Start With the Exposure Question
A concentration–time profile can be summarized in several defensible ways, but those summaries do not answer the same question. AUC describes accumulated exposure over a stated interval; Cmax describes the highest concentration recorded; Tmax describes when that recorded maximum occurs; and partial AUC focuses on exposure within a specified time window. Choosing among them is therefore not a matter of finding one universally best metric. It is a matter of deciding which feature of exposure is relevant to the question.
That distinction matters when two profiles cross, when one reaches its peak later, or when a short-lived high concentration occurs. In each situation, an overall summary can conceal a feature that another metric makes visible. Conversely, focusing on a peak or a selected time window can omit exposure elsewhere in the profile. A useful diagnostic begins by stating the question in words, then checking whether the metric and its time interval actually represent that question.
These definitions distinguish observed summaries from features of an unobserved, continuous profile. In particular, observed Cmax is the largest recorded value, not necessarily the true continuous peak between samples. Observed Tmax is likewise limited to the sampling times: it identifies when the recorded maximum occurred, not necessarily the exact time of the underlying peak. For tied observed maxima, report the tie or use a stated convention rather than implying a uniquely determined peak time.
Before comparing profiles, check that the question and data make the comparison meaningful. The profiles should concern the same analyte and use compatible concentration and time units, a common time origin, and comparable dose and sampling conditions. State the interval used for total AUC and the bounds of any partial AUC. The sampling schedule should be adequate for the feature being assessed. These conditions do not guarantee that a metric answers the question; they prevent avoidable differences in measurement or definition from being mistaken for differences in exposure.
A Practical Selection Rule
Translate the question into the part of the profile it concerns. If it asks about accumulated exposure over the full interval of interest, total AUC is the direct summary. If it asks about exposure within a clinically or scientifically relevant time window, partial AUC is more targeted. If it asks about the height of the observed peak, use Cmax. If it asks when that peak occurs, use Tmax. More than one metric may be useful when the question has more than one component, but each should have a stated purpose.
| Exposure question | Metric to consider | What the metric does not establish by itself |
|---|---|---|
| How much exposure accumulated over the declared overall interval? | Total AUC | Whether exposure was concentrated early, late, or near the peak |
| How much exposure occurred during a specified interval? | Partial AUC for that interval | What happened outside the selected window |
| What was the highest sampled concentration? | Cmax | Overall exposure or the duration of elevated concentrations |
| At what sampled time did the maximum occur? | Tmax | Peak height, overall exposure, or the exact continuous peak time |
A short justification should therefore name both the question and the metric: for example, “Use partial AUC from 0 to 2 hours because the question concerns exposure during the first two hours.” Naming a metric without identifying its relevant time interval, where applicable, leaves the decision incomplete.
Worked Example: Crossing Profiles and a Time-Window Question
Hypothetical data. Two profiles are measured in mg/L at the same times after dosing. The question is whether exposure is greater during the first four hours, not which profile has greater exposure over the full observation interval.
| Time (h) | Profile A (mg/L) | Profile B (mg/L) |
|---|---|---|
| 0 | 0 | 0 |
| 2 | 8 | 3 |
| 4 | 10 | 8 |
| 6 | 4 | 7 |
| 8 | 1 | 1 |
The profiles cross between 4 and 6 hours: Profile A is higher at 4 hours, but Profile B is higher at 6 hours. Thus, neither profile is uniformly higher over the whole interval. For this illustrative calculation, use the trapezoidal area between adjacent samples: interval area equals the time width multiplied by the average of the two endpoint concentrations.
For the requested partial AUC from 0 to 4 hours, Profile A gives \(2(0+8)/2 + 2(8+10)/2 = 8 + 18 = 26\) mg·h/L. Profile B gives \(2(0+3)/2 + 2(3+8)/2 = 3 + 11 = 14\) mg·h/L. Under this calculation, Profile A has the greater exposure in the specified early window.
For context, the corresponding total AUCs from 0 to 8 hours are \(8+18+14+5=45\) mg·h/L for A and \(3+11+15+8=37\) mg·h/L for B. The full-interval result also favors A here, but that agreement is not guaranteed in crossing profiles. The metric choice follows the question about early exposure; it is not justified merely because the total AUC happens to point in the same direction.
