Dr. James Roerig is a clinical pharmacist at the Regions Hospital in St. Paul, MN, where his primary interest is in clinical research and psychopharmacology. His talk today is the potential for drug interaction and other considerations in choosing pharmacotherapeutic agents.

My task today is to talk about drug interactions, and so I'm going to focus on drug interactions with the new antidepressant compounds. I also want to start out talking a little bit about some of the background to some of the drug interaction research, and kind of familiarize ourselves with the P450 system as well as other systems. One thing to keep in mind as we start out is that there are two types of drug interactions. There are two types of drug interactions and we can classify them into two groups: the pharmacokinetic interactions, which we will be talking more about today; but also the pharmacodynamic interactions. And one example of pharmacodynamic interactions is a serotonin syndrome - what we are going to talk about most of the time period is about moving plasma levels by other drugs. But this is an example of a dynamic receptor type of interaction.

Let me give you a real brief case example. We had a patient who was in the community and was receiving paroxetine as their antidepressant, their community physician wanted to put the person on some prenatal vitamins, wrote a prescription for Prenate^®. They went to the community pharmacy, got it filled, and it was filled with Parnate instead of Prenate^®. She then took several doses, presented at midnight with a full-blown serotonin syndrome, with all of these types of symptomatologies, and proceeded to die. While we are worried about drug levels moving we also have some other type of drug interactions one has to keep in mind.

If we look at where these type of drug interactions come from, the serotonin syndrome primarily is associated with monoamine oxidase inhibitors combined with some type of serotonergic compound, usually an SSRI, but maybe other types of serotonin compounds. We have rarely seen reports of this interaction with other types of combinations, but the vast majority are with monoamine oxidase inhibitors.

Well, when we look at metabolism we look at two phases of metabolism. First is a phase 1 type of oxidative-reductive reactions, and then phase 2 the glucuronidation steps; and most of our drug interaction data is going to be associated with the phase 1 type of systems. There are some glucuronidation interactions, and I will list a couple of those in a few minutes.

Well, there are several different systems that are involved with our metabolism of exogenous substances; one is the cytochrome P450 system, which we are learning quite a bit about right now, but I just want to point out there are other systems. The flavin monooxygenase system is also responsible for metabolizing drugs, and we don't know as much about this system. We know that there are probably five of these; we know that they are not inducible. And an example of a drug that has an FMO enzyme involved is olanzapine or Zyprexa^®, which has two P450 enzymes involved in its metabolism, but also the FMO enzymes. So, one of the reasons why we have such variability in drug interactions are there are blank spots in our knowledge about what is contributing to these types of interactions.

Well, to show what a dynamic area this is, I just have listed here, back in 1992 there were about 221 cytochrome genes identified. By 1995 there were about 481, and it's been increasing more and more. There are substantially less than 50 of these, however, that are involved in metabolism, and we're going to talk about 3, 4 or 5 of them today.

These are the types of things that happen with the oxidative reactions. We can take methyl groups off; we can hydroxylate; we can oxidate the ring; we can dealkylate; we can do numbers of things that are all involved in this oxidative reaction that we will be talking about.

The P450 activities obviously have some positive effects. They metabolize off toxins; they metabolize off drugs; so indeed that is a very beneficial effect. However, there are some things that the P450 system does that we might prefer it not do. One of those things is activating pro-carcinogens. The amines formed on charbroiled meat are not yet carcinogenic but are metabolized by P450 enzyme, particularly the 1A2 enzyme, to the carcinogens. Another example is the 2E1 enzyme that metabolizes acetaminophen to its hepatotoxic metabolite. The important part here is that the 2E1 enzyme is very susceptible to induction by, guess what, alcohol. And so this is one of the reasons why people who ingest large amounts of alcohol have a higher risk of liver toxicity if they use acetaminophen for the hangovers or headache.

We also have learned that some people do not have the enzymes. There is variability between individuals, and we generally term this "genetic polymorphism." We will divide people into extensive metabolizers and poor metabolizers, and basically the poor metabolizers don't have the gene. Extensive metabolizers do have the gene. There is more subdividing that we can do, because we have identified that there may be ultra-rapid metabolizers that may have more than one gene, or a duplicate gene; the regular rapid metabolizers; and then a slow group - while they still metabolize better than poor metabolizers, they have a partially functioning gene. So again, data suggests that there is extreme variability between individuals and as much as we would like you to nail this down to an exact science, it is fairly unpredictable at times.

