Monday, March 24, 2008

A Gene for Bipolar Disorder? We Wish.

Here's an interesting news item from Associated Press. John Kelsoe, a respected psychiatric genetics researcher at U.C. San Diego, has formed a company, Psynomics, to market a genetic test for bipolar disorder. The science of genetic research is ridiculously complicated, but you can learn a lot more by reading this excellent article published in the journal Science (but you have to subscribe to access it). The last time I covered this topic in The Carlat Psychiatry Report was in November 2005, and you can read that issue, including an interesting interview with UCSF's Samuel Barondes here (free full access).

The bottom line is that this new test appears to be a bit of a scam, and is based on a notoriously unreliable methodology called "linkage analysis." In psychiatry, this technique has resulted in a litany of dramatic "findings" in psychiatric genetics, most of which have been retracted because of failures to replicate. The Science article quotes Francis Collins, head of the National Human Genome Research Institute, as saying that there is as yet no legitimate genetic marker for bipolar disorder. And efforts to replicate Dr. Kelsoe's marker using the state-of-the-art technique of whole genome scanning have failed.

What we have, then, is a $399 test that will inform some patients that they have double the risk of bipolar disorder. They will respond by believing they have a "bad" gene and will presumably request more intensive psychiatric care, even if they have few symptoms of a mood disorder. Within a year or two, as more data is published, the company will be forced to admit that the test has no merit. Here's hoping Psynomics has thought hard about their refund policy.

Hat tip to: Bernard Carroll, M.D.


Bernard Carroll said...

The Psynomics people do make a couple of valid points in their material on the website. First, without specifically labeling it as such, they employ Bayesian principles of test interpretation: the meaning of a result is influenced by whether the person already has psychiatric symptoms. Second, they correctly say that a test doesn't have to be perfect, only that it needs to add meaningful information to supplement what can be established from standard clinical assessment. As we know, standard clinical assessment leaves a lot to be desired in the case of bipolar disorder, with long lags between onset of symptoms and eventual diagnosis.

On the evidence so far presented, however, the Psynomics test fails on the second of these issues.

Anonymous said...

Why does it not surprise me that the genetics "angle" is now being promoted. Pry into the relationships into this company, and I'll bet there is a pharmaceutical company lurking in the backround, suggesting that there is a benefit to their drug being utilized for this alleged genetic presentation.

Let's see in order of the usual suspects: Lilly, Bristol Myers, Abbott, Jannsen, Pfizer.
Interestingly enough, it's the atypicals that come to mind first.

When do we learn of the genetic sequence to define who is the cad?

jennifer said...

I'm very very leery of any test like this.