From Disease Phenotype to PBPK Parameter
Renal or hepatic disease is not itself a PBPK parameter. It is a clinical state that may alter several distinct physiological or drug-specific quantities. A model should represent the changes that are relevant to the drug and the mechanism being modelled, rather than multiply every clearance term by one disease-severity factor. For example, reduced glomerular filtration directly affects the filtration component of renal clearance, while a change in hepatic blood flow enters the liver model through hepatic perfusion. Neither change automatically determines what happens to tubular transport, binding, or intrinsic metabolic capacity.
The practical task is to connect a phenotype to a parameter and state the assumptions behind that connection. The examples below use hypothetical values. Each is an illustration of parameter translation, not a clinical estimate for a particular disease or drug.
The distinction between a measured value and a model assumption matters. A reported decrease in organ function does not, by itself, establish that every organ process has decreased by the same percentage. Nor does a decrease in albumin concentration imply a particular numerical change in a drug's unbound fraction without a binding model or relevant binding data.
Renal Filtration Is One Component of Renal Clearance
For a drug that is freely filtered and has no net tubular secretion or reabsorption in the calculation, the filtration clearance can be represented as \(CL_{R,\mathrm{fil}}=GFR\,f_{u,p}\). Here \(GFR\) is expressed as a volume of plasma per unit time and \(f_{u,p}\) is the unbound fraction in plasma. Their product has units of clearance. This relation assumes that filtration is proportional to the unbound drug concentration in plasma and that the GFR used is appropriate for the modelled individual.
This expression concerns the filtration contribution, not necessarily total renal clearance. If a model includes secretion or reabsorption, those processes need their own terms and assumptions. A change in GFR alone does not establish that these other processes change in the same proportion.
Worked Example: Updating Filtration Without Scaling All Clearance
Suppose a hypothetical drug has \(f_{u,p}=0.10\). In a reference subject, \(GFR=120\ \mathrm{mL/min}\). The filtration contribution is
For a hypothetical disease phenotype, take \(GFR=48\ \mathrm{mL/min}\), with the same unbound fraction. Then
The filtration term is 40% of its reference value: \(4.8/12=0.40\). Now suppose, solely for illustration, that reference total clearance consists of 12 mL/min filtration, 8 mL/min secretion, and 30 mL/min nonrenal clearance. If secretion and nonrenal clearance are assumed unchanged, disease total clearance is \(4.8+8+30=42.8\ \mathrm{mL/min}\), compared with \(50\ \mathrm{mL/min}\) at reference. The total clearance ratio is \(42.8/50=0.856\), not 0.40. The result depends on the explicit assumption that the other terms remain unchanged; it is not a general claim about renal disease.
Hepatic Blood Flow and the Well-Stirred Liver Model
The well-stirred liver model makes the roles of hepatic blood flow and intrinsic clearance explicit. Use blood-based concentrations consistently: \(Q_H\) is hepatic blood flow, \(f_{u,b}\) is the unbound fraction in blood, and \(CL_{\mathrm{int},H}\) is intrinsic hepatic clearance defined consistently with the unbound blood concentration. This convention avoids substituting a plasma unbound fraction into an equation based on whole-blood concentrations without a blood-to-plasma conversion.
The model assumes a well-mixed liver, steady flow through the organ, and drug removal proportional to unbound concentration within the liver. It also assumes that the specified flow and clearance terms adequately represent hepatic handling for the modelled pathway. Under these assumptions, the blood concentration leaving the liver is \(C_{b,\mathrm{out}}\), and that entering is \(C_{b,\mathrm{in}}\).
The result follows from a mass balance. At steady state, the drug removed from blood per unit time equals the unbound-concentration-driven intrinsic removal within the liver:
Rearranging gives \(C_{b,\mathrm{out}}/C_{b,\mathrm{in}}=Q_H/(Q_H+f_{u,b}CL_{\mathrm{int},H})\). The extracted fraction is therefore \(1-C_{b,\mathrm{out}}/C_{b,\mathrm{in}}\), or \(f_{u,b}CL_{\mathrm{int},H}/(Q_H+f_{u,b}CL_{\mathrm{int},H})\). Multiplying this fraction by inflowing hepatic blood flow gives the stated clearance equation.
Worked Example: Translating a Hepatic Blood-Flow Change
Consider a hypothetical drug with \(f_{u,b}=0.05\) and \(CL_{\mathrm{int},H}=2000\ \mathrm{mL/min}\). Let reference hepatic blood flow be \(90\ \mathrm{mL/min}\). The well-stirred equation gives
Now suppose the disease phenotype is represented by \(Q_H=54\ \mathrm{mL/min}\), while \(f_{u,b}\) and intrinsic clearance are held fixed. Then
The hepatic clearance ratio is \(35.06/47.37=0.740\), even though flow has fallen to \(54/90=0.600\) of reference. The extraction ratio rises from \(47.37/90=0.526\) to \(35.06/54=0.649\). Thus, in this example, clearance does not fall in direct proportion to flow. The calculation assumes no simultaneous change in binding, intrinsic clearance, or other relevant processes.
Albumin and Functional Liver Mass: Distinct Model Changes
Albumin concentration can affect the binding of drugs that bind appreciably to albumin, but concentration is not itself the drug's unbound fraction. A binding model, drug-specific binding data, or a justified estimate is needed to translate a change in albumin into \(f_{u,b}\) or another consistently defined unbound fraction. The relationship need not be a simple inverse proportion across the full range of albumin concentrations. Other binding components and the drug's affinity also matter.
