Worked Example: Integrating MABEL and Toxicology

Why Integrate Pharmacology and Toxicology?

A first-in-human starting dose should reflect more than one line of evidence. Pharmacology can identify a dose expected to produce only a small, biologically relevant effect; toxicology can provide a dose anchored to an animal no-observed-adverse-effect level (NOAEL). These approaches answer different questions. A MABEL (minimum anticipated biological effect level) is driven by the dose expected to begin producing a biological effect in humans, while a toxicology-based estimate begins with animal tolerability and translation to a human equivalent dose (HED).

Neither estimate automatically resolves whether the proposed human exposure is acceptably separated from animal exposure at the NOAEL. Comparing predicted human and animal exposures adds useful context, particularly when Cmax and AUC tell different stories. The aim is to assemble these pieces transparently, not to turn them into a universal formula that replaces scientific judgment.

Definition: A pharmacology-based MABEL is a dose estimate grounded in the anticipated onset of a biological effect, using relevant pharmacology and its translation to humans. A toxicology-based dose estimate starts with an animal NOAEL converted to an HED and reduced by a prespecified safety factor. An exposure margin is the ratio of an appropriate animal exposure at the NOAEL to the predicted human exposure at a candidate dose, calculated separately for endpoints such as Cmax and AUC.

A Framework for Combining the Evidence

First state each estimate and the assumptions behind it. The MABEL calculation should already reflect the pharmacology relevant to the intended human population, including the biological endpoint and how the animal or in vitro evidence translates. The earlier material on receptor-occupancy MABEL provides one example of deriving a pharmacology-based dose. For the toxicology estimate, use the NOAEL-derived HED and the safety factor justified for the program; do not confuse the HED itself with the reduced starting-dose estimate.

D_{\text{tox}}=\frac{\mathrm{HED}_{\text{NOAEL}}}{F}

Here \(D_{\text{tox}}\) is the safety-factored toxicology dose and \(F\) is the applied safety factor. A straightforward dose-based comparison is to consider the lower of \(D_{\text{MABEL}}\) and \(D_{\text{tox}}\). This comparison is a conservative organizing principle, not a claim that the two estimates are interchangeable: one is pharmacology-based and the other is toxicology-based.

Exposure comparisons provide a separate check. For an endpoint \(j\), such as Cmax or AUC, define the margin at candidate human dose \(D\) as:

M_j(D)=\frac{E_{j,\text{animal at NOAEL}}}{E_{j,\text{human at }D}}

The numerator and denominator must refer to the same exposure metric and compatible units. A margin of 10 means that the animal exposure at the NOAEL is ten times the predicted human exposure at that dose for that endpoint. It does not mean a tenfold probability of safety, and it is not a universal acceptance threshold. Any target margin must be justified in the context of the molecule, endpoint, study quality, and program.

Deriving an Exposure-Based Dose Constraint

If human exposure is approximately proportional to dose over the range being considered, an exposure-margin criterion can be expressed as a dose constraint. Let \(D_r\) be a reference human dose with predicted exposure \(E_{j,\text{human}}(D_r)\), and let \(M_{j,\min}\) be the minimum margin selected for endpoint \(j\). Under dose-proportional exposure, the predicted exposure at dose \(D\) is \(E_{j,\text{human}}(D_r)D/D_r\). Requiring the margin to be at least \(M_{j,\min}\) gives:

\frac{E_{j,\text{animal at NOAEL}}} {E_{j,\text{human}}(D_r)D/D_r} \geq M_{j,\min}

All quantities are positive, so rearranging yields an upper dose bound:

D\leq D_r \frac{E_{j,\text{animal at NOAEL}}} {M_{j,\min}E_{j,\text{human}}(D_r)}

This result is conditional on dose-proportional exposure and on the selected margin criterion. With nonlinear pharmacokinetics, calculate predicted exposures at each candidate dose using the appropriate model rather than scaling exposure by a dose ratio. In either case, evaluate Cmax and AUC separately: a candidate must be considered against each relevant endpoint, not just the endpoint that gives the larger margin.

