Statistics in Drug Research.pdf

372
STATISTICS IN DRUG RESEARCH METHODOLOGIES AND RECENT DEVELOPMENTS SHEIN-CHUNG CHOW Statplus, Inc. Yardley, Pennsylvania and Temple University Philadelphia, Pennsylvania JUN SHAO University of Wisconsin-Madison Madison, Wisconsin Copyright © 2002 by Marcel Dekker, Inc. All Rights Reserved.

Transcript of Statistics in Drug Research.pdf

  • STATISTICS IN DRUG RESEARCH

    METHODOLOGIES AND RECENT DEVELOPMENTS

    SHEIN-CHUNG CHOW Statplus, Inc.

    Yardley, Pennsylvania and Temple University

    Philadelphia, Pennsylvania

    JUN SHAO University of Wisconsin-Madison

    Madison, Wisconsin

    Copyright 2002 by Marcel Dekker, Inc. All Rights Reserved.

  • ISBN: 0-8247-0763-X

    This book is printed on acid-free paper.

    HeadquartersMarcel Dekker, Inc.270 Madison Avenue, New York, NY 10016tel: 212-696-9000; fax: 212-685-4540

    Eastern Hemisphere DistributionMarcel Dekker AGHutgasse 4, Postfach 812, CH-4001 Basel, Switzerlandtel: 41-61-261-8482; fax: 41-61-261-8896

    World Wide Webhttp://www.dekker.com

    The publisher offers discounts on this book when ordered in bulk quantities. For more infor-mation, write to Special Sales/Professional Marketing at the headquarters address above.

    Copyright 2002 by Marcel Dekker, Inc. All Rights Reserved.

    Neither this book nor any part may be reproduced or transmitted in any form or by anymeans, electronic or mechanical, including photocopying, microfilming, and recording, orby any information storage and retrieval system, without pemaission in writing from thepublisher.

    Current printing (last digit):10987654321

    PRINTED IN THE UNITED STATES OF AMERICA

    Copyright 2002 by Marcel Dekker, Inc. All Rights Reserved.

  • Series Introduction

    The primary objectives of the Biostatistics series are to provide useful ref-erence books for researchers and scientists in academia, industry, and gov-ernment, and to offer textbooks for undergraduate and/or graduate coursesin the area of biostatistics. This book series will provide comprehensiveand unified presentations of statistical designs and analyses of importantapplications in biostatistics, such as those in biopharmaceuticals. A well-balanced summary will be given of current and recently developed statisticalmethods and interpretations for both statisticians and researchers/scientistswith minimal statistical knowledge who are engaged in applied biostatis-tics. The series is committed to providing easy-to-understand state-of-the-art references and textbooks. In each volume, statistical concepts andmethodologies will be illustrated through real examples.

    Pharmaceutical research and development are lengthy and expensiveprocesses, which involve discovery, formulation, laboratory work, animalstudies, clinical studies, and regulatory submission. Research and develop-ment are necessary to provide substantial evidence regarding the efficacyand safety of a pharmaceutical entity under investigation prior to regula-tory approval. In addition, they provide assurance that the pharmaceuticalentity will possess good characteristics, such as identity, strength, quality,purity, and stability after regulatory approval. Statistics plays an importantrole in pharmaceutical research and development not only to provide a validand fair assessment of the pharmaceuticals under investigation prior to reg-ulatory approval, but also to ensure that the pharmaceutical entities possessgood characteristics with desired accuracy and reliability. This volume cov-ers several important topics in pharmaceutical research and development,such as pharmaceutical validation, including assay and process validation;dissolution testing and profile comparison; stability analysis; bioavailabilityand bioequivalence, including the assessment of in vivo population and indi-vidual bioequivalence and in vitro bioequivalence testing; and key statisticalprinciples in clinical development, including randomization, blinding, sub-stantial evidence, bridging studies, therapeutic equivalence/noninferioritytrials, analysis of incomplete data, meta-analysis, quality of life, and med-

    Copyright 2002 by Marcel Dekker, Inc. All Rights Reserved.

