Exploring the Advantages of Innovative Early Phase Trial Design

Feature
Article

In early phase trials innovative designs offer potential for improvements in efficiency and decision-making.

Tim Clark

Tim Clark, Dip. Stat., PhD
Vice-president of clinical sciences
ICON

Determining the design that best fits a trial’s purpose is a critical challenge. In fact, an ICON survey of 149 biotech professionals found that it was a top concern for biotechs preparing to progress from preclinical to clinical testing. Specifically, protocol development to generate the right scientific data required for further investment into an investigational product was selected as a key challenge by 45% of respondents. And, with some estimates placing the average out-of-pocket expenses for early phase trials at $7.1 million, these trials can be quite costly.1

For sponsors considering how to most efficiently and effectively allocate resources, innovative trial designs offer advantages. In early phase trials—specifically, phase I and II trials involving safety, dosing, and proof of concept—innovative designs offer potential for improvements in efficiency and decision-making, as well as lowering costs. Innovative trial designs, such as adaptive trials, Bayesian methodologies and master protocols, can be utilized to add value when used in early phases.

Adaptive trials

At a basic level, adaptive trials allow for modifications to elements of the trial design while the trial is ongoing, based on evaluations of the incoming study data. All adjustments are based on predetermined criteria at set points throughout the course of the trial. Elements that may be adjusted include sample size, sub populations to be enrolled, and study dose or regimen.

Karen Smith

Karen L Smith
Director of biostatistical consulting services
ICON

Because they allow for decision-making based on data as it accumulates, adaptive trials have meaningful potential to conserve resources. In early phases, they offer the ability to identify optimal clinical benefits and make informed decisions regarding safety and efficacy earlier in the clinical trial process. If conclusive data is collected, a trial can be terminated early—whether to curtail the use of resources on an unsafe or ineffective drug candidate, or to move forward to the next phase. Further, creating a study design that integrates adaptations can reduce the time and money spent on protocol amendments, which typically require a significant amount of input and financial resources to implement.

To be effective and minimize any potential bias, adaptive trials must be planned carefully, with predetermined criteria that trigger specific adaptations. Challenges to implementing adaptive trials may include the need for complex statistical analysis and the requirement for frequent, real-time communication between trial sites for data transmission and the implementation of adaptations. Despite these challenges, adaptive trials are versatile and may be used in combination with other innovative trial designs, such as Bayesian statistical methods and master protocols.

Bayesian statistical methods

Typically, clinical trials utilize a frequentist statistical approach, wherein the focus is on determining the probability of obtaining the observed data, assuming the null hypothesis (i.e. no difference between the groups) is true. Such an approach utilizes only the data gathered in the clinical trial. On the other hand, in a Bayesian analysis, prior beliefs, which can be existing data and/or clinical opinion, about the parameters of interest (e.g. population mean) are directly used in the analysis. The prior distribution is combined with the data to obtain a posterior distribution, which represents the updated beliefs about the parameters of interest after observing the data. The probability of a difference between the groups is calculated directly from the posterior distribution.

Bayesian statistical methods can improve the efficiency of early and later phase trials by using data collected from external clinical studies, real-world data, and/or clinical opinion. The utility of this approach can be seen in areas such as dose selection, where prior information—including preclinical pharmacokinetic and pharmacodynamic studies, as well as past external clinical trials—can be used to model the dose-response relationship. Precise estimation of the dose-response curve is the foundation for a precise estimate of the minimum effective dose and the range of effective doses. Furthermore, Bayesian models can be used in dose-escalation studies to estimate the probability of toxicity at each dose level. The model can be updated with data from each patient, and the probability of toxicity can be used to guide the dose escalation decisions.

Using Bayesian statistics offers additional benefits, including smaller sample sizes compared to traditional frequentist methods. This is particularly helpful in cases such as the study of rare diseases, which inherently have a limited pool of potential participants. However, it is critical to carefully construct the prior distribution and demonstrate through simulation studies that the Bayesian model has acceptable operating characteristics (e.g. the type 1 or false positive error).

Master protocols

Constructed to test multiple hypotheses or assets, master protocols provide one overarching set of protocols that can guide several sub-studies. There are many different types of master protocols, each of which is optimal for different purposes: Basket studies allow the testing of a single therapy across multiple indications; umbrella trials enable the study of multiple therapies in a single indication, often divided into sub-studies for different disease subtypes or expressions; and platform studies allow different treatments to enter and exit trials over an extended period of time.

Master protocols are employed in both early and later phases. In early phase and exploratory contexts, master protocols are particularly useful in proof-of-concept studies. Many of the benefits of master protocols relate to the shared resources among sub-studies—for example, infrastructure such as clinical sites and data monitoring committees. One of the key advantages of these designs is the possibility to use a single, shared control arm across all sub-studies. Although this requires careful consideration of which participants are included in each comparison with some statistical adjustments to be made, doing so means that, proportionally, more participants may receive an investigational product, and fewer total participants need to be recruited compared to the number required if each arm were an independent study. As a result, recruitment costs and time are reduced. Additionally, master protocols often allow for a more efficient screening process, which can help recruit participants across closely related indications, also reducing recruitment time.

Further, in early phases, master protocols can aid in determining the most advantageous path forward. For instance, an umbrella trial can seek to discern which investigational product is most beneficial to patients, such as by gathering early safety and efficacy data for multiple versions of a therapy. This allows sponsors to select the candidate most likely to be successful to advance to further clinical trials.

A promising road ahead

In the past, sponsors have hesitated to employ innovative trial designs out of concern for regulatory approval. However, regulatory bodies are increasingly accepting of these less traditional approaches. Additionally, a growing number of trials have successfully employed innovative designs, setting a positive precedent for their use. As long as these trials are carefully planned with clear reasoning, there is no reason not to explore the potential benefits of these trial designs for early phase studies.

Authors

Tim Clark, Dip. Stat., PhD, Vice-President, Clinical Sciences, ICON

Karen L Smith, Director, Biostatistical Consulting Services, ICON

Sources

  1. Sertkaya, Aylin, et al. “Costs of Drug Development and Research and Development Intensity in the US, 2000-2018.” JAMA Network Open, vol. 7, no. 6, June 2024, p. e2415445, https://doi.org/10.1001/jamanetworkopen.2024.15445.
Related Content