Preliminary concept paper

Table 1. Examples of in vivo substrate, inhibitor and inducer for specific CYP enzymes have been recommended for study (oral administration) ^(1,2)

CYP	Substrate	Inhibitor	Inducer
1A2	theophylline, caffeine	fluvoxamine	smoking^(3)
2B6	efavirenz		rifampin
2C8	repaglinide, rosiglitazone	gemfibrozil	rifampin
2C9	warfarin, tolbutamide	fluconazole, amiodarone (use of PM subjects)^(4^)	rifampin
2C19	omeprazol, esoprazol, lansoprazol, pantoprasol	omeprazole, fluvoxamine, moclobemide (use of PM subjects)^(4)	rifampin
2D6	desipramine, dextromethorphan, atomoxetine	paroxetine, quinidine, (use of PM subjects)^(4)	None identified
2E1	chlorzoxazone	disulfirum	ethanol
3A4/ 3A5	midazolam, buspirone, felodipine, simvastatin, lovastatin, eletriptan, sildenafil, simvastatin, triazolam	atanazavir, clarithromycin, indinavir, itraconazole, ketoconazole, nefazodone, nelfinavir, ritonavir, saquinavir, telithromycin, voriconazole	rifampin, carbamazepine

⁽¹⁾ substrates for any particular CYP enzyme listed in this table are those with plasma AUC values increased by 2-fold or higher when co-administered with inhibitors of that CYP enzyme; for CYP3A, only those with plasma ACU increased by 5-fold or higher are listed. Inhibitors listed are those that increase plasma AUC values of substrates for that CYP enzyme by 2-fold or higher. For CYP3A inhibitors, only those increase AUC of CYP3A substrates by 5-fold or more are listed. Inducers listed are those that decrease plasma AUC values of substrates for that CYP enzyme by 30% or higher.

⁽²⁾ note that this is not an extensive list; for an updated list, see URL???

⁽³⁾ a clinical study may be conducted in smokers as compared to non-smokers (in lieu of an interaction study with an inducer), when appropriate

⁽⁴⁾ a clinical study may be conducted in poor metabolizers (PM) as compared to extensive metabolizers (EM) for the specific CYP enzyme (in lieu of an interaction study with an inhibitor), when appropriate.
If the initial study is negative with the most sensitive substrates, it can be presumed that less sensitive substrates will also be unaffected.
CYP3A inhibitors may be classified based on their in vivo fold-change in the plasma AUC of oral midazolam or other CYP3A substrate, when given concomitantly. For example, if an investigational drug increases the AUC of oral midazolam or other CYP3A substrates by 5-fold or more (>5-fold), it may be labeled as “strong” CYP3A inhibitor. If an investigational drug, when given at the highest dose and shortest dosing interval, increases the AUC of oral midazolam or other sensitive CYP3A substrates by between 2- and 5 fold ( > 2- and <5-fold) when given, it may be labeled as “moderate” CYP3A inhibitor. When an investigational drug is determined to be a “strong” or “moderate” inhibitor of CYP3A”, its interaction with “sensitive CYP3A substrates” or “CYP3A substrates with narrow therapeutic range” (see Table 2 in section V for a list) may be described in various sections of the labeling, as appropriate.
When an in vitro evaluation cannot rule out that an investigational drug is an inducer of CYP3A (section VI), in vivo evaluation may be conducted using the most sensitive substrate (e.g., oral midazolam). When midazolam has been co-administered following administration of multiple doses of the investigational drug, as may have been conducted as part of an in vivo inhibition evaluation, and the results are negative, it can be concluded that the investigational drug is not an inducer of CYP3A (in addition to the conclusion that it is not an inhibitor of CYP3A). In vivo induction evaluation has often been conducted with oral contraceptives. However, as they are not the most sensitive substrates, negative data may not exclude the possibility that the investigational drug may be an inducer of CYP3A.
2. Investigational Drug as Substrate of CYP Enzymes
In testing an investigational drug for the possibility that its metabolism is inhibited or induced (i.e., as a substrate), selection of the interacting drugs should be based on in vitro or other metabolism studies identifying the enzyme systems that metabolize the drug. The choice of interacting drug should then be based on known, important inhibitors of the pathway under investigation. For example, if the investigational drug is shown to be metabolized by CYP3A and the contribution of this enzyme to the overall elimination of this drug is substantial, the choice of inhibitor and inducer could be ketoconazole and rifampin, respectively, because of the substantial effects of these interacting drugs on CYP3A metabolism (i.e., they are the most sensitive in identifying an effect of interest). If the study results are negative, then absence of a clinically important drug-drug interaction for the metabolic pathway could be claimed. If the clinical study of the strong, specific inhibitor/inducer is positive and the sponsor wishes to claim lack of an interaction between the test drug and other less potent specific inhibitors or inducers, or give advice on dosage adjustment, further clinical studies would generally be recommended (see Table 1 for a list of CYP inhibitors and inducers and Table 3, section V for additional 3A inhibitors). If a drug is metabolized by CYP3A and its plasma AUC was increased by 5-fold or higher by CYP3A inhibitors, it is considered a “sensitive substrate” of CYP3A. The labeling may indicate that it is a sensitive CYP3A substrate and its use with strong or moderate inhibitors may be cautioned based on the drug’s exposure- response relationship (see section V for labeling implications). Certain approved drugs are not optimal selections as the interacting drug. For example, cimetidine is not considered an optimal choice to represent drugs inhibiting a given pathway because its inhibition affects multiple metabolic pathways as well as certain drug transporters.
3. Investigational Drug as Substrate or Inhibitor of P-gp Transporter
In testing an investigational drug for the possibility that its disposition may be inhibited or induced (i.e., as a substrate of P-gp), selection of the interacting drugs may be based on whether the investigational drug is also a CYP3A substrate. If it is also a substrate of CYP3A, it may be appropriate to use a dual inhibitor of both CYP3A and P-gp, such as ritonavir. If the investigational drug is not a substrate of CYP3A, it may be appropriate to use a strong inhibitor of P-gp, such as cyclosporine or verapamil.
In testing an investigational drug for the possibility that it may be an inhibitor of P-gp, selection of digoxin or other known substrates of P-gp may be appropriate.

