If you’re predicting disease, you should be regulated

Last week’s Congressional hearings on the direct-to-consumer genetics industry (featuring a provocative GAO report based on covertly recorded phone calls made to major DTC companies) have spurred plenty of discussion, including Daniel MacArthur’s post here at Genomes Unzipped and Dan Vorhaus’ post at Genomics Law Report.

But we disagree with some other Genomes Unzipped members about the regulatory future of the industry, and in particular we believe that medical interpretation of genetic data should be regulated.

We are both of the view that in general, it’s preferable to avoid government regulation because it imposes a cost on innovation and reduces market freedom. On the other hand, we accept government regulation over some things in the interest of public safety (e.g. drugs, automobiles), or law and order (e.g. guns, nuclear material). It’s one price we pay for the benefits of living in an orderly society.

In applying these principles to this question Matt Hurles found a good balance: light- or self-regulation on providing people with readouts of their genomes, but heavier regulation on selling a medical interpretation of those data. In other words, if you want your raw genotype data — all those As, Cs, Gs, and Ts, and nothing else — and you’re willing to pay for it, then you should be able to get it. The testing labs already have to pass CLIA certification to ensure they provide high quality data, and we believe all the Unzippers agree that regulation of this aspect of DTC genetics is appropriate.

The trickier question (where disagreement arises) is how to regulate the interpretation of those results, whether by a private company or a medical practitioner. Whereas the technical accuracy side of DTC genetics is carefully monitored, at the moment, the interpretations provided by DTC companies are sold in what is effectively a free market. For other commodities we would generally allow the market to find the right balance between the companies acting in the interest of the customer (provide them with only validated, useful information) and in their own interest (increase profits by providing a sexy story about customers’ genomes). Society doesn’t, however, allow the practice of medicine to operate in an open market because we place a special value on it, and therefore impose regulations to ensure a minimum standard of care.

The key question therefore is whether the medical interpretation of DTC genetic information should be held to that same standard. The GAO report, for all its flaws, has at least shown that DTC genetics companies, by providing inaccurate information about risks, are not currently meeting that interpretive standard. Furthermore, some are following unethical practices (e.g. encouraging people to send in someone else’s DNA without consent), which falls well short of the conduct expected from the medical profession. To us, at least, that’s sufficient grounds for heavier regulation of these aspects of the industry.

Finally, we agree with many comments that we should be careful not to immediately conflate regulation with over-regulation.  Nothing we’ve seen so far suggests that some form of sensible regulation can’t be implemented; let’s wait to see the proposed regulations before passing judgement.

  • Digg
  • StumbleUpon
  • del.icio.us
  • Facebook
  • Twitter
  • Google Bookmarks
  • FriendFeed
  • Reddit

49 Responses to “If you’re predicting disease, you should be regulated”


  • Daniel MacArthur

    Hey guys,

    I’m clearly towards the anti-regulation end of the spectrum among members of the group, but I think there’s more agreement than disagreement here. As a starting point, for instance, we all agree that people should have unfettered access to their raw genetic data.

    We also agree to a large extent on the need for consumer protection: I don’t believe (and have never argued) that DTC genetics companies should be entirely unregulated, but rather than we need to bear in mind that regulation comes at a cost – and that if that cost is too excessive, the consequences for the whole field of genetic medicine could be dire.

    I fully agree that some level of regulation is required here; and indeed, this is already the case. As you note, CLIA currently provides some level of protection on the technical accuracy front, but that could be strengthened and streamlined to better fit the fast-changing world of genomic technology. Similarly, there is already a body that would be well-suited to clamping down on the false claims of the bottom-feeders in the industry: the Federal Trade Commission. I’m all for loosing the FTC on the industry as a mechanism for weeding out the bogus claims and unethical marketing.

    However, I think we differ on the best mechanism for legislating on interpretation. To me, the best way forward is to increase transparency: by forcing companies to reveal the details and the scientific basis of their tests, a mandatory database of testing companies (preferably based on the NIH’s proposed Genetic Testing Registry) would provide consumers with the details needed to make an informed choice. The NIH and other academic bodies also need to play a role in engaging with companies to refine risk prediction algorithms, and to educate consumers and clinicians in the nuances of genetic data. Rather than the paternalistic approach – block consumers from accessing results considered to be too dubious – we instead need to empower them to make an informed decision.

    Am I being naive? Perhaps. But in the absence of compelling evidence indicating that (accurate) genetic test results harm participants, I think the default balance should be towards regulation that is as moderate as possible.

    Finally, there’s the issue of whether I’m being too pessimistic in my assessment of the shape of regulations to come. Perhaps that’s true – and certainly cooler heads than mine have suggested that. But the FDA has certainly not been making noises suggesting their goal is a compromise; instead, they actively colluded in the farcical Congress show trial last week, and told reporters that “they would set a high, if not impossible, bar for companies seeking to sell genetic tests directly to consumers“. Perhaps this is just grand-standing for the cameras, and more moderate views will emerge as the paperwork is drawn up, but I think I have reason to be skeptical.

  • Without any interpretation at all? But having sequence without any interpretation is useless. If something is to be sequenced then in some way it needs to be interpretated otherwise it’s a complete waste of time.

  • I think most of us are agreed that some form of regulation is a good idea, but that it should not be at the expense of completely stifling innovation or unnecessarily limiting individual choice.

    Clearly the interpretation is the difficult part, either to do or to regulate, and we already regulate the interpretation of medical tests through professional bodies that regulate services (such as the General Medical Council in the UK, which regulates doctors). The technical interpretation of a standard assay – say, whether the results of a blood cholesterol test fall outside of the ‘normal’ reference range – is regulated through professional laboratory standards, and the clinical interpretation is regulated through physician registration. This kind of ‘soft’ (i.e. non-legislative) regulation is great so long as you have a named professional who is responsible for the interpretation (and therefore professionally liable if they get it wrong). So this is certainly an option that the regulators and DTC companies could adopt.

