At the DPHARM conference, Eisai Inc. showcased their groundbreaking use of Bayesian methods to achieve accelerated approval for their Alzheimer’s drug, LEQEMBI™ (lecanemab-irmb). Led by Dr. Shobha Dhadda and Dr. Lynn D. Kramer, the session delved into the intricacies of Bayesian clinical trial design and its significant impact on drug development, offering a compelling case study in innovation.

The Evolution of Bayesian Design

Dr. Lynn D. Kramer began by providing a comprehensive overview of Bayesian theory, tracing its development from the 1970s to its eruption in the 2000s. He highlighted the contributions of Don Barry, a pioneer in Bayesian design, who played a crucial role in advancing the methodology. During the initial decades, Bayesian theory was primarily in the development and discussion phase, with researchers trying to understand its potential applications. By the early 2000s, Bayesian methods began to gain traction, particularly in clinical trials. Kramer pointed out several key studies that utilized Bayesian designs, such as the I-SPY trials and the Trulicity trials, demonstrating this approach’s practical benefits.

The FDA also started developing strategies for using Bayesian designs, recognizing their potential to improve trial efficiency and outcomes. Kramer mentioned that the FDA had even released draft guidance focusing on the advantages of Bayesian design. However, it was still in the draft phase, which is common with many FDA guidelines. According to Kramer, one of the significant advantages of Bayesian design is its ability to conserve placebo patients. In traditional trials, a significant number of participants are assigned to placebo groups, which can be a deterrent for patient enrollment. Bayesian adaptive designs, however, can adjust the allocation of patients based on interim results, increasing the likelihood that participants receive potentially therapeutic drugs rather than placebos. This improves patient satisfaction and enhances the ethical aspects of clinical trials.

The Case of LEQEMBI™: A Bayesian Success Story

Dr. Shobha Dhadda explained why Eisai chose a Bayesian design for their Alzheimer’s trials. She noted that the shift towards early Alzheimer’s disease phases, such as mild cognitive impairment (MCI) and mild Alzheimer’s disease (AD), necessitated a robust proof-of-concept study before advancing to large, expensive Phase III trials. Traditional designs have led to numerous failures, with over 30-40 drugs failing in the last two decades due to inadequate Phase II studies. Dhadda elaborated that one of the key reasons for these failures was the lack of a strong clinical proof of concept before moving into Phase III trials. Many studies had relied solely on biomarker data, which, while useful, did not provide a comprehensive understanding of the drug’s clinical efficacy.

Eisai was determined to avoid this pitfall by ensuring a robust Phase II study for lecanemab. Dhadda explained that there were still many unknowns about lecanemab at the time, including the optimal dose and regimen. The team considered several doses and regimens and needed to determine the study’s duration, especially given the chronic and slow-moving nature of early-stage Alzheimer’s disease. In such a scenario, where the placebo curve might not show significant movement, identifying the right dose and demonstrating a treatment effect becomes particularly challenging. To address these challenges, Eisai decided to employ a Bayesian adaptive design. Dhadda highlighted that this approach allowed them to identify the optimal dose and regimen while minimizing costs and sample size.

The Mechanics of Bayesian Adaptive Design

Dr. Dhadda elaborated on the mechanics of Bayesian adaptive design, explaining the concept of response adaptive randomization. Unlike traditional fixed allocation, Bayesian design allocates more subjects to the most effective dose arms based on an algorithm. This approach allows for multiple interim analyses, focusing on safety and efficacy. Dhadda explained that in a traditional design, the allocation of subjects to different arms is fixed at the beginning of the study. For example, in a study with two or three arms, the randomization might be set at a 1:1 or 1:2 ratio, and this allocation remains constant throughout the study.

In contrast, a Bayesian adaptive design uses an algorithm to adjust the allocation of subjects based on interim results. This means the dose arm showing the most promise receives more subjects, while less effective dose arms receive fewer subjects. Dhadda clarified a common misconception about Bayesian adaptive designs: once subjects are assigned to a dose arm, they remain in that arm for the study’s duration. The adaptive allocation only applies to new subjects entering the study. This ensures that the study maintains its integrity and results are reliable. The Bayesian design also allows for multiple interim analyses, which can be conducted for various purposes, such as assessing safety or efficacy.

Predicting Phase III Success

The Bayesian Phase II trial for lecanemab perfectly predicted the outcomes of the Phase III trial, known as Clarity AD. The Phase II trial’s adaptive design identified the most effective dose regimens validated in the larger Phase III trial. This seamless transition from Phase II to Phase III was a testament to the power of Bayesian methods in drug development. Dr. Kramer explained that the Phase II trial was designed to examine the dose-response and dosing frequency based on the pharmacokinetic (PK) and pharmacodynamic (PD) models. The team wanted to understand whether similar responses could be achieved with different dosing frequencies, such as biweekly versus monthly.

This was particularly important for an antibody drug like lecanemab, which needs to access the brain, potentially leading to differences in PK. The Phase II trial also aimed to explore the consistency of responses across various endpoints, including cognitive and functional measures. The team used a novel endpoint, ADAS-Cog, which they developed to detect early changes in the disease. This endpoint was particularly sensitive to early changes, allowing the team to identify the most effective dose regimens quickly. Kramer highlighted that the Phase II trial was not unblinded; the adaptive randomization and interim analyses were conducted by a computer algorithm, ensuring that the study remained blinded to the researchers.

Key Takeaways and Future Implications

Dr. Kramer concluded the session by summarizing the key takeaways. The Bayesian design allowed Eisai to answer more questions accurately and inexpensively, ultimately improving patient outcomes. The Phase II trial’s success in predicting Phase III outcomes demonstrated the potential of Bayesian methods to revolutionize clinical trials. Kramer emphasized that the Bayesian design enabled the team to make early decisions, conserve placebo patients, and quickly identify the most effective dose regimens. This approach not only accelerated drug development but also improved the ethical aspects of clinical trials by reducing the number of patients exposed to placebo.

The session underscored the importance of innovative trial designs in accelerating drug development, particularly for complex diseases like Alzheimer’s. As Bayesian methods gain more acceptance, they promise to make clinical trials more efficient and effective, benefiting patients and the pharmaceutical industry. The success of the LEQEMBI™ trials serves as a compelling example of how adaptive designs can transform the landscape of clinical research, offering new hope for patients and setting a new standard for future drug development.

Summary

Eisai’s successful deployment of Bayesian methods in their Alzheimer’s trials is a compelling case study of the power of adaptive clinical trial design. By embracing innovation, Eisai not only accelerated the approval of LEQEMBI™ but also set a new standard for future drug development, demonstrating the transformative potential of Bayesian methods in clinical research.

Website | + posts

Moe Alsumidaie is Chief Editor of The Clinical Trial Vanguard. Moe holds decades of experience in the clinical trials industry. Moe also serves as Head of Research at CliniBiz and Chief Data Scientist at Annex Clinical Corporation.