The BEST Platform. A Modular Early-Phase Platform for Seamless Dose Finding and Cohort Expansion Laiya Consulting, Inc. 2018

Transcription

1 The BEST Platform A Modular Early-Phase Platform for Seamless Dose Finding and Cohort Expansion Laiya Consulting, Inc Introduction The Bayesian early-phase seamless transformation (BEST) platform provides a fast, efficient, and powerful solution for early-phase drug development. The BEST platform allows for 1) seamless transition from dose finding (Phase 1a) to cohort expansion (Phase 1b), and to a proof-of-concept (POC) stage (Phase 2a) if needed, 2) simultaneous expansion of multiple doses in multiple indications, 3) interim decision making to graduate or terminate a dose-indication arm adaptively, 4) powerful data analysis for RP2D selection, and 5) flexible selection of different modules tailored for customized design. The statistical innovation of BEST centers at a proprietary Bayesian hierarchical model (BHM). With this model, BEST is able to empower a master protocol for early drug development that incorporates phase 1a and phase 1b into a single trial allowing for highly efficient exploration of efficacious and safe doses in multiple indications. Due to the novel BHM, the BEST platform can improve the overall study power in selecting the promising doses and indications for late-stage drug development and eliminate toxic or inefficacious doses quickly without wasting resources. This leads to increased probability of success for the entire drug development, speeds up the process, and reduces the cost for sponsors. Scheme and Modules Figure 1 presents the framework of BEST s seamless dose finding, cohort expansion and POC. The entire platform offers a suite of innovative and efficient designs for different objectives.

2 Figure 1. The BEST scheme vs the traditional sequential scheme for early phase drug development. Designs offered in BEST 1. The mtpi-2 design (modified Toxicity Probability Interval Design version 2, Guo et al., 2017a): a simple, safe and efficient phase 1 dose finding design, in which all decisions can be transparently tabulated for examination before trial begins. 2. R-TPI (Rolling Toxicity Probability Interval Design, Guo et al., 2018): a fast and efficient phase 1 dose finding trial, that incorporates a rolling enrollment scheme, aiming to accelerate the trial conduct without sacrificing safety and desirability. 3. PITE (Probability Interval Design based on both Toxicity and Efficacy): a transparent, efficient and powerful solution for immuno-oncology (IO) phase 1 dose finding trials, which incorporates efficacy outcomes together with toxicity outcomes to inform dosing decisions to optimize efficacy and safety simultaneously. 4. Dual-agent Drug Combination Dose Finding designs: the state-of-art designs such as the AAA design (Lyu et al., 2018) in which multiple adaptive scheme (e.g. adaptive dose insertion and parallel patient enrollment at multiple doses) can be incorporated. 5. MUCE (MUltiple Cohort Expansion): the key feature of BEST platform that allows for expansion of multiple doses in multiple indications. With the innovative BHM, MUCE is able to save the patient resources without sacrificing the chance of finding the efficacious doses and indications.

3 6. SCUBA (Subgroup ClUster Based Bayesian Adaptive Design, Guo et al., 2017b): a powerful precision-medicine solution for phase 2 subgroup enrichment trials aiming to find optimal subgroups that benefit from precise treatment and increase the probability of success for the whole trial. Other Modules 1. Interim analysis that allows early stopping for futility, superiority, or toxicity. 2. Comparison with different reference rates for different indications. 3. Flexible sample size calculation for the entire master protocol or for each component such as dose finding only, or cohort expansion only. 4. Adaptive or equal randomization across different doses during cohort expansion or POC testing. For more details, see the Figure 2 and contact us to design a trial tailored for your own needs.

4 Figure 2. The list of BEST modules and design choices. Benefits of BEST 1. [Smaller sample size] Compared to the conventional approach (independent frequentist test for each dose-indication arm), the BEST platform is able to save about 20%-30% sample size for the trial without sacrificing the chance of finding the efficacious doses and promising indications. 2. [Higher power] Compared to the conventional approach (independent frequentist test for each dose-indication arm), the BEST platform has a larger (up to twice many) chance to find the efficacious dose and promising indications. 3. [Flexibility] The BEST platform is highly flexible, providing investigators interim analyses that allow early stop of cohort expansion for inefficacious or overly toxic dose. 4. [Time-saving] It accelerates drug development due to the seamless strategy and sample size reduction.

5 Services and Prices (MUCE as an example): Service Content Price (US Dollars) Part 1: Design Consultation 20, a. Tailored simulation study; b. A report of simulation results that demonstrates the operating characteristics of the BEST based on the trial settings and summarizes the design recommendation; c. Meeting to explain the results. Part 2: Trial Services 15, basic; a. Protocol writing (for BEST related); b. Statistical analysis (for BEST related); 2,500.00/analysis for c. Report writing (for BEST related); interim analysis d. Final decision making on RP2D and indications; e. Interim analysis [optional] Decision on early stop for futility, efficacy, etc.; RP2D triggering analysis. Part 3: Meetings and Teleconferences a. With FDA, CDE, PI, and SPONSOR; b. Study training to CRO /hr for CSO; /hrs. for senior+ statisticians. Part 4: Add-on Services: a. Protocol review 1, b. Trial summary report review 1, c. Adaptive randomization for patient allocation; Per patient cost d. Other adaptation scheme (e.g. sample size re-estimation) /hr Note: for other products listed above or a novel brand-new design tailored for your trial, the services and prices for Parts 1 and 2 will vary. Contact us to get detailed information. References: 1. Guo, W., Wang, S. J., Yang, S., Lynn, H., & Ji, Y. (2017a). A Bayesian interval dosefinding design addressing Ockham's razor: mtpi-2. Contemporary clinical trials, 58, Guo, W., Ji, Y., & Li, D. (2018). R-TPI: Rolling Toxicity Probability Interval Design to Shorten the Duration and Maintain Safety of Phase I Trials. Journal of Biopharmaceutical Statistics. 3. Lyu, J., Ji, Y., Zhao, N., & Catenacci, D. V. (2018). AAA: triple adaptive Bayesian designs for the identification of optimal dose combinations in dual agent dose finding trials. Journal of the Royal Statistical Society: Series C (Applied Statistics). 4. Guo, W., Ji, Y., & Catenacci, D. V. (2017b). A subgroup cluster based Bayesian adaptive design for precision medicine. Biometrics, 73(2),