Introduction to Physiologically Based Pharmacokinetic Modeling

1. What Is Physiologically Based Pharmacokinetic Modeling?

Physiologically based pharmacokinetic (PBPK) modeling is a mechanistic approach to pharmacokinetics that represents the body using physiological structures and processes and combines them with drug-specific physicochemical and biochemical properties.

Unlike a conventional compartmental model, where compartments are primarily mathematical constructs, a PBPK model typically represents organs or tissues explicitly and describes drug movement between them using physiological quantities such as organ blood flows, tissue volumes, and tissue partitioning.

The FDA describes PBPK models as frameworks that integrate drug and system information, while the EMA describes PBPK as modeling drug concentration over time in tissues and blood using the interplay between physiological, physicochemical, and biochemical determinants. FDA PBPK Program and the EMA PBPK reporting guideline provide regulatory context for these concepts.

2. Why Use a Physiologically Based Model?

A conventional PK model can summarize observed concentration-time data very effectively. PBPK modeling addresses a different need: it attempts to represent the mechanisms that generate those observations.

This distinction becomes especially useful when the objective is to make predictions under conditions that differ from the original clinical study.

PBPK does not eliminate the need for clinical data. Rather, its value depends on whether the underlying drug and system information are sufficiently characterized and whether the model has demonstrated appropriate predictive performance for its intended use.

3. The Two Fundamental Components of a PBPK Model

A useful conceptual framework is to divide a PBPK model into two interacting components:

1. The drug model: describes the properties and processes specific to the compound.
2. The system model: describes the physiology of the organism or population in which the drug is administered.

Drug-specific information

• Molecular weight
• Lipophilicity
• Ionization and pKa
• Aqueous solubility
• Plasma and tissue protein binding
• Membrane permeability
• Transporter substrate or inhibitor properties
• Metabolic enzyme substrate, inhibition, or induction properties
• Renal clearance mechanisms
• Formulation and dissolution characteristics

System information

• Organ and tissue volumes
• Organ blood flows
• Blood and plasma composition
• Enzyme and transporter abundance
• Physiological pH values
• Body weight and body composition
• Age-dependent physiological changes
• Population-specific physiological characteristics
• Disease-related changes when adequately characterized

4. How Does a PBPK Model Represent the Body?

Most PBPK models divide the body into compartments corresponding to organs or tissues. Each organ can have a specified volume, blood flow, and drug distribution behavior.

A simplified organ model can be represented as a tissue compartment receiving drug from arterial blood and returning drug to venous blood.

The mathematical structure becomes a system of mass-balance equations. A generic tissue equation can be written conceptually as:

where \(A_t\) is the amount of drug in the tissue, \(Q_t\) is tissue blood flow, \(C_a\) is arterial concentration, \(C_{v,t}\) is venous concentration leaving the tissue, and \(R_t\) represents additional drug input or removal processes such as metabolism when applicable.

The exact equation depends on the tissue model and assumptions about perfusion, permeability, binding, metabolism, and distribution.

5. Physiology Is Part of the Model

One of the defining features of PBPK modeling is that physiological quantities are explicitly represented rather than hidden entirely inside fitted PK parameters.

This physiological structure is what makes PBPK particularly useful for certain extrapolation problems. If the target population differs physiologically from the population used to establish the model, system parameters can potentially be modified while preserving drug-specific properties.

6. Drug-Specific Properties in PBPK

The second major component is the drug. PBPK models use physicochemical and biochemical properties to determine how the compound behaves within the physiological system.

Physicochemical properties

Properties such as lipophilicity, ionization, solubility, molecular size, and permeability influence absorption and distribution. For example, ionization can affect membrane permeability and pH-dependent solubility.

Binding

Protein binding affects the fraction of drug available in plasma and can influence distribution and clearance relationships. Depending on the model, binding may be represented using experimentally determined parameters or mechanistic submodels.

Metabolism

Drug metabolism can be represented using enzyme-specific pathways. Intrinsic metabolic clearance may be scaled from in vitro systems to the whole organism using information about enzyme abundance, microsomal or cellular protein content, organ size, and other scaling factors.

Transport

Transporters may contribute to uptake, efflux, absorption, biliary disposition, renal processes, and tissue distribution. When transporter information is sufficiently characterized, these mechanisms can be explicitly represented.

7. From In Vitro Data to In Vivo Prediction

A major application of PBPK modeling is the translation of experimental observations obtained in vitro into predictions of human PK. This is often referred to as in vitro-to-in vivo extrapolation (IVIVE).

