The Mayo Clinic offers blood tests that screen for genetic variants that influence the way the body and brain handle psychoactive substances. The most important genes on the panel code for enzymes that break down medications. If you have too little enzyme (or none), a normal drug dose might cause excessive side effects. If you have too much, that same normal dose would prove homeopathic.
These differences can be striking. Certain North Africans (and a number of Somali refugees have settled near Mayo, in Minnesota) have multiple repeats of genes for enzymes that break down Prozac, Paxil, and Effexor; normal doses of these medications are almost useless for patients from this population. Almost half of the Japanese, in contrast, are primed for high blood and brain levels of the same substances. A different enzyme cluster handles Celexa, Lexapro, and Elavil. So in theory, it should help to know your profile.
David Mrazek, the Chair of Psychiatry at Mayo and an accomplished researcher, reviewed pharmacogenomics recently for the trainees and faculty here at Brown. The presentation was impressive. It is reasonably clear that within a decade we should all arrive at our doctor’s office with a smart card or a password for a Web-based file that contains our entire genome. To make use of that data, practitioners will turn to computer programs that take into account laboratory data and gene-gene interactions and then make clinical recommendations about such decisions as drug dosing. (The finances of this operation are not clear; the profits may be in the interpretation software more than the gene ascertainment. Institutions like Mayo are trying to place themselves well in this marketplace.)
Looking at the power-point images, I could not decide how important the technology is here and now. The ranges for dose recommendations in the literature, based on enzyme variations, tend to run in the 50% to 150% range. That is, some patients should be given only half the usually recommended dose, and some half again as much. Doctors who are already in the “start low and go slow” camp may do reasonably well without gene data. And since the current algorithms do not take into account complex interactions, it’s not clear how accurate they are in the individual case. Mrazek’s lecture did make me think of a column I wrote for psychiatrists, back in 1993, about patients who do well on very low doses of medication. In a few days, I will put that essay up on line, as a posting on this blog. The new genomics confirms the old clinical wisdom.
My sense for the moment is that I would reserve gene testing for patients who appear to need medication but who do not respond to it or who seem “sensitive” to every drug they try. I do wonder how much that leaning on my part has to do with cost concerns. (The tests seem to run about $300 for one gene and $600 for a panel; near Mayo, insurance covers the cost.) If the tests were free, would I want them in the chart of every patient? As I say, the day will come. Perhaps in three or four years, we will see gene panels for psychiatry that predict drug responsiveness in a more refined manner, with information that goes beyond enzymes to transporters and receptors active in the brain. And then, on the horizon is the whole genome and the challenge of making that data serve us.