Learn what impact better trial design could have on your disease area and how our trial design software can help you explore different possible designs.
This presentation introduces some key elements of the Tessella Berry FACTS software, showing the Dose-Finding component for a Phase II trial where the key adaptive endpoint is a continuous measure. For the purpose of the example, a fictional diabetes trial is designed measuring reduction in Fasting Blood Glucose.
Aspects covered include dose-response modeling with a Normal Dynamic Linear Model (NDLM), longitudinal modeling to make use of earlier visit data where it is considered predictive of the final endpoint, the setting up of various scenarios for the simulator and the concept of specifying different targets such as finding the dose nearest to the ED90.
Read the white paper for an in depth examination of the design choices for a HIV trial.
The trial objective is to study 4 different doses of the study drug, comparing their effectiveness to a ‘standard of care’ control arm. The clinical team want to find the dose with the best tolerability and response and whether it has a good chance of being able to be successfully tested in subsequent phase 3 trial.
The primary endpoint will be ‘no detectable viral load’ after 3 months and the treatment tolerated for the full three months. The outcome is dichotomous, subjects starting with observable viral load at baseline and monitored after 1, 2 and 3 months. Effect and tolerability endpoints could have been analyzed separately but here they are combined in a single score.
This is an example of a trial design for a phase 2 trial in Alzheimer’s. The primary endpoint, which will be used for final decision-making and to drive the adaptation, will be the change from baseline in subjects’ ADAS-cog score after 12 weeks.
It is determined that the minimum clinically significant difference (CSD) compared to placebo would be a 1.75 points greater reduction in ADAS-cog score. The clinical team believe that a credible good performance of the compound would be to achieve 2.2 points greater mean reduction than the control arm.
The aim of the trial is to study 6 different doses of the study drug, comparing their effectiveness to a ‘standard of care’ control arm and determine if there is a good chance of a response that exceeds that on control by the CSD. In this trial there is not the time or number of subjects to study long term tolerability or safety, so, to inform our choice of dose (or doses) for phase 3, we wish to estimate the minimum efficacious dose (MED) – that is, the lowest dose that beats control by the CSD.
This is an example of a comparison of two phase 1 trials: the CRM vs 3+3 in an Oncology phase 1 setting. Tessella Berry FACTS has been used to establish this comparison.
The principal aim of an Oncology phase 1 trial is the identification of the Maximum Tolerated Dose (MTD). The majority of oncology phase 1 trials are run using a design called “3+3. Under this design, subjects are treated in cohorts of 3, and based
on the number of dose limiting toxicities seen in that cohort, decisions on which dose to give the next cohort and whether to stop the trial are made.
Reiner at al argue that this design does not give very good operating characteristics. “the probability of recommending the [correct] MTD at the end of the trial … never exceeds 44% and is most often closer to 30%”. Despite this, the 3+3 design remains in common usage due to its simplicity, the straightforward (and appealing) nature of its decision rules and its familiarity. Possibly one other reason is the difficulty of doing a lot better. In this essay we look at using the oft touted alternative to the 3+3, the Continuous Reassessment Method (CRM) proposed by O’Quigley in 1990 and given some modest but important design tweaks by Goodman in 1995.