Quantitative Drug Safety and Benefit Risk Evaluation: Practical and Cross-Disciplinary Approaches

Cover
Half Title
Series Page
Title Page
Copyright Page
Table of Contents
Editors
List of Contributors
Introduction Chapter
Part A: Regulatory Landscape and Interdisciplinary Collaboration in Safety Evaluation
Chapter 1: The Emergence of Aggregate Safety Assessment across the Regulatory Landscape
1.1 Introduction
1.2 Food and Drug Administration (FDA) Regulations
1.3 EMA Regulations
1.4 Japan Pharmaceuticals and Medical Devices Agency (PMDA) Regulations
1.5 China National Medical Products Administration (NMPA) Regulations
1.6 Post-Marketing Aggregate Safety Analysis
1.7 Real-World Health Data
1.8 Discussion
References
Chapter 2: Monitoring of Aggregate Safety Data in Clinical Development: A Proactive Approach
2.1 Background: The Need for Pre-Marketing Safety Surveillance
2.2 The Aggregate Safety Assessment Plan (ASAP)
2.2.1 Section 1: Value Proposition and Governance
2.2.2 Section 2: Safety Topics of Interest and Pooling Strategy
2.2.3 Section 3: Data Analysis Approaches
2.2.4 Section 4: Analysis of Key Gaps
2.2.5 Section 5: Ongoing Aggregate Safety Evaluation (OASE)
2.2.5.1 Multiplicity in Safety Data
2.2.5.2 Some Methods for Monitoring and Signal Detection
2.2.6 Section 6: Communication of Safety Information
2.3 The ASAP as a Modular Template-Driven Approach
2.4 The ASAP as a Safety Planning Tool to Respond to a Pandemic Such as COVID-19
2.4.1 General Effect on Data Collection
2.4.2 Unanticipated Infections of Subjects in the Clinical Trials Due to the Pandemic
2.4.3 Using the ASAP To Respond to a Pandemic:
2.5 Anticipated Hurdles toward Implementation of the ASAP
2.6 Foundation: Intelligent Data Architecture
2.6.1 Consistent Implementation of Derivation Rules
2.6.2 Consistent MedDRA Coding
2.6.3 Data Traceability and Single Source of Truth
2.7 Foundation: Background Rates
2.8 Trends Likely to Influence the Further Direction of the ASAP
Note
References
Chapter 3: Safety Signaling and Causal Evaluation
3.1 How to Detect a Signal
3.1.1 Preclinical Findings
3.1.2 Single-Case Reports
3.1.3 Aggregate Data – Clinical Development
3.1.3.1 Adverse Events
3.1.3.2 Laboratory Data
3.1.4 Aggregate Data – Post-Marketing
3.2 How to Evaluate a Signal
3.2.1 Confounding and Bias
3.2.2 Causal Association
3.3 How to Determine When a Signal Represents a Product Risk
3.3.1 Strength of Association
3.3.2 Consistency
3.3.3 Specificity
3.3.4 Temporality
3.3.5 Biologic Gradient
3.3.6 Plausibility
3.3.7 Coherence
3.3.8 Experimental Evidence
3.3.9 Analogy
3.4 What Tools Are Available?
3.4.1 Common Graphical Displays for the Adverse Event Review
3.4.2 Common Graphical Displays for the Laboratory Data Review
3.4.3 Other Tools, Platforms, and Resources
3.5 Summary
Notes
References
Chapter 4: Safety Monitoring through External Committees
4.1 Introduction – Safety Monitoring and Assessment during Clinical Development
4.2 Role of DMCs
4.3 Evaluating the Need for a DMC
4.4 Composition of the DMC
4.5 DMC Operations
4.6 DMC’s Scope for Monitoring Trial and Safety of Participating Patients
4.6.1 Investigational New Drug (IND) Safety Expedited Reporting to Health Agencies
4.6.2 Expanded DMC to Serve as a SAC (or Safety Assessment Entity)
4.6.3 Quantitative Analysis of ADRs to Support the SAC (or Safety Assessment Entity)
