Worked Translation Case From Animal Exposure to Human Response

From Animal Data to a Human Response Profile

A translation analysis is a chain of linked predictions. Animal pharmacokinetic (PK) information is used to specify a human concentration-time profile; pharmacodynamic (PD) parameters are then applied to that exposure profile to predict response over time. Each link depends on assumptions, so the final curve is a conditional projection—not a direct observation of human response.

This hypothetical case uses a simple one-compartment intravenous bolus model and a direct inhibitory Emax relationship. Its purpose is to make the calculations and their dependencies visible. It does not establish a clinical dose, a clinically meaningful response, or the likelihood that an individual patient will respond.

Scaling Animal PK to a Human Concentration Profile

Suppose a hypothetical 10-kg animal has an estimated volume of distribution \(V_a=5.00\) L and clearance \(CL_a=0.2651\) L/h. The animal receives 11.90 mg intravenously, equivalent to 1.190 mg/kg. The human projection is for a 70-kg adult receiving the same dose per kilogram. Thus, the weight ratio is \(70/10=7\), and the projected human dose is \(1.190\times70=83.3\) mg.

Under the stated scaling rules, \(V\) is proportional to body weight and \(CL\) is proportional to body weight raised to 0.75. Applying those rules gives the human volume and clearance:

The elimination rate constant is \(k=CL/V\), so the projected human value is \(1.141/35.0=0.03260\ \mathrm{h}^{-1}\). Immediately after the bolus, the one-compartment concentration is dose divided by volume: \(C_0=83.3/35.0=2.38\) mg/L. The concentration model is therefore approximately \(C(t)=2.38\exp(-0.03260t)\) mg/L, with time \(t\) in hours.

These values are model-based concentrations, not measurements in people. In this example, the scaling rule also implies a longer elimination half-life in the larger species: \(t_{1/2}=\ln(2)/k\). For the projected human \(k\), this is \(0.6931/0.03260\approx21.26\) hours. The half-life is a consequence of the selected clearance and volume scaling assumptions, not an independent validation of them.

Translating Concentration to Response

Suppose the hypothetical animal PD analysis supports a normalized baseline \(R_0=1.00\), a maximum fractional inhibition \(I_{\max}=0.250\), and an \(IC_{50}=0.410\) mg/L. For a direct inhibitory Emax model, the fractional inhibition at concentration \(C\) is \(I_{\max}C/(IC_{50}+C)\). The predicted response is baseline multiplied by one minus this fractional inhibition:

This formulation assumes that concentration is nonnegative, \(IC_{50}>0\), and that \(I_{\max}\) is the maximum fractional reduction from baseline. At \(C=0\), the model returns \(R_0\). As \(C\) becomes very large, the response approaches \(R_0(1-I_{\max})\), which is 0.750 in this case. The equation specifies an immediate concentration-response relationship; it does not itself represent a delayed effect or a changing baseline.

Checking Sensitivity and the Translation Chain

A point prediction can conceal how much it depends on transferred parameters. For example, at a hypothetical concentration of 0.4979 mg/L at 48 hours, compare the response under two illustrative \(IC_{50}\) values, keeping \(R_0=1.00\) and \(I_{\max}=0.250\) fixed. This is a parameter sensitivity check, not evidence that either value is more plausible.

A translation case is best checked as a sequence rather than judged only by its final response curve. The following checks make the main links explicit.

What the Projection Does—and Does Not—Establish

The main alternative to simple allometric scaling is a more mechanistic PK projection, such as a physiologically based pharmacokinetic model. That approach can represent species-specific physiology and processes such as tissue distribution or metabolism, but it requires appropriate structural assumptions and input data. The present case is easier to calculate and audit, but compresses those processes into clearance and volume. A further alternative is to estimate or update the human concentration-response relationship using human data when available; simply transferring an animal \(IC_{50}\) does not test whether sensitivity is conserved.

Several failure modes can break the apparent chain. If clearance is nonlinear over the relevant exposure range, a dose-proportional concentration profile may be wrong. If the drug is highly protein-bound and binding differs by species, applying total concentration directly may misrepresent the active exposure. If response lags behind concentration, the direct Emax curve may predict the wrong timing even when its concentration-response shape is reasonable. A species-dependent baseline or disease state can also make an otherwise correctly scaled exposure produce a different response. The earlier material on translating baseline and disease-state effects and translating delayed effects addresses those model features; they cannot be resolved by adjusting PK scaling alone.

Useful diagnostics include plotting observed animal concentrations against the fitted PK profile, inspecting residuals over time and concentration, and checking whether PD residuals show systematic time patterns or species-specific direction. Predictions that depend strongly on a transferred parameter should be shown across plausible values, as in the \(IC_{50}\) comparison. More elaborate models are not automatically better: their additional parameters need data or defensible external information. Conversely, a close fit in the animal does not guarantee accurate human prediction if the cross-species assumptions fail.