In its Aug 12 issue, Nature evaluates how Francis Collins, who has taken on the mantle of hyperbolist-in-chief for turning medicine into genetics, has done in his first year as director of NIH. Of course, he's had his genome tested for risks (3 times! Replication is always good). And of course he's credited direct to consumer risk services as showing the way to personalized medicine.
What he's reported to have found is that he's at elevated risk for adult-onset diabetes. So he's hit the exercise trail (though he's still riding his famous motorcycle to the office instead of walking or running) and he's lost 11 kilos, watched his diet, and remarkably, hasn't come down with diabetes during the year.
That this is proclaimed as a success for genotype-based medicine is a travesty of the truth. And here are several reasons for saying so:
First, Francis is 60 years old and not diabetic. The population risk for diabetes by his age is something around 10%, and he doesn't have it, which is actual data not a statistical prediction. Whether the risk estimated for his genotype is above average or not, at his age such information is largely useless.
Second, given his age and that he's escaped so far, if he ever does get diabetes it can hardly be called premature or due to some special risk.
Third, his health regimen was apparently not based on any clinical finding like a glucose tolerance test, so it can't be called therapeutic--it's not personalized 'genomic medicine'. If it was based on a clinical test it was personalized medicine, to be sure, but the same kind that's been the job of medicine since Hippocrates.
Fourth, and above all, if you're overweight, or have diabetes in your family, or don't get enough exercise, then watching diet and getting exercise is a great preventive thing to do regardless of any genotype information. It doesn't have to be 'personalized'. Why? Most diabetes cannot be predicted by genetic risk at all, or not by identified genes, as the GWAS results have very clearly and repeatedly shown. In fact, the low predictive power of specific genetic variants is actually one of the more positive real findings of GWAS!
You don't need genotyping to do what he's done, and his experience gives no support for genetically personalized medicine (it doesn't say personalized genomic medicine is useless, either, of course).
Personalized medicine has its place, as we have said before. Its place has to do with those who really do have identifiable genetic risk and some reason to suspect that. We discussed that yesterday, in the context of genetic counseling, the legitimate medical use of genotypic data.