One of the comments on the submitted article on its website nails it:
"What 23 and me is saying is not accurate. The reason why the FDA has a problem with them is not the fact that they are decoding DNA for people, it is the fact that they are interpreting the results to diagnose specific medical conditions direct to the public.
"This is against FDA regulations unless it is waived (CLIA waived). In order for a test to be CLIA waived it has to be 'simple laboratory examinations and procedures that have an insignificant risk of an erroneous result.' Examples include glucose tests and pregnancy tests. DNA testing is FAR from that.
"I am a member of 23 and me and they scour scientific research papers to assign medical conditions to certain genomes. They rate the quality of the research from 1 to 4 stars based on how established the research is. How they determine this is unknown. Genomics is rapidly expanding and it is often hard to determine disease correlation with a genetic marker."
Hear. Hear. I have a local friend whom I grew up with who is an electrical engineer who works on medical devices. By any reasonable reading of the FDA regulations adopted under statutory authority from Congress, the product and service combination on offer from 23andMe falls within the scope of FDA's regulatory authority, and the company's consumer-facing product claims are extraordinary claims for which the company has not provided evidence. That's bad.
An article from Fast Company previously submitted to HN and discussed here points out that 23andMe's business model is basically to gather a lot of cell samples and a lot of personal health data from consumers, and then gradually infer genetic relationships to disease risk that might be valid and actionable. But that's not how it is marketing itself to consumers today.
I posted that comment on NYT then submitted this article to HN. I used to work in the medical laboratory business before getting into software development. I do not think 23 and me is going to escape enforcement from the FDA. The comments they made to NYT were about the fact that the FDA has to look at each test individually to determine if it can be CLIA waived. They imply that this is an archaic way of doing things as they are testing millions of correlations, and compare it with the challenges in the tech community with slow regulatory change.
They are wrong about this. The FDA does have to evaluate every diagnosis that they are making as it is that.. a diagnosis. Diagnoses that are going straight to the consumer have an even higher bar to meet not unlike a drug becoming OTC. Medical laboratories are not allowed to perform direct to patient testing for non waived tests.
The purpose of this is that they do not want patients to interpret tests incorrectly where there is gray area, and possibly make adverse decisions based on the diagnosis.
I was just popping in to post essentially the same thing. I have worked on two class III medical devices (systems work with a focus on software) over the past eight years (lead developer and project manager). It is obvious to anyone who has an ounce of experience dealing with the FDA that the interpretation is the problem and that such a product should be regulated in some way.
Honestly, their comments come off as a bit flippant, and really, they should be a bit more fearful. Right or wrong, the FDA is not to be trifled with.
According to the FDA's letter of action, they spent years discussing the company's marketing of their product with 23andme. Surely the company, in that time, could have chosen to rebrand. Instead, it is plastered with quasi-medical advice.
Not that I think this is even bad—but I do think that this puts the company's marketing claims justifiably under the FDA's purview.
All they need to do is slap a disclaimer all over the place and instead of mentioning specific diseases say something non-specific like "this test result indicates your heart/lung/bone/etc is unhealthy/risky/bad/etc". The completely bullshit vitamin/herbs/supplements/alternatives industry has been making similar claims without a peep from the FDA. You gotta know the rules to the game even if they are arbitrary.
From the FDA letter to 23andMe describing why they have regulatory oversight:
"because it is intended for use in the diagnosis of disease or other conditions or in the cure, mitigation, treatment, or prevention of disease, or is intended to affect the structure or function of the body."[1]
No mention of disease (or structure or function) then there would be no problem.
The FDA makes many arbitrary distinctions because of the long and twisted legal and political history of "health" industry regulation[2].
If 23andMe marketed their service purely for, say, genealogical purposes, there wouldn't be an obvious problem. Instead, they not only report on disease-related SNPs, but try to interpret those into a coherent diagnostic message.
In my opinion, this is stepping into the FDA's realm of regulation.
Can't disagree with you there. Seems like their original plan of testing for genetic risk factors for a large number of diseases will require lengthy approvals for each and perhaps require conditions such as counseling that are not compatable with a kit.
My original comment was more a snark about the free pass the FDA gives to the supplement industry. This is a much less settled area so it is unclear if some fudge like I suggest could work either with the FDA or in the market.
Perhaps they will even seek legislative of legal relief like the supplements industry.
1. Drop all interpretation of the results from the results pages. Instead, just spew out a bunch of semantically-tagged gene tables.
2. Take all the interpretation map-data they have so far, and dump that into a FOSS Github project for a Firefox/Chrome Extension, which parses the results page and re-renders it into... exactly what it is now.
