Why Dose Timing Is Part of DDI Study Design
A drug-drug interaction (DDI) study can use an appropriate perpetrator and victim yet still miss the interaction if the doses are poorly timed. A victim dose given before the perpetrator has reached an effective exposure may underestimate the interaction. Conversely, a schedule designed only to capture a brief maximum can misrepresent how the drugs are used together in practice. The dosing plan should therefore connect the expected time course of the interaction to the question the study is intended to answer.
The central planning task is to choose both a perpetrator lead-in—the time between starting perpetrator dosing and administering the victim—and a victim dose time that exposes the victim during the intended interaction state. For a rapidly acting interaction, that state may be reached soon after a perpetrator dose. For an interaction that builds as perpetrator exposure accumulates, the lead-in may need to be longer. The timing decision should come from the predicted interaction profile, not from a rule that every perpetrator must be given for a fixed number of days.
The ratio is a prediction for a particular victim dose and perpetrator schedule, not a property of perpetrator concentration alone. It can depend on the victim’s absorption and elimination, the time-varying perpetrator exposure during the victim’s concentration profile, and any active perpetrator species included in the model. For this reason, a perpetrator concentration peak is not automatically the time of maximal victim AUC change.
Translate the Prediction Into a Dosing Schedule
Start by specifying what the study should represent. If the aim is the largest expected interaction, schedule the victim when the predicted interaction is near its maximum and sufficiently stable to be reproducible. If the aim is clinically relevant coadministration, use the intended clinical dosing times and evaluate the interaction over that regimen—even if those times do not coincide with the peak perpetrator concentration. When both questions matter, simulations can support a primary schedule and a separate timing assessment.
A practical planning sequence is to simulate the proposed perpetrator regimen, evaluate candidate victim dose times, and compare the resulting exposure ratios. The perpetrator lead-in should be long enough for the relevant interaction to be established at the selected victim time. A prespecified criterion makes “long enough” testable; for example, one may require the predicted interaction to be at least 90% of its eventual change. This threshold is a design choice, not a universal biological constant.
Specify whether the study seeks a near-maximal effect, an intended clinical coadministration pattern, or both. Define the exposure endpoint and the criterion for an adequately established interaction.
Use the planned perpetrator doses and a suitable model to predict perpetrator exposure and the victim exposure ratio across candidate start times.
Select the earliest time that meets the target criterion, then verify that the chosen victim dose overlaps the intended interaction window.
Compare the selected schedule with intended use, tolerability, adherence, and the study’s sampling plan. If they conflict, make the trade-off explicit rather than silently changing the question.
Why Repeated Doses Approach a Plateau
A simple accumulation calculation can give an initial estimate of perpetrator lead-in time. Assume linear one-compartment elimination with first-order rate constant \(k\), repeated doses every \(\tau\) time units, and no dose changes. At a fixed time after each dose, each earlier dose contributes a fraction that decays geometrically. If \(N\) doses have been given, the fraction of the eventual repeated-dose concentration at that same point in the dosing interval is:
Here \(t_{1/2}\) is the elimination half-life. To justify the fraction, write the concentration after \(N\) doses as a finite geometric sum of dose contributions, proportional to \(1+q+\cdots+q^{N-1}\), where \(q=e^{-k\tau}\). The corresponding eventual sum is proportional to \(1/(1-q)\). Their ratio is \((1-q^N)\), which equals \(1-e^{-Nk\tau}\). This calculation assumes linear elimination and identical dosing intervals; it estimates concentration accumulation, not necessarily the time course of the DDI effect.
If a 90% concentration plateau is the chosen proxy, solve \(1-e^{-Nk\tau}\geq0.90\), giving \(N\geq\ln(10)/(k\tau)\). A PBPK prediction of the victim exposure ratio remains the more direct basis for scheduling when the interaction does not track perpetrator concentration proportionally.
Worked Example: Estimating a Lead-In for a Long-Half-Life Perpetrator
Hypothetical scenario. A perpetrator is planned once daily, its relevant half-life is estimated to be 36 hours, and the protocol uses 90% of eventual concentration at the same dosing phase as an initial lead-in target. How many dosing intervals are needed?
The elimination rate is \(k=\ln(2)/36=0.01925\ \mathrm{h}^{-1}\), and \(\tau=24\ \mathrm{h}\). Thus \(k\tau=0.4621\). The target condition is:
Because the regimen uses whole daily dosing intervals, round up to \(N=5\). At five intervals (120 hours), the predicted fraction is \(1-e^{-5(0.4621)}=1-e^{-2.3105}=0.9008\), or about 90.1%. Under the stated assumptions, five intervals meet the concentration target. This is a starting estimate only: the victim exposure ratio must still be checked because concentration accumulation and interaction accumulation need not be identical.