For example, an in vitro experiment may provide an estimate of intrinsic metabolic clearance. A PBPK model can combine that information with physiological properties of the relevant organ to predict systemic clearance.

The exact scaling approach depends on the experimental system and biological pathway. Microsomal systems, hepatocytes, recombinant enzymes, transporter assays, and other experimental platforms require different interpretations.

PBPK provides a framework in which these inputs can be connected explicitly rather than collapsing every mechanism into a single empirical clearance parameter.

8. Representing Absorption, Distribution, Metabolism, and Excretion

A PBPK model attempts to represent the major processes governing drug disposition:

The important point is that PBPK does not require every possible mechanism to be modeled. The model should include mechanisms that are relevant to the scientific question and sufficiently supported by evidence.

9. A Simple Example: Hepatic Clearance

Many PBPK models represent hepatic metabolism using a mechanistic liver model. A simplified well-stirred representation can be written as:

where \(Q_H\) is hepatic blood flow, \(f_u\) is the unbound fraction in blood or plasma as appropriate to the model, and \(CL_{\mathrm{int}}\) is intrinsic hepatic clearance.

This equation illustrates the interaction between physiology and drug-specific properties. Hepatic blood flow is a system property, while intrinsic metabolic clearance is primarily drug-specific. Protein binding links the two through the fraction available for hepatic extraction.

More sophisticated PBPK models can represent individual metabolic pathways, enzymes, transporters, and zonation or other mechanistic features rather than treating intrinsic clearance as a single aggregate parameter.

10. How Does PBPK Represent Drug Distribution?

Distribution is one of the areas where PBPK can provide substantially more mechanistic detail than a simple compartment model.

A tissue concentration may depend on blood flow, tissue composition, membrane permeability, binding, and the physicochemical characteristics of the drug. In a simplified perfusion-limited tissue model, equilibrium between tissue and plasma may be represented using a tissue-to-plasma partition coefficient:

where \(C_t\) is tissue concentration, \(C_p\) is plasma concentration, and \(K_{p,t}\) is the tissue-to-plasma partition coefficient.

More mechanistic tissue models can explicitly represent membrane permeability and intracellular processes when equilibrium assumptions are not sufficient.

11. Modeling Oral Absorption

For orally administered drugs, the PBPK model must describe how drug moves from the gastrointestinal tract into the systemic circulation.

A simplified oral absorption framework may include:

• Gastric emptying
• Intestinal transit
• Dissolution
• Solubility
• Intestinal permeability
• pH-dependent ionization
• Enterocyte metabolism
• Intestinal transport
• First-pass hepatic extraction

The resulting systemic exposure depends on both the rate and extent of absorption.

where \(F_a\) represents the fraction absorbed from the gastrointestinal tract, \(F_g\) represents the fraction escaping intestinal metabolism, and \(F_h\) represents the fraction escaping hepatic first-pass extraction in this commonly used conceptual decomposition.

The exact definitions and implementation depend on the PBPK platform and model structure.

12. From Model Structure to Simulation

A PBPK model is generally represented as a system of differential equations describing drug amounts or concentrations throughout the body.

Conceptually:

where \(\mathbf{A}(t)\) represents drug amounts across compartments, \(\boldsymbol{\theta}_{\mathrm{drug}}\) represents drug-specific parameters, \(\boldsymbol{\theta}_{\mathrm{system}}\) represents physiological parameters, and \(u(t)\) represents drug administration or other inputs.

Numerical differential-equation solvers then calculate concentrations over time.

This is an important conceptual difference from simply fitting a curve to concentration data. The PBPK simulation is generated by a network of mechanistic equations whose parameters have physiological or drug-specific interpretations.

13. Worked Example: A Simplified PBPK Tissue Model

Consider a hypothetical tissue with a volume of 2 L. Assume the tissue receives a blood flow of 1 L/h and that the tissue-to-plasma partition coefficient is 3. Suppose the plasma concentration at a particular time is 4 mg/L and the tissue is assumed to be at instantaneous distribution equilibrium.

Step 1: Calculate the tissue concentration

Step 2: Calculate the amount of drug in the tissue

Step 3: Interpret the result

Under these simplified assumptions, the tissue contains approximately 24 mg of drug when plasma concentration is 4 mg/L.

The important point is not the numerical result itself. The example illustrates how PBPK links a plasma concentration to tissue concentration through a tissue-specific distribution parameter and then links concentration to drug amount through tissue volume.

14. Representing Different Populations

One of the most important applications of PBPK is simulation in populations that differ from the population used to generate the original clinical PK data.

