23-not-for-me: My thoughts on direct-to-consumer genetic testing

As a geneticist, I’ve been studying the genomes and genetics of other individuals (and species) for years now. From time to time, I have been curious about learning about my own genome. What’s hidden in the base pairs of my DNA sequence and what does it say about me? Do I really want to know? What would I do with that information if I had it?

Now, anyone who’s willing to fork out $99* can get their genetic information easily through companies like 23andme. If you haven’t heard of 23andme yet, you will as they’re now advertising on television. It works like this: you spit in a tube and a month or so later, you find out if you carry genetic variants in your DNA that may be associated with increased risk for certain diseases, may affect your response to certain drugs or a number of other traits. You also find out information about your ancestry.

Just to be clear, you don’t find out the sequence of your entire 3 billion base-pair genome. You get sequence information for about 1 million sites that are known to vary among individuals (called Single Nucleotide Polymorphisms or SNPs, pronounced ‘snips’) and have been studied for association with various phenotypes (things we can detect in people, like physical traits or health conditions, that are a result of genetics or genotypes).

That sounds like a lot of data. And I love data!

So, am I going to spit in a tube and wait for my own genetic data?

I don’t think so! At least, not yet.

The major reason is that we have limited information to really understand the impact of different variants. Yes, we’ve been able to identify a lot of variation among humans and, yes, we’ve even done studies that link certain variants with an elevated risk for common health problems like diabetes and heart disease. But most of these studies show weak associations with modestly elevated risk of disease.

With a few exceptions, there is very little actionable information that one can learn beyond current recommendations for a healthy lifestyle. What would the recommendation be for someone who carries a variant that’s associated with elevated risk for diabetes or heart disease? The answer: healthy diet and exercise. What’s the recommendation for the general population to reduce risk of diabetes and heart disease ? The answer: healthy diet and exercise. And what if you find out that you’re not at risk for heart disease or diabetes based on the variants that were tested, does that mean you won’t get heart disease or diabetes? No, not necessarily! So how much does knowing you have these low impact variants actually help?

Let’s talk about data again. It is a lot of data. But is it enough? If the genetic testing tells you that you don’t have the variants that put you at risk for breast cancer, are you really safe? Having worked in this field, I know that there are literally thousands of mutations across tens of genes (that we know of so far) that can put you at risk for breast cancer. A negative result from a few variants tested in such tests may give someone with family history of breast cancer a false impression that they are ‘safe’ from the disease.

A big concern is that, for some folks, this kind of testing (direct-to-consumer) may take the place of traditional genetic testing which is done under the guidance of a physician and a genetic counselor, who can better interpret and explain risk of disease. Based on your family history, your physician and genetic counselor can determine if you need to get genetic testing done for disorders known to have strong genetic basis (like predisposition to breast cancer or Huntington’s disease). I would argue that knowing and informing your physician of your family history is paramount.

So what about you? Should you send away spit for your genetic information?

Sure, if you want. But, I would think of it mostly as entertainment only (for now). It could be fun to find out about your ancestral genetic background. My own family history can be traced back for a few hundred years but I might be curious to see who my ancestors were further back in time. And, one may argue that gathering genetic data from a large number of individuals may actually help us better understand impact of genetic variants. But as far as I know, there’s no medical history or medical tracking that’s associated with your genetic information, which would be needed those kinds of studies.

Bottom line, if you decide to fork out a hundred bucks, just be informed about how much weight to put in the information you get. I wouldn’t base any major health decisions based only on these sorts of tests. Personally, I would wait until there’s more information available that will help us better understand the impact of genetic variants that contribute modestly to disease states. And if you suspect you may carry a disease-causing mutation based on family history–talk to your doctor!

*Actual cost can vary depending on how long you want access to the data. The cost of the kit and testing is $99.

4 thoughts on “23-not-for-me: My thoughts on direct-to-consumer genetic testing”

  1. Hi,

    Your arguments of why you’d not be interested in getting your 23andMe results are logical and everyone should consider the points you make about what the data will and won’t be able to reveal (now or in the future). And I definitely agree with your advice that a person should consult their doctor and/or a genetic counselor if they are concerned about their health or genetic risk for a disease.

    However, you also say “And, one may argue that gathering genetic data from a large number of individuals may actually help us better understand impact of genetic variants. But as far as I know, there’s no medical history or medical tracking that’s associated with your genetic information, which would be needed those kinds of studies.”

    Are you aware of the research 23andMe does? You can learn more here: https://www.23andme.com/research/ and read about some of the findings they have made here: https://www.23andme.com/about/factoids/. Like many skeptics you might scoff at some of the more “fun” things they have investigated, but one should not dismiss findings like the Parkinson’s risk variant association published here: http://www.plosgenetics.org/article/info%3Adoi%2F10.1371%2Fjournal.pgen.1002141

    Full disclosure: I used to work at 23andMe, though not on the research side of things.

    1. Thanks for your comment ED! I definitely don’t scoff at the fun stuff–I think it’s interesting! With respect to medically related info–I do agree that there is a lot of research potential if the appropriate data are gathered by following customer health records/history over a long period of time. The Parkinson’s study you refer to is interesting but the methods indicate that most of the ‘cases’ were recruited and had been previously diagnosed by a physician. So again, this means the majority of folks that get testing done for their own personal genetic info aren’t involved in those studies (except perhaps as controls?).

      1. There’s lot of surveys about lots of different stuff (some clinical, some “fun”). If you don’t have the thing under study, then yes, you could be a control. There are lots of common things that people might have and be able to contribute data on that don’t require that they be specially recruited as was the case with the Parkinson’s community.

        Someone more knowledgeable than me would be needed to debate the merits of the exact research methodologies used. All I wanted to get at was that when you say that there’s no medical history or medical tracking being associated with the genetic data, that’s not really true. There’s actually quite a bit of self-reported information collected.

        Thank you for the thoughtful post!

        1. Yeah, absolutely there are a lot of surveys for medical information at the time of the test. By medical tracking, I was referring to continuous tracking of medical history/record beyond the snapshot at the time of the test. For example, how many of the folks who carry a ‘higer risk’ allele for diabetes and don’t have it at the time of the test actually go on to develop diabetes (in the 23andme dataset)–so longterm follow-up. If long-term tracking is implemented, those studies would provide a lot of valuable information.

Leave a Reply