C. Ambery, J. Robertson, D. Austen
Clinical Pharmacology Modelling and Simulation, GSK, UK
Objectives: A test compound is being developed as a potential oral anti-inflammatory. The compound will be investigated for the first time in patients in a long term study. However the tablet manufacturing process for the compound has been modified, and although in vitro data indicates the tablets should perform similarly, given the study duration and known variable PK profile it was considered prudent to perform an interim PK check to de-risk the long term trial. As a first proposal an interim PK assessment was proposed in 16 subjects. The objectives of this work were to determine the minimum number of patients required to give sufficient confidence that exposure is unchanged, the minimum number of patients required to give confidence to dose adjust and to quantify the risk of potential PK exposure changes ahead of study start.
Methods: PK exposure data from 30 healthy subjects was used to calculate the posterior distribution for the PK parameters of interest. The posterior probability that the ‘True’ exposure is within criteria of interest (<0.8, 0.8<x<1.25,>1.25) was determined for a range of potential observed ratios (0.333x to 3x) and sample sizes (N = 2n-1, with n = 1 to 6). The operating characteristics of our decision rule for dose adjust (rule out two-fold change in PK exposure) were determined by simulation to quantify how often a dose adjustment would be proposed if in truth exposure was unchanged. The probability of exceeding PK exposure limits was determined for a range of potential PK changes.
Results: If the true ratio (test-patient/reference-healthy) is 1, i.e. the distribution is the same in both groups, then 8 subjects would provide the team with sufficient confidence they would make the dose adjustment an acceptable percentage of the time. Whereas, if there is great disparity between the groups, very little data is required. For a sample size of 8, if the true ratio (test/reference) is 1, then less than 5% of simulated studies would require dose adjustment. If the true reduction is 20% (ratio = 0.8) then for a sample size of 8 subjects 30% of studies would result in dose adjustment. The probability of exceeding the safety limit in a future subject if the PK exposure is unchanged is 0.11%. The probability of exceeding the safety limit if PK exposure is 50% increased is 8.6%.
Conclusions: By utilising Bayesian decision theory the sample size to provide sufficient confidence to rule out a two-fold change in PK exposure compared with historic data was half that typically studied. Safety risk ahead of study start was quantified.
Reference: PAGE 23 (2014) Abstr 3228 [www.page-meeting.org/?abstract=3228]
Poster: Methodology - Study Design