FDA vs. 23andMe: How do we want genetic testing to be regulated?

Yesterday the US Food and Drug Administration sent a letter to the human genetic testing company 23andMe giving them 15 days to respond to a series of concerns about their products and the way they are marketed or risk regulatory intervention. This action has set off a lot of commentary/debate about the current and future value of personal genomics, whether these tests should be available direct to consumers or require the participation of a doctor, and what role the government should play in regulating them.

I am a member of the Scientific Advisory Board for 23andme, but I am writing here in my individual capacity as a geneticist who wants to see human genetic data used widely but wisely (although I obviously have an interest in the success of 23andme as a company – so I can not claim to be unbiased).

I see a wide range of opinions from my friends on this matter – ranging from “F**k the FDA – who are they to tell me what I can and can not learn about my DNA” to “Personalized genomics is snake oil and it’s great that the FDA is stepping in to regulate it”. I fall somewhere in the middle – I think there is great promise in personalized genetics, but at the moment it is largely unrealized. Looking at your own DNA is really interesting, but it only rarely provides actionable new information. I don’t think the FDA should restrict consumer access to their genotype or DNA sequence, but I do think the government has an important role to play in ensuring that consumers get accurate information and that the data are not oversold in the name of selling products.

As people try to decide what kinds of tests and information should be available and how the government should regulate them, I think it’s useful to ask a series of questions.

1) Should a person be able to have their DNA sequenced and get the data?

Putting aside any questions about how useful this information is right now and how it is marketed, do you think companies should be able to offer a service where consumers send in a spit or blood sample and a few hundred dollars and get their genome sequenced in return? (23andme currently provides SNP genotyping, not whole genome sequencing, but we’re very close to the point where human genome sequencing is cheap and reliable enough to make this possible.)

I think the answer is obviously yes. I can’t see any good argument for why we should prevent people who to from obtaining their own DNA sequence.

Which leads to:

2) Should a person who has had their genome sequenced be able to access scientific literature relevant to their genome? 

Again, putting aside questions about the accuracy or utility of this information, there is a lot of published scientific literature that is potentially relevant to people with a particular genotype (including genome-wide association studies as well as a lot of classical human genetic literature and other functional studies). Assuming someone has their own genome sequence, it would be hard to argue they shouldn’t have access to information that would allow them to understand what their genome means.

Which leads to:

3) Is there a role for third parties in helping people interpret their genome sequence? 

The problem with the previous question is that it would be next to impossible for someone to actually interpret their genome simply by perusing the scientific literature (and I’m not even going to get started on the fact that much of this literature is behind paywalls). Even trained human geneticists wouldn’t do that. They’d go to some website – OMIM, DECIPHER, etc… – and use various automated tools to extract what is known about their genotype.

But few people have the technical savvy to be able to analyze their own genome in this way. So, assuming there is interest, there is a great niche for third parties to step in an provide services to people to help them interpret their own DNA. Is this a bad thing? Again, I don’t see how it is – assuming that these third parties provide accurate information (more on this below).

Should this third party be a doctor, as some (mostly doctors) are arguing? There are certainly doctors out there who have a great grasp of human genetics. But there aren’t a lot of them. And even the doctors who do know the world of human genetics inside and out aren’t in a position to help people navigate every nook and cranny of their genome. This is a job for software, not for people.

If you accept points 1,2 and 3 above – which to me seem inarguable – then you accept the right for companies like 23andme to exist. You might not think they provide a valuable service. You might not think they do a good job at providing these services. But you can’t argue – as many are now doing – that direct-to-consumer genetic testing companies should be shut down.

Should direct-to-consumer genetic testing companies be regulated? 

I think this is also a no-brainer. Obviously they should be regulated – and fairly tightly so in my opinion. Few consumers have the capacity to judge on their own whether the genetic testing services provided by a company are accurate and whether interpretive information provided by third parties is valid. It is vital that the FDA protect consumers in two ways: 1) by validating the tests and the companies that provide them, and 2) by monitoring genetic information that is provided by consumers – especially if it is being used to market tests or other products. The former seems relatively easy – validating genotyping and sequencing is well-trodden turf. The latter is a bit more complicated.

If genetics were simple and our understanding of it were complete, companies could provide accurate reports that say “based on your genotype, your age and personal history, you have a 7.42% chance of developing ovarian cancer in the next 10 years”. However, we are far, far, far away from this. We have an incomplete catalog of human genetic variation; known genetic variation can explain only a small fraction of the heritable component of most phenotypes of interest; we have a poor understanding of how different genetic variants interact to affect disease risk or other phenotypes; and we have essentially no capacity to incorporate environmental effects into predictive models. In many cases current, incomplete, data may point to someone having an elevated risk of some disease, when they really have a lower than average risk. And, to top it all off, there are very few cases where knowing your risk status or other phenotype points to genotype-specific actions (with the BRCA status referred to in the FDA letter a notable exception).

The data are, at this point in time, very very messy. I don’t think anyone disagrees with that. The question is what to do about that. One the one side you have people who argue that the data are so messy, of so little practical value, and so prone to misinterpretation by a population poorly trained in modern genetics that we should not allow the information to be disseminated. I am not in this camp. But I do think we have to figure out a way for companies that provide this kind of information to be effectively regulated. The challenge is to come up with a regulatory framework that recognizes the fact that this information is – at least for now – intrinsically fuzzy.

