Navigating Drug–Drug Interactions in Clinical Drug Development: A Tutorial

ABSTRACT

This tutorial provides essential guidelines and insights, in addition to regulatory guidance, for the evaluation of drug–drug interactions (DDIs), with a focus on the tools and timing of assessment critical for effective management of inclusion/exclusion criteria during drug development and accurate labeling of investigational drugs. It aims to equip researchers with the knowledge and methodologies required to identify and evaluate DDIs, addressing both victim (investigational drugs that are affected by concomitant medications) and perpetrator (investigational drugs that cause changes in the PK of concomitant medications) interactions. The tutorial explores the mechanistic basis of DDIs, explaining how such interactions can alter drug absorption, distribution, metabolism, and excretion, potentially influencing the benefit‐to‐risk profile. The tutorial further highlights the tools and methodologies employed in assessing drug–drug interactions (DDIs) across various stages of development. It outlines the importance of in vitro studies for early‐stage screening of enzyme‐ and transporter‐mediated interactions to identify potential perpetrators, followed by in vivo studies to confirm these findings. Additionally, it explores the application of computational modeling approaches, such as physiologically based pharmacokinetic (PBPK) modeling and population pharmacokinetic (popPK) approaches, where feasible, to predict both victim and perpetrator interactions prior to or in addition to clinical trials. The aim of this tutorial is to serve as a comprehensive resource for DDI considerations when developing a new molecular entity, supporting researchers during clinical drug development.

1. Introduction

Characterizing drug–drug interactions (DDIs) is essential for optimizing dosing and preventing adverse events due to increased drug exposure because of inhibition or decreasing efficacy because of induction in patients receiving coadministered drugs. Multiple concomitant medications, which can be substrates, inhibitors, or inducers of drug‐metabolizing enzymes (DMEs) and transporters, may be administered in each disease. This tutorial guides clinical scientists through design and interpretation considerations for investigating DDIs to optimize the dosing of concomitant medications with investigational drugs.

1.1. Importance of Drug Interactions

Many drugs have an adequate benefit: risk at a specific dose in patients when taken alone but may lead to substantially more adverse events (AEs) when coadministered with various drugs. Thus, it is imperative that the product label outlines the need for dosing modifications when drugs are concomitantly administered. The importance of dose modifications due to DDIs was recognized in the 1990s and early 2000s, where several approved drugs were withdrawn from the market because of increased toxicity in the presence of a DDI. Many of the withdrawn drugs (e.g., terfenadine, astemazole, cisapride) were cytochrome P450 (CYP)3A4 substrates with off‐target binding to human ether‐a‐go‐go‐related gene (hERG); when concomitant CYP3A4 inhibitors were coadministered with these products, exposure increased, and the risk of arrhythmias or sudden death increased [1, 2, 3].

Today, concomitant administration of more than one approved drug continues to increase patient risks. For example, severe adverse drug reactions increase the rate of hospital admissions. A recent study suggests that 40% of patients admitted to the hospital experienced at least one drug interaction and had a longer length of stay compared to those with no drug interactions [4]. Evaluating these DDI risks is essential given the rising incidence of polypharmacy (those taking ≥ 5 medications) in patients with comorbidities, particularly in patients who are hospitalized or have advanced age [5, 6, 7].

Given polypharmacy, there is a vast number of potential DDIs in the treatment of many diseases. The potential for DDIs is further complicated by concurrent use of over‐the‐counter (OTC), complementary, and alternative medicines [8]. Notably, studying every potential drug interaction is not feasible.

1.2. Key Approach to Evaluating Drug Interactions

A scientific risk–based approach has been developed to evaluate the potential for DDIs using in vitro and in vivo studies, along with model‐based approaches, such as physiologically based pharmacokinetics (PBPK) and population pharmacokinetics (popPK) to optimize the benefit‐to‐risk profile throughout drug development. This risk‐based approach involves evaluating whether concomitant drugs can alter the exposure of an investigational drug by inhibiting or inducing DMEs or transporters, known as victim DDIs. Conversely, it assesses whether the investigational drug can affect the exposure of concomitant drugs through the induction or inhibition of DMEs or transporters, referred to as perpetrator DDIs. The thorough evaluation of both victim and perpetrator DDIs for an investigational drug is outlined in the recent International Council for Harmonization (ICH) M12 Drug Interaction Guidance [9, 10]. Notably, the scientific justification of potential DDIs is continually evolving (Figure 1) with the emergence of new scientific information about DMEs and transporters, which must be considered.

FIGURE 1.

Illustration of the strategy to evaluate drug interactions for the investigational drug as a victim (upper panel) or as a perpetrator (lower panel). The tools required to implement the strategy are outlined, with two‐way arrows indicating the need for an evolving strategy based on new information.

DDIs have been reviewed extensively [8, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20]. This tutorial focuses on the tools (e.g., in vitro studies, modeling approaches, clinical studies) used in the evaluation of DDIs and the strategy (e.g., when and how to use these tools throughout drug development) for scientists new to evaluating DDIs. First, we will discuss the investigational agent as a victim, covering tools and then strategy. This approach will also be used when discussing the investigational drug as a perpetrator. DDI evaluation of biological (e.g., monoclonal antibodies, antibody–drug conjugates, oligonucleotides, small‐interfering ribonucleic acids, and peptides) interactions is out of scope, but DDIs of monoclonal antibodies and therapeutic proteins have been reviewed elsewhere [21, 22, 23, 24, 25].

2. Evaluation of Drug Interactions That May Increase or Decrease Exposure to an Investigational Drug (i.e., Investigational Drug as a Victim)

The key to understanding the potential for DDIs of an investigational drug to be affected by a concomitant medication is to identify whether an investigational drug is a substrate for various DMEs and transporters and how important these DMEs are to the overall elimination pathway for the investigational drug.

2.1. Tools Used in Evaluating Victim Drug–Drug Interactions

2.1.1. In Vitro Metabolism Studies

The potential for victim DDIs is initially evaluated using in vitro tools to characterize whether the investigational agent is a substrate for various cytochrome P450 (CYP) isoenzymes, human UDP‐glucuronosyltransferase (UGT), and other Phase 2 enzymes (e.g., sulfotransferases). The ICH M12 guidance on DDIs describes the types of in vitro studies needed to decide on the clinical DDI strategy for an investigational drug [9, 10]. If an investigational drug is a substrate for a DME, one must understand the magnitude of importance of the individual pathway. In general, the M12 guidance suggests that an enzyme that is estimated to account for 25% of the total elimination of the investigational drug generally needs a clinical victim DDI study to characterize DDI risks [9, 10].