Potential system changes include:

• Age and maturation
• Body weight and body composition
• Organ size
• Organ blood flow
• Enzyme abundance
• Transporter abundance
• Renal function
• Disease-associated physiological changes
• Pregnancy-associated physiological changes

For example, a pediatric PBPK model can incorporate age-dependent physiological development rather than treating a child simply as a smaller adult.

Similarly, renal or hepatic impairment models can represent changes in relevant physiological or functional parameters when those changes are sufficiently characterized.

15. PBPK and Drug-Drug Interactions

Drug-drug interaction (DDI) modeling is one of the important applications of PBPK because many interactions can be described mechanistically through changes in enzymes and transporters.

For example, a perpetrator drug may inhibit an enzyme responsible for clearance of the victim drug. A PBPK model can represent the perpetrator concentration over time and use it to modify the activity of the relevant pathway.

A more detailed model may allow \(V_{\max}\), \(K_m\), enzyme abundance, inhibition constants, induction parameters, or transporter parameters to change according to the interacting drug and its exposure.

This mechanistic structure can be useful when predicting DDIs involving doses, concentrations, or populations that were not directly studied clinically.

However, DDI predictions remain highly dependent on the quality of the perpetrator and victim drug models and on the characterization of the relevant enzymes and transporters.

16. How Is a PBPK Model Developed?

1. Define the intended use. Begin with the scientific or regulatory question the model is intended to answer.
2. Collect drug-specific information. Assemble physicochemical, binding, permeability, metabolism, transport, formulation, and other relevant information.
3. Define the system. Select the physiological population and appropriate organ, flow, enzyme, transporter, and other system information.
4. Construct the structural model. Represent the relevant absorption, distribution, metabolism, and excretion processes.
5. Parameterize the model. Use experimental and literature information to assign values or distributions to model parameters.
6. Simulate observed studies. Compare model predictions with available clinical and experimental PK data.
7. Evaluate predictive performance. Determine whether the model adequately predicts relevant observations.
8. Refine transparently. Update assumptions when justified by evidence and document the changes.
9. Apply the qualified model. Use the model for the intended prediction while clearly documenting assumptions and uncertainty.

The FDA's PBPK guidance recommends a structured presentation of regulatory PBPK analyses, including an executive summary, introduction, materials and methods, results, discussion, and appendices. The EMA guideline similarly emphasizes documentation of the model, platform, and predictive performance. FDA PBPK guidance and EMA PBPK guideline.

17. Verification, Qualification, and Predictive Performance

A PBPK model should not be considered credible simply because it produces plausible concentration-time curves. Its performance needs to be evaluated for the intended application.

Verification

Verification asks whether the model has been implemented correctly. Examples include checking equations, units, numerical behavior, parameter transformations, and software implementation.

Qualification

Qualification concerns whether the PBPK platform and associated system knowledge are adequate for a particular intended use. The EMA specifically describes qualification of PBPK platforms for intended uses in its reporting guideline.

Predictive performance

Predictive performance asks how well the model predicts observations that are relevant to the intended application. This can include comparisons between predicted and observed PK across studies, doses, populations, or interaction scenarios.

18. Sensitivity Analysis and Uncertainty

PBPK models can contain many parameters, and not every parameter contributes equally to the final prediction.

Sensitivity analysis examines how changes in model inputs affect model outputs.

For a model output \(Y\) and parameter \(p\), a local sensitivity can be expressed conceptually as:

In practice, sensitivity analysis may involve varying parameters over plausible ranges and observing the effect on exposure, concentration, clearance, or another output.

Uncertainty can arise from:

• Experimental measurement error
• Uncertain physiological parameters
• Uncertain drug-specific properties
• Scaling assumptions
• Structural model assumptions
• Interindividual variability
• Limited clinical validation data

Understanding which assumptions drive a prediction is often as important as the predicted value itself.

19. PBPK Versus Conventional Compartmental PK

These approaches are not mutually exclusive. A compartmental model can be highly useful for describing clinical PK, while PBPK may be more appropriate when the scientific question requires mechanistic extrapolation.

20. What Can PBPK Models Be Used For?

PBPK models can support a broad range of drug-development questions when the model is appropriate and sufficiently qualified for the intended use.

• First-in-human dose selection
• Pediatric drug development
• Drug-drug interaction assessment
• Renal impairment simulations
• Hepatic impairment simulations
• Formulation and absorption assessment
• Exposure prediction in special populations
• Bridging across populations
• Exposure-response and PK/PD modeling
• Support for dose selection and clinical trial design

The FDA states that PBPK analyses may be submitted to support applications including INDs, NDAs, BLAs, and ANDAs. The EMA guideline describes PBPK reporting in regulatory submissions including marketing applications, pediatric investigation plans, and clinical trial applications. :contentReference[oaicite:1]{index=1}

21. PBPK in Drug Development and Regulatory Submissions

PBPK models are increasingly used in drug development because they can provide a mechanistic framework for integrating information from preclinical studies, in vitro experiments, clinical pharmacology studies, and physiological databases.