D. Route of Administration
The route of administration chosen for a metabolic drug-drug interaction study is important. For an investigational agent used as either an interacting drug or substrate, the route of administration should generally be the one planned for in product labeling. When multiple routes are being developed, the necessity for doing metabolic drug-drug interaction studies by all routes should be based on the expected mechanism of interaction and the similarity of corresponding concentration-time profiles for parent and metabolites. If only oral dosage forms will be marketed, studies with an intravenous formulation would not usually be needed, although information from oral and intravenous dosings may be useful in discerning the relative contributions of alterations in absorption and/or presystemic clearance to the overall effect observed for a drug interaction. Sometimes certain routes of administration can reduce the utility of information from a study. For example, an intravenous study may not reveal an interaction for substrate drugs where intestinal CYP3A activity markedly alters bioavailability. For an approved agent used either as a substrate or interacting drug, the route of administration will depend on available marketed formulations, which in most instances will be oral.
E. Dose Selection
For both a substrate (investigational drug or approved drug) and interacting drug (investigational drug or approved drug), testing should maximize the possibility of finding an interaction. For this reason, the maximum planned or approved dose and shortest dosing interval of the interacting drug (as inhibitors or inducers) should be used. For example, when using ketoconazole as an inhibitor of CYP3A, dosing at 400 mg QD for multiple days would be preferable to dosing at lower doses. When using rifampin as an inducer, dosing at 600 mg QD for multiple days would be preferable to dosing at lower doses. Doses smaller than those to be used clinically may be needed for substrates on safety grounds and may be more sensitive to the effect of the interacting drug.
F. Endpoints
1. Pharmacokinetic Endpoints
The following measures and parameters are recommended for assessment of the substrate: (1) exposure measures such as AUC, Cmax, time to Cmax (Tmax), and others as appropriate; and (2) pharmacokinetic parameters such as clearance, volumes of distribution, and half-lives. In some cases, these measures may be of interest for the inhibitor or inducer as well, notably where the study is assessing possible interactions between both study drugs. Additional measures may help in steady state studies (e.g., trough concentration (Cmin)) to demonstrate that dosing strategies were adequate to achieve near steady state before and during the interaction. In certain instances, an understanding of the relationship between dose, blood concentrations, and response may lead to a special interest in certain pharmacokinetic measures and/or parameters. For example, if a clinical outcome is most closely related to peak concentration (e.g., tachycardia with sympathomimetics), Cmax or another early exposure measure might be most appropriate. Conversely, if the clinical outcome is related more to extent of absorption, AUC would be preferred. The frequency of sampling should be adequate to allow accurate determination of the relevant measures and/or parameters for the parent and metabolites. For the substrate, whether the investigational drug or approved drug, determination of the pharmacokinetics of important active metabolites is important. This concept paper focuses on metabolic drug-drug interactions, however, protein binding determinations are considered necessary to distinguish between induction or stimulation of metabolism and displacement from protein-binding site. The latter is not considered to be a source of clinically important drug interactions because unbound drug concentrations are unaffected.
2. Pharmacodynamic Endpoints
Pharmacokinetic measures are usually sufficient for metabolic drug-drug interaction studies, although pharmacodynamic measures can sometimes provide additional useful information. Pharmacodynamic measures may be needed when a pharmacokinetic/pharmacodynamic relationship for the substrate endpoints of interest is not established or when pharmacodynamic changes do not result solely from pharmacokinetic interactions (e.g, additive cardiovascular effect of quinidine and tricyclic antidepressants). When an approved drug is studied as a substrate, the pharmacodynamic impact of a given change in blood level (Cmax, AUC) caused by an investigational interaction should be known from other interaction studies about the approved drug, with the possible exception of older drugs.
