European Statistical Forum - Programme

The aim of this session is to:

• Understand the progress made by EMA in its ACT-EU initiative focusing on Priority Action 8 (PA8) on methodology.
• Highlight the importance of promoting innovative statistical approaches and providing guidance to support uptake and implementation in clinical trials.
• Identify areas where additional guidance in methodology could advance and impact drug development.

Innovation in clinical trials has been a significant focus in recent years. Various guidance documents and publications have emerged on novel designs and advanced methods to help accelerate drug development.

The Accelerating Clinical Trials in Europe (ACT-EU) initiative launched in early 2022 defined 10 priority areas including Methodologies. Perspectives will be shared on the progress made in ACT-EU and key areas of focus expected in the next few years. Innovation requires taking a new invention and commercializing it so it can be broadly implemented and becomes fully adopted. Perspectives will be shared on why promoting innovative statistical approaches is vital to accelerate drug development and what activities are planned to support implementation in clinical trials.

Finally, regulatory agencies have released work plans on what new guidance will be developed and where existing guidance will be updated. Perspectives will be shared on how these activities will help to advance and impact drug development including opportunities to contribute to these activities.

Abstract available soon

Platform trials are implemented to assess multiple research questions under a common operational and inferential infrastructure in an efficient way. The key research question of interest is typically a question regarding the efficacy and safety of investigational interventions vs. a common control arm. Those investigational interventions may all come from different intervention owners, or from the same company. Platform trials affect the interests of numerous stakeholders, including patient communities, investigators, regulators and certainly the intervention owners. As such, designing platform trials results in a complex process including discussions on design requirements and properties among all those stakeholders. This design process can be well supported by clear articulation of individual perceived benefits, challenges, and their prioritization for the design problem at hand. Clinical trial simulations support in addition a thorough understanding of operating characteristics, allowing for informed decision-making on study design elements and their value in addressing stakeholder needs. Focus of this presentation will be a cross-functional platform trial design assessment framework. The proposed generic framework helps in identification of design optimization opportunities to maximize the value of eventual platform trials for all stakeholders involved. The framework addresses real and perceived concerns and limitations on platform trials. Using an example case of a platform trial, we’ll discuss how the cross-functional assessment framework enables focused trial design development.

In this talk we will discuss the impact of methodological aspects on the robustness and credibility of derived data from platform trials in view of regulatory decision making.

The talk will cover randomization aspects (e.g. changing allocation ratios over time due to opening and closing of arms, response adaptive randomization, and 2-stage randomization), the use of shared controls and the need for type 1 error control.

The impact of shared controls on type 1 error and blinding will also be discussed and recommendations will be given where possible. In this regard, we will share some recent findings on the acceptability of platform trials from an analysis of scientific advice procedures at the EMA (Nguyen et al, 2024) and a simulation study on the impact of data leakage in survival platform trials on type 1 error (Nguyen et al, 2025).

Finally, the current status of the EMA reflection paper on methodological aspects of platform trials will be shared.

References:
• Nguyen QL, Kästel H, Hees K, Hofner B (2024). The Use of Complex Clinical Trials: A Regulatory Review. Submitted.
• Nguyen QL, Posch M, Hofner B, König F (2025). Impact of information leakage in platform trials with survival endpoints on type I error control. In preparation.

Basket trials are increasingly used for the simultaneous evaluation of a new treatment in various patient subgroups. Eligible patients typically share a commonality (e.g., a genetic aberration or clinical symptom), on which the treatment may potentially improve outcomes. Sophisticated analysis models, which feature borrowing of information between subgroups, have been proposed for enhanced estimation of the treatment effects. Yet the development of methods to choose an appropriate sample size appears to fall behind. A widely implemented approach is to sum up the sample sizes, which are calculated as if the subtrials are to be carried out as separate studies.

In this talk, I will introduce a novel Bayesian modelling strategy that characterises the complex data structure of basket trials.
More specifically, this approach reflects considerations that the treatment effect in some subtrials may be more commensurate between themselves than with others. Closed-form sample size formulae are further derived to enable pairwise borrowing of information between commensurate subtrials.

Our approach ensures that each subtrial has a specified chance of correctly deciding whether the new treatment is superior to or not better than the control by some clinically relevant difference. Given fixed levels of pairwise (in)commensurability, the subtrial sample sizes are solved simultaneously.