    However, I am uncertain about the practicalities of regulating only ‘medical’ interpretation of a genome outside of the clinical context. What do we define as medically relevant in the genome? It would be possible to list some variants and diseases that are definitely medical – cystic fibrosis, BRCA mutations, etc – and perhaps a list of these validated, clinically actionable variants should be produced and require clinical advice to interpret. But where do we draw the line? What about a SNP that increases the risk of disease 1.5-fold? It is not (currently) clinically useful, but it is medically relevant. What about genes involved in food metabolism, or blood pressure, or body mass? And, for that matter, what about any recessive mutation that has no direct influence on that individual’s health – like a heterozygous carrier for cystic fibrosis, for example? Surely we shouldn’t go back to the “two-bucket theory of disease” – one bucket for genetics and one for everything else? This doesn’t really make sense! I can go to the chemists today and buy a DTC cholesterol testing kit (which is a medical device, not a service, and regulated as such…) and I am free to interpret (or misinterpret) it. Such a test could potentially diagnose an underlying genetic condition such as familial hypercholesterolaemia, so why would this be different from a genetic test (a hypothetical DTC FH genetic testing kit, say)?

  • Vincent Plagnol

    I find it hard to argue against the fact that a medical test should be regulated. But the matter is in the details: the line is not clear to me between a “readout” of the genome and a “medical interpretation” of these data. In your view (and going over your post it is not clear to me) is a 30% increase in diabetes risk a medical information? This is typically a risk increase that can be estimated from family history. This does not look like something one wants to regulate but this is nevertheless somewhat medically relevant. I think 23andMe and other companies have played on this grey zone to promote their products without the need for an official accreditation.

    And, as you wrote, how to implement regulation is a difficult question. The field of genetics is moving fast and government agencies are notoriously slow to put an official stamp on a medical procedure. If a “breaking news” story with important medical application is published in a scientific journal, a person may (rightfully) want to know the implications right away. There will probably be a demand for interpretation in this context that will be difficult to regulate, whatever this regulation may be. Scientists are already hardly capable of handling the deluge of genetic findings through peer review, things will only get worse if one expects the government to intervene.

    To summarize, any regulation that promotes a quality product sounds great but it will, in my opinion, be difficult to implement something sensible without a very heavy regulatory hand that would take most of the value out of the product.

  • I agree with Caroline: what constitutes “medical” and who makes that determination? What is “predicting disease?” Consider APOE: it plays a role in Alzheimer’s, cardiovascular disease and macular degeneration. In AD it is clearly important; in the other two, less so. Are we going to apply different regulatory standards to different predictions attached to the same alleles? I don’t see how.

    Now suppose it’s even more complicated: what if APOE were part of a chromosome-19 haplotype that was highly predictive or even “diagnostic” of a particular ancestry? Does 23andMe redact it? What about Oxford Ancestors and Family Tree DNA and every population genomics research endeavor?

    I think this will be like eating soup with a fork.

  • I’m on the “regulation by transparency” side. I hate the fact that there are companies out there making $millions/yr on dubious (“fraudulent2 according to GAO) products – and they are not making medical claims so would not be touched. Transparency might weed them out, via a mandatory GTR which would shame the FTC into taking action against companies that have been operating for years like this.

    I agree also with the difficult line to draw on where does medicine stop. We have OTC and prescription only medicine but even there something that is deemed to be prescription only all of a sudden becomes OTC (e.g. ranitidine) – this often coincides with the end of patent protection but that’s another story.

    Also the pragmatic argument is that if I can get me gene sequence data I will find many ex-USA companies willing to interpret it for me, there maybe one in Iceland where I could upload my data to. There will undoubtedly be many – some good and some bad. This tide is coming in and there is nothing we can do about it, at least the GTR route gives the consumer (which includes medical practitioners by the way, they can be mislead too)some sort of clarity.

    …standing on your head (see Misha above)

  • The very act of offering something for sale carries a tacit claim as to that product’s utility. That claim is an interpretation that even the most reputable companies engage in, and which is suspect even for the few genes with the most empirical support for their effects. For example, a customer has a SNP indicating a population-wide increased disease risk of 30 percent. But to date, no one knows the actual risk increase in any individual’s specific genomic context. Indeed, actual risk is already an abstraction based on the subject having an abstractly average genomic background. Actual risk may be vastly higher or lower due to epistatic genes at other loci whose effect is yet cryptic. Selling the data with the tacit claim that it is usefully informative about *that disease* in *that customer* is false advertising, notwithstanding claims that the practice is only recreational or educational or empowering in some very subjective sense. The bottom-feeders differ from the good guys more in the rhetorical baldness of their claims than they do in the implied logic behind the offering.

  • One thing I’ve been wondering is that how the regulation might affect international companies, e.g. if risk prediction regulation makes it very hard, complex or expensive for US companies to do, is there a possibility that consumers would buy the sequencing/genotyping service from US company, and get the interpretation from an international company through their website (this company of course possibly being owned by the US company)? Well, I assume that the regulation would also make it hard for international companies to advertise their services in the US, but I think that there certainly are some loop-hole possibilities..

  • I’m not seeing anything about the security/privacy of the data in this discussion.

    Maybe none of you have had to get insurance in the US free market. Or maybe y’all have so much money it doesn’t matter to you. It sucks. If I had some allele that suggested diabetes or breast cancer or whatever, and I answer “no pre-existing conditions” on the insurance form–can then drop me later? What if they found out on a discussion group that I talked about that allele and knew it? I wouldn’t discuss my alleles with anyone at this point. And that’s part of the reason I haven’t done any of these.

