Cytel’s New Horizons Webinar Series introduces you to the latest innovations in statistical trial design. This webinar from the series is presented by Dr. Yuan Ji, a consultant for Cytel. Yuan is the founder of Laiya Consulting and currently is the Professor of Biostatistics at The University of Chicago. In his presentation, Professor Ji introduces the U-Design version 1.4, which mainly consists of a new module of dose-finding trial designs with joint efficacy and toxicity outcomes.
Click the button to register for the next webinar in this series, presented by Cytel's Ursula Garczarek. Ursula will be presenting a case study on the value of detailed clinical trial simulations for rare diseases.
U-Design is a comprehensive Software as a service (SaaS) platform used for trial designs, developed by Laiya Consulting. It offers head-to-head comparison of trial designs based on massive simulations. The software also provides Intelligent reports for submission packets and trial protocols, and a comprehensive user manual with in-depth descriptions of the statistical models.
The webinar introduces U-Design version 1.4, which mainly consists of a new module for Single-agent cohort-based dose-finding trials with efficacy and toxicity endpoints. This new module includes five advanced Bayesian designs, all of which aim to identify the optimal biological dose (OBD) instead of the maximum tolerated dose (MTD). The OBD is a dose with tolerable toxicity but maximum efficacy. Whereas, in MTD, toxicity is the primary endpoint and is used for patient allocation and decision-making during the trial.
Eff &Tox Dose-Finding Module: Designs with Binary Eff & Tox Endpoints
In the Toxicity-only dose finding designs, we are familiar with designs such as 3+3, CRM and mTPI. All these designs use the binary toxicity outcomes. The Eff &Tox dose-finding designs use joint efficacy and toxicity outcomes. The five designs incorporated in this module are:
1. Ji3+3 (Lin and Ji, 2020a) – A model-free design which is transparent to physicians and is simple to implement. It is observed to perform well compared to other designs.
2. PRINTE (Lin and Ji, 2020b) – It is built upon the previous work in TEPI design and does not require a physician-elicited decision table. Like Ji3+3, PRINTE is also transparent to physicians and is simple to implement.
3. TEPI (Li et al., 2016) - A natural extension of mTPI by adding the efficacy interval into the dose-finding model. It is simple and transparent, with a pre-specified decision table.
4. EffTox (Thalland Cook, 2004) – An outcome-adaptive, model-based Bayesian design. It models the dose-efficacy and dose-toxicity relationship respectively using two different dose-response curves. It can also evaluate the desirability of a dose by a family of contours characterizing the trade-off between Efficacy and Toxicity.
5. UBOIN (Zhou et al., 2019) – A model-based design that jointly models toxicity and efficacy using a multinomial-Dirichlet model. It employs a utility function to measure dose risk-benefit trade-off.
The new module follows the familiar user interface in existing U-Design modules and allows users to compare multiple designs side by side via simulations. Advanced visualization allows users to examine the simulation results and operating characteristics, so that the best design can be selected for each trial. The platform provides Intelligent report that can be downloaded to speed up the report/protocol writing for users. Additionally, it provide decision tables generation and download, as well as a function for OBD estimation at the end of the trial.
Click the button to watch the detailed demo and review of U-Design v1.4 by Yuan Ji.
U-Design will continue to evolve and serve the community and we welcome feedback of versions 1.3 (basket trials) and 1.4 (eff/tox dose-finding).