2.1.2. Human Mass Balance Study

In vitro data may identify the DMEs, predominantly contributing to metabolism, but a human mass balance (hADME) study will be needed to confirm the metabolic pathways and develop an understanding of the contribution of other elimination pathways of an investigational drug [9, 10]. Prior to the conduct of a hADME study, in vitro data is used to drive the evaluation of DDIs, but once the hADME study is completed, the victim DDI strategy should be updated to include all metabolites that account for more than 10% of the radioactivity in humans and at least 25% of the area under the curve (AUC) for parent drug or if there is an active metabolite that may contribute substantially to efficacy or safety [9, 10].

2.1.3. In Vitro Transporter Studies

Another important tool is the vitro evaluation of whether an investigational drug is a substrate for various transporters. The International Transporter Consortium (ITC, https://www.itc‐transporter.org/) should be consulted when evaluating what transporters should be evaluated, given the evolving nature of this field. Whether an investigational drug is a substrate for drug transporters is dependent on the absorption, distribution, metabolism, and excretion (ADME) pathways of the investigational drug, typically identified by the hADME study. If intestinal absorption is limited, then an investigational agent may be a substrate for efflux transporters (P‐glycoprotein [P‐gp] or breast cancer resistance protein [BCRP]). If biliary excretion is an important pathway, then P‐gp, BCRP, and multidrug resistance protein (MRP‐2) should be considered. When an investigational drug undergoes significant active renal secretion (25% of clearance), then the investigational agent may be a substrate for organic anion transporter (OAT)1, OAT3, organic cation transporter (OCT)2, multidrug and toxin extrusion (MATE)1, and MATE‐2 K [9, 10].

2.1.4. PBPK Modeling

PBPK models are advanced computational tools that predict the ADME of drugs. These models integrate detailed physiological and biochemical data to simulate how inhibitors or inducers affect the pharmacokinetics of a victim drug, including interactions with key enzymes (e.g., CYPs) and transporters (e.g., P‐gp). By combining in vitro, in vivo, and clinical data, PBPK models provide a comprehensive framework to evaluate the impact of perpetrator drugs on the PK of an investigational drug (Figure 2, Ref. [26]).

FIGURE 2.

Integrated workflow to illustrate in vitro/preclinical and clinical data generation and modeling considerations to identify, characterize, and predict human pharmacokinetics in plasma and tissues for molecules where transporter plays an important role. Taken from Giacomini et al. [26]. DDI, drug–drug interaction; PD, pharmacodynamic; PET, positron emission tomography; PK, pharmacokinetic.

Key elements for successful “high‐impact” PBPK modeling in DDI studies are: (i) platform qualification, (ii) drug model validation for the intended mechanism and use, (iii) input parameters derived from experimentally measured, predicted, or estimated data, (iv) model development guided by training datasets and verified with independent datasets, (v) sensitivity analyses of uncertain parameters, and (vi) patient risk evaluation based on PBPK predictions and associated uncertainties [27].

2.1.5. Clinical Victim Studies

A stand‐alone clinical victim DDI study involving the administration of an investigational drug alone and in combination with an index inhibitor (i.e., concomitant drugs that is known to be a potent or strong inhibitor of a single metabolic pathway) or inducer to healthy volunteers is considered the gold standard for evaluating victim DDIs. Table 1 summarizes the main points for consideration when designing most clinical victim DDI studies, some of which have been summarized previously [17].

TABLE 1.

Clinical design of drug–drug interaction studies of inhibition of metabolism or transport of an investigational drug.

Requirement	Clinical considerations
Study design	Randomized crossover study (A → A + I vs. A + I → A) may be appropriate for an investigational drug and a specific inhibitor with a short half‐life
	Sequential crossover study is the most common type of clinical DDI study, where the investigational agent is administered alone and then with an inhibitor. This study is appropriate for most investigational agents with a reasonable half‐life (≤ 3 days) of the investigational drug or inhibitor
	Parallel study may be needed when the investigational drug has a very long half‐life
Choice of inhibitor	Choose a potent inhibitor that is specific for pathway of interest (refer to M12 guidance or Certera DIDB)
	Use a strong “index” inhibitor so that results can be extrapolated to other drugs metabolized by the same DME
	There should be little or no overlap with other pathways, which may not be possible for some UGTs or transporters. In this case, you may have to choose 2 inhibitors to study
Dose of inhibitor	Dose should produce maximum inhibition based on literature
	Can be dosed as a single dose (if investigational drug has a short half‐life) or at steady‐state (if investigational drug has a longer half‐life). If repeat doses of the inhibitor are administered, the dose and duration should be within the approved package insert
	The inhibitor must be present at clinically relevant inhibitory concentrations for a duration of blood sampling that can measure 3–5× the half‐life of the investigational agent
Dose of investigational agent	The dose should be lower than the maximum tolerated dose studied to date. Choosing the dose may be difficult depending on which DME is involved and how sensitive a substrate the investigational drug is
	Magnitude of CYP inhibition >> UGT inhibition
	Understanding the magnitude of drug interactions of other DME substrates is important in determining the dose of investigational agent to be studied. PBPK modeling may help in identifying the most appropriate dose to study
Duration of dosing of investigational agent	Typically a single dose is adequate (> 98% of cases), unless there is a metabolite that is has a concentration–time profile that cannot be measured after a single dose, but can be measured at steady‐state (rare case)
Timing of administration	If a single dose of inhibitor and investigational drug are administered, the inhibitor should be administered first and the investigational drug should be administered around the T max for the inhibitor
	If the inhibitor is given to steady‐state, the investigational drug and the inhibitor can be administered simultaneously once the inhibitor is at steady state
Blood sampling schedule	Should be adequate to characterize 3–5 half‐lives in the presence and absence of an inhibitor, considering that the half‐life may be longer in the presence of an inhibitor
Number of subjects	The sample size should be sufficient to demonstrate reliable estimates of the magnitude and variability of the interaction. Typically, a 2–3 times larger sample may be needed to demonstrate a lack of DDI than the number of subjects needed to characterize the magnitude of an expected interaction

In addition to stand‐alone clinical studies, victim DDIs can sometimes be evaluated as part of a nested study or as part of popPK analysis. For investigational drugs that cannot be administered to healthy subjects, a substudy (i.e., nested study) may be done within a Phase 2/3 study in patients for which a DDI is not the primary objective. In addition, if some patients with a disease are expected to be concomitantly administered a specific inhibitor with the investigational drug, the victim DDI may be evaluated as part of a popPK analysis, but only if there are a sufficient number of subjects who receive the investigational drug with and without the inhibitor.

