Jessica L. Gören, PharmD, BCPP
Associate professor, University of Rhode Island
Instructor in Psychiatry, Harvard University
Clinical psychiatric pharmacist, Cambridge Health Alliance
Daniel Carlat, MD
Associate clinical professor, Tufts University School of Medicine
Dr. Goren and Dr. Carlat have disclosed that they have no relevant relationships or financial interests in any commercial company pertaining to this educational activity.
During the heyday of the SSRI wars, every pharmaceutical sales rep was educating us about drug interactions. Zoloft and Celexa reps would gloat about how “clean” their drugs were, while Paxil reps would try to shift the conversation to a discussion of social anxiety disorder. Now that most SSRIs have gone generic, the reps have stopped pushing them, and we have been hearing a lot less about drug interactions—but that doesn’t mean they’ve gone away. In fact, drug interactions are common in psychiatry. The task of keeping track of interactions has become less daunting with the advent of free software from companies like Epocrates (www.epocrates.com) and Medscape (www.medscape.com), which allow you to type in every drug your patient is taking and find out if there is a potential interaction. But there are various problems with such computerized databases. For one, they tend to be overly inclusive, often listing every conceivable interaction, no matter how unlikely. For example, citalopram (Celexa), an SSRI considered by most of us to be a pretty safe choice in combination with just about any drug, looks pretty dangerous in the Epocrates database. Cross referencing it with just about any mood stabilizer, antipsychotic, or antidepressant yields a host of red flag messages, often involving an increased risk of serotonin syndrome, neuroleptic malignant syndrome, and that apparently common citalopram side effect—SIADH! Second, since your computer has not personally evaluated your patient, it can’t know what kinds of symptoms to look for, and which potential interactions to focus on. For example, if your patient is jittery and tremulous after having started an SSRI, you will know to focus on drugs that can cause serotonin syndrome—but your computer will not. In this article we’ll survey those drug interactions that are most likely to become troublesome in day to day psychiatric practice. But first, we’ll begin with a primer on the two large categories of drug interactions possible: pharmacodynamic and pharmacokinetic.
Pharmacodynamic Interactions Pharmacodynamic interactions operate at the level of neurotransmitters and mechanisms of action. For example, clonazepam (Klonopin) makes people sleepy by stimulating GABA receptors. Quetiapine (Seroquel) also makes people sleepy, probably by blocking histamine receptors. Combine the two, and patients become really sleepy. Other times, pharmacodynamic interactions may cause two drugs to oppose one another. Antipsychotics work by blocking dopamine receptors. Stimulants enhance dopamine release. So what happens when they are used together? Well, the answer depends on many different factors (eg, tightness of drug-receptor binding, relative concentrations of the drugs at the site of action, etc). So in some patients, the antipsychotics may, at least theoretically, be antagonized by the pro-dopamine effect of stimulants. (For a more detailed discussion of this issue, see http://bit.ly/f5YLkH.) While we may not realize it, we account for pharmacodynamic interactions on a regular basis in our clinical practices by doing things like lowering doses, choosing alternative medications, and increasing visit frequency. For example, when a patient who has been on long term clonazepam for generalized anxiety presents with depressive symptoms, we generally try to avoid sedating antidepressants in order to prevent daytime sleepiness. Here’s a fancier example of using knowledge of pharmacodynamics to our advantage: Levodopa is a good treatment for Parkinson’s disease, but it can cause psychosis by revving up dopamine. Rather than decreasing the dose of levodopa, clinicians often turn to quetiapine to antagonize levodopa’s pro-psychosis effect, while sparing its positive effects on movement. In our experience, many psychiatrists consider these adjustments as simply the “art of prescribing,” not realizing just how skilled they are at understanding and managing pharmacodynamic interaction.
Pharmacokinetic Interactions Pharmacokinetic interactions are hard to predict since they are unrelated to the pharmacologic action of drugs. The effects of the interaction depend on where and when two or more drugs come in contact during drug processing. Drugs can interact with one another at four different junctures:
absorption (that is, the process of getting the drug into the bloodstream),
distribution (ferrying drugs to different tissues once they’ve been absorbed),
metabolism (dismantling drugs into simpler components), or
excretion (sending drugs into the sewage system).
We’ll discuss each one in turn, focusing on some common examples in psychopharmacology.
Absorption. Drug-food, rather than drug-drug, interactions are most relevant during absorption. For example, ziprasidone (Geodon) absorption is halved when taken without food, which is why we instruct our patients to take this drug after a full meal (at least, we should be doing this!). Food also speeds absorption of both sertraline (Zoloft) and quetiapine, but only by 25% or so, usually not enough to be clinically relevant. Meanwhile, food famously slows absorption of erectile dysfunction drugs such as sildenafil (Viagra) and vardenafil (Levitra)—but not tadalafil (Cialis).
