Why Match Exposure Rather Than Dose per Kilogram?
An animal dose and a human dose can be expressed in the same units, but the same number of milligrams per kilogram does not imply the same drug exposure. Clearance, bioavailability, absorption, and other pharmacokinetic properties differ across species. Exposure matching instead asks: what human dose is predicted to produce an exposure comparable to the exposure measured in the animal?
The exposure measure must be specified. AUC summarizes drug exposure over a stated interval, while Cmax is the maximum concentration over the relevant dosing period. Either can be compared across species when the measurements are defined consistently and a suitable human pharmacokinetic prediction is available. Exposure matching is a calculation, not by itself a recommendation that the resulting human dose is appropriate to administer.
The Exposure-to-Dose Relationship
For a drug with linear pharmacokinetics over the dose range of interest, AUC over a dosing interval is proportional to dose under a fixed regimen. For a single dose, the usual relationship is \(AUC = F D/CL\), where \(D\) is dose, \(F\) is the fraction reaching systemic circulation, and \(CL\) is clearance. For repeated dosing at steady state, the corresponding relationship over one interval is \(AUC_{\tau}=F D/CL\), with \(D\) denoting the dose per interval and \(AUC_{\tau}\) the AUC over that interval.
Thus, matching an animal AUC requires a human dose that produces that same AUC according to the human prediction. If human clearance and bioavailability are available, the relationship gives a direct calculation. If a human model instead predicts exposure at a reference dose, its exposure-per-dose relationship can be used, provided proportionality is justified. For Cmax, the relationship is less universal: it depends on the route, absorption, formulation, dose interval, and concentration-time profile. A Cmax prediction at a human reference dose is therefore often more useful than trying to infer Cmax from clearance alone.
A short derivation shows why the AUC calculation is based on human pharmacokinetics, not a direct animal-to-human dose ratio. Suppose the target is the measured animal AUC, \(AUC_a\), and human exposure is described by \(AUC_h=F_hD_h/CL_h\). Setting human exposure equal to the target and solving for human dose gives:
This result assumes that the human \(CL_h\) and \(F_h\) used in the calculation are appropriate for the dose, route, and population being considered, and that the AUC relationship applies over the dose range. It does not require the animal dose to be converted by body weight. The animal dose identifies the conditions under which the target AUC was observed; the human dose is obtained from the predicted human exposure relationship.
Worked Example: Matching Animal AUC With a Human Clearance Prediction
In a hypothetical animal study, a dose of 30 mg/kg produced an AUC of 120 mg·h/L. A human pharmacokinetic prediction gives clearance of 3.0 L/h and bioavailability of 0.75 for the route and regimen being evaluated. Calculate the human dose predicted to match the animal AUC.
Use \(D_h=AUC_a CL_h/F_h\). Substituting the values gives:
The units reduce to milligrams: \((\text{mg·h/L})(\text{L/h})=\text{mg}\). Under the stated assumptions, a 480 mg human dose is predicted to produce an AUC of 120 mg·h/L. As a check, the predicted AUC at that dose is \(0.75(480\ \text{mg})/(3.0\ \text{L/h})=120\ \text{mg·h/L}\). The animal dose of 30 mg/kg is useful context for the observed exposure, but it is not multiplied by a human body weight to obtain this estimate.
This is an exposure-equivalence calculation only. It does not establish that the route, regimen, or dose is suitable for use in people, nor does it account for uncertainty in the animal measurement or human clearance prediction.
Matching Cmax and Using Human Reference-Dose Predictions
Cmax matching follows the same general logic: find the human dose whose predicted Cmax equals the animal Cmax. If human Cmax is proportional to dose under the relevant conditions, a reference-dose prediction supplies the conversion. Define \(q_h=Cmax_h(D_{ref})/D_{ref}\), the predicted human Cmax per unit dose for that regimen. Then the dose predicted to match an animal target \(Cmax_a\) is \(D_h=Cmax_a/q_h\). This shortcut is conditional: a change in dose can alter absorption, peak timing, or other processes, so proportionality must not simply be presumed.
Worked Example: Matching Cmax From a Human Reference Dose
In a hypothetical animal study, Cmax was 2.4 mg/L after a 20 mg/kg dose. A human model predicts Cmax of 0.30 mg/L after a 100 mg dose using the same intended route and dosing conditions. Assume the human Cmax prediction is dose proportional over the range needed for the calculation. Find the human dose predicted to match 2.4 mg/L.
The human Cmax per milligram is \(0.30\ \text{mg/L}/100\ \text{mg}=0.0030\ \text{mg/L per mg}\). Therefore:
Equivalently, proportional scaling gives \(D_h=100\ \text{mg}\times(2.4/0.30)=800\ \text{mg}\). The units are milligrams of human dose, and the predicted Cmax at that dose is \(0.0030\times800=2.4\ \text{mg/L}\). The result depends on dose proportionality and comparability of the concentration-time conditions; it is not a conclusion that 800 mg is an appropriate clinical dose.
A model may predict a nonlinear exposure-dose relationship. In that case, the target is obtained by solving for the dose in the predicted relationship, rather than applying a dose ratio. Interpolation within a well-supported model range is generally more defensible than extrapolation far beyond observed or otherwise justified doses.
Worked Example: Matching AUC With a Nonlinear Human Prediction
Suppose a hypothetical animal dose produced a target AUC of 45 mg·h/L. A human model predicts AUC values of 20 mg·h/L at 100 mg, 45 mg·h/L at 200 mg, and 80 mg·h/L at 300 mg. What human dose in this model matches the animal target?
