-counter genetic test, saying it’s being sold illegally to diagnose diseases, and with no proof it actually works.

I did not see this coming at all. Guess I was too focused on thinking about informed consent issues.

Advertisements

39 Responses to “FDA shuts down 23andMe”


  1. I generally like regulation and wish the FDA would get on the case of mystics selling Chinese and Indian “traditional” cures — those can actually poison people. But instead they want to interfere with people’s right to know their genetic sequence — that’s not cool at all. Follow the money on this one — who is profiting?

    Like

  2. drugmonkey Says:

    I think you need to think more broadly about harms. Quack remedies that do nothing and take your money? And what about life changing or expensive options taken on the basis of misunderstanding strength of the evidence? Caveat emptor?

    Like

  3. Dr24hours Says:

    Seems like a reasonable thing to me. The letter states plainly that 23&me didn’t adequately respond to previous FDA issues.

    They’re dealing with a test that is unambiguously designed to diagnose. Oversight is warranted. They didn’t properly secure it. There’s no controversy here.

    Like


  4. Are they going to shut down genetic counseling? It’s exactly the same issue. Yes, I know the argument is “well, those are trained professionals”. But *they* can collect your saliva and send it off to a similar service and change thousands for it rather than a $100-$200? Bullshit.

    Like


  5. @Dr24hours
    No, it isn’t “unambiguously designed to diagnose” anything. It just tells you what SNPs you have and if they are associated with anything. They make it quite clear that correlation isn’t causation. Anyway, they are a lot more interesting reasons to look at your SNPs than disease — looking at your ethnic makeup and where your ancestors is actually why I did it, for example.

    Like

  6. DrIgg Says:

    23andMe and other PGS are an important leap in personal consumer health information. I think the FDA recognizes this, but is under strong pressure to set precedence of how these services are regulated. 23andMe is going to be central to these regulatory decisions.

    As you mentioned DM, and the FDA outlined in their letter, the consequences of false positives or negatives could be huge. Without oversight, consumers are left in the dark as to the accuracy of the tests. This company might have the highest accuracy testing and the utmost ethical company standard, but without regulation, the door is open to the inevitable “me too” companies that possibly won’t. The FDA’s hand is forced on this one.

    Like

  7. geeka Says:

    My thought is that if you took your 23&me sequence to your PCP and demanded treatment, they’d do whatever test they’d normally do if you had the symptoms of whatever the SNP says you’d have a proclivity for. Should the FDA regulate this? Probably not, it’s not a basis for treatment. It’s like the FDA having to regulate every microscope in every lab (not just the ones used for diagnostics).

    Like

  8. dr24hours Says:

    @Jonathan Badger,

    Sorry, should have said “unambiguously marketed to diagnose…”.

    Like

  9. zb Says:

    The article suggests sloppy, naive, interactions on the part of 23 and me (which is related to the informed consent issues DM has complained about as well).

    They are a tech/entertainment company moving into medical diagnosis and research and in doing so, they’ve ignored the ethics and regulation of those fields. If Genentech markets a genetic test, they have a clear idea of the FDA rules they have to follow and know they are a highly regulated industry. A geneticist doing research at a university goes through all kinds of hoops in order to get their study approved.

    23 and me is the reality show equivalent of psychology research and crossed into the zone that the FDA regulates and didn’t take the FDA seriously. Glad to see that the FDA used its power to regulate. I think 23 and me will loose this one unless they get with the program and hire the same expensive lawyers companies like Genentech use.

    Like

  10. zb Says:

    “I think 23 and me will loose this one unless they get with the program and hire the same expensive lawyers companies like Genentech use.”

    Oh, and listen to them.

    Like

  11. labrat Says:

    @J Badger: “Are they going to shut down genetic counseling?”

    Because that is done by – yes – trained medical professionals, which have a state issued license and send it to a facility / lab that uses FDA-approved tests and equippment (which is why the stuff is more expensive). They can still sell their stuff for ancestral tracing but not to give you information about medical conditions.

    Like


  12. @labrat
    Will they have to mask SNPs that are correlated with disease? That’s kind of the issue here — its all about information access, not about anything requiring special medical-grade equipment or training. As a genomicist myself, I’d consider “medical professionals” as laymen regarding anything genomic anyway. Personally what I liked about 23andMe is that you could download the data yourself and run third-party analysis tools or brew your own if you are bioinformatically inclined.

