Monday, March 3, 2008

New England Journal, Prostate Cancer and Babel

Remember when I said that all of these association studies had weak Odds Ratios? I also said in the Sherpa's golden rules of genome wide association study that any OR less than 2 is probably not better than a family history. Here we have a study in the NEJM listing a powerful combination of SNP data AND Family History. This was e-published back in January, but I draw your attention to it again as it deserves notice.

In a significant meta analysis it is shown that the OR if you have a first degree relative with prostate cancer is 2.5 I hope Genome-Boy and his trusty side kick Prosty are reading!

Well, this study and its shortcomings...There are some. This study Blows mere family history out of the water. This study, dubbed CAPS, evaluated Prostate Cancer in Sweden.

The analysis of SNPs revealed 5 SNPs which had significant risk implicated...Here's the kicker, if a person has 4 SNPs and Family History, then your Odds Ratio for Having Prostate Cancer is.....get this 9.46 compared to the men who had none of these factors.

Take That PSA and Digital Rectal Exam!

Now where does this study have shortcomings?

1. It is retrospective and this is subject to bias, therefore needing prospective analysis before we will use it.

2. This population is a relatively homogeneous population that breeds nationally

3. Only one of the SNPs has an identifiable gene. Without a gene, we can only guess what role the SNP may play let alone devise a medication or treatment to offset these effects

The Sherpa Says:
This is what I am talking about! When replicated prospectively...and this will be, this will be a powerful tool to use for risk stratification. To my journalistic friends, please don't report the Odds Ratio as if it were a relative risk! To my prostate prone friends....cheer up. Prostate Cancer is rarely a killer.


Misha said...

Maybe a brief tutorial on OR vs RR is in order.

Steve Murphy MD said...

Brief??? I concur...