The FDA wants to classify genetic tests like those offered by 23andme as medical devices, and to apply the appropriately strict criteria used for medical devices to genetic tests. But the problem with this is that contemporary genetic tests will almost certainly fail to meet these criteria, and I don’t see who benefits from that scenario. Genetic tests are simply not – at least not yet – medical devices in any meaningful sense of the word. They are far closer to family history than to an accurate diagnostic. The FDA and companies like 23andme need to come up with standards for accurately and honestly describing the current state of knowledge for genotype-phenotype linkages and their application to individual genotypes. They need to establish what generic statements can and can not be used to market genetic tests so that people don’t purchase them with unrealistic expectations about the kinds of information they will provide. Let’s hope this flareup between the FDA and 23andme is the spark that finally makes this happen.


This entry was posted in genetics and tagged , . Bookmark the permalink. Both comments and trackbacks are currently closed.


  1. DrugMonkey
    Posted November 27, 2013 at 12:12 am | Permalink

    Well put. I agree.

  2. Linda Avey
    Posted November 27, 2013 at 4:00 am | Permalink

    Your points about “very very messy” data and an “incomplete catalog of human genetic variation” are–as you know, from that first meeting long, long ago in your office–what sparked the start of 23andMe in the first place. If all 23andMe was interested in doing was genotyping a small set (hundreds) of known markers and explaining their role in human disease, the costs would be less and there might be more margin to direct toward regulatory expenses. But the over-riding mission to conduct research requires the generation of a much larger dataset per person, therefore dictating higher cost of goods. Even though the new v4 dataset has fewer markers (600k or so), it’s still enough to drive this research effort, a point that seems to get lost in all this regulatory debate. It doesn’t negate the need for regulatory oversight but I wish this unique aspect of the service at least came up in the discussion.

    • Posted November 27, 2013 at 7:45 am | Permalink

      Of course. But the FDA isn’t trying to regulate that service (yet).

  3. David States
    Posted November 27, 2013 at 4:31 am | Permalink

    I am with you until the final paragraph. In some cases, yes, genotype information is analogous to family history, but in other cases it is highly specific. Are you arguing that 23andMe should specifically censor high penetrance medically important loci like sickle cell, Huntington’s and CF? Where do you draw the line? ApoE? Dopamine receptor polymorphisms? An alternative view is that 23andMe should be divided into two distinct entities, a technical genotyping service and the publisher of a genotype indexed journal. The technical accuracy of genotype IMHO can and must include SNP specific confidence estimates. As long as reasonable accuracy targets are established and 23andMe meets them, I am OK with offering direct to consumer genotyping. As far as the genotype indexed journal goes, that exists right now, just type “delta F508 JAMA” into Google. No one argues that the FDA should regulate journal content, and no one wants to keep a cystic fibrosis patient from reading about their disease. And journal articles about CF are quite specific and personal for a patient who has the disease.

    • Posted November 28, 2013 at 10:00 am | Permalink

      But my point is that if you’re ok with public genotyping – which you are – and you’re ok with people interpreting their genotypes against the scientific literature – which you are – how can you not be ok with a third party helping people to interpret their genotypes with respect to the scientific literature?

  4. drworms
    Posted November 27, 2013 at 5:34 am | Permalink

    Great post, I feel similarly about the role of the FDA. Largely, that it should be to protect consumers from direct harm caused by the company/product. While I agree a much better framework forinterpreting the results is necessary, I’m not sure the FDA is nimble enough to reevaluate each new genotype/phenotype association we find. Perhaps a new regulatory body is in order?

    Second, and I’m not the first to make this observation, its frustrating to see the FDA go after 23andme when I can walk into a store and find a homeopathy section with shit marketed to idiots that is unregulated. Why allow people to sell Fresh Organic Unicorn Tears as a way to realign your Chi with your Chakras without regulation?

    • Posted November 27, 2013 at 7:50 am | Permalink

      I guess I wasn’t clear about this. I DO NOT think the FDA should regulate every statement made by 23andme. That would never work. Rather I think there need to be some general standards for how to report information in the literature to patients that capture both the current state of our understanding and what we do not know. 23andme has tried to do this all along, and I don’t think they would object to FDA reporting standards that would provide a framework for how to report this information.

      And it’s not the FDA’s fault that they can’t go after homeopathy and supplements, etc… – they’re not allowed to by law.

  5. gimpy
    Posted November 27, 2013 at 6:11 am | Permalink

    Here is a list of all the diseases 23andme test for, many of these are monogenetic.

    How on earth can you argue that they aren’t promoting viable diagnostic tests?

    More generally, you aren’t dealing with the issue of informed consent. 23andme do not inform their customers of the risks of genetic testing, nor do they ensure they have an adequate understanding of these before accepting their money. In fact, there is a conflict of interest between their motivation (profit) and their customer’s motivation (understanding their genome with respect to health risks). Ethics demands an impartial intermediary, whether medical professional or genetic counsellor, to advise people on the risks and benefits of testing before they consent to it.

    You need to consider this.

    • Posted November 27, 2013 at 7:59 am | Permalink

      Yes. There are lot’s of viable genetic tests for monoallelic conditions carried out by 23andme. Interestingly, the FDA did not seem concerned about these – rather it was more worried about the cases where the information is not so complete.

      And I do not agree with you about the need for intermediaries. I don’t think people need to be protected from information, especially about themselves.

      • gimpy
        Posted November 27, 2013 at 8:18 am | Permalink

        It’s not about protecting people from information, it’s about making sure that, before they consent to testing, they have the ability to adequately understand the risks and benefits of knowing that information because it has the potential to cause harm.