Functional liver mass is a separate physiological consideration. A model may scale intrinsic hepatic capacity with functional mass if it assumes that the remaining functional tissue has the same drug-metabolizing activity per unit mass as the reference tissue. Under that assumption, a functional-mass factor \(r_M\) gives \(CL_{\mathrm{int},H,\mathrm{disease}}=r_M CL_{\mathrm{int},H,\mathrm{ref}}\). This is not an automatic consequence of a change in anatomical liver size or a general disease label. If per-mass activity also changes, the mass factor alone is insufficient.
Worked Example: Combining Binding and Functional-Mass Assumptions
For a hypothetical drug, let \(Q_H=90\ \mathrm{mL/min}\), reference \(f_{u,b}=0.020\), and reference \(CL_{\mathrm{int},H}=2000\ \mathrm{mL/min}\). The reference hepatic clearance is
Suppose, hypothetically, that a binding model or measurement supports changing \(f_{u,b}\) to 0.040, and that functional liver mass is represented as 70% of reference. Assume per-unit-mass intrinsic activity is unchanged, so \(CL_{\mathrm{int},H}=0.70(2000)=1400\ \mathrm{mL/min}\). Holding \(Q_H\) fixed, the disease clearance is
Under these particular assumptions, the calculated total-blood-concentration hepatic clearance is higher than the reference value, despite lower intrinsic capacity. The result occurs because the assumed increase in unbound fraction changes the product \(f_{u,b}CL_{\mathrm{int},H}\) from 40 to 56 mL/min. It does not establish that hypoalbuminemia generally increases clearance, nor does it by itself determine unbound exposure. The assumptions and the concentration basis must be reported with the result.
Building and Checking a Disease-Specific Parameter Set
A useful parameter set records each translation separately. The table shows the type of mapping to consider; it is not a universal prescription to change every listed parameter in every patient.
| Phenotype information | Candidate PBPK change | Key condition or caution |
|---|---|---|
| Reduced GFR | Update the filtration contribution using GFR and \(f_{u,p}\) | Do not infer proportional changes in secretion, reabsorption, or nonrenal clearance without evidence or an explicit assumption. |
| Changed hepatic blood flow | Update \(Q_H\) in the hepatic model | Recalculate clearance using the model; do not assume clearance has the same ratio as flow. |
| Changed albumin-related binding | Update a suitable drug-specific unbound fraction or binding representation | Albumin concentration alone does not specify the numerical binding change. |
| Reduced functional liver mass | Scale intrinsic capacity if the model assumes unchanged activity per unit functional mass | Separate the mass assumption from any change in per-mass activity. |
Several checks can reveal a faulty translation. Confirm that flow, GFR, and clearance values use compatible units; a product such as \(GFR\,f_{u,p}\) must have units of clearance. Confirm whether the model uses blood or plasma concentrations before choosing the corresponding unbound fraction. Check that a mass scaling rule is applied to the intended capacity term rather than to blood flow or all clearances indiscriminately. Finally, inspect whether the predicted direction and size of change are plausible under the stated assumptions. A calculated clearance that changes unexpectedly is a reason to examine the component terms, not to force a global multiplier onto the output.
Alternative approaches can be useful for different purposes. A uniform clearance multiplier is simple and may serve as a rough empirical summary when only an overall clearance change is available, but it cannot identify which PBPK mechanism changed and may transfer poorly to another drug. A component-based model is more mechanistic and can distinguish filtration, hepatic perfusion, binding, and functional capacity, but it requires more inputs and stronger assumptions. If the available phenotype data do not support a mechanistic parameter change, the uncertainty should be stated rather than hidden by a precise-looking multiplier.
Common Pitfall and Limits of Interpretation
Wrong reasoning: “GFR is 40% of reference, so total clearance must be 40% of reference,” or “albumin is lower, so hepatic clearance should be multiplied by a chosen factor.” Both claims jump from one phenotype measurement to an unsupported change in an aggregate model output.
Correction: Apply the GFR change to the filtration term under its stated assumptions, and preserve or separately justify other clearance terms. For hepatic handling, enter the supported changes in \(Q_H\), \(f_{u,b}\), and intrinsic capacity into the model, then calculate the resulting clearance. A phenotype may alter several parameters at once, but each mapping needs a rationale. Model predictions also remain limited by the quality of the inputs, the validity of the well-stirred assumption, and whether the selected model includes the relevant pathways.
Check Your Understanding
For each item, identify the relevant parameter and check the reasoning against the solution.
- A hypothetical drug has \(f_{u,p}=0.25\), and GFR changes from 100 to 60 mL/min. Under the filtration-only assumptions, what are the reference and disease filtration clearances? Solution: \(100(0.25)=25\ \mathrm{mL/min}\) and \(60(0.25)=15\ \mathrm{mL/min}\). Only the filtration term is shown to fall by 40%; total clearance cannot be calculated without its other components.
- In the well-stirred model, which unbound fraction belongs in the hepatic clearance equation when hepatic flow and concentrations are expressed on a whole-blood basis? Solution: Use \(f_{u,b}\), the unbound fraction in blood, with consistently defined blood-based terms. Substituting \(f_{u,p}\) requires a justified blood-to-plasma conversion or an explicit assumption that makes the fractions equivalent.
- A model assumes functional liver mass is 80% of reference and per-unit-mass intrinsic activity is unchanged. If reference intrinsic clearance is 500 mL/min, what disease value follows from this assumption? Solution: \(0.80(500)=400\ \mathrm{mL/min}\). This scales the intrinsic capacity term only; it does not imply that hepatic blood flow or every clearance term is also 80% of reference.
- Why is a 50% decrease in albumin concentration not, by itself, enough to calculate a drug's new unbound fraction? Solution: The concentration change does not specify the drug-specific binding relationship or account for other binding contributions. A binding model or suitable data is needed to justify the change in unbound fraction.