Bam. "23andMe" is no longer giving medical advice; some FOSS project is. (Even though the developers, individually, could be ongoing contributors to the very same FOSS project.)
This is actually probably for the better, either way. Other front-ends could make use of 23andMe's genetic-data "API", and the FOSS interpretation-map could be maintained by people (doctors!) all over the world.
But you've just open sourced their entire value-proposition, which is that correlation part based on all the data. Mapping your DNA is now the 'easy' part, and offers them no differentiation and competitive advantage. (Ok, they still have lots of data to use to do their own lookups of all customers using the FOSS and improve that project, but again, they're adding value outside their company)
I think that could still be tenable, if the marketing was rearranged so that the FOSS project was the initial point-of-contact for people looking for gene mapping (picture it as a whole app, rather than just an extension), and then it does lead-gen for "23andMe, the premium gene mapping service compatible with our gene explorer."
The FOSS project still might get forked, of course... but as long as the 23andMe people are actually the leading experts in doing those correlational studies, their fork will stay the canonical/preferred one. Seems more honest. (Of course, no company wants to have to be "honest" in that way, but if the only other option being legislated out of existence, it might still be a profitable strategy.)
Creating hyperlinks between genes and research papers is not what I'd call diagnosing. You find evidence of ABCD gene, you search all research papers that cite ABCD, you curate and rank papers by quality, summarize related conditions with statistical correlations.
This is just a high-tech version of me using WrongDiagnosis.com, typing in 5 or 6 symptoms, and having it return all possible articles that cite those symptoms, ranked by number of simultaneous occurrences.
The potential hack to defeat the FDA here is to simply split the company in 2 parts.
Company #1 sells a $99 service which gives does the test.
Company #2 is a partner of Company #1 through an API. Consumers can elect to OAuth access of their gene data from Company #1 to Company #2.
Company #2 takes gene data over an API and sells you linked and ranked reports on it. Company #2 sells no devices, performs no tests. They are pure software bioinformatics. They take standard file formats and produce reports.
Consumers can elect not to use Company #2's services, get their own private data in raw format and write their own python scripts if they want.
Or, consumers can buy Company #2's services, but supply their own data from any company that offers personal genomics raw-data dumps.
The idea that me reading ranked reports to research papers linked to my personal health data is 'extraordinary claims' and a danger to me is frankly offensive. I would hope that a company offering software that parsed and processed genetic data and turned it into linked reports from crawled research papers could be considered a first amendment right.
But if not, the github approach is certainly a worthwhile approach.
Again, as I said in other posts, the FDA doesn't regulate supplements, doesn't regulate Dianetics/Scientology, didn't stop "The Secret" from killing people with heat-stroke therapy. It's enormously offensive to me that the government spends so much time going after stuff like this, pot smokers, et al, instead of massively beefing up enforcement over the hundreds or thousands of hucksters selling harmful bullshit health products on radio and late night TV ads.
Does no one remember http://en.wikipedia.org/wiki/HeadOn ?
Coldeze?
Five Hour Energy, which may have actually killed people, and still gets away with deceptive ads.
What about Hoodia? Yeah, the FDA advises people not to purchase it, but does nothing to stop consumers from buying it based on false claims of efficacy.
People don't like this style of reasoning, but in my opinion, the government regulatory agencies have limited resources. I want the SEC going after big banks and investment firms, not small time insider traders. I want the IRS auditing people who avoided millions or billions in taxes, not the guy who owes $1000. I want the FDA going after things which are much much more likely to be ripping off and harming consumers, either financially, or physically, and on the scale of possible things that can harm you, this is way way down the list based on a really loose theory of someone avoiding treatment by false negative, when in reality, if anything, the majority of people are going to be having more conversations with their doctor over this product.
It's like complaining about the people who lost under Obamacare and ignoring the fact that tens of millions more couldn't get any health insurance in the first place.
Your "hack" wouldn't work. The first company would still have to validate that their tests were accurately detecting SNPs. Then the second company would still have to validate that their "diagnoses" were accurate. Medical devices aren't restricted to physical tests. If it is used in the diagnosis of disease, even if it is just software, then it still might be a medical device. This is especially true if it is marketed directly to the consumer.
23andMe does more than just create hyperlinks between genes and research papers. It tries to assign risk.
History has shown that a little regulation in this field is a good thing (and just might make the 23andMe product better).
It's not in the FDA's remit to be regulating research, that would be an infringement of academic freedom, it can however regulate commercial entities using that research to sell devices to the general population, who don't have the tools to judge the research -- published research isn't absolute truth, just a report by a researcher(/group) of what they've observed and what they think. The extra context they add around the research goes beyond just 'hyperlinking' and it's right to regulate. Do you think quacks should be able to present legit research as supporting their claims?
