Direct to consumer (DTC) genetic tests: Here Comes the FDA

Could you guess what this is from one of its hairs? That’s about how predictive current genetic tests are for complex diseases like prostate cancer or heart disease.

As I talked about in my last blog, direct to consumer (DTC) genetic tests are taking a beating right now from the FDA. Part of the problem has to do with some snake oil salesmen contaminating the whole field. But part of it has to do with the data that is available right now and how it is presented.

In this blog, I’ll deal with the data itself and why the FDA has such a problem with it. Next blog I’ll try to come up with ways to still get that data to consumers without ruffling any regulatory feathers.

A big problem with some of the current tests is that people will get different results based on the company they use. For example, the FDA reported that the same man was told he had an increased risk, the usual risk and a decreased risk for prostate cancer depending on the company he used. The FDA hates this kind of stuff.


To them (and many people), these kinds of results say that these tests aren’t working correctly. And given their experience with medical testing, they’re right.

A real medical test should give consistent results. If you take a test that looks at your cholesterol levels, your heart attack risk should be the same no matter who administers the test. Or if you take a genetic test to see if you are a carrier for sickle cell anemia, then again the results should be the same no matter who does the testing.

The genetic test for prostate cancer isn’t anything like these tests because it isn’t a medical test at all. It would be kind to even call it a work in progress. It is really just some interesting information at this point.

This is because we don’t have a good handle on the genetic risks for prostate cancer risk. There are almost certainly lots of different glitches in lots of different genes involved in increasing someone’s risks for getting prostate cancer. All of these variations need to be factored into a man’s risk for prostate cancer. And we only know about a few and those few have a pretty limited impact.

To add to the problem, different companies also look at different glitches or SNPs. So one company might look at one or two SNPs and say a man has a 1.2 fold increase in getting prostate cancer. And another might look at a different two and say he has a 1.1 fold decrease in getting prostate cancer. Different results but neither is particularly meaningful in predictive sort of way.

It is a little like trying to figure out what an elephant looks like from just the tip of its tail. And comparing that result to someone using an elephant’s eye lash to figure out the same thing. Odds are you’re not going to end up with the same animal. Just like you won’t end up with the same result from these kinds of genetic tests.

So when you have one of these tests done for complicated diseases like prostate cancer, diabetes, back pain, heart disease, etc., you’re only getting one piece of genetic information when you need 10 or 100 to figure out your actual risk. You have the eyelash or the tip of the elephant’s tail instead of the whole skeleton.

Now I don’t want you to walk away thinking that I am advocating that these companies shouldn’t give customers these results. I’m not. People should have a right to this information if they want it.

What I am advocating is that these companies come up with some way to explicitly state that these kinds of results are not useful medically. They need to be careful how they present the tests they offer and how they present the results. They don’t want to inadvertently lure customers into taking a test that can’t give them what they want.

37.7749295 -122.4194155