FDA to regulate genetic tests as medical devices

It looks as though the FDA is swooping down on the direct-to-consumer genetic testing industry in a serious way, sending formal letters to five companies informing them that their tests will be regulated as medical devices:

WASHINGTON -- The Food and Drug Administration is issuing regulatory letters to five genetic test makers, the first sign that the government is cracking down on companies that claim to use DNA samples to predict inheritable diseases.

The FDA letters notify each company that their tests are considered medical devices and therefore must be federally approved as safe and effective. None of the companies have submitted their products for approval, according to the FDA.

The letters posted online deal with specific tests marketed by: 23andMe Inc., deCODE Genetics, Illumina, Navigenics and Knome Inc.

FDA asks each of the companies to contact the agency to make arrangements for submitting their tests for review.

Depending on the details, this may be disastrous for the nascent personal genomics industry and its customers - excessive regulation would negatively impact on innovation in the field by increasing the barrier to entry for new products, as well as increasing costs for consumers.
It's also worth noting that there's no real evidence that FDA regulation would have prevented the sample mix-up by 23andMe last week. As Dan Vorhaus noted in an excellent post on the incident, this mistake had nothing to do with the first-to-consumer nature of 23andMe's product; it's a problem that could have occurred to any genetic testing company. So the FDA's singling out of direct-to-consumer testing companies and their providers looks to be driven more by publicity than by a genuine desire to protect test consumers.
For more as this story evolves, keep an eye on this blog and Genomics Law Report.
Update: the letters are all available here (thanks to Dan Vorhaus). Here's an excerpt from the letter to 23andMe:

23andMe has never submitted information on the analytical or clinical validity of its tests to
FDA for clearance or approval. However, your website states that the 23andMe Personal
Genome Serviceâ¢is intended to tell patients in advance how they will respond to certain
medications including warfarin and clopidogrel. It also states that the data generated from the 23andMe Odds Calculator, a feature ofthe 23andMe Personal Genome Serviceâ¢, includes the contribution of single-nucleotide polymorphisms (SNPs) to disease risk. Consumers may make medical decisions in reliance on this information.


More like this

Update 14:35 EDT 04/08/10: I just received a phone call from a 23andMe representative indicating that these rumours are not true; more details to follow once I have a written statement. Update 17:26 EDT 04/08/10: Here is the written statement from 23andMe's PR firm: We will continue to work…
Two days ago I reported a rumour that the FDA might have convinced genotyping chip provider Illumina to stop providing its products to direct-to-consumer genetic testing companies - a move that would effectively prevent these companies from being able to operate. The rumour seemed plausible at the…
The brief Golden Age of direct-to-consumer genetic testing - in which people could freely gain access to their own genetic information without a doctor's permission - may be about to draw to a close. In a dramatic week, announcements of investigations into direct-to-consumer genetic testing…
(Cross-posted to Genomes Unzipped.) Today's US Congress Committee on Energy and Commerce hearing into the direct-to-consumer genetic testing industry was a vicious affair. Representatives from testing companies 23andMe, Navigenics and Pathway faced a barrage of questions about the accuracy and…

A PR move more than anything else perhaps? I agree that this probably would not have affected last week's mix-up, but a move to "crack down" on these companies may pacify those who find this technology frightening due to misinformation/lack of information.

All I read about are the downsides to FDA regulation, but there clearly must be some argument from the other side. Would it prevent shady companies from selling fake products? Would it ensure minimum quality standards? What's the opposing perspective?

Why is that appalling news? Personal genetic information is not entertainment, it's medical information. Genetics tests are medical tests. These companies were practicing medicine without a license.

It was inevitable that this would happen, and it's a good thing that it did.

Chris: Right, you always hear the argument that regulations are causing people to die by prolonging the time that it takes useful drugs to enter the market. What you don't hear is that for every useful drug there are 99 useless and downright dangerous drugs that are prevented from entering the market thanks to these regulations. Without regulations, we would be awash in bullshit.

This is terrible news. I like Obama, but it's overreaching government regulations like this that will doom his administration.

Stick to regulating things that can actually hurt us (oil wells, Vioxx, Madoff), and lay off regulating behavior by consenting adults.

Hi Chris,

If the FDA's priority was to prevent shady companies from selling fake products, it would be targeting companies like My Gene Profile, not Illumina or 23andMe. The industry needs measures to clear out the bottom-feeders, but there are better approaches than the FDA's clumsy sledgehammer.


