Dangerous Liaisons

With a large portion of the US population taking multiple prescription drugs and supplements, the increased risk of drug interactions and side effects drives the need for better testing before the medicines reach patients.

All illustrations © raquel aparicio

My mother-in-law moved in with us when she was 82. As her physical condition gradually deteriorated, the number of medications she was taking for various ailments increased: two for high blood pressure, two to promote gastric motility, one for congestive heart failure, one synthetic thyroid hormone, an expectorant, and two inhalers for chronic obstructive pulmonary disease (COPD). In addition, there was the occasional antibiotic for recurrent pneumonia. The drugs were prescribed by at least three different groups of doctors, none of whom communicated with the others. It soon became difficult to tell a new malady from a side effect of one of the drugs, or a potentially harmful interaction...

A year or two into her time with us, she started to have an irregular heartbeat, an arrhythmia. After EKGs, a Holter monitor, and stress tests, the arrhythmia was diagnosed as a side effect of the cisapride that she was taking for gastric reflux. When her doctors replaced the cisapride with another gastric reflux medication, the drug caused a tremor so pronounced that her primary physician thought she had Parkinson’s disease.

Shortly after she stopped taking cisapride, it was removed from the US market for causing cardiac arrhythmia in a number of patients. But the situation was a little more complicated; the serious, potentially fatal arrhythmia that led to the withdrawal of the drug (following at least 80 reported deaths) was more likely to occur in patients who also took another drug that blocked a liver enzyme that eliminates cisapride from the body. With the natural elimination of cisapride blocked, the body would accumulate the drug to dangerous levels. The problem was that this liver enzyme wasn’t just inhibited by one drug. A wide variety of therapeutics could block it, including common antibiotics such as erythromycin and HIV antiretroviral drugs such as ritonavir. If my mother-in-law had been prescribed erythromycin while she was still taking cisapride, it is likely that she could have died from a stroke or from a fall after fainting.

Physicians have hundreds of drug–drug interactions (DDIs) to keep track of, and even that long list is not complete, generally covering interactions that have been experienced by patients, reported, and recognized for what they were. As more drugs become available for various ailments, the potential for drug interactions increases, especially in retirement-age adults who are the highest consumers of prescription medications per capita.

According to a recent study, about 2.2 million adults in the United States between the ages of 57 and 85 take multiple medications, and could be at risk for drug–drug interactions.1 A patient’s thin line of defense consists of the pharmaceutical companies’ requirement to test for dangerous combinations of drugs before they reach the market. But the tests are limited. For example, there are no methods for testing the interaction of 10 drugs concurrently. Instead, we test specific types of interactions that are the most common and harmful. With prescription patterns becoming more complex, many companies, including ours, have started developing new tests to try to capture other types of interactions before new drugs reach the clinical trial stage of development.

Every drug that makes it to the pharmacist’s shelf has been tested and dosed based on how rapidly and extensively it is absorbed, how quickly it gets to its site of action at an effective concentration, and how long it stays active before being cleared by the liver or kidneys—measures that are collectively called a drug’s pharmacokinetics.

So far, so good for individual drugs. But detecting DDIs is not so straightforward. It is rarely a simple matter of one chemical interfering directly with another; the link between drugs is often indirect and complicated by cellular and biochemical compartmentalization. Before the 1990s, the only way to spot a DDI was in a patient.2

The enzyme involved in the DDI between cisapride and erythromycin was discovered back in the 1980s, when the first nonsedating antihistamine, terfenadine, came on the market. Over the course of several years, clinicians realized that the sudden death of some patients taking terfenadine was associated with the drug’s combination with antibiotics as well as other drugs. Eventually, researchers realized that the antibiotics and other drugs were inhibiting cytochrome P450 (CYP) 3A4, a liver enzyme that oxidizes and inactivates terfenadine as part of the body’s normal metabolism of the antihistamine. The CYP family is a hugely important group of metabolic enzymes involved in the synthesis of hormones, membrane lipids, bile acids, and vitamins. It also eliminates cellular toxins and drugs. With CYP3A4 blocked, the terfenadine build-up resulted in a severe cardiac arrhythmia in many patients.

