Study Design Considerations for Phase IIA Proof-of Concept Trials

How do you stop a failing treatment early, before it costs you patients and resources?

Phase IIA trial design is one of the most consequential decisions in early drug development, since it determines how quickly a sponsor can stop an ineffective treatment before committing further patients or budget.

Phase II clinical trials sit at a critical juncture in drug development, the point at which we decide whether a candidate treatment shows enough early evidence of activity to justify a large, expensive Phase III programme. Because of this, Phase II designs need to be both statistically rigorous and ethically efficient, allowing sponsors to stop early when a treatment is unlikely to succeed.

In this free webinar, Brian Fox, Research Statistician at nQuery, has provided an overview of Phase II clinical trial design with a particular focus on adaptive and multi-stage designs.

These include Simon's Two-Stage Design, Simon's Admissible Designs, Bryant & Day's (1995) design, Litwin et al.'s (2007) design, and Lin & Shih's (2004) design.

Worked examples throughout illustrate how each design is applied in oncology and other early-phase settings.

Learning objectives of this webinar:

Phase IIA proof-of-concept trials exist to screen out ineffective treatments as early and efficiently as possible, so that resources and patients are directed toward the candidates most likely to succeed. Achieving this requires designs that build in interim looks and pre-specified stopping rules, rather than committing to a fixed sample size upfront.

Simon's Two-Stage Design has long been the workhorse of single-arm Phase II oncology trials, and its Admissible Designs extension gives trialists a spectrum of options to balance expected and maximum sample size. More recent extensions adapt this framework to toxicity, survival, and adaptive response endpoints.

Whether it's choosing between an optimal and a minimax design, jointly screening for response and toxicity, or incorporating progression-free survival, there is a clear need to tailor the two-stage framework to the specific demands of a Phase IIA trial.

1. Overview of Phase II Clinical Trials and Where Phase IIA Trial Design Fits

We begin by examining the role of Phase II within the wider development pathway:

• The bridge between first-in-human Phase I and confirmatory Phase III trials
• The split between Phase IIA (proof-of-concept) and Phase IIB (dose-finding)
• Why early stopping for futility is central to Phase IIA design
• Single-arm vs. controlled Phase II designs
• How screening decisions at this stage shape the rest of the development programme

This context sets the stage for understanding why multi-stage design is the natural fit for Phase IIA's screening role.

2. Design Considerations for Adaptive & Multi-Stage Phase IIA Trial Design

Effective Phase IIA design requires careful pre-specification of several key elements:

• Stage boundaries and interim stopping rules for futility
• Trade-offs between expected sample size and maximum sample size
• Choice of primary endpoint (response, toxicity, progression-free survival)
• Optimal vs. minimax design philosophies
• Extending the two-stage framework to more complex or adaptive endpoints

Getting these elements right at the design stage is what allows a Phase IIA trial to fulfil its purpose: screening candidates efficiently and ethically.

3. nQuery Practical Demonstrations

The webinar includes hands-on worked examples in nQuery, covering:

• Simon's Two-Stage Design (Optimal & Minimax)
• Simon's Admissible Designs
• Two-Stage Design for Response & Toxicity (Bryant & Day)
• Two-Stage Design for Progression-Free Survival (Litwin)
• Two-Stage Design for Adaptive Response (Lin & Shih)

We conclude with a comparison of the designs covered and practical guidance on how to select the most appropriate design for a given clinical scenario.

What Is a Phase IIA Proof-of-Concept Trial?

Phase II trials form the bridge between first-in-human Phase I studies and confirmatory Phase III trials. They are commonly split into two stages: a proof-of-concept (PoC) stage and a dose-finding stage, often labelled Phase IIA and Phase IIB respectively.

Phase IIA's goal is to screen out ineffective treatments as early and efficiently as possible, so that resources and patients can be directed toward the most promising candidates. Because of this screening role, Phase IIA trials place a strong emphasis on interim analyses and early stopping for futility, rather than the fixed, single-look designs more typical of confirmatory studies.

This is why multi-stage designs, rather than a single fixed-sample study, have become the standard tool for this phase of development.

What Is Simon's Two-Stage Design and Why Is It the Standard for Single-Arm Phase II Trials?

Simon's Two-Stage Design has long been the workhorse of single-arm Phase II oncology trials. It offers a simple framework for stopping a trial early when a treatment fails to show a basic level of efficacy needed to warrant further development or evaluation.

The design works in two stages. In the first stage, a small number of patients are treated and their response assessed. If too few respond, the trial stops for futility and the treatment is not developed further. If enough respond, the trial proceeds to a second stage, enrolling additional patients before a final efficacy decision is made.

