Theory of Drug Development presents a formal quantitative framework for understanding drug development that goes beyond simply describing the properties of the statistics in individual studies. It examines the drug development process from the perspectives of drug companies and regulatory agencies.
By quantifying various ideas underlying drug development, the book shows how to systematically address problems, such as:
- Sizing a phase 2 trial and choosing the range of p-values that will trigger a follow-up phase 3 trial
- Deciding whether a drug should receive marketing approval based on its phase 2/3 development program and recent experience with other drugs in the same clinical area
- Determining the impact of adaptive designs on the quality of drugs that receive marketing approval
- Designing a phase 3 pivotal study that permits the data-driven adjustment of the treatment effect estimate
- Knowing when enough information has been gathered to show that a drug improves the survival time for the whole patient population
Drawing on his extensive work as a statistician in the pharmaceutical industry, the author focuses on the efficient development of drugs and the quantification of evidence in drug development. He provides a rationale for underpowered phase 2 trials based on the notion of efficiency, which leads to the identification of an admissible family of phase 2 designs. He also develops a framework for evaluating the strength of evidence generated by clinical trials. This approach is based on the ratio of power to type 1 error and transcends typical Bayesian and frequentist statistical analyses.
A Theory of Evaluating Drugs
Clinical Drug Development Phases 1 through 3
Stages of Clinical Development
Bevacizumab
Choosing Drugs to Develop
Probability of Technical Success
Uncertainty Surrounding Expected Future Cash Flows
Maximize the Value of the Company Today or Tomorrow?
Decision Rules for Phase 2
Phase 2/3 Strategy
Model
When Is a Phase 2/3 Strategy Better Than a Phase 3 Trial Alone?
How Much Can Efficiency Be Improved?
Admissible Phase 2 Trial Designs
Projects That Are Not Least Attractive
Example: Bevacizumab
Example: Rituximab
Example: TNK
Maximize the Minimum Efficiency
Single-Arm Phase 2 Trial
Phase 2 Trials Based on Surrogate Endpoints
Impact of a Surrogate on the Efficiency of Drug Development
Estimation of the Potential Impact of a Specific Surrogate on Efficiency
Dose Selection and Subgroups: Phase 2 as a Pilot Trial
Relative Efficiency for Selecting a Dose
Properties of Relative Efficiency for Selecting a Dose
Relative Efficiency for Selecting a Subgroup
Evaluating the Marker Hypothesis
Multistage Screening
Efficiency
Order of Tests in Drug Development
Adverse Events
A Theory of Evidence in Drug Development
Preference for Simple Tests of Hypotheses over Model-Based Tests
Control Maximum Risk
Variance of a Model-Based Estimate of Treatment Effect
Comparison of a Simple Difference in Means with a Model-Based Estimate of Treatment Effect
A Study Design That Permits Data-Driven Model Adjustment of the Treatment Effect Estimate
Quantifying the Strength of Evidence from a Study
Ratio of True Positives to False Positives
Studies with Interim Analyses
A Boundary with a Constant Ratio of Power to Type 1 Error
O’Brien-Fleming Boundary
Bayesian or Frequentist?
Quantifying the Strength of Evidence: A Few Additional Comments on Interim Analyses
Wald’s Likelihood Ratio Test
Pocock Boundary
Confirmatory Trials
Can Evidence from Phase 2 Trials Be Combined with Evidence from Phase 3?
Example: Phase 2 in Rheumatoid Arthritis
Design a Phase 3 Trial to Account for Evidence against the Global Null Hypothesis
Evidence from Phase 3 Trials
Example
Additional Topics
Maximize Efficiency Subject to a Constraint on True+/False+
Power of the Log Rank Test to Detect Improvement in Mean Survival Time and the Impact of Censoring
Setup
Minimizing the Log Rank Test
Examples
Censoring
Survival Benefit in the Bevacizumab Phase 3 Colorectal Cancer Trial
Adaptive Phase 2/3 Designs
Impact of Adaptive Designs on Drug Company Behavior
Net Effect of Adaptive Phase 2/3 Designs on the Ratio of True to False Positives
Size of the Phase 3 Trial
Sizing a Phase 3 Trial Based on the Minimum Clinically Meaningful Difference
Using Phase 2 Results to Size the Phase 3 Trial
Extending the Model of Clinical Drug Development
Maximizing Net Present Value (NPV)
Picking the Best Dose in Phase 2
Targeted Therapies
Appendices
References appear at the end of each chapter.
Biography
Eric B. Holmgren is a consultant and statistical scientist. He previously worked at Genentech and Hoechst Roussel Pharmaceuticals. He received a Ph.D. in mathematical statistics from Stanford University.
"’In each chapter, author provides appropriate statistical formulas that readers can actually utilize. Since this book handles many mathematical formulas, and contains many real good examples, this book would be very useful for statisticians who work at pharmaceutical companies and are deeply involved with drug development … Overall, this book covers necessary and important aspects for drug development, and would be quite useful to clinical statisticians."’
—Byung-Ho Nam, PhD, Department of Cancer Control and Policy, Graduate School of Cancer Science and Policy, National Cancer Center, Korea, in Biometrics"The given book presents a theory of drug development that is based on maximizing the efficiency with which drugs that truly provide clinical benefits are identified. The author shows how to optimize the drug development process at its three main stages (Phases 1, 2, 3), and at some transitional sub-stages, so that the number of molecules that result in a positive final Phase 3 clinical trial per investment is maximized."
—Fatima T. Adylova in Zentralblatt MATH