Worked Timing Decisions
Worked Example: Scheduling a Victim Dose Near a Rapid Interaction Window
Hypothetical scenario. A PBPK simulation for a rapidly acting perpetrator predicts that its relevant concentration is highest about 1.5 hours after each dose. For a particular victim regimen, the largest predicted AUC change occurs when the victim is administered at that time. The study aims to capture this near-maximum effect. The perpetrator is dosed at 08:00 and 20:00. Choose a victim dose time for the morning assessment.
The predicted concentration peak is \(08{:}00+1.5\ \mathrm{h}=09{:}30\). The simulation’s stated victim-specific result therefore supports administering the victim at 09:30, rather than simply giving both drugs at 08:00. If the model also predicts that the interaction remains within 5% of its maximum from 09:00 to 11:00, 09:30 falls inside that near-maximum window.
The conclusion depends on the supplied simulation result: the victim dose at 09:30 is predicted to capture the largest effect for this schedule. The perpetrator peak alone would not establish that conclusion. Before finalizing the schedule, the team should verify that the model’s input includes the intended formulation and dosing conditions and that the study can administer and sample the victim at this time.
Worked Example: Choosing Between a Near-Maximum and a Longer Lead-In
Hypothetical scenario. Candidate simulations predict the following victim AUC ratios after starting a repeated perpetrator regimen: day 3, \(R=1.52\); day 6, \(R=1.74\); day 8, \(R=1.78\). The predicted eventual ratio is \(R_{\max}=1.80\). The protocol defines an adequate near-maximum interaction as attaining at least 90% of the eventual change in the ratio’s excess above 1. Which candidate first meets the criterion?
The proportion of the eventual excess effect is calculated as \((R(t)-1)/(R_{\max}-1)\). At day 3, this is \((1.52-1)/(1.80-1)=0.52/0.80=0.650\), or 65.0%. At day 6, it is \(0.74/0.80=0.925\), or 92.5%. Day 6 is therefore the earliest listed candidate meeting the 90% criterion. At day 8 the value is \(0.78/0.80=0.975\), or 97.5%.
If the study seeks the earliest adequate near-maximum effect, day 6 is supported by these predictions; the additional predicted gain by day 8 is \(1.78-1.74=0.04\) AUC-ratio units. If the clinical question specifically concerns day 8 coadministration, the study should instead preserve that regimen. The arithmetic helps make the design trade-off visible; it does not decide which clinical question matters more.
Diagnostics, Alternatives, and Limitations
Review the full predicted time course rather than only the selected time point. A proposed schedule is questionable if the exposure ratio is still changing substantially during the victim’s relevant concentration profile, if small timing shifts produce large changes in the predicted ratio, or if the chosen victim time does not match the intended clinical regimen. Also check whether the selected lead-in follows from the modeled dosing history, including missed or delayed doses if these are relevant to the study question.
The main alternatives are to administer the victim after a fixed short lead-in, to wait for a concentration-based plateau, or to select a time directly from the predicted victim exposure-ratio profile. A fixed short lead-in is simple but may miss accumulation. A concentration plateau is transparent and useful as an initial planning rule, but may be a poor proxy when the interaction response is nonlinear or delayed relative to concentration. Selecting from the exposure-ratio profile is more directly tied to the endpoint, but depends on the adequacy of the model and its inputs. In each case, the clinical regimen and the target estimand should guide the choice.
A common mistake is to assume that giving the victim at the perpetrator’s concentration peak must produce the maximal DDI. That reasoning ignores the victim’s own time course and the relationship between perpetrator exposure and interaction magnitude. The correction is to evaluate candidate victim doses using the predicted victim exposure endpoint under the complete dosing schedule. If the DDI changes over time in a way that a concentration-only schedule cannot explain, a simple accumulation calculation is inadequate; the relevant time-dependent process must be represented before the schedule can be justified.
Check Your Understanding
Answer each question, then compare your reasoning with the solutions below.
- A perpetrator has a 24-hour half-life and is dosed every 24 hours. Using the concentration-accumulation approximation, what fraction of its eventual concentration is reached after three dosing intervals?
- A predicted victim exposure ratio changes from 1.00 without perpetrator to an eventual value of 2.00. A candidate schedule predicts a ratio of 1.70. What fraction of the eventual excess effect has been reached?
- Why is a perpetrator concentration peak not sufficient evidence that a victim dose given then will capture maximal victim AUC change?
- A schedule produces a near-maximum predicted ratio but does not resemble the intended clinical dosing times. What design issue must be resolved?
Solutions. 1. Here \(k\tau=\ln 2\), so \(f_3=1-e^{-3\ln 2}=1-1/8=0.875\). The estimate is 87.5% of the eventual concentration at the same dosing phase, under the stated linear-elimination assumptions. 2. The fraction is \((1.70-1)/(2.00-1)=0.70\), or 70%. 3. Victim AUC reflects the perpetrator’s time course over the victim concentration profile, not just perpetrator concentration at one instant; the result also depends on the modeled relationship between perpetrator exposure and interaction magnitude. 4. The investigators must decide whether the primary aim is the near-maximum interaction or clinically representative coadministration, and make any difference between those aims explicit in the schedule and interpretation.