This is an example of the effect of genetic polymorphism on a couple of very common analgesic drugs. These compounds have to be metabolized to their active components in order to provide analgesia for the patient. The enzyme that does this is a 2D6 enzyme. If they don't have that enzyme they can take codeine but they're not going to be getting the morphine generated and therefore they will not be getting much analgesia. So if a person on Tylenol^® No. 3 doesn't have this gene, they will basically be on Tylenol^®. We can change people into poor metabolizers by blocking this enzyme, and we'll talk about several of our antidepressant compounds that block 2D6, and again render these drugs and some other narcotic compounds relatively ineffective. And I have run into that. One particular patient came in with a diagnosis of chronic pain. He was on Percodan^®, and taking a lot of Percodan^®, and one of the reasons he was taking a lot of Percodan^® is that again, he was being treated with a paroxetine a 2D6 inhibitor.

This is an example of a couple of glucuronidation interactions and these are somewhat more rare. We certainly are very concerned, especially in psychiatry, with lamotrigine-valproate interactions because often times lamotrigine is used in refractory bipolar patients and they very likely will already be on valproate. So we have to use a much slower titration schedule, because valproate is going to inhibit the glucuronidation of this compound, and as we know, if this compound's plasma levels move up too quickly the patient is at risk for severe rash. Another example of a lamotrigine interaction are two cases reported with sertraline, and it's interesting, the cases go in different directions. In this particular situation the addition of sertraline caused a doubling of the lamotrigine plasma level with resulting toxicity, although not a rash. And in this situation the sertraline was taken away and plasma levels dropped, despite a fairly hefty dose increase. Again, these are only two case reports, but because of the drug that was involved here, I think that it's important to keep these things in mind and watch literature for more data on those.

Foods can cause problems with our drugs and most of you are probably familiar with grapefruit juice. If your patients are going to start or stop taking grapefruit juice it is a very good idea to look at what kind of drugs they are on. And grapefruit juice is an inhibitor of two enzymes: the 1A2 and the 3A4. Now, I use this interaction to my benefit because when I stop at the gourmet coffee place that's on my drive into work. If I have grapefruit juice at home I don't get quite as big a cup of coffee. I can last longer on that one cup of coffee rather than getting another one and really be fairly buzzed up for most of morning. It can be a problem. If you were on terfenadine and then decided to go on grapefruit juice - in fact, there is a case in Canada where a death occurred because of this - or if you're taking triazolobenzodiazepine or a number of other compounds. And these are just some of the 1A2 compounds. There are also the 3A4 substrates like triazolam and midazolam.

If you like broccoli, brussel sprouts, those types of things, you will induce the 1A2 enzyme and you might actually cause some inhibition of the FMO enzyme. Again, charbroiled beef induces 1A2, and also smoking induces 1A2, and it's from the cigarette smoke and not from the nicotine specifically, so a patch will not do that. We have particular problems in our hospital because we are smoke free. When people come in with schizophrenia for instance, who smoke very heavily, they'll stop smoking in hospital and get the patch. Few of them take the cure so when they go back out again they start smoking again and screw up the metabolism of the drugs that we are giving them.

We even can affect metabolism by what the infant is exposed to in utero, and there's some data regarding a phenobarbital exposure that has been documented to alter enzymes.

Well, when we consider drug interactions and we look at the whole body of data there are a couple of things that we have to keep in mind. One is the clinical significance versus the statistical significance. When you read literature or review data in the literature, you find that sometimes they will identify that there is a statistical difference between area under the curves, or something like that, but you also have to think, is it going to be clinically significant? Variability is involved in all of these types of things: the initial inhibitory constant or the affinity for the enzyme; what type of plasma levels are present; how much drug is there; how much is protein bound and how much is free - it's only the free drug that is going to interact; what is the liver partition coefficient; how much of the drug really gets into the liver where the enzyme may be located; and then the effect of metabolites. And I will show some examples of these things. One example, sertraline, and its dose, it has a dose-related effect on one of the enzymes that I'll show you in a moment.