For regulatory submissions, however, the model should be presented as an evidence-supported analysis rather than simply as a simulation generated by software.

The FDA's 2018 guidance recommends six major sections for a PBPK study report:

1. Executive Summary
2. Introduction
3. Materials and Methods
4. Results
5. Discussion
6. Appendices

The FDA also notes that acceptance of PBPK results in place of clinical PK data is determined case by case, taking into account the intended use and the quality, relevance, and reliability of the PBPK analysis. :contentReference[oaicite:2]{index=2}

The EMA guideline emphasizes both qualification of the PBPK platform for the intended use and assessment of the predictive performance of the specific drug model. :contentReference[oaicite:3]{index=3}

22. What PBPK Models Do Not Tell Us Automatically

PBPK modeling can be highly mechanistic, but mechanistic does not mean assumption-free.

• Physiological parameters are not known exactly. Organ volumes, blood flows, enzyme abundance, and other properties vary among individuals.
• Drug-specific measurements can be uncertain. In vitro measurements may not perfectly represent in vivo behavior.
• Scaling assumptions can be important. IVIVE may introduce uncertainty when experimental systems differ from human physiology.
• Model structure matters. Different assumptions about tissue distribution, metabolism, or transport can produce different predictions.
• More detail does not automatically mean greater accuracy. Additional mechanisms can introduce additional parameters and uncertainty.
• Extrapolation is conditional. A model validated in one context may not automatically be adequate for a substantially different application.
• Software does not replace model evaluation. A commercial platform does not automatically qualify every drug model built with it.

23. A Practical PBPK Modeling Workflow

1. Define the question. Identify exactly what the model needs to predict.
2. Define the intended population. Specify healthy adults, patients, children, elderly subjects, organ impairment, pregnancy, or another target population as appropriate.
3. Collect drug-specific data. Assemble physicochemical, binding, permeability, metabolic, transporter, formulation, and clinical PK information.
4. Define the physiological system. Select appropriate organ volumes, blood flows, enzyme and transporter information, and population characteristics.
5. Construct the PBPK model. Represent relevant absorption, distribution, metabolism, and excretion mechanisms.
6. Check units and implementation. Confirm mass balance, numerical stability, parameter values, and model equations.
7. Simulate known studies. Compare model predictions against relevant observed PK data.
8. Evaluate predictive performance. Determine whether the model performs adequately for the intended use.
9. Perform sensitivity and uncertainty analyses. Identify assumptions that materially influence the prediction.
10. Document the model. Record inputs, sources, assumptions, parameterization, software, simulations, diagnostics, and limitations.
11. Apply the model. Simulate the target scenario while clearly distinguishing model-based prediction from direct observation.

Where to Go Next

A natural progression is to study the individual components of PBPK modeling in greater detail:

• Physiological system models
• Drug-specific physicochemical inputs
• Mechanistic tissue distribution
• Hepatic clearance and IVIVE
• Renal clearance
• Oral absorption and gastrointestinal PBPK
• Enzyme and transporter modeling
• Drug-drug interaction modeling
• Pediatric and special-population PBPK
• Model verification and qualification
• PBPK applications in regulatory submissions

The next tutorial can build directly on this foundation by examining physiological system models and organ compartments and showing how blood flow, tissue volume, partitioning, and organ-specific clearance combine to generate a whole-body PBPK model.

References

1. U.S. Food and Drug Administration. Physiologically Based Pharmacokinetic Analyses — Format and Content Guidance for Industry. September 2018. FDA guidance.
2. European Medicines Agency. Guideline on the reporting of physiologically based pharmacokinetic (PBPK) modelling and simulation. EMA/CHMP/458101/2016. Effective July 1, 2019. EMA guideline.
3. European Medicines Agency. Guideline on the reporting of physiologically based pharmacokinetic (PBPK) modelling and simulation. Scientific guideline PDF. EMA PDF.
4. U.S. Food and Drug Administration. Program of Physiologically-Based Pharmacokinetic and Pharmacodynamic Modeling (PBPK Program). FDA PBPK Program.

These references provide regulatory context for PBPK modeling and reporting. The equations and conceptual examples in this tutorial are simplified teaching representations and should not be interpreted as a complete specification of any particular PBPK platform or regulatory model.