  • iv Series Introduction

    ical imaging. This volume provides a challenge to biostatisticians, phar-maceutical scientists, and regulatory agents regarding statistical method-ologies and recent developments in pharmaceutical research, especially forthose issues that remain unsolved.

    Shein-Chung Chow

    Copyright 2002 by Marcel Dekker, Inc. All Rights Reserved.

  • Biostatistics: A Series of References and Textbooks

    Series EditorShein-Chung Chow

    President, U.S. OperationsStatPlus, Inc.

    Yardley, PennsylvaniaAdjunct ProfessorTemple University

    Philadelphia, Pennsylvania

    1. Design and Analysis of Animal Studies in Pharmaceutical Devel-opment, edited by Shein-Chung Chow and Jen-pei Liu

    2. Basic Statistics and Pharmaceutical Statistical Applications, James E.De Muth

    3. Design and Analysis of Bioavailability and Bioequivalence Studies,Second Edition, Revised and Expanded, Shein-Chung Chow andJen-pei Liu

    4. Meta-Analysis in Medicine and Health Policy, edited by Dalene K.Stangl and Donald A. Berry

    5. Generalized Linear Models: A Bayesian Perspective, edited by DipakK. Dey, Sujit K. Ghosh, and Bani K. Mallick

    6. Difference Equations with Public Health Applications, Lemuel A. Moy6and Asha Seth Kapadia

    7. Medical Biostatistics, Abhaya Indrayan and Sanjeev B. Sarmukaddam8. Statistical Methods for Clinical Trials, Mark X. Norleans9. Causal Analysis in Biomedicine and Epidemiology: Based on Minimal

    Sufficient Causation, Mikel Aickin10. Statistics in Drug Research: Methodologies and Recent

    Developments, Shein-Chung Chow and Jun Shao

    ADDITIONAL VOLUMES IN PREPARATION

    Copyright 2002 by Marcel Dekker, Inc. All Rights Reserved.

  • Preface

    Pharmaceutical research and development is a lengthy process involvingdrug discovery, laboratory development, animal studies, clinical develop-ment, regulatory registration, and postmarketing surveillance. To ensurethe efficacy, safety, and good characteristics of pharmaceutical products,regulatory agencies have developed guidances and guidelines for good phar-maceutical practices to assist the sponsors and researchers in drug researchand development. Even after a pharmaceutical product is approved, it mustbe tested for its identity, strength, quality, purity, and reproducibility be-fore it can be released for use. This book provides not only a comprehensiveand unified presentation of designs and analyses utilized at different stagesof pharmaceutical research and development, but also a well-balanced sum-mary of current regulatory requirements, methodology for design and anal-ysis in pharmaceutical science, and recent developments in the area of drugresearch and development.

    This book is a useful reference for pharmaceutical scientists and bio-statisticians in the pharmaceutical industry, regulatory agencies, andacademia, and other scientists who are in the related fields of pharma-ceutical development and health. The primary focus of this book is on bio-pharmaceutical statistical applications that commonly occur during variousstages of pharmaceutical research and development. This book providesclear, illustrated explanations of how statistical design and methodologycan be used for the demonstration of quality, safety, and efficacy in phar-maceutical research and development.

    The book contains 12 chapters, which cover various important topics inpharmaceutical research and development, such as pharmaceutical valida-tions including assay and process validation, dissolution testing, stabilityanalysis, bioavailability and bioequivalence, randomization and blinding,substantial evidence in clinical development, therapeutic equivalence/non-inferiority trials, analysis of incomplete data, meta-analysis, quality of life,and medical imaging. Each chapter is designed to be self-explanatory forreaders who may not be familiar with the subject matter. Each chapter

    Copyright 2002 by Marcel Dekker, Inc. All Rights Reserved.

  • vi Preface

    provides a brief history or background, regulatory requirements (if any),statistical design and methods for data analysis, recent development, andrelated references.

