Why Compare Interaction Ratios Across Doses?
A drug interaction is often summarized by an exposure ratio: exposure with a perpetrator drug divided by exposure in a matched control condition. That summary can conceal an important feature. If the ratio at a low probe dose differs from the ratio at a high probe dose, the interaction is not dose-invariant over the tested range for that endpoint. This pattern can arise from nonlinear disposition, dose-dependent bioavailability, changing contributions of parallel clearance pathways, or features of the study and exposure estimate.
The mismatch is a diagnostic signal, not a mechanism. A larger ratio at low dose does not, by itself, prove enzyme saturation; nor does a ratio that approaches 1 at high dose establish that the interaction disappears. The interpretation requires checking how the endpoint was measured, whether the conditions are comparable, and what changes in the PBPK model could generate the observed pattern.
The comparison is most interpretable when dose, formulation, sampling window, food conditions, and perpetrator regimen are matched within each dose pair. If the drug is studied at steady state, compare AUC over the same dosing interval; for single-dose studies, use the same defined AUC endpoint, such as AUC to infinity when it is estimable. The two dose levels need not be the same as each other, but each dose must have its own valid control and perpetrator comparison.
What a Dose-Specific Ratio Represents
For a drug with linear disposition over the relevant range, AUC after a dose can be written as \(F D/CL\), where \(F\) is the fraction of dose reaching systemic circulation and \(CL\) is systemic clearance. At steady state, the corresponding interval AUC has the same dose-to-clearance relationship when the absorbed systemic dose is \(F D\). For an oral drug, therefore, the ratio can reflect both a change in bioavailability and a change in clearance:
This expression is an organizing relation, not a guarantee that clearance and bioavailability are constant. If either depends on dose or concentration, the ratio can change across doses. For an intravenous dose, \(F=1\), so the bioavailability term is absent; however, nonlinear systemic disposition can still make the clearance contribution dose-dependent. For oral data, an observed change in AUC ratio cannot generally be assigned to clearance alone.
A useful descriptive contrast is \(Q=R_{\mathrm{AUC}}(D_{\mathrm{high}})/R_{\mathrm{AUC}}(D_{\mathrm{low}})\). If \(Q=1\), the point estimates are equal across those doses. If \(Q\) differs from 1, the point estimates differ; uncertainty in the estimates still matters when deciding whether the difference is persuasive. \(Q\) is a comparison of two interaction ratios, not the high-dose AUC divided by the low-dose AUC.
Mechanistic Clues and a Simple Derivation
Consider two parallel elimination pathways whose effective contributions to clearance at a particular dose are \(CL_1\) and \(CL_2\). Suppose the perpetrator inhibits pathway 1 by leaving a fraction \(q\) of its control contribution, with \(0\leq q\leq1\), while pathway 2 is unchanged. Under the linear effective-clearance representation at that dose, the ratio is:
This follows because control AUC is proportional to \(D/[CL_1(D)+CL_2(D)]\), while AUC with perpetrator is proportional to \(D/[qCL_1(D)+CL_2(D)]\). Dividing the latter AUC by the former cancels the dose and reverses the clearance quotient:
If the relative contribution of the inhibited pathway changes with dose, the ratio changes even when the perpetrator’s fractional inhibition \(q\) is constant. For instance, saturation of one pathway can reduce its effective contribution at higher concentrations, making another pathway relatively more important. Saturation of the inhibited pathway can therefore reduce the observed interaction at a higher dose. Other nonlinear mechanisms can produce other directions of change, so the direction is a clue to investigate rather than a universal rule.
Dose-dependent oral bioavailability offers a separate route to mismatch. Saturation of intestinal or hepatic first-pass processes, or changes in absorption or transport, can alter \(F\) differently in the control and perpetrator conditions. When \(F_{\mathrm{with}}/F_{\mathrm{control}}\) varies across doses, the AUC ratio can vary even if the clearance ratio is stable. A PBPK explanation should account for the processes represented in the model and be checked against the relevant observations; a ratio pattern alone does not distinguish these possibilities.
