Drug-Drug Interactions: Studies In Vitro and In Vivo

Help on accessing alternative formats, such as Portable Document Format (PDF), Microsoft Word and PowerPoint (PPT) files, can be obtained in the alternate format help section.

Contact: Office of Clinical Trials

September 21, 2000

Foreword

Guidance documents are intended to provide assistance to industry and health care professionals on how to comply with the Therapeutic Products Programme policies and governing statutes and regulations. They also serve to provide review and compliance guidance to Therapeutic Products Programme staff, thereby ensuring that the Programme's mandate is implemented in a fair, consistent, and effective manner.

Guidance documents are administrative instruments not having force of law and, as such, allow for flexibility in approach. Alternative approaches to the principles and practices described in this document may be acceptable provided that they are supported by adequate scientific justification. Alternative approaches should be discussed in advance with the Programme to avoid the possible finding that applicable statutory or regulatory requirements have not been met.

As a corollary to the above, it is equally important to note that the Programme reserves the right to request information or material or define conditions not specifically described in this guidance, in order to allow the Programme to adequately assess the safety, efficacy, or quality of a therapeutic product. The Therapeutic Products Programme is committed to ensuring that such requests are justifiable and that decisions are clearly documented.

Table of contents

• Introduction
  • Basic Considerations
  • Drug Metabolism
• Study of Metabolic Drug Interactions in Vitro
  • In Vitro Studies of Drug Metabolism and Enzyme Inhibition
    • Subcellular Fractions of Human Liver Tissue
    • Whole Cell Models
    • Heterologous Expressed and Purified Human Drug-Metabolizing Enzymes
    • Pharmacological Probes
    • Immunochemical Probes
  • In Vitro Studies of Enzyme Induction
  • Choice of Drug Concentration Range
  • Position in the Time Course of Drug Development
• Role of in Vivo Animal Studies
• Study of Metabolic Drug Interactions in Vivo in Clinical Trials
  • Design
  • Population
  • Choice of Potentially Interacting Drugs
  • Route of Administration
  • Dose Selection
  • Endpoints
  • Adverse Events
  • Statistical Considerations
• Correlation Between Studies in Vitro and in Vivo
• Factors Influencing Drug Interactions
• Product Monograph
• Literature References
• Website References
• Appendix A

Introduction

The Therapeutic Products Programme subscribes to the position that an adequate pre-marketing investigation of the safety and efficacy of a new pharmaceutical or therapeutic biological (hereafter to be referred to as drug) should include characterization of its metabolism and exploration of its interactions with other drugs. This guidance document provides suggestions to industry scientists and Therapeutic Products Programme evaluators concerning current approaches to the conduct and regulatory review of in vitro and in vivo studies addressing drug interactions. The suggestions are not intended to be requirements, but rather are offered as items of consideration for scientists and medical practitioners involved in the research or regulatory assessment of drug interactions. The Therapeutic Products Programme recognizes that the investigational approach used for a particular drug may have to be individualized depending on the pharmacodynamic, pharmacokinetic, and safety characteristics of the product as well as its proposed clinical application. As the methodological approaches to the study of drug metabolism and drug interactions are undergoing rapid evolution, regular consultation of the scientific literature is recommended to determine how the status of research in this field has changed since the issuance of this guidance document.

Basic Considerations

Drug interactions are pharmacodynamic, pharmacokinetic, or clinical responses to the administration of a drug combination that differ from the known effects of the individual drugs administered alone. The clinical consequences of drug interactions may be antagonistic, additive, synergistic, or idiosyncratic, resulting in treatment failure, increased pharmacologic effect, or toxic reactions, which may be serious or fatal.

Pharmacodynamic drug interactions result in alteration of the response to one or both drugs without affecting their plasma concentrations (e.g. displacement from receptor binding sites). Pharmacokinetic drug interactions are a consequence of altered levels of exposure to the drug or its metabolites through one or more of the following mechanisms ^{1 2}:

• altered absorption: Drug absorption may be altered by chelation or complex formation (e.g. cholestyramine can form complexes with warfarin, nonsteroidal anti-inflammatory drugs, or sulfonamides), prokinetic or antimotility effects (e.g. the absorption rate of acetaminophen is increased by metoclopramide and delayed by propantheline), or pH effects (e.g. increased pH due to H₂-blockers and antacids can reduce the absorption of ketoconazole) ³.

• altered distribution: Competition of two drugs for the same plasma protein-binding sites can result in an increased free plasma concentration of the lower affinity drug. Significant interactions are possible if an extensively protein-bound drug of low volume of distribution is co-administered with a high affinity displacing drug that is present at a concentration approaching or exceeding the molar concentration of the protein-binding sites (e.g. displacement of warfarin by trichloroacetic acid, the major metabolite of chloral hydrate) ¹.

• altered transport: Drug interactions can result from displacement of drugs from the ATP-binding cassette (ABC) transporter proteins such as the MDR-1 gene-encoded P-glycoprotein, an energy-dependent multidrug efflux pump located on the luminal surface of enterocytes and biliary hepatocytes and the brush border of proximal renal tubule epithelium (e.g. the P-glycoprotein inhibitor, quinidine, increases plasma levels of digoxin) ³, ⁴.

• induction: Induction or increased synthesis of one or more drug-metabolizing enzymes leads to enhanced metabolism and hepatic clearance of all substrates for those particular pathways ⁵. Numerous drugs have been identified as inducers of drug-metabolizing enzymes ^1A.

• inhibition: Inhibition or decreased metabolism and hepatic clearance of substrates for a specific drug-metabolizing enzyme can result from competition between drugs for the enzyme's binding sites, covalent modification of the enzyme by a reactive metabolite, or formation of a catalytically inactive reversible complex between the enzyme and a drug or its metabolite ⁵. Inhibitory effects on one or more drug-metabolizing enzymes have been demonstrated for a range of drugs ^1A.

• altered excretion: Potential mechanisms for the altered renal excretion of drugs include competition for renal anion or cation transport systems (e.g. probenecid inhibits the tubular secretion of the penicillins), changes in urinary pH (e.g. sodium bicarbonate increases renal elimination of phenobarbital), and inhibition of renal metabolism ⁶.

The objective of this guidance document is to address the issue of metabolic drug interactions. Elimination of a drug may occur through metabolism to one or more active or inactive metabolites, excretion of the parent drug by renal or biliary routes, or by a combination of excretory and metabolic elimination pathways. If elimination occurs primarily through metabolism, identification of the principal metabolic route may have important safety implications. Some drugs and dietary xenobiotics have the ability to influence the metabolism of concomitantly administered drugs by acting as inhibitors or inducers of drug-metabolizing enzymes. Thus, drug interactions may have the potential to result in substantial increases or decreases in the blood and tissue concentrations of the drug under investigation or certain of its metabolites.

The clinical significance of the interactions resulting from inhibition or induction will be dependent on the relative pharmacodynamic activity and toxicity of the parent drug and its metabolites. In the case of a prodrug that is converted to an active metabolite (e.g. codeine), pharmacodynamic activity will be decreased by inhibitors and increased by inducers. For drugs that are inactivated by the metabolic process, inhibitors will tend to enhance pharmacodynamic activity, while inducers may result in therapeutic failure. In situations where a drug interaction results in variations in the relative levels of a parent drug and metabolite(s) that are approximately equipotent in terms of efficacy and safety considerations, inhibition and induction may be of little therapeutic consequence.

Drugs that are substrates for metabolism by more than one drug-metabolizing enzyme generally have a decreased likelihood of clinically significant drug interactions due to the availability of compensatory metabolic pathways if one is inhibited. Drug interactions are likely to be particularly important in the following situations ⁵, ⁷:

• when elimination of a drug occurs primarily through a single metabolic pathway

• when a drug is a potent inhibitor or inducer of a drug-metabolizing enzyme

• when one or both of the interacting drugs has a steep dose-response curve

• when one or both of the interacting drugs has a narrow therapeutic range

• when inhibition of the primary metabolic enzyme or induction of a secondary metabolic enzyme results in diversion of the drug into an alternative metabolic pathway that generates a metabolite having toxic or modified pharmacodynamic activity

• when a drug has non-linear pharmacokinetics or when the interaction results in conversion from linear to non-linear pharmacokinetics

Drug Metabolism ^{5, 7, 8}

The biotransformation of drugs is generally a two phase process. Phase I reactions involve oxidation, reduction, or hydrolysis of the parent compound, typically producing metabolites of increased polarity, which may be subjected to excretion or further biotransformation. For metabolites undergoing phase II of the biotransformation process, the polar groups present on the intermediate undergo conjugation with glutathione, acetate, glucuronate, sulfate, or glycine to produce excretable compounds. Some drugs undergo phase II conjugation reactions without prior phase I biotransformation.

The majority of phase I drug metabolism occurs through the cytochrome P450 system (CYP), a superfamily of heme-containing isoenzymes located primarily in hepatocytes, within the membranes of the smooth endoplasmic reticulum. The enterocytes of the small intestine are the principal extrahepatic source of CYP isoforms, with smaller CYP quantities being present in the kidneys, lungs, and brain. The human CYP system has been subclassified into over 70 unique families based on amino acid sequence homology.

• The gene family name is denoted by an Arabic numeral (e.g. CYP3). CYPs within a given family have greater than 40% sequence homology.

• CYP families are further differentiated into gene subfamilies denoted by an upper case letter (e.g. CYP3A). CYPs within a particular subfamily have sequence homology in excess of 55%.

• Gene numbers of individual enzymes are denoted by a second Arabic numeral following the subfamily letter (e.g. CYP3A4).

The isoforms of the CYP system are responsible for the oxidative metabolism of endogenous substances such as steroid hormones, prostaglandins, lipids, and fatty acids and for the detoxification of exogenous compounds such as drugs and dietary xenobiotics. The major CYP isoforms involved in human drug metabolism are CYP3A4, CYP2D6, CYP2C9, CYP2C19, CYP1A2, and CYP2E1. CYP2A6 and CYP2B6 likewise play a significant role in the metabolism of certain drugs.

Enzymes involved in phase II conjugation reactions include glutathione S-transferases, UDP-glucuronosyl transferases, sulfotransferases, N-acetyltransferases, acyltransferases, and methyltransferases Appendix A.

Study of Metabolic Drug Interactions in Vitro

In order to investigate drug interactions in vitro, it is desirable first to identify all of the major metabolic pathways involved in the biotransformation of the test drug, the specific enzymes responsible for these biotransformation reactions, and the metabolites generated by the biotransformation process. While many CYP inhibitors are also substrates for the affected enzyme (e.g. erythromycin is both a substrate and inhibitor of CYP3A4), it should be recognized that a drug may be resistant to metabolism by a given CYP isoform, yet still be capable of inhibiting the ability of the same isoenzyme to metabolize other drugs (e.g. quinidine is a potent inhibitor of CYP2D6, but is metabolized by CYP3A4). Availability of this information will enable the rational anticipation and exploration of drug interactions whereby the metabolism of the test drug is affected by other agents or vice versa ^{2A, 5, 8}.

In Vitro Studies of Drug Metabolism and Enzyme Inhibition

The following experimental model systems and probes have proved useful for the study of drug interactions:

Subcellular Fractions of Human Liver Tissue

Human liver microsomes (vesicles of endoplasmic reticulum) are a common system in which to screen for the effects of a new drug on CYP pathways and to generate preliminary information on potential drug-drug interactions ^{9, 10}. Hepatic microsomes are a subcellular tissue fraction obtained by differential high-speed centrifugation of homogenized liver. Important drug-metabolizing enzymes present within the microsomal fraction include the cytochrome P450 superfamily, the flavin-containing mono-oxygenases, epoxide hydrolases, and a variety of transferases (e.g. the UDP-glucuronosyltransferases). Microsomes from multiple donors should be used, either as individual or pooled preparations, in order to avoid conclusions based on microsomes that may exhibit one or more aberrant metabolic pathways. However, in certain situations, determination of metabolism using aberrant microsomes may be a specific objective of the study.

