Thursday, August 12, 2010

Benefits to a 3-tier system for DTC genetic testing

The popular genetic testing company 23andMe has two rankings for genetic associations present in the scientific literature: Established Research Reports and Preliminary Research Reports.  The relatively recent GAO report on genetic testing claimed that 23andMe provided "reports that showed conflicting predictions for the same DNA and profile, but did not explain how to interpret these different results" (and the response from 23andMe can be viewed here).  However, I think this system provides a useful basis to improve education about genomics research and genetic testing.  In fact, I think it would be even better to provide 3-tier system to describe genetic associations.

In particular, I think genetic associations can be classified as "Based upon Preliminary Evidence," "Based Upon Reproducible Evidence," or "Therapeutically Useful."  In this case, I would consider "Reproducible" associations to be equivalent to 23andMe's "Established Research Reports."  I would consider "Therapeutically Useful" associations to be those that have been proven useful in terms of significantly reducing patient mortality or morbidity.

The "Therapeutically Useful" classification is important because there could be many reasons why individuals with a particular mutation may have a increased risk of dying from cancer that is statistically significant, but information about that particular mutation may not be important for informative for making medical decisions.  For example, models for genetic predisposition may be complicated for certain diseases, and it may be necessary to incorporate currently unknown information about other mutations in order to provide an accurate estimate of risk to develop a given disease.   Also, there are cases where environmental factors may be more important than genetic factors.  And the list goes on.

In practice, I think the FDA could play a role in helping define the third category of genetic tests, and I can think of at least two ways to implement this.  First, genetic testing companies could post some sort of "FDA-approved" icon for tests that have shown to produce positive results when applied in a clinical setting.  Second, the FDA could post a listing of genetic tests that have been proven useful through clinical trails and provide a list of appropriate treatments that correspond to a given test result.

There are benefits to having access to genetic information that doesn't necessarily meet the criteria for a "Therapeutically Useful" test.  For example, there may be no "FDA-approved" diagnostic for a particular problem and an experimental prediction may be the best possible resource.  Furthermore, the FDA has indicated that it does not see a need to regulate the release of "raw genetic information," and it is acceptable to communicate information about genetic associations though other means.  For example, the FDA would never censor an article in the New York Time about a new discovery or preliminary result.  Genetic tests that are"Based upon Preliminary Evidence" or "Based Upon Reproducible Evidence" are basically indicating that "Hey, you have this mutation that has been described in the scientific literature."  If it is too confusing to provide separate predictions for each tier of genetic associations, then genetic testing companies should at least be able to provide a list of publications describing a mutation of interest (and possibly provide a brief summary of the findings).

Information from DTC genetic testing can also fundamentally increase understanding about human genetics and contribute to scientific research, as indicated by 23andMe's publication in PLoS Genetics.  In fact, 23andMe could actually help establish "Therapeutically Useful" tests if customers could upload clinical information that is directly incorporated into their models.  Likewise, companies like PatientsLikeMe could help test the therapeutic value of genetic tests if patients could upload their genetic information

In general, I think individuals should take a much time as they reasonably can to research a topic prior to making a life-altering decision.  Even if a diagnostic is 98% accurate, what if you happen to be in the minority that gets a false-positive?  Even well-established tests can have false positives.  Important information can be gained from independent tests, consulting with a physician or genetic counselor (or getting second opinions from multiple professionals), or even talking to friends who might have went though similar situations.

Although I understand that a "3-tier" system for genetic testing may be confusing for people at first,  I think this system could be a useful tool to educate the public about genetic testing and encourage individuals to take a more active role in making medical decisions.  In fact,  a recent post by John Timmer concluded with the suggestion that heavy regulation of the DTC testing industry will probably not be necessary if a sufficiently large proportion of the general public took the initiative to better educate themselves.


  1. Quite interesting and nice topic chosen for the post Nice Post keep it up.Excellent post.I want to thank you for this informative post. I really appreciate sharing this great post. Keep up your work.
    Genetic test company

  2. Nice blog post... This blog clearly shows the importance of genetic testing. It is really important to know where to get genetic testing for best and accurate results.

  3. Hi Archibald,

    You touch on some important points in terms of questions about how to define “best and accurate results,” but I have to admit that I almost deleted this comment because I am concerned that a) I believe your link was for one company and b) the description of the “tests” made me concerned that you were providing information about something that may be worse than 23andMe.