G. Sample Size and Statistical Considerations
For both investigational drugs and approved drugs, when used as substrates and/or interacting drugs in drug-drug interaction studies, the desired goal of the analysis is to determine the clinical significance of any increase or decrease in exposure to the substrate in the presence of the interacting drug. Assuming unchanged PK/PD relationships, changes may be evaluated by comparing pharmacokinetic measures of systemic exposure that are most relevant to an understanding of the relationship between dose (exposure) and therapeutic outcome.
Results of drug-drug interaction studies should be reported as 90% confidence intervals about the geometric mean ratio of the observed pharmacokinetic measures with (S+I) and without the interacting drug (S).³ Confidence intervals provide an estimate of the distribution of the observed systemic exposure measure ratio of S+I versus S alone and convey a probability of the magnitude of the interaction. In contrast, tests of significance are not appropriate because small, consistent systemic exposure differences can be statistically significant (p < 0.05) but not clinically relevant.
When a drug-drug interaction is clearly present (e.g., comparisons indicate twofold or greater increments in systemic exposure measures for S+I) the sponsor should be able to provide specific recommendations regarding the clinical significance of the interaction based on what is known about the dose-response and/or PK/PD relationship for either the investigational agent or the approved drugs used in the study. This information should form the basis for reporting study results and for making recommendations in the package insert with respect to either the dose, dosing regimen adjustments, precautions, warnings, or contraindications of either the investigational drug or the approved drug. FDA recognizes that dose-response and/or PK/PD information may sometimes be incomplete or unavailable, especially for an approved drug used as S.
Second, the sponsor may wish to make specific claims in the package insert that no drug-drug interaction is expected. In these instances, the sponsor should be able to recommend specific no effect boundaries, or clinical equivalence intervals, for a drug-drug interaction. No effect boundaries define the interval within which a change in a systemic exposure measure is considered not clinically meaningful. There are two approaches to define no effect boundaries.
Approach 1: No effect boundaries can be based on population (group) average dose and/or concentration-response relationships, PK/PD models, and other available information for the substrate drug. If the 90% confidence interval for the systemic exposure measurement in the drug-drug interaction study falls completely within the no effect boundaries, the sponsor may conclude that no clinically significant drug-drug interaction was present.
Approach 2: In the absence of no effect boundaries defined in (1) above, a sponsor may use a default no effect boundary of 80-125% for both the investigational drug and the approved drugs used in the study. When the 90% confidence intervals for systemic exposure ratios fall entirely within the equivalence range of 80-125%, standard Agency practice is to conclude that no clinically significant differences are present.
The selection of the number of subjects for a given drug-drug interaction study will depend on how small an effect is clinically important to detect, or rule out, the inter- and intrasubject variability in pharmacokinetic measurements, and possibly other factors or sources of variability not well recognized. In addition, the number of subjects will depend on how the results of the drug-drug interaction study will be used, as described above.
This concept paper should not be interpreted by sponsors as generally recommending the inclusion of some number of subjects in a drug-drug interaction study such that the 90% confidence interval for the ratio of pharmacokinetic measurements falls entirely within the no effect boundaries of 80-125%. This approach, however, could be deemed appropriate by a sponsor, after considering the expected outcome of a drug-drug interaction study, the anticipated magnitude of variability in pharmacokinetic measurements, and the desired label claim that no clinically significant drug-drug interaction was present.