Simon's framework offers two standard variants:

Optimal design: Minimises the expected sample size when the treatment is truly ineffective, reducing patient exposure to a treatment that isn't working.

Minimax design: Minimises the maximum possible sample size, which can be preferable when the total number of patients available or affordable is the primary constraint.

The choice between these reflects a trade-off: the optimal design is more efficient on average, while the minimax design caps the worst-case trial size.

What Are Simon's Admissible Designs and How Do They Extend the Two-Stage Framework?

Simon's Admissible Designs extend the classical Two-Stage Design by giving trialists a spectrum of designs that balance expected sample size against maximum sample size, rather than forcing a binary choice between the optimal and minimax extremes.

Rather than being limited to a single optimal design or a single minimax design, admissible designs allow a sponsor to select a design anywhere along the trade-off curve between these two extremes, choosing the balance of average efficiency and worst-case size that best suits the practical constraints of the trial: recruitment feasibility, budget, and the acceptable level of risk associated with a larger-than-expected trial.

This flexibility makes admissible designs a practical extension for sponsors who find that neither the strict optimal nor the strict minimax design fits their operational needs.

How Do Two-Stage Designs Extend Beyond a Single Response Endpoint?

Simon's original framework was built around a single binary response endpoint. Several extensions adapt the same two-stage logic to other endpoints and more complex early-phase questions:

Bryant & Day's (1995) design extends the two-stage framework to jointly monitor response and toxicity, allowing a trial to stop early not only when efficacy is inadequate, but also when toxicity is unacceptably high. This reflects the reality that Phase IIA decisions are rarely about efficacy alone.

Litwin et al.'s (2007) design adapts the two-stage approach for progression-free survival, a time-to-event endpoint increasingly used in place of, or alongside, binary response in oncology trials.

Lin & Shih's (2004) design extends the framework to adaptive response settings, allowing greater flexibility in how interim decisions are made as data accumulate.

Each of these designs preserves the core logic of Simon's framework, an early opportunity to stop for futility, while tailoring the statistical machinery to the endpoint and clinical question at hand.

How Do You Choose the Right Design for a Given Phase IIA Trial?

With several two-stage frameworks available, the choice of design should be driven by the primary endpoint and the practical constraints of the trial.

A single binary response endpoint, with no strong constraint on maximum vs. expected sample size, is best served by Simon's Two-Stage Design.

A single response endpoint with more flexibility in the sample size trade-off may be better served by an Admissible Design, allowing the trial to sit at the point on the trade-off curve that best matches recruitment and budget constraints.

Toxicity and efficacy considered jointly points toward Bryant & Day's design, which monitors both simultaneously.

A time-to-event endpoint such as progression-free survival points toward Litwin's design.

A need for greater adaptivity in interim decision-making points toward Lin & Shih's design.

In each case, the underlying question is the same: what is the primary endpoint, and what is the acceptable trade-off between stopping early to protect patients and resources, versus continuing to gather sufficient evidence of activity? Sample size software capable of modelling each of these frameworks is an essential tool for evaluating that trade-off explicitly at the design stage.

Frequently Asked Questions About Phase IIA Trial Design

What is the difference between Phase IIA and Phase IIB trials? Phase IIA trials are the proof-of-concept stage, focused on screening out ineffective treatments as early as possible. Phase IIB trials follow, focusing on dose-finding once a treatment has shown sufficient evidence of activity.

What is Simon's Two-Stage Design used for? Simon's Two-Stage Design is used in single-arm Phase II oncology trials to allow early stopping for futility, minimising the number of patients exposed to a treatment that isn't working.

What is the difference between an optimal and a minimax design? An optimal design minimises the expected sample size when the treatment is ineffective. A minimax design minimises the maximum possible sample size, which matters most when total recruitment capacity is the binding constraint.

When should I use Bryant & Day's design instead of Simon's Two-Stage Design? Bryant & Day's design should be used when a Phase IIA trial needs to monitor both response and toxicity jointly, rather than efficacy alone.

Which two-stage design is appropriate for a progression-free survival endpoint? Litwin et al.'s (2007) design extends the two-stage framework to progression-free survival and other time-to-event endpoints.

About nQuery
nQuery helps make your clinical trials faster, less costly and more successful.
It is an end-to-end platform covering Frequentist, Bayesian, and Adaptive designs with 1000+ sample size procedures. 

nQuery Solutions
Sample Size & Power Calculations
Calculate for a Variety of frequentist and Bayesian Design