Worked Example: A Large Low-Dose Ratio That Shrinks at High Dose
Suppose hypothetical dose-normalized AUCs are consistent with the following effective clearances. At the low dose \(D_L\), control clearance is 10 L/h and clearance with perpetrator is 1.9 L/h. At the high dose \(D_H\), the corresponding values are 10 L/h and 9.1 L/h. For this illustration, assume the same bioavailability in the two conditions at each dose and that AUC can be represented as \(D/CL\). Calculate both interaction ratios.
At the low dose, control AUC is \(D_L/10\) and AUC with perpetrator is \(D_L/1.9\). Thus:
At the high dose, control AUC is \(D_H/10\) and AUC with perpetrator is \(D_H/9.1\). Thus:
The high-to-low ratio contrast is \(Q\approx0.2088\). In this hypothetical example, the AUC increase with perpetrator is much larger at the low dose than at the high dose. Saturation of the inhibited route or increased contribution from an alternative route could be considered, but neither is established by these calculations. The effective clearances are illustrative summaries, not a unique mechanistic explanation.
Worked Example: Dose-Dependent Bioavailability Can Change the Ratio
Consider a hypothetical oral probe. Assume control clearance is 10 L/h and clearance with perpetrator is 8.333 L/h at both dose levels, giving a clearance ratio \(CL_{\mathrm{control}}/CL_{\mathrm{with}}=1.2\). At the low dose, let \(F_{\mathrm{control}}=0.40\) and \(F_{\mathrm{with}}=0.80\). At the high dose, let \(F_{\mathrm{control}}=0.80\) and \(F_{\mathrm{with}}=0.90\). Use the oral AUC ratio relation to calculate the interaction ratios.
At the low dose, the bioavailability ratio is \(0.80/0.40=2.0\). Therefore:
At the high dose, the bioavailability ratio is \(0.90/0.80=1.125\). Therefore:
The clearance contribution is held constant in this constructed example, yet the AUC ratio falls from 2.400 to 1.350 because the relative bioavailability effect is smaller at the high dose. These numbers do not identify a particular absorption, transport, or first-pass mechanism. They show why an oral AUC mismatch should not automatically be interpreted as dose-dependent systemic clearance.
Worked Example: A Shift in Parallel Pathway Contributions
Suppose, hypothetically, pathway 1 is inhibited by a perpetrator such that \(q=0.10\), while pathway 2 is unchanged. At the low dose, let the control effective contributions be \(CL_1=8\) L/h and \(CL_2=2\) L/h. At the high dose, let them be \(CL_1=2\) L/h and \(CL_2=8\) L/h. These values represent a change in relative pathway contribution with dose; they are not measurements from a real study. Calculate the AUC ratio at each dose.
At the low dose, total control clearance is \(8+2=10\) L/h. With perpetrator, the contributions are \(0.10(8)=0.8\) L/h and 2 L/h, for a total of 2.8 L/h. Therefore:
At the high dose, total control clearance is \(2+8=10\) L/h. With perpetrator, the contributions are \(0.10(2)=0.2\) L/h and 8 L/h, for a total of 8.2 L/h. Therefore:
The inhibited pathway accounts for most of the control clearance at low dose but little at high dose, so inhibiting it has a smaller effect on total clearance at the high dose. This pathway-switching pattern is one plausible explanation for the mismatch. A mechanistic model must still support the assumed pathway contributions and their dose dependence.
A Practical Diagnostic Sequence
When ratios differ across probe doses, proceed from the exposure calculation toward the mechanism. This order helps avoid explaining a measurement or design artifact with unnecessary mechanistic complexity.
For every dose, calculate perpetrator-condition AUC divided by control AUC. Confirm that the AUC metric, units, sampling window, and analyte are consistent across conditions.
Confirm that each comparison uses the same probe dose, formulation, schedule, and relevant conditions, and that perpetrator exposure is appropriate and comparable. A mismatch in these features can masquerade as dose dependence.
