Translating Ratios into Dose Changes Under Proportional Pharmacokinetics

From an Exposure Ratio to a Candidate Dose

A predicted drug-drug interaction (DDI) exposure ratio can help translate a model result into a candidate dose adjustment. The calculation is direct only under a key condition: exposure must change in proportion to dose over the range being considered, and the interaction ratio must remain applicable at the adjusted dose. Under those conditions, a dose can be scaled to target a chosen exposure. The result is a quantitative candidate, not by itself a clinical recommendation.

First define the direction of the ratio. In this tutorial, $R$ is the exposure under the interacting condition divided by the exposure under the reference condition, with both conditions evaluated at the same dose. For example, $R=1.50$ means the predicted exposure under the interacting condition is 1.50 times the reference exposure at that dose. A ratio below 1 means exposure is lower under the interacting condition.

Definition: Let $D_0$ be a reference dose, $A_0$ the reference exposure at $D_0$, and $R$ the interacting-to-reference exposure ratio at the same dose. Let $A^*$ be the target exposure under the interacting condition. If exposure is dose-proportional and $R$ is dose-independent over the relevant range, the continuous candidate dose is $$ D^* = D_0 \frac{A^*}{A_0}\frac{1}{R}. $$ If the target is the reference exposure, $A^*=A_0$, so $D^*=D_0/R$.

Here, “exposure” must refer to a specified metric, such as AUC over a dosing interval or $C_{\max}$, with the relevant dosing regimen and conditions held consistent. A dose calculated to match AUC does not automatically match $C_{\max}$: the two metrics can have different interaction ratios. The calculation also does not establish what exposure is clinically appropriate. That target must be specified separately.

Why the Reciprocal Appears

Suppose reference exposure is proportional to dose over the relevant range. At a dose $D$, reference exposure is then $A_0(D/D_0)$. If the interacting-to-reference ratio remains $R$, the interacting-condition exposure at that dose is

A_{\mathrm{int}}(D)=R A_0\frac{D}{D_0}.

To find the dose that reaches the target $A^*$, set this expression equal to the target and solve for $D$:

A^*=R A_0\frac{D^*}{D_0} \quad\Longrightarrow\quad D^*=D_0\frac{A^*}{A_0}\frac{1}{R}.

This derivation shows both the reciprocal adjustment and its conditions. If exposure rises by a factor $R$ and the target is unchanged, the dose is divided by $R$, not multiplied by $R$. If the target is some other specified multiple of reference exposure, that target multiple must also appear in the calculation.

Assumptions to Check Before Calculating

The equation is useful only when its assumptions are reasonable for the dose and conditions at issue. In particular:

Dose proportionality: the selected exposure metric changes in direct proportion to dose over the interval from the reference dose to the candidate dose. In this range, doubling dose doubles that exposure metric.
Ratio applicability: the predicted ratio applies to the same interacting condition, regimen, population, and exposure metric used for the dose calculation. It is assumed not to change materially as dose changes.
Comparable conditions: the reference and interacting exposures refer to comparable dosing intervals and steady-state or single-dose conditions. A ratio from one regimen is not automatically transferable to another.
Defined target: the desired exposure is specified. Matching the reference exposure is one possible objective, not a default proof that the resulting exposure is safe or effective.
Usable regimen: the calculated dose can be represented by an available formulation and schedule that can actually be administered as intended.

These are assumptions to assess, not consequences of the equation. For example, a model prediction at one dose does not alone demonstrate proportionality across a much wider dose range. If the relevant model or evidence does not support these conditions, the reciprocal calculation should not be treated as a reliable dose adjustment.

Worked Example: Reducing a Dose After an Exposure Increase

Consider a hypothetical drug taken once daily at a reference dose of 200 mg. A PBPK prediction gives an interacting-to-reference AUC ratio of $R=1.50$ at the same dose. The target is the reference AUC, and, for this example, assume AUC is dose-proportional over the candidate range and the ratio remains 1.50 there.

