Rosenkrans says he worries about the impact of oversimplifying the interpretation of genetic diagnostic tests. For example, Medicare Australia, a government organization that subsidizes health costs for Australians, requires evidence that a lung cancer patient has the EGFR mutation for them to subsidize the drug. "That’s an example of a health authority using diagnostics in an inappropriate way," says Rosenkrans.

What's more, dealing with the complexity of diagnostic tests will become an even bigger issues as the tests become more sophisticated. The EGFR test is relatively simple, for example, detecting a mutation in a single gene. Future diagnostic tests will likely rely on multiple factors, to predict the best treatment or to give a partial prediction of how a patient will react to a drug (see "Heading off Heart Attacks").

According to Brian Spear, director for genomic and proteomic technologies at Abbott Laboratories in North Chicago, IL, the pharmacogenomics community needs to figure out how to deal with the complexity of genomic data in the clinic. "Should we be faithful to the truth, which can be very complicated and difficult to implement?" he asks. "Or should we smooth out the edges to a place where people can use it?...I don’t know the answer, but it’s a choice we’ll have to make."

Lawrence Lesko, director of the FDA's office of clinical pharmacology and biopharmaceutics, says doctors using these tools should weigh the results of genetic tests with other risk factors, such as weight and age. "I see diagnostics as an adjunct to therapy," says Lesko. "People talk about responders and nonresponders, but I think we’re really talking about a spectrum of probability."

Test makers, meanwhile, are working on educating doctors about their diagnostic tests. Swiss drug-maker Roche is launching an educational campaign to teach physicians about a new test, approved by the U.S. Food and Drug Administration in January 2005, which predicts how people will metabolize many popular drugs, including antidepressants, pain medications, and beta-blockers. The test has only recently been distributed to clinical laboratories that process these tests for physicians -- so it's too soon to say how it will fare in clinical practice.