Utilizing of Platform Clinical Trial to Help Make

Utilizing of Platform Clinical Trial to Help Make Faster Decisions ASA Biopharmaceutical Section Regulatory. Industry Statistics Workshop J. Kyle Wathen Director Statistical Modeling and Methodology Rhonda Fenwick, Time is Now I Through her art, Rhonda has explored psoriasis, a chronic skin disorder she has lived with since the age of six.

Outline § § Very brief introduction to platform trials Bayesian decision framework Package website Future plans 2

Platform Trial - Introduction § An experimental infrastructure to evaluate multiple treatments and/or combinations of treatments in heterogeneous patient populations. • Not all interventions are included, or even known, at the start of the platform • Utilize a Master Protocol – no compound specifics • New compounds are added with an Intervention Specific Appendix (ISA) - details about the compound § Key benefits include – improvements in patient recruitment and borrowing of patient data across ISAs 3

Typical Timeline - POC Compound 5 Compound 4 Compound 3 Compound 2 Compound 1 Jan 2020 Start Platform Jan 2022 Jan 2024 Jan 2026 4

Sharing Information Between ISAs ISA 2 ISA 1 Start Platform 55

Sharing Information Between ISAs ISA 3 ISA 2 ISA 1 Start Platform 66

Sharing Information Between ISAs ISA 4 ISA 3 ISA 2 ISA 1 Start Platform 77

Sharing Information Between ISAs ISA 5 ISA 4 ISA 3 ISA 2 ISA 1 Start Platform 88

Design and Simulation § Many similarities between ISAs – Each ISA may require different designs, eg single vs multiple doses § One ISA can impact another – Recruitment – Placebo/Control sharing § ISAs may be concurrent or consecutive § Currently no commercial software or package that can accomplish all of the requirements § Started with R code for first platform trial then extended and started a package in R for the second platform 9

Bayesian Go-No Go Decision Rules § § Dual Criteria for each Outcome MAV – Minimum Acceptable Value, TV Target Value m. P, m. E parameter of interest for P and E, respectively Calculations – Pr( d = m. P - m. E > MAV | data ) > LB – Pr(d = m. P - m. E > TV| data ) > UB § Cutoffs: LB, UB Pr(d > MAV | data ) ≥ LB Pr(d > MAV | data ) < LB Pr(d > TV | data ) ≥ UB Graduate Indeterminate/Continue Pr(d > TV | data ) < UB Indeterminate/Continue Drop

Pr( d > MAV | Data ) Decision Making Pr( d > TV | Data ) Pr(d > MAV | data ) ≥ 0. 9 Pr(d > MAV | data ) < 0. 9 Pr(d > TV | data ) ≥ 0. 05 Graduate Indeterminate/Continue Pr(d > TV | data ) < 0. 05 Indeterminate/Continue Drop

Combining Outcomes Design 2 Outcome 1 Design 1 Graduate Continue Drop Continue Graduate Outcome 2

Why R Package for Platform Simulation? § OCTOPUS – Optimize Clinical Trials on Platforms Using Simulation § Not all features are available commercially –desire to test and reuse work if possible § Having the R source code allows for new options and extensions § Transparent to the nature of what is going on in the trial § Many statisticians are familiar with R § Freely available at – https: //kwathen. github. io/OCTOPUS 13

OCTOPUS - Key Features § § § § Random or fixed entry times for ISA Each ISA can have different modeling and decisions Information sharing across ISAs, each ISA can be different Monitoring of ISA can be setup as minimum information with subsequent analysis at predetermined interval or the amount of information needed for each interim analysis Flexibility to add new patient simulator, analysis, randomizers, ect Any number of outcomes Some default plots can be created Covariates and subgroup specific decisions is in development 14

R Package + Project Specific Files R Package Project Specific Core components Built on generic functions Tested Generalized functions from projects Community driven development in future versions Define trial design element Define simulation design element Define any project specific functions Key Advantages – Tested code, reuse general parts, speed up development, learn across projects, project details remain in the project specific files, extendable, generic concepts can be moved from projects to package 15

16

Shiny App Compare Recruitment § Quickly compare options of 2 POC vs Platform with 2 ISAs in terms of recruitment Image or graphic goes here 17

Image or graphic goes here 18

19

Conclusion and Future Plans § Have had positive feedback from several people on wanting to use the package § Have utilized the package to simulate more than 5 platform studies and 3 non-platform § Developers guide § Task list § Streamline the simulation and creation of results for “standard” output § Training – In person and videos – Training for someone use the package and for anyone interested in helping to extend the package 20

Thank You! Kyle Wathen kwathen@ITS. JNJ. COM 21

Backup 22