Justification. The common sampling times and units support a direct descriptive comparison, and the requested window matches the stated 0-to-4-hour partial AUC. If the scientific question instead concerned the whole measured interval, total AUC would be the more direct summary. Cmax or Tmax would not answer either accumulated-exposure question.
Delayed Peaks: Separate Peak Height From Peak Timing
A delayed peak is a reason to distinguish Cmax from Tmax, not a reason to treat them as interchangeable. Two profiles can have the same observed Cmax but reach it at different times; they can also have different peak heights but the same observed Tmax. AUC adds a different perspective because it summarizes area rather than peak location. If the timing of exposure matters, an overall AUC alone may fail to show the delay.
Worked Example: Equal Peak Height, Different Peak Time
Hypothetical data. Concentrations are measured in mg/L after dosing, at the same sampling times for both profiles. Suppose the question asks whether the observed peak is delayed, while also checking whether peak heights differ.
| Time (h) | Profile A (mg/L) | Profile B (mg/L) |
|---|---|---|
| 0 | 0 | 0 |
| 1 | 4 | 2 |
| 2 | 10 | 6 |
| 4 | 5 | 10 |
| 6 | 1 | 4 |
The largest observed concentration is 10 mg/L for each profile, so both have \(Cmax=10\) mg/L. Profile A reaches that observed maximum at 2 hours, so \(Tmax=2\) h; Profile B reaches it at 4 hours, so \(Tmax=4\) h. The observed peak height is the same, but the observed peak time for B is 2 hours later.
An AUC calculation provides context but answers a different question. Using trapezoids, Profile A has \(1(0+4)/2 + 1(4+10)/2 + 2(10+5)/2 + 2(5+1)/2 = 2+7+15+6=30\) mg·h/L from 0 to 6 hours. Profile B has \(1+4+16+14=35\) mg·h/L over the same interval. The total area does not negate the timing difference; it summarizes accumulated exposure over 0 to 6 hours.
Justification. The profiles use the same sampled times, so their observed maxima and peak times are directly comparable. The conclusion is about the sampled profiles only: a denser schedule could locate a peak between these times or change the observed Cmax and Tmax. Since the question is about delay, Tmax is essential; Cmax is useful as a separate description of peak height.
Brief High Concentrations: What Cmax Can and Cannot Show
A brief high concentration can dominate Cmax while contributing relatively little to total AUC. This is not a defect in Cmax: it is what makes Cmax useful for questions about peak magnitude. The limitation arises when a peak summary is used to imply something about cumulative exposure or the duration of elevated concentrations. Conversely, a similar AUC does not show that peak concentrations are similar.
Sampling is especially important here. A narrow peak that falls between sampling times may not be recorded, causing observed Cmax to understate the actual peak and observed Tmax to mislocate its timing. A sudden isolated high value may also warrant checking assay and data-quality information. Neither situation is resolved simply by switching to total AUC: each metric has its own target, and the adequacy of the data must be assessed against that target.
Worked Example: Similar Total Area, Different Observed Peaks
Hypothetical data. Consider two profiles sampled hourly. Concentrations are in mg/L, and the question is whether one has a higher observed peak, while also asking whether its total measured-interval exposure is much larger.
| Time (h) | Profile A (mg/L) | Profile B (mg/L) |
|---|---|---|
| 0 | 0 | 0 |
| 1 | 2 | 6 |
| 2 | 20 | 8 |
| 3 | 2 | 6 |
| 4 | 0 | 2 |
| 5 | 0 | 0 |
The observed peaks are \(Cmax_A=20\) mg/L at \(Tmax_A=2\) h and \(Cmax_B=8\) mg/L at \(Tmax_B=2\) h. Thus, A has the higher observed peak, while both observed maxima occur at the same sampled time.
The trapezoidal total AUCs from 0 to 5 hours are \(1+11+11+1+0=24\) mg·h/L for A and \(3+7+7+4+1=22\) mg·h/L for B. The total areas are relatively close in these hypothetical data even though the observed Cmax values differ substantially. This illustrates why Cmax is appropriate for a peak-height question, while AUC is appropriate for the accumulated exposure question.