    Someone I know also asked me this: if I had something that indicated bipolar disorder, could a landlord refuse to rent to me because I’m a risk? I couldn’t assure them they were safe.

    I am very pleased we have GINA legislation, but it hasn’t been tested yet.

  • A jaded but perhaps practical approach to government regulation of DTC

    Once a few entities lock-up the reimbursement path for fed dollars related to providing medically relevant genetic information …. won’t the issue US government regulation of the sector will go away? I think so.

    What we watched with GAO on July 22, 2010 in Washington is a battle over who gets to make money in this sector – proof the personal genetic data market is finally real.

    Here are the questions that we need government and others help answering:

    What genetic testing info would a health insurance provider (i.e. the US Federal Government) be required to offer reimbursement for due to the test having sufficient validity/utility?

    What are the hoops that entities go through to flag them as being a legitimate provider of the above services? (I hope the criteria are based on transparency of the genetic markers in question – not capital barriers that promote a monopoly).

    Everything else will probably not get regulated in due time but be subject to the usual silly lawsuits that any entity working with consumers in the US are subject to.

    Where are the Republicans in Washington that want the open marketplace to self-regulate when you need them?

  • I’d like to see the FDA shift focus to protecting consumers by enforcing access to information predicting disease.

  • I’d argue that much of the potential for misuse of the DTC test data lies in the a priori expectations of those who seek it out. These expectations often reflect the one-gene-one-protein/Mendelian ball-and-chain that we are still stuck with (no offense to Mendel, by all accounts he was a cool dude) and which our public and medical education systems continue to perpetuate. Many of us are culturally conditioned to view the primary value of genetic data as predictive/deterministic/causal, expecting it to produce a nice discrete variable that we can plug into our unique, linear narratives about health risk across the lifespan. In a genomic context, this is obviously not a good paradigm a) for the reasons Misha highlights and b) for the reasons that Stephen highlights. I do think that Stephen’s point is particularly salient, if inconvenient, for us to think about. As a genetic counseling student, time and time again I have informed women over 35 about their population-based fetal aneuploidy risks and have felt downright guilty about using this data to come up with a hypothetical abstraction of the risk to their current pregnancy. Sure, I realize that it’s all we have, and in an American prenatal setting, patients often seem to really value predictive information based on large cohorts of pregnant women. I have, however, caught several practitioners forgetting the epistemological basis of their prognostications. So, on a large scale, rampantly using the “population-data-to-infer-individual-risk” thing without much oversight does worry me a little.

    What I personally find exciting about this data is not its predictive value but what we can learn from it about disease pathology. However, that is not a perspective that I can reasonably expect everyone to share.

    Mandatory use of the test registry seems the most feasible and fair means of quality control for the time being, and I think that this kind of transparency should extend to all labs, not just the DTC companies. I do wonder how this sort of blanket transparency would affect market structure over the long run?

  • “The GAO report, for all its flaws, has at least shown that DTC genetics companies, by providing inaccurate information about risks”

    The report did not show that. It showed that the DTC genetics companies are precariously balanced at the cutting edge of science, exactly like medical professionals in assessing risks. If genetic scientists have provided a clear, straightforward answer to a particular question e.g. if you have gene X, you will have disease Y, then both DTC testing companies and their customers and medical professionals ordering a genetic test for a patient know exactly what the result means. That’s the easy stuff.

    An example might be a couple wanting to find out if they are both carriers for a disease like sickle-cell anaemia. If they are both heterozygotes for the relevant allele, then the risk is a simple matter of Mendlian calculation – on average, we would expect one in four of their offspring to be unfortunate homozygotes with the lethal anaemia.

    But things get more complicated when numerous genes could be involved, not all of which have yet been identified, and/or environmental or lifestyle hazards play a part. Heart disease and cancer are good examples.

    Each DTC company has its own method of assessing risk in these cases. I can only speak for 23andMe and deCODEme. Both are transparent in their systems. The crucial factor leaps out at any intelligent person. 23andMe will only take a finding into consideration if it has been replicated. So they tend to include fewer studies in their risk assessments than deCODEme. That caution may be good practice, but the result I find is that deCODEme’s risk assessments overall better fit my own and my family’s medical history, except in the one case where 23andMe was including more studies.

    Science does not stand still. New studies are being published all the time, which will gradually make all genetic testing more useful and accurate, whether DTC or not. That is why both 23andMe and deCODEme provide regular updates online. Any good company should do that. They should also make clear the instances where other genetic factors are known for a disease, which they do not test for. 23andMe does that for example with breast cancer.

    What I should like to see is light regulation to ensure that other companies measure up to the highly professional standards set by these two.

  • Daniel MacArthur

    Stephen:
    For example, a customer has a SNP indicating a population-wide increased disease risk of 30 percent. But to date, no one knows the actual risk increase in any individual’s specific genomic context. Indeed, actual risk is already an abstraction based on the subject having an abstractly average genomic background. Actual risk may be vastly higher or lower due to epistatic genes at other loci whose effect is yet cryptic. Selling the data with the tacit claim that it is usefully informative about *that disease* in *that customer* is false advertising, notwithstanding claims that the practice is only recreational or educational or empowering in some very subjective sense. The bottom-feeders differ from the good guys more in the rhetorical baldness of their claims than they do in the implied logic behind the offering.

    Oh, come on. All risk factors are contingent on individual modifiers; following this rationale, doctors shouldn’t tell people their blood lipid levels are worryingly high (after all, they might carry a rare genetic variant that protects them from excess lipids). The argument is rendered even more spurious in this case by the fact that there’s currently zero convincing evidence for widespread epistasis between common risk variants, despite people looking very hard for it.

