Wednesday, June 10, 2009

Consumer Genomics Spin


Thank God for Dan Vorhaus AKA @GenomicsLawyer on Twitter and a nice wrap up by Emily Singer

But like a groupie believing everything one reads, I want to caution those who attend that conference.

I heard major exaggerations and one might even say misstatements from the people presenting.......

The biggest one upset me. I am a fan of Jeff Gulcher's. I think he is a great guy. And his 1 in 100,000 story about finding aggressive prostate ca because of his SNP risk is very real. But, he claimed something which had been proven incorrect.

It had been proven incorrect months ago......
Maybe he did what DTC does best.....they imply things...... Jeff implied that their SNPs for CVD actually reclassify women's risk......

In order to prove that he would need a cohort and he would have to control for all other things AND he would have to have been studying this cohort for about 20 years.....which to my recollection has not happened with deCode.....

More importantly, the SNP they discovered was not found to add ANYTHING useful to the current classification system for risk of heart attack....


However, there is something which reclassifies approximately 40% of women and 17-20 percent of men......the Reynolds Risk Score, studied by Paul Ridker et.al. up at Harvard....


In this great example, clinical risk factors are combined with genetic factors, in this case Family History and it outperforms Framingham Risk Stratification........


But to state that their tests perform the same or even to insinuate this, well, that's just intellectually dishonest and I wouldn't expect that from Jeff......which means he "must have been mistaken"



So today I look forward for more exaggeration at CGS as I call in to the ICOB meeting for the Coriell Personalized Medicine Collaborative......I implore others to follow Dan @genomicslawyer and look for the same.


The Sherpa Says: We must always listen with a critical ear to what is said by people with money or power or both......

No comments: