Written by Yuan Ji, Professor of Biostatistics at The University of Chicago and Mansha Sachdev, Senior Marketing Manager, Content

The need for innovative and efficient trial designs has become increasingly apparent in the evolving landscape of oncology drug development. Traditional clinical trials often focus on a single cancer type, requiring multiple individual trials to assess a treatment’s efficacy across different cancer subtypes. This approach can be both resource-intensive and time-consuming. Basket trials, however, offer an innovative solution by allowing simultaneous evaluation of a single therapy across multiple cancer types or subtypes that share common molecular characteristics. This method promises to enhance precision and efficiency in oncology drug development, particularly when combined with Bayesian statistical methods.

Here, we outline the potential transformative role of Bayesian basket trials in oncology drug development.

Understanding Basket trials in the context of oncology

A Basket trial is essentially a multi-arm Phase 2 or Phase 3 study investigating a treatment for multiple diseases or sub-diseases. For instance, you might include three to five indications in the ‘basket,’ all sharing a fundamental characteristic, such as the same mutation, molecular biomarker, or disease pathogenesis. A basket trial would allow you to evaluate these three to five indications at a lower cost than running five separate single-arm trials, leading to a substantial increase in efficiency. The key advantage of basket trials is the ability to borrow information across different cancer types, potentially improving statistical precision and power in estimating and detecting treatment effects. This design is particularly useful in oncology, where biomarker-targeted therapies are becoming more prevalent.

The Role of Bayesian Methods

Bayesian methods are especially suited for basket trial designs due to their ability to naturally borrow information via hierarchical models. These methods also naturally allow for adaptive designs that can respond to interim results, making real-time adjustments to the trial. Bayesian hierarchical models use prior distributions that assume exchangeability, meaning that treatment effects are expected to be similar but not identical across different cancer subtypes.

By incorporating information from all subtypes in the analysis, Bayesian methods can enhance the precision of treatment effect estimates and improve decision-making processes during the trial. This approach is not only more efficient but also ethically advantageous, as it allows for interim decisions to stop the trial for futility if no or very low efficacy signals are observed.

Recent Developments in Bayesian Methods for Basket Trials

Several Bayesian approaches have been proposed for basket trials, each with unique features and advantages:

Bayesian Hierarchical Model (BBHM): This model allows for differential information borrowing across subgroups, enhancing the ability to detect true treatment effects while controlling for potential false positives.

Calibrated Bayesian Hierarchical Model (CBHM): This model links the strength of information borrowing to the homogeneity of treatment effects across subgroups, providing better control of type I error rates.

Exchangeability-Nonexchangeability (EXNEX) Method: This approach distinguishes between subgroups that are exchangeable and those that are not, allowing for more tailored borrowing of information.

Multiple Cohort Expansion (MUCE) Method: This method is designed for trials with multiple arms and includes mechanisms for multiplicity control, reducing sample size requirements while maintaining statistical power.

Robust Bayesian hypothesis testing method (ROBOT): This approach uses a Bayesian semiparametric model under a hypothesis testing framework, inducing flexible and data-drive borrowing and better control of statistical error.

Adaptive Dose Optimization Platform (ADOPT): This approach uses a Bayesian hierarchical model in dose-optimization trials in oncology that seamlessly combine dose escalation, expansion, and optimization into a single trial.

Simulation Studies and Practical Implications

Simulation studies have demonstrated the impact of information borrowing on the operating characteristics of Bayesian methods in basket trials. These studies show that Bayesian approaches can significantly improve the power to detect treatment effects and reduce the sample sizes needed for conclusive results. This has profound implications for drug development, potentially accelerating the timeline for bringing new therapies to market and reducing overall costs.

Real-World Examples

Several real-world examples illustrate the successful application of basket trials in Phase 1 and 2 settings. For instance, the Vemurafenib basket trial investigated the efficacy of this BRAF inhibitor across different cancer types with the BRAF V600E mutation. While Vemurafenib showed remarkable efficacy in melanoma, its effect was limited in colorectal cancer with the same mutation, highlighting the importance of understanding heterogeneity in treatment responses.

Similarly, the Glivec basket trial for chronic myeloid leukemia (CML) and other rare cancers with the same genetic abnormality demonstrated the potential of basket trials to expedite the approval of targeted therapies based on robust molecular evidence.

Final Takeaways

Basket trials represent a significant advancement in oncology drug development, offering a more efficient and precise approach to evaluating new therapies. The integration of Bayesian methods further enhances the potential of these trials by leveraging information across multiple cancer types, improving statistical power, and enabling adaptive designs. As the field of precision medicine continues to evolve, basket trials are poised to play a crucial role in the development of targeted cancer therapies, ultimately benefiting patients through more personalized and effective treatment options.

Join our upcoming webinar for an in-depth review of cutting-edge statistical methods in tumor-agnostic clinical trials. We will delve into the most recent advancements in basket trials, with a special focus on innovative Bayesian approaches.

Share Via