Clinical Trial Simulation to Estimate the Sample Size for Investigation of the Impact of a Drug A on the Pharmacokinetics of Methotrexate, a common co-medication used in Rheumatoid Arthritis (RA)
Shuying Yang and Misba Beerahee
GlaxoSmithKline, Clinical Pharmacology Modelling & Simulation, UK
Objectives: Due to the widespread background use of methotrexate (MTX) in RA patients and the observed systemic exposure related adverse events driven by MTX, there is a need to evaluate potential impact of any add-on therapy, such as Drug A, on the PK of MTX. Although risk assessment on metabolic pathways of these two drugs (MTX and Drug A) indicates unlikely drug-drug PK interaction, there is a clinical/regulatory need to evaluate any interaction before embarking in a large patient study. Therefore, we proposed to design a sub-study within a large phase IIb study to address any drug interaction liability between the two drugs thereby obviating the need to perform sequential studies. A clinical trial simulation strategy based on population PK modelling with sparse sampling schemes is used to estimate the sample size required to adequately examine the possible impact of Drug A on the PK of MTX.
Methods: Preliminary assessment indicated that any likely (if at all) effect of Drug A on the PK of MTX, would possibly be from a bioavailability standpoint. A population PK model for MTX was developed using historical in house data in RA patients. Using this model an optimal design strategy was applied to determine the optimal sampling windows within the design scope of the multi-centre phase IIb study. Clinical trial simulations based on a parallel design, in conjunction with the population PK modelling were subsequently applied to evaluate the sample size needed to assess the possible influence of Drug A on the bioavailability of MTX under the standard bioequivalence criteria.
Results: A two-compartment, first order absorption model with proportional residual error adequately described the systemic PK of MTX. With the five population PK parameters, five optimal sampling windows were selected using the optimal design softwares, PopDes and PopED. Under parallel design with equal number of subjects allocated to two arms: Drug A+ MTX and Placebo+MTX, total number of subjects in each trial ranging from 40 to 100 were studied. The steady state PK data of MTX for each subject was simulated using the population PK model. Individual simulated data for each trial were subjected to population PK analysis to estimate the relative bioavailability of MTX between the two groups using standard bioequivalence criteria. Additional explorations included estimation of power for the corresponding sample size and influence of PK variability of MTX on the sample size.
Conclusions: It is possible to apply population PK modelling, in conjunction with optimal design and clinical trial simulation to determine the sample size to explore possible drug-drug interactions. The sample sizes determined based on the method proposed are significantly less than that based on the conventional study design using non-compartmental analysis, under same assumptions.