The continuous candidate dose is

D^*=\frac{D_0}{R} =\frac{200\ \mathrm{mg}}{1.50} =133.3\ \mathrm{mg}.

Suppose the hypothetical product is available in 25 mg increments. The nearest practical choices are 125 mg and 150 mg. Check each against the target by calculating its predicted interacting-condition AUC relative to the reference AUC:

\frac{A_{\mathrm{int}}(125)}{A_0} =1.50\frac{125}{200} =0.9375, \qquad \frac{A_{\mathrm{int}}(150)}{A_0} =1.50\frac{150}{200} =1.125.

The 125 mg regimen predicts 93.75% of the reference AUC, or 6.25% below it; 150 mg predicts 112.5%, or 12.5% above it. Of these two choices, 125 mg is closer to the target on the exposure-ratio scale. This comparison does not establish that 125 mg is clinically suitable; the product's permitted use, dosing schedule, and other clinical considerations still matter.

Worked Checks for Other Targets and Ratios

Worked Example: Increasing a Dose After an Exposure Decrease

In a hypothetical case, the reference dose is 80 mg once daily. Under an interacting condition, the predicted AUC ratio at the same dose is $R=0.40$. The target is the reference AUC. Assume AUC remains dose-proportional up to the candidate dose, the ratio remains applicable, and the hypothetical product can provide 20 mg increments.

The ideal dose is

D^*=\frac{80\ \mathrm{mg}}{0.40}=200\ \mathrm{mg}.

The factor of $1/0.40=2.5$ means the candidate is 2.5 times the reference dose. Check the predicted AUC relative to reference:

\frac{A_{\mathrm{int}}(200)}{A_0} =0.40\frac{200}{80} =0.40(2.5) =1.00.

Thus, under the stated assumptions, 200 mg predicts the reference AUC. It is representable in 20 mg increments, but that arithmetic check does not authorize a 200 mg regimen. The assumed proportional range, formulation instructions, and clinical appropriateness would still need to support it.

Worked Example: Targeting Less Than the Reference Exposure

Suppose a hypothetical reference dose is 120 mg, and the interacting-to-reference ratio for the selected exposure metric is $R=1.25$. The specified target under the interacting condition is $A^*=0.80A_0$, rather than the full reference exposure. Assume dose proportionality and ratio applicability over the candidate range. Available strengths are 10 mg and 20 mg, and combinations of these strengths can be used in this hypothetical example.

The continuous candidate dose is

D^* =120\ \mathrm{mg}\times\frac{0.80A_0}{A_0}\times\frac{1}{1.25} =120\ \mathrm{mg}\times0.80\times0.80 =76.8\ \mathrm{mg}.

A practical 80 mg dose can be formed from four 20 mg units. Its predicted exposure relative to reference is

\frac{A_{\mathrm{int}}(80)}{A_0} =1.25\frac{80}{120} =0.8333.

The target is $0.80A_0$, so 80 mg predicts $0.0333A_0$ above the target, which is about 4.17% above the target value. For comparison, 70 mg, formed from three 20 mg units and one 10 mg unit, predicts $1.25(70/120)=0.7292A_0$, about 8.85% below the target value. Of these two choices, 80 mg is closer to the specified target. The target itself must have a clinical justification; the calculation only translates it into a candidate dose.

Checking the Candidate and Recognizing Failure

After calculating a dose, substitute it back into the exposure equation. For a candidate $D_c$, the predicted interacting exposure relative to the reference is

\frac{A_{\mathrm{int}}(D_c)}{A_0}=R\frac{D_c}{D_0}.

Compare that value with $A^*/A_0$, not automatically with 1. This check catches arithmetic errors and makes rounding to a practical strength visible. It also clarifies whether the candidate is above or below the chosen target.

Dose strength selection is a second step, not part of the continuous calculation. The mathematically nearest dose may be unavailable, unsuitable for the prescribed schedule, or inconsistent with product instructions. When several feasible doses exist, compare their predicted exposures with the target and state the discrepancy. The exposure equation can rank candidates relative to that target, but it cannot decide which discrepancy is clinically acceptable.

