Why the Exposure Metric Must Match the Question
A concentration-time profile contains more information than any single exposure summary can retain. One question may concern concentration when an effect is measured; another may concern the total exposure over a treatment interval, the highest concentration reached, or the duration for which concentration exceeds a relevant threshold. Choosing a metric is therefore part of specifying the pharmacologic question, not a neutral calculation performed after the data are collected.
The earlier material on exposure-effect questions and exposure windows establishes the need to align exposure and effect in time. Here, the practical task is to decide which feature of the profile represents the proposed pharmacologic driver. No metric is best for every drug, effect, or endpoint. A metric can be calculated correctly and still be a poor answer to the question if it summarizes the wrong feature of exposure.
A Practical Matching Framework
Begin by stating what feature of exposure is hypothesized to matter and when the effect is measured. Then choose the metric that represents that feature over the relevant window. The table gives a concise guide; it is a guide to matching questions and summaries, not evidence that a particular drug follows the suggested mechanism.
| Pharmacologic question | Candidate metric | What it summarizes |
|---|---|---|
| What concentration coincides with an effect measured at time \(t^*\)? | \(C(t^*)\) | Concentration at that specified time |
| Does total exposure over a defined interval relate to the effect? | AUC over that interval | Concentration accumulated with time |
| Does average concentration over the interval relate to the effect? | \(C_{\text{avg}}\) | AUC divided by the interval duration |
| Could the highest exposure be relevant to an effect or safety question? | \(C_{\max}\) | Maximum concentration in the defined window |
| Could the lowest concentration before a dose relate to loss of coverage? | \(C_{\min}\), often the predose trough | Minimum concentration in the defined window or dosing interval |
| Does the duration at or above a pharmacologically justified threshold matter? | Time above threshold | Time for which \(C(t)\) exceeds the specified concentration |
AUC and average concentration are related but not interchangeable in every comparison. For a window of duration \(T=t_1-t_0\), average concentration is AUC divided by \(T\). If all subjects have the same window duration, the two metrics differ only by a constant scaling. If durations differ, the same AUC can correspond to different average concentrations, and AUC may also reflect differences in how long the window spans.
Before selecting any summary, check that the exposure window corresponds to the effect window, the concentration scale and units are consistent, and the profile is observed densely enough to estimate the chosen feature. If concentration measurements are sparse, for example, a measured maximum may miss the true peak. If the effect reflects prior exposure or delayed response, concentration at the observation time may not capture the exposure relevant to that effect.
Worked Example: Matching Concentration at a Time to a Timed Effect
In a hypothetical study, an effect is measured at 8 hours after a dose. Two participants have the following estimated concentrations at 8 hours: participant A, \(C(8)=1.8\) mg/L; participant B, \(C(8)=3.1\) mg/L. The question is whether the effect measured at 8 hours varies with the concentration present at that same time.
The directly matched exposure metric is \(C(8)\), because both exposure and effect refer to the 8-hour time point. The contrast is
Thus participant B's measured 8-hour concentration is 1.3 mg/L higher than participant A's. This comparison identifies the concentration contrast relevant to the stated question; it does not establish that concentration caused any difference in effect. If the question instead concerned total exposure over the first 8 hours, \(C(8)\) alone would not answer it, and an AUC over that interval would be a more appropriate candidate.
Worked Profiles: Cumulative Exposure and Threshold Duration
Calculations from a profile make clear why the metric must be chosen before interpreting a result. The following hypothetical examples use linear interpolation between measured concentrations to approximate the profile between sampling times. That interpolation is a calculation assumption, not a claim that real concentrations always change linearly.
Worked Example: Choosing AUC or Average Concentration
A hypothetical concentration profile is measured at 0, 2, 4, and 8 hours, with concentrations 0, 4, 2, and 1 mg/L, respectively. The pharmacologic question concerns cumulative exposure from 0 to 8 hours. Estimate AUC using the trapezoidal rule.
For each interval, the trapezoidal area is the interval duration multiplied by the mean of its endpoint concentrations:
The interval duration is \(8-0=8\) hours, so the average concentration is
AUC is the direct candidate for the question about cumulative exposure. Average concentration is also interpretable, but answers a question about exposure averaged over the 8-hour window. If two participants have different window durations, reporting only AUC can obscure that difference in duration; report the window and consider whether average concentration better matches the scientific comparison.
Worked Example: Selecting Time Above a Threshold
Use the same hypothetical profile: concentrations are 0, 4, 2, and 1 mg/L at 0, 2, 4, and 8 hours. Suppose, for illustration, the pharmacologic question concerns time above a justified threshold of 2 mg/L. Estimate the duration above the threshold, assuming linear interpolation between samples.
On the rising segment from 0 to 2 hours, concentration increases from 0 to 4 mg/L. It reaches 2 mg/L halfway through that segment:
On the next segment, concentration decreases from 4 mg/L at 2 hours to 2 mg/L at 4 hours. It equals the threshold at 4 hours. Therefore it is above the threshold from 1 hour until just before 4 hours, for an estimated duration of
This estimate is appropriate only if the threshold has pharmacologic meaning for the question and the interpolation is adequate. It would not be justified to choose 2 mg/L simply because it produces an appealing duration or an observed association.
Worked Example: When Peak and Trough Answer Different Questions
Two hypothetical dosing profiles have the following estimated values within the same dosing interval: profile A has \(C_{\max}=8\) mg/L and a predose trough of 1 mg/L; profile B has \(C_{\max}=5\) mg/L and a predose trough of 3 mg/L. Suppose one question concerns whether a transient high concentration coincides with a short-lived adverse effect, while another concerns whether concentration is maintained before the next dose.
For the first question, the peak contrast is
Profile A has the higher estimated peak. For the second question, the trough contrast is
Profile B has the higher predose trough. Neither comparison shows that either effect actually occurs; the profiles only identify which summary aligns with each question. If the concern is total exposure, neither peak nor trough alone is sufficient to describe AUC or the full profile.
Justification, Diagnostics, and Alternatives
The relationship between AUC and average concentration follows directly from the definition of a time average. For a concentration profile over a finite window of duration \(T\), the time average is the integral of concentration divided by the length of the interval. Since the integral is AUC, the average is AUC divided by \(T\). This also shows why the two summaries rank profiles identically when every window has the same duration, but need not do so when durations vary.
Diagnose a metric choice by checking the assumptions behind its use and whether the observed data can support its calculation. Sparse sampling can miss a true peak, obscure a threshold crossing, or make trapezoidal AUC sensitive to interpolation. A concentration recorded at a nominal time may be poorly aligned with the actual effect measurement. A threshold metric is especially sensitive when concentrations lie near the threshold: small measurement errors can shift the estimated crossing time. AUC also depends on the specified window; changing its start or end changes what is accumulated.
The main alternatives are not competing calculations of one identical quantity. They summarize different profile features. \(C(t^*)\) is local in time; AUC summarizes concentration accumulated across a window; average concentration rescales that accumulation by duration; peak and trough describe extremes; time above threshold describes duration relative to a specified cutoff. Compare alternatives only when the pharmacologic question plausibly allows more than one feature. If multiple metrics are explored, distinguish the metric specified by the scientific rationale from additional exploratory summaries.
Common Mistake
Wrong reasoning: “AUC is widely used, so it is the right exposure metric for every effect.” This treats a conventional summary as though it answered every pharmacologic question. If an effect is measured at one specified time, a contemporaneous concentration may better match the question; if duration above a justified threshold matters, AUC alone does not report that duration.
Correction: State the pharmacologic question first, including the effect timing and exposure window. Select the metric that represents the exposure feature implicated by that question, and explain the calculation assumptions. If the mechanism is uncertain, compare scientifically plausible summaries transparently rather than selecting whichever one gives the strongest observed relationship.
Check Your Understanding
Answer each question, then compare your reasoning with the solutions.
- An effect is measured at 6 hours after dosing, and the question concerns concentration present at that measurement. Which metric is the direct candidate?
- Over a 12-hour window, AUC is 30 mg·h/L. Calculate average concentration and state its units.
- A pharmacologic question concerns duration above a validated concentration threshold. Why is peak concentration alone insufficient?
- Two profiles have equal AUC, but one has a higher peak and lower trough. Does equal AUC establish that they have the same exposure profile?
- Why can sparse sampling undermine an estimated time above threshold?
Solutions.
- Use \(C(6)\), the concentration at 6 hours, because the metric is aligned with the time of the effect measurement.
- \(C_{\text{avg}}=30\ \text{mg}\cdot\text{h/L}/12\ \text{h}=2.50\ \text{mg/L}\).
- A peak gives the maximum concentration but not how long concentration remains above the threshold. The duration requires locating the threshold crossings over the profile.
- No. Equal AUC means equal accumulated concentration over the specified window, but profiles can differ in peak, trough, and the timing of concentrations.
- With few samples, a crossing may occur between widely spaced observations, and interpolation may misestimate its timing. The resulting duration can therefore be inaccurate, especially when concentrations are near the threshold.