        Perhaps it is a cultural thing, in the UK genetic counselling is an integral part of the medical system. E.g. http://www.nhs.uk/Conditions/Genetics/Pages/genetic-testing-and-counselling.aspx

    • Kevin McKernan
      Posted November 27, 2013 at 8:17 am | Permalink

      Hayek won the nobel prize in economics pointing out that both parties profit in any given economic exchange or the exchange doesn’t happen. Anyone familiar with how to balance a chemical equation should get this. Why the fear mongering on profit motives which then need a monopolistic oversight agency? I fear the agency more than the company who has competition. With 23&Me not earning a profit yet, it would imply the customer is getting a larger share of that profit than the company.
      There are no reported victims of 23 & Me. There are many victims for killing an industry like this.

      • Nick
        Posted November 27, 2013 at 10:03 am | Permalink

        “Hayek won the nobel prize in economics pointing out that both parties profit in any given economic exchange or the exchange doesn’t happen.”

        Yes, this is obvious. Both parties obviously believe that they will profit or they wouldn’t choose to exchange and since everything everyone believes to be true is in fact true, the result follows immediately.

        It’s hard to believe that someone got a Nobel Prize for something so simplistic! Whoops, sorry, I meant simple.

      • gimpy
        Posted November 28, 2013 at 2:49 am | Permalink

        If we’re throwing Nobel prizes in economics at each other then the more recent one for the description and proof of Information Asymmetry is surely more relevant here?

        • Salvation regulation
          Posted December 1, 2013 at 7:01 pm | Permalink

          That’s a Möbius strip of confusion you just replied with. One could argue the NSA has asymmetric information about every transaction. They are publically on record with the fact that they share their asymmetric information with other government agencies like the FBI and DEA. There is nothing to stop them from sharing this with the FDA although that’s never been reported.

          What’s your point? That monopolistic regulatory bodies which can be captured by corporate interests and have asymmetric information are going to make angel like regulators?


          • Hermenauta
            Posted December 2, 2013 at 7:08 pm | Permalink

            You know, when a regulatory agency deals with assymetric information phenomena, there are some tricks that you can use to minimize the risk of capture. For example, they usually do public consultations to ensure all points of view are took in account, what rises the cost of being irrational. So your caracterization of the behaviour of a regulatory body is not totally fair.

    • Posted November 27, 2013 at 8:23 pm | Permalink

      gimpy: “23andme do not inform their customers of the risks of genetic testing”

      I’m not sure when you last visited their site, but there are disclaimers and explanations all over the place. At ordering, at every step of looking at your results, and for the more sensitive ones, there are even more steps that you have to read and click through. 23andme does more to educate people than you may realize.

      gimpy: “nor do they ensure they have an adequate understanding of these before accepting their money.”

      I’ve had genetic counseling the traditional way. I asked questions of the counselor, who seemed annoyed at some questions and tried to steer me away from asking others. I’d much rather a web page, backed by experts curating information form the literature. I can’t have a conversation with a web page, but the utility of that, in my experience, isn’t quite as great as you seem to suggest.

      What it comes down to for me is that for all the specter of harm raised, no one (and surely not for lack of trying) has been able to point to any actual harm caused in the years they’ve been offering their service.

  6. Posted November 27, 2013 at 6:44 am | Permalink

    I had a similar experience with the FDA over another test which wasn’t genetic but also wasn’t testing yes/no for a condition. They basically made us write a 510(k) on the basis that we were saying yes/no cancer even though it was widely inappropriate and all our test offered was more data that would help a clinician make that decision.

  7. Posted November 27, 2013 at 7:20 am | Permalink

    I absolutely agree that we need regulation for DTC genetic testing.

    What if we had regulation to incentivize insurance companies to only covered diagnostic/clinical tests where data is shared ethically, with informed consent from the patient?

    Data sharing is the bigger problem. 23andMe can only test for three of thousands of mutations on BRCA1 and BRCA2 because the company that formerly had a patent on our community’s genes still treats our data as a trade secret. Here is my take, from the perspective of a patient / BRCA carrier: ow.ly/rdoOR

    • Jonathan
      Posted November 27, 2013 at 9:41 am | Permalink

      Ironic, then, that 23andMe has applied for and been granted gene patents of its own, no?

      Gene variant databases such as Myriad’s being locked up as trade secrets are indeed worrying, but the industry will argue that without some commercially defensible intellectual property, it is impossible to attract the investment necessary to develop and commercialize the research. How do we square this circle? Perhaps your incentive idea is a way forward?

  8. 897
    Posted November 27, 2013 at 7:21 am | Permalink

    Is it fair to characterize an action by FDA in interest of public health as a ‘flareup’ with company X? Aren’t they just doing what they are mandated to do?

    Similarly 23andme saying in their most recent blog they are having a ‘dialogue with the FDA’ suggests friends chatting cosily over coffee. It isn’t.

  9. Kevin McKernan
    Posted November 27, 2013 at 7:34 am | Permalink

    The age old fallacy: “People are so stupid we need a monopolistic government agency to make them safe from themselves. Meanwhile those stupid people can still go and vote for more agencies.” Circular logic you would expect most scientists would be immune to.

    The market is the best regulator of safety.
    When the FDA makes a mistake they get a bigger budget.
    They have no competition. Under writers laboratory is a private example that works and showcases how regulators can be competitive and don’t have to be a gov monopoly. UL approval is used for electrical equipment which could burn your house down. It’s not a gov monopoly.
    How many people have burned their house down or been harmed by 23&me? There isn’t a single case reported. How many people die from FDA approved drugs or die waiting for the FDA to approve a drug..100,000s. You will never get safety through monopolistic regulatory bodies.