    From Marcel Dekker, Inc., we thank Acquisitions Editor Maria Allegra,for providing us with the opportunity to work on this project, and Pro-duction Editor Theresa Stockton for her outstanding efforts in preparingthis book for publication. We are deeply indebted to StatPlus, Inc. andthe University of Wisconsin for their support. We would like to expressour gratitude to Mrs. JoAnne Pinto of StatPlus, Inc. for her administrativeassistance, and Mr. Yonghee Lee and Hansheng Wang of the University ofWisconsin for their considerable assistance in most of the numerical workduring the preparation of this book.

    Finally, we are fully responsible for any errors remaining in this book.The views expressed are those of the authors and are not necessarily thoseof their respective company and university. Any comments and suggestionsthat you may have are very much appreciated for the preparation of futureeditions of this book.

    Shein-Chung ChowJun Shao

    Copyright 2002 by Marcel Dekker, Inc. All Rights Reserved.

  • Contents

    Series Introduction

    Preface

    Chapter 1. Introduction1.1 Process of Drug Develop

    1.1.1 Drug Discover1.1.2 Formulation 1.1.3 Laboratory Developme1.1.4 Animal Studies1.1.5 Clinical Developme1.1.6 Regulatory Registratio

    1.2 Regulatory Requiremen1.2.1 Regulatory Mileston1.2.2 International Conference on Harmonizatio1.2.3 U.S. Regulation1.2.4 Institutional Review Board (I1.2.5 investigational New Drug Application (IN1.2.6 New Drug Application (N1.2.7 Advisory Committ

    1.3 Good Pharmaceutical Practic

    1.4

    1.3.11.3.21.3.31.3.4Good Statistics Practice

    Good Laboratory Practice (GGood Clinical Practice (Current Good Manufacturing Practice (cGMP) . FDA Inspection

    Copyright 2002 by Marcel Dekker, Inc. All Rights Reserved.

  • viii Contents

    1.5 Aim1.5.11.5.21.5.31.5.41.5.51.5.6

    of the Book and Outline of,Practical IssueThe Aim and Scope of the BPharmaceutical ValidatiDissolution TesStability Anal sBioavailability and BioequivaleRandomization and Blindi

    1.5.7i.5.81.5.91.5.101.5.111.5.12

    Substantial Evidence in Clinical DevelopmeTherapeutic Equivalence and Noninferiority Analysis of Incomplete Meta-Analysis Quality of Statistics in Medical Im

    Chapter 2.2.12.2

    Pharmaceutical ValidationRegulatory RequirementStandard Cu2.2.1 Model Selecti2.2.2 Weight Select

    2.3 Calibration and Assay Re2.4 Assay Validatio

    2.4.1 Accuracy 2.4.2 Precision 2.4.3 Other Performance Characteristics

    2.5 In-process Controls and Validat2.6 Multiple-Stage

    Chapter 3. Dissolution Testing3.1 USP/NF Dissolution 3.2 Probability of Passing the Dissolution 3.3 Dissolution Profile and Simil3.4 Methods for Assessing Similar

    3.4.1 Model-Dependent Approa3.4.2 Model-Independent Appro3.4.3 The f2 Similarity FaChow and Kis Met3.5

    Copyright 2002 by Marcel Dekker, Inc. All Rights Reserved.

  • Contents ix

    3.6

    3.5.1 The Case of Constant Mean Relative Dissolution Rate3.5.2 The Random Effects Model ApproacChow and Shaos Metho

    Chapter 4. Stability Analysis4.14.24.3

    Statistical Model and DTesting for Batch-to-Batch VariShelf-Life Estimat4.3.1 Fixed-Batches Appro4.3.2 Random-Batches Approa4.3.3 Discussion

    4.4 Two-Phase Shelf-Li4.4.1 A Two-Phase Linear Regression Mo4.4.2 The Second-Phase Shelf-Lif4.4.3 Discussion

    4.5 Discrete Respo4.5.1 The Case of No Batch-to-Batch Variatio4.5.2 The Case of Random Batch4.5.34.5.44.5.5

    4.6 Multiple Components/Ingredients 4.6.1 Shelf-Life Estimation with Multiple Responses .4.6.2 Shelf-Life Estimation with Multiple Ingredients

    Testing for Batch-to-Batch VariatiAn Example Ordinal Respons

    Chapter 5. Bioavailability and Bioequivalence5.15.25.3

    5.4

    Average, Population, and Individual Bioequivalence Statistical Design and Statistical Tests Suggested by the 5.3.1 Testing for 5.3.2 Testing for 5.3.3 Testing for

    Alternative Designs for 5.4.1 The 2 3 Crossover De5.4.2 The 2 3 Extra-Reference Des5.4.3 Comparisons

    Copyright 2002 by Marcel Dekker, Inc. All Rights Reserved.