The model predicts \(AUC_h(200\ \text{mg})=45\ \text{mg·h/L}\). Thus the matching dose is 200 mg. The predicted values show that exposure is not proportional to dose: doubling the dose from 100 mg to 200 mg raises AUC from 20 to 45 mg·h/L, not to 40 mg·h/L. For example, proportional scaling from 100 mg would give \(100(45/20)=225\) mg, which is inconsistent with the stated model prediction at 200 mg.
The appropriate calculation here is to use the model’s exposure-dose relationship, not to impose proportionality. The answer is supported directly at a model-evaluated dose; if the target fell between model points, interpolation would require a specified and defensible model or interpolation rule.
Assumptions, Diagnostics, and Failure Modes
Before using an exposure match, check whether the target and the human prediction refer to comparable quantities. AUC over one interval is not automatically interchangeable with AUC over a different interval or after a single dose. Cmax depends on the dose interval and the concentration-time profile. Route and formulation differences can alter exposure, especially peak concentration. The prediction must also correspond to the intended population and dosing conditions.
When proportional scaling is used, inspect whether exposure divided by dose is reasonably stable across the relevant dose range. Dose-escalation data or model predictions at multiple doses can reveal curvature. For AUC, departures from dose proportionality can arise when clearance or bioavailability changes with dose. For Cmax, changes in absorption rate or peak timing can make a reference-dose ratio misleading. A ratio calculated from one dose cannot diagnose these patterns on its own.
Also check whether the estimated human pharmacokinetic parameters are uncertain. Because \(D_h=AUC_a CL_h/F_h\), uncertainty in clearance, bioavailability, and the measured animal AUC carries directly into the calculated dose. Report the assumptions and, when possible, examine how reasonable changes in key inputs affect the estimate. A single precise-looking result can conceal a wide range of plausible matched doses.
- Metric mismatch: Comparing AUC over different intervals or comparing an observed Cmax with a prediction defined under a different regimen makes the equality hard to interpret.
- Unjustified proportionality: Multiplying dose by an exposure ratio assumes a stable exposure-per-dose relationship. If the relationship is nonlinear, use the predicted dose-exposure curve instead.
- Route or formulation mismatch: A change in bioavailability or absorption can alter AUC or Cmax. Do not treat exposure predictions from different conditions as interchangeable without justification.
- False precision: A calculated dose is conditional on inputs and model assumptions. Report meaningful precision and explain material uncertainty rather than presenting the result as an exact biological constant.
How Exposure Matching Differs From Other Conversions
Body-surface-area conversion and allometric scaling use cross-species relationships to transform dose or predict pharmacokinetic parameters. Exposure matching instead uses an observed animal exposure as the target and a human exposure-dose relationship to calculate a corresponding human dose. These approaches answer related but different questions: a dose conversion does not itself guarantee matched exposure, while an exposure match depends on a credible human pharmacokinetic prediction.
| Approach | Primary calculation | What it requires | Important limitation |
|---|---|---|---|
| Body-surface-area conversion | Transforms an animal dose using species-specific factors | Animal dose and conversion factors | Does not directly establish matched AUC or Cmax |
| Allometric scaling | Predicts a human pharmacokinetic quantity from cross-species relationships | Relevant multispecies data and a justified scaling relationship | Predictions can be unreliable when species differences disrupt the relationship |
| Exposure matching | Finds the human dose predicted to match a specified animal exposure | Animal exposure and a human dose-exposure prediction | Depends on the exposure metric, conditions, and accuracy of the human prediction |
These approaches can inform an evidence assessment together, but they should not be treated as interchangeable calculations. In particular, an exposure-matched dose is not automatically a starting-dose recommendation. Its interpretation depends on the purpose of the comparison, the reliability of the underlying data, and other safety and pharmacology considerations.
Common Pitfall: Treating Exposure Matching as a Dose-per-Kilogram Rule
A common error is to reason, “The animal received 20 mg/kg, so the human should receive 20 mg/kg to match exposure.” This confuses dose with the resulting concentration-time profile. The same dose per kilogram can produce different AUC and Cmax when clearance, bioavailability, or absorption differ. The correction is to identify the target exposure and calculate the human dose from the human exposure-dose relationship. If that relationship is unavailable or too uncertain, the exposure match is not well supported; a body-weight calculation does not fill the gap.
The practical value of exposure matching is that it makes the comparison explicit: the target metric, the human prediction, the assumptions, and the calculation can all be inspected. Its limitation is equally important: it cannot repair a poor human prediction, an incompatible exposure metric, or substantial species-specific pharmacology.
Check Your Understanding
Answer each question, then compare your reasoning with the solutions.
- A hypothetical animal AUC is 50 mg·h/L. Human clearance is predicted to be 2.0 L/h and bioavailability is 0.80. What human dose matches the AUC under the linear model?
- A human model predicts Cmax of 0.20 mg/L at 50 mg. Under dose proportionality, what dose is predicted to match an animal Cmax of 1.2 mg/L?
- Why is it not justified to use a dose ratio when predicted AUC rises from 20 mg·h/L at 100 mg to 45 mg·h/L at 200 mg?
- Name two conditions that should be checked before comparing animal and human Cmax values.
Solutions. 1. \(D_h=AUC_aCL_h/F_h=(50\ \text{mg·h/L})(2.0\ \text{L/h})/0.80=125\ \text{mg}\). 2. The predicted Cmax per dose is \(0.20/50=0.0040\ \text{mg/L per mg}\), so \(D_h=1.2/0.0040=300\ \text{mg}\). This answer depends on dose proportionality. 3. Doubling dose does not double AUC: the increase is from 20 to 45 mg·h/L. A simple dose ratio assumes proportionality, so the human dose-exposure prediction should be used instead. 4. Check that route and formulation are comparable and that dosing conditions, including the dose interval, support a meaningful Cmax comparison. Also verify that the Cmax definitions and measurement or prediction conditions are consistent.