Reporting Model Structure and Assumptions

Why Structure and Assumptions Need to Be Reported

A PBPK prediction is interpretable only when a reviewer can tell what system was represented and what rules governed its simulation. A statement such as “a PBPK model predicted exposure in the target population” leaves crucial questions unanswered: which tissues were represented, how drug moved between them, how the dose entered the system, and what was assumed about the population and physiology? A model diagram alone is not enough if its compartments and arrows are not defined; equations alone are not enough if the simulated regimen and initial conditions are unclear.

This tutorial focuses on describing those elements. It does not replace an executive summary or a discussion of the evidence supporting a model. The reporting task here is to make the model’s implemented structure and assumptions legible: a reviewer should be able to trace the path from dose, through the represented system, to the predicted quantity.

Describe the Model as a Connected System

A useful model diagram identifies each compartment, the substance or quantity represented in it, and the direction of each relevant transfer. It should show the route of administration and the route by which drug reaches systemic circulation, as well as elimination or metabolism pathways represented in the model. Labeling a box “liver” is not sufficient if the equations treat it as a single well-mixed compartment; the report should say so. Likewise, distinguish compartments that represent blood from those representing tissue, and state whether a diagram depicts anatomical connections, mathematical connections, or both.

The diagram and its accompanying description should agree with the implemented model. If the model includes a gut absorption compartment, a portal pathway, and a liver compartment, the reported structure should make clear how material moves among them. If a tissue is omitted, combined with another compartment, or represented by a lumped residual compartment, say so. These descriptions help a reviewer determine whether a reported prediction corresponds to the system implied by the model rather than to a different, more detailed system suggested by a generic diagram.

State Equations and the Meaning of Simplifications

For each dynamic compartment, report the state variable and its governing equation, or provide an equivalent specification that makes the implemented relationships unambiguous. A common starting point is mass balance: the rate of change of drug amount equals the rate entering minus the rate leaving, including any metabolism or other modeled loss. For a perfusion-limited tissue compartment \(i\), one possible equation is:

Here \(C_i\) is the tissue concentration, \(V_i\) is tissue volume, \(Q_i\) is blood flow to the tissue, \(C_a\) is arterial blood concentration, and \(K_{p,i}\) is the tissue-to-blood partition coefficient used in the model. Under the stated equilibrium assumption, the venous concentration leaving the tissue is \(C_i/K_{p,i}\). The term \(CL_{\mathrm{int},i}C_i\) represents a modeled metabolic loss when \(CL_{\mathrm{int},i}\) is expressed as a clearance acting on the tissue concentration. If metabolism is not included in that tissue, the loss term is absent. The report must define its own state variables and concentration conventions; symbols alone do not establish what the implementation means.

The equation follows directly from mass balance. The amount in the tissue is \(V_iC_i\), so its rate of change is \(V_i\,dC_i/dt\) when \(V_i\) is fixed. Blood enters at rate \(Q_iC_a\) and leaves at rate \(Q_iC_i/K_{p,i}\). Subtracting the leaving rate and the metabolic loss gives the equation above. This derivation also exposes assumptions: constant tissue volume, the stated relationship between tissue and venous concentrations, and the specified flow and metabolism terms.

Define the Population, Regimen, and Simulation Conditions

A report should identify the population represented by the simulation in clinically meaningful terms, including relevant physiological characteristics that define the modeled group. It should say whether the model represents, for example, adults with normal organ function or a specified patient group, rather than relying on a broad label such as “the clinical population.” The report should not imply that a simulation represents people or circumstances that were not specified in its setup.

For each simulated regimen, give the administered amount, route, timing, duration or number of administrations, and any relevant formulation or input assumptions. State whether the listed amount is the administered dose or the amount treated as available to the model. Also identify the simulated output and its units, such as a plasma concentration-time profile or an exposure summary. These details are needed to interpret what the model’s result describes.

Initial conditions and boundaries are equally important. Specify the initial amount or concentration in each dynamic compartment, any pre-dose or steady-state condition, and the simulation time interval. State how doses are introduced, what concentration or flux conditions apply at the system boundary, and how modeled material leaves the system. If a compartment begins empty, say so rather than leaving that condition implicit.

Make Assumptions and Boundaries Reviewable

List assumptions that materially define the simulated system, rather than relying on a generic statement that “standard PBPK assumptions” were used. Examples include instantaneous tissue-blood equilibrium, well-mixed compartments, fixed or time-varying physiology, linear versus saturable processes, and whether a pathway or process is omitted. For each, briefly state where it applies and what it excludes. A concise, specific statement is more useful than an unqualified claim that the model is physiologically realistic.

The boundary conditions and assumptions should be internally consistent with the equations. For example, a model that assumes fixed tissue volume should not silently use a time-varying volume in its mass balance. A model that represents venous concentration as tissue concentration divided by a partition coefficient should explain that relationship rather than leaving the reviewer to infer it from the diagram. If external input, physiological conditions, or dosing change during a simulation, state when and how those changes are imposed.

Check for Completeness and Common Failure Modes

A practical review asks whether a reader can reconstruct the model’s conceptual operation from the report. The diagram, equations, population description, regimen, and initial and boundary conditions should be mutually consistent. Their purpose is not to reproduce every implementation detail in prose, but to ensure that key choices are identifiable and that the predicted quantity has a clear meaning.

• Unlabeled arrows. A diagram shows connections but not whether an arrow represents blood flow, absorption, or another transfer. Label the process and its direction.
• Compartments without definitions. A box named for an organ may conceal whether it represents tissue, blood, or a combined space. State the represented quantity and mixing assumption.
• Implicit dose handling. A nominal dose is reported without clarifying its route, input compartment, timing, or fraction treated as available to the model. Specify each item.
• Unstated starting conditions. A concentration-time profile is shown, but the reader cannot tell whether it begins from zero, after a prior dose, or at steady state. Report the initial state.
• Equations and diagrams that disagree. A pathway is shown but has no corresponding transfer in the equations, or an equation contains a process absent from the diagram. Reconcile the two descriptions.

A common mistake is to assume that naming a recognized PBPK structure is enough: “a whole-body model with standard compartments.” That phrase does not specify which compartments were included, how they connect, or which processes were simplified or omitted. The correction is to describe the implemented structure and its conditions explicitly. This tutorial’s focus is structural reporting; a detailed accounting of where parameter values came from belongs in the separate reporting of parameter provenance and supporting evidence.

The main alternatives are not simply “more detail” and “less detail.” A model may represent many organs separately or use a reduced structure in which some tissues are grouped. Either can be described adequately if its compartments, connections, equations, and intended simulated population are clear. A more detailed diagram does not by itself demonstrate that the model is appropriate for its use, and a concise model description does not excuse leaving assumptions unstated. Structural transparency lets a reviewer understand what was simulated; it does not, on its own, establish the adequacy of the choices.