    Gene-environment interactions are potentially more interesting; but can companies be blamed for not taking these into account when they’re not actually known yet?

    There are some reasonable arguments against the risk models used by personal genomics companies, but this isn’t one of them.

  • “If you’re predicting disease, you should be regulated” – by that metric the FDA should regulate my interviews with my biological grandparents on my own family’s medical history. Since, in most instances that is a far better disease risk predictor than a SNPs.

  • @ Daniel MacArthur
    >The argument is rendered even more spurious in this case by the fact that there’s currently zero convincing evidence for widespread epistasis between common risk variants, despite people looking very hard for it.

    Since common risk variants of strong effect are so hard to find generally, it’s not surprising that interactions between them would not be apparent either. Any statement that a trait is polygenic is admission of cryptic epistasis. Indeed, the lack of statistical power is partly *due* to epistasis, as well as such things as whether common SNPs accurately tag functional SNPs, etc.

    And, of course, a variant that protects against high lipid levels need not necessarily be rare. Is there any DNA marker for a complex trait yet that has the predictive power of high blood lipids? Isn’t blood lipid level a useful surrogate for the combined effect of lots of epistatic genes? This becomes more of an argument for picking highly predictive evidence of whatever kind, which to date will usually be a matter of family history, biomarkers, etc., not single SNPs.

  • @Dave

    The relevant analogy here would be a company that takes your family tree and all known diseases in it, and produces a personalised health report with risks for various diseases, should be regulated, to ensure that they are doing it sensibly.

    @Stephen

    It is not just that we haven’t found specific gene-gene interactions, we also know that including gene-gene interactions in predictive models don’t significantly increase our predictive power (i.e., for prediction, pointwise effects are far more important than gene-gene interactions). See, for exaple, David Clayton’s work on diabetes.

  • Why do we need centralized regulation for interpreting genomes? Shouldn’t there be competition to come up with the best methods? Shouldn’t patients and doctors shop around for the best service and make informed individual decisions?

    “Nothing we’ve seen so far suggests that some form of sensible regulation can’t be implemented”

    Regulators aren’t angels, they are humans. Their incentives are to minimize the risk of being negatively portrayed in the media for letting something slip through, which could cost them their jobs or future career advancement. But they don’t feel to the costs of over-regulation — costs that fall mostly on patients who don’t have access to innovative technologies. Regulation is a ratchet that ordinarily only turns in one direction. Think about your last trip to the airport. That’s what regulations of genomics is going to be like — not at all sensible.

    Getting specific, look at the Administrative Procedure Act (http://en.wikipedia.org/wiki/Administrative_Procedure_Act) — ostensibly the law that regulates how new “sensible” regulations come into existence. Now contrast the requirements of that Act to FDA’s actions regarding DTC and LDTs. Where is the formal rulemaking process? If you want to lobby for something, lobby for FDA to comply with federal law and use an APA-compliant rulemaking process. They should be publishing new regulations in the CFR, not the WSJ.

  • @Luke Jostins

    I agree with that analogy; however is 23andMe (as an example) offering an interpretation of the data or a concise summary of peer reviewed publicly accessible data? Given a rs# and a genotype I could easily find the same exact information 23andMe provides to their customers.

  • Think about your last trip to the airport. That’s what regulations of genomics is going to be like — not at all sensible.

    That needs to be put in 128 point font and blazed from the rooftops.

    Regulation is NOT peer review. Has anyone here ever had a bad peer reviewer? You take for granted your right to criticize that reviewer in your rebuttal, should they make a flagrant mistake. And if push comes to shove you can simply resubmit to another journal.

    Now realize that with the FDA there is no possibility of criticism and no hope for resubmission. Moreover, a technical error is punished as a criminal matter, involving fines, forfeitures, and jail time.

    You’d really have to be bloody daft to attempt to take genetic discoveries to market in the United States in this environment.

    http://www.fdanews.com/newsletter/article?issueId=13891&articleId=128911

    FDA Says Medtronic Addressing Concerns, Sends Letter Anyway

    In what some devicemakers are referring to as the “new FDA way,” Medtronic has received a warning letter even though the company has adequately addressed the agency’s concerns. The FDA acknowledges receiving several responses from Medtronic to a Feb. 4 Form 483. “You appear to be addressing our concerns,” the agency says in the warning letter. Medtronic provided responses to all the observations, company spokesman Brian Henry said. No date has been set for a follow-up inspection.

  • There is a worrying amount of vagueness to this discussion and it would benefit greatly from concrete facts.

    FACT: The standard fee for a PMA submission is $217,787

    http://www.fda.gov/MedicalDevices/DeviceRegulationandGuidance/HowtoMarketYourDevice/PremarketSubmissions/PremarketApprovalPMA/UCM048161#fees

    FACT: Every revision of a device under PMA triggers a panel-track fee of $163,340

    http://www.fda.gov/medicaldevices/deviceregulationandguidance/guidancedocuments/ucm089726.htm#_Toc33587904

    Therefore, while the agency continues to receive and review stakeholder comments on this issue, applicants should submit a panel-track supplement and pay the associated fee for:

    a new indication for use (i.e., patient population/disease state); or
    a change in device design or performance that could significantly affect clinical outcome.
    For the types of changes listed above, a new clinical trial is generally necessary to demonstrate reasonable assurance of safety and effectiveness.

    FACT: Jeffrey Shuren wants DTC to be regulated as PMA:

    http://online.wsj.com/article/BT-CO-20100716-711161.html

    Top Food and Drug Administration officials suggested they would set a high, if not impossible, bar for companies seeking to sell genetic tests directly to consumers.