    Meanwhile, what are the unseen consequences of shutting this down. 23&mes service alerted Sergei Brin of his LRRK2 mutation which has now brought more money into Parkinson’s research. How many other high net worth individual wont get engaged as a result of this silly decision?

    • Posted November 27, 2013 at 8:10 am | Permalink

      I never said people are stupid and we need a government agency to protect them. To the extent that the FDA is trying to protect people from information, I think they are way off base. I think people should be able to order up genetic tests on their own without any interference from the government or medical professionals.

      I think 23andme has done a very good – although not always perfect – job in presenting information in an accurate way and providing context for the information they provide. However, as you surely appreciate, there are lots of incentives for companies to provide misleading information to consumers to promote products, especially when they have to do with health. But there are already lots of genetic tests on the market that are completely bogus, and consumers are generally not equipped to distinguish among them. I would love it if there were independent organizations that certified information in genetic tests, or, more importantly, information from pharma about drugs. But there aren’t, and until the are, I don’t have a problem with the FDA playing a role – provided that they view their role as protecting the integrity of information consumers receive, not restricting their ability to receive it.

      • Nysaa53
        Posted November 27, 2013 at 11:04 am | Permalink

        I have a 23&me DNA report. I find it very useful to see what health problems MIGHT occur and what issues I need to watch for. For example I’m overly sensitive to warfarin. This is good to mention to my doctor that this is something we need to be aware of should I need this medication. I am also intrigued by the ancestry information. BUT in no way am I taking this as a guide to self diagnosis. The FDA should keep out of this company. This should be no more health regulated than the webmd and mayo websites that provide information on health conditions.

      • Posted December 3, 2013 at 10:41 am | Permalink

        “I would love it if there were independent organizations that certified information in genetic tests,…” “But there aren’t, and until there are, I don’t have a problem with the FDA playing a role – provided that they view their role as protecting the integrity of information consumers receive, not restricting their ability to receive it.”

        Michael, might you or another poster here be able to expand on the above reply you made? Perhaps cite or link to what other counties are or aren’t doing in this area. Personally, I think this needs to be taken on from an International Perspective. For this certified information on genetic tests … are you speaking of 1. genetic tests for ancestry – criminal DNA forensics – etc. purposes? 2. genetic tests for health – diet – exercise – beauty – etc. purposes? 3. neither 1. 0r 2? 4. both 1. and 2. Why don’t you, Linda Avey, me and others study this issue more – get a grant to study about the framework of such an “international” “independent” “genetic” “genetic” “organization” (IIGO – sort of like a NGO for Human Genetics) which could “certify” genetic information – then seek funding to establish that organization. One of my avocations is the History of the Red Cross / 1864 Geneva Convention and the great woman, Clara Barton, who made this happen in the U.S. on May 21, 1881 at 1326 I Street NW, Washington D.C. It took nearly 17 years for this. Perhaps Linda Avey can step up to the plate and be our modern day Clara Barton for Human Genetics.

    • gimpy
      Posted November 27, 2013 at 8:11 am | Permalink

      Just on the specific Parkinson’s thing, and this is wholly separate from the FDA questions, but 23andme actually risked harming Parkinson’s research by patenting a variation linked with LRRK2 variations to modify risk of PD.

      The last thing research needs is to have areas marked as no go to researchers because of patent issues.

      • Kevin McKernan
        Posted November 27, 2013 at 8:42 am | Permalink

        Both fair points.
        Im involved in a CLIA and CAP certified genomics lab which only accepts tests from physicians and most colleges are stunned at my support of 23&Me. They are the Kahn academy of genomics and without them the 3,000 genetic counselors in the US will become a life threatening bottleneck. I have no problem if the FDA wants to publicly contest the labeling of the product. While there are many great scientists in the FDA, I don’t trust their monopolistic structure or bottleneck to ever provide justice to this field. I think a consumer reports of various genetic test out there is a worthwhile effort and will likely emerge (DNA Direct I think does this now) the moment there is a market for it. I suspect this market wont emerge until someone is actually harmed by the data.

        As for patents, blame the game not the player. 23&Me didn’t create the patent system but I can assure you that its nearly impossible to get a biotech company funded without having some patent applications. In fact the patent system was invented to address the fallacy of “market failure”. Ironically it creates exactly the opposite of protection from market failure. The goal of protecting the little guy almost certainly selects against them with the current patent system. The center for innovative freedom writes prolifically about the pitfalls of IP and they are IP attorneys. http://c4sif.org

        • gimpy
          Posted November 27, 2013 at 9:59 am | Permalink

          Speaking specifically on the patent issue, and I don’t want to distract from the thought provoking discussion on the specific FDA issue much more, but while it is fair to blame the game, it also has to be considered that patenting a test for naturally occurring sequences is generally only possible in the USA, and patents will only apply in the USA, but the effects of the patent will have impact on research elsewhere in the world as nobody with links to the USA will want to be involved.

          In this case, it is a purely selfish act that doesn’t benefit anyone but the patent holder, and damages people and research outside the USA.

          This is a reason to strongly discourage attempts by any personal genomics company to patent data derived from its customers. In fact a moratorium on the whole concept would be sensible.

          • Kevin McKernan
            Posted November 27, 2013 at 2:52 pm | Permalink

            Its better known as his “Pricing mechanism” work.

