Worked Example: Allometric NOAEL-Based Starting Dose

From a Toxicology NOAEL to a Human Starting Dose

A NOAEL-based starting-dose calculation is a sequence of linked judgments, not a single conversion. The toxicology package must first identify an appropriate species and a dose at which adverse effects were not observed. That animal dose can then be converted to a human-equivalent dose (HED), reduced by a safety factor, and checked against predicted human exposure. Each step answers a different question: species selection concerns biological relevance, HED conversion concerns dose scaling, and exposure prediction concerns how much drug the human is expected to encounter.

The calculation below is a hypothetical illustration, not a recommendation for any investigational drug. It assumes a systemically available drug with sufficiently characterized pharmacology and pharmacokinetics to support the stated comparisons. The NOAEL is the highest tested dose in a specified study at which no treatment-related adverse effect was observed under that study's conditions. It is not proof that the dose has no biological effect or that it is safe in humans.

Selecting the Species and NOAEL

Species selection precedes arithmetic. A useful species should be pharmacologically relevant to the intended human target or mechanism, and the toxicology study should provide interpretable dose and exposure information. The choice also considers the pattern and severity of findings, study duration, route, and whether the observed effects can be assessed in humans. If two species are relevant, the more conservative result may inform the starting dose, but “choose the species with the lowest NOAEL” is not a sufficient rule: doses in different species cannot be compared directly without considering scaling and exposure.

For this example, suppose a hypothetical compound has pharmacological activity in both rats and cynomolgus monkeys. Adequate repeat-dose studies report a rat NOAEL of 100 mg/kg/day and a monkey NOAEL of 30 mg/kg/day. At the next higher tested doses, each study reports treatment-related adverse findings. Both studies use a route relevant to the proposed clinical route, and both provide exposure measurements. These assumptions make both species informative; the monkey is selected as the initial controlling species because its converted HED is lower, and its toxicology findings and exposures are considered relevant to the planned comparison.

Converting the NOAEL to an HED

In the body-surface-area method, the \(K_m\) factor is body weight divided by body surface area. It therefore expresses the relationship between a dose per kilogram and a dose per unit surface area. To match the animal dose per unit surface area to a human dose per unit surface area, the animal dose is multiplied by the animal \(K_m\) and divided by the human \(K_m\). In practice, the conventional factors used here are treated as dimensionless conversion factors.

This conversion does not claim that animal and human plasma concentrations, AUC, or pharmacodynamic responses are equal at the HED. It is a dose-scaling convention. Human exposure requires a separate pharmacokinetic prediction or measurement.

Predicting Human Exposure and Applying the Safety Factor

To connect a dose to exposure, the example uses a simple linear pharmacokinetic relationship. For a dose given over an interval, the interval AUC at steady state is \(F \times \mathrm{Dose}/CL\), where \(F\) is the fraction of the dose reaching systemic circulation and \(CL\) is clearance. The same expression gives AUC from a single dose extrapolated to infinity when the model and assumptions apply. Here, assume once-daily dosing, linear pharmacokinetics over the dose range, \(F=0.80\), and a predicted human clearance of 5.00 L/h. Under these assumptions, predicted AUC is proportional to dose. Real starting-dose decisions may also need to consider other exposure measures, such as Cmax, and nonlinear or time-dependent pharmacokinetics can invalidate this simple scaling.

A safety factor is applied to the HED as a dose divisor. Its rationale should reflect the evidence and residual risks, including the quality and relevance of toxicology, pharmacological activity, exposure comparisons, and uncertainty in translation. The arithmetic below uses a factor of 10 solely as a stated hypothetical choice; the factor is not automatically appropriate for every program.

Justification, Diagnostics, and Limitations

The order of operations matters. The HED conversion first places the animal NOAEL on a human dose-per-kilogram scale; the safety factor then reduces that HED; and the PK calculation estimates exposure at the resulting dose. With linear pharmacokinetics, reducing the dose by a factor of 10 reduces predicted AUC by a factor of 10. This follows directly from the exposure equation:

The justification is conditional on \(F\) and \(CL\) remaining constant over this range. If absorption is dose-dependent, clearance is nonlinear, or the prediction is outside the supported data range, the proportionality need not hold.

Before accepting a result, check both the scientific assumptions and the arithmetic. Verify that the chosen species is pharmacologically relevant, that the reported NOAEL belongs to the stated route and schedule, and that the dose units match the \(K_m\) conversion. Check that the body weight is used only when converting mg/kg to a total dose. In the exposure calculation, use compatible dose and clearance units, and state whether the estimate represents a single-dose AUC or an interval AUC. Compare the predicted human exposure with relevant toxicology exposure measures rather than assuming HED implies exposure equivalence.

A NOAEL-based calculation is one input to first-in-human dose selection, not a stand-alone guarantee of safety. The HNSTD and MABEL approaches answer related but distinct questions; the appropriate basis depends on the pharmacology, toxicology, and evidence available. A dose based on a toxicology NOAEL may not be sufficiently conservative if strong human pharmacology is expected below the NOAEL-derived exposure. Conversely, a large safety factor does not repair an irrelevant species, a poorly characterized NOAEL, or a flawed exposure prediction.