4.7 Safety Monitoring in Studies with Multiple Drugs or Diseases – Complex Designs
4.8 Safety Monitoring in Pragmatic Trials
4.9 Graphical Assessment of Safety Endpoints
4.10 Summary
References
Part B: Statistical Methodologies for Safety Monitoring
Chapter 5: An Overview of Statistical Methodologies for Safety Monitoring and Benefit-Risk Assessment
5.1 Introduction
5.2 A Perspective of Statistical Methods in Safety Monitoring
5.2.1 Blinded versus Unblinded Safety Assessments
5.2.2 Bayesian and Frequentist Approaches
5.2.3 Post-Marketing PV versus Pre-Marketing Evaluation
5.2.4 Static versus Dynamic Evaluation
5.2.5 Analyses from Patient-Level Data to Integrated Safety Evaluation
5.3 Benefit-Risk Assessment Methodologies
5.4 Concluding Remarks: Safety Monitoring and Benefit-Risk Methodologies
Bibliography
Chapter 6: Quantitative Methods for Blinded Safety Monitoring
6.1 Introduction
6.2 Defining Blinded Safety Monitoring
6.3 Challenges in Blinded Safety Monitoring
6.4 Related Regulatory Consideration
6.5 Frequentist, Bayesian, and Likelihood Approaches to Inference
6.6 Statistical Methods for Blinded Safety Monitoring
6.7 Some Practical Considerations in Implementing Blinded Safety Monitoring
6.8 Linking Blinded and Unblinded Safety Monitoring
6.9 Tools and Software for Quantitative Safety Monitoring
6.10 Illustrative Example of Quantitative Blinded Safety Monitoring
6.11 Conclusion
References
Chapter 7: Bayesian Safety Methodology
7.1 Introduction
7.1.1 Bayesian Statistical Methods in Safety Analysis
7.1.2 Bayesian Statistics
7.1.2.1 Illustrative Example of Bayesian Analysis
7.1.2.2 Hierarchical Models
7.1.3 Chapter Flow
7.2 Bayesian Approaches to Safety Monitoring of Blinded Data
7.2.1 Bayesian Framework
7.2.2 Methods for the Event Rate (Number of Events/Exposure Time)
7.2.2.1 Example 1: The Method of Schnell and Ball
7.2.3 Methods for Proportion of Subjects with an Event
7.2.4 Bayesian Detection of Risk Using Inference on Blinded Safety Data (BDRIBS) Method
7.2.4.1 Example 2: BDRIBS Method
7.2.4.1.1 BDRIBS Assuming the Historical Event Rate Is Fixed
7.2.4.1.2 BDRIBS Assuming the Historical Event Rate Follows a Gamma Distribution
7.2.5 Discussion
7.3 Bayesian Hierarchical Modeling Approaches to Safety Monitoring of Unblinded Data
7.3.1 Literature Review
7.3.2 Bayesian Hierarchical Model
7.3.2.1 Three-level Bayesian Hierarchical Mixture Model
7.3.2.1.1 Choice of Prior
7.3.2.1.2 Example
7.3.2.2 Discussion
7.4 Bayesian Meta-Analytic-Predictive(MAP) Prior and Empirical Meta-Analytic-Predictive (eMAP) Prior Approach to Borrowing Historical Information
7.4.1 MAP Prior
7.4.2 Robust MAP (rMAP) Prior
7.4.3 eMAP Prior
7.4.4 Illustrative Example Applying MAP Priors
7.4.4.1 Discussion
7.5 Chapter Discussion
References
Chapter 8: Likelihood-Based Methods for Safety Monitoring
8.1 Introduction: Importance of Safety Monitoring during Development and Post-Approval Pharmacovigilance
8.1.1 Experiences That Led to Further Enhancement of Monitoring
8.1.2 The State of Monitoring
8.1.3 Looking Forward
8.2 General Framework for Statistical Methods in Dynamic or Continuous Safety Monitoring
8.2.1 A Framework in the Context of Randomized Studies
8.2.2 Excessive and Untenable Levels of Risk and Levels of Evidence
8.2.3 Probability of Misleading Evidence: P-Values in Repeated Safety Testing
