Genetic testing to predict medication side effects

Follow me on Twitter @RobShmerling

Medication side effects are a big problem. It's estimated that about half of filled prescriptions are not taken as directed, and a major reason for this is side effects. If you've ever had diarrhea, felt sleepy, or developed a rash after taking a new medication, you know how unpleasant side effects can be. And sometimes it's much worse than unpleasant: drug side effects can cause permanent damage and even be deadly.

Predicting success… and side effects

Wouldn't it be great if your doctor could predict which medication is most likely to work for you and least likely to cause side effects? Pharmacogenetics — the use of genetic information to predict the risks and benefits of a medication — could do just that. The idea is that your genes may provide helpful clues regarding which medication is best in your particular case. There are already examples of this, such as:

• Azathioprine: this is an immune-suppressing medication that some people have trouble metabolizing due to the genes they inherited; a blood test prior to the start of treatment can identify those most at risk.
• Allopurinol: certain ethnic groups (e.g., those of Han Chinese or Thai extraction) are more likely to carry a gene that increases the risk of a severe allergic reaction to allopurinol, a medication primarily used to treat gout.

While these examples deal with medication risks, individual genetic testing may also be able to identify which medications are most likely to help a person based on their genes.

A new study looks at statins

Statin drugs are among the most widely prescribed medications in the world. They lower cholesterol, reduce inflammation, and have been proven to reduce the risk of heart attack and stroke in those at high risk for these conditions. However, a limiting side effect is muscle pain, an annoying symptom that may require discontinuation of the drug. (A more serious muscle disease may develop, especially when statins are combined with other drugs, but fortunately these more serious reactions are rare.) As there are several formulations of statin drugs, for any given person one statin drug might cause trouble while another might not. These variations might also be determined, at least in part, on that person's genes.

Prior research has suggested that people who carry certain genes are more likely to develop muscle pain when taking statins, and certain statins might cause more trouble than others for people with a higher-risk gene. These genes direct the synthesis of a protein involved in transporting drugs into liver cells.

A new study enrolled 159 people who had previously developed muscle pain when taking a statin to determine whether sharing the results of their genetic tests could be helpful in choosing a statin drug that would not cause muscle pain.

The researchers divided study subjects into two groups:

• One group was provided with the results of their genetic testing. If a high-risk gene was found, they were offered a statin considered to be less risky; for those without the high-risk gene, the group was offered any of several statins.
• The other group (the "usual care" group) wasn't told their genetic test results until the study was completed. For this group, decisions regarding statin choice were based on "standard guidance regarding statin selection and dosing."

In the first three months, nearly 60% of those in the first group decided to take a statin; only a third of those in the other group did so. As a result, within eight months cholesterol levels tended to be better in those receiving their genetic test results. The impact of this approach could be large, as all of the study subjects had previously stopped statin medications due to side effects.

Is it in the genes… or the "nocebo effect"?

One interesting aspect of this study is that the "nocebo effect" could have been responsible for at least some of the study subjects' past side effects. The nocebo effect is a phenomenon in which the expectation of a side effect makes it more likely to occur, similar to how the expectation of benefit may make a placebo more likely to work. People who had previously had muscle pain with a particular statin might have the expectation of recurrence with any statin, but armed with genetic information that might help reduce risk, that expectation of trouble might be lessened. Genetic testing could lead to fewer side effects, not only by directing the choice of medications but also through a reduction in the nocebo effect.

We're not there yet

Here's the part where I'm obligated to mention the limitations of using genetic testing to direct drug treatment. First, in most cases, prediction isn't perfect. Some people with a high-risk gene are fine when they take the medication; similarly, those lacking the high-risk gene can still react badly to the drug. One reason for this is that the benefits and risks of drugs are rarely determined by a single gene and many other factors matter, such as other medications taken and other medical problems. Another concern is cost. Many genetic tests are costly and it's often unclear whether the benefits (which may be modest) are worth the expense. It's possible that as genetic testing becomes more common and extensive, costs will come down; and as more genes are studied, the benefits of testing may become clearer (and, hopefully, more robust).

I think it's clear that genetic testing to individualize medication choices holds great promise. It seems likely that at some point in the future, it will be routine.