            So if profit is a measure of win-win transactions why do we need to regulate it with a monopoly of force? Threads here imply that profit motivation is dangerous and it needs to be regulated with guns? I know that sounds extreme but if Anne doesnt listen to their position, make no mistake, they raid raw milk farms with guns they will raid silicon valley with the same procedures.

          • Kevin McKernan
            Posted November 27, 2013 at 3:37 pm | Permalink

            Preaching to the choir on gene patents

            But this still doesnt negate that Sergey has pumped money into Parkinson as a result.

          • gimpy
            Posted November 30, 2013 at 2:30 am | Permalink

            I have no idea what you’re on about with your talk of regulation with guns? Is this a USA specific turn of phrase?

            All industries need to be regulated one way or another to ensure markets operate with limited abuse such that people and companies are not taken advantage of or taking advantage. It’s why there are laws to regulate everything from the sale of property to the minimum standards of food. In the case of medical products there are various regulatory bodies to ensure that, as best they can, products work and harm is minimised. The debate over the FDA is one of the extent to which this is necessary with respect to genetic testing. Nothing to do with guns.

          • Salvation regulation
            Posted December 1, 2013 at 1:45 pm | Permalink

            Gimpy- you have presented a Logical fallacy known as argumentum populum. Just because regulation has been done or is the norm doesnt make it pragmatic or ethical. http://en.m.wikipedia.org/wiki/Argumentum_ad_populum

            I believe the previous post was to remind those who believe in coercive regulation that they are unethical and this applies to the current architecture of the FDA.

            Many ethical product review and certification companies exist. UL is one, US news and world reports on colleges is another. These reviewers have competition and as a result are more responsive to their customers and more accountable. Customers can leave if they do a poor job. This voluntary nature of the contract is what brings excellence and quality into the picture.

            There is no recourse or accountability for the FDAs mistakes. There is no accountability for people harmed awaiting their drug approvals. Their track record on drugs they do approve is impossible to measure as they have no competition (in the US).

            Their edicts are enforced with the tacit threat of a gun. If you don’t comply, people with guns will arrive at your door. Here is an example.

            The FDA and everything the government does is funded through the same violent threat if there is a failure to pay taxes.

            So the people they are claiming to protect are forced at the threat of violence to use their services. The companies they ‘regulate’ have no free market selection over that contract and are also forced to work with them.

            As admitted by Nick, the voter has no control over FDA policy. This is why the FDA is viewed as a monopoly. It claims to protect those it represents and fails more frequently than a free market service would deliver in its absence.

            Fact-log order more people die from FDA approved drugs than illegal ‘unregulated’ drugs.

            Fact- the FDA has no competition and its prices have risen faster than inflation.

            Fact- Their ‘regulation’ is 100% backed by bullets and is not a credible authority despite the well intentions or excellent science in the organization.

            So I think you are not seeing the difference between ethical and voluntary market interactions which have been shown to produce excellent regulation compared to regulation which is by force of a non representative government. The latter never works and not surprisingly it is all based on violence.
            This is what’s usually called a protection racket.
            So once again, why do I need the FDA to read my genome? What gives you the right to expose me to that lack of freedom? It’s your cultish belief in the state which has proven over history to be far more dangerous than 23 & Me. 262M people have been killed by their OWN government in the last 100 years (google RJ Rummel). Your cultish belief in the state perpetuates this violence with no evidence of its neccesity. This belief system now restricts my freedom to purchase a 23 & me test. While you argue over how much FDA there should be in reading my own genome I am suggesting the only ethical answer is zero.

    • Nick
      Posted November 27, 2013 at 10:30 am | Permalink

      The age old fallacy: “People are so stupid we need a monopolistic government agency to make them safe from themselves. Meanwhile those stupid people can still go and vote for more agencies.” Circular logic you would expect most scientists would be immune to.

      There are two levels of wrongness to this statement:

      1) People can be uninformed about medicine and still manage to elect people who select informed people to work in regulatory agencies. We’re not talking about the entire nation directly voting on FDA regulations.

      2) Even if we were doing such a thing, that vote would be decided by the median voter, not its most uninformed. A population composed 90% of Mike Eisens and 10% of Deepak Chopras would probably produce a very nice regulatory framework to protect the Deepaks from themselves (which, yes, I’d say they need).

      • Kevin McKernan
        Posted November 27, 2013 at 8:44 pm | Permalink

        I appreciate your dissection here but you do realize you have just presented a strawman on point 1. You have restated the question in a new analogy but not challenged the premise. That premise being that uninformed people vote for ‘informed’ people to protect them from their ignorance. How can an uninformed person detect or judge who is informed enough to do this correctly? Should we elect a community organizer or Francis Collins? How would an uninformed person know the difference? Once voted for, who regulates those elected regulators and at what frequency? You just admitted that’s the voters job to start but clearly they don’t vote on FDA level policy?

        The point of highlighting the circular logic in voting for monopolies that regulate companies with competition is to showcase this fundamental fallacy. One could expand your point and make this a more ironic discussion. Does the voter have direct access to the policies of the FDA or direct access to the decision made to continually drone Pakistan, hold gitmo open, and spy on every American? This is a very relevant point. It’s not a non-sequitor. Are you suggesting the individual voter is not responsible for those atrocities because s/he’s uninformed in those matters? If you accept this…Do you mean to imply the representatives don’t always represent the voters wishes because they are smarter than the voter? How could they represent if the voter is not a specialist who can really judge their representative? Your comments declare those too dumb to known whats good for what they are most vested in (themselves) is best served by those who once elected also bomb foreign countries through that mechanism of making 23 and Me safe? Brilliant.