V. LABELING IMPLICATIONS

All relevant information on the metabolic pathways and metabolites and pharmacokinetic interaction should be included in the PHARMACOKINETICS subsection of the CLINICAL PHARMACOLOGY section of the labeling. The clinical consequences of metabolism and interactions should be placed in DRUG INTERACTIONS, WARNINGS AND PRECAUTIONS, BOXED WARNINGS, CONTRAINDICATIONS, or DOSAGE AND ADMINISTRATION sections, as appropriate. Such information related to clinical consequences should not be included in detail in more than one consequences related section, but rather referenced from one section to other sections as needed. When the metabolic pathway or interaction data resulted in recommendations for dosage adjustments, contraindications, warnings (e.g., co-administration should be avoided), that were included in the BOXED WARNINGS, CONTRAINDICATIONS, WARNINGS AND PRECAUTIONS, or DOSAGE AND ADMINISTRATION sections, these recommendations should also be included in the corresponding “HIGHLIGHTS” section of the labeling with appropriate referencing of other labeling sections. Refer to the guidance for industry “Labeling for Human Prescription Drug and Biological Products – Implementing the New Content and Format Requirements” and “Clinical Pharmacology and Drug Interaction Labeling” for more information on presenting drug interaction information in labeling.

The following general principles affect labeling for specific metabolism or drug interaction data.

• 
  In certain cases, information based on clinical studies not using the labeled drug under investigation can be described with an explanation that similar results may be expected for the labeled drug. For example, if a drug has been determined to be a strong inhibitor of CYP3A, it does not need to be tested with all CYP3A substrates to warn about an interaction with “sensitive CYP3A substrates” and “CYP3A substrates with narrow therapeutic range”. Table 2 lists examples of “sensitive CYP3A substrates” and “CYP3A substrates with narrow therapeutic range”.
  

Table 2. Examples⁽¹⁾ of sensitive CYP3A substrates or CYP3A substrates with narrow therapeutic range

Sensitive CYP3A substrates^(2)	CYP3A Substrates with Narrow therapeutic range^(3)
budesonide, buspirone, eletriptan, felodipine, imatinab, lovastatin, midazolam, saquinavir, sildenafil, simvastatin, triazolam	Alfentanil, astemizole(a), cisapride(a), cyclosporine, diergotamine, ergotamine, fentanyl, irinotecan, pimozide, quinidine, sirolimus, tacrolimus, terfenadine(a)

⁽¹⁾ note that this is not an extensive list; for an updated list, see URL???

⁽²⁾ “sensitive CYP3A substrates” refer to drugs whose plasma AUC values are increased 5-fold or more when co-administered with CYP3A inhibitors

⁽³⁾ “CYP3A substrates with narrow therapeutic range” refer to drugs whose exposure-response data are such that increases in their exposure levels by the concomitant use of CYP3A inhibitors may lead to serious safety concerns (e.g., Torsades de Pointes); (a) not available in US

• 
  If a drug has been determined to be a sensitive CYP3A substrate or a CYP3A substrate with a narrow therapeutic range, it does not need to be tested with all strong or moderate inhibitors of CYP3A to warn about an interaction with “strong” or “moderate” CYP3A inhibitors. Table 3 lists examples of “strong CYP3A inhibitors” and “moderate CYP3A inhibitors”. Similarly, if a drug has been determined to be a sensitive CYP3A substrate or a CYP3A substrate with a narrow therapeutic range, it does not need to be tested with all CYP3A inducers to warn about an interaction with CYP3A inducers. Examples of CYP3A inducers include rifampin, rifabutin, rifapentin, dexamethasone, phenytoin, carbamazepine, phenobarbital and St. John's Wort.