5 ms·

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

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. http://www.fastcompany.com/3018598/for-99-this-ceo-can-tell-you-what-might-kill-you-inside-23andme-founder-anne-wojcickis-dna-r http://www.fastcompany.com/3018598/for-99-this-ceo-can-tell-...

kefka 13y ago
Fine. Then label it for "Entertainment purposes only".
- carbocation 13y ago
  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.
  - spikels 13y ago
    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.
    - carbocation 13y ago
      > "this test result indicates your heart/lung/bone/etc is unhealthy/risky/bad/etc" It's pretty unlikely they would escape the FDA's review simply by making their medical diagnoses slightly more vague.
      - spikels 13y ago
        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]. [1] http://www.fda.gov/ICECI/EnforcementActions/WarningLetters/2013/ucm376296.htm http://www.fda.gov/ICECI/EnforcementActions/WarningLetters/2... [2] http://www.fda.gov/NewsEvents/Testimony/ucm112576.htm http://www.fda.gov/NewsEvents/Testimony/ucm112576.htm
        carbocation 13y ago
        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.
specialp 13y ago
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. http://wwwn.cdc.gov/CLIA/Resources/WaivedTests/default.aspx http://wwwn.cdc.gov/CLIA/Resources/WaivedTests/default.aspx
EpicEng 13y ago
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.
btian 13y ago
It's wrong. 23andme does not diagnose disease. It links DNA to published research papers and provides confidence level based on a number of factors.
- EpicEng 13y ago
  ...provides confidence level based on a number of factors You have just described a medical diagnosis.
- potatolicious 13y ago
  In the same way that Lyft rides are paid with "donations" and not "fares", yes? Fortunately corporate doublespeak does not make a duck stop waddling or quacking.
derefr 13y ago
Simple patch, in my mind: 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.
- LogicX 13y ago
  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)
  - derefr 13y ago
    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.)
cromwellian 13y ago
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 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.

kefka 13y ago

Fine. Then label it for "Entertainment purposes only".

carbocation 13y ago
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.
- spikels 13y ago
  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.
  - carbocation 13y ago
    > "this test result indicates your heart/lung/bone/etc is unhealthy/risky/bad/etc" It's pretty unlikely they would escape the FDA's review simply by making their medical diagnoses slightly more vague.
    - spikels 13y ago
      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]. [1] http://www.fda.gov/ICECI/EnforcementActions/WarningLetters/2013/ucm376296.htm http://www.fda.gov/ICECI/EnforcementActions/WarningLetters/2... [2] http://www.fda.gov/NewsEvents/Testimony/ucm112576.htm http://www.fda.gov/NewsEvents/Testimony/ucm112576.htm
      - carbocation 13y ago
        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.

specialp 13y ago

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. http://wwwn.cdc.gov/CLIA/Resources/WaivedTests/default.aspx http://wwwn.cdc.gov/CLIA/Resources/WaivedTests/default.aspx

EpicEng 13y ago

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.

btian 13y ago

It's wrong. 23andme does not diagnose disease. It links DNA to published research papers and provides confidence level based on a number of factors.

EpicEng 13y ago
...provides confidence level based on a number of factors You have just described a medical diagnosis.
potatolicious 13y ago
In the same way that Lyft rides are paid with "donations" and not "fares", yes? Fortunately corporate doublespeak does not make a duck stop waddling or quacking.

derefr 13y ago

Simple patch, in my mind: 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.

LogicX 13y ago
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)
- derefr 13y ago
  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.)

cromwellian 13y ago

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 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.