        A valid non sequitor to this discussion is that ‘when’ the informed class wants to listen to their uninformed voting body, they use cotton-gin like technologies of the hanging chad further befuddled through the electoral college at a breakneck pace of every 4 years. Meanwhile, in the ‘dangerous’ free market (that lacks access to this break through technology called voting:) we have flash trading on the microsecond time scale. How does this dichotomy persist when the government even has the technology to read everyones email? Why collect a vote when they have subsecond access to your thoughts? The vote is really a religious ceremony of prayer impotent to real representation. Even if your vote did count, it would only count if there was a tie and there has never been a tie. The same organization that has technology to read every email and drop smart bombs down chimneys through encrypted internet channels greets its uninformed voters with a representative technology called ‘paper ballet voting’ which hasn’t advanced since ancient Greece. Any attempt to modernize is arrested over the fear of electronic corruption. Really? Hope their IT systems governing their nukes are not as vulnerable as their voting technologies.

        At this point you have two dead end arguments.
        1)Voting can properly find experts and fund them at a monopolistic regulatory body like the FDA to keep the 23 and Me customer safe from harm despite the fact that voting is a joke in terms of modern forms of voicing your concerns. In the market place this wish is voted on with your dollars everyday (flash trades are microsecond timescales). If the same uninformed person can select someone to vote for, they can also select a company that is responsible with genetic information. You simply cant split this dichotomy with any reason as the vote enables a monopoly with poor representative frequency and the free market options do not.

        2)Voting isn’t a joke and the right way to keep a person safer from companies with competition is to have the customer vote for an organization without competition (gov). You can change the former by voting with your $ for a new company’s improved service or by not buying their service but you can’t change the latter (at least for 4 years). Any assumption that government is magically nobel and uncorruptable only needs to reference RJ Rummel which shows its more dangerous than small pox for its own citizens. And the organization you are relying on in government happens to have a former Monsanto executive at the helm. This revolving door isnt a conflict and will benefit all voters in all circumstances.

        So we’re back to the circular logic of “people are uninformed so we need these same people to vote for who will protect them from themselves”.

        Your second point suggests we are a democracy. This is historically inaccurate. We are a constitutional republic but either case admits that 2 wolves and a 1 sheep can vote on what is for dinner and that’s moral. I would suggest one stop worrying about how to ‘protect’ other people from themselves and ask yourself if you morally have the right to.

        Thats a lot to chew on. Enjoy the holiday and lets be thankful the FDA hasnt shut all sequencing down. As I have said before, intelligent specialists do exist in the FDA. My comments are not to discount their exceptional science. Its their gov framework of no competition or accountability I find to be a fallacy of protection for most things they regulate.

  10. Dean Sequera
    Posted November 27, 2013 at 7:52 am | Permalink

    I work in R&D or a company that develops clinical diagnostic assays which are subject to FDA regulation. Thus I understand the position the FDA is taking. But the FDA is seriously behind the technology curve and slow to formulate appropriate regulations, not only for direct to consumer DNA testing but also for new analysis techniques. The FDA is under political pressure to exercise its authority more vigorously. But they are using old tools. When the only tool the government has is a hammer, everything looks like a nail.

    There are some deep pockets at 23andMe to back a legal case. It has been brewing for quite some time as this site documents:


    I do think it is reasonable of the FDA to ask for the methodology and statistics if their kit was marketed as a diagnostic test. I got the kit myself and never looked at it as a diagnostic kit and didn’t see any language that would have led me to that conclusion.

    However, the FDA has been somewhat inconsistent as to what it has regarded as its regulation domain. For example they let Genomic Health market its Onco type DX test for years without intervening. The FDA IVDMIA (in vitro diagnostic multivariate index assay) guidelines have been languishing for over 7 years.

    Looking at SNP profiles in the genome is akin to DNA fingerprinting with something like RFLP. From this one can build dendrograms of relatedness. But like FBI fingerprinting, there must be a large database. So 23andMe is following the trend of “Big Data” to identify patterns. This is not surprising given its origins in Silicon Valley and the relationship with Google which uses similar techniques with Google Translate. Given enough input (DNA SNPs) one can map to outputs (tendency to disease states). I can understand the reluctance of 23andMe to share those proprietary “black box” algorithms. But is it necessary to understand the complex mathematics behind the combination of neural networks, fuzzy logic and whatever else they might employ? Do they need to go through expensive, time-consuming, inefficient clinical trials on a vastly smaller subset of patients for each condition that may be elucidated? It’s the wrong model to apply.

    The information 23andMe provides is one more piece in the puzzle to be able to get an overall picture of disease states. There is so much more than DNA. Epigenetic and environmental factors must be considered. When a broad DNA profile is coupled with specific molecular and immuno diagnostic tests (very much the domain of the FDA) patients are better prepared to work with their doctors to chart a course of appropriate action to ensure optimum health.

  11. Posted November 27, 2013 at 8:00 am | Permalink

    Only one thing to add:

    The FDA wants to classify genetic tests like those offered by 23andme as medical devices […] The FDA and companies like 23andme need to come up with standards

    It’s my understanding that the FDA explicity offered 23&Me the opportunity to do just that, to work with them on developing novel standards. This is a biotech/pharma dream come true and 23&Me almost deserve to be shut down for being stupid enough to ignore it.