According to a recent study, about 2.2 million adults in the United States between the ages of 57 and 85 take multiple medications, and could be at risk for drug-drug interactions.

In the 1990s, when researchers realized that CYP inhibition was a property of multiple existing drugs, the FDA responded by requiring a series of preclinical in vitro tests for all investigational new drugs.3,4 One of the most effective testing mechanisms that came out of that guidance was the requirement to use human in vitro systems such as human liver microsomes to test for CYP inhibition. Microsomes are prepared by homogenizing human liver tissue and separating out the subcellular components via high-speed differential centrifugation. The pellet contains the endoplasmic reticulum membranes—the site of CYP-mediated metabolism of drugs. This organelle suspension provides a sensitive and effective system for detecting inhibition of CYPs by a drug, as researchers can readily tell when the concentration of a probe substrate diminishes (with active CYP) or remains constant (with CYP blocked).

Such interactions didn’t preclude drugs from being prescribed together; clinical pharmacologists could develop dosing schemes that took those interactions into account. For example, inhibitors could be classified by the strength of their inhibition of different CYPs, and appropriate dose adjustments could be made to other drugs that were administered at the same time.

Today we know that we also have to take the inhibiting drug’s mechanism of action into account: is it a reversible vs. irreversible inhibitior? Reversible inhibitors block CYP function only as long as the inhibitor is present in the bloodstream, whereas irreversible inhibitors inactivate an enzyme permanently, knocking out its function until the cell produces more enzyme, which could take hours or days.

The FDA relied on these and other in vitro tests, saying that results indicating no interaction are sufficient to rule out the need for a clinical DDI study. Positive or borderline in vitro results, on the other hand, indicate the need for such a clinical study in healthy human volunteers. While these assays, which remain the industry standard today, were certainly an improvement over discovering DDIs in patients, they still fall short of catching all of the clinically observed interactions.

By the year 2000, researchers thought they had a pretty good handle on which metabolizing enzymes had to be tested for DDIs. The FDA required in vitro data on the interactions with human drug-metabolizing enzymes, with the understanding that members of the CYP superfamily were involved in most cases. But metabolizing enzymes were not the only mechanism for eliminating drugs from the body.

One clue, although nobody knew it at the time, was the discovery in 1976 that drug-resistant mutants of a mammalian cell line expressed more of a membrane-bound protein called P-glycoprotein (P-gp) than wild-type cells. Before long, it became clear that P-gp was an efflux transporter that ejected drugs from a cell and played a role in the resistance of cancer cells to chemotherapeutic drugs. The human gene encoding that protein was named MDR1 in recognition of its role in multidrug resistance in cancer.5 A number of other mammalian efflux transporters, or pumps, have since been discovered, first in cancer cells and eventually in normal cells. These pumps are important for drug uptake, distribution, and clearance (collectively called a drug’s “disposition”), specifically in pharmacologically crucial organs such as the small intestine, blood–brain barrier, liver, and kidneys.

Once researchers appreciated the central role played by overexpression of P-gp and other transporters in drug-resistant tumors, they turned a spotlight on this class of proteins. Soon drug companies were looking for ways to block the activities of the efflux pumps in order to retain the chemotherapeutic agents in the tumor cells longer—thus improving the efficacy of the drug. In a sense, they were trying to create an intentional drug–drug interaction. The problem was that, due to the fact that P-gp is expressed in multiple pharmacokinetically important locations, the concentration of the chemotherapeutic agent increased everywhere in the body, leading to side effects so intolerable that an increase in therapeutic efficacy became a moot point. To this day, no company has successfully developed a P-gp inhibitor that achieves the desired effect on tumors without the undesirable systemic effects.

It gradually became clear that these proteins played a role in the disposition of many types of drugs by the body—not just chemotherapies—which meant that they could also play a role in unintentional drug–drug interactions, and might be a way to explain interactions that had to date gone unidentified.

Drug Metabolizing Enzymes
Researchers first learned the role of metabolizing enzymes in drug-drug interactions when some patients died from the combined administration of an antihistamine and an antibiotic. The antihistamine terfenadine was normally metabolized by the enzyme cytochrome P 450 (CYP) 3A4—a fact taken into account to establish a safe dose. But when certain antibiotics—or other drugs—were administered at the same time, they blocked CYP3A4 activity, which caused a dangerous buildup of terfenadine.

