From Predicted Ratios to a Candidate Dose
A predicted drug-drug interaction (DDI) ratio describes a change in exposure; it does not, by itself, prescribe a dose. A ratio can classify the magnitude of a predicted change, while dose adjustment asks a different question: what dose might bring exposure toward a chosen reference under specified assumptions? This distinction matters because matching AUC does not necessarily match Cmax, and neither calculation alone establishes that the resulting regimen is safe or effective.
The cases below use hypothetical predictions. In each, the interacting drug is present in the new regimen, and the reference is the victim drug given without that interacting drug. The proposed dose is a calculation to evaluate, not an automatic clinical recommendation. Its relevance depends on the therapeutic target, the reliability of the prediction, and whether the assumptions used in the calculation hold over the proposed dose range.
For illustration, use the common fold-change convention: \(F<1.25\) is no effect by this convention, \(1.25\leq F<2\) is weak, \(2\leq F<5\) is moderate, and \(F\geq5\) is strong. Regulatory conventions and the exposure metric used for a particular decision can differ; therefore, state the convention and report AUC and Cmax separately. A magnitude category is not a statement about clinical consequence.
Proportional Dose Adjustment
A simple candidate adjustment is available when AUC is proportional to dose over the relevant range. Assume that the interaction's relative effect is unchanged after the dose change, that the victim drug is at steady state when the comparison requires it, and that clearance, bioavailability, adherence, and other relevant conditions do not change with dose or regimen. Let \(D_{\mathrm{old}}\) be the dose under the interacting-drug regimen before adjustment and \(D_{\mathrm{new}}\) the candidate dose afterward. Define \(m=D_{\mathrm{new}}/D_{\mathrm{old}}\). Under dose-proportional AUC, the adjusted AUC ratio is \(R_{\mathrm{AUC}}m\). Setting that ratio to 1 gives:
This calculation targets the reference AUC, not an individualized therapeutic target. It also does not guarantee that Cmax returns to its reference value. Only if Cmax is also proportional to dose and the ratio remains unchanged does the adjusted Cmax ratio equal \(R_{\mathrm{Cmax}}m\). If dose proportionality is uncertain or known to fail, a linear calculation can be misleading; use an appropriate dose-exposure relationship instead.
Worked Example: Moderate AUC Increase, Smaller Cmax Increase
Suppose a hypothetical prediction for a 100 mg once-daily victim-drug regimen gives \(R_{\mathrm{AUC}}=2.40\) and \(R_{\mathrm{Cmax}}=1.80\). Assume AUC and Cmax are both proportional to dose in the range considered, and assume the interaction ratios stay constant after the dose change. Classify each ratio and calculate a candidate dose that targets the reference AUC.
For AUC, \(F=\max(2.40,1/2.40)=2.40\), which is moderate under the stated convention. For Cmax, \(F=\max(1.80,1/1.80)=1.80\), which is weak. The metrics therefore indicate different fold-change categories; they should not be collapsed into one unqualified label.
The dose multiplier targeting the reference AUC is \(m=1/2.40=0.4167\). Thus, \(D_{\mathrm{new}}=100\text{ mg}\times0.4167=41.67\text{ mg}\), approximately 41.7 mg once daily. This is a 58.3% reduction from 100 mg.
Under the additional assumption of Cmax proportionality, the adjusted Cmax ratio is \(1.80\times0.4167=0.7501\), approximately 0.750. The calculation predicts AUC at the reference level but Cmax about 25.0% below the reference. It does not show that this Cmax is clinically acceptable, nor that a 41.7 mg dose can be prescribed using available dosage forms.
Decreases and Nonproportional Exposure
For a predicted decrease, the same formula can imply a dose increase. The result must be treated cautiously: the interaction may not remain constant as dose changes, and an increased dose can raise exposure beyond the intended range if the prediction or proportionality assumption is wrong. Classifying a decrease by its reciprocal fold change makes its magnitude easier to compare with an increase, but the direction must still be reported.
When dose-exposure relationships are nonproportional, an empirically supported power relationship provides one possible calculation. Let \(b_{\mathrm{AUC}}\) describe the dose dependence of the interaction-adjusted AUC ratio, and let \(b_{\mathrm{Cmax}}\) describe that of the Cmax ratio. Under the assumed model, if the dose is multiplied by \(m\), the ratios become \(R_{\mathrm{AUC}}m^{b_{\mathrm{AUC}}}\) and \(R_{\mathrm{Cmax}}m^{b_{\mathrm{Cmax}}}\). These exponents must be justified for the relevant dose range and regimens; they are not universal constants. Choosing a multiplier to target AUC gives:
After calculating \(m\), evaluate the predicted Cmax separately using its own exponent and ratio. AUC targeting does not require Cmax to return to its reference value. If the power model is unsupported, extrapolates beyond observed doses, or fails to represent the relevant mechanism, the resulting number should not be presented as a reliable adjustment.
Worked Example: Nonproportional AUC and Cmax
In a hypothetical case, a 200 mg regimen with the interacting drug has predicted ratios \(R_{\mathrm{AUC}}=1.90\) and \(R_{\mathrm{Cmax}}=1.40\). Suppose dose-ranging information supports, over the proposed range, the power relationships with \(b_{\mathrm{AUC}}=1.30\) and \(b_{\mathrm{Cmax}}=1.10\). Assume these relationships and the interaction ratios remain applicable to the adjusted regimen.
