Translating Human Pharmacology into an Exposure Target

From Human Pharmacology to an Exposure Target

A pharmacological response is driven by exposure at a relevant biological site, not by dose alone. Human-relevant concentration–response evidence can therefore help identify an exposure at which a specified response may begin, or at which a chosen response level is expected. For first-in-human planning, that exposure is a useful point of comparison: it helps describe what predicted human exposure might mean pharmacologically.

An exposure target is not itself a recommended starting dose. A dose also depends on the predicted human pharmacokinetics, route, and the safety constraints established from nonclinical evidence. This tutorial focuses on defining and interpreting the pharmacology-based exposure target, not on deriving a final dose or a mechanistic minimum anticipated biological effect level.

Concentration–Response Model and Assumptions

For a response that increases monotonically with concentration and approaches a maximum, a standard Emax model is

Here \(C\) is the concentration used in the analysis, \(E_0\) is the response without drug, \(E_{\max}\) is the maximum increase above baseline, \(EC_{50}\) is the concentration producing half of that maximum increase, and \(h\) is the Hill coefficient controlling the curve’s steepness. For an inhibitory effect, the response can instead be expressed as the magnitude of inhibition, with the same fractional-response form. The meaning of \(C\) must be explicit: it might be measured total concentration or unbound concentration, but the assay and the projected human exposure must use compatible concentration definitions.

Using this model to choose a target assumes that the measured concentration represents the concentration relevant to the response, that the response is adequately described by a monotone Emax curve over the range of interest, and that the assay conditions provide a meaningful basis for human pharmacology. It also assumes that the selected effect level has a clear interpretation for the intended decision. If the response is time-dependent, strongly affected by delayed turnover, or not monotone, a single concentration–response curve may not be enough to define an exposure target.

Solving for the concentration corresponding to a chosen fractional response gives

This follows by multiplying both sides by the denominator, rearranging to obtain \(fEC_{50}^{\,h}=C^h(1-f)\), and taking the positive \(h\)th root. For example, with \(h=1\), \(EC_{10}=EC_{50}/9\). This is a property of the fitted model, not a universal biological rule: a different Hill coefficient produces a different ratio.

Choosing the Response Level and Concentration Basis

There is no single low-effect fraction that is suitable for every drug or decision. A target could be defined at a prespecified fraction of the modeled response, or at a concentration corresponding to a separately justified pharmacological threshold. The rationale should explain why that response level is informative for first-in-human planning. In particular, “low” is not a substitute for “safe”: a small response may matter for a sensitive biological system, whereas a larger response may be tolerable in another context.

The concentration basis can materially change the numerical target. If an assay’s response is linked to unbound concentration, but projected human exposure is reported as total plasma concentration, the two should not be compared directly. A conversion using the unbound fraction may be reasonable only if the relevant binding assumptions are supported and the assay and human matrix are sufficiently comparable. An unsupported conversion creates precision without evidence.

The concentration–response model also does not decide whether the relevant human exposure metric is a peak, an average, or an integrated exposure. That depends on the pharmacological response and its time course. If a transient peak drives the response, comparing a modeled concentration threshold with predicted \(C_{\max}\) may be appropriate. If the response depends on sustained exposure, an average concentration, AUC, or time above a concentration may be more informative. The exposure metric should be selected consistently with the pharmacology and the dose-limiting exposure rationale, not merely because it is readily available.

Uncertainty and Conservative Interpretation

A point estimate of potency can understate uncertainty. Assay variability, parameter uncertainty, differences in biological context, and uncertainty in translating an assay concentration to a human exposure can all affect the target. A conservative analysis should identify the uncertainty that matters to the decision and show how it changes the target. For a decision concerned with avoiding pharmacological activity, a lower plausible threshold concentration can be more cautious than using the fitted point estimate. That does not mean every uncertainty should be combined mechanically into one arbitrary safety factor.

Relating the Target to Projected Human Exposure

Once the pharmacology target is defined, it can be compared with predicted human exposure. A simple dose-to-exposure relationship may be used for illustration when the relevant pharmacokinetic prediction is available and its assumptions are appropriate. For a proportional relationship between dose and the selected exposure metric, an exposure at a dose can be estimated by multiplying the dose by the predicted exposure per unit dose. This comparison helps show whether a proposed dose range may approach a pharmacology target; it does not replace safety-based starting-dose selection.

Diagnostics, Alternatives, and Limitations

Before using a concentration target in a decision, check whether the fitted curve and its inputs support the interpretation. Inspect whether the observed concentrations span the response transition and approach enough of the maximum to estimate \(EC_{50}\) and \(h\) reliably. Compare model predictions with the observed data, and investigate systematic departures, influential observations, or a response that does not approach a plateau. A numerically precise \(EC_{10}\) can still be unreliable if it is extrapolated far below the measured concentration range.

• Concentration mismatch: comparing total human plasma concentration with an unbound assay potency without a supported conversion can misstate the exposure margin.
• Time-course mismatch: a static concentration–response model may not describe delayed, cumulative, or transient effects. Consider whether a concentration at one time represents the relevant exposure.
• Uncertain curve shape: a poorly estimated Hill coefficient can change low-effect concentrations substantially. Show sensitivity to plausible values rather than treating one fitted slope as exact.
• Overinterpreting low-effect levels: an \(EC_{10}\) is a model-defined response level, not a universal no-effect concentration or a guarantee of safety.
• Unrepresentative pharmacology: if the assay context is not relevant to the intended human interpretation, a mathematically good fit does not make its potency estimate applicable.

A pharmacology-based target complements, rather than replaces, a safety-based approach. NOAEL- and HNSTD-based evidence and the associated human exposure comparisons address different questions from a concentration–response target. A target inferred from human-relevant pharmacology can help interpret predicted human exposure, but it does not remove the need to evaluate nonclinical safety, pharmacokinetic uncertainty, and the rationale for the actual starting dose. Mechanistic evidence and specialized potency-to-MABEL procedures require additional reasoning beyond the scope here.