The people that are at most risk are obviously those people that are older, have multiple disease states, and are taking multiple drug therapies. People can have quite complicated drug regimes so it really is important to identify what the new drug, the new antidepressant compound, for instance, is going to do with those types of things, again, people with renal or hepatic disease, people on single drugs with very potent inhibitory effects, or people on multiple enzyme blockers. So for instance, grapefruit juice, and a couple of other drugs, may be blocking more than one enzyme at any one time and may be particular risk for drug interactions.

This is just a nomenclature of the P450 system; the family, sub-family, gene. So in this example we have a family and a family member No. 2 that's greater than 40 percent amino acid commonality, the actual sub-family is D, and then the individual gene is 6. These are very old or very ancient enzymes that evolved over a long, long time and are very adept at doing what they are supposed to do, and that is metabolizing off foreign substances.

One of the things you see with this is that if you look at the breakdown of the enzymes that are present in the liver, the 2D6 enzyme, which we will talk about for a fair amount, is really a relatively minor amount of the total enzyme. One of the biggest portions of the enzymes are the 3A's enzymes, 3A4 particularly, but also note that there's a large other here, there's a lot of room for us to learn new things about these enzymes.

One of the things that helps us out with a lot of drugs is that quite a few of our agents are metabolized by a variety of enzymes, so in this example imipramine is metabolized by these three enzymes plus 2D6. And so if we go in and inhibit this particular enzyme, these other ones can take up the slack, however if you are working with a drug like desipramine for instance, that is only metabolized by the 2D6 enzyme, if we inhibit 2D6 here, we have a greater impact on the desipramine plasma levels.

This just shows some of the characteristics of the enzymes in terms of their polymorphism, their inducibility, and if you could see which enzymes they are there, you would find that the 2D6 enzyme and, not the 3A4 but some of the 2C enzymes here, are the ones that are polymorphic. And the inducible enzymes are 1A2, the 2E enzyme, which is also polymorphic, and then the 3A enzyme.

This is a brief review of the pharmacokinetics of our newer antidepressants. The basic thing to keep in mind here is the half-life issues. We have a range of half-life's with the very long half-life of fluoxetine and its active metabolite to the very short half-life of venlafaxine and nefazodone. One of the things to keep in mind about half-life and drug interactions is that the drug is going to hang around for a while after you discontinue the drug and you can measure that or get a handle on that by looking at the half-life. It's going to hang around for five half-lives. Now in the case of switching to an MAO inhibitor, we really have to wash out the previous SSRI therapy completely, so that means washing a person out on Prozac^® for 5 weeks or more, and the other SSRIs probably, to be on the safe side, 2 weeks. For the pharmacokinetic drug interactions, it may not be necessary to get all the drug out, you still may be able to use a drug combination, but you'll have to keep in mind based on the half-life how long you're going to have to use a reduced dose of the second agent.

Well, let's move into the enzymes. This is a 2D6 enzyme and it would be very helpful if you could read these drugs over here, however I think I can remember them. They're in your handout. This is the example of the prototype inhibitor of 2D6, this is quinidine, and if we look at our SSRIs that are primary inhibitors of 2D6 they are going to be Prozac^® and Paxil^® and norfluoxetine, a fluoxetine metabolite, sertraline, primarily those drugs.

If we look at these drugs on the right as substrates that are being affected by these drugs on the left, so these drugs on the right have their plasma levels go up. Lets just look at a few that may be significant: donepezil - Aricept^® is involved with 2D6 metabolism; the secondary tricyclics like desipramine and nortriptyline - if you're in a situation where, as Dr. Kelsey mentioned, you want to add a norepinephrine effect to an SSRI - keep in mind that these compounds are going to have to be used in much lower doses with the SSRIs that are indeed 2D6 inhibitors. Here again are the narcotics that I mentioned to you before, and these drugs again are rendered ineffective because parent molecules go up in plasma level but they are not the analgesic component of these drugs. Another compound you may be using in the elderly population is risperidone. Risperidone has a dose-related extrapyramidal side effect risk, so if its plasma levels tend to go up you might begin to see a person developing more EPS.

