The number of subjects to be enrolled in the study depends on the magnitude of the effect to be detected that is considered to be clinically relevant, the inter and intrasubject variability in the PK measurements and any other factors that might affect the outcome of the study.
The most common statistical design for pharmacokinetic drug interactions is the crossover design accounting for half of all the studies submitted to the Agency from 1987 to 1997. More recently an increased reliance on a fixed sequence design (where a subject receives a drug for a fixed period and the second drug is introduced at a certain time in the dosing period). Such a design is considered to be a variation of the crossover design. A parallel design is most useful in situations where one of the studied drugs or its metabolites have a long half-life.
According to the FDA guidance, the results of the drug-drug interaction studies should be reported as 90% confidence intervals about the geometric mean ratio of the observed PK measure with and without the interacting drug. Confidence intervals will provide an estimate of the distribution of the observed systemic exposure with and without the interacting drug and thus conveying a probability of the magnitude of the interaction. On the other hand, tests of significance are not appropriate for such studies due to the fact that clinically insignificant exposure differences can achieve statistical significance without having to recommend dosing adjustments or contraindications.
Moreover, the FDA guidance recommends that in a drug-drug interaction study, the sponsor of the investigational drug should be able to provide specific dosing recommendations based on what is known about the PK/PD relationship or the dose-response relationship. Unfortunately such information is not always available especially for drugs that are already on the market.
If the sponsor intends to make a specific claim in the package insert that no drug interaction is present, the sponsor should be able to recommend specific "no effect boundaries" or clinical equivalence intervals defined as the interval within which the change in a systemic exposure measure is considered to be clinically not relevant.
The guidance recommends three approaches in defining these no effect boundaries:
The no effect boundaries are based on population average dose-response or exposure-response relationships and any other available information for the drug under study. If the 90% confidence interval for the systemic exposure measure falls within the no effect boundary, then it may be concluded that no clinically significant drug-drug interaction is present.
The no effect boundary may also be based on the concept that a drug-drug interaction study addresses the question of switchability between the substrate given alone and in combination with an interacting drug. In this case, a sponsor may wish to use an individual equivalence criterion to allow for scaling of the no effect boundary.
In the absence of no effect boundaries as defined in Approach 1 or 2, a sponsor may use a default no effect boundary of 80-125% for both the investigational drug and the approved drugs used in the study. When the 90% confidence intervals for systemic exposure fall entirely within the equivalence range, the Agency in most cases will conclude that clinically significant interaction is present.
It is to note that Approach 3 does not necessary imply that the sponsor needs to always power the study in a way that the 90% confidence interval for the ratio of pharmacokinetic measurements falls entirely within the no effect boundary resulting in an increased number of subjects for each study.
Was this article helpful?