Adaptive Designs provide greater efficiency and flexibility for Investigators/pharmaceutical companies by minimizing the number of patients in each development phase and reducing the time needed for the whole process.
Study design is early phases is extremely important. Safety and efficacy, ethical considerations and long patient recruitment all need to be taken into account. Patient recruitment retention is challenging, and biostatisticians should be involved in the beginning to define protocol requirements. There will generally be vast amounts of data to analyze, and therefore the biostatistician will need access to real-time data in order to make go/no-go decisions. Investing time in the proper design and set-up of an oncology trial in the early phases is essential to increase success rates in later phases.
In Phase I trials, the 3+3 design is the most commonly used design to evaluate the highest dosage with acceptable toxicity. But this design can be very insufficient in terms of number of patients and/or precision of the dosage. Bayesian designs like the Continuous Reassessment Method and its modifications offer the chance to reach the Dose Limiting Toxicity (DLT) as well as the advantage of treating more patients with a possibly effective dosage.
While the 3+3 design can be evaluated by physicians themselves, the CRM Methods need the support of a statistical program, which might be a drawback.
The goal of Phase II development is to examine the preliminary evidence of a new compound and exclude compounds found to be ineffective from further research. Adaptive designs, like the Simon’s 2-stage design, have been used for a long time in Phase II.
A patient’s hetergeneity limits the evidence of study results. To overcome this problem, Randomized design methods are used more frequently whenever the number of patients available allow. Randomized “Parallel Non-comparative Regimens” compare different treatment regimens, characterized either by different dosages or different application intervals, in an effort to avoid various types of bias, including patient selection bias.
In contrast to other indications where Phase I trials are often conducted with healthy volunteers and patients with the targeted disease are only involved in Phase II, in Oncology diseased patients are already involved in Phase I. Flexible 2-stage designs have been proposed (Seamless Phase I/II trials) which can minimize the number of patients needed to come to a “go/no-go” decision.
Looking for biostatistics methodology and design support for an upcoming oncology study? Contact one of our experts at email@example.com