    • Jared
      Posted November 27, 2013 at 11:14 am | Permalink

      There’s no hope for you if you believe that story…
      Doctors run the FDA. Doctors want their cut. It’s as simple as that. No “working together” bullshit is going to fix that situation.

  12. Mary
    Posted November 27, 2013 at 9:44 am | Permalink

    I don’t think 23andme is a bad actor in this space. Besides the information I got from my results, I have never been spammed with offers for other products that will work with my sequence info to improve or affect my health. That said, I do think the info they provide should be quality controlled and validated at some level.

    But this makes me ill: http://mygenewizepro.com/ And this is the kind of genomic snake-oil that will arise if the FDA lets the field just drift along. Actually, I’m sure it will arise anyway off shore, but it would be wise for quality providers to be on the side of the FDA and distinguish themselves from charlatans.

  13. Brian
    Posted November 27, 2013 at 10:03 am | Permalink

    I’m thinking the genie is out of the bottle. I am not an expert in any of these fields and specifics have already been eloquently posted. I will send my spit to China if I can’t do it at competitive cost at an FDA approved lab. I will expect (soon) a standardized digital container for my DNA. With DRM of course. Is there a service that can also analyze my DNA (FDA approved of course, a “root” of trust) that can provide me with a key to my genome data file? Not only would it be used to unlock it for analysis, but it would verify the file is human (in case the fly by night lab is real dirty…) and specifically verify it is my DNA. Now I have complete control over who I submit it to for opinions.

    Like Martin Luther 23andMe is pioneering translating the software of man into a format man can understand. I expect large institutions will find this problematic.

    Like in the media industry your DNA will be torrented.

  14. Posted November 27, 2013 at 12:10 pm | Permalink

    Michael, you state that “Genetic tests are simply not – at least not yet – medical devices in any meaningful sense of the word.” The statutory definition of medical device includes any article “intended for use in the diagnosis of disease or other conditions.” Congress has charged FDA with overseeing any article the meets this definition and, as a regulatory professional myself, I agree with FDA that 23andMe’s Personal Genome Service meets this definition. The standards that FDA expects 23andMe to meet are the same that other companies that offer genetic tests with medical interpretations are expected to meet.

    You also state that “The FDA wants … to apply the appropriately strict criteria used for medical devices to genetic tests. But the problem with this is that contemporary genetic tests will almost certainly fail to meet these criteria.” However, several companies offering human genetic tests with medical interpretations have already received approval or clearance for their devices, so it appears that your statement is incorrect.

    • Jared
      Posted November 27, 2013 at 3:06 pm | Permalink

      Genetic testing in no way diagnoses disease. What are you talking about?

      Genetic testing allows you to read scientific studies with relevance to your DNA. It does not seek to diagnose disease. While your connection sounds plausible, it is completely wrong. You cannot tell if someone has a disease from a DNA test. End of story.

  15. BenK
    Posted November 27, 2013 at 2:23 pm | Permalink

    I agree we need genomic testing regulated to a degree. However, I do not think that the FDA is headed in the best direction on this one.

    The labs are already regulated by CLIA. This should mean that the sample makes it in and the data makes it out without mix-ups. This is critical regulation and is challenging for some groups to achieve.

    There is a second level – is the data what is being advertised? I’d argue that SNP screening is not at all what 23andMe advertises to the average Joe, and in this, I feel they are acting poorly. They either need to make it very clear it isn’t a genome… or include copy number variants, etc. This is not easy and not cheap.

    In addition, now we know more about the diversity of sequences within tumors, the brain, and so on; this needs to be made more clear to people (and scientists).

    There is also the issue of the microbiome – and the fact that the eukaryotic genome interacts with environment, microbiome, medical history, and so on. This is ignored too glibly.

    Thus we reach the interpretation. Here I perceive a battle for control, in which doctors want to regulate access to the data so that they can select the vendor, control the price, take a bit off the top, etc. Perhaps I’m cynical. I don’t think the genome will be as predictive as people think; and I believe that everyone in the chain will try to profit from the patient. Patients often receive unneeded x-rays because medical records aren’t transferred well and physicians just order another test. Genomics should not be the same way.

    I’d prefer that there be professional certified genetic counselors; that interpretation and laboratory performance be broken apart. The individual should be able to purchase the sequence and take it to any counselor, or physician – or to the library and consult textbooks, if he would like. He may end up getting what he pays for in the analysis; but at least he can control the data.

  16. Jay Essay
    Posted November 27, 2013 at 2:31 pm | Permalink

    The problem with Mr. Eisen’s proposal, that the FDA and 23andMe and other such companies come up regulations outside the medical device definition, is that it ignores the very clear statutory scheme of the Food, Drug, and Cosmetics Act and the accompanying case law. Food and drug law revolves around classifications, be it food, drugs, dietary supplements, cosmetics, medical devices (and within that, class I, class II, and class III devices), etc. Mr. Eisen’s proposal seems to call for a new category, but that can’t come from the FDA-that has to come from Congress. It can be done, the paper I’m working on editing is about the addition of dietary supplements as a defined category in the statutory firmament (and how that was a terrible decision), but it can’t be done by the FDA.

    23andMe’s tests claim to diagnose the potential for getting diseases, and under the FDCA, that makes it a medical device. As a new medical device, one that can’t follow the 510(k) process by being equivalent to a previous device, it requires premarket approval which in turn requires going through a scientific review of its safety and efficacy. Given the FDA’s claim that 23andMe stopped responding to the FDA and has not demonstrated safety/efficacy, of course the FDA is going to order them to halt sales. I’m surprised they were selling it in the first place without FDA approval.

