Consumer Genetic Testing Company 23andMe Faces Its Own Test From the FDA

In response to a letter from the FDA, a direct-to-consumer (DTC) genetic testing company in Mountain View, California called 23andMe has agreed to stop providing health data on new purchases of its $99 genetic tests.  Part of the reason the FDA has gone after 23andMe is because the company did not deal with them in a timely fashion.  If you poke the bear, be prepared to get mauled!  But there is much more to this than a fit of pique by the FDA.

The FDA is genuinely concerned that 23andMe is overselling their product.  And for some of what 23andMe tests, they are almost certainly right.  But for a lot of what they report, 23andMe is spot on.  The tricky part will be to tweak the parts that are not well presented without eliminating the many wonderful things the company can tell you.

This also isn’t just a story about an upstart genetic testing company doing battle with the FDA either.  It is the first step in a discussion about how people’s genetic information will be communicated to them once their DNA has been completely read.  And given how cheap and easy this is becoming, that future will be here sooner rather than later.

A Little Bit of Knowledge

Many people are surprised at how little scientists understand about how our DNA defines who we are and what our future health risks might be.  There is so much going on in terms of multiple genes and the environment that we are currently incapable of making good predictions about most people’s risks for complex diseases like Type 2 diabetes or Alzheimer’s just from their DNA.

Now this isn’t to say we know nothing.  We are very good (although still not perfect) at rare, fairly simple genetic diseases like cystic fibrosis and sickle cell anemia.  And we know a few things about the more complex diseases and are learning more all the time.  But most people can’t get tell from their genetic results whether they will get Type 2 diabetes or not.  Frankly, they can’t really even tell whether or not they are at a higher risk for the disease.  Which brings us to 23andMe.

23andMe delivers a set of results based on your DNA that indicates your risk for a whole set of simple and complex diseases.  Presented in the right way, this could be a boon to anyone who takes the test that can handle getting results like this (especially if they are adopted and so have no information on their family history of disease).

They’d learn about what simple genetic diseases they and their partners might pass down to their future children and they’d see the continuing progress on understanding the genetics behind the more complex diseases.  And as an added bonus they’d learn about their ancestry and about some harmless traits like red hair and blue eyes.  How cool is that!

The problem in all of this is how 23andMe reports data about complex diseases.  The FDA is genuinely concerned that 23andMe is overselling this aspect of their testing.  And up to a point, they may be right.

You can see this from a caller’s comments from a recent broadcast of KQED’s radio show FORUM.  This 23andMe customer called in to say that her 23andMe results assured her that she doesn’t have a big risk for Alzheimer’s.  Her 23andMe results do no such thing.

What her results do tell her is that she does not have a common marker, APOE e4, that increases a person’s risk for Alzheimer’s.  It says nothing about the other unknown markers that she might have that could increase her risk for getting this most common form of dementia.  And there are undoubtedly plenty of these.  APOE e4 is partly to blame for many but by no means all cases of Alzheimer’s.

It is hard to know if she came to this conclusion because of the way 23andMe presents its data or just because of a misunderstanding about what genetic results mean for complicated results like this.  But certainly there is always room for improvement in data presentation.

As you can see here, there are an awful lot of words on 23andMe’s page on Alzheimer’s disease and the point about APOE e4 being one of many potential factors is buried pretty deep.  It would be nice if that point were brought up to the top so people wouldn’t either overreact or underreact to their results.  It should be made clear that if they have the marker they are indeed at an increased risk but that if they don’t, they may still be at a higher risk because of other unknown markers.

This sort of misunderstanding can have real consequences.  In the FDA’s letter they suggested that women may go and have their breasts removed because of a genetic result similar to Angelina Jolie’s.  This isn’t a particularly useful example as very few insurance companies would authorize the expense of this kind of surgery based on a single direct to consumer genetic test.

Imagine instead that Type 2 diabetes runs in your family and you get your results back and interpret them to say that you are not at a higher risk for the disease.  So you relax and get tested less often even though you are at a higher risk.  No good will come from that.

It is scenarios like these that are a big part of the FDA’s concern.  People will react to what they think are meaningful results and some harm will come to them.  (I don’t have the space to go into another concern of the FDA—that the test will give a wrong result. Click here for a nice summary of this part of the problem.)

Now having gone through all of this it is important to reiterate that this is only part of what 23andMe tells its customers based on their genetic test.  There is a ton of good information that we don’t want to lose by shutting down the whole test.

Perhaps 23andMe can resolve some of its issues with the FDA by making it clearer that science doesn’t allow them to predict much of anything yet about most complex diseases.  So maybe, for example, make it clear that not having APOE e4 does not mean you aren’t at a higher risk for Alzheimer’s but instead that you don’t have a marker known to increase your risk.  Or indicate upfront that their data cannot accurately predict their future risk for diseases like Type 2 diabetes.  This isn’t as sexy but is still useful information to have.

One possibility would be for 23andMe to have two bins of health data.  One is for the information that is solid.  For example, this bin might include details about your carrier status for simple, Mendelian genetic diseases like cystic fibrosis, sickle cell anemia and so on.  If you and your partner carry one of these, then each of your kids would have a 25% chance of getting the disease.  This is a useful thing to know that 23andMe makes simple and affordable.

The other bin would be for genetic results from good studies that deal with complex genetic diseases that are not yet fully understood.  These results would be presented as what they are—works in progress that might one day be meaningful.   This bin would be filled with much of what 23andMe reports.

Even with these tweaks there may still be misunderstanding because of the complex nature of these genetic tests.  But 23andMe should not necessarily be held accountable for every possible misunderstanding.  With these sorts of tweaks, people taking a 23andMe test will have a better understanding than many people get from their doctor.

As the person who answers genetics questions from the public at Ask a Geneticist, I get an awful lot of questions from people who don’t understand the results they get from their doctor about a genetic test their doctor prescribed.  For example, one person wrote in that her doctor told her she was A1298C and not to worry.  She was given no information about the gene that was tested or what the numbers and letters even mean. No matter what its shortcomings, if 23andMe offered such a test it would do far better than this!

As I said earlier, this isn’t just about 23andMe either.  This discussion may set the precedent for how people will receive their genetic information in the future. A big question is whether people should have direct access to these results or if they should be delivered through an intermediary who can help them better understand what their results really mean.  This question needs to be answered soon.

My guess is that here in the U.S. anyone who wants it will be eventually be able to get direct access to their genetic information.  Given that, the best course of action might be to make sure people understand what these tests can and can’t tell you.

A high school class similar to a required health course might be the answer for the next generation.  The tricky part will be arming people 18 and over with the information they need to deal with this complex information.  And 23andMe has taken a step in the right direction with this last point.  They have been involved in a new, free Udacity online course that should be a great start in bringing people up to speed.

(Unfortunately even with this too long a piece I couldn’t cover everything.  I left out the important issue of genetic counseling as well as what FDA regulation of direct to consumer tests means for other medical devices like that pedometer on your iPhone.)

Nice summary of the FDA vs. 23andMe