Microsomal preparations require the addition of exogenous cofactors, including a source of NADPH. Transferase activity can be studied by supplementing microsomal preparations with conjugating moieties. Predictions of drug interactions from cell-free systems such as microsomes may be irrelevant if marked in vivo differences between plasma concentrations and intracellular hepatocyte concentrations are not taken into account. Microsomes are generally not an appropriate system in which to study sequential metabolic reactions (i.e. the coupling of phase I and II reactions) owing to the disruption of the natural orientation between subcellular components. However, advantages of the system include ease of preparation, commercial availability, and long-term stability during cryopreservation.

The S9 subcellular fraction (supernatant of liver homogenate resulting from precipitation of the nuclei and mitochondria by centrifugation at 9,000-20,000g) is a useful preparation in which to study drug metabolism involving both microsomal and cytosolic enzymes ¹⁰.

Whole Cell Models

Biopsy samples and harvested livers not used for transplantation represent sources of human liver tissue for experimental purposes. The characteristics of these preparations will vary depending upon the age, health, and genotypic status of the donor, as well as factors such as diet, alcohol or tobacco consumption, and medication usage. Isolated hepatocytes in suspension or primary culture ^{11, 12} and precision-cut liver slices ¹³ offer numerous advantages over subcellular fractions, including a full complement of hepatic drug- metabolizing enzymes, endogenous cofactors, preservation of the natural orientation for linked enzymes, and a more accurate reproduction of potential differences between extracellular and intracellular drug concentrations across the plasma membrane barrier. They also afford the opportunity to study the role of alternative metabolic routes for a substrate in the presence of drugs that inhibit the principal drug-metabolizing pathway.

Liver slices have the additional advantages of preserving the tissue cytoarchitecture and cell- to-cell communications. However, clearance predictions from liver slices may be lower than those from hepatocytes or microsomes if, due to slice thickness, equilibration is not achieved between the cells of the slice and the incubation medium ⁵).

Short-term stability of enzymatic activities represents the major problem with hepatocytes and liver slices (typically less than 3-4 h for suspensions, less than 24 h for cultures or slices). The viability of these preparations varies depending on culture conditions and the relative stability of the isoforms under investigation.

Caco-2 cell monolayers serve as a surrogate model for absorption and metabolism at the level of the human intestine ¹⁴. Caco-2 cells are derived from human colon cancer cells. When cultured on porous membranes, they differentiate spontaneously into monolayers of polarized cells. Caco-2 cell monolayers are a useful system in which to investigate drug interactions resulting from inhibition of the P-glycoprotein efflux mechanism. Disadvantages of Caco-2 cell monolayers include the underexpression of metabolizing enzymes and a time-dependent loss of enzyme activity in culture.

Heterologous Expressed and Purified Human Drug-Metabolizing Enzymes

The cDNAs for common CYPs have been cloned and the recombinant human enzymatic proteins expressed in a variety of cells with low intrinsic cytochrome P450 enzyme activity including bacteria, yeast, insect cells, mammalian cells, and human lymphoblastoid cells or HepG2 human hepatoma cells ^{15, 16}. Lysates of these transgenic cells are subjected to subcellular fractionation. The activity of the heterologously expressed enzyme can be studied either in microsomes prepared from the transgenic cell lysates or in reconstitution systems containing electrophoretically purified enzyme. Because of the low level of endogenous CYPs in these cell types, adequate protein yields with high specificity for the heterologously expressed enzyme are possible. In contrast, purification of CYPs from human liver microsomes is a complicated process in which it is often difficult to achieve acceptable resolution of CYPs of the same subfamily. cDNA-expressed enzymes generally exhibit affinity properties (K_m, K_i) that are representative of native enzymes, though rates of metabolism (V_max) can be more variable.

Recombinant models have the advantage of offering a single enzyme system in which to investigate isoform-specific metabolism of a drug or the ability of the drug to inhibit substrate metabolism through that isoform. However, the absence of competing pathways prevents these systems from supplying information on the relative contribution of the isoform in question to the overall metabolism of the drug in vivo. Nevertheless, cDNA-expressed and purified enzymes can be a useful system in which to confirm results obtained with native human liver tissue or microsomes. Purified preparations of the following heterologously expressed human drug-metabolizing enzymes are now commercially available: CYPs, flavin-containing mono-oxygenase, epoxide hydrolase, and glutathione S-transferase.

The study of metabolically competent recombinant enzymes in transfected cells is an emerging technology.

Pharmacological Probes

The metabolic pathways for the test drug in subcellular fractions or whole cell models can be demonstrated through the use of selective chemical inhibitors of specific CYP enzymes ^{5, 8, 17, 18}. Inhibitory activity can be expressed either as an inhibition constant (K_i) or the 50% inhibitory concentration (IC₅₀). The IC₅₀ is an estimate of the concentration of the drug inhibiting the maximum rate of metabolism of a fixed concentration of substrate by 50% and, as such, is a measure of the inhibitory potency. IC₅₀ determinations have the advantage of being independent of the biochemical mechanism of enzyme inhibition. However, the IC₅₀ is applicable only to the substrate concentration studied. Therefore, extrapolation to in vivo situations may be unreliable if plasma concentrations of the substrate differ markedly from the substrate concentration studied in vitro. For competitive inhibitors, the IC₅₀ will approximate the K_i only if the substrate concentration studied is considerably below the K_m (the substrate concentration at which the rate of metabolism is half-maximal).

Knowledge of the in vitro inhibition constant of a drug (K_i) for a particular CYP isoform is, therefore, considered more useful in assessing the probability of a drug interaction. The K_i is a measure of the affinity of the inhibitor for the enzyme. K_i determinations require the study of inhibition at a range of concentrations for both the inhibitor and substrate. While K_i values are theoretically dependent on the binding characteristics of the inhibitor, different K_i values may be obtained using different substrates. It is, therefore, desirable to perform independent K_i estimations using multiple substrates. Positive controls should be used to establish assay sensitivity. The effect of concentration on the selectivity of inhibition should be considered.