Well the next slides I'm going to go through rather quickly. This is just data on some of the newer drugs to show that there actually is some data behind them. This is a sertraline/fluoxetine study with people at steady state on desipramine, and added either 50 mg of sertraline or 20 of Prozac^®, and as you can see in measuring the area under the curve, we had a big effect with Prozac^®. We had some effect on sertraline - keep in mind, this is a 50-mg dose, not a 100-mg dose or 150 or something like that. This bar is important to note - it's about 14 days after they stopped the individual test drugs, so even 14 days later after the discontinuation of fluoxetine there were still significant effects that it was having. In fact, the effect of fluoxetine was increasing all throughout the study until the termination, as you might expect because fluoxetine never came to steady state in this 28-day study.

This just shows a number of studies looking at sertraline's effective dose, at least in the middle-aged population. And a fair amount of data and several different types suggest that the effective dose may be greater than 50 mg, so the question would be, what happens with sertraline at higher doses than 50 mg.

And there are several different pieces of data, this concerns me in that as the dose or plasma level got high, sertraline started to approach what the other two SSRIs do to desipramine.

This shows some in vitro data with venlafaxine showing really no effect at all here compared to fluoxetine on the generation of hydroxy desipramine or hydroxy imipramine.

This shows a PROBE study with dextromethorphan and again, this is looking at a 28-day study with a 2-week washout, and you see compared to Prozac^®, venlafaxine has very little activity.

Here's another venlafaxine study doing the same type of thing with a PROBE, the dextromethorphan PROBE.

This is the last study. This is actually with the substrate, so we've moved out of the Petrie dish, we've moved out of the PROBE study in healthy volunteers and now moved into risperidone. And you can see here with venlafaxine, if you look at the total active moiety combination with a metabolite and its parent compound, there is no significant difference. So a number of things suggesting that venlafaxine may be clean.

This is a new SSRI, citalopram. This compound has been - we're finally getting some drug interaction data with this. When it first came on the market we really didn't have very good data. This shows that warfarin did not have a problem, nor did digoxin or theophylline.

If we go to these compounds, we see a little bit effect with imipramine and this is thought to be because of some mild 2D6 inhibition.

Lastly, I wanted to show you this slide. These are a number of patients who were on a tricyclic antidepressant and an SSRI and what had happened to their blood levels. I just want to point out the variability here. And some of these people had very little, if any, increase and some had major increases, I mean we're getting up into the very severe range here about 1000 ng per ml from way down in the 200 range. The people with the stars here are the people that reported new side effects to their physicians, so some of these people had major increases but they did not perhaps tell their prescriber anything that would have clued them in on the fact that this might have been happening. So we're suggesting, as Dr. Kelsey mentioned, monitoring plasma levels in these individuals.

I'm going to bypass this so we don't get off time. This is a 3A4 enzyme, the prototype is ketoconazole, and you have to truncate this because it goes on very far. By the way, these are the graphs of the inverse of the binding constant, and for the 3A4 system, we primarily have nefazodone and fluvoxamine, and a little bit of an effect from Prozac^®'s metabolite. You can see fluoxetine itself has very little effect here but norfluoxetine does have some effect, almost as much as fluvoxamine, so it may depend on that particular individual, how much norfluoxetine they're generating, whether they have an interaction at this particular enzyme.

Well, again if we look in our substrates here we have a lot of drugs and this keeps growing and growing. We have the nonsedating antihistamines, which are off the market now, but you may have just gotten your Dear Dr. letters about cisapride or Propulsid^®. Again, Propulsid^® is a 3A4 substrate. Our triazolobenzodiazepines as well as carbamazepine and calcium channel blockers. What I become more interested in now is that I'm taking Lipitor^®, and the statin drugs are metabolized by 3A4. So when I'm drinking that grapefruit juice in the morning, trying to stay healthy, I have to make sure I take my Lipitor^® at night, perhaps, and I may not get much of an interaction. There are reports of nefazodone causing muscle toxicity here. If you work with any patients that are HIV positive, the protease inhibitors also inhibit this enzyme but they, in addition, are substrates. One thing I'll point out, St. John's Wort affects this enzyme. If you have people that are using herbals, what it does to the enzyme is not block it but it induces the enzyme so it has almost a phenobarbital-like effect or a Tegretol^®-like effect, and that may then lower these plasma levels. Again you see the donepezil here at the bottom.