    As one positive example, I think 23andMe may have actually provided better information about getting information from a genetic counselor (from than the information for my insurance company (even though I have a PPO, they only listed one genetic counselor providing information on the phone: I would eventually like to see what both options have to say about possibly providing another perspective for interpretation of my genotyping / sequencing data, but I personally have some other things I need to focus on scheduling appointments or calling during available hours (so, I currently can’t say much more at this time).

    That said, I think it is EXTREMELY important that there are warnings about being “For Research Purposes.” I believe 23andMe has made progress in trying to communicate this (although, as is often the case, there is still some room for further progress). However, my own experience is that many bioinformatics methods require testing and critical assessment for each project (especially in the research context, where there often is no “best” solution, and accuracy may be hard to define), and even the combinations of experimental protocols (and different strategies that can be used to analyze the data) can give some noticeably different results. For example, this New York Times article mentions that somebody’s African ancestry estimations could vary from 3% to 45% in somebody who would identify that as their primary ancestry (with, as I understand, higher and more similarity in later tests, but if only defining ethnicity at the broadest level):

    Best Wishes,

  4. I just wanted to follow up with a couple notes:

    1) At first, I didn't see Informed DNA from the link from 23andMe. However, I could find them with the following steps:

    a) Go to
    b) Select "By Phone"
    c) Select your state and specialization as "Adult"

    So, there is overlap between what Aetna recommended and the site linked out from 23andMe. In fact, I initially selected a company that didn’t offer support for non-specific questions (not related to family planning or cancer risk), so the Aetna recommendation may have saved some time to be able to schedule an appointment to discuss a “Genetics Info Session” for 23andMe reports from Informed DNA.

    2) Just to be more clear, here are some partial the 23andMe and Ancestry statistics reported in the NYT article:

    23andMe --> "43.4 percent sub-Saharan African"
    AncestryDNA --> "0 to 58 percent [confident] was from Africa" and "10 percent Benin/Togo, 9 percent Mali, 8 percent Ivory Coast/Ghana"

  5. I am still waiting to confirm the costs for Informed DNA (I was authorized for $99 with Aetna insurance, but there was a "Service Terms (Self Pay) form mentioning "The out-of-pocket cost for your genetic counseling with InformedDNA is $390.00. This covers your initial appointment
    and all follow-up appointments for the same hereditary condition for six months from initial date of service" that I interpreted as a cost without insurance, and I have currently been billed $0).

    Also, I had the following notes from my ~45 minute session:

    -Genetic Counselor was more familiar with the newer 23andMe interface than I was

    -Genetic Counselor informed me about Genetics Home Reference (a resource that I don't believe I previously knew about).

  6. Just as another update:

    1) For Informed DNA genetic counseling, I was authorized for a charge of $99 on 12/7/2018 (while this didn’t appear in the “billing” section for the patient portal, there was an Informed DNA “message” for a transaction receipt in the patient portal, and I confirmed the charge on my credit card). However, I did not receive any additional charges, so I believe talking to a genetic counselor was successfully covered by my Aetna PPO insurance. So, that is good :)

    2) One interesting thing that I didn’t mention in the last update is that my 23andMe report indicates that I am a cystic fibrosis carrier but that wasn’t indicated in my GET-Evidence reports (after uploading my data to the Personal Genome Project)
    -->a. I think this is an issue with the indel format for 23andMe’s “raw data” genotypes and the freebayes .vcf for the Vertias WGS sample (in terms of the GET-Evidence interpretation).
    -->b. Nevertheless, I visualized my .bam alignments (in IGV) with my Vertias WGS data and Genos Exome data, and I could clearly see that I was heterozygous for the 23andMe cystic fibrosis variant.
    -->c. Also, I could confirm this was in fact a rare variant using dbSNP

  7. As a brief additional note, I checked the ANNOVAR annotations (free to those with non-profit e-mail) for ClinVar (those that were "dropped" as a filter), and I could identify the Cystic Fibrosis pathogenic variant (with position "chr7 117149182 117149183 TT - het") from the provided (freebayes) .vcf


Creative Commons License
My Biomedical Informatics Blog by Charles Warden is licensed under a Creative Commons Attribution-NonCommercial-NoDerivs 3.0 United States License.