    In an interview, Jeffrey Shuren, head of FDA’s medical-devices division, and Alberto Gutierrez, who oversees the FDA unit that regulates certain laboratory tests, said they considered direct-to-consumer genetic tests “high risk” [aka PMA] and have already informed companies selling such tests through their websites that they must seek FDA approval.

    Although the FDA hasn’t ordered existing tests off the market, Shuren suggested it could be difficult to gain agency approval.

    FACT: PMA or “high-risk” is the same level of regulation required for pacemakers

    http://www.fda.gov/medicaldevices/deviceregulationandguidance/howtomarketyourdevice/premarketsubmissions/premarketapprovalpma/default.htm

    If the device is a high risk device (supports or sustains human life, is of substantial importance in preventing impairment of human health, or presents a potential, unreasonable risk of illness or injury) and has been found to be not substantially equivalent (NSE) to a Class I, II, or III [Class III requiring 510(k)] device, then the device must have an approved PMA before marketing in the U.S.

    FACT: The fees for PMA are actually dwarfed by the costs of clinical studies and the approval process:

    http://www.bloomberg.com/news/2010-06-23/cytori-breast-repair-device-gets-higher-hurdle-from-fda-as-rules-pondered.html

    The cost of getting a device cleared through the current fast-track [510(k)] process can range from less than $1 million to as much as $50 million, compared with $50 million to $150 million under the rules for higher-risk [PMA] devices , said Alexander of Alquest.

    FACT: 23andMe is just the tip of the iceberg. Every clinical lab in the United States, including “noncombatants” at the Mayo Clinic and University of Wisconsin that had nothing to do with DTC, is now under the gun for multimillion dollar FDA costs, amounting to a massive unfunded mandate:

    http://www.genomeweb.com/blog/reimbursement-questions-dawn-ldt-reg-changes-loom

    So my question is: If labs are suddenly regulated as medical device makers, and will be required to file 510(k) or PMA approval submissions, how can they expect a return on what will essentially be an unfunded mandate?

    More to the point, what steps can clinical labs take to ensure that reimbursement policies keep pace with the expected burden of additional regulatory compliance costs? The short answer is not much, at least for now.

    …However FDA’s new LDT regs shake out, almost every clinical lab will be on the hook, and many will stop offering certain tests — almost certainly the most clinically important, and costly ones.

    A reasonable person might believe that these actions do not amount to “sensible” regulation. Where is the money for FDA compliance going to come from in such a tough economy? How are the small labs going to survive?

    The answer is simple: they will not. A wave of bankruptcies and shutterings will commence as hospitals find it infeasible to play the FDA game, which requires dedicated lobbyists and “regulatory affairs” specialists. The result will be the centralization of laboratory services in the hands of Labcorp and Quest. It’s not like it will be easy for LC and Quest to survive this assault, at a cost of $1-5 million dollars per FDA cleared test, but at least they have some chance of making it through.

    As for your small hospital lab or startup? Not a chance.

    Who benefits from this? Big business and big government. Large companies like Genentech regulate their competitors out of existence and the FDA carries their water all the way along. Witness Jeffrey Shuren of CDRH’s statements during the hearings, where he carries water for the “traditional IVD manufacturers”, multibillion dollar companies who have in his mind somehow been scared away from genomics by tiny startups.

    Big government also profits. For every startup reading this thread, I urge you to go and hire someone from the FDAAA (the FDA Alumni Organization, http://fdaaa.org). Hiring a former FDA official will allow you to navigate the bureaucracy, will give you a heads up on incoming regulation, and is often the only way to dodge one of the FDA’s arbitrary crackdowns. The bureaucrats always look well on it when you are enriching one of their fellows, because they know that they can take some time at the trough in a few years. It’s called the “revolving door” or “public/private treadmill” for a reason, folks.

    Still think this regulation is “sensible”?

  • Earlier comments brought up the prospect of driving DTC genomics companies serving the U.S. market outside of the U.S., and thus beyond the practical reach of American law.

    Or, least the disease-risk-interpretation parts of these enterprises.

    It seems to me that the online gambling industry could provide a pretty good analogy for this scenario.

    Online gambling was unknown, experienced explosive growth, attracted press exposure, came under regulatory scrutiny, and was found to be illegal as practiced. Gamblng websites then closed, found offshore ISPs, and re-opened.

    I don’t know enough about the state of online gambling today to offer an opinion as to whether this is a cautionary tale or a salutory one. Anybody?

  • @reality I couldn’t agree more that this debate should be based on facts being presented on both sides of the argument. So I present one which you neglected to mention:

    FACT: the $217,787 fee for the first PMA submission from a company making < $30 million in sales or receipts is waived.

    I leave it as an exercise for the reader to find a DTC genetics company with sales amounting to greater than 30 million dollars.

  • Not all of the DTC companies are in the US. deCODe are in Iceland. There is also another company called GenePlanet:

    http://www.geneplanet.com

    They are based in Ireland but their research centre is in Slovenia. Does anyone have any experience of using this company?

  • A great discussion, as always. A few quick points:

    1) Wait and Talk. On the topic of “don’t knock it until we see it,” while I think Jeff and Kate’s point is well-taken, I would also suggest that it is incumbent upon the regulated community – which includes companies, investors, customers and even clinicians – to speak up about what does and does not make sense when it comes to a regulatory solution. The FDA’s public conversation was part of that process, but that was (a) not directed specifically at the regulation of DTC testing and (b) should mark the beginning, not the end, of the dialogue. There are still a number of options on the table, and not all of them are equal. Waiting passively for regulators to act may not be the best strategy.

