Should Mary Carmichael Get DTC Genetic Testing? My Full Post

Mary Carmichael at Newsweek is running an interesting week-long series of posts about deciding whether or not to buy DTC genetic testing for herself.  Carmichael series She asked me to contribute an “about 300 word” post on “How do I know if I can trust the results and should I be worried about what I might find out?” Monday, I sent her both a 389 word version and a 300 word version. Tuesday, she sent me back a 308 word version for posting, which I accepted. Today (Wednesday), the Newsweek site has posted a 230 word version.  Carmichael:  Day 3 It captures much of what I wanted to say, but not all. So here’s the original 389 word version.

* * *

You can be confident you can’t trust the results.

SNP chips, though useful in research, aren’t good at providing health information. First, as the recent GAO report vividly showed, companies often interpret the same SNP results differently. The same DNA conferred an above-average risk of developing leukemia at one company, but a below-average risk for leukemia at another. You can’t trust any result unless there is a consensus on its interpretation.

You also can’t trust the reports as useful about your health. Personalized medicine isn’t just about genetics. If your test shows you have a genetically low risk of lung cancer but you smoke 4 packs a day, your actual risk is much higher than average.

Finally, even true SNP effects aren’t very big. 23andMe’s web page shows sample results. For breast cancer, its report shows that a woman with a 9.1% (genetic) chance of breast cancer, compared with an average for women of European ancestry of 12.5%. Will that 3.4 percentage point difference change your life – or be worth paying good money for? (Oh, and if you aren’t of European ancestry, good luck getting any meaning out of these tests – the research behind the results was almost always done with white people.)

You should not be very worried about what you find out – as long as you realize that it is almost always useless. But these companies do sell information on some powerful SNPs: APOE, linked to Alzheimer disease, and three mutations in BRCA 1 and BRCA 2 that powerfully raise a woman’s risk of ovarian and breast cancer.

You might worry about reactions to bad news about those genes. Two smart and knowledgeable people, James Watson and Steven Pinker, both excluded only APOE from their public whole genome sequences. Studies show negative psychological reactions are uncommon, but those studies included extensive genetic counseling – not required by these firms.

Without counseling, you could misinterpret results in dangerous ways. If you test negative for those BRCA 1 and 2 mutations, you might decide to avoid mammograms. That could be fatal. Not having those mutations lowers your genetic risk only from about 12.5% to about 12.4%.

Knowledge can be powerful, both when understood properly and – dangerously – when misunderstood. As a first step the FDA should require that every health-related genetic test be prescribed by, and be returned through, your doctor.

4 Responses to Should Mary Carmichael Get DTC Genetic Testing? My Full Post
  1. Two points in your remarks, both od which were in the online version of Mary’s artcle that I read.
    First: (Oh, and if you aren’t of European ancestry, good luck getting any meaning out of these tests – the research behind the results was almost always done with white people.)
    Isn’t this a somewhat dated and maybe inappropriate statement to make. From the information I have read there are three project that I have read about, Genographic, Sorensen and Genome project where testing of “white” people are of a minority. From discussions I have followed on several of the information list re DNA and studies I would also have to say that there it is highly probable that this is also incorrect.
    Second: As a first step the FDA should require that every health-related genetic test be prescribed by, and be returned through, your doctor.
    Why? Is it require that I need a doctor to order every medication I can take? Is it required that I need a doctor to refer me to everybody offering every type of medical treatment?
    To my point of view it is like saying I need a mechanic to inform me the I need to refuel my car and refer me to the fuel depot or that if I have a headache a doctor has to presccribe panadol for me to take.
    The laboritries that these test are done in to the best of my knowledge need to be licenced and the technitician and scientist working there need to have the qualifications the do the test and report on the results.
    If the lab is prepared to undertake a test for me for $x and I am prepared to pay $x for that test why shouldn’t that retail transaaction be able to take place?
    I have spend 20 years of my life in the service of my country as a soldier so that you and I would have the freedom to expres our views on this matter and I do get a bit concerned when someone tries to take a freedom away from me because someone else doesn’t think I can handle the news that that freedom might reviel to me or that I haven’t the common sence that if I don’t understand something I would not go and ask someone be it a doctor a friend or whoever about it.
    Thank you for listening to my views, and I understand we may not agree but in our countries we can agree to disagree.


