Table 4 Graves-PCD

Graves-PCD (ISRCTN81162400) is an early phase dose-ranging study coordinated by Newcastle CTU. It is testing four doses of daratumumab against placebo for the treatment of severe Graves’ disease, an autoimmune disorder of the thyroid. The final design involves up to 30 participants will be randomised, split into two stages. After 15 participants (3 per arm) have had primary outcome assessed (change in Serum TRAb antibodies from baseline to 12 weeks) an interim analysis will be conducted. Up to two doses of daratumumab and placebo will continue in the second stage, with the selected doses dependent on a three-parameter Emax model [27] fitted to stage 1 outcome data. Prior to the final design being decided, possible considerations for each of the steps are provided below:

Step 1: As an early phase CTIMP, substantial regulatory oversight is required regardless of the design used. The trial required regulatory approval and robust procedures including pharmacovigilance and data monitoring. With a low total number of participants, data errors could cause disproportionate impact on data monitoring. As a dose-ranging study, the final analysis will be statistically complex

These factors mean the CTU and statistical resource required is likely to be high even without an AD. Using a template costing tool allows mapping this out

Step 2: An AD that allows updating the stage 2 doses based on stage 1 outcome data has several consequences on trial tasks: (1) some set-up tasks such as protocol development, ethical, and regulatory approval, creating PIS, specification and testing of CDMS and randomisation systems might be more complex; (2) an interim SAP would need developing, at least one interim analysis conducted and any adaptations implemented in a timely manner; and (3) the final statistical analysis and reporting of the trial would be more complex

Step 3: More statistical time is required for developing an interim SAP and to conduct interim analyses. More data management time is needed to provide clean data in a timely manner for the statistician at interim analyses. At the time, Newcastle CTU was outsourcing CDMS and randomisation to a third party, and the prospect of a more complex design could mean a higher fee for these services and more staff resource required for testing. More trial management time is required for more complex set-up tasks

The AD initially proposed would allow doses post-interim to depend on results up to that point, leading to uncertainties on which doses are needed and implications on pharmacy support. If stage 2 occurred, the randomisation system requires updating. If there was no evidence of dose–response after stage 1, the trial would stop early, and this would influence the resources required to end the trial (lessening them but still requiring some to close and report the trial)

Step 4: The trial was being submitted to MRC DPFS [28] for funding. The panel requires clear description of the statistical properties of the trial, meaning initial statistical simulations were required prior to submission. MRC DPFS requires specification of milestones with staged funding, so having the interim analysis completion as a milestone would allow consideration of the costs incurred at different parts of the trial

Step 5: After considering the timelines, costs, and statistical properties it was decided that a single interim analysis would provide most of the benefit possible from the approach without additional costs of more interim analyses