    • Jared
      Posted November 28, 2013 at 12:47 am | Permalink

      Isn’t it a pretty big leap to say potential for getting a disease is the same thing as diagnosing a disease? I’m pretty sure it is.

  17. CwazyBill
    Posted November 27, 2013 at 3:22 pm | Permalink

    IMO, the FDA should stay out of anything other than providing the best information possible to consumers. Labeling, including possible recommendations, and where to find out more, would also be helpful.

  18. Sperry
    Posted November 27, 2013 at 4:33 pm | Permalink

    I think it would behoove the author to “follow the money”. There is a significant amount of cash to be made by this type of genomic testing. Big Med and Big Pharma have a very, very, very large interest in this.

    I suspect if one looks into the lobbying of the FDA on this issue, one will find the incredibly deep pockets of BM and BP having a notable influence on the FDA’s actions. They do not like the little guys on their turf, scooping up the dollars they think they have a right to.

    Just sayin’, follow the (lobbying) money.

  19. Posted November 27, 2013 at 5:16 pm | Permalink

    Nice post mike.

    Do companies that calculate disease risk based on life style and family history exist? If so how does the FDA view such companies?

    I could for example setup a company calculating a person’s relative risk of cancer based on weighting the cancer risk of the relatives appropriately by relatedness and heritability (along with making clear statements that this is not a diagnosis of having cancer, and merely informational). I could provide as evidence of this working the huge number of studies justifying my methodology, detailed workflow and test cases of my analysis. This seems very similar to what 23andme is doing (as you note). It seems like a company that merely advised you on your disease risk based on a set of genotypes you provided, would be exactly analogous to this family history approach. If the company doing the genotyping is doing it to medical standards (as I think 23and me are), it seems like this is also the correct analogy.

    • Posted November 28, 2013 at 10:02 am | Permalink

      I don’t think there are any companies that do this on a genome-wide basis, but I am pretty sure there are companies that do this for specific traits. I’ll poke around.

  20. David States
    Posted November 28, 2013 at 7:53 am | Permalink

    How accurate does the data need to be? Let’s assume for simplicity that there are 100 highly significant loci (CF, sickle cell, Huntington’s, etc.) in the genome and 23andMe gets to 1 million customers, then the company will be delivering 10e8 highly significant SNP calls. If the confidence for an individual SNP call is 99.99%, an error rate of 1e-4, then roughly 10,000 people will be given the wrong answer at an important loci with high confidence. This could lead people to make serious life changing decision (whether to have children, perform mastectomies, etc.). Is this an acceptable error rate?

    • Posted November 28, 2013 at 10:04 am | Permalink

      That’s not the right way to look at it. You have to establish error tolerances for specific tests. If 99.99% is acceptable for a single test, you can’t punish a company for simply doing more of them, because the alternative – that each of these tests is done separately but with a similar error rate – doesn’t help anyone.

  21. mtnrunner2
    Posted November 30, 2013 at 10:35 am | Permalink

    The efficacy of tests, or the knowledge level of the consumer, have absolutely no bearing on whether we have the right to prevent people from buying the tests. We simply haven’t the right to stop them; it is the individual’s life, and the individual’s decision, not ours or the government’s.

    That would be analogous to saying the we can’t have free speech until the ideas have been pre-approved by the government (after all, they do such a great job of that :\). And ideas are just as dangerous as misused medical technology; see Nazi Germany or anywhere it is advocated that the state owns us.

    In purely practical terms, the lack of widespread knowledge about these tests should be handled like we handle other areas of knowledge: by the testimony of experts.

    But in general, rights don’t guarantee results, and they are not contingent on them, and never should be.

    Ultimately the only reason this discussion even occurs today is because we don’t have constitutional protections on commerce like we do for speech. If we did, the notion of regulating commerce would be properly regarded as just as absurd as regulating thought.

  22. Robert
    Posted December 3, 2013 at 9:56 am | Permalink


    I agree with just about everything that you have written and do not wish to comment on the FDA. My rant is about the accuracy and value of the risk estimates that 23andme gives to clients.

    A small minority of users are likely to learn something useful about their carrier status for Mendelian disorders or BRCA status, etc., but most clients will be most interested in their risk for the major multifactorial disorders, heart disease, cancer, type II diabetes, etc. I don’t believe that the values 23andme provides are accurate or have any clinical usefulness. You alluded to the obvious, genetics is complicated. Risk values based upon SNPs won’t be meaningful. There are additional relevant SNPs left to be discovered and the risk values don’t include epistasis, rare variants, copy number variants, epigenetic status or somatic mutations and somatic recombination events to name just a few of the variables involved. This precludes any accuracy possible in risk assignments.

    The risk of suffering from one of these major diseases is high for all of us and tweaking these risks with genetic testing is of little use. What does it matter if my risk for diabetes is 12% or 35% assuming that the risk could be accurately obtained which it can’t?

    The “solution” for most of us is essentially the same and has been known for over 40 years…eat a moderate diet, get vigorous exercise, don’t smoke, watch your weight, get enough sleep, etc. Risk values from 23andme doesn’t change these recommendations.

    I am also an optimist that in time personal genomics with become more clinically useful. However the waiting period may be much longer than many currently anticipate and it doesn’t help to over-hype what is available today. What 23andme says on its website is not the same as what it says on TV ads and other presentations to the public.

  23. Salvation regulation
    Posted December 6, 2013 at 4:56 am | Permalink

10 Trackbacks