Real quacks present do this all the time and no one stops them, and they are far more dangerous. People are making money selling anti-vax books and videos after all, but we can't regulate those, but we can shown that millions of children have potentially been harmed by these claims far far more than 23andme.
Basically what you're saying is, consumers are idiots, so they don't have the rights to judge such information.
Really, if a consumer is found to have BRCA, and shown links to academic research that have a statistical correlation between BRCA and breast cancer, what do you think they're going to do, go cut off their breasts themselves? No, they're going to go visit a real doctor and say "look what 23andme said about me, what do you think doc?" And the doctor is going to order more thorough tests.
In almost every scenario you can imagine, someone reading a 23andme report about themselves, even one loaded with false positives, is going to generate a visit to a medical professional for interpretation and advice. 23andme is not telling anyone to self-treat based on such information.
The only scenario anyone can claim would be truly dangerous would be someone getting a false negative clean profile and going "welp, I'm perfectly healthy, no need to bother with my regular checkup". Now really, do you believe someone who paid $99 to 23andme and is actively seeking personal medical information is the kind of person who avoids doctors?
We have lots of problems in this society, and the problems of people avoiding doctors can be boiled down to either a lack of health insurance, or issues surrounding personality that have little to do with 23andme. The FDA is preemptively regulating a new industry based on relatively weak theories of harm, and my point is, they simply have better things to do.
At least let 23andme get off the ground and you have actual data on people's response to it before you start hammering them and triggering stupid class action lawsuits from people who haven't even been harmed. The FDA is potentially killing a startup over administrivia.
> The potential hack to defeat the FDA here is to simply split the company in 2 parts.
The easier "hack" is to just stop making unwarranted health claims. Or, if the claims are not unwarranted, to provide the support for them.
> Again, as I said in other posts, the FDA doesn't regulate supplements
Yes, it does, but the scope of its regulatory authority set by Congress over supplements is different than the scope it has over food, drugs, and medical devices.
> doesn't regulate Dianetics/Scientology
Dianetics/Scientology is neither a food, drug, supplement, medical device, nor anything else vaguely related to the FDA's scope of regulatory authority.
> didn't stop "The Secret" from killing people with heat-stroke therapy.
As with Dianetics/Scientology, this is not within the scope of FDA's regulatory authority.
> It's enormously offensive to me that the government spends so much time going after stuff like this, pot smokers, et al, instead of massively beefing up enforcement over the hundreds or thousands of hucksters selling harmful bullshit health products on radio and late night TV ads.
I don't think the government is actually spending all that much time on the things you are complaining about (I think you are confusing the amount of time that people spend on the internet talking abou thte government doing something with the amount of time/effort the government spends doing it), but if you want to advocate that Congress should expand the FDA's regulatory authority so that the things you complain that they aren't doing are actually within their scope fo regulatory authority, feel free.
I still don't see that as an argument that the FDA shouldn't act thoroughly within the domain they are charged with regulating now.
>Yes, it does, but the scope of its regulatory authority set by Congress over supplements is different than the scope it has over food, drugs, and medical devices.
It does a piss poor job of it. Stop wasting time on 23andme, and pour resources in that. The FDA admitted that a whopping 70% of all supplements violate their rules.
>Dianetics/Scientology is neither a food, drug, supplement, medical device, nor anything else vaguely related to the FDA's scope of regulatory authority.
Point is, these things make unwarranted health claims and cause consumers of them to take actions which directly harm themselves. If someone published health recommendations that tell people to drink bleach to "cleanse" themselves, the FDA can do nothing, but if someone tells them they have the BRCA gene and here's a link to resource that shows it increases risk of breast cancer, well then, shut em down.
> I still don't see that as an argument that the FDA shouldn't act thoroughly within the domain they are charged with regulating now.
Right, the law says smoking pot is illegal, so there are millions of people who have had their lives ruined by the criminal justice system because of that rule, and millions more who have to pay billions for prison funds. But yes, that is not an argument that cops and judges shouldn't prosecute pot smokers to the fullest extent of the law and not prioritize, because, well, it's the law! Ditto for illegal immigrants. Just ship em all back because it's the law. Obama has been asked to de-prioritize going after pot smokers and illegals, do you think that is a bad idea and instead, he should prosecute them to the fullest extent of the law up until the last minute when the law is changed?
