Mapping PK Parameters Across Species

Why Parameter Mapping Requires More Than a Body-Weight Ratio

A cross-species pharmacokinetic prediction is only as credible as the mapping assigned to each parameter. Clearance and volume are often candidates for body-size scaling, but their exponents need not be the same. Absorption rate and bioavailability describe different processes and generally cannot be obtained by multiplying an animal estimate by a body-weight ratio. A useful translation therefore specifies, parameter by parameter, what is being transferred, what scaling rule is used, and what evidence supports that rule.

The introductory material on uncertainty in physiological scaling explains why a prediction depends on its scaling assumptions. Here the focus is the mapping itself: define a parameter consistently across species, select a defensible rule for it, and check the implications before using it in a human prediction.

Mapping Clearance and Volume

A common empirical mapping relates a parameter to body weight using a power law. For clearance, the rule can be written as follows, where \(A\) denotes the animal species and \(H\) the human species:

Here \(CL\) is clearance, \(BW\) is body weight, and \(b_{CL}\) is the clearance scaling exponent. A frequently used starting assumption for small-molecule clearance is \(b_{CL}=0.75\), but this is not a universal biological constant or a guarantee of accurate prediction for a particular drug. The exponent may be estimated from suitable multispecies data, supported by a relevant mechanistic model, or used as an explicit assumption when data are limited. The body-weight ratio is dimensionless, so the predicted clearance retains the units of \(CL_A\), such as L/h.

Volume requires its own mapping. For example, a chosen volume parameter \(V\) may be mapped as

A proportional-to-body-weight rule corresponds to \(b_V=1\), but that too is an assumption to justify rather than a default that applies in every case. Specify whether the parameter is, for example, central volume or steady-state volume, and use the same definition in each species. A model's central volume and steady-state volume are not interchangeable merely because both are reported in litres.

Clearance and volume should be mapped separately. Their ratio determines the elimination rate constant in a one-compartment model, \(k_e=CL/V\), and the corresponding half-life is \(t_{1/2}=0.693V/CL\). Consequently, changing the assumed exponent for either parameter can materially change the predicted time course, even when the predicted clearance seems plausible. A half-life is not an independent substitute for checking the mappings of \(CL\) and \(V\).

Absorption and Bioavailability Need Different Decisions

The absorption rate constant \(k_a\), when a first-order absorption model is appropriate, has units of inverse time. It describes the rate of drug entering the systemic circulation from an absorption site; it is not a clearance or a volume. A body-weight power rule for clearance does not establish a scaling rule for \(k_a\). Absorption can depend on formulation, dissolution, gastrointestinal transit, local physiology, and other species-dependent processes. Unless a validated model supports a specific mapping, use a species-specific estimate or a justified human value or range rather than mechanically scaling animal \(k_a\).

Absolute bioavailability \(F\) is the fraction of an administered dose that reaches the systemic circulation unchanged. It is dimensionless and lies between 0 and 1. For an oral dose, it can reflect the fraction absorbed and losses before systemic entry, including intestinal and hepatic first-pass loss. These processes can differ by species and by formulation. Thus an animal \(F\) is informative evidence, but should not automatically be treated as the human \(F\).

For linear pharmacokinetics and comparable conditions, absolute bioavailability can be calculated from oral and intravenous exposures:

Dose-normalized exposure is used because the oral and intravenous doses need not be equal. This relationship assumes exposure is proportional to dose over the doses used, that the AUCs represent the same relevant exposure interval, and that clearance is comparable between the route assessments. It should not be used without qualification if pharmacokinetics are nonlinear or the conditions make clearance differ.

Under the same linear assumptions, oral AUC is \(F\times Dose/CL\), whereas intravenous AUC is \(Dose/CL\). Therefore, after oral administration, \(F\) affects the total systemic exposure while \(k_a\) primarily affects the rate and shape of the concentration-time profile. In the simple model with complete characterization of AUC, changing \(k_a\) alone does not change total AUC. It can, however, change peak concentration and time to peak, which matter when predicting a concentration-time profile or a peak-related response.

How to Evaluate a Proposed Mapping

Before applying a mapping, check that both the source and target parameters have the same meaning. Clearance estimated from plasma concentrations may not be directly comparable with clearance defined using whole-blood concentrations without considering the blood-to-plasma relationship. Likewise, an oral apparent clearance \(CL/F\) is not the same quantity as systemic clearance \(CL\). A mistaken parameter identity cannot be repaired by choosing a more precise allometric exponent.

A useful evaluation checks the biological and empirical basis of each rule, then examines the resulting PK profile. For clearance, compare the prediction with available species data and relevant information on metabolic or excretory pathways. For volume, check whether the mapped value is consistent with the drug's distribution properties and the specific volume parameter in the model. For oral absorption, confirm that route and formulation are relevant; for \(F\), consider whether species differences in absorption and first-pass loss could be material.

Warning signs include markedly inconsistent human predictions from different source species, an implausible implied half-life, a predicted \(F\) outside its allowable range, or a simulated absorption profile that conflicts with observed timing or formulation behavior. These findings do not identify the correct alternative by themselves. They indicate that the mapping assumptions, parameter definitions, data quality, or model structure need review. A mapping that reproduces one exposure summary may still fail to reproduce the concentration-time profile.

Choosing Among Mapping Approaches

Empirical allometry is convenient when several species provide suitable data and the parameter follows a reasonably coherent body-size relationship. Its strengths are transparency and ease of calculation. Its weaknesses are dependence on the species set, the fitted or assumed exponent, and the drug's biology; a power law can summarize observed relationships without explaining them.

Mechanistic scaling uses relevant physiological quantities and drug-specific processes to predict a parameter. For example, a clearance prediction may use information about organ blood flow and metabolic capacity rather than body weight alone. This approach can be more informative when species differences in the elimination pathway are understood, but it requires appropriate inputs and assumptions. A complicated mechanistic calculation is not automatically more reliable than a simple rule.

A third option is to retain a species-specific value, particularly for absorption rate and bioavailability, or when no credible transfer rule is supported. For human prediction, that may mean using an explicit human scenario or range based on relevant evidence rather than carrying over the animal estimate. The trade-off is that the prediction may be less numerically precise, but it avoids implying that an unvalidated scaling relationship is known.

Common Mistake and Practical Limits

No mapping removes limitations in the underlying data. Sparse species information may not identify a reliable exponent, and study differences in dose, route, assay, formulation, or sampling can masquerade as species effects. A human prediction also depends on whether the model's structural assumptions adequately represent the drug's disposition. Therefore report a mapped value as a conditional prediction: it follows from the stated data, parameter definitions, and assumptions, rather than being a direct measurement of human PK.