Distribution. Valproic acid (Depakote) is highly protein bound, and it is only the unbound portion (the “free fraction”) of the drug that has a therapeutic effect. Aspirin is also highly protein bound, so if your patient combines the two drugs, the aspirin will kick some of the valproic acid off its proteins, causing the free fraction of the drug to increase. Standard valproic acid levels do not account for the difference between free and bound fractions, so your patient’s serum level might appear normal, but the actual functioning valproic acid can be very high, potentially causing side effects. One way to account for this interaction is to order a free valproate level (with the normal therapeutic range being about 5 mcg/ml to 10 mcg/ml, much less than the total valproic acid therapeutic range of about 40 mcg/ml to 100 mcg/ml).
Excretion. Lithium, unlike almost all other drugs in psychiatry, is not metabolized by the liver. Instead, it is excreted unchanged by the kidneys. Because of this, various drugs that affect kidney function can severely affect lithium levels. Coffee, for example, speeds up kidney functioning and can lead to lower lithium levels. On the other hand, both ibuprofen (along with other NSAIDs) and ACE inhibitor can decrease lithium excretion and lead to toxicity.
Liver metabolism. Most drug-drug interactions take place in the liver, where drugs are processed in order to render them water soluble, which allows the body to more easily excrete them, either in the urine or feces. There are two phases of liver metabolism. Phase I involves the famous cytochrome P-450 enzymes, or CYP450. These enzymes attack drugs in a variety of ways, such as “hydroxylation” (adding a hydroxyl group), “dealkylation” (taking away an alkyl group), and several others. Unfortunately for those of us trying to remember drug interactions, there are many subfamiliesof CYP450 enzymes, including CYP 1A2, 2C19, 2D6, and 3A4. Phase II metabolism continues the process of biotransformation, relying mainly on glucuronidation—which is rarely a factor in drug interactions in psychiatric practice.
Practical Implications of Drug-Drug Interactions To understand drug-drug interactions, you’ll need to refamiliarize yourself with some basic terms. Drugs are “substrates” of specific enzymes. An “inhibitor” is a drug that binds more tightly to an enzyme than the current resident. This “victim” drug then gets stuck in a game of metabolic musical chairs as it scurries around looking for a free enzyme system to break it down. Since this drug is not getting metabolized as quickly as it otherwise would, its serum levels become higher than expected. “Induction” happens when the inciting drug stimulates the production of extra enzymes. With more enzymes around, the victim drug is broken down more rapidly, leading to lower levels. But since it takes a while for all this extra enzyme synthesis to occur, induction, unlike inhibition, does not happen immediately, but takes place over a one to three week period. Now that you know the basics, how can you most efficiently apply them to your practice? Here are some tricks.
Identify the 10 drugs that you most commonly prescribe, and memorize the major drug interactions for each one.
Antidepressants, antipsychotics, antibiotics, antiretroviral, and older anticonvulsants have a high likelihood of significant drug interactions—so be particularly vigilant if your patient is taking any of these.
Recognize the drugs with narrow therapeutic windows, ie, when the toxic dose is not much higher than the therapeutic dose. Commonly used narrow therapeutic window drugs include lithium, carbamazepine (Tegretol), warfarin (Coumadin), digoxin, phenytoin (Dilantin), and phenobarbital (Luminal).
Recognize drugs that have serious side effects and outcomes if blood levels are significantly decreased or increased (eg, oral contraceptives, lamotrigine (Lamictal), clozapine, TCAs, warfarin).
Drugs with long half-lives (eg, diazepam (Valium), aripiprazole (Abilify)) can be particularly troublesome when involved in drug interactions, because metabolic inhibitors can make them ultra long lasting. Be cautious with any new or rarely prescribed drugs, simply because neither you nor anybody else has had much experience with them, and unreported drug interactions can appear.
The risk of drug interactions increase as the number of drugs increases. Setting a threshold to check for interactions is helpful (eg, any patient on three or more drugs).
Another important concern with drug interactions is timing. Inhibition happens quickly. It can occur with the first dose of a medication and it can subside quickly. How long it takes to subside depends on the inhibitor’s half-life—generally, the inhibition will stop after five half-lives. For induction to occur, the body has to synthesize more CYP450 enzymes, and this can take up to four weeks. This accounts for the delayed “auto-induction” of carbamazepine. Conversely, for induction to subside, these extra enzymes need to be broken down. That process can take weeks to occur. As a general rule of thumb, any drug prescribed with an inhibitor should be started at half the usual dose and titrated more slowly. Conversely, a drug prescribed with its inducer may need to be dosed higher after the few weeks it takes for induction to occur.
For a useful table with common drug interactions, click here.
Useful References for Drug Interactions The following programs allow you to input a group of drugs to check for interactions: