11

Drug Interactions

Analyzing AEs and ascribing the causality to a particular drug can be quite difficult. This difficulty, however, is magnified when the patient also takes additional drugs. This is sometimes known as “polypharmacy.” It is generally believed that the more drugs taken, the greater the risk of AEs and the greater the risk of drug–drug interactions.

In such cases, it may not be possible to ascribe the AE to one particular drug. In most situations of regulatory reporting, the reporter or the company is generally required to specify one or more “suspect drugs” and, if present, one or more “concomitant drugs.” The former are presumed to have a suspected causative role in the AE and the latter not.

Further complicating matters are drug interactions, a situation that occurs when two (or more) drugs are taken that influence each other directly or indirectly. That is, the pharmacokinetics (e.g., blood levels) or pharmacodynamics (effects in the body) of one or all of the drugs may be altered.

For example, the coadministration of desloratadine (Clarinex) and erythromycin, ketoconazole, azithromycin, or fluoxetine in pharmacology studies produced increased plasma concentrations (Cmax and AUC0-24h) of desloratadine and its major metabolite but did not produce clinically relevant changes in the safety profile (Clarinex Package Insert, February 2007). This is an example of a drug–drug interaction producing changes in pharmacokinetics (the plasma levels) but not in the pharmacodynamics (no clinical safety untoward effects).

A patient may suffer from a pharmacodynamic interaction when taking several products that share the same adverse reaction. For example, when simultaneously taking aspirin and clopidogrel (both reduce the clotting mechanism) plus a nonsteroidal anti-inflammatory drug like piroxicam and, unknowingly, another nonsteroidal anti-inflammatory drug like ibuprofen (both weaken the gastric lining and promote bleeding), the patient is at great risk of gastrointestinal bleeding.

At the other end of the spectrum is a drug like warfarin, which can be lifesaving but which has more than 55 potential drug interactions listed by drug class and more than 150 drugs and dozens of botanicals (some of which have anticoagulant properties) by specific name in the U.S. labeling. In addition, several “disease–drug” interactions are listed whereby these specific diseases may produce increases in the Prothrombin Time/International Normalized Ratio (PT/INR). The interactions may produce elevations or decreases in the PT/INR. In some cases, the same drug with Coumadin may produce an elevation in these levels in one patient and a decrease in another patient (Coumadin Package Insert, January 2010). These changes have the potential to produce significant clinical effects by putting the patient at risk for hemorrhage or clotting.

It is impossible, in the course of testing new drugs, to run drug–drug interaction studies against all drugs or even all classes of drugs. At best, sponsors run selected interaction studies against

• the most commonly used drugs that the exposed patients would be likely to take because of their age, diseases, sex, and so on.
  
• drugs that might be expected to produce interactions based on pharmacology data (e.g., cytochrome P450 metabolism) or based on historical data from similar drugs in the class.

These studies tend to be done on healthy patients in short-term clinical pharmacology trials. Drugs that are suspected of producing interactions but are significantly toxic by themselves (e.g., cancer drugs) generally cannot be studied in this manner because of ethical considerations. It is hoped that one day pharmacogenetics may provide better means of answering drug–drug interaction questions.

Most data on drug–drug interactions are based on study of the Cytochrome P450 (CYP) system. CYP represents a large group of enzymes whose function is primarily to catalyze the oxidation of organic compounds, in particular drugs but also lipids, hormones, and other chemicals. The enzymes are found primarily in mitochondria or endoplasmic reticulum in cells and are found throughout the body. The enzymes we are most concerned about are found mainly in the liver and handle the biotransformation/metabolism of drugs in preparation for excretion.

Various drugs may increase or decrease the activity of one or more CYP enzymes by inducing the synthesis of the enzyme or inhibiting the enzyme’s activity. Thus, if a drug inhibits an enzyme, then another drug that is metabolized by this enzyme may accumulate to toxic (ADRproducing) levels. Conversely, synthesis of more enzyme may increase metabolism of the second drug, lowering its levels and thus producing less efficacy (and perhaps fewer AEs). This may be unimportant if a drug has a wide therapeutic window but may be life-threatening if the window is small and critical concentrations of the drug are needed for efficacy.

Action on the CYP system is not limited to drugs but can be caused by herbals, smoking, and some foods. A good example is the herbal Saint John’s wort, which is a potent inducer of CYP3A4. If Saint John’s wort induces more CYP3A4, drugs metabolized by this enzyme (the drugs are referred to as substrates), such as cyclosporine or innadivir, may be cleared more rapidly and have less efficacy. (See Risk of drug interactions with St John’s wort. JAMA 2000;283:1679. There are tables of substrates, inhibitors, and inducers published. See Web Resource 11-1, for example. In addition, the labeling of drugs will discuss drug–drug interactions.)

Note the marked complexity of the issue here if patients are taking multiple drugs. The tables and measurements, which are usually based on studies done in normal individuals during phase I, are really qualitative. They do not indicate whether the changes (induction or inhibition) will be large or small. This is usually due to the great variability in individuals that is seen in the clinical trials. Add on multiple drugs, some of which may inhibit, some of which may induce, some of which may do either, depending on the individual, and it becomes clear that the tables of interactions are at best guides and alerts to pay attention to the possibility of drug interactions. It is necessary, particularly in polypharmacy, to monitor the effects of the drugs, adverse events, and any other clinical issues and changes in the patient. In effect, once more than two drugs are being taken by a patient, it is very hard if not impossible to predict what interactions may occur and their intensity.

For further information, see the excellent section in Stephens’ Detection of New Adverse Drug Reactions, 5th edition, edited by John Talbot and Patrick Waller (Wiley, 2004). Also see the U.S. FDA’s consumer information on drug interactions at Web Resource 11-2. Textbooks on drug interactions are also available. The EMA has an excellent and thorough review of interactions (drug, food, and herbals), covering both in vitro and clinical studies, available at Web Resource 11-3.

As noted above with Coumadin, certain patients who are either debilitated or suffer from certain diseases (e.g., autoimmune disorders, cardiovascular disease, gastrointestinal disease, infection, psychiatric disorders, respiratory disorders, seizure disorders, and others) may be at greater risk for drug interactions, and the more severe the underlying disease, the greater the risk. It should also be noted that there are possible drug–food (e.g., grapefruit juice), drug–nutrient, drug–disease, drug–herbal, and drug–alcohol interactions. It is also worth keeping in mind that drug–OTC interactions may be missed if the over-the-counter products a patient is taking are not asked for by the questioner or remembered by the patient. There have been attempts to mine data on large databases, looking for drug interactions. One method uses the disproportionality scores (see Chapters 7 and 19) for the two drugs in question for an ADR suspected of worsening by an interaction. Individual scores are calculated and then an “interaction” score is determined. This method has not been too successful or widely used. Other methods calculate confidence intervals for each drug and compare them to a confidence interval for a combined “virtual drug” of the two drugs combined in an attempt to estimate the drug interaction effect (see Leone, Magro, Moretti, et al., Identifying adverse drug reactions associated with drug–drug interactions: data mining of a spontaneous reporting database in Italy. Drug Safety 2010;1;33 [8] :667–675; and van Manen, Fram, DuMouchel, Signal detection methodologies to support effective safety management. Expert Opin Drug Safety 2007;6[4]:451–464).

Stay updated, free articles. Join our Telegram channel

Join