Steve Murphy is up in arms about a recent email from 23andMe to its customers advertising the use of genetic variants on its V2 chip to predict individual risk of statin-induced myopathy and breast cancer.
Of course, Steve does have a strong financial interest in 23andMe staying as far away as possible from the area of clinical diagnostics, but I share his unease here.
So far personal genomics companies have by and large done their best to steer clear of being seen as “doing medicine”, but this move would seem to put 23andMe explicitly over that line. In the case of the breast cancer variants this was very much an active decision. These are not markers that just happen to be present on 23andMe’s SNP chip; they were deliberately added during the design of the V2 product, along with risk variants for several other moderate to severe diseases (e.g. Bloom’s syndrome and Glycogen Storage Disease type 1a).
Intriguingly, both of these latter diseases and the three tested breast cancer variants are all much more common in Ashkenazi Jews than in the general population – suggesting that 23andMe may well have a particular target audience in mind.
In a post a month ago I explained why testing for “high-penetrance” disease variants (variants with near-deterministic association with disease, as opposed to the probabilistic increases in risk for the common variants 23andMe currently tests) might be a bad move for 23andMe in particular:
It’s unclear whether 23andMe intends to move more heavily into carrier
testing by expanding its range of conditions and associated variants to
lower and lower frequencies. However, it seems to me there are good
reasons to avoid such a move: doing so would seriously undermine the
company’s ability to characterise its service as non-clinical in
nature, substantially increase the risk of litigation for false
results, and potentially catalyse unpleasant regulatory consequences
both for the company and for the field as a whole. In addition, it’s
unclear to me whether extensive carrier screening would sit well with
the overall image of the company – there’s just not much that’s fun about the possibility of discovering potential child-killing horrors lurking in your genome.
Large-scale carrier testing is a product that would no doubt appeal to
many prospective parents, but it would feel pretty jarring sitting next
to cartoon movies about human evolution in 23andMe’s repertoire.
23andMe has carefully and skilfully carved a niche out for itself in recreational genomics, an area where a considerable market exists – as an illustration, Blaine Bettinger has recently noted that one company alone has sold over half a million kits for testing genetic ancestry. With the huge number of markers generated by their genome scan products, consumer genomics providers like 23andMe have the capacity to predict genetic ancestry better than anyone else in the market – and other features, such as the ability to trace the inheritance of simple traits through a family tree, are also highly attractive to the public.
23andMe has both the PR clout and the expertise in complex data visualisation required to completely dominate this substantial market; I’m not sure why it would risk all that by making an overt (and IMO premature) move into the clinical arena.
Of course, 23andMe knows a hell of a lot more about the incipient regulatory landscape of consumer genomics than I do – perhaps this move is not as risky as it seems?