The law can be wrong. In this case, the FDA has been given the power to regulate something that straddles an entirely new industry and they are preemptively intervening before any harm has been shown. My argument is, regardless of the law, one can shift around one's priorities. The FDA claims 70% of supplement makers are violating rules. Companies like Ranbaxy are shipping adulterated generics. These are actually harming people and the FDA has at least some authority.
There's an entirely new industry cropping up around the "Quantified Self". Individuals are gaining more and more power to monitor their biological systems, and personal genomics are just another in a long line of this evolution. The traditional power structures in the medical community do not like this decentralization towards consumers being able to get medical information about themselves without consulting doctors first, without doctors ordering tests, without doctors first interpreting the results.
I am stone cold libertarian on this. When it comes to my body, I demand the right to do anything to it, even if it may be bad for me, and to have any information on it, and to contract with those who can provide such information. My own concern is real fraud, that is, if someone says here's a blood pressure monitor, that doesn't actually measure BP, or here's someone selling a DNA test, which actually gives me a totally false randomized DNA profile. My concern is not over someone giving my a dump of my DNA and then generating a series of correlated links to potential diseases. I will ultimately take that information and do the research myself. Anymore than I am not concerned with a health site that takes my medical records or personal survey, and gives me a bunch of self-diagnoses diseases that match.
In short, where you and the FDA see "health claims", I see a search engine.
The problem there is that FDA has statutory authority to regulate some things and not others. (The notorious exemption is "natural supplements.") But the FDA takes a remarkably thorough approach to what it does regulate. Let me give two examples.
Example one: I have been in two FDA-regulated medical trials. One was a placebo-controlled trial of taking fexofenadine daily for indoor allergies. I had no idea if my numbered supply of tablets was placebo tablets or genuine tablets. I had a schedule for taking them, and had to keep a detailed log of symptoms. The second clinical trial I did was a trial of consumer-facing store displays for a proposed over-the-counter medicine for controlling cholesterol. At the test center, there was a mock store display that looked just like the shelves of a pharmacy, and I read the mock labeling and then was asked several questions about my understanding of the purpose and use of the drug. I was even given a brief IQ test as part of the trial, to see how my understanding of the store display would compare with that of someone with a different level of IQ. The FDA is thorough, and it relentlessly uses carefully gathered data and statistics analyzing those data to make regulatory decisions.
Example two: My son the hacker had a summer job while still a university student writing code to control the display of a test device to be used by medical doctors in a hospital or clinic setting. He had to think through all the user interactions medical doctors would have with the device, and how to code the device so that it would have high usability for doctors and high reliability. As he neared the end of his summer job, medical doctors came to the medical device company where he worked to do carefully monitored usability testing of the product. My son told me at the end of the summer that the product he worked on would have to be tested with regard to all aspects of hardware and software for several more years, including a complete line-by-line code review of all his code by FDA specialists in software. He heard from his bosses that time to market for that product would be as much as five years after the summer when he worked on it (in other words, still in the future today). The FDA is very thorough in its regulation of medical devices. I also know this from an electrical engineer who is a childhood friend of mine, who has long worked in another medical device company in our town, which has major medical device companies headquartered here. Safety is job one at the FDA. Effectiveness is job two. Both have to be proven before a new drug or medical device goes to market.
Simply put, the people at 23andMe have been doing what a mathematician calls "hand-waving," concluding things that they haven't proven. That helps develop a kind of zeal for the task for the long haul, but it doesn't ensure that the task gets done right.
i should make clear, i absolutely think 23andMe are in the wrong. very very far into the wrong.
the fact that the FDA are effectively impotent in many areas they are not is largely irrelevant, but I enjoyed seeing it brought up since it is very true and also a real world problem with a measurable impact on the economy and quality of life.
First, calling the FDA "a very important partner" is overly glib. The FDA is not a "partner" in the sense that they will accompany you or assist you on your journey. They are only there to put the brakes on you. They are not merely "important," they are _vital_.
Second, the article is skewed to make it sound like people aren't entitled to our own DNA information. That's not what is contested. Of course, you can have access to your own DNA data, as long as the proper checks are made to ensure you're actually getting an accurate reading.
The issue is whether telling you have a disease with life-scale consequences requires validation of these claims.
There is something to be said for their tactic of giving a large number of answers instead of validating single tests, but it still requires a thorough level of validation to protect people from exploitation and it wouldn't hurt their cause to present themselves more respectfully to the authorities. Perhaps they're trying to fake casualness, that's certainly my reading, but I hope they don't screw this all up but not thinking it through and throwing around lots of bullshit.
