Why the Exposure Metric Matters
An animal NOAEL is not fully informative for first-in-human dose selection until the exposure associated with it is characterized. Two regimens can produce the same total exposure but different peak concentrations; conversely, similar peaks can accompany different exposure over time. If toxicity is related to the peak, the total exposure may be a poor stand-in. If injury develops with sustained or cumulative exposure, a peak alone may miss the concern.
The earlier material on using animal pharmacokinetics to convert NOAEL to human exposure establishes how to connect an animal dose with measured exposure and a predicted human exposure. This tutorial addresses the next decision: which exposure measure should constrain the dose? The answer depends on the toxicologic endpoint and its plausible relationship to exposure, not on which measure is easiest to calculate.
These measures have different units and meanings. \(C_{\max}\) has concentration units, such as mg/L; AUC has concentration-times-time units, such as mg·h/L; duration above a threshold has units of time. They cannot be compared by their numerical values alone. Any comparison must be between a metric-specific exposure limit and a human prediction of that same metric under a specified regimen.
Relating the Endpoint to a Metric
A metric is most defensible when the toxicology supports its connection to the endpoint. A rapid, transient effect that appears near peak concentrations and resolves as concentrations fall may support a peak-related concern. Progressive injury during repeated dosing, especially when it tracks total exposure or improves as exposure declines, may support AUC or a cumulative measure. A toxicity that requires concentrations to remain above a biologically relevant level may be better represented by time above threshold or trough concentration.
These are evidence-based clues, not automatic rules. Toxicity can relate to more than one measure, and a single observed pattern rarely establishes mechanism by itself. Consider dose timing, onset and recovery, exposure-response patterns across animals, and whether the endpoint was observed after single or repeated dosing. Verify that concentration sampling was frequent enough to capture the peak and that the AUC interval matches the endpoint’s relevant period.
Suppose \(E(D)\) is a predicted human exposure metric at dose \(D\), and \(E_{\mathrm{lim}}\) is a justified limit for that same metric. Under dose proportionality, write \(E(D)=kD\), where \(k\) is exposure per unit dose. The dose that reaches the limit is then:
The limit must be appropriate to the metric and should be justified by the evidence and the broader risk assessment. This calculation alone does not establish that an exposure is safe, nor does it determine any safety factor. When separate, justified limits apply to multiple metrics, calculate a dose constraint for each and use the most restrictive constraint for the proposed regimen. Do not invent a limit for a metric merely because its value is available.
Worked Dose Calculations
Worked Example: A Peak-Linked Acute Endpoint
Hypothetical data. In a repeat-dose animal study, a transient neurologic sign occurred near the measured peak and resolved as concentrations declined. A separate assessment has specified an illustrative human \(C_{\max}\) ceiling of 1.20 mg/L. An independently justified AUC ceiling of 6.00 mg·h/L is also being checked. For the proposed human regimen, predicted exposure is dose-proportional: \(C_{\max}=0.100D\) mg/L and AUC \(=0.700D\) mg·h/L, where \(D\) is in mg.
Peak constraint. Solving \(0.100D=1.20\) mg/L gives:
AUC check. Solving \(0.700D=6.00\) mg·h/L gives:
If both ceilings are justified and apply to this regimen, the more restrictive calculated constraint is 8.57 mg, because \(8.57<12.0\). Thus the AUC check, not the peak-linked endpoint by itself, limits the dose under the stated assumptions. If the AUC ceiling were not independently supported, it should not be imposed just because it produces the lower number. These illustrative ceilings are not claimed to be suitable clinical limits.
Worked Example: An Exposure-Accumulation Concern
Hypothetical data. A liver enzyme abnormality increased over repeated dosing and tended to resolve during recovery. Across animals, its pattern was more consistent with sustained exposure than with a brief peak. For illustration, assume the risk assessment has specified an AUC ceiling of 8.00 mg·h/L per dosing interval and a separate \(C_{\max}\) ceiling of 2.40 mg/L. Predicted human exposures under the intended interval are AUC \(=0.500D\) mg·h/L and \(C_{\max}=0.0800D\) mg/L.
The AUC-based constraint is:
The peak-based check is:
If both limits apply, the AUC constraint is more restrictive: the candidate dose must not exceed 16.0 mg on these assumptions. The endpoint pattern supports examining AUC, but it does not prove that AUC is causal; the exposure-response and recovery evidence should be reviewed. The AUC is specifically per dosing interval, so it must not be substituted with an AUC measured over a different interval without justification.
Worked Example: When Time Above a Threshold Is More Informative
Hypothetical data. A toxicity is thought to require concentration to remain above a toxicologically supported threshold of 0.800 mg/L. For a simplified single-dose prediction, assume a monoexponential decline after an instantaneous peak, with \(C_{\max}=0.400D\) mg/L and an elimination rate constant of \(0.200\ \mathrm{h}^{-1}\). The specified limit is no more than 4.00 h above the threshold. An independently justified AUC ceiling for the same dose is 10.0 mg·h/L.
Under the assumed decline, concentration after the peak is \(C(t)=C_{\max}e^{-kt}\). Setting \(C(t)=C_{\mathrm{th}}\) and solving for time above threshold gives \(t_{\mathrm{above}}=\ln(C_{\max}/C_{\mathrm{th}})/k\), provided \(C_{\max}>C_{\mathrm{th}}\). The 4.00 h limit therefore requires:
Since \(C_{\max}=0.400D\) mg/L, the dose constraint is:
For this model, AUC after the peak is \(C_{\max}/k=(0.400D)/0.200=2.00D\) mg·h/L, so its constraint is \(D_{\mathrm{AUC}}=10.0/2.00=5.00\) mg. If both limits apply, time above threshold is more restrictive at 4.45 mg. This result depends on the stated PK simplification and threshold; a different absorption profile or a threshold without toxicologic support would make this calculation inappropriate.
Justification, Diagnostics, and Alternatives
The dose conversion is a direct consequence of the assumed exposure-dose relationship. If \(E(D)=kD\), the condition for remaining at or below the limit is \(kD\leq E_{\mathrm{lim}}\). For \(k>0\), dividing both sides by \(k\) gives \(D\leq E_{\mathrm{lim}}/k\). Thus \(D_{\mathrm{lim}}\) is the largest dose satisfying that metric-specific constraint under the model. If exposure is nonlinear in dose, this algebra does not apply; instead, solve the modeled inequality \(E(D)\leq E_{\mathrm{lim}}\) using the nonlinear exposure prediction.
Review onset, severity, reversibility, dosing schedule, and exposure patterns across animals. State what evidence supports a peak, integrated, sustained, or other exposure relationship.
Define whether the comparison uses \(C_{\max}\), AUC over a dosing interval, trough, or time above threshold. Use the same relevant interval and schedule in the animal evidence and human prediction.
Confirm that the human exposure estimate is suitable for the proposed dose range and regimen, and that dose proportionality is a reasonable approximation if using a linear conversion.
Convert each justified exposure limit to a dose. If several independently justified limits apply, select the most restrictive dose constraint and document which metric controls.
Useful diagnostics include checking whether observed toxicity follows peak timing or develops with continued exposure; whether recovery tracks declining exposure; whether higher-dose animals show a consistent exposure-response pattern; and whether sampling could have missed the true peak. Compare individual-animal data where possible: group averages can hide animals with high peaks or unusually prolonged exposure. Review whether the AUC window captures the relevant exposure, and whether single-dose PK is an adequate proxy for the repeated-dose regimen.
A mismatch between the chosen metric and the proposed regimen is a common failure mode. For example, using single-dose \(C_{\max}\) to constrain a repeated-dose effect may miss accumulation, while comparing AUC over 24 hours with AUC over a shorter dosing interval may distort the dose calculation. If PK is nonlinear, clearance changes with dose, or the timing of exposure is poorly observed, the simple proportional calculation may be unreliable; use an appropriate PK model and state the uncertainty.
| Metric | What it summarizes | When it may be informative | Key limitation |
|---|---|---|---|
| \(C_{\max}\) | Peak concentration during a defined interval | Rapid or transient effects plausibly linked to peak exposure | Can miss duration and cumulative exposure |
| AUC | Integrated concentration over a defined interval | Effects associated with sustained or total interval exposure | Different concentration-time profiles can have the same AUC |
| Trough or time above threshold | Residual concentration or duration above a specified level | Effects requiring persistent exposure or threshold exceedance | Requires an appropriate interval and, for a threshold metric, a supported threshold |
| Cumulative exposure | Exposure summed over multiple doses or time | Effects that build over a course of treatment | May obscure the timing and spacing of exposure |
No metric is universally conservative. Choosing whichever metric gives the lowest dose can appear cautious, but can be misleading if that metric has no supported limit or is calculated over the wrong interval. Conversely, relying on only one metric can overlook a second plausible exposure-related hazard. The appropriate comparison is endpoint-specific and should preserve the distinction between evidence, assumptions, and risk-management choices.
Common Mistake
A metric-based dose calculation is one component of starting-dose selection, not a stand-alone declaration of safety. Its usefulness depends on the quality of the toxicology, the exposure measurements, and the human PK prediction. Record the endpoint, selected metric, interval, assumptions, and any other justified constraint so that the controlling calculation can be reviewed.
Check Your Understanding
For each question, first identify the relevant metric and assumptions. Solutions follow each prompt.
- A transient adverse sign occurs close to the animal concentration peak and resolves rapidly. What metric is a reasonable primary candidate, and what would you check before using it?
Solution: \(C_{\max}\) is a reasonable candidate because the effect is temporally associated with peak exposure. Check that sampling captured the peak, that the pattern is consistent across animals, and that the human prediction represents the proposed regimen. - A human prediction gives \(C_{\max}=0.0500D\) mg/L and a justified peak ceiling is 1.00 mg/L. Under dose proportionality, calculate the dose constraint.
Solution: \(D=1.00/0.0500=20.0\) mg. This is the dose reaching the specified \(C_{\max}\) ceiling under the stated prediction, not a general safety conclusion. - Why is a 24-hour AUC not automatically interchangeable with AUC over a 12-hour dosing interval?
Solution: They integrate concentration over different periods and can have different values. The limit and prediction must refer to a matching, toxicologically relevant interval. - Suppose supported AUC and \(C_{\max}\) limits imply dose constraints of 14 mg and 18 mg, respectively. Which constraint controls if both apply to the regimen?
Solution: The 14 mg AUC-based constraint is more restrictive because \(14<18\). This choice is warranted only if both limits are justified and apply to that regimen. - What changes if the predicted exposure is not proportional to dose?
Solution: Do not use \(D_{\mathrm{lim}}=E_{\mathrm{lim}}/k\) as a linear conversion. Use an appropriate nonlinear exposure model to find the dose satisfying \(E(D)\leq E_{\mathrm{lim}}\).