  • x Contents

    5.5 Tests for 5.5.1 The 2 2 Crossover De5.5.2 The 2 4 Crossover De5.5.3 The 2 3 Desig

    5.6 In Vitro Bioequivalen5.6.1 Six Tests in the FDA Guida5.6.2 Nonprofile Anal5.6.3 Profile Anal

    5.7 Sample Size Determinati5.7.1 Sample Size for IBE Tes5.7.2 Sample Size for PBE Testi5.7.3 Sample Size for In Vitro Bioequivalence Testi

    Chapter 6. Randomization and Blinding6.1 Randomization Mo

    6.1.1 The Population M6.1.2 The Invoked Population M6.1.3 The Randomization Mo

    6.2 Randomization Meth6.2.1 Complete Randomizatio6.2.2 Permuted-Block Randomizatio6.2.3 Adaptive Randomization6.2.4 Discussion

    6.3 The Number of Cente6.4 Effect of Mixed-Up Treatment C6.5 Blinding 6.6 The Integrity of Blin6.7 Analysis Under Breached Blind

    Chapter 7.7.17.2

    7.3

    Substantial Evidence in Clinical DevelopmentReproducibility ProbabilThe Estimated Power Appro7.2.1 Two Samples with Equal Varianc7.2.2 Two Samples with Unequal Variance7.2.3 Parallel-Group DesiThe Confidence Bound Approach

    Copyright 2002 by Marcel Dekker, Inc. All Rights Reserved.

  • Contents xi

    7.4 The Bayesian Appro7.4.1 Two Samples with Equal Varianc7.4.2 Two Samples with Unequal Varianc7.4.3 Parallel-Group Des

    7.5 Applications and Extensi7.5.1 Substantial Evidence with a Single Tr7.5.2 Sample Size Adjustmen7.5.3 Trials with Different De7.5.4 Reproducibility Probabilities for k Tri7.5.5 Binary Dat

    7.6 Generalizability 7.6.1 The Frequentist Appro7.6.2 The Bayesian Appro7.6.3 Applications

    Chapter 8. Therapeutic Equivalence and Noninferiority8.1 Equivalence/Noninferiority T

    8.1.1 Selection of Cont8.1.2 Hypotheses of Equivalence/Noninferiority 8.1.3 Equivalence Limits and Noninferiority Margins . 8.1.4 Design Strateg

    8.2 Assessing Therapeutic Equival8.2.1 Two Commonly Used Approaches8.2.2 Clarification of Conf8.2.3 Equivalence between Proportio

    8.3 Active Control T8.3.1 Establishment of Drug Efficacy in tive Control

    Trials 8.3.2 Flemings Appro

    8.4 Assessment of Drug Efficacy in Active Control Trial8.4.1 Two-Group Parallel Design with Normally Distributed

    8.5

    Data and a Common Varianc8.4.2 Two-Group Parallel Design with Unequal Variances8.4.3 The General CaActive Control Equivalence T8.5.1 Active Control Equivalence Tri

    Copyright 2002 by Marcel Dekker, Inc. All Rights Reserved.