For example, comparing themselves to Google is ludicrous. First, google's indexing search engine HAD been doing before, by altavista, webcrawler, etc., etc. Second, there was no question that whether what they were doing could negatively influence people's health and well-being. Perhaps the scholar/books scanning was a bit controversial. Perhaps some were surprised at Streetview, but that was actually legal and in both cases, they took several key measures, voluntarily, to keep themselves well inside the law. BTW, I'm no google fanboy, but to compare their IT stunts and magic tricks with potentially delivering low-quality information about life-threatening health conditions "isn't in the same ballpark. It isn't even the same goddamn sport." -Jules.
Is anyone actually contesting that 23andMe doesn't get an accurate enough reading of DNA? I thought the problem was all with interpretation.
https://www.23andme.com/more/science/ states "Data quality starts with the laboratory that processes your DNA. Our contracted laboratory is licensed by the federal government to perform clinical testing. It uses standardized processes and procedures that ensure highly accurate genetic data." and so far I haven't seen anyone contesting this. That is, DNA reading part is already regulated.
Yes. The FDA is contesting that 23andMe doesn't get an accurate enough reading of DNA. FTFA:
"Last week, the F.D.A. ordered 23andMe to stop marketing its genetic test because it said the company had not provided adequate evidence that the tests were accurate,"
"genetic test (of diseases)" is quite different from reading DNA, and presumably includes interpretation. I don't understand how can reading DNA part can be inadequate, given that it's the same lab doing the same process as other clinical testings. 23andMe is using off-the-shelf SNP array, it's not like it's doing anything special with respect to reading DNA.
How long before 23andMe sells your personal DNA information up the river to advertisement corporations?
Don't trust corporations with your personal information because, with few exceptions, they are more than willing to sell you out to make a quick buck. And then they claim that their slogan is 'Don't be evil.'
I believe one of their goals is to crowd source medical information and possibly sell the results. Right now they have a "23 and We" research program where you can voluntarily opt in. Even if you don't when you go on their site they try to get you to answer questions about yourself from time to time. I have never answered them and they seem very innocuous such as is your earwax wet?. I am not sure if they dig deeper if you answer those questions in future questions
Philip Greenspun (of the famous "tenth rule of programming"!) had an interesting question [1]:
Question: would it be illegal for 23andme to relocate to Singapore, for example, and have people mail DNA samples to Singapore and then email them the results? The company’s web site indicates that customers in Singapore (and a bunch of other countries) can send their DNA to have it tested. Would the FDA have jurisdiction to prevent a Singapore-based company from accepting packages from consumers in the U.S.?
You really trust a large government bureaucracy that's been beholden to large pharmaceuticals to get this right? Really? The danger of the information is that great? I'm just not buying it and the comments in this thread so pure "anti-disruption" bias which apparently is the new hipster anti-hipster sentiment. 23andme provides information that can be used in conversation qualified professionals. You might was well shut down every medical related forum, Yahoo Answers, WebMD and the rest if you really think that information or misinformation is the danger. Really people take a step back the current memeset at look at the facts.
I think this comment is conflating "regulation" with "shutting down".
As a society, in the USA, we have created the FDA and empowered it to cast a watchful eye over certain types of marketing and manufacture. Regulation limits the rate of innovation in exchange for attempting to limit harm.
"What 23 and me is saying is not accurate. The reason why the FDA has a problem with them is not the fact that they are decoding DNA for people, it is the fact that they are interpreting the results to diagnose specific medical conditions direct to the public.
"This is against FDA regulations unless it is waived (CLIA waived). In order for a test to be CLIA waived it has to be 'simple laboratory examinations and procedures that have an insignificant risk of an erroneous result.' Examples include glucose tests and pregnancy tests. DNA testing is FAR from that.
"I am a member of 23 and me and they scour scientific research papers to assign medical conditions to certain genomes. They rate the quality of the research from 1 to 4 stars based on how established the research is. How they determine this is unknown. Genomics is rapidly expanding and it is often hard to determine disease correlation with a genetic marker."
Hear. Hear. I have a local friend whom I grew up with who is an electrical engineer who works on medical devices. By any reasonable reading of the FDA regulations adopted under statutory authority from Congress, the product and service combination on offer from 23andMe falls within the scope of FDA's regulatory authority, and the company's consumer-facing product claims are extraordinary claims for which the company has not provided evidence. That's bad.
An article from Fast Company previously submitted to HN and discussed here points out that 23andMe's business model is basically to gather a lot of cell samples and a lot of personal health data from consumers, and then gradually infer genetic relationships to disease risk that might be valid and actionable. But that's not how it is marketing itself to consumers today.
http://www.fastcompany.com/3018598/for-99-this-ceo-can-tell-...