This solution resembles the frequentist formulation of the problem in yielding comparable sample sizes for circumstances of no borrowing. When borrowing is permitted, a considerably smaller sample size is required. I will illustrate the application using data examples based on real basket trials. A comprehensive simulation study shows that the proposed approach can maintain the true positive and false positive rates at desired levels. Added efficiency through mid-course modification to basket trials, such as sample size reassessment, will be discussed.

Perspectives will be also given on future methodology development, especially on the multiplicity correction, for basket trial designs.

Umbrella trials evaluate multiple targeted treatments in a single disease setting, each targeting a unique disease subgroup. To date, umbrella designs are commonly designed and analysed as a series of independent subtrials under a master protocol framework, thus providing only operational efficiency gains.

We propose a two-stage Bayesian adaptive umbrella design that permits borrowing of information across the common treatment arms of a randomised umbrella trial. Specifically, we leverage the benefits of information borrowing using a commensurate prior approach that we have previously proposed in the basket setting. The design features adaptive assignment in favour of an experimental treatment after the first interim analysis (where borrowing is permitted) if it outperforms the control. The Bayesian predictive power is used to decide how far a deviation from equal allocation is considered in the second stage.

Our simulation results show that the proposed Bayesian adaptive design can increase allocation to experimental arms whilst maintaining desirable statistical power and error rate control. As such, we demonstrate substantial statistical efficiency gains through information borrowing and adaptive allocation in the umbrella setting.

Malaria remains a major global health challenge, with an estimated 263 million infections annually, leading to approximately 597,000 deaths, predominantly in children under 5 years (WHO, 2023). Addressing the urgent need for innovative treatments, the PLATINUM trial investigates novel anti-malarial agents in patients with uncomplicated malaria, emphasizing learning across four critical dimensions to overcome key challenges in drug development.

The first dimension focuses on treatment regimen optimization, progressing from monotherapy in adults to combination therapies in children. The second dimension addresses stepwise population expansion, enabling a structured transition from adults to children while ensuring safety and efficacy.

The third-dimension leverages Bayesian analyses, using informative priors to integrate existing knowledge and data from earlier trial stages. This approach reduces the sample size required, accelerating the pathway to actionable insights. Finally, the fourth dimension emphasizes efficient evaluation of multiple compounds, enabling parallel and quick assessment of novel agents and combinations, aligned with the WHO vision of a “world free of malaria.”

Through this adaptive, multi-part platform study, the PLATINUM trial provides a robust framework for rapidly and effectively developing new treatments, in the light of growing resistances to existing therapy.

As clinical trials embrace innovative approaches like Bayesian adaptive designs and master protocols, the array of design choices can be daunting. This presentation explores the nuances of developing and simulating adaptive platform trials, emphasizing the evolving role of statisticians in this dynamic framework.

Using an adaptive platform trial currently in development as an example, the talk will demonstrate how statisticians have transitioned from a focus on traditional data analysis to becoming pivotal contributors in shaping trial designs. With increasing design complexity, extensive simulations are now indispensable, enabling statisticians to play a critical role in refining trial methodology.

The session will provide an overview of adaptive platform trials, including their terminology, associated risks, and benefits. It will showcase how simulations are performed using OCTOPUS, a freely available R package (https://kwathen.github.io/OCTOPUS/), enhanced with trial-specific adaptations, to accommodate the flexibility of Bayesian analysis. Several visual tools will be presented to illustrate trade-offs and assess the frequentist performance metrics of a Bayesian design.

The presentation highlights how intuitive visualizations and cutting-edge simulation techniques can support teams and stakeholders in understanding trial workflows and predicting outcomes across diverse scenarios. By fostering informed decision-making, these methods aim to optimize trial design, saving time and resources while ultimately improving the design and understanding of expected performance.

Adaptive platform trials offer a flexible approach to drug combination development.

The presentation focuses on the Piranga adaptive platform trial, a case study in drug combination development for infectious diseases, specifically Hepatitis B Virus (HBV). It contrasts traditional development methods with the adaptive platform design, highlighting advantages in assessing multi-agent treatments and combination therapies.

The presentation highlights efficiencies, and benefits of an adaptive platform design, while also acknowledging the challenges in implementing such a design.

Ultimately, the presentation argues that adaptive, data-driven platform designs are more nimble, and effective compared to traditional approaches.