    Like

  13. labrat Says:

    There was a good Radiolab sequence (#509) on the “written word and its consequences” where a family wrestles with a medical test (genomic screening for a disease) that has a very clear, but meaningful diagnosis – Chorea. I think it is important to receive test results like these in the presence of an expert who can provide counseling right there and then. Now, I am not sure if 23anMe did those tests but in general genomic information with a wide array of consequences should be ]handled responsible, approved tests that are verified are one part of it.

    Like

  14. labrat Says:

    I am a physician and listening to this episode made it very clear to me how it is to be on the other side – same as when we had to make a decision based on probabilities regarding amniocentesis yes or no after a triple test during pregnancy. Which risk probability are you willing to bear. Its easy to say there is only a 10% chance to a patient – but if that is suddenly YOUR 10% the world looks different.

    Like

  15. zb Says:

    I’m guessing where they really ran afoul is in the marketing — not simply providing the SNPs, but arguing that those SNPs would give useful information about health outcomes.

    They want to market themselves for health outcomes, and argued to the FDA that they would provide acceptable scientific evidence at some time in the future, that the studies were ongoing. They haven’t done so, by the scientific standards acceptable to the FDA.

    So, now, they can try to provide that evidence (a big undertaking), or change their marketing materials. I don’t think they’d be required to mask the disease SNPs, but, presumably, the buyer would have to go find their own information about the SNPs, rather than get a poorly evidenced list of correlations from the company.

    Like

  16. drugmonkey Says:

    The article suggests sloppy, naive, interactions on the part of 23 and me (which is related to the informed consent issues DM has complained about as well).

    They are a tech/entertainment company moving into medical diagnosis and research and in doing so, they’ve ignored the ethics and regulation of those fields.

    Pretty clear that 23andme is not acting in ignorance at this point but has been fighting/refusing the FDA.

    FDA has spent significant time evaluating the intended uses of the PGS to determine whether certain uses might be appropriately classified into class II, thus requiring only 510(k) clearance or de novo classification and not PMA approval, and we have proposed modifications to the device’s labeling that could mitigate risks and render certain intended uses appropriate for de novo classification. Further, we provided ample detailed feedback to 23andMe regarding the types of data it needs to submit for the intended uses of the PGS. 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. 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.

    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.

    Like


  17. @labrat
    “Which risk probability are you willing to bear. Its easy to say there is only a 10% chance to a patient – but if that is suddenly YOUR 10% the world looks different.”

    Of course, but the only person who can make the decision is yourself — there isn’t any amount of “counseling” that changes that.

    @DM
    “Pretty clear that 23andme is not acting in ignorance at this point but has been fighting/refusing the FDA”

    Good for them. Disruptive innovations are disruptive. Automobiles were once banned from many streets because they interfered with horses. And the regulations banning them didn’t change by themselves.

    Like

  18. drugmonkey Says:

    Automobiles were once banned from many streets because they interfered with horses. And the regulations banning them didn’t change by themselves.

    Automobile companies didn’t start including seatbelts or airbags or crash protection design out of the goodness of their hearts or even because the market demanded it. They did so because of regulations.

    Like

  19. Dave Says:

    The biggest problem for SNP-based “testing” is that assigning ANY kind of causality is impossible. We don’t even understand what these (mostly) intronic and non-coding SNPs represent – are they simply markers or flagposts? Are they acting on cis or trans genes? Are they functional in any significant way? The basis for their association with diseases, especially complex polygenic disorders that we are primarily talking about here, is purely statistical in most cases. This was the promise and failure of GWAS studies.

    So it was always premature for these companies to be offering these tests and I am happy they have been shutdown. The science is just not there yet.

    Genetic counseling is a lot different. Often we are talking about non-synonymous mutations which are known to be causal and so testing is valid and the results are more interpretable.

    Like

  20. dr24hours Says:

    @DM “Automobile companies didn’t start including seatbelts or airbags or crash protection design out of the goodness of their hearts”

    Actually, the airpag was invented and patented by Mercedes-Benz, who then did not enforce the patent because the technology saved lives. They had the loegal right to say “Only Benz has airbags” and didn’t.