Compare dose-normalized AUCs within control and perpetrator conditions and review whether sampling adequately represents the chosen AUC. Dose-normalized AUC can reveal nonproportional exposure, but by itself does not identify the cause. Cmax and Tmax can provide additional absorption-phase clues, but they are different endpoints and do not replace the AUC comparison.
For oral dosing, consider dose-dependent bioavailability as well as clearance. Consider saturable elimination, transporter or enzyme effects, and changing contributions of parallel pathways where supported by the model and data. For intravenous dosing, bioavailability does not explain the mismatch, so investigate systemic disposition and design or estimation issues.
Ratios are estimates, not exact constants. Use the uncertainty information available for each estimate before describing a cross-dose difference as compelling. Agreement of point estimates with a mechanistic story is not independent validation of that story.
Comparisons, Limitations, and Common Pitfalls
| Approach | What it helps answer | Key limitation |
|---|---|---|
| Compare AUC ratios across doses | Whether the observed interaction is dose-invariant for the chosen AUC endpoint | Does not identify the mechanism by itself |
| Compare dose-normalized AUC within each condition | Whether exposure is proportional to dose in control or perpetrator conditions | Nonproportionality may have several causes; it is not the same as a changing interaction ratio |
| Compare oral and intravenous probe data | Whether an oral-only pattern is consistent with a bioavailability contribution | Requires suitable, comparable data; interpretation still depends on disposition assumptions |
| Use a mechanistic PBPK model | Whether specified physiological and drug processes can reproduce the dose-specific pattern | Several parameter or pathway combinations may fit the same ratios; fit alone does not prove a mechanism |
A frequent error is to divide control AUC by perpetrator AUC while describing the result as the perpetrator-to-control ratio. For example, if AUC with perpetrator is \(D/1.9\) and control AUC is \(D/10\), the defined ratio is \((D/1.9)/(D/10)=10/1.9\), not \(1.9/10\). Writing the numerator and denominator explicitly before simplifying prevents this reversal.
Another pitfall is to call a ratio change across doses “enzyme saturation” without checking alternatives. The same pattern can arise from dose-dependent bioavailability, pathway switching, differences in perpetrator exposure, incomplete AUC estimation, or variability in the estimates. Conversely, equal ratios at two doses do not prove linear pharmacokinetics; distinct dose-dependent processes can offset one another. The available data and model assumptions determine how far a mechanistic conclusion can go.
Check Your Understanding
Work through these questions, then compare your reasoning with the solutions.
- At one dose, hypothetical control AUC is 50 mg·h/L and AUC with perpetrator is 125 mg·h/L. What is the AUC interaction ratio, and how is it interpreted?
- For an oral drug, the clearance ratio is constant across doses, but the ratio \(F_{\mathrm{with}}/F_{\mathrm{control}}\) falls at the higher dose. What direction of change in the AUC interaction ratio does this imply?
- In the parallel-pathway example, explain why the perpetrator has a smaller effect on AUC at the high dose.
- Two dose-specific point estimates differ, but their uncertainty is large. What can be concluded from the point estimates, and what should not be concluded?
- Why does dose-normalized AUC not, by itself, identify the cause of a cross-dose interaction-ratio mismatch?
Solutions. (1) \(R_{\mathrm{AUC}}=125/50=2.500\). AUC with perpetrator is 2.500 times control AUC at that dose, or 150% higher. (2) With the clearance ratio fixed, a falling bioavailability ratio makes \(R_{\mathrm{AUC}}\) fall, by the oral AUC ratio relation. (3) At high dose, the inhibited pathway contributes only 2 of 10 L/h control clearance, compared with 8 of 10 L/h at low dose. Inhibiting a smaller share of total clearance produces a smaller increase in AUC. (4) The point estimates differ, but the data may not establish a reliable cross-dose difference; do not claim a mechanism or a definite dose-dependent interaction without accounting for uncertainty. (5) Dose-normalized AUC can show that exposure is not proportional to dose, but bioavailability, clearance, absorption, and other processes may contribute. Additional evidence is needed to distinguish them.