Well what kind of data do we have with the 3A4 system? This is venlafaxine data looking at terfenadine showing really no differences.

I think the next one shows the plasma levels are superimposable, so this suggest that again, there's not a 3A4 effect with venlafaxine.

If we move to looking at alprazolam, another 3A4 substrate, again with venlafaxine in this case we did not see an effect.

If we move on to look at carbamazepine, another 3A4 substrate, again with venlafaxine there isn't a change that was found.

We look at citalopram and triazolam again a 3A4 substrate, you'll see again a superimposable curve here, suggesting that citalopram does not have a 3A4 effect, which is good news.

We move on to 2C enzymes, and have some very important drugs like phenytoin and the S form of warfarin, as well as perhaps the NSAIDs. The major inhibitors over here are fluoxetine and fluvoxamine as the antidepressant compounds, and in the back of your handout there are various examples of some of these interactions occurring.

We go to the other 2C; we have some significant drugs here also. The beta-blockers are metabolized by a number of different compounds. Topiramate is metabolized by this compound so we would be concerned that if this level went up too fast, we might get some of the CNS toxicity. Again we're looking at fluvoxamine and fluoxetine, so Luvox^® and Prozac^® being somewhat active here - a little bit of sertraline activity in this case.

1A2 is the last major enzyme and we only have one antidepressant that affects this enzyme and that is Luvox^® or fluvoxamine, and it's very potent in inhibiting this enzyme.

So if we look at the types of problems we have, again theophylline, caffeine will rise in their plasma levels, there's the possibility of tacrine levels going up that may lead to cholinergic side effects. We've had clozapine levels go up, and here, again, we're seeing an antipsychotic, haloperidol so if that's being used, then Luvox^® would have been added or, again, grapefruit juice or fluoroquinolones, produce these types of elevations with resulting side effects profile.

Lastly I just wanted to point out that this 2E1 enzyme is inhibited by disulfiram or Antabuse^®. Ethanol is the substrate of this enzyme and ethanol is a major inducer of this enzyme. Again, the acetaminophen, as I mentioned, metabolizes to a toxic metabolite, so induction of this enzyme will make this more toxic. If you have anybody that takes benzene on a regular basis, benzene is going to have a toxic metabolite. So theoretically those people in that situation might benefit from a dose of disulfiram to block the 2E enzyme so the toxic metabolites aren't generated.

So, to review the potentially significant interactions, paroxetine is primarily a 2D6 inhibitor. It does not affect any of the other enzymes in any large degree, and these are some of the drugs that are going to be involved. And there are some others that I don't have listed here, but these are some of the examples of the drugs that could be affected by paroxetine.

Fluoxetine has several different enzymes. We mentioned the 2C, we mentioned again 2D6, similar to paroxetine, and the 3A enzymes, as we discussed, and their corresponding substrates.

With sertraline we have a little bit of 2C effect, we have some 2D6 effect that's dose dependent. On a low dose you may not have any effect at all. At higher doses you may have more effects. So we don't really think we have a 3A4 effect with this compound. And if we look at glucuronidation, we may have an effect with glucuronidation.

Fluvoxamine is a 1A2 inhibitor and a 3A4 inhibitor, so again, a dirtier drug hitting two enzymes, and some C effect, so really hitting two to three or four enzymes.

If we look at nefazodone, it's primarily a 3A4 inhibitor, so one enzyme but a very, very important enzyme.

Citalopram has a little bit of 2D6 effect that's weaker than sertraline and therefore is not going to have much effect on these compounds.

Mirtazapine we really don't know. There's really not a whole lot of data with mirtazapine for us to look at. It may be a weak inhibitor of a couple of different enzymes, but if that's clinically significant we just don't know yet.
And then we have venlafaxine that really doesn't have these drug interactions. A little bit of effect as they've documented in their package insert on desipramine.

Thank you very much.

Return to Agenda