2.1.6. Pharmacogenetics

Pharmacogenetics can be used as a tool to characterize victim DDIs in some cases. If a drug is metabolized by a polymorphic enzyme, then the magnitude of a DDI will be the same as the difference between extensive metabolizers and poor metabolizers. For example, if an investigational agent is a CYP2D6 substrate, exposure in a CYP2D6 poor metabolizer represents the magnitude of a drug interaction in a CYP2D6 extensive metabolizer receiving a concomitant CYP2D6 inhibitor. Therefore, if a clinical study of CYP2D6 extensive and poor metabolizers has been conducted, a clinical DDI study of the investigational drug with a CYP2D6 inhibitor may not be needed [9, 10].

2.2. Developing the Victim DDI Strategy

The victim DDI strategy evolves throughout drug development as new information is gained. The initial strategy begins by using in vitro data, which should be updated when data from a hADME study are available, and/or when data from PBPK and clinical study results are available (Figure 1). Below is a description of the considerations for developing and implementing the victim DDI strategy.

2.2.1. Preclinical Through Phase 2

In vitro studies evaluating whether an investigational agent is a substrate for DMEs (see Section 2.1.1) and some transporters (see Section 2.1.3) should be conducted prior to the investigational drug being administered to patients with a disease, considering that concomitant medications are likely to be administered to these patients and one must optimize the benefit‐to‐risk profile in early studies by reducing any DDI risks [9, 10]. If the route of metabolism is unknown, then patients taking drugs with the potential to inhibit or induce any enzyme or transporter must be excluded, making a Phase 2 study less feasible to conduct. Therefore, it is important to conduct these in vitro studies to identify metabolic pathways prior to or during a first‐time‐in‐human (FIH) study in healthy volunteers.

In some circumstances, in vitro studies may not give the appropriate information needed to assess the need for clinical DDI studies. For example, if a drug is only slowly metabolized in hepatocytes or if a drug is unstable in human plasma, serum, or hepatocytes, it may be difficult to rule out that an investigational agent is a substrate for a specific DME. If a drug is not a substrate for standard CYP or UGT enzymes, then one should consider less common DMEs (e.g., CES, alcohol dehydrogenase, epoxide hydrolases) [9, 10]. Also, additional information may be gained from the hADME study (see Section 2.1.2) to indicate what further metabolic enzymes need to be evaluated.

Once the in vitro metabolism and transport data are available, the key concept for the victim DDI strategy is identifying what stand‐alone clinical studies are needed, when PBPK can eliminate the need for a clinical study or contribute to the design of clinical studies, and when popPK (or nested studies) may contribute to the victim DDI information [9, 10], with the overall goal to identify when dosage adjustments of the investigational drug are needed.

The second key concept is the timing of these studies. There is no one‐size‐fits‐all approach for the timing of PBPK and clinical victim DDI studies. The timing will be highly dependent on whether excluding subjects taking certain specific inhibitors in a Phase 2 study is feasible based on their use in the disease of interest. If the use of specific inhibitors is common in a patient population of interest, then, at a minimum, PBPK (see Section 2.1.4) should be conducted prior to the conduct of a Phase 2 study. In some cases (or depending on the PBPK results), clinical victim DDIs (see Sections 2.1.4 and 2.1.5) may need to be conducted prior to the start of a Phase 2 study, particularly if the magnitude of a DDI is predicted to be high (e.g., CYP3A4 or CYP2D6 substrates). If the predicted magnitude of a DDI is expected to be relatively low (e.g., 2‐fold or less based on PBPK), then a clinical study can be conducted concurrently during Phase 2 if there are appropriate safety margins for the investigational drug. Ideally, it is best to begin conducting clinical DDI studies prior to Phase 3, but not all clinical victim DDI studies have to be conducted before the start of Phase 3.

It is important to consider that PBPK modeling (see Section 2.1.4) has been extensively used in making international regulatory decisions around the potential for DDIs and the clinical consequences [28, 29, 30, 31, 32, 33, 34, 35], with at least ten reported examples of when PBPK modeling was considered “high impact” (e.g., where PBPK eliminates the need for a clinical DDI study) [27].

Early on during drug development, PBPK may be used to evaluate the magnitude of a victim DDI (where there is a potent and specific index inhibitor or inducer) before a clinical study is conducted, to ensure dose selection of the investigational drug in the presence of victim DDIs is optimized (see Section 2.1.4). Once the clinical victim DDI study is completed, the PBPK model can be updated and the use of an investigational drug concomitantly with moderate or weak inhibitors from this pathway can be predicted [27].

PBPK models are particularly valuable in assessing DDIs involving substrates of well‐known interactions (e.g., CYP3A4, CYP2D6, CYP2C19, P‐gp), offering insights into how enzyme or transporter inhibition/induction affects the exposure to an investigational agent. These models guide dose adjustments, improve safety, and optimize therapy by predicting clinical scenarios proactively [25, 26]. By identifying and mitigating risks, PBPK models refine study designs for DDI assessments and enhance benefit: risk profiles in drug therapy (see Section 2.1.4).

Other uses of PBPK may include when an investigational drug is a substrate for P‐gp or organic anion transporting protein (OATP), but are much less likely to be successful at eliminating the need for a clinical DDI study; PBPK models of transporter interactions likely contribute to information on the study design of a DDI (e.g., particularly the dose of the investigational drug that can be studied safely). A great example of this use of PBPK is described for the effect of OATP1B1 inhibitors (e.g., rifampicin, cyclosporin) on statins [36, 37]. When clinical data become available, PBPK models can be validated by comparing model predictions with actual clinical outcomes to confirm the reliability and applicability of PBPK models in predicting DDIs (see Section 2.1.4).

2.2.2. Phase 2 Through NDA Submission

As stated previously, the victim DDI strategy evolves throughout drug development. For example, in vitro data can usually identify the DMEs predominantly contributing to the metabolism, but a hADME study conducted concurrently with Phase 2 will be needed to confirm the major elimination pathway and develop an understanding of the contribution of other metabolic pathways (e.g., renal, biliary) to the elimination of an investigational drug and what transporters should be evaluated (see Section 2.1.3). Thus, once the hADME study results are available, it is important to plan the clinical studies needed and assess whether PBPK can provide insights for a clinical study or replace a clinical study [19, 27, 30, 31].