Why is personal genetic information automatically medical information? Because doctors tell us it is? 23andMe doesn't hand out disease diagnoses or advise behaviour changes on the basis of its risk predictions; it is not, and never has been, practicing medicine.

So would the ancestry version of the 23andMe test not fall under the 'medical device' category despite being the same assay only with a different annotation database, which is generated anyway from publicly available information?

No Martin, what's bullshit is a government that treats everyone like idiots.

What would happen if we applied these same principles to newspapers which report on medical studies? Is reporting on an association between obesity and childbirth problems medical advice that should be regulated?

By Madeleine (not verified) on 11 Jun 2010 #permalink

I agree that personal genetic information is not automatically medical information, but look again at the excerpt from the FDA letter to 23andMe:

[Y]our website states that the 23andMe Personal Genome Service⢠is intended to tell patients in advance how they will respond to certain medications including warfarin and clopidogrel. It also states that the data ... includes the contribution of single-nucleotide polymorphisms (SNPs) to disease risk.

It's hard to see how that is anything but medical information. If these are the kinds of claims that these companies are making, they were very naive to think FDA wouldn't act. One may disagree with the current regulatory framework for medical tests, but FDA's actions here seem perfectly consistent with the current rules.

Gutierrez quote from http://www.newsweek.com/blogs/the-human-condition/2010/06/11/why-the-fd… :
"Software is a medical device, and theyâre making medical claims. Theyâre taking results and making medical claims that come out of those results."

Does this mean that Promethease is subject to regulation by the FDA? The FDA's position, taken to its logical conclusion, seems quite ridiculous.

Stick to regulating things that can actually hurt us (oil wells, Vioxx, Madoff), and lay off regulating behavior by consenting adults.

red pepper,

Information about how I might respond to warfarin or clopidogrel could easily hurt me. If I rely on it to decide my dosage, and it's wrong, it could kill me.

Why does it make sense to regulate Vioxx but not medical tests? Isn't talking Vioxx simply a behavior by consenting adults?

If it's logical to regulate drugs to make sure they work as intended, why is it illogical to regulate medical tests to make sure they also work as intended?

I've offered my own summary and analysis of the letters over at the GLR: http://www.genomicslawreport.com/index.php/2010/06/11/what-five-fda-let…

There is likely a need for expanded regulation across the genetic testing marketplace (not just DTC), but the application of regulation does not appear to be all that intelligently tailored at this point (see, e.g., Daniel's point above re: My Gene Profile). I still think, at this point, we need transparency first, regulation second.

- Dan

If the FDA's priority was to prevent shady companies from selling fake products, it would be targeting companies like My Gene Profile....

I'm afraid that's a non sequitur. Assuming the linked blog post is accurate, I agree that My Gene Profile is making bogus claims. But they're not making bogus claims about drugs, medical devices, or medical tests. Accordingly, FDA has no jurisdiction. FDA can't shut them down any more than they could shut down a dishonest used car dealer. (Perhaps the Federal Trade Commission could, though.)

"Why is personal genetic information automatically medical information?"

Why would it automatically not be medical information? What genetic information would you consider to be medical information?

"Because doctors tell us it is?"

No, because the substance and intended application of the information is intrinsically medicine. There does exist an exclusion for "research purposely only." Is that what you believe these genomic products to be doing? "Research only?" What would a genetic testing provider who claims to be for "research only" have to do to satisfy you that the test was not intended for "research only" in good faith despite the claim otherwise?

"23andMe doesn't hand out disease diagnoses"

Tay Sachs is a disease diagnosis. BRCA mutation is a diagnosis itself which also makes one susceptible to breast cancer and other cancers. You could argue about the definition of "diagnosis," but these two particular examples had existing universal diagnosis definitions with a history of clinical application.

Tay Sachs
ICD-10: E75.0
ICD-9: 330.1

(I don't know the ICD-10)
ICD-9: V84.01, V84.02

"or advise behavior changes on the basis of its risk predictions"

In addition to the contents of the genomic test results themselves which I assume to be self-evident medical advice, you know and I know that 23andMe has always intended to "advise" people how to change their behavior based on the contents of the test results, and 23andMe has actively advised users to share their results with medical doctors through means including the 23andMe Twitter feed.