    2) Line-Drawing. It’s clear that whatever regulatory system emerges – and this applies to the much broader category of LDTs, as well as to DTCs – the issue that Caroline, Misha and so many others have identified is going to be one of, if not the, most difficult to solve: what counts as a medical interpretation worthy of heightened regulatory scrutiny?

    Regulating all genomic interpretations is incompatible with providing broad access to genomic information, since only a handful of people can interpret the raw data on their own (and I’m certainly not one of them). But determining where to draw that line, particularly given the interconnectedness of gene-trait associations and the rapid rate of change in the science (both discussed above), is going to be fiendishly difficult. The key will be to come up with some standard-setting body that strikes the proper balance between involving stakeholders (this is an issue that should not be determined by the FDA – or any other agency – acting on its own) and minimizing the number of voices so that the body is nimble enough to respond to changes in the science. I’m not sure there are any perfect solutions here.

    2a) Useful to Whom?. A sub-point here is whether that line-drawing incorporates (i) analytical validity, (ii) clinical validity and/or (iii) clinical utility. I think that there is widespread agreement on (i), a general agreement that (ii) is important, although no clear agreement on how to assess it, and very little agreement about whether (iii) is appropriate. The issue clinical utility gets at one of the core tensions here: is my genomic information useful when I say it is useful, or only when my doctor (or some standard-setting body) says it is useful?

    3) What is special about DTC? When it comes to the regulation of DTC genetic tests – as a subset of all other genetic tests – I think we (and regulators) need to be much clearer than at present in explaining where, exactly, the concern lies. Is it that individuals are being misled by untrustworthy businesses? If so, then the FTC (or possibly the FDA acting in a similar role; the GPMA would have the two agencies work together) has a clear role to play. Is it that genetic tests are not appropriately evaluated beyond analytical validity? If so, then it’s not clear that DTC tests should be regulated differently than the broader category of LDTs. Or is that individuals cannot be trusted to appropriately handle certain types of genetic information and interpretations? If so, then it may be that certain types of information simply cannot be presented directly to consumers, although then we return immediately to the line-drawing problem discussed above.

    It is insufficient to know that the FDA intends to regulate DTC genetic tests (as they clearly do); what is needed is a clearer explanation as to whether, why and how DTC genetic tests are to be regulated differently from other types of tests.

    4) Transparency First. Finally, I can only echo Daniel’s comments above that, no matter what, there is a need for greater transparency. By this I think we both mean a more proactive approach to data collection (i.e., what genetic tests are available, how they are marketed, how they are used by individuals and what impact – positive or negative – they have on end users, both with respect to and apart from influencing medical decisions) and greater regulatory openness. Without such transparency it is difficult to see how any government entity can expect to craft a regulatory framework that strikes the proper balance between the protection of the public and the preservation of innovation and long-term growth in the area of genetic testing and personalized medicine.

    - Dan

  • Jeff Barrett:

    That would seem good, wouldn’t it? Seems so reasonable. But the devil is in the details.

    http://fda.gov/downloads/CombinationProducts/MeetingsConferencesWorkshops/UCM116739.pdf

    The FDA encourages applicants who are
    uncertain as to whether single or separate applications should be submitted for a
    combination product to discuss the issue with the lead reviewing Division

    It is not at all obvious that the FDA is going to consider a multivalued output test based on the human genome to be a “single application”. The human genome is the mother of all combination products!

    Moreover, it should be readily apparent that if every new disease risk prediction triggers a $163000 fee and multimillion dollar clinical trial, that there is a major disincentive to updating a product in response to new information.

    Finally and most importantly, the cost of the fees is just the beginning. The full PMA application process takes years and costs $50-150 million, as the citation above shows. Do not ask me, go and ask a venture capitalist with some experience in this space. Ask the people involved with Solexa who are now at Illumina. Ask the guys at Oxford Nanopore, or better yet ask their lawyers.

    Ask anyone who’s taken a device through the FDA’s valley of death.

    But don’t ask the FDA. For obvious reasons they shy away from any discussion of the actual explicit costs of regulation. Go back and look, Shuren dodged that question maybe three times in the government webcast.

  • @Dan Vorhaus

    This IS the crux of the argument here. I have been playing with this one for 3 years now since I started posting on these companies.

    “What is medical testing/diagnosis?”

    Is pre-disease the new disease?

    Pre-Diabetes. Pre-Hypertension. These exist in the medical lexicon and will be considered diagnosis in the ICD-10

    We use diagnoses to guide treatment and to cajole payors into paying for services and medications.

    Seriously. If Medicare doesn’t have the right diagnosis, they will not pay for certain lab tests.

    What is a medical diagnosis? I as a board certified physician cannot say with certainty. “What exactly is a medical diagnosis”

    I can say what is not. Things which have zero influence over health are not medical diagnoses. What might those be. Well, certainly asparagus pee sniffer is in that category.

    The FDA will regulate. As I said, DTCG sucked LDTs into this pit. Most Notably after 23andME went off the reservation and offered BRCA testing. It only took 1 year for the FDA to come up with a strategy to evaluate this.

    BRCA mutations ARE a diagnosis.
    PGX testing is a medical test.

    It is funny most commenters accept that now. A week ago, most would call me crazy, paternalistic and a luddite. Now we are debating the predisposition testing, which means that some for some reason, reason has filled the “Oppostion to regs” camp.

    Yes the FDA PMA will kill DTCG, but not some labs that do LDT, Myriad, LabCorp, Quest, ClinicalDx. But yes, mom and pop labs will get crushed.

    I think the FDA will create an exemption for orphan tests, which DTCG will not fall into.

    This will keep the clinically valid testing in play.

    But now we have to ask, what IS clinically valid testing?
    Hence the problem here. I think it will require a panel of clinicians and scientists to truly come up with what is and what is not. We all will likely disagree on 1/3, agree on 1/3 and be undecided on 1/3.