When planning clinical studies, the need for a standard clinical victim DDI study (see Section 2.1.5) is determined using a risk‐based approach using data gained from in vitro studies of metabolism and transport, PBPK, and results from a human hADME study, when available [9, 10]. The clinical victim DDI study (see Section 2.1.5, Table 1) is then conducted with an index inhibitor for an important elimination pathway for the investigational drug. If there is no DDI when an investigational agent is coadministered with a potent and specific inhibitor of the predominant pathway, then a DDI with any drugs inhibiting that specific pathway can be ruled out. The same approach can be used for specific inducers. Thus, using index inhibitors or inducers reduces the need for multiple potential clinical victim DDI interaction studies.

For standard clinical victim DDI studies (see Section 2.1.5), a DDI can be ruled out if the 90% confidence limit for the geometric mean ratio (investigational drug + concomitant inhibitor/investigational drug alone) is within 80%–125% (i.e., the bioequivalence criteria). The M12 guidance allows for less strict acceptable 90% confidence limits based on the therapeutic range for the investigational drug, the type of toxicity expected if maximum plasma concentration (C _max) or area under the curve (AUC) is high, and the consequences of the investigational drug being ineffective. Importantly, the acceptable 90% confidence no‐effect limits need to be prespecified [9, 10].

If an interaction is not ruled out in a clinical DDI study (see Section 2.1.5) with a potent index inhibitor, one can decide to use a PBPK model or popPK model to evaluate the magnitude of a DDI with weak or moderate inhibitors, if needed. For example, if the investigational agent is a CYP3A4 substrate and there is a 5‐fold increase in the exposure of the investigational agent in the presence of a potent inhibitor like itraconazole, the PBPK (see Section 2.1.4) or popPK approach (see Section 2.1.5) may help in identifying the magnitude of an interaction and clinical relevance when the investigational agent is concomitantly administered with a weak or moderate CYP3A4 inhibitor and whether any dose adjustments are needed.

Not all DDIs can be evaluated using index inhibitors. For some enzymes (UGTs) or transporters, there is no substrate specificity, and there is no index inhibitor. In these cases, it is best to choose an inhibitor based on the magnitude of the interaction seen with other substrates for the specific transporter or DME (see Appendix G of the M12 guidance) [9, 10]. For example, most P‐gp inhibitors also are strong CYP3A4 inhibitors (e.g., itraconazole, clarithromycin, telaprevir) or moderate CYP3A4 inhibitors (e.g., verapamil, erythromycin). Some P‐gp inhibitors also induce P‐gp with repeat dosing (e.g., ritonavir, verapamil). The effects of different inhibitors on P‐gp substrates suggest that itraconazole has the largest effect on dabigatran, but itraconazole also inhibits CYP3A4 [38]. Thus, evaluating the most appropriate inhibitor for an investigational drug that is a P‐gp and a CYP3A4 substrate is complicated. The availability of metabolites helps in the interpretation of the results from a study with itraconazole. If a victim DDI is just due to inhibition of P‐gp, then one would expect that the parent and metabolite concentrations would both increase in the presence of itraconazole. However, if the DDI was due to CYP3A4 inhibition, then the parent exposure would increase and the metabolite exposure would decrease in the presence of itraconazole.

Other clinical victim DDI studies conducted include the evaluation of induction of DMEs and transporters on the PK of the investigational drugs. The timing of an inducer study depends on the feasibility of excluding inducers during Phase 2/3 studies; generally, induction studies can typically be conducted concurrently with Phase 3. In addition, in most cases (except combinations of antivirals or agents used in the treatment of tuberculosis), a dosage recommendation for an investigational product administered with an inducer cannot be determined. In this case, the exposure to the parent drug decreases, but exposure to the metabolites increases substantially, making it difficult to assess the tolerability of higher metabolite exposure without further preclinical or clinical studies. Studies to evaluate the effect of an inducer on the PK of an investigational drug are conducted as sequential crossover studies (Table 1) or as parallel studies. The inducer should be administered to steady state and for enough time to have maximum induction of DMEs and transporters (e.g., 6–8 days) prior to administration of the investigational drug, and administration of the inducer should be continued throughout the sampling for the investigational drug.

There are other types of studies that may need to be considered to evaluate victim DDIs for an investigational drug. For example, drug interaction studies may be conducted to evaluate whether an acid reducing agent affects the rate or extent of absorption of an investigational agent. The issues relevant to interactions with acid lowering agents and the need for clinical studies are outlined elsewhere [39]. In addition, any pharmacodynamic interactions between a concomitant medication and the investigational drug need to be considered but are outside the scope of this tutorial.

The long‐term goal of the victim DDI strategy is to provide physicians with labeling guidance based on in vitro data, clinical studies, PBPK, and/or popPK studies once a drug is approved. At the end of Phase 3, there may still be untested scenarios (e.g., combinations of drugs inhibiting multiple pathways) that are also important. PBPK can play a crucial role in evaluating these untested scenarios and supporting product labeling, particularly for drugs metabolized by CYP3A and CYP2D6 [30]. For example, a drug may primarily be a CYP3A4 substrate, with CYP2D6 as a secondary pathway. One would expect that the magnitude of a DDI with a CYP3A4 inhibitor would differ in CYP2D6 poor metabolizers (which are unfeasible to study) and CYP2D6 extensive metabolizers (with available data). Thus, PBPK can be used for an investigational drug to predict the magnitude of a DDI when a CYP3A4 inhibitor is coadministered with the investigational drug in CYP2D6 poor metabolizers.

DDI information is included in the US product label in at least 2 sections (each with a different format of similar information): Section 7 DRUG INTERACTIONS and Section 12.3 Pharmacokinetics Drug Interactions (Figure 3, Ref. [40]). Generally, the more detailed results from PBPK or clinical studies will be included in Section 12.3 Pharmacokinetics Drug Interactions, typically including a forest plot illustrating the magnitude of an interaction (Figure 3, Ref. [40]); in limited cases, statements or tables will be presented instead of forest plots. In addition, in vitro data showing that an investigational drug is not a substrate for specific DMEs (see Section 2.1.1) or transporters (see Section 2.1.3) is important data to be included in this section of the product label.

FIGURE 3.

Proposed format for drug‐drug interaction information in Section 7 and 12 of the product label. Adapted from Tran et al. [40].

A table will almost always be presented within Section 7 DRUG INTERACTIONS (Figure 3, Ref. [40]), focusing on the recommendation to practicing physicians. It is also important at this stage to determine whether the recommendations include a change in the dose, a contraindication, or additional Warnings and Precautions (all with links back to Sections 7 and 12.3). These recommendations will be determined based on the benefit‐to‐risk profile for the investigational drug and the safety risks for the concomitant medication, not using a one‐size‐fits‐all approach.

