Friday, February 18, 2011

Indecent Exposure

We thought we were parodying what's already absurd enough the other day in our post about the atomic bomb being the first Too-Big-To-Fail mega research project.  Actually, nobody suspected that this was the hydra whose heads couldn't be cut off fast enough, but the future was latent in its name: the atOmic bomb project.

Hydra, from Wikimedia Commons
Our past cynicism notwithstanding, Nature tops all that this week, reporting something we should have foreseen but even we in our cynicism did not.  It's the 'exposome'.  Researchers are proposing to wire people up and photograph or otherwise measure every breath they take, every bite they eat, every chemical they come into contact with, collecting both 'external' and 'internal' exposomes, to "reveal the effects of diet, toxins and other exposures", in order to determine which exposures cause which diseases.  

It's a bit of a technological challenge but if there's anything we excel at it's overcoming technological challenges, so we assume the measuring devices will be made (funds will  now likely be diverted to years of R and D to do that), and study subjects will be convinced to wear, swallow, implant or otherwise port cameras, breathalyzers, air quality monitors and so on for months or years at a time, and of course donating blood, urine and fecal samples at specified intervals for genetic and exposure analysis.  Ready your every orifice!

The challenge of creating and maintaining the huge databases this kind of research will yield will be met, and statisticians will figure out new ways to analyze the data and it will all keep many people busy for many years to come.  So much better of a grant bonanza even than the old-fashioned biobank idea, trivially small by comparison.   

Exposomics. Brilliant! 

The tiny little problem with all this is that it's already very easy to predict the results. Let's not even consider the problem that yesterday's risks, being all that we can estimate, are an unreliable predictor of tomorrow's risks. Just as with GWAS, these studies will find effects, but they will be small and explain little and will not be useful for predicting disease.  The few strong ones will be hyped to death (more material for Nature to trumpet, naturally), but most of them will or could have been identified by less exotic means.

Risk factors with major effects are not difficult to identify.  Any Joe Blow, even without a micro-array, can do it.  Single-gene diseases or major environmental risk factors like smoking or lead paint or cholera are readily revealed by current genetic and epidemiological methods.  As the late curmudgeon David Horobin, founder of the non-conformist journal Medical Hypotheses, once wrote, if you can't detect something in small samples (we think he may have said 30), then it's not worth detecting.

That may be going too far, but we already know that when there are multiple factors at play, each with a small effect, be it genes or environmental risk factors, we move into an arena where small samples won't do: either our methods fail, or the answers are unhelpful in any clinical or public health sense, for the same reason: if they are too small to be detectable they are too small, and ephemeral or fickle to be that useful.  Even if we can detect them, which GWAS, biobanks (and, yes, Exposomics) will occasionally do, it is far from obvious that the cost is worth the game.

When risks are very small, as for example, in dental x-rays, and we know that but can't really estimate them, by far the cost-effective approach is simply to restrain use to situations when something that is important is at stake.  That may not be the 100% best-in-principle approach, but in practice it will save far more than it costs.  And the research money can go to providing important dental x-rays for those who can't otherwise afford them.

Omics-itis is bound to spread.  We expect our local deli soon to have a placard outside saying:

Here now!  PeanutButterAndJellySandwichOmics!
Exposomics?  Really, now!

No comments: