Short answer: Physiologically based pharmacokinetic (PBPK) modeling uses models and simulations that combine physiology, population and drug characteristics to describe a drug’s pharmacokinetic (PK) behaviour mechanistically. (RAPS, quoting FDA) A PBPK model represents the body as organs connected by blood flow, so it can predict exposure under conditions that have not been studied clinically. Regulators decide case by case whether to accept PBPK results in place of clinical PK data, based on the intended use and the quality, relevance and reliability of the analysis. (FDA, PBPK Analyses: Format and Content)
That last point drives planning. A PBPK model is not accepted or rejected in general. It is judged against the specific question it is meant to answer. This guide explains how the models are built, how they differ from other pharmacometric models, where they help generic development, and what determines how much weight they carry.
How a PBPK model works
Classical PK models use abstract compartments with fitted rate constants. A PBPK model instead uses organ and tissue compartments linked by blood flows and parameterised with physiological values. (Jones and Rowland-Yeo, CPT Pharmacometrics Syst Pharmacol 2013) Its inputs fall into three groups:
| Input group | Examples |
|---|---|
| System (physiology) | Organ volumes and blood flows, gastrointestinal transit and pH, enzyme and transporter abundance, and how these vary by age, disease or pregnancy |
| Drug | pKa, logP, pH-solubility profile, permeability, protein binding, in vitro metabolism and transport data, tissue-to-plasma partition coefficients |
| Formulation and trial design | Dissolution profiles, particle size, dose, regimen, fed or fasted state, population characteristics |
Tissues are usually modelled as perfusion-limited, where blood flow controls distribution (typical for small lipophilic drugs), or permeability-limited, where membrane transport controls it (typical for larger polar drugs). Virtual populations are generated by varying system parameters, which lets the model predict variability as well as average exposure. (Jones and Rowland-Yeo 2013)
When formulation and dissolution data drive the absorption part of the model, the approach is often called physiologically based biopharmaceutics modeling (PBBM). FDA’s draft biopharmaceutics guidance notes that several terms are used and covers them under “PBPK analyses”. (FDA draft guidance, 2020)
Building, verifying and validating a PBPK model
- Collect data: physicochemical properties, in vitro dissolution and available clinical PK data.
- Develop the model: integrate drug, formulation and physiological information.
- Verify: compare simulated PK with observed clinical data that were not used to build the model.
- Refine: adjust parameters where predictions and observations disagree, and document why. This “learn, confirm and refine” loop is the recommended strategy. (Jones and Rowland-Yeo 2013)
- Apply: simulate the scenario of interest, such as a new formulation in a virtual population.
- Test sensitivity: show which parameters drive the conclusion and whether plausible changes would alter it.
EMA’s reporting guideline also expects documentation that qualifies the PBPK platform for its intended use. (EMA, EMA/CHMP/458101/2016)
PBPK vs popPK, compartmental models and QSP
| Approach | Structure | Built from | Typical question |
|---|---|---|---|
| PBPK | Organ compartments connected by blood flow | Mainly physiological and in vitro data (“bottom-up”), verified against clinical data | What exposure should we expect in a situation not yet studied (new formulation, interacting drug, special population)? |
| Compartmental PK (e.g., 1- or 2-compartment) | Abstract central and peripheral compartments with rate constants | Fitted to observed concentration data | What parameters describe the observed profile? |
| Population PK (popPK) | Compartmental model with statistical terms for between-subject variability and covariates | Fitted to clinical data from many subjects (“top-down”) | What explains variability in observed exposure, and what dose fits a subgroup? |
| Quantitative systems pharmacology (QSP) | Mechanistic model of disease biology and drug effect | Biological pathway data and clinical response data | How does target engagement translate into clinical response? |
In practice these approaches are often combined, and many programmes use PBPK and popPK side by side.
Where PBPK modeling is used
FDA’s PBPK programme has published regulatory experience across drug-drug interactions (DDI), paediatrics, organ impairment and pregnancy. (FDA, PBPK Program) In generic development, the most relevant uses are:
- Formulation development and risk assessment: predicting how changes in dissolution or particle size could affect Cmax and AUC before a clinical study.
- BE study design: exploring likely outcome and variability, and the effect of fed or fasted conditions.
- Manufacturing and formulation changes: FDA’s draft biopharmaceutics guidance covers PBPK analyses for oral product development, manufacturing changes and controls in IND, NDA and ANDA submissions and supplements. It is a draft and not for implementation. (FDA, draft 2020)
- Virtual bioequivalence (VBE): simulating test and reference products in a virtual population and evaluating Cmax and AUC statistics.
- Troubleshooting a failed BE study: testing mechanistic explanations for why a formulation failed.
A documented regulatory example
FDA reports that PBPK modeling supported approval of a generic diclofenac sodium topical gel, 1% (ANDA 211253), and describes it as the first ANDA approval for which a PBPK model supported the BE assessment. The dermal PBPK model was used in place of the comparative clinical endpoint BE study in the product-specific guidance. Approval still rested on a totality of evidence: formulation sameness with the reference product, an in vivo PK BE study, and the virtual BE assessment. (FDA; Tsakalozou et al., 2021) This is one case. It does not show that PBPK can generally replace clinical studies.