General guidance around product labeling of DDIs is outlined in an FDA guidance for the US and in an EMA guidance for Europe [9, 10, 41, 42]. For the US, there is some controversy about when a lack of a drug interaction should be included in Section 7. The guidance states “The primary focus of the description of the clinically significant DI and its prevention or management instructions”; Many interpret this to state that only positive DDIs are clinically relevant and should be mentioned in Section 7. However, from a different perspective, the lack of a DDI can also be clinically relevant given that it allows the concomitant use of two drugs without any changes to dosing regimens [42].

Overall, the strategy presented above allows victim DDIs to be fully understood, along with their consequences (e.g., exclusion criteria or dose adjustments), considering available data and appropriate tools (see Sections 2.1.1, 2.1.5) to optimize inclusion/exclusion criteria throughout drug development, product labeling, and the benefit‐to‐risk profile for an investigational drug to treat disease with unmet medical needs.

3. Evaluation of Drug Interactions Where the Investigational Agent May Increase or Decrease Exposure to Concomitant Medications (i.e., Investigational Drug as a Perpetrator)

The key to understanding the potential for DDIs based on the ability of an investigational drug to affect the PK of concomitant medications (i.e., investigational drug as a perpetrator) is to identify whether an investigational drug and its important metabolites (e.g., account for more than 10% of the radioactivity in humans and at least 25% of the AUC for parent) induce or inhibit drug metabolism and transport.

3.1. Tools Used in the Evaluation of Perpetrator Drug–Drug Interactions

3.1.1. In Vitro Studies for Induction

The cornerstone to evaluating perpetrator DDIs is in vitro studies evaluating the induction or inhibition of DMEs and transporters [18, 43]. Studies of induction typically include whether parent and metabolites can induce CYP1A2, CYP2B6, and CYP3A4. Induction observed with CYP3A4 applies to CYP2Cs and P‐gp, given the common mechanism of induction through transcription factors pregnane X receptor and constitutive androstane receptor [44], while induction of CYP1A2 is associated with the transcription factor aryl hydrocarbon receptor. The M12 guidance suggests that CYP2C8, CYP2C9, and CYP2C19 are less inducible than CYP3A4 [9, 10]. However, at least one case has shown that an investigational drug produced weak induction of midazolam (CYP3A4) and omeprazole (CYP2C19) to the same magnitude (44%) [45]. Thus, in general, if there is induction of CYP3A4, then induction of CYP2Cs should also be evaluated as part of PBPK or in a clinical study.

3.1.2. Equations for Assessing Clinical Relevance of Induction

The M12 guidance describes several ways to identify whether induction in vitro will lead to clinically relevant induction based on the magnitude of change in mRNA expression, correlation methods, or a basic kinetic model [9, 10]. Generally, if all three donors have < 2‐fold change in mRNA at 50× the unbound Cmax (C _max,u) in the presence of the investigational drug, then induction can be excluded. For all other cases, one can use the correlation method (a relative induction score compared with well‐known probes) or the basic model. The basic model uses the concentration producing 50% of maximum induction (IC₅₀) and the C _max,u to calculate an R value; if the R value is less than 0.8, then a clinical study is needed to evaluate whether an investigational drug produces clinically relevant induction [9, 10]. Calculations should be done for parent and metabolite after steady‐state dosing with the relevant clinical dose(s). The simplest methods may be used first, followed by PBPK modeling for those cases that are more difficult to interpret.

3.1.3. In Vitro Studies for Inhibition

In vitro studies can evaluate whether the investigational drug and its metabolites can inhibit CYP isoenzymes, first by looking at reversible inhibition and then by time‐dependent inhibition (TDI). Because an investigational drug that inhibits drug metabolism or transporters does not have to be a substrate of the particular DME or transporter, inhibition of multiple DMEs needs to be evaluated, including a broad range of CYPs and transporters [9, 10]. The magnitude of interactions due to UGTs is generally smaller than those due to inhibition of CYPs. Therefore, the M12 guidance suggests that the in vitro studies for UGTs should be conducted when an investigational drug is used concomitantly with a drug metabolized by UGTs [9, 10].

3.1.4. Equations for Assessing Clinical Relevance of Inhibition

Many investigational drugs demonstrate in vitro inhibition of DMEs or transporters, but not all lead to clinically relevant DDIs. To assess clinical relevance, the inhibition constant (Ki) must be evaluated in the context of C _max,u, accounting for the parent drug and key metabolites at steady‐state for the clinical dosing regimen [9, 10]. Early in drug development, the clinical dose(s) and time to steady‐state may be unknown. In such cases, conservative assumptions should be used for predictions. As clinical development progresses and more PK data become available (e.g., PK of parent drug and metabolites, the maximum tolerated dose, and the dose likely to achieve efficacy), the assessment of clinically relevant in vitro drug inhibition should be refined. This can lead to revised predictions. Importantly, the perpetrator DDI strategy is dynamic and evolves throughout drug development (Figure 1) [46, 47].

The M12 guidance outlines the equations used to evaluate the potential for clinically relevant inhibition or TDI of CYPs with cutoff values (referred to as R values) that suggest that a clinical study is needed. For example, the risk of clinically relevant reversible inhibition of a CYP isoenzyme can be ruled out if the k _i is at least 50 times higher than the C _max,u [9, 10].

Inhibition of the following transporters should be evaluated in vitro for parent and important metabolites: P‐gp, BCRP, OATP1B1, OATP1B3, OAT1, OAT3, OCT2, MATE1, and MATE2‐K. The clinical relevance of inhibition of these transporters by an investigational drug can be determined based on the equations outlined in Table 1 of the M12 guidance [9, 10]. Because P‐gp and BCRP are found in the GI tract, the equations consider the amount of drug in the GI tract along with the IC₅₀. It is important to make sure the units are the same when using this equation (i.e., nM is equivalent to nmoles/L; therefore, 250 mL is converted to 0.25 L). OATP1B1 and OATP1B3 utilize the estimated unbound concentrations at the liver inlet. All other transporters utilize the C _max,u, but the cutoff value for clinical relevance differs among transporters. It is also important to evaluate any new transporters identified by the ITC (https://www.itc‐transporter.org/).

3.1.5. PBPK Modeling for Induction and Inhibition

As seen for victim DDIs (see Section 2.1.4), PBPK can be very useful in understanding the effect of an investigational drug on the PK of concomitant medications. PBPK is especially helpful in evaluating the induction or inhibition by an investigational drug, given the availability of well‐studied probe substrates (midazolam for CYP3A4, bupropion for CYP2B6, and caffeine for CYP1A2) [48, 49].

3.1.6. Clinical Perpetrator Studies

A stand‐alone clinical study with a relevant index substrate administered before and after administration of steady‐state dosing with the investigational drug (i.e., sequential DDI study) is the gold standard for the evaluation of perpetrator DDIs. The highest clinical dose of the investigational drug (i.e., the dose expected to be approved) should be administered for a time period that allows the investigational drug and its metabolites to reach the steady state.

One or two separate stand‐alone studies can be conducted if an investigational drug only inhibits one or two DMEs. However, in many cases, the investigational drug may have the potential to induce or inhibit multiple DMEs and transporters. For this case, a single study can be conducted using a probe cocktail or a microdose probe cocktail instead of multiple stand‐alone studies.

3.1.7. Probe Cocktail Study

Probe cocktails involve the simultaneous administration of specific substrates for multiple enzymes and/or transporters to study subjects. The key to success is to ensure that the substrate drugs do not interact with each other and are, therefore, considered validated.

Validated cocktails include: “The Pittsburgh cocktail,” “Cooperstown cocktail”, “Basel and Geneva cocktail,” “Greenford‐Ware cocktail,” “Inje cocktail,” and “The Oita Combination cocktail” (Refs. [50, 51, 52, 53, 54, 55, 56, 57, 58, 59], Table 2). To date, most of the proposed cocktails include index drugs for up to six probes. Available validated probe cocktails are reviewed elsewhere [56, 58, 59]. It is also important to note that interactions at the transporter level among substrates have not always been evaluated.

TABLE 2.

Clinically validated cocktails involving CYP‐mediated DDIs.

	Pittsburgh cocktail	Zhu cocktail	Karolinska cocktail	Blakey cocktail	Cooperstown 5 + 1 Cocktail	Inje cocktail	Turpault cocktail	Basel cocktail	Geneva cocktail
Publication date	September 2006	November 2001	June 2003	February 2004	November 2003	November 2007	December 2009	March 2014	September 2016
CYP1A2	Caffeine 100 mg	Caffeine 100 mg	Caffeine 100 mg	Caffeine 100 mg	Caffeine 2 mg/kg	Caffeine 93 mg	Caffeine 100 mg	Caffeine 100 mg	Caffeine 50 mg
CYP2B6								Efavirenz 50 mg	Bupropion 20 mg
CYP2C9	Flurbiprofen 50 mg		Losartan 25 mg	Tolbutamide 250 mg	Warfarin 10 mg	Losartan 30 mg	Warfarin 10 mg	Losartan 12.5 mg	Flurbiprofen 10 mg
CYP2C19	Mephenytoin 100 mg	Mephenytoin 100 mg	Omeprazole 20 mg		Omeprazole 40 mg	Omeprazole 20 mg	Omeprazole 20 mg	Omeprazole 10 mg	Omeprazole 10 mg
CYP2D6	Debrisoquine 10 mg	Metoprolol 100 mg	Debrisoquine 10 mg	Debrisoquine 5 mg	Dextromethorphan 30 mg	Dextromethorphan 30 mg	Metoprolol 100 mg	Metoprolol 12.5 mg	Dextromethorphan 10 mg
CYP3A4	Dapsone 100 mg	Midazolam 7.5 mg	Quinine 250 mg	Midazolam 0.025 mg/kg (i.v.)	Midazolam 0.025 mg/kg (i.v.)	Midazolam 2 mg	Midazolam 0.03 mg/kg (i.v.)	Midazolam 2 mg	Midazolam 1 mg
CYP2E1	Chlorzoxazone 250 mg	Chlorzoxazone 250 mg		Chlorzoxazone 250 mg

Note: The dosing route is oral if not specified. Modified from Yu et al. reference [59].

Abbreviations: DDI, drug–drug interaction; IV, intravenous.

Despite the advantages of using multiple probes in a probe cocktail, the dosing, safety monitoring, and sampling schedule can become complicated. First, the study design needs to consider formulations and availability of each substrate. For example, debrisoquin and mephenytoin are not available in the US. One must consider safety issues including the need for back‐up vitamin K if warfarin is used; glucose monitoring if tolbutamide is used; blood pressure monitoring if metoprolol is used; and monitoring for sedation if midazolam or debrisoquin are used. One should consider the ability to characterize the full plasma concentration–time profile for each probe substrate with or without metabolites in the presence and absence of an inhibitor for 3–5 times the half‐life, using a sensitive and specific assay.

Some alternatives to full plasma concentration–time profiles have been advocated, including the plasma or urinary ratio of parent‐to‐metabolite for dextromethorphan and caffeine and the fraction of metabolite in urine over 8 h for mephenytoin or debrisoquine [56]. These methods are more variable and require a larger number of subjects to rule out a DDI.

3.1.8. Microdose Probe Cocktail Study

Microdose probe cocktails have been introduced to reduce the dose of the substrates and remove potential for DDIs among substrates in the cocktail. Microdoses (dose < 1/100th of the pharmacologically active) avoid the pharmacological effects of substrates themselves, improving safety in the presence of large increases in probe substrate exposure. These microdose probe cocktails have been described elsewhere [56, 59, 60, 61, 62].

A challenge remains in extrapolating findings to clinically relevant doses, particularly when the substrate exhibits nonlinear pharmacokinetics. Magnitudes of DDIs were similar for microdoses and standard doses of midazolam, tolbutamide, caffeine, and fexofenadine [60], midazolam [63], omeprazole [64], and pitavastatin, atorvastatin, midazolam, and rosuvastatin [62]. In contrast, dabigatran microdose studies showed some nonlinearity in the magnitude of the interaction between standard and microdoses, with a higher magnitude of an interaction at lower doses. Since P‐gp in the GI tract is known to be saturable (microdose therefore represents the worst‐case scenario) and only small (< 2‐fold) interactions have been reported for P‐gp substrates in the presence of P‐gp inhibitors. Thus, the dabigatran microdose represents the worst‐case scenario and can be used to address the question of whether there is a clinically relevant DDI based on in vitro data.

One must also consider the availability of a solution formulation and the need for the solutions to be produced onsite with relevant stability until the time of administration.

3.1.9. Endogenous Biomarkers

Endogenous compounds can assess the impact of an investigational drug on specific pathways, without requiring administration of index substrates. Endogenous biomarkers such as 4β‐hydroxycholesterol have been used to characterize CYP3A4 induction [65]. This biomarker can identify whether an investigational drug is a potent inducer, but not a weak inducer, given high intersubject variability.

The ITC is currently evaluating the use of endogenous biomarkers for the evaluation of inhibition of transporters by an investigational drug [66] Coproporphyrin I (CPI) has been validated to increase with OATP1B1/OATP1B3 inhibition [67]. Other endogenous biomarkers of renal and hepatic transporters have not been fully validated but could be collected in clinical phase I studies as supportive evidence when in vitro studies suggest inhibition of transporters [66].

4. Developing the Perpetrator DDI Strategy

The perpetrator DDI strategy evolves throughout drug development. The initial strategy begins by using in vitro data, which should be updated when data from a hADME study are available, and/or when data from PBPK and clinical study results are available (Figure 1). Below is a description of the considerations for developing and implementing the perpetrator DDI strategy using the tools provided.

4.1. Preclinical Through Phase 2

In vitro studies (Refer to Sections 3.1.1 and 3.1.3 ‐ In Vitro Studies for Induction and In vitro Studies for Inhibition) should be conducted prior to the investigational agent being administered to patients with a specific disease, considering that concomitant medications are likely to be administered to these patients to reduce risks to patients. Initially, these studies should be conducted for the parent and for any major metabolites based on metabolism (see Sections 2.1.1 and 2.1.2) [9, 10]. If these studies are not conducted, then patients taking concomitant medications that are substrates for any DMEs or transporters must be excluded, making a Phase 2 proof of concept study less feasible to conduct. Therefore, like for victim DDIs, it is important to conduct these perpetrator in vitro induction and inhibition studies to identify whether the investigational drug has the potential to affect the PK of concomitant medications prior to or during a typical FIH study in healthy volunteers [68].

Calculations should be conducted (Section 3.1.2 and 3.1.4) with the understanding that these calculations will be updated as data emerges. During preclinical development, the C _max at steady state is based on predicted PK, which is updated once single‐dose PK data is available, and again when repeat‐dose PK data is available. At these early stages, the Phase 3 dose is unknown; therefore, conservative estimates (i.e., high end of potential dose range) of the dose should be used for the calculation of clinical relevance [9, 10, 66, 69].

After conduct of the hADME study (Section 2.1.2), there is a better understanding of the metabolic pathway of the investigational drug. The in vitro inhibition studies (and calculation of clinical relevance) should be conducted for metabolites only if the AUC of the metabolite is greater than or equal to 25% of AUC for the parent and also accounts for at least 10% of the drug‐related material in circulation. Once again, there may be little information about the metabolite Cmax after steady‐state dosing with the investigational drug. Therefore, conservative estimates of the metabolite Cmax should be used and the calculations should be updated when more metabolite PK data is available [9, 10, 66, 69].

Once in vitro data and R values (based on C _max,u in humans at steady‐state) are known for DMEs and transporters, a decision tree may help in deciding what clinical studies or PBPK modeling are needed (Figure 4) and when the clinical studies should be conducted.

FIGURE 4.

Decision tree for the evaluation of the need for clinical drug–drug interaction studies or PBPK based on in vitro data and evolving information.

If the R values are under the cutoff criteria, then concomitant medications that are substrates for these DMEs and/or transporters do not require exclusion from future clinical trials, and statements in the product labeling will usually be made in Section 12.3 (Pharmacokinetics: Drug Interactions) and in Section 7 (Drug Interactions), stating that no drug interaction is expected [40].

For DMEs and transporters that have an R value above the cutoff, there are different consequences depending on what phase of drug development the investigational drug is in, the disease being treated, the concomitant medications used to treat the disease or comorbid conditions, the feasibility of conducting a future clinical efficacy study excluding concomitant medications metabolized or transported by the relevant DMEs or transporters, and whether the clinical dose is established. For example, the clinical dose is unknown in an early Phase 2 study, and multiple concomitant medications that are substrates of relevant DMEs/transporters may be administered in the disease of interest. PBPK modeling may be particularly useful at this stage to help understand the magnitude of the DDI at the highest clinical dose of the investigational drug, whether relevant concomitant medications need to be excluded, or whether a clinical DDI study is needed prior to starting the Phase 2 study (see Section 3.1.5, Figure 1).

When a clinical perpetrator DDI study is planned, it must be conducted at the highest well‐tolerated clinical dose (e.g., after the single ascending dose and multiple ascending dose studies have been conducted in Phase 1) or at the clinical Phase 3 dose (or at the highest Phase 3 dose). If there is a very wide potential range of doses and results from PBPK modeling suggest a low DDI risk, one may choose to wait until the Phase 3 dose is known to conduct the stand‐alone DDI clinical studies. Another approach is to conduct the studies at the highest clinical dose in Phase 1 and then conduct PBPK modeling to look at the clinical relevance of lower doses if the Phase 3 dose is much lower than the highest dose studied in Phase 1.

The decision of whether to use individual studies, a probe cocktail study, or a microdose probe cocktail study should be based on the number of DMEs and transporters with potentially clinically relevant inhibition or induction by the investigational drug. If there are only up to 3 DMEs or transporters inhibited, then single stand‐alone studies are feasible. If more DMEs or transporters are inhibited, or if there is induction across CYPs, it is more efficient to use a probe cocktail or microdose probe cocktail approach (Sections 3.1.7 and 3.1.8). These types of studies may require more than one cohort if the data for use across substrates either do not exist or there is the potential for known DDIs. For example, bupropion is a CYP2B6 substrate but inhibits CYP2D6; therefore, if an investigational drug induces CYP2B6 and inhibits CYP2D6, the evaluation of these effects of an investigational drug on CYP2B6 and CYP2D6 should be in two separate cohorts.

It is important to use sensitive index substrates for these clinical perpetrator DDI studies (Appendix F of the M12 guidance [9, 10]). Sensitive substrates for a specific pathway are those that result in an increase in AUC ≥ 5× with a strong index inhibitor. Moderately sensitive substrates are drugs with an increase in AUC ≥ 2× but < 5×. If a clinical study of a sensitive index drug alone and after administration of the investigational drug results in the 90% confidence interval of the geometric mean ratio (probe substrate + investigational drug/probe substrate alone) being within 80% to 125%, then a drug interaction can be ruled out.

The key difference between a clinical victim and clinical perpetrator studies (see Section 2.1.5 and 3.1.6) is the number of subjects needed. For clinical perpetrator studies, the study needs to be powered to exclude an effect of the investigational drug on a specific substrate, given that in the majority of cases, the in vitro/in vivo correlation of DDIs is not absolute. In contrast, one expects an increase or decrease in the investigational drug with an enzyme inhibitor or inducer for clinical victim studies, so the number of subjects can be smaller to describe the magnitude of the DDI. The number of subjects needed for a clinical perpetrator DDI study should be based on data from well‐known index substrates or previously conducted probe cocktail or microdose probe cocktail studies.

When there are effects of an investigational drug on the PK of an index substrate, there are clear definitions for the interpretation. For example, if the 90% confidence interval for the geometric mean ratio (probe substrate + investigational drug/probe substrate alone) is outside the 80%–125% rule, then the investigational agent can be classified as a potent (≥ 5× increase in AUC), moderate (≥ 2× to < 5× increase in AUC), or weak inhibitor (< 2× increase in AUC) of specific pathways, depending on the magnitude of the interaction. Similarly, for inducers, an investigational drug would be considered a strong inducer (≥ 80% decrease in AUC), moderate inducer (< 80% to ≥ 50% decrease in AUC) or weak inducer (< 50% decrease in AUC) of a sensitive index substrate.

It is important to be aware that even with the use of index drugs, there is some overlap between substrate specificity. For example, omeprazole is metabolized by CYP2C19 and CYP3A4, and repaglinide is metabolized by CYP2C8 and CYP3A4. To overcome these limitations, collection of metabolite PK can identify which pathways are inhibited or induced.

Studies of inhibition or induction of transporters are more complex. There are no index drugs for most transporters because there is an overlap in specificity. A substrate for P‐gp for a clinical perpetrator study must be chosen based on evidence that the change in AUC is 2× after administration of itraconazole, quinidine, or verapamil; it is an in vitro substrate for P‐gp and is not extensively metabolized. Potential P‐gp substrates to be used in a clinical perpetrator study have been summarized by Chu et al. [38]. Substrates for other transporters are much less well established but are described in Table 18 of Appendix F of the M12 guidance [9, 10].

If an investigational drug inhibits transporters, then the use of endogenous biomarkers in a clinical stand‐alone DDI study or a Phase 2 study can be evaluated, depending on the number of subjects needed (see Section 3.1.9). It is important to understand the variability in the endogenous biomarker prior to the decision to use an endogenous marker [70]. If variability is high, then evaluation of whether the investigational drug affects the endogenous biomarker may require large subject numbers (e.g., as a secondary or exploratory endpoint of a Phase 3 study). If variability is low, biomarkers may be studied in a stand‐alone clinical DDI study [66].

One special perpetrator DDI study is the use of an investigational drug with hormonal contraceptives. Because ethinyl estradiol is metabolized by multiple pathways (CYP3A4, CYP2C9, UGT1A1, and SULT1E1), the effects of an investigational drug on hormonal contraceptives are complex. If an investigational drug inhibits ethinyl estradiol [71], there may be an increased risk of serious adverse effects. If an investigational drug induces metabolism (even as a weak inducer), then breakthrough bleeding and unintended pregnancy can occur. Issues related to DDI's with hormonal contraceptives are described elsewhere [71, 72].

4.2. Phase 2 Through NDA Submission

Because the Phase 3 dose and the contribution of metabolites are better understood than prior to Phase 2, the calculations evaluating whether there is the potential for parent and metabolite to induce or inhibit metabolism and transport (Sections 3.1.2 and 3.1.4) should be updated, along with any information available from clinical studies (see Section 3.1.6) and PBPK (see Section 3.1.5) to develop the final perpetrator DDI strategy.

If clinical perpetrator studies have been completed, PBPK modeling characterizes how the Phase 3 dose changes the interpretation of the results from these studies. For instance, if the clinical perpetrator studies were done at a dose higher than or lower than the Phase 3 dose, PBPK modeling can be conducted to evaluate the magnitude of an interaction at the Phase 3 dose.

While almost all clinical perpetrator DDI studies (except in oncology) are conducted in healthy subjects, there may be differences in the magnitude of the DDIs due to genotype, age, or specific disease states. PBPK models may be used to better understand the magnitude of DDIs in specific populations (e.g., pediatrics, elderly, smoking, exercise) [73, 74, 75]. Additionally, PBPK models are being developed to predict DDIs in pregnant and lactating women. These models take into account the physiological changes that occur during pregnancy and lactation, helping to assess the safety and efficacy of medications in these populations and ensuring that both the mother and the fetus or infant are protected from harmful DDIs [76, 77]. Furthermore, PBPK models can simulate different dosing regimens of the perpetrator drug to understand the dose–response relationship and its impact on the index substrate, helping to optimize dosing strategies without conducting numerous clinical trials.

During Phase 4 development (e.g., new indications, new formulation, new routes of administration), PBPK becomes even more useful to evaluate the magnitude of perpetrator DDI interactions given the new doses, formulations, or routes of administration [78].

Similar to victim DDIs, PBPK models can be used to study complex DDIs involving multiple drugs, providing insights into whether interactions are cumulative or synergistic, information that may not be apparent when studying each drug individually [73, 79, 80] Additionally, PBPK models are valuable for understanding TDI and the effects over time, such as the combination of inhibition after a single dose and induction following repeated doses [20, 80].

Pharmacodynamic interactions also should be considered in the perpetrator DDI strategy. For example, a combination of drugs can increase the sedative effects in an additive or synergistic manner. Other drugs may decrease the seizure threshold or may increase the QT risk. These types of interactions are outside the scope of this tutorial.

By the end of Phase 3 (and at the time of NDA submission), consequences of perpetrator DDIs should be fully understood using a combination of data from PBPK models and clinical studies. Placement of these interactions in the product label follows the same guidelines as outlined for victim DDIs and will be determined based on the benefit‐to‐risk profile for the investigational drug and the safety risks for the concomitant medication, not using a one‐size‐fits‐all approach (Figure 3, Ref. [40]). Importantly, any changes in dosing requirements for concomitant medications need to be considered as they relate to their respective package insert.

5. Summary and Conclusions

In summary, given the incidence of polypharmacy in patients with comorbidities, it is important to evaluate DDIs to optimize the benefit‐to‐risk profile for an investigational drug. Strategies for the evaluation of victim DDIs and perpetrator DDIs for an investigational drug need to be developed early in drug development and continually updated as new data (e.g., clinical dose, PBPK, metabolites) become available. The most appropriate method for DDI evaluation is the use of a scientific risk‐based approach based on in vitro data, clinical data, and model‐based approaches (PBPK and popPK, where feasible) to optimize inclusion/exclusion criteria throughout drug development, to optimize dosing recommendations, and to support product labeling.

Conflicts of Interest

The authors declare no conflicts of interest. A.A.‐H. declared no competing interests for this work. V.D.S. declared no competing interests for this work. J.L.D. declared no competing interests for this work.

Lohy Das J., Ajavon‐Hartmann A. T., and Schmith V. G. D., “Navigating Drug–Drug Interactions in Clinical Drug Development: A Tutorial,” Clinical and Translational Science 18, no. 9 (2025): e70342, 10.1111/cts.70342.