Here is a video of Anne Wojcicki herself announcing to TEDMED 2009 exactly how 23andMe intends for their service to be used to modify medical behavior ---specifically Warfarin dosing.


"FDA's clumsy sledgehammer"

How was the FDA clumsy? Yah, OK Steven Murphy. The FDA may be clumsy, but you're obligated to state why, because as far as I can see, everything else you written about this particular subject has been clumsy itself.

You are making a mistake, Daniel. 23andMe could have done great things, but they were so antagonistic that now there is a special FDA task force for genomics specifically. That doesn't happen overnight. It happens over years of public abuse by people who can't handle the power but feel obligated to prove that they can achieve the success of their peers (or spouses) without "help." Your ire is misplaced. Your responsibility as an authority on this subject now is to not alienate the medical community with sour grapes of your own... unless you want to be your own Steven Murphy of genomic scientists and be laughed at yourself.

By Andrew Yates (not verified) on 12 Jun 2010 #permalink

The FDA is asking 23andMe et al. to explain why their products should not be regulated as in vitro diagnostic devices.

It may not be clear to all readers that, in practice, it would not be possible for 23andMe (or the others) to submit applications for such regulatory clearance (in the industry, these are called "PMAs" and "510-ks").

These filings are very demanding: time-consuming and expensive. They require that procedures and methods be set out exhaustively, and that results be validated with clinical trials. Among other things!

The delays and regulatory costs would likely drive 23andMe to bankruptcy, and/or force the company to make its products so expensive as to be unattractive to almost all customers. (This latter sentence is simply my opinion; I haven't made a go of the numbers.)

Perhaps some new, as-yet-unspecified regulatory regime for consumer genomics would represent an achievable hurdle for these companies to overcome with their current business models. But not the current regulations.

The FDA actually comes across as somewhat naïve and amateurish in their recent attempts to flex their muscles in this area. Take, for example, their letter to Illumina, in which the FDA faults them for supplying 23andMe and deCODE genetics with the Illumina Infinium HumanHap550 array. The problem is that the deCODEme service doesn't actually use this array (see http://www.decodeme.com/faq ). If the FDA wants to show they are better able to monitor this field than the consumers of these products, they should at least be able to correctly identify the "medical devices" in question. Their recent actions only seem to add weight to the idea that, in general, the medical establishment is less knowledgeable in this area than the informed consumer.

The delays and regulatory costs would likely drive 23andMe to bankruptcy, and/or force the company to make its products so expensive as to be unattractive to almost all customers.

Well there's at least one simple solution: 23andMe et al. could simply not advertise their products as useful for making medical treatment decisions.


The problem is that the deCODEme service doesn't actually use this array (see http://www.decodeme.com/faq ).

Perhaps, but according to FDA's letter to Illumina:

During a meeting between Illumina and FDA on July 31, 2009, [Illumina] stated that 23andMe and deCODE genetics both order Illumina arrays for use in their personal genome diagnostic tests.

FDA isn't being naive or amateurish at all. It's fulfilling its regulatory duty under the law. As their letters explain, Congress passed laws decades ago that require approval of medical tests.

All this animus against FDA is misdirected. If you think 23andMe should be able to sell unapproved genetic tests that provide info on drug dosing and disease risk, it's Congress you need to be concerned with.

Why didn't FDA send letters to Affymetrix and BGI (who did the sequencing for Knome)???

@qetzal, I think greg is saying that FDA should say 1M-Duo chips are used by deCODE.

By Geneticist fro… (not verified) on 13 Jun 2010 #permalink

Oligopoly loves regulation. Make the barriers high and the little guys can't compete. Big money special interests win again.

Is there any evidence that open source information is harmful ?

Anything? Anybody?


23andMe et al. is not "open source" because the terms of service specifically limits and forbids particular intrinsic applications of the information derived from that service.

You are welcome to argue that such a contract is a necessary evil in protest to existing government in drugs and medicine. I would agree with a reasonably presented argument given that presentation. However, such an argument would necessarily be the flagrant rebellion from existing federal enforcement policies regarding drugs and medicine in the United States ---and likely elsewhere. To make such an argument would require the courage, honesty, and leadership and 23andMe, Navigenics, and Pathways have apparently lacked as evidence by their failures as businesses and inability to keep cohesive leadership and investment teams for even a few short years.

