The Cytel Trial Design Innovations (CTDI) Webinar Series recently hosted a webinar on designing event-based studies. Such studies are essential to designing high-efficiency clinical trials in certain therapeutic areas, but they add a number of challenges to the already complex landscape of adaptive trials.
The webinar was held on January 23rd, featuring Biostatistician and pioneering Bayesian trial-designer Pantelis Vlachos. We had the opportunity to sit down with Dr. Vlachos and speak about innovative trial designs and their benefits, adaptations and interim looks in oncology and cardiovascular, the challenges of designing event-based studies more generally, and how Cytel’s array of software tools, particularly East®, has enabled trial sponsors to fully consider their options in the design of high-efficiency clinical trials.
There is a lot of excitement in the industry around innovative trial design. What are some of the benefits?
This is a high-risk, high-reward industry. The cost of drug development, failure rate and human cost associated with prolonged participation in a trial turn out to be steep in case of an ineffective trial. Innovative trial designs can often answer the research questions in a scientifically valid manner, with fewer study subjects. Overall, I think it goes well with the whole adaptive business.
Typically, when we think about the study there are different parameters for product optimization. For example, the probability of success from the study, financial aspects like expected net present value and the duration of the study. Complex innovative designs actually provide us with the potential to maximize the efficiency of the study and lower the cost of drug development. Which, in the long run, accelerates the access of patients to efficacious drugs and enables the sponsors to become more responsible stewards of patients’ health.
Progress in treatments for cancer has been in the news lately. What makes designing and managing an oncology or cardiovascular trial particularly difficult?
The main issue in the case of these studies is the endpoint. What is measured as the response of a subject is how long it would take to observe an event, like death, to establish the efficacy of the study design. For example, there are studies where you can get to the endpoints faster like in the case of pain studies. You have the drugs administered and you may know within hours if the pain killers have an effect. However, in some of the oncology studies, the primary endpoint is event-based. Which means your endpoint is tied to a particular event, like the time to death or progression time of the disease, and you have to wait for this event to occur. While you are waiting for this event to occur, subjects may drop out or you may experience censoring of observations. This can complicate things quite a bit.
What has Cytel done to help ease these challenges?
With both East® and OK GO, Cytel offers the capability to explore the different types of study designs. We can see through the power of simulation how sensitive those designs are to departures from certain assumptions that are usually made while creating these designs. These assumptions can be in relation to the expected distribution of the subjects, on how the subjects will arrive, any accrual patterns or any drop out patterns it may have. However, these assumptions may not always be correct and so, our software products offer you the capability to test how the departures from these assumptions affect the results and the probability of your success. The Cytel software products provide flexibility with the design that can vary from something simple to anything complex that utilizes some of the more recent adaptive and innovative methods.
Can you share some event-based studies that you have worked on in the past?
I recently worked in the area of triple-negative breast cancer which is a rare disease and requires the use of complex innovative designs. In the case of rare diseases, the regulators are also way more open to having a design or procedure that will utilize as few subjects as possible and could get to efficacious results quickly.
What are some exciting new features in East® that those planning event-based studies can anticipate?
We have worked on providing upgrades in almost all of the features of East®. The structure of East® is modular. There are different modules that deal with normal, fixed, binomial, time-to-event studies, fixed and group sequential designs, and adaptive designs for when you have multiple endpoints. There can be multiple measures that you need to consider or designs for multiple arms when you have multiple treatments compared to a control.
We have made advancements in all of these areas. For example, for time-to-event studies in oncology, we have simulations that allow you to have an interim look where a Go/No-Go criterion based on a surrogate endpoint can be used. This interim look can be based on an endpoint which is different but correlated from your primary endpoint. The primary endpoint in these cases is a time-to-event endpoint, such as overall survival (time to death) or a similar event. For these types of events, they will not lend themselves to having an interim look, conducting either stop for efficacy or futility early on, as you will have to wait for the event to occur.
Typically, you would want to perform one of these interim looks based on an endpoint that is related to the final endpoint. A kind of endpoint that you can observe at a much earlier stage and is known as ‘a surrogate endpoint’. At Cytel, we have provided this capability through simulation and can perform a stop for futility based on such an endpoint. As I mentioned, this (surrogate) endpoint and the final or primary endpoint are correlated, and they will enable the sponsor to have an early exit if the study results are not as expected.
Cytel has also introduced another module in the area of oncology for an adaptive population enrichment. It is an extension of one of our seminal modules in East® that allows us to view sample size re-estimation. It is an adaptive procedure which allows you to increase the sample size midway through the study if your results look promising. Additionally, if the results fall in the enrichment zone, it offers the capability to increase the sample size or focus the remainder of the study only on a subset of the population which has a beneficial prospect to the therapy.
What benefits can the users expect from the new developments in East®?
These new developments allow us to put to design a study that will take into account several interim looks. This will enable you to adjust your study and continue or stop the study based on the results that you have obtained.
You also have the capability to perform different types of adaptations. You can increase your sample size, you can drop an arm and focus only on a certain subset of subject arms or treatment arms, or you can focus on a specific part of the population that seems to have the greatest benefit.
The Cytel Strategic Consulting team outlines the new face of adaptive trials in modern drug development in a white paper.
In a nutshell, we can achieve the goal of increasing the efficiency of a study by accelerating the access of patients to life-altering therapies. Cytel also has plans to have a methodology for multiple arms compared to control, as well as multiple stages, for time-to-event endpoints. Currently, this methodology is only available for normal and binomial types of endpoint. We are also working on incorporating the information from multiple endpoints and not just multiple arms.
In last week’s webinar, Pantelis further explores this topic and describes advanced techniques that have the potential to help you avoid unnecessarily prolonging a study, make better use of limited R&D resources, improve the experience of your patients, and help them more rapidly access effective therapies.
About Pantelis Vlachos
Pantelis is Director/Strategic Consultant for Cytel, Inc. based in Geneva. He joined Cytel in January 2013. Before that, he was a Principal Biostatistician at Merck Serono as well as a Professor of Statistics at Carnegie Mellon University for 12 years. His research interests lie in the area of adaptive designs, mainly from a Bayesian perspective, as well as hierarchical model testing and checking although his secret passion is Text Mining. He has served as Managing Editor of the journal “Bayesian Analysis” as well as editorial boards of several other journals and online statistical data and software archives.