[Paper Review] Decision rules for identifying combination therapies in open-entry, randomized controlled platform trials

Elias Laurin Meyer, Peter Mesenbrink+4 more|arXiv (Cornell University)|Mar 17, 2021

Statistical Methods in Clinical Trials44 references26 citations

TL;DR

This paper proposes decision rules for identifying effective combination therapies in open-entry, randomized controlled platform trials using Bayesian and frequentist methods. It demonstrates that data sharing, decision rule specification, and prior assumptions about treatment efficacy critically influence operating characteristics such as power, type I error, and trial efficiency, with implications for trial design and stakeholder alignment in drug development.

ABSTRACT

Platform trials have become increasingly popular for drug development programs, attracting interest from statisticians, clinicians and regulatory agencies. Many statistical questions related to designing platform trials - such as the impact of decision rules, sharing of information across cohorts, and allocation ratios on operating characteristics and error rates - remain unanswered. In many platform trials, the definition of error rates is not straightforward as classical error rate concepts are not applicable. For an open-entry, exploratory platform trial design comparing combination therapies to the respective monotherapies and standard-of-care, we define a set of error rates and operating characteristics and then use these to compare a set of design parameters under a range of simulation assumptions. When setting up the simulations, we aimed for realistic trial trajectories, such that e.g. a priori we do not know the exact number of treatments that will be included over time in a specific simulation run as this follows a stochastic mechanism. Our results indicate that the method of data sharing, exact specification of decision rules and a priori assumptions regarding the treatment efficacy all strongly contribute to the operating characteristics of the platform trial. Furthermore, different operating characteristics might be of importance to different stakeholders. Together with the potential flexibility and complexity of a platform trial, which also impact the achieved operating characteristics via e.g. the degree of efficiency of data sharing, this implies that utmost care needs to be given to evaluation of different assumptions and design parameters at the design stage.

Motivation & Objective

• To define and evaluate error rates and operating characteristics in open-entry, exploratory platform trials comparing combination therapies to monotherapies and standard of care.
• To assess how decision rules, data sharing strategies, and assumptions about treatment efficacy impact trial performance.
• To guide the design of efficient, reliable platform trials in complex therapeutic areas like NASH, where multiple combination therapies are evaluated simultaneously.
• To support regulatory and clinical decision-making by quantifying trade-offs between statistical power, type I error, and trial duration under realistic simulation conditions.

Proposed method

• The study employs a Bayesian hierarchical model with dynamic borrowing of information across cohorts to estimate treatment effects.
• Decision rules are defined using posterior probabilities (e.g., P(θ > 0) > γ) and non-inferiority margins (δ < 0) for pairwise comparisons.
• Simulations are conducted under a stochastic cohort entry mechanism, where the number of cohorts and treatment arms is not fixed a priori.
• The platform trial design includes four arms per cohort: combination therapy, backbone monotherapy (A), add-on monotherapy (B), and standard of care (SoC).
• Interim analyses are performed after each cohort’s enrollment, with stopping rules for futility or efficacy based on predefined thresholds.
• The simulation framework allows for 100+ million platform trial trajectories and evaluates over 10,000 parameter combinations to assess operating characteristics.

Experimental results

Research questions

• RQ1 How do different decision rules (Bayesian vs. frequentist) affect the type I error rate and statistical power in platform trials for combination therapies?
• RQ2 To what extent does data sharing across cohorts influence the operating characteristics of a platform trial, including type I error and power?
• RQ3 How do assumptions about treatment efficacy and allocation ratios impact the efficiency and reliability of combination therapy identification?
• RQ4 What is the impact of cohort entry dynamics and stochastic recruitment on final sample size and trial duration?

Key findings

• The method of data sharing significantly affects operating characteristics, with greater borrowing improving power and reducing type I error when assumptions are correctly specified.
• Decision rules based on posterior probabilities (e.g., P(θ > 0) > 0.95) provide better control of type I error and higher power compared to simpler frequentist thresholds.
• Assumptions about treatment efficacy—especially the true response rate of monotherapies and combination therapies—have a strong influence on the probability of correctly identifying effective combinations.
• The inclusion of more cohorts increases the complexity of data sharing and can lead to higher final sample sizes, especially when concurrent data sharing is used.
• The trial design is most efficient when allocation ratios are balanced and decision rules are finely tuned to the expected treatment effect sizes.
• The simulation framework reveals that even small deviations in prior assumptions or decision rule thresholds can lead to substantial changes in operating characteristics, highlighting the need for robust design-phase evaluation.