  2. Thanks for the comment.

    As to the first point, yes, much more research is being done with people who aren’t of European ancestry now, but the vast bulk of the work linking SNPs to disease has been done with Europeans. At least two of the projects you mention, the Genographic Project and the Sorensen project, are expressly aimed at understanding human history and genealogy, respectively. They are looking both within and beyond Europe. I’m not sure what you mean by the Genome Project. The Human Genome Project’s human genome sequence was mainly, though not entirely, based on people of European descent (much of it on one European-American named Craig Venter). The HapMap, a major genomics project, did involve people of European ancestry, people from China, people from Japan, and members of the Yoruba people from Nigeria, roughly 1/3 each.

    But don’t take my word on it. The recent GAO report on Direct to Consumer genetic tests talks about this at pages 10 and 11. It says

    “Many of these studies the companies use to make risk predictions apply only to those of European ancestry. Consequently, our fictitious Asian and African American donors did not always receive risk predictions that were applicable to their race or ethnicity, although the companies either did not disclose these limitations prior to purchase or placed them in lengthy consent forms”.

    It continues, giving examples from responses it received from four leading DTC companies.

    “• Company 1 provided Donor 1’s fictitious African American profile with test results based on her race for just 1 of the 15 diseases we compared: type 2 diabetes. For the remaining diseases, Company 1 provided a risk prediction but included a disclaimer, such as “this result applies to people of European ancestry. We cannot yet compute more precise odds” for those of African American descent….
    • Company 2 claimed on its Web site that it had “better coverage [of genes] associated with the most important diseases for all ethnicities” than its competitors. However, the company provided Donor 2’s fictitious Asian profile with test results for just 6 of the 15 diseases we compared….
    • Company 3 sent Donor 3’s fictitious African American profile results for just 3 of the 15 conditions we compared.”

    I think I’m right on this one.

    On the second point, opinions differ a lot on this one, but, yes, in the United States and most rich countries, you do require a doctor to order (to prescribe) most drugs for you. You even have to get a prescription to buy eyeglasses or contact lenses. The drug exceptions are drugs like acetaminophen (Tylenol), where the FDA has decided they are safe and effective enough that they can be used without a doctor’ intervention – “over the counter” or OTC. I don’t think health-related genetic tests are that safe and effective. I


  3. (continued – I posted the last one mistakenly before I was done)

    In general I have a libertarian bias, but much less so in the area of health. Most people have a hard time understanding medical issues and most people aren’t at their best when they need health care – they are often in pain and scared. I wouldn’t try to regulate computer programs or tv shows because people might misunderstand them and done harmful things, but health, to me, is different. And not just to me – our whole health care system is full of unusually strong regulations. Practicing medicine without a license is a crime; so is selling a prescription drug without a prescription. Maybe the whole health care system is wrong in being paternalistic – and sometimes I think it has been, and continues to be, too paternalistic. But, for me, not here.

    Reasonable people can certainly differ on that point, though. Thanks for posting.


  4. In response to your response to Bob’s post.

    I would argue that direct to consumer genetic testing is different from prescriptions for eyeglasses and medications for one very clear reason. With medications and even eye glasses you need to have a prescription because you are actually treating your body with something. If you have the wrong dose of a medication you can actually do the body harm. However, I see direct to consumer genetic testing as different.

    Direct to consumer genetic testing is about providing information to an individual. Not all this information is even medical in nature, some of it concerns ancestry or basic traits like eye color. I feel people should be able to have such freedom to access their data. Companies like 23andMe provide the raw data as well so one can use second party programs like SNPedia to learn more if they wish. I don’t feel like one should be denied access to their raw genetic information, that information truly belongs to the individual.

    One should not be required to have a prescription to get access to their own genetic data. I don’t see how such information is any more harmful to a patient then say family history. I know diabetes runs in my family for example. How is it any different knowing this than finding out the same from 23andMe?


Leave a Comment