Matching Exposure Metrics Across Model Components

Why Exposure-Metric Definitions Must Match

A population PK model may report exposure for each individual, an exposure-response model may use a summary as its predictor, and a trial simulation may generate that summary under a dosing regimen. The handoff is meaningful only if the models refer to the same exposure quantity. AUC over one dosing interval, AUC over a calendar day, and AUC from time zero to infinity are not interchangeable merely because each is called “AUC.”

Matching metrics is a definition and alignment task: specify what concentration is summarized, over what time window, under which dosing conditions, and relative to what time origin. This tutorial focuses on those choices. It assumes that the concentration-time profiles and model outputs have already been generated; it does not treat uncertainty in estimating exposure from sparse observations.

Specify the Metric Before Passing It Between Models

Let \(C(t)\) denote the concentration of a specified analyte at time \(t\), with units such as mg/L. For a defined interval from \(t_0\) to \(t_1\), the area under the concentration-time curve is

If concentration is in mg/L and time is in hours, AUC is in mg·h/L. The subscripts matter: \(\mathrm{AUC}_{0-24}\) means the area from hour 0 to hour 24, whereas \(\mathrm{AUC}_{0-\tau}\) means the area over a dose interval of length \(\tau\), provided hour 0 is defined as the relevant dose time. For repeated dosing at steady state, \(\mathrm{AUC}_{\tau}\) is often used for the area over one steady-state dosing interval. It must not be assumed to mean the same window for different values of \(\tau\).

The maximum concentration in a specified interval is \(C_{\max}=\max_{t_0\leq t\leq t_1} C(t)\). Its value depends on the interval and profile; the time of that maximum, \(T_{\max}\), is a different quantity. Trough concentration usually means the concentration immediately before the next dose. At steady state with dosing interval \(\tau\), it is commonly denoted \(C_{\min,\mathrm{ss}}\) or described as the pre-dose concentration at \(t=\tau^{-}\). A concentration measured at a scheduled visit is not automatically a trough: its sampling time must actually correspond to the pre-dose time.

Before connecting components, record at least: analyte and concentration type; units; dose and dosing interval; time origin; integration or evaluation window; steady-state status; and whether the metric is individual-level or population-level. These details determine whether a numerical value can be passed without changing its meaning.

Why Window Length Changes the Meaning of AUC

A useful derived quantity is average concentration over a specified window:

This follows directly from the definition of an average: the constant concentration that would yield the same area over a window of length \(t_1-t_0\) is the area divided by that length. Its units are concentration, unlike AUC. Dividing by the window length can make averages interpretable across different interval lengths, but it does not make two profiles equivalent: profiles with the same average can have different peaks, troughs, and timing.

For a periodic steady-state profile with dose interval \(\tau\), the area in each full interval is the same, so the average concentration over any such interval is \(\mathrm{AUC}_{\tau}/\tau\). That conclusion requires the same steady-state periodic regimen and a complete dosing interval. It does not justify treating a partial interval, a non-steady-state window, or a calendar-day area that cuts across differently timed doses as the same metric.

Align the Time Window, Not Just the Metric Name

A particularly consequential mismatch occurs when one component uses a dose-interval window and another uses a fixed clock window. Consider a regimen with doses every 12 hours. If a model reports exposure separately over the first and second 12-hour periods after a reference dose, those areas need not be equal. Their sum is the area over the full 24-hour period, but neither half-period area alone is the 24-hour AUC.

The same care applies to windows spanning dose changes, missed doses, or a change in dose interval. “Daily AUC” can mean a fixed 24-hour calendar window, while “AUC over one dosing interval” means a window whose duration is \(\tau\). They coincide only when the interval and reference times make them coincide.

Check Regimen and Reference Time for Cmax and Trough

Cmax and trough are especially sensitive to when the interval begins and ends. Under repeated dosing, the steady-state peak and pre-dose minimum describe a recurring profile for a particular regimen. If the regimen changes, the profile and its extrema may change—even when the analyte and units stay the same. A peak defined over a whole day may also differ from a peak defined within a single dose interval if the day contains multiple doses.

A Practical Alignment Procedure and Its Failure Modes

A reliable handoff makes each metric explicit before fitting or simulating the receiving model. The following sequence is useful when reviewing component specifications and output tables.

Warnings include unexplained changes in units; AUC values with no endpoints; “trough” values recorded at variable times after dosing; peaks reported from different windows; and differences between model outputs and receiving-model inputs that cannot be reconciled by a stated conversion. Such discrepancies may indicate a definition mismatch rather than a substantive pharmacological result.

Choosing Among Metrics and Recognizing Their Limits

AUC summarizes cumulative concentration over a window; average concentration rescales that area by window duration; Cmax describes the largest concentration in a defined interval; and trough describes a pre-dose concentration under a defined regimen. These summaries emphasize different features. Choosing one to connect models should reflect the quantity actually defined in the components, not a belief that one metric is universally superior. If a response relationship was specified using AUC, substituting Cmax changes the predictor and its interpretation.

Metrics also discard information. Two profiles can share the same AUC but differ in peak, trough, and timing; two profiles can share Cmax but differ in total exposure. Even a fully matched metric does not establish that one model component is clinically adequate or that the exposure-response relationship is causal. It ensures only that the quantity handed across components has a consistent definition.

A common mistake is to treat “AUC” as a universal, regimen-independent number. For example, reasoning that two values are directly comparable because both have units of mg·h/L ignores differences in window length and dose schedule. The correction is to inspect the time subscripts and regimen first. If the windows differ, report the distinct AUCs or, when scientifically appropriate, calculate average concentration over each explicitly stated window; do not silently relabel one metric as another.

When reporting an integrated analysis, include the metric definition alongside the numerical result. A concise specification such as “steady-state total-plasma \(\mathrm{AUC}_{0-12}\), one 12-hour interval after the scheduled dose, mg·h/L” is much more transferable than “AUC.” Matching definitions is a necessary condition for coherent model integration, but not a substitute for checking that the models are suitable for their intended uses.