I continue to be confused by the entertainment of these scorched earth "Napster of medicine" rapid deployment strategies driven by gluttonous mob support. Let me ask you: is the world more or less hostile to the freedoms of individual artists given the fantastic powers we enjoy to transmit any data worldwide almost immediately for free? Do you think that consequences for medicine will be any different? Especially when the revolution was founded on something so duplicitous as "contractually enforced non-actionable" medical reports?

Everybody here does know or should know that ---if any coherent strategy can be distilled from the hangover from this glutinous orgy in medical genomics at all--- that a complete break and subsequent independent existence from the "healthcare establishment" has always been the strategy for 23andMe et al.

The most depressing fact is that this COULD have worked, but it DIDN'T. Why? And now who suffers? The independence and freedom to innovate of all BUT the most powerful institutions of the "healthcare establishment" ---the biggest victor of all the federal government itself.

And why? I would like to know WHY given ALL the power and talent and access in the world that 23andMe FAILED and NOW the world is worse for it when it COULD have been that much better. I want some answers, because I don't it to happen again, and I'm sick and tired of everybody being such a goddamn coward and not to name names or accept actual responsibility with facts and reasons instead of hiding behind silly internet rants published by handles like "dorkface."

By Andrew Yates (not verified) on 14 Jun 2010 #permalink

If the same companies would offer a "home antibody screening test" to detect HIV, XMRV, morgellons and lyme, no one would doubt the FDA's right to jump all over it. The companies are violating the main quack miranda rule of stating "not designed to diagnose or treat diseases". You cannot do that and hope the FDA will stay away in the long run.

How is this a PR move? This action is overdue. if you make claims like that (being able to tell if someone responds to Warfarin for example) then you better be prepared to deal with the FDA.

I am amazed that anyone thinks we should turn our bodies over to the FDA. That anyone thinks the government has the right to oversee what I do, what I learn and what choices I make with that information has been seriously conned. Who are all of you that think you should give up your rights to your medical knowledge?

Obviously you have never diagnosed and cured something the medical mafia failed to do. I have. And once you learn how to do this, you don't walk back into the lion's den.

No one with any good publications goes into government. You go to a hot startup or to academia (and in that order, these days, for the most ambitious/skilled).

Why is the FDA is more qualified to judge sequencing than Illumina? Does anyone really think they are the "experts" here? When is the last time someone at the FDA actually created a wonder drug, built a new prosthetic, or invented a sequencing machine?

When you think FDA, think TSA not CSI.

Why is the FDA is more qualified to judge sequencing than Illumina?

Reading and/or comprehension fail.

It isn't about judging sequencing. It's about judging medical claims. And in that area, yes, FDA is the expert.

Oh really, qetzal? And why pray tell is the FDA more qualified to judge "medical claims" than the actual PhDs and MDs who discovered and published the breast cancer and warfarin associations in the first place? You know, like Russ Altman from 23andMe or Kari Stefannson from DeCode?

I must have missed the part where Alberto Gutierrez taught genomic medicine at Harvard Medical School.

Genomic conservatives like yourself will be the death of us all (literally).

Also qetzal I guess you may not actually know how the FDA works. They cannot regulate the practice of medicine (though god knows they've extended their grubby paws time after time, only to be beaten back by the threat of a class action). Moreover because of that little technicality known as the US Constitution, technically they can only regulate "interstate commerce" -- so it's possible that someone might decide to just operate in California or Texas. It'd be a hit for sure but not as bad as having them declare 23andMe illegal overnight, without benefit of vote, debate, etc.

If you don't want 23andMe's product, don't get it. But hands off my genome!

@Orilly #31 - The Federal Food, Drug, and Cosmetic Act gives the FDA the responsibility of regulating medical tests. It doesn't have to cross state lines, it just has to be a medical test.

I'm all for companies being allowed to sell you your genome - the raw data, that is. If they want to score and sell medical diagnostics they need to go through CLIA and 510K II.

Now, if someone wants to distribute free software on the web that will extract medically relevant info from the data you purchased and interpret it for you: no one can stop 'em. Caveat emptor.
On the positive side, if the software includes the option of adding relevant bits of patient history to the risk calculations, it will probably do a better job of it than your own doctor would. Eddy, Gigerenzer, and others have tested MDs and found that most don't understand how test sensitivity, base rate, false positives/negatives, conditional probability, etc, are factored into interpreting diagnostic tests.