    Hence, uncertainty. Thus no investment in the next round of this DTCG fad. Which means, the snake oil and respectable companies will have to become international to avoid regs. Which is a shame, because regs are what is needed to assure quality, since it is obvious to me that the companies don’t want to or haven’t had the drive to self regulate.

  • Katherine Morley

    I agree with Dan, and everyone else who’s raised this issue, that drawing the line between “medical” and “non-medical” will be one of the most difficult aspects of any regulation of the industry. Certainly I don’t have any easy solution to the problem. But as Keith mentioned, access to some other types of health care products is regulated in a variable way (prescription-only versus OTC drugs, or OTC diagnostic tests versus those requiring a visit to the doctor). I would rather see some line drawn (hopefully an informed rather than arbitrary one, and not necessarily by the FDA) than a blanket ban on DTC genetics.

  • @Katherine,

    There will not likely be a ban. But, given the nature of medical information, it is likely to be a High Risk device according to the FDA. This will incur quite the cost and test the mettle of DTCG investors. Who won’t reap investment rewards for another decade.

  • @Debbie Kennett

    Re: http://www.geneplanet.com
    They are based in Ireland but their research centre is in Slovenia. Does anyone have any experience of using this company?

    I don’t have any experience but I did notice them a few months ago. The first question on their FAQ is:

    Why would I want these service offered by GenePlanet?

    A good question, why indeed? Cost is $549, there is a very limited demo on the site, could not find info about how they calculate risk etc, overall quite opaque. So no good reason to prefer them over 23andMe, I just hope that the FDA do not create the reason…

    Medical vs. Non-medical – again… will never be resolved and it is not the question really. It’s all medical (or at least something to do with healthcare, as long as we’re not dealing with ancestry etc) but maybe some should be delivered through a practitioner and some not (I prefer to leave all up to us to choose but that doesn’t matter).

    I can read in a newspaper that BMI > 30 is obesity and that it increases risk of heart disease, diabetes, etc. The “is obesity” bit is a fact, the “increases risk” bit is an interpretation, it is “predicting disease” as in the title of this post. It’s just the same as attributing a risk to a gene variant, it’s based on population research and is applied to individuals (as with Cholesterol and other biomarker tests as mentioned by Daniel above). If I had a BMI>30 I would accept that I would be at higher risk and also that it may not actually happen to me, and further that I could do something about it. The fact that this particular disease prediction is widely accepted makes no difference, the interpretation is not regulated, nor could it be.

    If I read in a newspaper that a new research study reported that people with a BMI of 22 who live near Naples in Italy have a 10-fold increased risk of stroke then that would be news to me and I would want to find out more, I might go to my doctor (actually I wouldn’t but others may) or I might look into the source – whatever I want to do I will be able to because it will be accessible in one form or another. Companies like 23andMe make this easier for genetics than it has ever been before. We know that they are trustworthy and competent (as are the other main companies), we don’t know much about Gene Planet, but on the other hand people in the Slovenia area may feel more comfortable using their services over those from USA or Iceland.

    What’s more important, regulation or transparency? My answer would be the latter but it does not seem to be that regulation will guarantee it. Strong FDA style regulation may damage the DTC companies which would be a high cost because they have been forced to be more or less totally transparent either through their own desire to be, pressure from consumers, commentators and critics, or a combination. Regulating and forcing some parts to go through practitioners may well reduce the need to be so transparent and we will all be worse off.

    If and when some harm is demonstrated then the questions can be asked, DTC has been on the market now for long enough to have been thoroughly monitored for harm, none has been reported (unlike for aspirin or paracetamol). No reports of people panicking, jumping off bridges etc, but many reports of benefits, including the “fantasy” of increased motivation for improving diet/lifestyle

  • Katherine Morley

    @ Steven Murphy

    To clarify: I agree that the likelihood of a direct ban is extremely low, but others have suggested that regulatory standards may end up being set so high as to effectively result in this (see Daniel’s first comment in this thread). I would rather see regulation applied to some subset of tests than have this occur.

  • Daniel MacArthur

    Dan’s comment is the most useful thing I’ve read so far on this issue – indeed, I think it warrants its own post (stay tuned).

    I have only one (unfortunately non-substantive) thing to add to Dan’s points:

    Finally, I can only echo Daniel’s comments above that, no matter what, there is a need for greater transparency.

    Since the transparency argument was first made in an eloquent post by Dan, and I’ve since shamelessly borrowed it, you just heard Dan echoing Daniel echoing Dan. We’re in dangerous metaphysical territory, Dan…

  • @AMac’s comment about the regulation of gambling as a lesson for genetic testing, this article in today’s New York Times caught my eye. I’m not familiar with the details of internet gambling regulation, but I would not be surprised to see something similar happen in the case of genetic testing over time, particularly if (a) the initial regulatory approach is relatively restrictive and (b) genetic information becomes as widespread as has been predicted.

    @Daniel: is there metaphysical territory of any other description?

  • Collin Burton

    Myriad Genetics, the company that tests the BRCA genes, isn’t regulated by the FDA and instead is regulated by the college of american pathologists (CAP) and the New York State department of health. Perhaps DTC genetics companies should be regulated by them?

    The very fact that these companies can’t agree on the correct result for the same sample, and in some cases even come up with different results within their own labs, means they need someone from the outside watching over them.

    It will be great to see the FDA start regulating some of these “nutritional supplement” companies that sell snake oil products. Those guys should be regulated out of business.

  • Hope the FDA are reading this column..

