|Year : 2011 | Volume
| Issue : 5 | Page : 565-567
An introduction to statistics in early phase trials
Department of Biostatistics, All India Institute of Medical Sciences, Ansari Nagar, New Delhi 110 029, India
|Date of Web Publication||26-May-2011|
R M Pandey
Department of Biostatistics, All India Institute of Medical Sciences, Ansari Nagar, New Delhi 110 029
Source of Support: None, Conflict of Interest: None
|How to cite this article:|
Pandey R M. An introduction to statistics in early phase trials. Indian J Med Res 2011;133:565-7
S.A. Julious, S.B. Tan & D. Machin (M/s John Wiley & Sons Ltd., Chichester, UK) 2010. 263 pages.
Early phase studies are known as learning and explaining phase of a drug development. During this phase of the trial, properties of the compound, dose to carry forward, inclusion-exclusion criteria, etc., are learned with limited resources. The knowledge gained from this phase of the trial has major impact on design of confirmatory phases of the trial. In the absence of any information about the properties of compound, the statistical methods used in early phase trials are different and more challenging than those used in the confirmatory phase trials. There is a paucity of textbooks on statistical tools used in early phase trials. This book fills the gap. Both the basic concepts and the statistical methods have been described well throughout the 15 chapters in the book.
Chapter 1 describes the basics in early development and clinical development plans. Both critical and non-critical paths have been elaborated. Basic concepts of pharmacokinetic and pharmacodynamics, dose proportionality Go/No Go and investment decisions, dose response which are critical in early phase trials, have been introduced.
Chapters 2 deals with derivation of pharmacokinetic parameters and nomenclature differences as much of the pharmacokinetic methodology is developed by non-statisticians. Derivation of parameters for single dose and for both compartmental and non-compartmental approaches for intravenous and extravenous dose in single compartment has been presented in a step-by-step manner. Repeat dose pharmacokinetics when it has reached steady state, is equally important. Estimation of repeat dose has been described in detail. Description of summary statistics of pharmacokinetic parameters and a checklist of reviewing a pharmacokinetic analysis are provided in this chapter.
Chapter 3 deals with sample size calculations for clinical trials. The focus of this chapter is on trials where the primary outcome is continuous to assess superiority as well as precision based studies. For a beginner, general principles and various aspects in sample size determination have been described. Methodology for sample size calculation is presented in a simplified manner to demonstrate statistically the presence of a specified clinical effect. However, in early phases of trial focus is on estimating any possible clinical effect with a view to perform later definitive study, therefore, in early phase trials focus is on estimation rather than on finding statistical significance. Precision-based sample requirements for early phase trials in both parallel and cross-over and superiority trials are presented in detail. Sample size based on feasibility in also provided. Sensitivity analysis followed by minimum sample size required for feasibility, precision about mean and variance, and regulatory considerations is discussed.
Chapter 4 is continuation of previous chapter. It deals with basic two period AB/BA cross-over trials. Sample size for two period AB/BA study deign and its appropriate analysis considering period effect and treatment effect for both parametric and non-parametric situation are described.
Chapter 5 is an extension of chapter 4. In this chapter, various aspects of superiority of multi cross-over studies, a common situation in early phase trials, is discussed. Sample size issues are worked out and examples have been presented for the above situations.
Chapter 6 deals with description of different types of specific early phase trials by introducing first-time-into-man (FTIM) studies which are conducted with primary objective of assessing the safety and tolerability of a range of doses to use throughout drug development. Given that FTIM i.e., standard healthy volunteer designs have limitations, the following five most common designs have been described: placebo interrupting design; placebo interrupting design with last dose carried forward; placebo replacement designs; interlocking placebo interrupting design; and interlocking cohort placebo replacing design. The chapter also provides 21 recommendations on design and analysis of FTIM studies. Worked out examples on FTIM studies are described in a simple manner.
Chapter 7 describes general concepts in frequentist and Bayesian approaches. Utility of Bayesian methods in FTIM studies is described with an illustration.
Chapter 8 deals with situation where due to toxic effects it is not possible to undertake trials in healthy volunteers and it becomes a necessity to investigate a new compound immediately in a patient population. Criteria for patients eligibility and choosing the doses to investigate have been covered. Some of the common designs such as C 33 D, BEST-OF-5, and Storer have been described in detail. Continual Reassessment Method (CRM) using various approaches to determine the level of dose limiting toxicity (DLT) has been described. The chapter ends with comparison of features of C 33 D, BEST-OF-5 and Storer design.
Chapter 9 deals with various aspects of bioequivalence studies. After providing sufficient background about the bioequivalence studies and regulatory guidelines from different countries and summary of assessing bioequivalence, sample size methodology and the analysis of bioequivalence studies for various designs with worked out examples and SAS code have given in detail. This chapter also describes adaptive designs and their utility in bioequivalent assessment.
Chapter 10 deals with the other Phase-I trials where the actual timings of the studies could be anywhere in the drug development process. Dose proportionality, one of the most important assumptions in drug development, has been dealt with. Design, sample size and analysis of dose proportionality, repeat dose assessment and drug interaction studies have been discussed. Various aspects of drug assessment in paediatric studies and relative potency have been covered.
Chapter 11 deals with general issues in Phase-II trials. The contents of this chapter can be found in any textbook on clinical trial. It is not very specific to the early trials.
Chapter 12 deals with one of the most important aspects of drug development i.e., dose response and consequent dose selection. Dose response profile with emphasis on the efficacy window has been well described. Statistical aspects of dose excalation studies, titration studies and designs for estimation have been covered in the subsequent sections. The chapter ends with power assessment of slope and a few worked out example.
Chapter 13 deals with Phase-II trails with toxic therapies. Brief description of single stage, two stage with single endpoint, dual endpoint and survival endpoint has been given. Bayesian approaches in the context of early trials have been discussed.
Chapter 14 deals with interpreting and applying early phase trials results in the context of overall development. Confidence intervals in estimation of population effect with worked out examples have been provided. Guidelines for selecting an effect size for future trial and application of the results have been briefly discussed.
Chapter 15 discusses the ways of constructing Go/No-Go criteria within a simple Bayesian framework. Bayesian approaches have been demonstrated as a useful way of combining prior beliefs with observed data to form posterior response for the outcome of interest. The methodology for constructing exact confidence interval and worked out examples of the proportion have been given. Formulae for prior response and construction of prior response and its effect on posterior distribution have been provided. The last section of this chapter contains Go/No-Go criterion using Bayesian approach, and worked out examples with SAS code have been given for the construction of Go/No-Go criterion.
The contents are relevant to the topics and the book is written in a very lucid manner. This book is useful for students in the fields of pharmacology and biostatistics. Non-statistician would appreciate the fact that statistical concepts have been explained in a simple manner. Statistician would find the statistical formulae and worked out examples very useful. For persons involved in early clinical trials, this book would be an extremely useful reference book.