Surprisingly, it’s often difficult to tell whether a transporter is involved in a DDI and, if so, to what degree. The few examples of a purely transporter-mediated DDI have involved drugs with a narrow therapeutic range (NTR) that are not metabolized by CYP3A4. NTR drugs have a very small window of efficacy—if the dose is too low, there is no effect and if the dose is even slightly beyond a threshold level, serious side effects can result. In these cases, inhibition of a transporter, resulting in even a slight alteration of the circulating concentration of the drug, could have severe consequences. The best example is digoxin, a drug taken for heart failure and atrial fibrillation. If exposure to the drug is increased by as little as 25% to 50%, side effects including cardiac arrest, nausea, vomiting and diarrhea can occur (with other drugs, increases in exposure of less than 100% aren’t generally considered significant to the patient). When digoxin is taken with a drug such as quinidine (an anti-arrhythmic), which inhibits P-glycoprotein, plasma concentrations of the drug increase on the order of 150%, enough to cause a digoxin overdose in some patients.6

But there may be other transporter-mediated interactions we haven’t pinned down yet. One example is the interaction that occurs in organ transplant recipients. In these patients, cyclosporine is given to prevent immune rejection in combination with statin drugs, which are administered to combat the hypercholesterolemia that is a frequent consequence of organ transplantation. Cyclosporine inhibits both CYP3A4, which metabolizes some statins, and multiple uptake transporters that some statins rely on for entry into liver cells. When taken together, the exposure to the statin can increase as much as 20-fold, with the consequences ranging from muscle discomfort to the potentially fatal condition, rhabdomyolysis—the rapid breakdown of muscle fiber.

One of the reasons that so few DDIs have been confirmed to be transporter mediated is that the in vitro tools available to study them have been less than definitive. A limited number of assay systems and cell lines exist, but each is problematic. For example, human cell lines contain multiple transporters, making it impossible to pinpoint a single receptor. Nonhuman cell lines that overexpress a single human transporter, on the other hand, function on a backdrop of animal transporters, which can also obscure findings, as animal and human transporters interact differently with some substrates and inhibitors. Another problem is that when one transporter is knocked out in a cell, another will often take over its function. In fact, because of the redundant function of many transporters, there may never be a toolbox of specific probe substrates and inhibitors of transporters, comparable to those available for drug-metabolizing enzymes.

Without information about which specific transporter is responsible for a given DDI, drug developers can’t design a definitive clinical trial to assess the implications of co-administration of drugs in a human. This is very different from the situation with drug-metabolizing enzymes of the CYP superfamily, where many drugs and other chemicals act as highly specific probe substrates or inhibitors to isolate the offending CYP enzyme.

Drug Transporters
Rather than metabolize drugs, transporter proteins simply shuttle them either into (uptake) or out of (efflux) a cell. For example, the efflux transporter P-gp shuttles digoxin—a drug given to patients with heart failure—from kidney cells into urine at a particular rate. When the transporter is blocked by the anti-arrhythmic quinidine, the renal clearance of digoxin is reduced, potentially resulting in an overdose.

In recognition of its clinical importance, the FDA has announced that it “expects” in vitro P-gp interaction data as part of any new drug application filed as of September 2006.4 Not that P-gp is the only important transporter; in fact, a new white paper from an international panel of experts, including some from the FDA, indicates the importance of a number of other key transporters.7 This is the next step in the evolution and eventual finalization of the FDA’s guidance on this topic, which has been in draft form since 2006.

In response, a number of companies—including our own, Absorption Systems—are developing better ways to define the role played by transporters in DDIs. Our approach tackles the problem of trying to identify the most important efflux transporter in a particular interaction by multiple-choice elimination. We start with a human intestinal cell line, in this case one called Caco-2, in which the expression and function of multiple efflux transporters is well characterized. When cultured under appropriate conditions, the cells differentiate into a polarized monolayer that mimics the epithelial cells lining the human small intestine. On the apical surface (i.e., the surface that would be facing the intestinal lumen in vivo), three efflux transporters are expressed: P-gp, BCRP and MRP2. By means of RNA gene silencing, we have knocked down the expression of one efflux transporter at a time.8 Unlike typical in vitro RNA interference, the “transporter knockdown” phenotype is long-lasting and stable. As a result, we now have a panel of cell lines, in each of which the expression of one efflux transporter is reduced.

