FDA and EMA Issue Guidelines for PBPK Modeling
An AAPS webinar discusses regulations in the U.S. and Europe.
By Mark Crawford
Ping Zhao, Ph.D., and senior program officer with the Bill and Melinda Gates Foundation, and Anna Nordmark, Ph.D. and pharmacokinetic assessor with the Medical Products Agency, the organization responsible for the regulation and surveillance of drugs and other medicinal products in Sweden, copresented an AAPS webinar entitled First-in-Class Regulatory PBPK Modeling Guidelines from Both Sides of the Pond.
In recent years, regulators have observed a significant increase in the various ways drug companies are using physiologically based pharmacokinetic (PBPK) models in their pharmaceutical submissions. Regulators are concerned with inconsistencies in how these models are validated or qualified. In fact, some report that qualification of the intended use is barely covered in these submissions; other concerns are that PBPK simulations do not contain enough detail, especially regarding lack of sensitivity and uncertainty analysis.
In response, U.S. and European regulators have developed draft guidelines to clarify their expectations for how PBPK should be used to streamline the approval process. In the first part of the webinar, Zhao reviews the 2016 Food and Drug Administration (FDA) PBPK draft guidance; in the second part, Nordmark reviews the European Medicines Agency’s (EMA) draft guidance on the qualification and reporting of PBPK modeling and simulation. Both presenters agree that confidence of use is a key issue when conducting PBPK modeling and requires increased scrutiny and evaluation by drug manufacturers before they make their submissions.
FDA Draft of PBPK Guidance
PBPK analyses are routinely submitted to FDA, usually for drug-drug interactions—for example, drugs as enzyme substrates or perpetrators, or transporter-based assessments, such as in vitro-in vivo extrapolation. PBPK can also be used for specific population studies, like pediatrics or organ impairments (hepatic and renal). Other possible applications include pregnancy, ethnicity, geriatrics, obesity, disease states, formulation changes, pH effects, and tissue concentrations.
However, PBPK does not always provide reliable or conclusive results. For example, when studying a drug as an enzyme perpetrator, PBPK can confirm the lack of enzyme inhibition, but additional evidence is required to confirm predictive performance for positive interactions. Another example is in vitro-in vivo extrapolation—results can be complicated by transporter-enzyme interplay, which interferes with predictive performance.
Thus, FDA wants to see more predictive performance from the PBPK data. They also want the rationale for the PBPK analyses, overview of the modeling strategy, demonstration of relevance and appropriateness for subsequent analyses, and components of model verification, modification, and application. Determining adequacy is essential: can PBPK models adequately describe the baseline PK of the drug?
To clarify this issue, FDA has developed guidelines for content and format of PBPK submissions in its 2016 FDA PBPK draft guidance. The goal is to facilitate efficient, timely, and consistent FDA review of applications using PBPK. “The guidance does not address methodological considerations and best practices for the conduct of PBPK modeling and simulation, or the appropriateness of PBPK analyses for a particular drug or a drug product,” says Zhao.
The draft received 10 comments from individual pharmaceutical companies, consortiums, regulatory agencies, and individuals within the two-month public comment period (closed on January 31, 2017). Questions were raised about confidence of use, fit-for-purpose pathways, variability assessments, simulation designs, data/model and software requirements, lack of nonclinical elements, and flexibility. In particular, companies want FDA to provide more clarity regarding confidence of use. For example, what acceptance criteria are required to show similarity between the simulated and observed data for model verification? Also, what level of confidence does FDA require for confirmation or rejection of the PBPK model? Even though confidence of use varies according to the purpose of the study, PBPK data should always establish predictive drug performance, through an appropriate model that is the best fit for the study.
EMA Draft of PBPK Guidance
Regulators in Europe have also witnessed an increase in PBPK submissions to EMA—as part of an initial package, in response to requests by regulators, or as postauthorization measures. In about 75 percent of the cases where a PBPK model is suggested or submitted, at least one of the purposes relates to drug-drug interaction (as victim or as perpetrator), especially for CYP3A4-mediated interactions. Other reasons for PBPK are to better understand PK and the role of enzymes/transporters, dose recommendations, food effects, effects of polymorphisms/ethnic differences, and PK in special population groups (renal/hepatic impairment).
Ideally, PBPK can be an effective tool for modeling the different pathways of metabolism, blood flows, and excretion routes, as well as a method for estimating dose/duration and optimizing study design. However, as with FDA, the EMA is finding that qualification of intended use is poorly developed, and PBPK simulations are deficient in details. Qualification is related to the PBPK platform—is there enough scientific support for a certain use for that particular platform? Qualification is especially important for high regulatory impact decisions: for example, using a PBPK model instead of clinical data.
Overall, the EMA is concerned about inconsistency in PBPK reports, especially the lack of quantitative assessment of precision of prediction. To improve this situation, the EMA released its guideline, Qualification and Reporting of Physiologically Based Pharmacokinetic (PBPK) Modeling and Simulation, in July 2016. A workshop was held in November 2016, and a public commentary period ended in January 2017.
Among the comments the EMA received, one recommended that the qualification dataset should be prespecified. Selection criteria for the drugs and the in-vitro and in-vivo parameters for these drugs should be described. Ideally, the dataset should cover a range of PK characteristics, such as permeability, extraction ratio, and protein binding, which could affect the outcome. Say, for example, the intended purpose is to use PBPK to predict the PK of drug X in young children, because the clinical PK data is very limited in this age group. Because this has high regulatory impact, the platform should be qualified for this intended use by using external/literature PK data from children at the same age range. The dataset should be able to predict the PK of compounds metabolized via the same enzymes and have similar absorption characteristics as drug X, with adequate performance.
Other comments received on the draft include a desire for more examples of diverse applications, including absorption PBPK modeling and simulation, a clearer separation of drug-independent and drug-dependent components, and more details on requirements for medium- and low-impact applications.
EMA Hopes to Finalize its PBPK Guideline by Mid-2018
“We want qualification of use so we can harmonize the assessment of PBPK applications across the European countries,” says Nordmark. “Presently not all aspects included in PBPK platforms are entirely scientifically justified, or suitable, for high regulatory impact decisions. They should be qualified first. From our view, this is not a restriction or hindrance, but will improve the acceptability of the submitted models by EU [European Union] regulators.”
To listen to this webinar, you may access the archive. Learn more about upcoming webinars or submit a proposal!
Mark Crawford, a freelance writer based in Madison, Wis., specializes in science and technology.