The AUC fold change is 1.90, weak under the stated convention; the Cmax fold change is 1.40, also weak. To target the reference AUC:
The candidate dose is \(200\text{ mg}\times0.61034=122.068\text{ mg}\), approximately 122 mg. Substitution checks the AUC target: \(1.90(0.61034)^{1.30}\approx1.00\).
The corresponding Cmax ratio is \(1.40(0.61034)^{1.10}\). The power term is approximately 0.58094, so \(R_{\mathrm{Cmax,new}}\approx1.40(0.58094)=0.8133\). Thus, under this model, the dose targets reference AUC while predicting Cmax about 18.7% below its reference. The calculation depends on both exponents applying at roughly 122 mg and with the interacting drug present; it does not establish that the dose is appropriate for a particular patient.
Worked Example: Predicted Exposure Decrease
Suppose a hypothetical 50 mg once-daily regimen with an interacting drug has \(R_{\mathrm{AUC}}=0.62\) and \(R_{\mathrm{Cmax}}=0.78\). Assume both measures are proportional to dose and the ratios remain constant. Classify the changes and calculate the candidate dose targeting reference AUC.
The AUC fold change is \(F=\max(0.62,1/0.62)=1.613\), approximately 1.61, so the AUC decrease is weak under the stated convention. The Cmax fold change is \(F=\max(0.78,1/0.78)=1.282\), approximately 1.28, also weak. The direction is a decrease for both metrics; the fold-change category does not communicate that direction on its own.
For AUC, \(m=1/0.62=1.6129\), and \(D_{\mathrm{new}}=50\text{ mg}\times1.6129=80.65\text{ mg}\), approximately 80.6 mg once daily. The predicted adjusted AUC ratio is \(0.62\times1.6129\approx1.00\). Under Cmax proportionality, the adjusted Cmax ratio is \(0.78\times1.6129\approx1.258\), or about 1.26. Thus, the proposed AUC-targeting increase would predict Cmax about 25.8% above the reference. A clinician would need to consider this discordance, available dosage forms, and the therapeutic context rather than assume that matching AUC settles the decision.
Checking the Calculation and Its Limits
A ratio-based adjustment is most interpretable when the reference and interacting-drug conditions are clearly defined and comparable. Check that numerator and denominator refer to the same victim drug, dose basis, route, and exposure metric, and that the ratio is not inadvertently inverted. Confirm whether the ratio is a prediction for the intended population and regimen. AUC and Cmax may respond differently because the interaction can alter the concentration-time profile, not merely scale it.
Useful warning signs include an adjustment that depends on extrapolating far beyond the doses supporting the dose-exposure relationship; a candidate dose outside the modeled range; implausible predicted exposure after adjustment; or a large discrepancy between adjusted AUC and Cmax. Also check whether the available dose strengths permit a practical regimen. A calculation yielding 41.7 mg does not mean that a 40 mg tablet, a split tablet, or an altered dosing interval will produce the same exposure. Formulation and schedule changes can affect peak and total exposure differently.
The main alternative to direct ratio-based calculation is to simulate or otherwise evaluate the proposed regimen under the mechanistic or empirical model, comparing both exposure metrics at the candidate dose. That approach can represent dose-dependent processes when the model has been supported for them, but its output still depends on model assumptions and input evidence. A simple proportional calculation is transparent and easy to audit, but cannot represent nonproportionality. A power relationship can represent curvature over a supported range, but should not be extrapolated merely because it produces a numerical answer.
The earlier material on proportional and nonproportional dose adjustment establishes why the dose-exposure relationship matters; the cases here apply that distinction while keeping classification separate from clinical action. The earlier material on choosing between dose adjustment, monitoring, and avoidance is also relevant to implementation: a candidate dose calculation cannot substitute for the overall risk-benefit judgment. Predictions have uncertainty, and a point estimate alone does not show how robust a proposed adjustment is near a decision boundary.
Check Your Understanding
Use the stated ratio definitions and assumptions in each problem. Solutions follow the questions.
- A predicted AUC ratio is 0.50 and a predicted Cmax ratio is 0.90. Classify each by fold change using the convention in this tutorial, and state the direction.
- With proportional AUC, an interacting-drug regimen is 80 mg and has \(R_{\mathrm{AUC}}=1.60\). Find the candidate dose that targets reference AUC.
- A regimen has \(R_{\mathrm{AUC}}=2.00\), \(R_{\mathrm{Cmax}}=1.50\), and both metrics are proportional to dose. If the dose is multiplied by 0.50, calculate both adjusted ratios.
- For \(R_{\mathrm{AUC}}=1.50\) and \(b_{\mathrm{AUC}}=1.20\), calculate the power-model dose multiplier that targets reference AUC. What assumption is essential to interpreting it?
Solutions. (1) AUC has \(F=\max(0.50,2.00)=2.00\), a moderate decrease. Cmax has \(F=\max(0.90,1/0.90)=1.111\), below 1.25, so it is no effect by this convention; its predicted direction is a 10% decrease. (2) \(m=1/1.60=0.625\), so \(D_{\mathrm{new}}=80\text{ mg}\times0.625=50.0\text{ mg}\). (3) The adjusted ratios are \(2.00\times0.50=1.00\) for AUC and \(1.50\times0.50=0.750\) for Cmax. AUC is targeted, while Cmax is predicted 25.0% below reference. (4) \(m=(1/1.50)^{1/1.20}\approx0.7132\). This is interpretable only if the power relationship and interaction effect apply at the adjusted dose and regimen; the exponent is not assumed to be universal.