A rationale should be provided for the choice of incubation time over which the test system is exposed to the investigational drug in combination with the pharmacological probe. The linearity of metabolic rate versus incubation time should have been confirmed in preliminary studies ¹⁸.

Immunochemical Probes

Alternatively, selective inhibition of certain CYP enzymes in microsome preparations can be achieved through use of polyclonal or monoclonal antibodies to specific isoforms ¹⁹. Inhibition of metabolite formation by an antibody specific to a particular cytochrome implicates that isoform in the metabolism of the test drug. Current problems with this approach include a lack of wide commercial availability of these antibodies, deficient antibody selectivity between subfamily members, failure to achieve maximal inhibition if the molar concentration of the antibody is not equivalent to the molar concentration of the enzyme, inability to achieve 100% inhibition (probably because the large size of the antibody molecule does not permit access to all isoform molecules), and a high degree of inter-laboratory variation in the extent of inhibition achieved.

In Vitro Studies of Enzyme Induction

Primary human hepatocyte cultures and hepatocyte cell lines (e.g. HepG2 human hepatoma cell line) have sometimes proved useful for the study of induction phenomena ¹¹. Artifacts are common, however, as cultured hepatocytes typically undergo a time-dependent loss of CYP expression and may also exhibit a diminished capacity for induction. The predictive value of hepatocyte cell lines is often limited by substantial phenotypic differences between these cells and the tissues from which they were derived. Enzyme induction can be measured by assaying for the activity of specific isoforms, immunodetection of isoform protein, or quantification of mRNA. The use of known inducers as positive control agents is necessary to verify the sensitivity of these systems.

Choice of Drug Concentration Range

In vitro drug interaction studies should employ clinically relevant concentrations of both the substrate and the inhibitor or inducer. Studies conducted at supratherapeutic drug concentrations may produce in vitro drug interactions that are not representative of the in vivo situation. Moreover, the major metabolic pathway for a given drug may be concentration-dependent. For instance, N-demethylation is the principal in vivo metabolic pathway for diazepam at therapeutic doses, while at high in vitro drug concentrations
(100 µM) 3-hydroxylation predominates.

Controversy exists as to whether the free (unbound) or total (unbound + bound) in vivo plasma concentration of the drug represents the most appropriate concentration for prediction of drug interactions from in vitro studies. Some compounds have high liver to plasma partition ratios despite extensive protein binding. Use of free fraction plasma concentrations for the prediction of metabolic drug interactions for such agents has resulted in an underestimation of the clinical outcome. In some instances, hepatic drug levels may even exceed the total plasma concentration by several fold ^{5, 17, 20}.

If the drug is to be developed as a resolved enantiomer, in vitro preclinical metabolism studies should be conducted with that enantiomer, rather than the racemate, as the metabolic pathway may be stereoselective (e.g. S-warfarin is metabolized by CYP2C9, while R-warfarin is metabolized by CYP3A4 and CYP1A2).

Position in the Time Course of Drug Development

The availability of in vitro metabolic studies prior to phase II clinical investigations should be encouraged. Appropriate in vitro studies of metabolism and drug interactions should have been completed prior to the initiation of phase III clinical trials.

Role of in Vivo Animal Studies

Whenever feasible, drug metabolism should be investigated in human tissues, cells, or subcellular fractions as interspecies and intraspecies differences in drug-metabolizing enzymes limit the ability to extrapolate between experimental animals and humans. While animal cell lines and liver microsomal fractions may be used to supply preliminary data, such information should always be confirmed in human liver preparations.

However, animal species may provide useful models in which to determine whether a new chemical species generated in vitro by human liver microsomes produces pharmacological or toxicological effects in vivo and how these compare with the effects of the parent compound. An appreciation of the pharmacodynamics and toxicology of the human metabolites of a drug is helpful in anticipating the clinical implications of drug interactions.

A knowledge of comparative drug metabolism in humans and animals can play a useful role in the rational selection of animal models for toxicology studies. Major discrepancies in the metabolism of a drug between laboratory test species and humans diminish the predictive value of toxicology data obtained from these animals.

The development of transgenic animal models for heterologous expression of human drug-metabolizing enzymes or for gene inactivation by homologous recombination represents an in vivo approach to the investigation of the role of specific enzymes involved in drug metabolism ^{21, 22}. However, owing to interspecies differences, the endpoints of such studies may not reflect the human situation ²¹.

Study of Metabolic Drug Interactions in vivo in Clinical Trials ^{3A, 6, 7, 23}

The clinical investigation of potential drug-drug interactions should be a systematic process of well designed trials performed upon appropriately selected cohorts of subjects or patients, receiving both interacting drugs under conditions of dosage and administration that are intended to reproduce the proposed or approved therapeutic dosing recommendations. Such studies should address relevant pharmacokinetic and pharmacodynamic endpoints and employ suitable statistical analyses.

Design

0pen label (unblinded) design is generally acceptable for studies of pharmacokinetic interactions, unless the interaction assessment includes pharmacodynamic measurements.

Possible designs include randomized crossover (randomization to A followed by A+B or A+B followed by A); one sequence crossover (A followed by A+B), or parallel group dosing (randomization to A or A+B). Through use of a "within individual" design, crossover studies offer greater statistical power for a given number of subjects than do parallel group studies. Parallel group studies require more subjects for a shorter duration of time, whereas crossover studies require fewer subjects for a longer time period. Crossover studies have the advantage of allowing individual subjects to serve as their own controls. Randomized crossover studies represent an improvement over one sequence crossover studies in that they are sensitive to sequence and carry-over effects as well as to period effects. Parallel group studies may be preferred for drugs with long elimination half-lives for which lengthy time intervals are required to achieve steady-state or to complete washout.