The utility of this system, which we call CellPort Technologies®, was demonstrated recently to help explain a clinical DDI that was partially responsible for the decision to withdraw ximelagatran, an anticoagulant, from the market. The study identified P-gp as the efflux transporter responsible for pumping ximelagatran and its active metabolite, melagatran, into bile.9 Co-administration of the common antibiotic erythromycin inhibited the transporter, leading to elevated levels of ximelagatran and melagatran, which was associated with liver damage.

While these assays were certainly an improvement over discovering DDIs in patients, they still fall short of catching all of the clinically observed interactions.

By knocking down one transporter at a time, we can test a new drug candidate in the parental cell line and each of the knockdown lines in turn, and by process of elimination see which of the three targeted transporters is responsible for efflux of the drug. Another advantage is that the results don’t rely on the available toolbox of transporter inhibitors, which are nondefinitive. Furthermore, it is an all-human system, unlike several other commonly used cell lines in which a given human transporter is overexpressed in a nonhuman cell line.

CellPort Technologies, as it currently stands, is not perfect. It won’t necessarily enable us to test for interactions among multiple drugs at one time (a feat no current assay system achieves), but it does offer the opportunity to screen for another class of interactions. As the number of drugs that the elderly take inevitably increases, and the more tools we have to interrogate potential drug interactions, the more capable we’ll be to catch interactions before they occur in patients.

It was frustrating to watch my once-vibrant mother-in-law decline, to watch as she was shuttled from one doctor to the next for test after test. By the time we reach our 80s, it is practically inevitable that we will be taking multiple drugs for what ails us. The more we know about how each of those drugs interacts with the transporters and enzymes that process drugs and toxins, the less likely it is we’ll be treating the side effect rather than the disease.

Chris Bode is a pharmacologist and the vice president of Corporate Development at Absorption Systems. He has been associated with the pharmaceutical industry for more than 20 years.

1. D.M. Qato et al., “Use of prescription and over-the-counter medications and dietary supplements among older adults in the United States,” JAMA, 300:2867–78, 2008.
2. B.P. Monahan et al., “Torsades de pointes occurring in association with terfenadine use,” JAMA, 264:2788–90, 1990.
3. U.S. Department of Health and Human Services, Food and Drug Administration, 1997. “Guidance for Industry, Drug metabolism/drug interaction studies in the drug development process: studies in vitro”; available online at http://www.fda.gov...
4. U.S. Department of Health and Human Services, Food and Drug Administration, 2006. “Guidance for Industry, Drug interaction studies—Study design, data analysis, and implications for dosing and labeling”; available online at http://www.fda.gov...
5. R.L. Juliano and V. Ling, “A surface glycoprotein modulating drug permeability in Chinese hamster ovary cell mutants,” Biochim Biophys Acta, 455:152–62, 1976.
6. W. Doering, “Quinidine-digoxin interaction: Pharmacokinetics, underlying mechanism and clinical implications,” N Engl J Med, 301:400–4, 1979.
7. K. Giacomini et al., “Membrane transporters in drug development,” Nature Rev Drug Disc, 9:215–36, 2010.
8. W. Zhang et al., “Silencing the breast cancer resistance protein expression and function in Caco-2 cells using lentiviral vector-based short hairpin RNA,” Drug Metab Disp, 37:737–44, 2009.
9. M. Darnell et al., “Investigation of the involvement of P-gp and MRP2 in the efflux of ximelagatran and its metabolites by using short hairpin RNA knockdown in Caco-2 cells,” Drug Metab Dispos, 38:491–97, 2010.

Interested in reading more?

Magaizne Cover

Become a Member of

Receive full access to digital editions of The Scientist, as well as TS Digest, feature stories, more than 35 years of archives, and much more!