What regulators look at before relying on PBPK evidence
| Authority | Document | What it sets |
|---|---|---|
| FDA | PBPK Analyses: Format and Content, final, 2018 | A six-section report: executive summary, introduction, materials and methods, results, discussion, appendices. It does not address modelling best practice; acceptance is case by case. (FDA) |
| FDA | PBPK for biopharmaceutics applications, draft, 2020 | Development, evaluation and use of PBPK for oral product development and changes; not for implementation. (FDA) |
| EMA | Guideline on the reporting of PBPK modelling and simulation (EMA/CHMP/458101/2016), effective 1 July 2019 | Report content and platform qualification for the intended use, for MAAs, paediatric investigation plans and clinical trial applications. (EMA) |
| ICH | M15 General Principles for Model-Informed Drug Development, Step 4, 29 January 2026 | Planning, model evaluation and documentation of model-informed evidence, with credibility linked to model influence and model risk. (ICH M15; RAPS) |
A practical reading of these principles, offered as editorial guidance rather than a rule:
| Factor | Supports reliance on PBPK | Weakens reliance on PBPK |
|---|---|---|
| Context of use | The model supports a decision alongside other evidence | The model is the only evidence for a high-impact decision |
| Verification | Verified against clinical data close to the scenario simulated | Little relevant clinical data for verification |
| Mechanism | Absorption and disposition well understood | Unexplained food effects or complex transporter effects |
| Extrapolation | Interpolation within the verified range | Extrapolation to new populations, doses or dosage forms |
| Consequence of error | Low clinical risk | Narrow therapeutic index or serious safety consequences |
PBPK modeling software
Commonly used commercial platforms include GastroPlus (Simulations Plus) and the Simcyp Simulator (Certara), and PK-Sim is available through the Open Systems Pharmacology suite. Regulators do not mandate a platform. FDA asks for the software and version to be documented in the report. (RAPS on the FDA guidance)
Limitations to plan for
- A model is only as reliable as its inputs. Weak dissolution or permeability data carry through to the prediction.
- Within-subject variability, which drives BE study power, is difficult to simulate credibly.
- Assumptions, parameter sources and software versions must be documented fully for review.
- Where the model carries significant weight in a decision, early discussion with the regulator is advisable.
How PharmaPrimes supports PBPK and virtual BE
PharmaPrimes provides PBPK/PBBM modeling and virtual bioequivalence simulations using GastroPlus. Applications include formulation optimisation, development risk assessment, BE study design, evaluation of formulation and manufacturing changes, scientific regulatory justification and troubleshooting of BE failures. Deliverables include model documentation, virtual population simulations, predicted PK profiles, BE statistical analysis and a VBE assessment report. See our PBPK/PBBM and virtual BE service, which can be combined with clinical BE/BA studies.
Frequently asked questions
What is PBPK modeling?
A mechanistic approach that simulates drug absorption, distribution, metabolism and excretion in organ compartments connected by blood flow, using physiological, drug and formulation data.
What does PBPK stand for?
Physiologically based pharmacokinetic.
What is the difference between PBPK and popPK?
PBPK is built mainly from physiological and in vitro data to predict untested situations. Population PK is fitted to observed clinical concentrations to describe and explain variability.
What is the difference between PBPK and QSP?
PBPK predicts drug concentrations in the body. QSP models disease biology and drug effect, linking exposure to response.
What are PBPK models used for in regulatory submissions?
FDA’s programme has published experience with drug interactions, paediatrics, organ impairment and pregnancy. Biopharmaceutics uses such as formulation changes are covered by a draft FDA guidance. Acceptance depends on the context of use.
Can PBPK modeling replace a bioequivalence study?
Not as a general rule. In the diclofenac gel case, PBPK replaced a clinical endpoint BE study, but approval also relied on formulation sameness and an in vivo PK BE study.
How is a PBPK model validated?
By comparing simulations with clinical data not used to build the model, testing parameter sensitivity, and documenting assumptions and platform qualification.
Explore a modeling strategy
Have a formulation question, a planned BE study or a failed one? Explore PBPK and virtual BE support with our modeling team.
References
- FDA. PBPK Analyses: Format and Content, 2018. https://www.fda.gov/regulatory-information/search-fda-guidance-documents/physiologically-based-pharmacokinetic-analyses-format-and-content-guidance-industry
- FDA. The Use of PBPK Analyses: Biopharmaceutics Applications, draft, 2020. https://www.fda.gov/regulatory-information/search-fda-guidance-documents/use-physiologically-based-pharmacokinetic-analyses-biopharmaceutics-applications-oral-drug-product
- EMA. Guideline on the reporting of PBPK modelling and simulation, EMA/CHMP/458101/2016. https://www.ema.europa.eu/en/reporting-physiologically-based-pharmacokinetic-pbpk-modelling-simulation-scientific-guideline
- ICH. M15, Step 4, 29 Jan 2026. https://www.pmda.go.jp/files/000279201.pdf
- FDA. PBPK Program. https://www.fda.gov/about-fda/cder-offices-and-divisions/program-physiologically-based-pharmacokinetic-and-pharmacodynamic-modeling-pbpk-program
- FDA. How Modeling Was Used to Support the FDA Approval of a Topical Generic Drug Product. https://www.fda.gov/drugs/regulatory-science-action/how-modeling-was-used-support-fda-approval-topical-generic-drug-product
- Jones HM, Rowland-Yeo K. CPT Pharmacometrics Syst Pharmacol. 2013;2:e63. https://pmc.ncbi.nlm.nih.gov/articles/PMC3828005/
- Tsakalozou E et al. CPT Pharmacometrics Syst Pharmacol. 2021;10(5):399-411. https://doi.org/10.1002/psp4.12600