CT PA Director of Biostatistics
Complex Innovative Designs are emerging within many clinical trial protocols. These designs often require adaptations and flexibility for the randomisation, similar to Adaptive Designs. Before entering the world of Complex Innovative Designs, it is important to understand the basic concepts of Adaptive Designs and how they may impact randomisation.
What are Adaptive Designs?
An Adaptive Design allows for a trial to adapt mid-study. These adaptations are planned and specified in the study’s protocol. The FDA defines adaptive designs as “a clinical trial design that allows for prospectively planned modifications to one or more aspects of the design based on accumulating data from subjects in the trial” in their 2019 guidance document Adaptive Designs for Clinical Trials of Drugs and Biologics.
What are Common Adaptations that Impact Randomisation?
Common adaptations that impact randomisation include:
- Introducing new treatments
- Dropping existing treatments
- Pausing and restarting treatments
- Adjusting treatment allocation ratios or assignment probabilities
- Sample size readjustment
How is Randomisation different from a Traditional Design compared to an Adaptive Design?
In a traditional randomisation design, the included treatments and allocation ratio are usually the same for the entire study. For instance, subjects are randomized to one of two treatments in an equal allocation ratio (1:1). One randomisation list is generated and used for subject randomisation across the study’s duration.
Whereas in an Adaptive Design, the randomisation design may change depending on the planned adaptations. For example, a study starts with three treatment groups assigned in a 1:1:1 ratio. Then based on the results of an interim analysis, one of those treatments could be dropped or a fourth treatment could be added. Therefore, the randomisation could change to either assigning in a 1:1 ratio or a 1:1:1:1 ratio. One randomisation list that is fixed with three treatments in a 1:1:1 ratio would not support the possible adaptations. Additional randomisation lists or some type of flexible randomisation would be needed.
What are Complex Innovative Designs?
Complex Innovative Designs are similar to standard Adaptive Designs, but typically include additional layers of complexity. Complex Innovative Designs have benefits such as increased flexibility and efficiency in drug development, the ability to share control arms, central electronic data capture systems and patient centricity. They are usually able to identify treatments that are effective and ineffective quicker than traditional trials.
Complex Innovative Designs include (but not limited to):
- Master Protocols (Basket, Umbrella, Platform)
- Biomarker-Targeted Treatments
- Bayesian Response Adaptive Randomisation
- Complex Dose Ranging / Dose Finding Cohorts
How do Complex Innovative Designs impact the Trial’s Randomisation?
Complex Innovative Designs that include randomisation adaptations are similar to standard Adaptive Designs, with additional dimensions of complexity. Think of standard Adaptive Designs as one dimensional where the same adaptations are applied to the entire study. If a treatment were dropped, it would be excluded from the study’s randomisation list. Whereas Master Protocols are studies that have multiple dimensions such as different subgroups, sub-protocols, sub-studies, etc. In this Complex Innovative Design case, a treatment may be dropped in one subgroup’s randomisation list, and in another subgroup, a treatment may be added. The adaptations need to be managed independently for each dimension’s (subgroup) randomisation.
Another example is the case of biomarker-targeted treatments, where only subjects who are biomarker-positive are eligible for that biomarker-targeted treatment, and those who do not have the biomarker are ineligible. Here, there is varying eligibility based on the included treatments that may or may not target specific biomarkers. The varying eligibility needs to be managed to allow or not allow assignment based on biomarker-targeted treatments and biomarker presence.
Due to the complexity involved, the implementation and management of randomisation for these types of designs often require a sophisticated Interactive Response Technology (IRT) system.
What is Successful Implementation for Randomisation of Complex Innovative Designs?
Successful implementation of Randomisation is all about being able to execute the adaptations with minimal disruptions since a key characteristic of Complex Innovative Designs is efficiency in drug development. Take the case of a study implementing a single randomisation list with the initial study parameters (e.g., included treatments / ratio) without accounting for any possible adaptations. With this set-up, if a new treatment is introduced, then a new randomisation list needs to be created and imported into the IRT system. While this approach works, it is disruptive since it incurs time and effort. A more efficient approach would be if the IRT included a flexible randomisation scheme with user ability to enter in the adaptations in real-time without having to create a new list.
What is the Optimal Level of Flexibility for a Complex Innovative Design’s Randomisation?
Every study is different, which means that each study’s optimal level of flexibility also differs. Planned adaptations are specified within the study’s protocol. Some protocols may explicitly detail the adaptations (e.g., which treatments could be added, which treatments could be dropped), while others may not. For example, a protocol states that new treatments can be introduced throughout the study’s duration as they are discovered. Since they are yet to be discovered, they cannot be explicitly specified in the initial protocol and will be included in an amendment once identified. If new treatments are expected, but not yet identified, the IRT randomisation could be built as flexible to allow new treatments to enter into the scheme dynamically through a user interface. If the protocol specifies that the ratio(s) can be adjusted, that should be incorporated into the IRT’s randomisation as well.
What is Recommended to Achieve the Optimal Level of Flexibility for a Complex Innovative Design’s Randomisation?
To be able to achieve the optimal level of flexibility, it is recommended to begin discussions on randomisation implementation early in the planning phase of the protocol. These discussions should include the key stakeholders such as the study’s clinical operations leaders, program managers, biostatisticians, clinical supply managers, along with the relevant vendor partner roles such as IRT’s biostatisticians, project managers, design, and programming experts. It is equally important to choose an IRT vendor that has the experience and expertise in implementing these complex innovative designs to be effective consultants and partners rather than simply order-takers.
Almac Clinical Technologies has the attributes necessary to successfully implement flexible randomisation for Complex Innovative Designs. Almac has a Biostatistics team that is 100% focused on IRT randomisation, and an Adaptive Design Center of Excellence comprised of cross-functional (development, testing, QA, design, project/program management) expertise to ensure every area is considered in achieving the optimal level of flexibility.