    While all above points are well taken, I am directly in sync with Daniel, Dan, and @reality regarding their staunch warnings about the need for continued vigilance on this issue. I can’t help but think that federal agencies, if left unchecked by serious criticism and continued challenge, will annihilate consumer genomics.

  • @Collin, look at the Dietary Supplement Health and Education Act of 1994. FDA can’t do a think about supplements.

  • Sorry for the misspelling, I meant thing.

  • Collin Burton

    @k They may not be able to regulate supplements, but they can regulate the bogus genetic tests some of these companies are using to help sell their supplements, for example genewize. I would hate to see legitimate DTC companies have their image sullied by such blatant pseudoscience. If regulators need to come in to keep DTC genetic tests out of the nutrtional supplements market then please bring them in.

  • @Robert West

    absolute vigilance – I second that

  • Re vigilance, just to add, the blogs are quite balanced but the press in general are just (lazily?) pushing the GAO report uncritically. Have not seen any balanced press yet, except Mark Hughes in The Times (UK)

    [I guess you mean Mark Henderson? DM]

  • Keith makes a really good point. How to appropriately influence the press?? We need some here..

  • Criticism is improtant fellas and ladies, but don’t you think the DTC companies could have done more than just publish a “White Paper” about standards and regs? Couldn’t they have pointed out each other’s shortcomings and worked together?

    The whole business adage of “All ships rise with a rising tide” certainly rings true here. In fact I heard it about a million times when my business buddies tried to get me to drop the blog as it was “bad for business”

    It seems to me that the answers to these debates are:

    1. Regulation is coming from the FDA not CAP, maybe even for Myriad.

    2. The self-regulation thing only works when the self has teeth, think medical licensing boards here…Not country club buddies.

    3. Transparency only works if the public can even understand what the hell you are saying. less than 35% of the US public is health literate let alone genomics literate

    4. No one is singling out DTCG, LDT is involved here too. Annihilation will occur because
    A. the chain of trust is broken, thanks YouTube
    B. There is no market
    C. Now the expense to operate will become too painful for any supposed hockey stick projection on a napkin

    5. The model for regulation is already in place. Amplichip. Speaking up is nice, but there is a model in place. Maybe not for predisposition, but I would say we can find some there as well.

    6. The press is on a wave of anger after PR people made them look stupid covering DTCG uncritically. Now they are going after DTCG uncritically. Nuance sells philosophy textbooks.

    The only debate left here is

    “What is medical and what is not?”

    ICD9? ICD10? Medical Scoieties? IOM? How will we really decide that one?

    One thing is for certain, it will not be decided by Silicon Valley or in a Genomics Lab.

    -Steve

  • @ Steve Murphy

    Re: The only debate left here is “What is medical and what is not?” ICD9? ICD10? Medical Scoieties? IOM
    How will we really decide that one?
    One thing is for certain, it will not be decided by Silicon Valley or in a Genomics Lab.

    Well, that is what most of the rest of us following this post are afraid of- “What is medical and what is not” involves a continuum of possibilities that is no more easily discerned by any of our current medical establishments, let alone by the FDA, than by those of us in Silicon Valley or in a Genomics Lab.

    BTW, can I borrow your philosophy text?

    Bob

  • Apologies to Mark HENDERSON, of course I did not mean to get you mixed up with the new Fulham manager, a terrible mistake!!!

    BTW – congrats to Genome Unzipped so far, the articles and comments have been outstanding

  • I agree with Steve Murphy about self regulation – it CAN work, but the companies have been dragging there feet, they have been criticised here and elsewhere for that. It’s about 2 years ago, or more, that the HGC in the UK began trying to get them together on this (they all met up in London in 2008). Now it’s time to get a move on, it would be enough to begin with for 23andMe, Navi, Pathway, and deCODE to set something up and a good start would be agreeing on the risk standards. Set something up that’s serious and make it easy for the FDA to back down a bit. Do the FDA really want to get into the whole mess of trying to regulate not just genomics but the whole LDT scene?

    Is the Personalised Medicine Coalition doing anything? I read in a December 2007 letter the HHS (bit.ly/bYmpDY pdf) that “PMC is ready to convene stakeholders who could propose such [voluntary] standards to educate consumers and guide the field”. What happened there? Activity on the PMC regulatory oversight section of their website seems to stop at 2007 (bit.ly/dzxQbn)

  • @Keith,
    Thanks.

    @Bob,
    It is by Kant….

  • Each direct-to-citizen genetic testing company uses a different gene-chip.

    In the case of Mendelian disorders, if a specific high-risk SNP is analyzed by different companies, then these different companies should report a similar, high risk of showing symptoms of the disorder. If there are SNPs with a similar high risk for the disorder, which are, however, not reported back to the citizen, then the report will state “selected”.

    The situation is different for multi-factorial diseases: each company will only test for some of the many SNPs known to have a, often weak, disease-association. Thus, the corresponding disease-risk predictions reported by different companies will naturally be non-identical.

    I see no need for a regulatory agency to get involved.

    The direct-to-citizen genetic testing companies only provide us with published, public knowledge (“according to the study published in … base … at SNP … is associated with a … higher/lower risk for disease …”). Many of the genome-disease-association studies, both for Mendelian disorders and multi-factorial diseases, were funded by the USA government itself via the NIH (National Institutes of Health). So one arm (FDA) of the government should not prevent the private sector from informing tax-paying citizens what discoveries were achieved using their taxes by another arm (NIH) of the government.

  • @Dirk
    “Direct To Citizen” is a misnomer bub. Clearly people who aren’t citizens of the US purchase these tests. Further, not all citizens pay taxes. So your argument has no legs.

  • It works the same way if you call something natural a cure. Look what happened to Kevin Trudeau.

Comments are currently closed.

Page optimized by WP Minify WordPress Plugin