Worked Example: Comparing MABEL, HED, and Both Exposure Margins

Hypothetical data. A pharmacology analysis gives a MABEL of \(0.012\) mg/kg. The animal NOAEL converts to an HED of \(0.40\) mg/kg, and the selected safety factor is 10. At \(0.012\) mg/kg, predicted human Cmax is \(0.24\) mg/L and AUC is \(1.8\) mg·h/L. Assuming dose-proportional exposure, animal NOAEL exposures are estimated as Cmax \(12\) mg/L and AUC \(90\) mg·h/L.

Toxicology-based dose. Apply the safety factor: \(D_{\text{tox}}=0.40/10=0.040\) mg/kg. This is above the MABEL of \(0.012\) mg/kg, so the MABEL is the lower of the two dose-based estimates.

Exposure margins at the toxicology-based dose. The dose ratio is \(0.040/0.012=3.333\). Under the stated proportionality assumption, predicted human Cmax is \(0.24\times3.333=0.800\) mg/L, and AUC is \(1.8\times3.333=6.00\) mg·h/L. Therefore: \(M_{\mathrm{Cmax}}=12/0.800=15.0\), and \(M_{\mathrm{AUC}}=90/6.00=15.0\).

Exposure margins at the MABEL. At \(0.012\) mg/kg, the margins are \(12/0.24=50.0\) for Cmax and \(90/1.8=50.0\) for AUC. Thus, the lower dose-based estimate also has larger exposure margins in this hypothetical case. Subject to the assumptions and the overall risk assessment, \(0.012\) mg/kg is the defensible starting-dose candidate from these calculations.

When an Exposure Criterion Adds a Constraint

A safety-factored HED is not guaranteed to produce a desired margin for every exposure endpoint. If the program has justified minimum exposure margins, calculate the dose bound for each endpoint and consider those bounds along with the MABEL and toxicology-based dose. The following example illustrates the calculation; its selected criteria are hypothetical and are not general standards.

Worked Example: Applying a Cmax Margin Criterion

Hypothetical data. Suppose \(D_{\text{MABEL}}=0.030\) mg/kg and \(\mathrm{HED}_{\text{NOAEL}}=0.60\) mg/kg. With \(F=10\), \(D_{\text{tox}}=0.60/10=0.060\) mg/kg. At a reference human dose \(D_r=0.030\) mg/kg, predicted Cmax is \(0.50\) mg/L and AUC is \(4.0\) mg·h/L. Animal NOAEL exposures are Cmax \(8.0\) mg/L and AUC \(120\) mg·h/L. Assume proportionality and, solely for this example, minimum margins of 10 for each endpoint.

Cmax constraint. The maximum dose meeting the Cmax criterion is: \(0.030\times[8.0/(10\times0.50)]=0.048\) mg/kg. The AUC-based bound is: \(0.030\times[120/(10\times4.0)]=0.090\) mg/kg. Cmax is the more restrictive exposure constraint.

Integrated decision. Compare \(0.030\) mg/kg (MABEL), \(0.060\) mg/kg (safety-factored HED), and \(0.048\) mg/kg (Cmax bound). The lowest is \(0.030\) mg/kg. At that dose, Cmax margin is \(8.0/0.50=16.0\) and AUC margin is \(120/4.0=30.0\); both exceed the example criteria. The margin criteria are useful here as explicit checks, not as substitutes for the MABEL or toxicology rationale.

Check the Exposure Model, Not Just the Dose Arithmetic

Dose-based scaling is convenient only when it is supported. Clearance changes, target-mediated disposition, saturable absorption, or other nonlinear processes can make exposure fail to increase in proportion to dose. In those cases, a dose ratio cannot validly generate the human exposure at another dose; use the relevant PK model predictions at the candidate dose. The assumptions behind animal-to-human comparisons also matter: species relevance, sampling, exposure measurement, and the relation between the toxicology finding and systemic exposure all affect how informative a margin is.

Worked Example: Nonlinear Exposure Predictions

Hypothetical data. A pharmacology-based analysis gives a MABEL of \(0.020\) mg/kg. The NOAEL-derived HED is \(0.50\) mg/kg and the safety factor is 10, giving \(D_{\text{tox}}=0.50/10=0.050\) mg/kg. A nonlinear human PK model predicts Cmax \(0.40\) mg/L and AUC \(3.0\) mg·h/L at the MABEL, but Cmax \(1.50\) mg/L and AUC \(7.5\) mg·h/L at \(0.050\) mg/kg. Animal NOAEL exposures are Cmax \(18\) mg/L and AUC \(150\) mg·h/L.

At \(0.050\) mg/kg, the margins are \(18/1.50=12.0\) for Cmax and \(150/7.5=20.0\) for AUC. At the MABEL, they are \(18/0.40=45.0\) for Cmax and \(150/3.0=50.0\) for AUC. The candidate doses satisfy \(D_{\text{MABEL}}<D_{\text{tox}}\), and the lower dose has larger predicted margins. Crucially, the example uses model-predicted exposures at each dose; it does not infer them by multiplying the MABEL exposures by the dose ratio. That distinction is necessary because the stated PK is nonlinear.

Assumptions, Diagnostics, and Limitations

Before relying on an integrated estimate, check that each input represents what its label implies. The MABEL should be connected to a relevant biological response and a defensible human dose translation. The toxicology dose should use an appropriate NOAEL and HED, with the safety factor stated rather than hidden. For exposure margins, verify that the animal and human values use the same metric and compatible units, and that the human prediction corresponds to the proposed dose and regimen.

A mismatch between endpoints is informative. If Cmax gives a much smaller margin than AUC, do not average the margins: investigate whether peak concentration is particularly relevant to the toxicology or pharmacology, and retain endpoint-specific comparisons. Sparse or uncertain exposure data, a weakly relevant animal species, or a steep response near the MABEL can limit how confidently any single calculated margin supports a starting dose. A numerical margin is a comparison, not proof that the human dose is safe.

The integrated framework also has limits. The lower-of-dose-estimates rule is a conservative way to organize the pharmacology and toxicology dose calculations; it cannot resolve every case where the underlying evidence differs in quality or relevance. Safety factors and minimum exposure margins require scientific justification, and applying multiple reductions without understanding their purpose can obscure the rationale or count the same concern more than once. Document what each factor or constraint is intended to protect against.

Common Mistake

Pitfall: “The HED divided by the safety factor is below the NOAEL-derived HED, so it must be the starting dose, regardless of the MABEL or exposure data.”

Correction: The safety-factored HED is one dose-based estimate, not an automatic selection. Compare it with the pharmacology-based MABEL, then examine relevant Cmax and AUC margins at the candidate dose. State assumptions and any criteria used; do not treat a nominal safety factor or exposure ratio as a guarantee of safety.

Key takeaway: Keep three elements distinct: the MABEL, the safety-factored NOAEL-derived HED, and endpoint-specific exposure margins. Use them together to justify a starting-dose candidate, while making clear which assumptions and risk judgments support that choice.

Check Your Understanding

Answer each question, then compare your reasoning with the brief solutions.

A hypothetical program has a MABEL of \(0.010\) mg/kg and an HED of \(0.30\) mg/kg with a safety factor of 10. What is the toxicology-based dose, and which dose-based estimate is lower?
Animal NOAEL Cmax is \(15\) mg/L, and predicted human Cmax at a candidate dose is \(0.75\) mg/L. Calculate and interpret the Cmax margin.
Why should Cmax and AUC margins be reported separately rather than averaged?
At a reference dose, predicted human AUC is \(5\) mg·h/L. Animal NOAEL AUC is \(100\) mg·h/L, and a hypothetical minimum AUC margin is 20. Assuming dose-proportional exposure, what is the largest dose relative to the reference dose that meets the criterion?

Solutions. (1) \(D_{\text{tox}}=0.30/10=0.030\) mg/kg; the MABEL, \(0.010\) mg/kg, is lower. (2) \(M_{\mathrm{Cmax}}=15/0.75=20\); animal NOAEL Cmax is 20 times the predicted human Cmax at that dose. (3) They describe different exposure properties and may have different relevance; averaging can hide a limiting endpoint. (4) The reference-dose exposure gives a margin of \(100/5=20\), exactly the criterion. Under proportionality, the largest dose is \(1.0\) times the reference dose; any higher dose would reduce the margin below 20.