  • xii Contents

    8.5.2 Active Control and Placebo Control Trial8.6 A Bayesian Approach for Active Control Tria

    8.6.1 Balanced Design with Known Varianc8.6.2 Unbalanced Design with Unknown Varianc

    8.7 Discussion

    Chapter 9. Analysis of Incomplete Data9.19.29.39.4

    9.5

    9.6

    Mechanism of Missing VaThe Power of ANOVA TImputation for Missing Crossover Desi9.4.1 Estimation of Treatment Eff9.4.2 Assessing Individual Bioequivalen9.4.3 Handling Informative Missingness by TransformationAnalysis of Last Observati9.5.1 Intention-to-Treat Anal9.5.2 Last Observation Carry-Forw9.5.3 Analysis of Last ObservatiAnalysis of Longitudinal 9.6.1 Selection M9.6.2 The Method of Groupi

    Chapter 10. Meta-Analysis10.1 Traditional Approaches

    10.1.1 The Approach of p-Va10.1.2 Treatment-by-Study Interacti10.1.3 The Average Treatment Differen10.1.4 Quantitative and Qualitative Interaction

    10.2 Testing of No Qualitative Interact10.2.1 Intersection-Union 10.2.2 An Aggregated

    10.3 Assessing Drug Eff10.3.1 Simultaneous 10.3.2 Two-Step 10.3.3 Evaluation of Equival

    10.4 Multiple Prod

    Copyright 2002 by Marcel Dekker, Inc. All Rights Reserved.

  • ~ontents Xll" 1"

    10.4.1 Parallel De10.4.2 Crossover D

    10.5 Bioequivalence St10.5.1 Meta-Analysis fer 10.5.2 Meta-Analysis fcr 10.5.3 Meta-Analysis for

    10.6 Discussic

    Chapter 11. Quality of Life11.1 Definition and Valida

    11.1.1 Definitio11.1.2 Validati11.1.3 Validit11.1.4 Reliabilit11.1.5 Reproducibilit

    11.2 Responsiveness and Sensiti11.3 QOL Score

    11.3.1 Number cf Composite Sc11.3.2 Subscales to Be Grouped in Each Composite Score 11.3.3 Optimal Weights cf Subsc11.3.4 An Exam

    11.4 Statistical M11.4.1 ~ime Series 11.4.2 Parallel Questionna

    11.5 Statistical 11.5.1 Multiplicit11.5.2 Sample Size Determinati11.5.3 Calibration with Life 11.5.4 Utility An

    Chapter 12. Medical Imaging12.1 Medical Imaging

    12.1.1 Contrast A12.1.2 Diagnostic Radicpharmaceutica12.1.3 Regulatory Requireme12.1.4 Statistical An

    Copyright 2002 by Marcel Dekker, Inc. All Rights Reserved.

  • xiv Contents

    12.2 ROC Analysi12.2.1 The Roe Cu12.2.2 Characteristics of the ROC 12.2.3 Comparison of ROC Cur

    12.3 The Analysis of Blinded-Reader Stud12.3.1 Design of Blinded-Reader Stud12.3.2 Assessing Interreader Agree

    12.4 Diagnostic Acc12.4.1 Diagnostic Accuracy with an Imperfect Reference

    Test 12.4.2 Diagnostic Accuracy without Gold Standard

    12.5 Sample

    References

    Copyright 2002 by Marcel Dekker, Inc. All Rights Reserved.

  • Chapter 1

    Introduction

    In recent years, a series of prescription drugs were pulled from the mar-ketplace due to safety problems. Table 1.1.1 lists drugs recalled by theU.S. Food and Drug Administration (PDA) between 1993 and 2000, andtheir reasons for recall. As it can be seen from Table 1.1.1, the mostcommonly seen reasons for recalls were safety concerns regarding cardiac/cardiovascular problems and liver toxicity/failure. This has renewed con-cerns that the drug development and regulatory approval process moves tooquickly. In the United States, however, drug development is a lengthy andcostly process. On average, it takes about 12 years to obtain regulatoryapproval for a new drug. The cost is estimated to be approximately 300to 450 million U.S. dollars. In many cases (e.g., the development of drugproducts for severe or life-threatening diseases such as cancer and AIDS), ithas been criticized that the drug development and regulatory process takestoo long for a promising drug product to benefit patients with severe orlife-threatening diseases. The lengthy and costly process of drug develop-ment is necessary, not only to ensure that the drug under investigation isefficacious and safe before it can be approved for use in humans, but also toassure that the drug product meets regulatory requirements for good drugcharacteristics of identity, strength (or potency), purity, quality, stability,and reproducibility after the drug is approved.