    So don’t tar everyone with the same brush.

    Like

  21. dr24hours Says:

    Although, a little research suggests that that story may not be 100% accurate. So I should fact check before spouting. Regardless, not there exist good corporate citizens.

    Like

  22. zb Says:

    I read that, too. But, I thought it was kind of like the producer of Dateline episode on bullying facing the questions of a Institutional Review Board. They wouldn’t understand each other.

    Like


  23. So are those microbiome goofballs next?

    Like

  24. Mike Says:

    FDA: “We believe that some of your advertising is misleading. Please address these concerns”
    23andMe: “Information wants to be free”
    FDA: “We believe that some of your advertising is misleading. Please address these concerns”
    23andMe: “Oh, that’s silly”
    FDA: “We believe that some of your advertising is misleading. Please address these concerns”
    23andMe: “Quiet! We’re collecting data! We still need more data. We’ll deal with all this hoop-jumping stuff later.”

    Like

  25. drugmonkey Says:

    So are those microbiome goofballs next?

    My prediction is that 23andme plays ball with the FDA now and gets their act in order. as someone mentioned above, this will be an important policy-setting case. the buttwipers at microbiome will probably (?) take notes? i mean how dumb would they have to be not to take it into account when 23andme gets spanked by the FDA?

    Like

  26. bluefoot Says:

    Interestingly, the 23andMe website has no indication of the news and the “order now” button for a test kit still works.

    I hadn’t looked at 23andMe’s website in a long time. They market their product saying the results will tell you about disease and health risks, and drug response (!). The FDA did the right thing. These aren’t validated clinical tests so they shouldn’t be marketed as such.

    Has anyone here used 23andMe’s services? Do they tell you the data behind the various SNPs that they call out ask risk factors, etc? I currently work in patient/risk stratification and to say it’s not that easy to predict disease course or drug response is a *serious* understatement.

    Like

  27. chris p Says:

    My read of it (after 20 years in pharma, medical device, and 5 years associated with a personalized medicine company):

    Technology innovators came up with a cool information-based product and sold it at first as a novelty. FDA was fine with the “for entertainment purposes only” nature of the information provided. Entrepreneurs enhanced their product in the last year to provide data that was truly diagnostic or related to therapeutic guidance (BRCA1, CYP processing) and started pushing this information out. FDA was not pleased with the validation studies that had been done on that information and said it needed to hit 510k qualifications. Management had an experienced Reg Affairs Director (from xDx who faced similar challenges ?) but decided to play tough. Why play tough? My speculation would be that the validation testing would swamp the company or they thought they were too big/popular/cutting edge for the FDA to stop them. Having worked at a company with multiple warning letters from the FDA that ended with a consent decree stopping sales for 1.5 years, I would say popularity does not work. If the FDA perceives the risk to patient safety, it is up to the company to PROVE it is safe rather than for the FDA to show evidence of the harm done. I am curious to see how the next chapter reveals itself.

    Like

  28. bluefoot Says:

    I should add: as a currently non-practicing geneticist, I’m all for people being able to sequence their own genome and analyzing the data, and volunteering to share that data if they want. But marketing the service as diagnostic across multiple diseases, risk factors and drug responses means the FDA *has* to step in.

    Like


  29. @bluefoot
    Has anyone here used 23andMe’s services? Do they tell you the data behind the various SNPs that they call out ask risk factors, etc?

    Yes. They actually give you your SNP data as a downloadable text file. That’s what is awesome about it. And they give references to the studies linking SNPs to phenotypes so you can check *yourself* how solid the correlation is. SNP analysis is *science* not mere medicine.

    And the influence of UnitedHealth on this decision is beginning to come out as HCP notes.

    Like

  30. DrugMonkey Says:

    They usually cite one or two studies IME. No more. It is not clear to me what threshold they use for asserting a useful link to a genotype. Is it discussed by a panel of experts?

    And we all know how troublesome it would be to assert a confident risk assessment on the basis of one or two papers.

    Like

  31. DrugMonkey Says:

    Who gives a frack whether a competitor who has to play by the rules is up in arms about 23andme’s shenanigans? What does this have to do with the very material issues being raised by people in the scitwittblogosphere who are not in COI?

