Basket Study

Biostatistics in Oncology Basket Study Designs: Unlocking the Potential of Personalized Cancer Therapy, a case study

Introduction

In the realm of oncology research, the quest for more effective treatments and personalized therapies has led to the development of innovative study designs, one of which is the basket study. Biostatistics plays a pivotal role in designing, analyzing, and interpreting the complex data generated by these studies. In this case study we look at our opportunities and challenges faced when designing and implementing a basket study.

The Role of Biostatistics in the Basket Study Designs

In oncology basket studies, our biostatisticians were and are essential in several different aspects:

1. Sample Size Calculation: Determining the appropriate sample size is crucial for the statistical power of the trial. Biostatisticians use their expertise to estimate the number of patients needed to detect a meaningful treatment effect. Because of the complexity of some basket study designs, often simulations will be required in order to determine operating characteristics under various scenarios, with the primary purpose of preserving type 1 error over potentially many cohorts and escalations.
2. Randomization and Treatment Assignment: Biostatisticians design randomization processes to ensure that patients are allocated to treatment arms in a fair and unbiased manner, minimizing potential biases. Given the general purpose of basket studies, if randomization is required, the different “baskets” (cohorts) should be a stratification factor in order to ensure balance, especially as sample sizes get larger and randomization is centralized.
3. Data Analysis: They employ sophisticated statistical methods to analyze the data generated from basket studies, comparing treatment outcomes, and assessing the effectiveness of targeted therapies. Most basket studies are adaptive with interim analyses to assess the safety and viability of the treatment across the different cohorts. This is why simulations are key for these studies.
4. Survival Analysis: Biostatisticians often employ survival analysis techniques to estimate the duration of time until an event of interest (e.g., disease progression or death) occurs, providing valuable insights into treatment efficacy. These analyses are key as they can provide crucial information for risk-benefit within each basket.
5. Adaptive Design: In trials with adaptive designs, biostatisticians help develop rules for modifying the trial based on interim data, optimizing the study's efficiency and ethical considerations. This further underscores the importance of performing simulations of the study design and operating characteristics to ensure the most appropriate outcomes for the study.

Challenges we Faced During our Process and Considerations we had to Implement

While basket studies hold immense promise, they also come with unique challenges for statisticians:

1. Data Complexity: The inclusion of multiple treatment arms and patient subgroups can result in complex data structures that require advanced statistical methods for analysis. Interim analyses can be utilized to check for patient safety and treatment benefit often. Depending on the phase and size of study, adaptive hypotheses would likely need to be specified to account for the various outcomes.
2. Multiplicity: With multiple hypotheses being tested simultaneously, the issue of multiple testing arises. Biostatisticians must employ correction methods to control the family-wise error rate.  Similar to the issues with data complexity, multiplicity must be accounted for in these trials. However, the key will be to determine the appropriate sample size and decision-making processes via simulation in order to ensure type 1 error control.
3. Patient Heterogeneity: The diverse patient population in basket studies can lead to heterogeneity in treatment responses, requiring advanced statistical modeling techniques.  These can be accomplished via subgroup analysis, survival modeling with covariates, and other methods that assess the validity of the models and results.
4. Longitudinal Data: Many basket studies involve the collection of longitudinal data over time, necessitating the use of specialized statistical methods to account for repeated measurements. For many endpoints, these can be accomplished with mixed effect models or models with generalized estimating equations to account for the within subject correlation.

Conclusion

From our experience, Oncology basket studies have represented a paradigm shift in cancer research, offering hope for more effective and personalized cancer treatments. Biostatistics played a vital role in designing our basket studies, analyzing complex data, and drawing meaningful conclusions.

Many of the basic statistical tenets regarding sample size calculation, data analysis, and hypothesis testing cannot specifically be applied “by the book”. This requires the attention of skilled statisticians and programmers who know how to run simulations, apply adaptive hypothesis testing, and ensure type 1 error control.

As technology and statistical methods continue to advance, we can anticipate even greater strides in the field of oncology, ushering in an era of precision medicine tailored to each patient's unique genetic profile.