The choice of single or multiple dose regimens for each of the drugs in an interaction study should be based upon (1) whether these agents are intended for acute or chronic use in their target patient populations, (2) the presence or absence of safety considerations precluding multiple dose use, and (3) pharmacodynamic/pharmacokinetic characteristics of the drugs (e.g. elimination half-life, lag time to pharmacodynamic response of interest, enzyme-inducing properties, etc.). On the whole, multiple dose regimens result in smaller confidence intervals at steady state. However, for drugs with low accumulation ratios, confidence intervals may be similar for single dose and multiple dose study situations.

The timing of drug co-administration may be an important variable in drug interaction studies (e.g. concurrent administration versus administration separated by 1, 2, 3, or more hours). The sequence of administration of the interacting drugs may likewise affect the magnitude of the interaction.

In the case of a drug that competitively inhibits the metabolism of a substrate, inhibition is usually maximal at the time when the inhibitor reaches its steady state level (4 or 5 half-lives) and when the inhibited drug reaches steady state at its new, longer half-life. For enzyme inducers, the time to maximal induction may extend beyond the attainment of steady state plasma levels, therefore, proving more difficult to predict. If the drugs are to be studied at steady-state, attainment of steady-state plasma concentrations should be documented for the parent drugs and their metabolites of interest by the analysis of samples collected over several days prior to the period at which the interactive effects are to be studied. It should be recognized that time-dependent changes in drug pharmacokinetics can occur over the course of chronic therapy, such that plasma exposure and metabolic profile may not stabilize until several months after the initiation of treatment.

Investigation of drug levels during the washout phase following discontinuation of the inhibitor or inducer may be necessary in order to determine the time period post-discontinuation during which the patient remains susceptible to drug interactions. The rate-limiting factor in the offset of induction may be the half-life of CYP protein turnover, rather than the half-life of the inducer.

Useful information on drug interactions can sometimes be obtained from plasma concentration measurements performed upon blood samples collected according to sparse or dense sampling schedules during phase II and III clinical trials ²⁴. These population pharmacokinetic studies have the advantage of providing data from the intended patient population. This approach allows identification of correlations between pharmacokinetic inter- and intra-individual variability and factors such as demographics, co-morbidity, or concomitant medication. Quantitative estimates of the magnitude of this variability can be generated. Such data may be of value in confirming an interaction suspected on the basis of in vitro findings or in the identification of unsuspected interactions. However, as the power of population pharmacokinetic screens to detect drug interactions is not well established, such data should not be used to dismiss the possibility of a suspected drug interaction, especially if positive in vitro or in vivo data are available from systematic drug interaction studies. False negative findings may be an issue if the study contains an insufficient number of patients taking the drug combination of interest. Furthermore, interpretation of population pharmacokinetic data may be difficult if the sequence and timing of co-administration were not standardized.

Population

The following subject or patient groups may represent appropriate populations for drug interaction studies:

• healthy subjects who meet restrictive eligibility criteria that increase the homogeneity of the study cohort
• subjects drawn from the general population, meeting expanded eligibility criteria that permit heterogeneity of the study cohort
• patients for whom the investigational drug is intended
• subjects who have been selected on the basis of phenotyping and/or genotyping for metabolic polymorphisms, predisposition to adverse drug reactions, or therapeutic outcome

Healthy volunteers as a subject group for drug interaction studies have the advantage of minimizing the confounding effects introduced by disease states or concomitant medications that may affect drug absorption, metabolism, distribution, or excretion. However, use of subjects drawn from the general population or patients for whom the investigational drug is intended may result in data that correlate better with the clinical situation. Phenotyping and/or genotyping during screening may be used for inclusion, exclusion, or stratification of subjects. Such approaches are useful in characterizing the effect of metabolic status (e.g. poor, extensive, ultra-rapid metabolizer) on the magnitude of the drug interaction, but introduce a bias which limits the generalizability of the results. Alternatively, the availability of phenotype and/or genotype information following study completion may be useful in the interpretation of pharmacokinetic outliers as well as inter-individual variability in pharmacodynamics and adverse events.

Choice of Potentially Interacting Drugs

The selection of potentially interacting drugs for drug interaction investigations should be based on a knowledge of the enzymes subject to inhibition or induction by the test drug or responsible for its metabolism, as identified in in vitro studies.

Selected drugs should be potent and specific inhibitors or inducers of the principal drug- metabolizing enzymes responsible for the biotransformation of the investigational drug.

Selected drugs should be sensitive substrates of the principal drug-metabolizing enzymes subject to inhibition or induction by the investigational drug or involved in its biotransformation.

Selection may favour approved drugs for which co-administration is likely given the indication and target population characteristics of the investigational drug.

Selection may be based on known interactions affecting chemically related drugs of the same therapeutic class.

Selection may favour drugs having low therapeutic indices for which interactions would be expected to have clinically significant consequences (e.g. digoxin, warfarin).

Selection of substances ingested socially (e.g. alcohol) or through the diet (e.g. grapefruit juice) may be advisable where these are anticipated to affect the metabolism of the test drug.

Route of Administration

The routes of administration used in the clinical investigation of drug interactions should correspond to those recommended in the proposed or approved Product Monographs for the drugs in question.

If multiple routes of administration are to be recommended in the Product Monograph, positive/negative drug interaction findings for one route should not necessarily be extrapolated to the others (e.g. an intravenous drug interaction study would not reveal an effect on intestinal CYP3A4 activity that might influence bioavailability).

Dose Selection

The highest proposed or approved doses of the test drugs and the shortest dosing intervals should be used in order to maximize the possibility of finding an interaction.

On occasion lower doses may be acceptable for safety reasons.

Exploration of differential effects over a range of doses may be useful to characterize the dose-dependency of a drug interaction.

Endpoints

• 1. Pharmacokinetic
  • Parameters that serve as useful endpoints in the analysis of pharmacokinetic drug interaction data include the area under the blood/plasma/serum concentration time curve (AUC), the peak concentration (C_max), the time to peak concentration ( t_max), the trough concentration (C_min), the clearance (CL), the volume of distribution (V_d), and the elimination half-life (t_1/2).
  • The fraction of drug existing unbound in the plasma (f_u) may be a useful parameter in the assessment of drug interactions for certain highly protein-bound agents that are distributed primarily in the plasma rather than in tissues (i.e. low V_d).
  • Pharmacokinetic parameters should be determined for the parent drug and important active or toxic metabolites.
  • The availability of pharmacokinetic data for both interacting drugs may be desirable, especially in situations where interactive effects are complex (e.g. concomitant use of an inhibitor and an inducer, co-administration of two drugs having reciprocal effects on each other's primary or secondary metabolic pathways through inhibitory activity on different drug-metabolizing enzymes, or use of an inhibitor having active or toxic metabolites that are substrates for enzymes affected by elevated levels of the interacting drug). In some instances, two independent studies may be required to adequately characterize the effects on an interaction on each of the drugs involved, while in other cases such information may be obtained from a single study (e.g. three period crossover approach).
• 2. Pharmacodynamic
  • Pharmacodynamic parameters may serve as useful endpoints when interactions are not wholly accounted for by pharmacokinetic effects (e.g. additive haemodynamic or anticoagulant effects).

Adverse Events

In clinical pharmacology studies addressing drug interactions, attention should be directed to adverse events occurring with greater frequency or severity during combination treatment than during treatment with either agent alone.

Clinical trial protocols should contain directions for the collection of blood samples from patients experiencing serious, severe, or unexpected adverse events. Plasma level determinations for concomitant medications should likewise be encouraged in such cases.

Statistical Considerations

The 90% confidence intervals should be provided for the mean ratios of the pharmacokinetic exposure measures of the drug in the presence and absence of the interacting agent.

Significance tests are of limited value as small, consistent differences in exposure may result in p-values that are statistically significant, but of doubtful clinical relevance.

Knowledge of the width of the equivalence interval beyond which a dosage adjustment is necessary is useful in determining sample size.

In the absence of other information to determine an equivalence interval, a standard interval of 80-125% can be employed for investigational and approved drugs in drug interaction studies.

Sample size should be sufficient such that the study has statistical power to detect a clinically important difference between the treatment groups. Sample size requirements will be increased if inter- or intrasubject variability in pharmacokinetic measurements is high.

Correlation Between Studies in vitro and in vivo

In assessing the correlation between in vitro and in vivo drug interaction studies, attention should be directed to the following considerations:

• In vitro drug interaction studies should be conducted at concentrations similar to those attained in vivo.

• The clinical importance of positive in vitro results should always be confirmed in vivo. Depending on the role of the investigational drug in the suspected interaction, the in vivo studies should be performed using one or more sensitive substrates or potent and specific inhibitors/inducers of the drug-metabolizing enzyme implicated. Results from such studies may be used to extrapolate qualitatively to other inhibitors, inducers, or substrates of the enzyme in question.

• In vitro drug interaction studies may not accurately predict in vivo interactions if alternative metabolic or excretory routes play a major role in the clearance of the drug in vivo.

• In the event of conflicting results between in vitro and in vivo studies, the in vivo studies should receive precedence, provided that these have been conducted under clinically relevant conditions.

Confounding factors that may result in failure to make accurate correlations between in vitro and in vivo studies include, but are not limited to, the following ²⁰:

• Free (unbound) in vivo drug concentrations in the plasma may be lower than the drug concentrations in the hepatic biophase as the hepatic uptake of many lipophilic compounds is not necessarily limited by protein binding.

• When a drug interaction involves elements of both induction and inhibition, the predominant effect on drug clearance can be either positive or negative and may be time-dependent (e.g. metabolic inhibition in single dose studies and induction at steady-state conditions).

• In vitro drug interaction results may underestimate the in vivo interactions for agents that inhibit the intestinal P-glycoprotein transporter, thus increasing the bioavailability of concomitantly administered drugs which are substrates for extrusion by this mechanism.

• High microsomal protein concentrations may result in overestimation of K_i values if the inhibitor becomes depleted due to microsomal metabolism or nonspecific binding to microsomal proteins ⁵).

  
  Top of Page

Factors Influencing Drug Interactions

Inter-individual susceptibility to drug interactions may be influenced by a wide range of factors including, but not limited to, the following:

genetic: Many drug-metabolizing enzymes, for example CYP2D6, CYP2C19, CYP2A6, CYP2C9, and N-acetyltransferase, are subject to genetic polymorphism such that large inter-individual differences exist in the ability to metabolize substrates for these enzymes ^{25, 26}. Some genetic polymorphisms show an increased incidence in certain ethnic groups. For example, poor metabolizers of CYP2D6 substrates represent 5 to 10% of the Caucasian population, but only 1 to 2% of the Asian population. Conversely, the incidence of poor metabolizers of CYP2C19 substrates is 18-22% in Asian populations, but only 2-6% in Caucasian populations ^{5, 25, 26}.

age: The magnitude and clinical consequences of a pharmacokinetic drug interaction may be age-dependent. Neonates exhibit reduced hepatic metabolism and renal excretion of drugs due to immaturity of liver and kidney function. The susceptibility of elderly individuals to drug interactions may be affected by age-related alterations in absorption, hepatic metabolism, renal clearance, or volume of distribution ²⁷.

gender: Studies of certain drugs have shown inconsistencies in pharmacokinetic parameters or pharmacodynamic endpoints between male and female subjects that are suggestive of gender-dependent differences in bioavailability, volume of distribution, drug-metabolizing enzyme activity, renal clearance, or physiological characteristics. Normalization of pharmacokinetic parameters for body weight may be advisable in assessing apparent differences between male and female patients in drug interaction studies ²⁸.

disease states: The effects of hepatic impairment on the magnitude and clinical consequences of drug interactions may be complex and difficult to predict. Patients with renal impairment may be at an increased risk for metabolic drug interactions, due to a diminished contribution of the excretory component of the drug elimination process.

social: Tobacco smoke is an inducer of CYP1A1, CYP1A2, and possibly CYP2E1 ²⁹. Chronic alcohol ingestion results in the induction of CYP2E1 ⁷.

dietary: Grapefruit juice contains chemicals that are potent inhibitors of CYP3A4 in the intestinal wall mucosa ³⁰. Cruciferous vegetables (e.g. brussels sprouts, cabbage, cauliflower) and hydrocarbons present in charcoal-broiled meat can induce CYP1A2 ⁷. The calcium present in dairy products has the potential to chelate drugs such as tetracyclines and fluoroquinolones ³. Irreversible, non-selective monoamine oxidase inhibitors reduce the metabolism of endogenous norepinephrine and exogenous tyramine ingested in foods and beverages that have undergone protein breakdown as a result of aging, fermentation, pickling, smoking, or bacterial contamination (e.g. beer, wine, certain cheeses and sausages). The increased bioavailability of tyramine in combination with the augmented endogenous norepinephrine stores can lead to an exaggeration of the indirect sympathomimetic effect of tyramine, resulting in hypertensive crises ³¹.

Product Monograph

The choice of whether to deal with a given drug interaction as a contraindication, warning, or precaution should be based on an assessment of the seriousness and severity of the clinical consequences of the established or suspected drug interaction. All documented and anticipated drug interactions should be included in the "Drug Interactions" sub-section of the "Precautions" section with appropriate cross-references to other sections in which they may also appear. Drug interactions should be presented as contraindications if they have the capacity to be life-threatening, cause permanent damage, or elicit other reactions that would prohibit concomitant administration. Interactions having the potential to cause serious or severe consequences that are reversible or not life-threatening should normally be included in the "Warnings" section together with recommendations for appropriate risk management measures. Drug interactions of unknown clinical significance or resulting in adverse effects that are merely bothersome can generally be adequately dealt with in the "Drug Interactions" sub-section of the "Precautions" section.

When describing the results of in vivo clinical drug interaction studies, the monograph should indicate the number of individuals studied, whether the subjects were healthy volunteers or patients, the dose and duration of treatment with the drugs in question, and the magnitude of the observed effect on important pharmacokinetic parameters (e.g. % increase or multiple of control value). Drug interactions identified through population pharmacokinetic approaches, clinical trial case reports, or spontaneous post-marketing adverse event reports should be identified as such. When sufficient data are available, comments should be provided on the mechanism of the interaction, the clinical manifestations, and appropriate actions to prevent or respond to an interaction (e.g. adjustments of dosage and/or dosing interval, timing and sequence of co-administration, recommended washout periods between administration of interacting drugs, plasma level monitoring, pharmacodynamic monitoring).

For drugs known to be potent and selective inhibitors or inducers or sensitive substrates of specific drug-metabolizing enzymes, standard labelling concerning drug interactions may be appropriate, even if specific data are not available concerning the interactions in question. For example, potent inhibitors of CYP3A4 and CYP2D6 are likely to decrease the metabolism of all substrates for these isoforms. However, drug interactions that are anticipated from in vitro studies should be identified as such and distinguished from drug interactions that have been demonstrated under in vivo conditions.

Warnings or precautionary statements based upon a documented interaction with a particular drug need not necessarily be extrapolated to other members of that drug's therapeutic class if a reasonable basis exists for believing that these drugs do not share a common metabolic pathway. For example, the benzodiazepines, triazolam and alprazolam, are substrates for CYP3A4 and undergo clinically significant interactions with inhibitors of this isoform, whereas lorazepam, a drug metabolized primarily by glucuronidation, is not subject to interactions with CYP3A4 inhibitors. Manufacturers wishing to gain exemption from therapeutic class labelling for drug interactions should convincingly demonstrate that the possibility of such interactions with their products has been adequately investigated and dismissed.

The "Pharmacokinetics" sub-section of the "Actions and Clinical Pharmacology" section of the Product Monograph should contain information on the major metabolic pathways involved in the biotransformation of the drug, the specific enzymes responsible for these reactions, and the principal metabolites generated. A comprehensive list of the metabolic enzymes tested should be provided.

When approval is granted to Product Monographs that contain information on clinically important new drug interactions, the manufacturers of the interacting drugs should be notified in order to ensure that consistent information appears in the Product Monographs for all agents involved in the interactions.

Prepared:

C. F. Strnad, Ph.D.

Reviewed:

W. Casley, Ph.D. B. Foster, Ph.D. I. Hynie, M.D., Ph.D. S. Robertson, M.D.

Literature References

¹ Kedderis, G.L. Pharmacokinetics of drug interactions. Adv. Pharmacol. 1997; 43:189-203.

² Hooper, W.D. Metabolic drug interactions. In "Handbook of Drug Metabolism" (T.F. Woolf ed., 1999), pp. 229-238. Marcel Dekker, Inc., New York, N.Y.

³ Fleisher, D. et al. Drug, meal and formulation interactions influencing drug absorption after oral administration. Clin. Pharmacokinet. 1999; 36(3):233-254.

⁴ Yu, D. K. The contribution of P-glycoprotein to pharmacokinetic drug-drug interactions. J. Clin. Pharmacol. 1999; 39:1203-1211.

⁵ Lin, J.H., Lu, A.Y.H. Inhibition and induction of cytochrome P450 and the clinical implications. Clin. Pharmacokinet. 1998; 5(5):361-390.

⁶ Bonate, P.L. et al. Drug interactions at the renal level. Clin. Pharmacokinet. 1998; 34(5):375-404.

⁷ Johnson, M. D. et al. Clinically significant drug interactions. Postgraduate Medicine 1999; 105(2):193-222.

⁸ Bertz, R.J., Granneman, G.R. Use of in vitro and in vivo data to estimate the likelihood of metabolic pharmacokinetic interactions. Clin. Pharmacokinet. 1997; 32(3):210-258.

⁹ Rodrigues, A.D., Wong, S.L. Application of human liver microsomes in metabolism-based drug-drug interactions. Adv. Pharmacol. 1997; 43:65-101.

¹⁰ Ekins, S. et al. In vitro metabolism: Subcellular fractions. In "Handbook of Drug Metabolism" (T.F. Woolf ed., 1999), pp. 363-399. Marcel Dekker, Inc. New York, NY.

¹¹ Li, A.P. Primary hepatocyte cultures as an in vitro experimental system for the evaluation of pharmacokinetic drug-drug interactions. Adv. Pharmacol. 1997; 43:103-130.

¹² Sinz, M. W. In vitro metabolism: Hepatocytes. In "Handbook of Drug Metabolism" (T.F. Woolf ed., 1999), pp. 401-424. Marcel Dekker, Inc. New York, NY.

¹³ Ferrero, J.L., Brendel, K. Liver slices as a model in drug metabolism. Adv. Pharmacol. 1997; 43:131-169.

¹⁴ Yee, S., Day, W.W. Application of Caco-2 cells in drug discovery and development. In "Handbook of Drug Metabolism" (T.F. Woolf ed., 1999), pp. 507-522. Marcel Dekker, Inc. New York, NY.

¹⁵ Crespi, C.L., Penman, B.W. Use of cDNA-expressed human cytochrome P450 enzymes to study potential drug-drug interactions. Adv. Pharmacol. 1997; 43:171-188.

¹⁶ Rodrigues, A.D. Applications of heterologous expressed and purified human drug-metabolizing enzymes: An industrial perspective. In "Handbook of Drug Metabolism" (T.F. Woolf ed., 1999), pp. 279-320. Marcel Dekker Inc., New York, NY.

¹⁷ von Moltke, L. L. et al. In vitro approaches to predicting drug interactions in vivo. Biochem. Pharmacol. 1998; 55(2)

¹⁸ Yuan, R. et al. In vitro metabolic interaction studies: Experience of the Food and Drug Administration. Clin. Pharmacol. Ther. 1999; 66(1):9-15.

¹⁹ Gonzalez, F.J. Overview of experimental approaches for study of drug metabolism and drug-drug interactions. Adv. Pharmacol. 1997; 43:255-277.

²⁰ Davit, B. et al. FDA evaluations using in vitro metabolism to predict and interpret in vivo metabolic drug-drug interactions: impact on labeling. J. Clin. Pharmacol. 1999; 39:899-910.

²¹ Friedberg, T. et al. In vivo and in vitro recombinant DNA technology as a powerful tool in drug development. In: "Handbook of Drug Metabolism" (T. F. Woolf ed., 1999) pp. 321-362. Marcel Dekker Inc., New York, NY.

²² McKinnon, R.A. & Nebert, D.W. Cytochrome P450 knockout mice: New toxicological models. Clin. Exp. Pharmacol. Physiol. 1998; 25:783-787.

²³ Huang, S.-M. et al. Assessment of the quality and quantity of drug-drug interaction studies in recent NDA submissions: Study design and data analysis issues. J. Clin. Pharmacol. 1999; 39:1006-1014.

²⁴ Sun, H. et al. Population pharmacokinetics. Clin. Pharmacokinet. 1999, 37(1):41-58.

²⁵ Ingelman-Sundberg, M. et al. Polymorphic human cytochrome P450 enzymes: an opportunity for individualized treatment. TIPS 1999; 20(8):342-349.

²⁶ Daly, A. K. Pharmacogenetics. In: "Handbook of Drug Metabolism" (T. F. Woolf ed., 1999) pp. 175-202. Marcel Dekker Inc., New York, NY.

²⁷ Hammerlein, A. et al. Pharmacokinetic and pharmacodynamic changes in the elderly. Clin. Pharmacokinet. 1998; 35(1):49-64.

²⁸ Beierle, I. et al. Gender differences in pharmacokinetics and pharmacodynamics. Int. J. Clin. Pharmacol. Ther. 1999; 37 (11):529-547.

²⁹ Zevin, S., Benowitz, N.L. Drug interactions with tobacco smoking. Clin. Pharmacokinet. 1999; 36 (6): 425-438.

³⁰ Ameer, B., Weintraub, R.A. Drug interactions with grapefruit juice. Clin. Pharmacokinet. 1997; 33(2): 103-121.

³¹ Stockley, I. H. Monoamine oxidase inhibitor drug interactions. In "Drug Interactions" (Stockley ed., 1999) pp. 594-615. Pharmaceutical Press, London, UK.

Website References

1A. Clinically Used Drugs Metabolized by Cytochrome P450.

2A. Guidance for Industry: Drug Metabolism/Drug Interaction Studies in the Drug Development Process: Studies In Vitro (1997) Food and Drug Administration

3A. Guidance for Industry: In Vivo Drug Metabolism/Drug Interaction Studies-Study Design, Data Analysis, and Recommendations for Dosing and Labelling (1999) Food and Drug Administration

Appendix A

Phase I Enzymes

Cytochrome P450 Superfamily
CYP1
CYP2
CYP3
CYP4
CYP5
CYP7
CYP11
CYP17
CYP19
CYP21
CYP27

Flavin-containing mono-oxygenases
Alcohol dehydrogenase
Aldehyde dehydrogenase
Dihydropyrimidine dehydrogenase
Butyrylcholinesterases
Cholinesterase
Hydrolases
Monoamine-diamine oxidase
Polyamine oxidase
Xanthine oxidase
Alkylhydrazine oxidase
Paroxonase
Prostaglandin synthetase-lipoxygenase
Aromatases
Azo and nitro reductases
Carbonyl reductase
Epoxide hydrolases

Phase II Enzymes
Glutathione S-transferases
GST A1-1
GST A2-2
GST M1a-1a
GST M1b-1b
GST M2-2
GST M3-3
GST M4-4
GST M5-5
GST P1-1
GST T1-1
GST T2-2
GST microsomal

N-Acetyltransferases
NAT1
NAT2

Uridine diphosphate-glucuronyltransferases
UGT1
UGT2

Methyltransferases
thiol methyltransferase
catechol-O-methyltransferase
thiopurine methyltransferase

Sulfotransferases
ST1A2
ST1A3
ST1A5
ST2A3

Acyltransferase

Acyl-CoA synthetases