    Like

  32. miko Says:

    The top hit in my “health” section is for a trait that occurs in 11.4% of my subpopulation. It assesses my my risk at 1.98x this average. There is a lengthy description of the symptoms and treatment.

    It is immediately caveated that this assessment is based only on published associations with variants that are represented on their chip. For each SNP, the effect size is shown, with an explanation of what this means.

    It then gives a twin study heritability estimate for the disorder (66-80%) along with a list of environmental/lifestyle factors shown to affect risk (with links to articles).

    Next, there is a timeline for the condition that starts with it’s first recorded description (Hippocrates), includes the twin studies in the 1970s, and links to every association study for the variants on their chip.

    At the bottom of the page, it says that the test is not FDA approved and is not for diagnostic purposes.

    On the discussion for this trait, people discuss their symptoms and various remedies. There are 2 comments (out of 18) that present wingnut theories. There are a few naive questions (how can my risk be decreased if my dad had it?) which are accurately answered by others.

    Overall, I am underwhelmed by the potential risks presented by the type of information 23andMe presents and the way they present it. I believe those who say their advertising and marketing information might be misleading….I have never seen any of it. Seems like an easy fix.

    Like

  33. respisci Says:

    Do the FDA concerns also involve questions regarding the reproducibility of the findings? Has anyone submitted their individual sample to 23andme various times and obtained identical results for each of these times?

    Would you not demand that of a genetic diagnostic test?

    Like

  34. sciencedude Says:

    It does not seem to be recognized by most people in this thread that 23andme not only looks at SNPs associated with various health risks, but also tests a few dozen SNPs that are directly associated with inherited conditions/diseases. For instance I discovered that I had a rare condition with variable penetrance for developing Emphysema. I went to see my regular doctor to have it confirmed. Fortunately, it is not a problem now, but would likely become one as I get older if I did not know about it. With that information in hand, I have made appropriate lifestyle changes and can get treated, if necessary, before it is too late. All the other the other SNP-associated health risks generally range between two-fold higher and two-fold lower than the general population. Mildly interesting, but so what.

    I agree with JB. The FDA has their head up their ass on this. Quoting from the story above, the FDA said: “For instance, if the BRCA-related risk assessment for breast or ovarian cancer reports a false positive, it could lead a patient to undergo prophylactic surgery, chemoprevention, intensive screening, or other morbidity-inducing actions…” Yeah, that sounds plausible. A woman finds she has one of the bad BRCA genes and immediately runs out to have her breasts removed before confirming the condition. That would also require a doctor willing to do the surgery based only on the 23andme results.

    Like

  35. Fe Says:

    It is just a scan for a subset of sequences out of the genome with some statistics provided that are correlated in some cases with published clinical data. Kind of like first pass type of project but tailored to individuals. Or a different approach to verify statiscal significance of some sequences for which functional data is not easily done. And a way to improve software? and to educate?

    Like

  36. Matt Says:

    I was and still am all for 23andMe.

    Any company actively creating a market for themselves by promoting the information of genetic code to a brand new target audience will do more good than harm. Most people glaze over when I start talking about genetics, genes, DNA; let alone a detailed description of single nucleotide polymorphisms, introns, exons, heritable traits, phenotype, genotype…

    At a basic level this test is helping inform people about the future of genetics and possibilities in one day understanding our individual genetic blueprints.

    But hey, let’s shut it down because we didn’t get our quarterly kick back on time.

    Typical.

    Like

  37. DrugMonkey Says:

    Interesting that you are so keen to educate on genetics and simultaneously so proudly ignorant and dismissive of ethics.

    Like

  38. Fe Says:

    I’m not dissmissive of ethics, but rather pointing out what SNPs by themselves are, and when solid functional and clinical data is not correlated with them. That’s why I threw in the possibilities of improving statistics or educating the public.

    Ethics involve values and moral issues, which should be backed up by clear laws that should be independent of political affiliation. Definetely ethics should be a BIG factor, because is the good thing but also because is wise. Science comes out better and applies longer term. Science is reasoned ethics.

    Like


Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s

%d bloggers like this: