Steady-State Bioequivalence for Multiple-Dose Products

Why Assess Bioequivalence at Steady State?

A repeated-dose product is often intended to be used over many dosing intervals. Its concentration profile after a single dose may not fully describe exposure during ongoing treatment: concentrations can accumulate, and both peak and trough concentrations may differ from those after the first dose. A steady-state comparison therefore evaluates exposure over a dosing interval during repeated administration, rather than assuming that a single-dose profile answers the same question.

This analysis focuses on two measures: the area under the concentration-time curve over one dosing interval, \(AUC_{\tau,ss}\), and the maximum concentration during that interval, \(C_{\max,ss}\). Here, \(\tau\) is the time between scheduled doses, and the subscript \(ss\) indicates steady state. For example, if the regimen is one dose every 12 hours, then \(\tau=12\) hours. The interval length is determined by the regimen; it is not automatically 24 hours.

In a typical comparison, participants receive test and reference products according to a randomized crossover design, with enough repeated dosing in each treatment period to characterize the steady-state profile. The treatment comparison is made for each measure separately. A favorable result for AUC alone does not establish bioequivalence if the confidence interval for \(C_{\max,ss}\) does not meet the criterion.

Calculating the Interval Measures

The profile should include concentration measurements across the entire interval, including its start and end. A common calculation for \(AUC_{\tau,ss}\) is the linear trapezoidal rule. For successive times \(t_i\) and \(t_{i+1}\), with measured concentrations \(C_i\) and \(C_{i+1}\), the area for that segment is the average of the endpoint concentrations multiplied by the elapsed time. The segment areas are summed. \(C_{\max,ss}\) is the largest observed concentration in the interval, with its observed time recorded as \(T_{\max,ss}\) if relevant.

Profile measurement and study design matter. The participant should have received the assigned regimen long enough for the profile to be representative of steady state; a fixed number of doses is not a universal substitute for considering the drug’s disposition and the observed concentration pattern. Repeated predose concentrations can help assess whether concentrations have stabilized. Sampling that misses the peak can underestimate \(C_{\max,ss}\), while missing or poorly characterizing the interval endpoints can affect the area estimate.

Model and Bioequivalence Decision

For a standard two-treatment, two-period crossover, analyze each endpoint on the natural-log scale. A common model includes treatment, period, and sequence as fixed effects, and participant nested within sequence as a random effect. The treatment contrast is the estimated mean log value for test minus the estimated mean log value for reference. Its standard error and degrees of freedom come from the fitted crossover model.

This analysis assumes the randomized crossover was conducted as intended, each participant received both treatments, and washout was sufficient to avoid carryover that would materially affect the next period. It also assumes that the log-transformed endpoint is suitable for the model: residuals are reasonably modeled by the assumed distribution, and the model’s variance structure is adequate. Observations from different participants are independent. These conditions should be considered using the design and study data; a model result cannot repair a deficient washout, unreliable concentration measurements, or a systematically incomplete profile.

The interval is formed on the log scale because exposure measures are positive and treatment effects are naturally interpreted as ratios. The following calculation shows why the ratio-scale interval is obtained by exponentiation: if the log-scale contrast lies from \(L\) to \(U\), then applying the increasing exponential function gives an interval from \(e^L\) to \(e^U\). The point estimate transforms in the same way, from \(\Delta\) to \(e^\Delta\).

Reading the Result in the Context of Repeated Dosing

At steady state, \(AUC_{\tau,ss}\) summarizes exposure over each dosing interval, while \(C_{\max,ss}\) summarizes the observed peak within that interval. When dose and interval are the same for test and reference, these endpoints address complementary aspects of the repeated-dose concentration profile. Under linear pharmacokinetics, interval AUC is related to bioavailability, dose, and clearance; it is not a direct measure of peak or trough. A similar interval AUC can coexist with a different peak shape, which is one reason both endpoints are considered.

A standard alternative is a single-dose bioequivalence comparison. It may be appropriate where the study question and product support that approach, but it measures a different dosing condition. A single-dose result should not be described as a direct measurement of steady-state peak and interval exposure. Conversely, a steady-state result applies to the regimen, conditions, and endpoints studied; it does not automatically establish equivalence for every dose, interval, patient group, or clinical outcome.

Diagnostics, Limitations, and Common Pitfalls

Before interpreting the intervals, examine whether the measurements and design support the analysis. Useful checks include whether the profile covers the full interval, whether the terminal measurement is consistent with the next predose measurement at steady state, and whether predose concentrations suggest stabilization. Review the distributions of log-transformed endpoints and model residuals for marked departures from the assumptions. Check for missing endpoint measurements and for protocol or timing deviations that could change the calculated interval measures. A single unexpectedly low peak may reflect sampling that missed the actual peak rather than a genuinely lower maximum.

These checks are diagnostic, not a license to choose whichever analysis yields a preferred conclusion. If a key assumption is doubtful, describe the issue and consider whether the primary estimate remains interpretable. The conventional limits used here are not universal for every product or regulatory setting; the applicable limits and analysis should be specified for the intended assessment.

• Using the wrong area. \(AUC_{\tau,ss}\) is the area over one dosing interval at steady state, not \(AUC_{0-\infty}\) after a single dose.
• Calling an in-range point estimate bioequivalent. The full confidence interval, not only the estimated ratio, must lie within the limits.
• Concluding from only one endpoint. AUC and peak concentration each need to meet the specified criterion for the two-endpoint conclusion.
• Treating steady state as a label rather than a condition. A profile collected before concentrations have stabilized may not represent repeated-dose exposure at steady state.

A common mistake is to reason, “The test/reference ratio is 1.19, which is below 1.25, so the products are bioequivalent.” This ignores uncertainty. In the example above, the confidence interval reaches 131%, outside the upper limit, so the endpoint fails the criterion. The correction is to calculate and interpret the full confidence interval for each endpoint and then state exactly which criteria were met.