A useful diagnostic is to calculate the implied dose-normalized exposure at more than one dose when suitable data or model outputs are available. Under proportionality, $A(D)/D$ should be approximately constant for a given condition over the relevant range. If it changes materially, the linear scaling assumption is not supported there. Likewise, if predicted interacting-to-reference ratios differ across doses, a single dose-independent $R$ is not an adequate basis for this calculation. A mismatch can also arise from comparing different exposure metrics, regimens, or conditions.

How This Calculation Relates to Other Approaches

The reciprocal rule $D_0/R$ is a special case of the target-based equation: it applies when the target is exactly the reference exposure. Writing the more general equation is helpful when the intended target is a specified fraction or multiple of reference exposure. Both are algebraic translations of a ratio under the same proportionality assumptions; neither independently validates the target or assumptions.

A second practical approach is to evaluate candidate doses directly in the PBPK model and compare their predicted exposure with the target. This can make the dose-strength check explicit and can test whether the simple scaling calculation is a reasonable approximation. If the model predicts dose-dependent changes in exposure or in the DDI ratio, the direct model result is not interchangeable with a reciprocal adjustment based on one ratio. Conversely, direct simulation does not remove the need to define the target, regimen, and exposure metric.

The ratio may itself be uncertain. A dose computed from a point prediction is therefore a candidate based on that point prediction, not a guarantee that the target will be achieved. Where uncertainty matters to the decision, the exposure predictions associated with plausible ratios can be checked using the same dose equation. This calculation does not eliminate uncertainty or establish an individual patient's exposure.

Common Mistakes and Limitations

Multiplying by the ratio when exposure is increased. If $R=1.50$ and the target is unchanged, using $1.50D_0$ would increase rather than offset the predicted exposure increase. The proportional adjustment is $D_0/1.50$.
Reversing the ratio direction. A ratio defined as interacting divided by reference must not be used as though it were reference divided by interacting. State the numerator and denominator before calculating.
Assuming any target is the reference exposure. If the target is $0.80A_0$, the target factor $0.80$ belongs in the formula. Omitting it calculates a different dose.
Treating an ideal dose as a recommendation. A continuous calculation may not map to an appropriate available regimen, and a close exposure match does not by itself establish safety or efficacy.
Assuming one exposure metric stands for all others. Matching AUC does not establish that $C_{\max}$ is also matched. Calculate and assess each relevant metric separately.
Extending proportionality beyond its support. If dose-normalized exposure or the interaction ratio changes over the relevant range, the linear calculation may mispredict exposure. Do not use it as though the assumption had been verified.

Key takeaway: Under dose-proportional pharmacokinetics and a dose-independent interaction ratio, the candidate dose is the reference dose multiplied by the target-to-reference exposure fraction and divided by the interacting-to-reference exposure ratio. Check the predicted exposure at the practical dose, and treat the result as conditional on its assumptions.

Check Your Understanding

For each question, define the ratio direction and use the stated assumptions. Brief solutions follow the questions.

A hypothetical 100 mg dose gives an interacting-to-reference AUC ratio of 2.0. If the target is the reference AUC, what is the ideal candidate dose under proportionality?
A hypothetical reference dose is 60 mg, $R=0.75$, and the target is $1.20A_0$. Calculate the continuous candidate dose.
For $D_0=150$ mg, $R=1.25$, and a 120 mg candidate, calculate interacting exposure relative to reference and state whether it is above or below $A_0$.
Name two assumptions that should be checked before applying the reciprocal dose adjustment to a PBPK prediction.

Solutions. (1) $D^*=100/2.0=50$ mg. (2) $D^*=60(1.20)/0.75=96$ mg. (3) $A_{\mathrm{int}}/A_0=1.25(120/150)=1.00$, so the predicted exposure equals $A_0$. (4) Examples include dose proportionality over the reference-to-candidate range and applicability of the same interaction ratio at the candidate dose; the exposure metric, regimen, target, and available formulation also need to be appropriate.