Is this going to get confusing?

Nah, I think people will understand.

So why are we subjecting people to this?

This week I read about the personal genome service ‘23andMe‘ being asked to “immediately discontinue” marketing of its genetic screening service, by the American FDA (Food and Drug Administration) and I thought that it warranted some discussion.

How nice… Again; why are we doing this? Isn’t this a pretty non-topical blog?

Well, for my 30th I had my genome mapped by 23andMe and in my previous commercial life, I had to read and understand pretty much all the FDA’s detailed regulations on getting a medical device approved. So given all this, I thought I was as qualified as anyone else to given an opinion.

Okay, so as I understand it, 23andMe will (for a $99 fee) take some saliva from you and tell you what traits/genes you have. The idea being that you get a profile of all the things that are likely to kill you, as well as what eye colour you should have.

This is a snapshot of some of my data, to give you some idea of what you get from 23andMe.

This is a snapshot of some of my data, to give you some idea of what you get from 23andMe. Click to make bigger

Err… well, that’s sort of how it works. The idea is a bit more complicated but in essence that’s what the service provides. But they do go to great lengths to stress the fluid/variable nature of the research, and in fact that the results are given as a probability of getting a certain condition based on the current research. For example, my results say that I am 2.22 times more likely to suffer from Bipolar Disorder – which sounds pretty bad. However, next to that quite scary number is the actual % risk, which reveals I have only a 0.23% risk compared to the average 0.1%.

You realise you’re using Bipolar Disorder as an example in a conversation with yourself right?

You’re confusing it with schizophrenia – for which I have ‘typical odds’, apparently (updated: see comment by Dan at the bottom of this post or go read this!). You’re missing the point – for each result you can see exactly how this is calculated, referencing which papers were used for that particular area. They couldn’t be clearer in showing their working.

Schizophrenia example

Okay, so why have the FDA got its knickers in a twist if 23andMe are so careful to explain these probabilities rather than diagnosis?

The actual warning letter (available to read here) focuses on things like ‘false positives’ and ‘false negative’ which are terms given to misdiagnosis. From what I can find, the FDA are essentially saying that 23andMe need to provide documentation to prove that their ‘medical device’ is up to code.

Wait, what ‘medical device’, isn’t 23andMe more than a single test?

Good question. But one that has a mind-numbingly long answer that I will try to summarise without boring people to death. Essentially (in this context) the FDA calls a ‘medical device’ a thing for testing for a disease. So each of the tests for each of the genotypes that 23andMe runs needs to prove that they can detect the disease to a certain standard in something called a 510(k) submission.

But if I understood you before, 23andMe don’t tell you you’ve got a disease – they say you have a genotype which in turn has a probability of causing a trait or disease.

Yes, and that’s the problem. The FDA guidance doesn’t really fit this sort of testing. I had exactly the same problem 5 years ago (I’m a hipster medical device designer) when I was trying to get a PSA test through FDA approval. The test we were developing didn’t say if you had prostate cancer or even if you had a prostate problem, it simply measured the level of PSA in the blood, and it was then up to a clinician (in conjunction with other data) to make a judgement on whether or not that value was significant. But we were sternly told by the FDA that basically, we had to submit the device as though we were telling people they had prostate cancer.

That sounds stupid.

Yup, it is.

So 23andMe are just a plucky company standing up to the stuffy inflexible regulations that can’t cope with their innovative ideas?

Well, no. 23andMe and are big company with a big war-chest of money and an army of lawyers.

Okay, so 23andMe are a big angry company that is standing up for the little guys and everyone is totally 100% behind them.

Also, no. The letter from the FDA really is broken, bureaucratic and stupid but it does bring to the forefront important debate about the service 23andMe offers. Many people are uncomfortable with the idea that they are essentially providing an at-home way of discovering you have, for example, the genotype for early onset Alzheimer’s. That can be pretty crushing news, especially to someone who has watched a loved one go though Alzheimer’s and now thinks they may have to suffer the same way. There is a growing call for this news to be given to people by a genetic counsellor, who can put this kind on thing in a context beyond simple percentages that you can read on the internet.

Right, how about “It’s a complex issue that has two sides…”?

Maybe. Personally, the debate over genetic counselling and the need for tighter regulations is one that I think should be happening all the time with work like this. The FDA warning letter really has nothing to do with that and has only really shone a big light on personal genomic services. Essentially they are two separate criticisms and it would be unfair to group a genuine desire for measured discussion with the bureaucratic pidgeon-holeing that is the FDA guidance.

Okay, last try… “The FDA are d**s.”

In this case, pretty much, yes.

You are never happy.

Hi there!

Sign up to receive awesome cartoony content in your inbox, every month.

We don’t spam! Read our privacy policy for more info.


Dan · 27 November 2013 at 16:34

Just wanted to say, there’s a common misconception that schizophrenia == multiple personality disorder (dissociative identity disorder). It’s not true. I know people with the former, and while some (not all) of them hear voices, none consider the voices to be themselves. Rather it’s as if someone else is whispering things to them, but they can never find the person doing the whispering. A lot of people with schizophrenia don’t hear voices at all, although they do still have a number of other issues that make it hard to function as a member of society. DID, as best as I can tell, is when you either hear voices and identify them as being ‘yourself’, albeit a different yourself to the one in control of the body currently, or when different personalities take control of the body at different times and have no memories of those times when another personality is in control.

Here’s an interesting paper/article about it: http://www.ncbi.nlm.nih.gov/pmc/articles/PMC1914490/

Matthew (@MCeeP) · 27 November 2013 at 16:48

You are quite right, I’m sorry for my lazy writing. The original edit said “hollywood schizophrenia” which isn’t much better either.

I’ve removed my earlier schizophrenia misconception and linked to that excellent article.

Thanks for pointing it out

Self publicising my paper · 10 July 2014 at 13:00

[…] it had a pretty luke-warm reception last time but I’m optimistic that despite not changing anything, people will love it. Besides, I […]

Leave a Reply