Application of Bayesian Borrowing Methods in Clinical Trials for Children with Type II Diabetes Mellitus

Area of Science:

• Clinical pharmacology
• Biostatistics
• Pediatric endocrinology

Background:

• Clinical trials for pediatric type II diabetes mellitus (T2DM) face recruitment challenges and greater treatment effect variability compared to adults.
• Larger pediatric studies are often required to achieve adequate statistical power, impacting feasibility and efficiency.
• Leveraging existing adult data can mitigate these challenges if scientifically justified.

Purpose of the Study:

• To introduce and demonstrate Bayesian borrowing methods for pediatric T2DM clinical trials.
• To enhance the feasibility and efficiency of anti-diabetic product development in children.
• To achieve scientifically sound and conclusive results with adequate study power.

Main Methods:

• Utilized Bayesian borrowing methods, incorporating external adult data into pediatric trial analysis.
• Employed a mixture prior to adjust information borrowing based on adult-pediatric T2DM similarity, mitigating prior-data conflicts.
• Pre-specified model parameters, identified external data sources, assessed operating characteristics, and defined borrowing weights.

Main Results:

• The application of Bayesian borrowing methods was successfully demonstrated in an anti-diabetic product development context.
• The methods provided a framework for obtaining conclusive results with adequate study power in pediatric T2DM trials.
• The mixture prior effectively managed the degree of information borrowed, balancing adult and pediatric data.

Conclusions:

• Bayesian borrowing methods offer a viable strategy to improve the efficiency and feasibility of pediatric T2DM clinical trials.
• Careful pre-specification of parameters, data sources, and borrowing strategies is crucial for successful implementation.
• This approach provides a valuable guide for developing anti-diabetic products for children with T2DM.