Genetic Variation as a PBPK Input

From Inherited Variation to a Model Input

A genetic result is not itself a clearance, concentration, or physiological space. It describes inherited variation that may alter a drug-relevant function, such as the amount or catalytic activity of an enzyme. PBPK modeling translates that functional difference into a parameter used by the drug model. The translation must be specific to the gene, the drug, the pathway, and the evidence supporting the assumed effect.

The key modeling distinction is between structure and parameterization. A genotype does not create a new liver or a new pharmacokinetic compartment. Instead, it may change an existing parameter, such as enzyme-specific intrinsic clearance in the liver. If the same enzyme acts in another tissue, a corresponding parameter there may also need adjustment; the genetic result does not imply that every tissue or every drug is affected equally.

For a diploid autosomal locus with two copies and no copy-number complication, a genotype commonly records the two alleles without saying which other variants lie together on each chromosome. A diplotype includes that chromosome-level pairing. A phenotype summarizes function rather than simply restating the sequence result. An activity score can support that summary, but its scale and interpretation depend on the gene and the rules used to assign allele function. Neither a score nor a phenotype is automatically a universal physical unit of enzyme activity.

What Must Be Assumed to Change a PBPK Parameter?

A modeler needs a defensible link from the genetic observation to a drug-relevant functional effect. For an enzyme-mediated pathway, a common implementation is to scale the reference enzyme-specific intrinsic clearance by a multiplier. The multiplier must represent the effect for the modeled drug and pathway, not merely a generic label such as “reduced function.”

Here \(CL_{\mathrm{int,ref}}\) is the reference intrinsic clearance attributable to the enzyme for the drug, \(CL_{\mathrm{int},g}\) is the corresponding value for genetic group \(g\), and \(m_g\) is a dimensionless relative functional multiplier. This form assumes that the multiplier is supported for the drug-enzyme pair, that the reference value represents the same pathway and conditions, and that changing the parameter does not require changing other model components. For example, the equation alone does not establish that binding, hepatic blood flow, or the abundance of other enzymes changes.

A useful mechanistic justification follows from enzyme kinetics. At concentrations well below the Michaelis constant, enzyme-mediated rate is approximately proportional to \(V_{\max}/K_m\), which is the low-concentration intrinsic clearance. If a genetic difference changes effective enzyme amount or catalytic capacity by a factor \(m_g\), while \(K_m\) is unchanged, then \(V_{\max,g}=m_gV_{\max,\mathrm{ref}}\), and therefore \(CL_{\mathrm{int},g}=m_gCL_{\mathrm{int,ref}}\). If the variant also changes substrate affinity, a single multiplier on intrinsic clearance may not reproduce the concentration dependence; \(K_m\) or a fuller kinetic description may need to change instead.

The effect is drug-specific because different drugs can be substrates to different degrees, can use multiple pathways, and can respond differently to a functional change in the same protein. A phenotype category is therefore not a universal clearance value. The mapping from genetic information to a model parameter needs to preserve the drug’s pathway structure and the evidence for the effect.

Preserve the Drug’s Pathway Structure

When a drug has more than one metabolic route, changing one enzyme-specific parameter should not silently scale the entire drug clearance. If pathways act in parallel and their intrinsic clearances are expressed on a compatible basis, their contributions can be added. A genetic multiplier is applied only to the affected contribution; unaffected routes remain as specified by the model.

This expression assumes the first term is the contribution of enzyme E and the second term represents all other modeled routes, unchanged across the genetic groups being compared. If the pathways interact, saturate, or are described by incompatible kinetic quantities, the simple sum is not sufficient.

Trace the Parameter Through the PBPK Model

A parameter change matters because it alters the drug model’s equations, not because genetic variation adds a new compartment. For a well-stirred liver with hepatic blood flow \(Q_H\), unbound fraction in blood \(f_{u,b}\), and total hepatic intrinsic clearance \(CL_{\mathrm{int}}\), the hepatic clearance is

This relation assumes the well-stirred liver representation and consistent clearance units. It shows why a change in intrinsic clearance does not generally produce the same proportional change in hepatic clearance: the result also depends on blood flow and unbound fraction. In an actual PBPK model, the adjusted enzyme parameter enters the relevant tissue process, and the model then propagates its effect through the existing compartments and equations.

Parameterization Choices, Diagnostics, and Limits

There are two common levels of implementation. A compact model can assign a genotype- or phenotype-specific multiplier to an enzyme-specific clearance parameter. This is efficient when the evidence supports a stable relative effect and the drug is adequately described by that parameter. A more mechanistic implementation can alter enzyme abundance, catalytic capacity, substrate affinity, or another component of the relevant kinetic process. That approach may be needed when the genetic effect is concentration-dependent or when the underlying mechanism is known well enough to represent directly. Neither approach should be mistaken for creating a genotype-specific organ compartment.

A categorical covariate model may be useful when only group-level clinical estimates are available, but its categories still need a clear parameter target and reference. Conversely, a biologically detailed enzyme model is not automatically better if the variant effect is poorly characterized. The appropriate choice is the simplest representation that matches the evidence and preserves the relevant pathways.

Several checks can reveal a faulty translation. Confirm that the genetic result identifies the relevant locus and that the functional interpretation applies to the modeled enzyme and drug. Check that a reported multiplier scales the intended pathway rather than total clearance. Verify units, reference values, and whether the parameter is already genotype-adjusted; applying an effect twice produces an unintended change. Compare predicted pathway contributions and exposure profiles with the direction and scale supported by available evidence. Implausible clearance, a discrepancy between pathway-specific and total effects, or an unexpected effect in an unrelated route can indicate a coding or structural error.

The simple multiplier has important limits. It may not represent altered substrate affinity, multiple functional variants on the same chromosome, gene copy-number changes, uncertain variant function, or variation in expression across tissues. A genotype-based functional classification may also differ from actual current activity because non-genetic influences can modify function. These possibilities are reasons to examine model assumptions and data, not reasons to assign an unsupported adjustment to every PBPK parameter.