    Drug development is not only used to efficiently identify promising com-pounds with fewer false negatives/positives, but its also utilized to scien-tifically evaluate the safety and efficacy of these compounds with certaindegrees of accuracy and reliability. To ensure the success of drug devel-opment, most regulatory agencies, such as the FDA, have issued numer-ous guidelines and/or guidances to assist the sponsors in fulfilling regu-latory requirements of safety and efficacy, and good drug characteristicsof identity, strength, purity, quality, stability, and reproducibility. The

    Copyright 2002 by Marcel Dekker, Inc. All Rights Reserved.

  • 2 Chapter i. Introduction

    sponsors are strongly recommended to comply with the regulations andguidelines/guidances set forth by the regulatory agencies for best pharma-ceutical practices. During the drug development process, practical issuesin design, execution, analysis, report, and interpretation of studies thatare necessarily conducted for regulatory submission, review, and approvalare inevitably encountered. These practical issues may include violationsor noncompliance of study protocol, invalid study design, misuse or abuseof statistics for data analysis, misinterpretations of the study results, andincorrectly addressed questions with wrong answers. These practical issuesoften result in a misleading conclusion for the evaluation of the drug underinvestigation.

    A brief description of the process of drug development is given in 1.1.Regulatory requirements for obtaining drug approval is briefly introducedin 1.2. 1.3 reviews the concepts of good laboratory practice (GLP), goodclinical practice (GCP), and current good manufacturing practices (cGMP)for good pharmaceutical practices. The role of good statistics practice(GSP) in good pharmaceutical practices is discussed in 1.4. The last sec-tion provides a description of the aim and scope of this book and an outlineof a number of practical issues in various stages of the drug developmentprocess that are studied in the subsequent .chapters.

    Table 1.1.1. Recalled Drugs Between 1993o2000

    Drug Name Drug Type Year Reason for RecallPropulsid Heartburn 2000Rezulin Diabetes 2000Raxar Antibiotic 1999Hismanal Allergy 1999Duract Arthritis 1998Posicor Blood pressure 1998Seldane Allergy 1998Redux Diet 1997Pondimin Diet 1997Manoplax Congestive heart 1993

    failure

    Cardiac problemsLiver toxicityCardiovascular problemsCardiac problemsLiver failureDangerous drug interactionDangerous drug interactionPossible heart valve damagePossible heart valve damageIncreased risk of death

    Source: U.S. Food and Drug Administration; compiled from AP wire reports

    Copyright 2002 by Marcel Dekker, Inc. All Rights Reserved.

  • 1.1. Process of Drug Development 3

    1.1 Process of Drug Development

    The ultimate goal of the drug development process is to produce new drugproducts for marketing. The drug development process, involving drug dis-covery, formulation, laboratory development, animal studies for toxicity,clinical development, and regulatory registration, is a continual process. Itconsists of different phases of development, including nonclinical develop-merit (e.g., drug discovery, formulation, and laboratory development), pre-clinical development (e.g., animal studies), and clinical development (e.g.,clinical studies). These phases may occur in sequential order or be over-lapped during the development process. To provide a better understandingof the drug development process, critical stages or phases of drug develop-ment are briefly and separately outlined below.