8.2.4 A Universal Bound
8.3 Overview of Likelihood-Based Statistical Methods
8.3.1 The Likelihood as a Basis for Comparison of Multiple Sets of Conditions
8.3.2 Wald Sequential Probability Ratio Test (SPRT)
8.3.3 Maximized Sequential Probability Ratio Test (MaxSPRT)
8.3.4 Conditional Maximized Sequential Probability Ratio Test (CMaxSPRT)
8.3.5 Sequential Likelihood Ratio Test (SeqLRT)
8.3.6 Longitudinal Likelihood Ratio Test (LongLRT)
8.3.7 Sequential Generalized Likelihood Ratio Test (SGLRT)
8.3.8 Summary of Likelihood Ratio Methods
8.4 Examples of Several Likelihood-Based Methods in Safety Monitoring
8.4.1 Monitoring for Fever and Neurological Effects of a Vaccine
8.4.2 Rotavirus Vaccine Monitoring for a Serious Intestinal Event
8.4.3 Example of Possible Applications of LongLRT
8.5 Comments on Other Methods and Software for LRT Methods
8.5.1 Non-LR Methods for Pre- and Post-Marketing Surveillance
8.5.2 A Brief Note on Software for LRT Methods
8.6 Summary and Discussion
References
Chapter 9: Meta-Analysis of Drug Safety Assessment
9.1 Introduction
9.2 Planning of a Meta-Analysis for Safety Assessments
9.3 Individual Participant and Aggregated Data Meta-Analysis of Safety
9.4 Meta-Analysis for Rare Events
9.4.1 The Case of Rare Events
9.4.2 Methods for Dealing with Rare Events
9.5 Multivariate Safety Endpoints
9.5.1 Why Multivariate Meta-Analysis?
9.5.2 Defining Multivariate Meta-Analysis
9.5.3 Multivariate Random Effects Model
9.5.4 Statistical Challenges in Multivariate Meta-analysis
9.6 Cumulative and Trial Sequential Meta-Analysis
9.6.1 Cumulative Meta-Analysis
9.6.2 Trial Sequential Meta-Analysis
9.7 Reporting of Meta-Analysis
9.8 Visualizing Meta-Analysis Results
9.9 Analysis Tools and Software for Meta-Analysis
9.10 Conclusions
References
Part C: Design and Analysis Considerations in Randomized Controlled Trials and Real-World Evidence for Safety Decision-Making
Chapter 10: Pragmatic Trials with Design Considerations for Cardiovascular Outcome Trials
10.1 Introduction
10.2 Design Considerations
10.2.1 Randomization in a Real-World Setting
10.2.2 Blinding the Treatment Interventions
10.2.3 Selecting a Comparator
10.2.4 Choice of Endpoints and Data Collection
10.2.5 Safety Evaluation and Monitoring
10.2.6 Design Tools
10.3 CVOTs: From RCT to Pragmatic Trials
10.3.1 Regulatory Guidance
10.3.2 CVOT Pragmatic Trial
10.4 Discussion
References
Chapter 11: Post-Market Safety Assessment Using Observational Studies and the Food and Drug Administration Sentinel System
11.1 Introduction
11.2 History and Role of the Sentinel System
11.3 The Organizational Structure of Sentinel
11.4 The Sentinel Common Data Model, Distributed Database, and Distributed Regression Analysis
11.5 Drug Utilization Using the Sentinel System
11.6 Signal Identification in the Sentinel System
11.6.1 Details of the TreeScan Method
11.6.2 Signal Identification in a Single-Exposure Group
11.6.3 Signal Identification in a Controlled Design
11.6.4 Signal Identification in a Self-controlled Design
11.7 Signal Refinement in the Sentinel System
11.7.1 The L1 Analysis
11.7.2 The L2 Analysis
11.7.3 The L3 Analysis
11.8 ARIA Sufficiency
11.9 Discussion: Future Outlook
References
Chapter 12: Analysis Considerations for Real-World Evidence and Clinical Trials Related to Safety