Justification. The hourly schedule records the stated observed peak, but it cannot establish the exact height or duration of a peak that may occur between samples. If the question specifically concerns a very brief concentration excursion, the schedule must be sufficiently dense around the expected peak. The reported Cmax is a property of these sampled observations, not proof that the continuous concentration never exceeded it.
Why Partial AUC Depends on the Window
Partial AUC can focus the analysis on early, late, or otherwise defined exposure, but the bounds are part of the meaning of the result. Changing \(t_1\) or \(t_2\) changes the quantity being summarized. A window chosen after inspecting which interval favors a particular profile risks turning a focused metric into a selective one. The interval should be justified by the exposure question and specified consistently for the profiles being compared.
The basic reason crossing profiles make window choice consequential follows from additivity of area. Suppose two profiles have areas \(A_1\) and \(A_2\) over an early interval, and areas \(B_1\) and \(B_2\) over a later interval. Then the full-interval difference is
If Profile A has greater early area, then \(A_1-A_2>0\). If Profile B has greater later area, then \(B_1-B_2<0\). The total difference combines these contributions, which can reinforce, reduce, or reverse the early-window difference. Therefore, a total AUC cannot be assumed to represent which profile is higher in a particular interval. This is why partial AUC can be informative when the question is explicitly time-specific, and why the bounds must be defensible.
Diagnostics, Alternatives, and Limits
Before settling on a metric, inspect the plotted profiles and ask whether the data support the feature the metric is meant to summarize. A profile plot can reveal crossings, delayed peaks, sparse sampling near a peak, or an isolated high value. Check units, time origin, common analysis windows, and the stated handling of the profile’s beginning and end. If a total AUC includes an extrapolated component, make clear that it is not simply area over the observed samples.
A choice is suspect when the metric does not match the question, when a partial-AUC window has no scientific justification, or when the sampling schedule is too sparse for a peak-related interpretation. It is also suspect to compare observed maxima collected on substantially different schedules as if they were equally capable of capturing the true peak. In these cases, revise the interpretation, improve the sampling or clarify the question; a different label for the same inadequate data does not solve the problem.
The main alternatives are complementary, not substitutes for every purpose. Total AUC is broad and cumulative but can hide when exposure occurred. Partial AUC is targeted but omits what falls outside its bounds. Cmax is direct for the highest recorded concentration but says little about duration or accumulated exposure. Tmax describes the timing of the recorded maximum but not its magnitude, and its precision is constrained by the sampling times. A concentration–time plot gives visual context, but it does not replace a numerical summary when a prespecified summary is needed.
Finally, a descriptive exposure metric is not a complete account of therapeutic similarity. AUC, Cmax, and Tmax summarize specified aspects of concentration profiles; choosing among them does not by itself establish a clinical conclusion. Keep the claim proportional to the metric: report what the summary shows, the interval and sampling basis for it, and what it does not show.
Check Your Understanding
For each question, identify the most directly relevant metric and give a brief justification. Solutions follow the questions.
- A study question concerns accumulated exposure from 0 to 12 hours. Which metric is the natural starting point, and what interval must be stated?
- Two profiles cross, and the question concerns exposure from 0 to 3 hours. Why might a total AUC alone be insufficient?
- Two profiles have the same observed Cmax, but one reaches it at a later sampled time. Which metric describes that timing difference?
- A narrow concentration peak may fall between widely spaced samples. What limitation should be stated when reporting observed Cmax and Tmax?
- Why is it incomplete to report a partial AUC without giving its time-window bounds?
Solutions. (1) Use total AUC over 0 to 12 hours; stating the interval defines the accumulated exposure being summarized. (2) Total AUC combines exposure across the full interval, so area outside 0 to 3 hours can offset or reverse the early-window difference; a justified 0-to-3-hour partial AUC directly addresses the question. (3) Tmax describes the sampled time of the observed maximum; Cmax describes its height. (4) The actual peak may occur between samples, so observed Cmax may miss its height and observed Tmax may not identify its exact time. (5) The bounds define which exposure is included, and changing them changes the metric’s meaning.