Registration for a live webinar on 'Chronic inflammation, immune cell trafficking and anti-trafficking agents' is now open.See webinar details
Design early phase drug combination trials: software
Published on May 29, 2017 33 min
Other Talks in the Series: Adaptive Clinical Trial Design
Phase II clinical trials - Bayesian methods
- Prof. Fei Ye
- Vanderbilt University Medical Center, USA
Hello, I am Ying Yuan. I'm a professor in Biostatistics at MD Anderson Cancer Center. In the first part, we have talked about the three designs to find a single MTD or MTD contour. So in this part we're going to focus on how to implement those three designs using our existing software. So we are going to go through each of them and show you how to use the software to design specific trial examples.
So right now, we switch to the software. So, so far we have talked about several designs to find the MTD or MTD contour. And from a practical point of view, the important thing is that we have the software to implement the design, and how can we use it in practice. So the important reason we chose those three designs is because it is very easy to use the software for those designs. And essentially they can be implemented using the R package "BOIN" available from the CRAN. And then in addition, if you don't like R, there is also Standalone, a graphic user interface-based software available from MD Anderson Biostatistics Software downloaded from the website. Here I give the link, if you are interested, you can download it. And in addition, we also provide a statistical tutorial and protocol template. So this is listed on my website - you can check it out.
Okay, so right now we are going to demonstrate how to use the software to design combination trials, step-by-step. The first design we are going to talk about is the linearization design. So this is the design used to find a single MTD for the combination trial. And because this linearization design uses the BOIN design, it is very simple, because essentially the first thing you need is to get a boundary. So recall, for this BOIN design, what you really need is that escalation boundary and a de-escalation boundary. Once you have those two boundaries, you can make a decision of dose assignment. So that is why the first function is "get.boundary." It is used to generate escalation and de-escalation boundaries So once we have a boundary, we can conduct a trial. Once we finished the trial, the only thing we need is to select the MTD. So that's why we have a second function called "select.mtd." So essentially what it does is it helps you select MTD at the end of trial based on isotonically transformed estimates. So, those two functions are enough to conduct the trial in practice, but when we prepare a protocol a lot of times we have to show the operating characteristics of the design. So essentially, we want to do a simulation to show the performance of the design. So that's why we have a third function called "get.oc." So this is used to generate the operating characteristics, you can use it for simulation purposes.