1st Edition
Bayesian Analysis with R for Drug Development Concepts, Algorithms, and Case Studies
Drug development is an iterative process. The recent publications of regulatory guidelines further entail a lifecycle approach. Blending data from disparate sources, the Bayesian approach provides a flexible framework for drug development. Despite its advantages, the uptake of Bayesian methodologies is lagging behind in the field of pharmaceutical development.
Written specifically for pharmaceutical practitioners, Bayesian Analysis with R for Drug Development: Concepts, Algorithms, and Case Studies, describes a wide range of Bayesian applications to problems throughout pre-clinical, clinical, and Chemistry, Manufacturing, and Control (CMC) development. Authored by two seasoned statisticians in the pharmaceutical industry, the book provides detailed Bayesian solutions to a broad array of pharmaceutical problems.
Features
- Provides a single source of information on Bayesian statistics for drug development
- Covers a wide spectrum of pre-clinical, clinical, and CMC topics
- Demonstrates proper Bayesian applications using real-life examples
- Includes easy-to-follow R code with Bayesian Markov Chain Monte Carlo performed in both JAGS and Stan Bayesian software platforms
- Offers sufficient background for each problem and detailed description of solutions suitable for practitioners with limited Bayesian knowledge
Harry Yang, Ph.D., is Senior Director and Head of Statistical Sciences at AstraZeneca. He has 24 years of experience across all aspects of drug research and development and extensive global regulatory experiences. He has published 6 statistical books, 15 book chapters, and over 90 peer-reviewed papers on diverse scientific and statistical subjects, including 15 joint statistical works with Dr. Novick. He is a frequent invited speaker at national and international conferences. He also developed statistical courses and conducted training at the FDA and USP as well as Peking University.
Steven Novick, Ph.D., is Director of Statistical Sciences at AstraZeneca. He has extensively contributed statistical methods to the biopharmaceutical literature. Novick is a skilled Bayesian computer programmer and is frequently invited to speak at conferences, having developed and taught courses in several areas, including drug-combination analysis and Bayesian methods in clinical areas. Novick served on IPAC-RS and has chaired several national statistical conferences.
SECTION I Background
1. Bayesian Statistics in Pharmaceutical Development
Introduction
Overview of Drug Development
Basic Research
Drug Discovery
Formulation
Laboratory Test Methods
Pre-Clinical Studies
Clinical Development
Translational Research
Chemical Manufacturing and Control
Regulatory Registration
Statistics in Drug Research and Development
Bayesian Statistics
Opportunities of Bayesian Approach
Pre-Clinical Development
CMC Development
Clinical Trials
Challenges of Bayesian Approach
Objection to Bayesian
Regulatory Hurdles
Concluding Remarks
2. Basics of Bayesian Statistics
Introduction
Statistical Inference
Research Questions
Probability Distribution
Frequentist Methods
Bayesian Inference
Selection of Priors
Bayesian Computation
Monte Carlo Simulation
Example
Markov Chain Monte Carlo
Computation Tools
BUGS and JAGS
SAS PROC MCMC
Utility of JAGS
Concluding Remarks
3. Bayesian Estimation of Sample Size and Power
Introduction
Sample Size Determination
Frequentist Methods
Bayesian Considerations
Bayesian Approaches
Power and Sample Size
Interim Analysis
Futility and Sample Size
Case Example
Modelling of Overall Survival
Maximum Likelihood Estimation
Futility Analysis
Concluding Remarks
SECION II Pre-Clinical and Clinical Research
4. Pre-Clinical Efficacy Study
Introduction
Evaluation of Lab-Based Drugs in Combination
Background
Statistical Methods
Antiviral Combination
Evaluation of Fixed Dose Combination
Bayesian Survival Analysis
Limitations of Animal Data
Current Methods
Bayesian Solution
Case Example
Concluding Remarks
5. Bayesian Adaptive Design for Phase I Dose-Finding Studies
Introduction
Algorithm-Based Design
3+3 Design
Alternate Algorithm-Based Designs
Advantages and Disadvantages of Algorithm-Based Designs
Model Based Designs
Continual Reassessment Methods
CRM for Phase I Cancer Trials
Escalation with Overdose Control
Escalation Based on Toxicity Intervals
Concluding Remarks
6. Design and Analysis of Phase II Dose-Ranging Studies
Introduction
Phase II Dose-Ranging Studies
Criticisms of Traditional Methods
Model-Based Approaches
Estimating Predictive Precision and Assurance for New Trial
COPD Study
Estimation Method
Concluding Remarks
7. Bayesian Multi-Stage Designs for Phase II Clinical Trials
Introduction
Phase II Clinical Trials
Multi-Stage Designs
Frequentist Approaches
Bayesian Methods
Bayesian Single-Arm Trials
Continuous Monitoring of Single-Arm Trials
Comparative Phase II Studies
Examples
Oncology Trial
Multi-Stage Bayesian Design
Concluding Remarks
SECTION III Chemistry, Manufacturing, and Control
8. Analytical Methods
Introduction
Method Validation
Background
Study Design for Validation of Accuracy and Precision
Current Statistical Methods
Total Error Approach
Bayesian Solutions
Example
Method Transfer
Background
Model
Linear Response
Case Example
Concluding Remarks
9. Process Development
Introduction
Quality by Design
Critical Quality Attributes
Risk of Oncogenicity
Bayesian Risk Assessment
Modeling Enzyme Cutting Efficiency
Bayesian Solution
Example
Design Space
Definition
Statistical Methods for Design Space
Bayesian Design Space
Example
Process Validation
Risk-Based Lifecycle Approach
Method Based on Process Capability
Method Based on Predictive Performance
Determination of Number of PPQ Batches
Concluding Remarks
10. Stability
Introduction
Stability Study
Shelf-Life Estimation
Current Methods
Bayesian Approaches
Examples
Selection of Stability Design
Bayesian Criterion
Setting Release Limits
Concluding Remarks
11. Process Control
Introduction
Quality Control and Improvement
Control Charts
Types of Control Charts
Shewhart I-MR Chart
EWMA Control Chart
CUSUM Control Chart
J-Chart
Multivariate Control Chart
Bayesian Control Charts
Control Chart for Data with Censoring
Control Chart for Discrete Data
Control Limit for Aberrant Data
Product Quality Control Based on Safety Data from Surveillance
Concluding Remarks
Biography
Harry Yang is Senior Director and Head of Statistical Sciences at MedImmune. He has 24 years of experience across all aspects of drug research and development and extensive global regulatory experiences. He has published six statistical books, 15 book chapters, and over 90 peer-reviewed papers on diverse scientific and statistical subjects, including 15 joint statistical works with Dr. Novick. Dr. Yang is a frequent invited speaker at national and international conferences. He also developed statistical courses and conducted training at the FDA and USP as well as Peking University.
Steven Novick is Director of Statistical Sciences at MedImmune. He has extensively contributed statistical methods to the biopharmaceutical literature. Dr. Novick is a skilled Bayesian computer programmer and is frequently invited to speak at conferences, having developed and taught courses in several areas, including drug-combination analysis and Bayesian methods in clinical areas. He served on IPAC-RS and has chaired several national statistical conferences.