12.1 Introduction
12.2 Case Study on Chronic Hepatitis C
12.3 Statistical Methods to Account for Confounding for Observational Data
12.3.1 Introduction
12.3.2 What Are PSs?
12.3.3 PS Methods and Other Approaches for Dealing with Confounding
12.3.3.1 PS Methods
12.3.3.2 Additional Considerations in the Use of PSs
12.3.3.2.1 Covariate Selection and Balance
12.3.3.2.2 Unmeasured Confounding
12.3.3.2.3 Machine Learning Approaches
12.3.3.2.4 Use of PSs in Multiple Treatments
12.3.3.2.5 Example Using PSs
12.3.3.3 Other Methods for Confounding
12.4 Synthesis of Multi-Source Data
12.4.1 Introduction
12.4.2 Linking Data Sources
12.4.3 External Control Arms
12.4.4 Meta-Analysis of Safety
12.4.5 Case Study on Safety Data Integration of RCTs and RWEs
12.5 Concluding Remarks
Acknowledgments
References
Part D: Safety/Benefit-Risk Evaluation and Visualization
Chapter 13: Trends and Recent Progress in Benefit-Risk Assessment Planning for Medical Products and Devices
13.1 Introduction
13.1.1 Overview
13.2 Evolution of Benefit-Risk Planning During the Medical Product Development Life Cycle
13.2.1 The Origins of Benefit-Risk Assessments
13.2.2 Current Approaches to sBRAs
13.2.3 Recent and Future Developments Applicable to Benefit-Risk Assessment Planning
13.3 Review of Published Recommendations Applicable to Benefit-Risk Assessment Planning
13.3.1 EMA Perspective on Benefit-Risk
13.3.2 FDA Perspective on Benefit-Risk
13.3.3 Planning for Use of RWD Suitable for Benefit-Risk Evaluation
13.3.4 Planning for Use of Patient Experience Data Suitable for Benefit-Risk Evaluation
13.4 Benefit-Risk Assessment Planning in Practice – Survey of the Industry Perspective
13.4.1 Understanding Industry’s Approach to Benefit-Risk Assessment and Planning
13.4.2 Overview of Industry Survey Design and Participants
13.4.3 Industry Survey Results: sBRA Implementation Measures
13.4.3.1 Coordination
13.4.3.2 Data/Information Sources
13.4.3.3 Tools
13.4.4 Industry Survey Results: sBRA Utilization for Decision-Making
13.4.5 Industry Survey Results: Challenges with sBRAs
13.4.6 Industry Survey Results: Future Opportunities for sBRAs
13.5 Future Work and Discussion
Acknowledgments
References
Chapter 14: Estimands in Safety and Benefit-Risk Evaluation
14.1 Introduction
14.1.1 Overview of the Estimand Framework
14.2 Safety Estimands and Challenges
14.2.1 Question for Safety
14.2.2 Estimand for Safety Assessment in Contrast with Efficacy Estimand
14.2.2.1 Treatment
14.2.2.2 Population
14.2.2.3 Variable (outcome)
14.2.2.4 Intercurrent Events
14.2.2.4.1 Missing Data for Safety
14.2.2.4.2 Intercurrent Event Due to Pandemics
14.2.2.5 Summary Measure
14.3 BR Estimands and Challenges
14.3.1 Estimands for Population-Level BRA: Demonstrated in the Context of CTD and PBRER
14.3.1.1 Attributes of Estimands for Efficacy
14.3.1.2 Attributes of Estimands for Safety
14.3.1.3 Attributes of Estimand for BRA
14.3.1.4 Other Considerations for Population-Level BRA
14.3.2 Estimand for Patient-Level BRA: Demonstrated in the Context of Patient’s Perspective on BRA
14.3.2.1 Attributes of Estimand for Individual BRA
14.4 Summary and Discussion
References
Chapter 15: Visual Analytics of Safety and Benefit-Risk from Clinical Trial Data
15.1 Introduction
15.2 Some General Considerations
15.2.1 Principles of Graph Creation and Visualization
15.2.2 Graph Construction Using the Grammar of Graphics
15.2.3 Question-Based Approach
15.2.4 Regulatory Guidance on the Use of Graphs
15.2.5 Defining Visual Analytics
15.3 Visual Analytics of Safety Data
15.3.1 Complexity of Safety Data
15.3.2 The Case for Visual Analytics in Safety Monitoring
15.3.3 Visual Analytics of Adverse Events Data
15.4 Visual Analytics of Benefit-Risk Data
15.4.1 Overview
15.4.2 Benefit-Risk Based on Patient-Level Data
15.4.3 Questions of Interest in Benefit-Risk
15.4.4 The Case for Visual Analytics in Benefit Risk
15.4.5 Some Examples of Visual Analytics Used in Benefit Risk
15.4.5.1 Value Trees
15.4.5.2 Forest Plots
15.4.5.3 Norton Plots
15.5 Conclusion
References
Index