    1.1.1 Drug Discovery

    Drug discovery consists of two phases, namely drug screening and drug leadoptimization. The purpose of drug screening is to identify a stable and re-producible compound with fewer false-negative and false-positive results.At the drug screening phase, the mess compounds are necessarily screenedto distinguish those that are active from those that are not. For this pur-pose, a multiple-stage procedure is usually employed. A compound mustpass all stages to be active. The commonly encountered problem in drugscreening for activity is the choice of dose. In practice, the dose is oftenchosen either too low to show activity or too high to exhibit a toxic effect.Drug screening for activities could be a general screening, based on phar-macological activities in animals or in vitro assays, or a targeted screeningbased on specific activities, such as that of an enzyme. Pharmacologicalactivity is usually referred to as the selective biological activity of chemicalsubstances on living matter. A chemical substance is called a drug if ithas selective activity with medical value in the treatment of disease. Leadoptimization is a process of finding a compound with some advantages overrelated leads based on some physical and/or chemical properties. To effi-ciently identify a stable and reproducible compound during the process ofdrug screening and lead optimization, statistical quality control based onsome established standards for acceptance and rejection is critical. The suc-cess rate for identifying a promisir[g active compound is relatively low. Asa result, there may be only a few compounds that are identified as promis-ing active compounds. These identified compounds have to pass the stagesof laboratory development and animal studies before they can be used inhumans. In practice, it is estimated that about one in 8 to 10 times 103compounds screened may finally reach the stage of clinical development forhuman testing.

    Copyright 2002 by Marcel Dekker, Inc. All Rights Reserved.

  • 4 Chapter 1. Introduction

    1.1.2 Formulation

    When a new drug is discovered, it is important to develop an appropri-ate dosage form /or formulation) for the drug so that it can be deliveredefficiently to the site of action of the body for the intended disease. It is nec-essary to develop appropriate formulation with an adequate dose to achieveoptimal therapeutic effect in patients with the intended disease. The com-monly seen pharmaceutical dosage forms include tablet, capsule, powder,liquid suspension, aerosol, cream, gel, solution, inhalation, lotion, paste,suspension, and suppository. Formulation is a synthesis of a new chemicalentity, which modifies the structure ,to enhance biologic activity. At theinitial development of formulation, a small amount of the drug productunder development/investigation is usually produced for laboratory testingsuch as solubility and accelerated stability. The optimization of formula-tions usually occurs in a scale-up from the laboratory batch to commercialor production batch, process control, and validation. The purpose of theoptimization of formulations is to optimize the responses of the drug de-livery system by identifying critical formulations and/or process variablestliat can be manipulated or a~e controllable. The optimization of formu-lation helps in meeting regulatory requirements for identity, strength /orpotency/, quality, purity, stability, and reproducibility of the drug product.

    1.1.3 Laboratory Development

    During the drug development process, laboratory development usually ap-plies to nonclinical safety studies, such as animal studies and in vitro as-says studies. For each newly discovered compound, the FDA requires thatanalytical methods and test procedures be developed for determining theactive ingredient(s) of the compound in compliance with USP/NF (UnitedStates Pharmacopedia and National Formulary) standards for the identity,strength, quality, purity, stability, and reproducibility of the compound. Ananalytical method is usually developed based on some instrument, such ashigh-performance liquid chromatography (HPLC). As a result, the estab-lishment of the standard curve for calibration of an instrument is criticalfor the assurance of accuracy and reliability of the results obtained from theanalytical method or test procedures. The FDA indicates that an instru-ment must be suitable for its intended purposes and be capable of producingvalid results with a certain degree of accuracy and reliability. An analyticalmethod is usually validated according to some performance characteristicssuch as accuracy, precision, linearity, range, specificity, limit of detection,limit of quantitation, and ruggedness as specified in the USP/NF standards.The instrument used for the development qf an analytical method or testprocedures must be calibrated, inspected, ahd checked routinely according

    Copyright 2002 by Marcel Dekker, Inc. All Rights Reserved.

  • 1.1. Process o[ Drug Development 5

    to written procedures for suitability, sensitivity, and responsiveness.

    1.1.4 Animal StudiesIn drug development, because certain toxicities, such as the impairment offertility, teratology, mutagenicity, and overdosage, cannot be investigatedethically in humans, the toxicity of the drug product is usually assessedin either in vitro assays or animal models. Animal models are often con-sidered as a surrogate for human testing, under the assumption that theycan be predictive of the results in humans. Animal studies involve doseselection, toxicological testing for toxicity and carcinogenicity, and animalpharmacokinetics. For the selection of an appropriate dose, dose-response(dose:ranging) studies in animals are usually conducted to determine effective dose, such as the median effective dose (EDs0). In addition, druginteraction, such as potentiation, inhibition, similar joint action, synergism,and antagonism, are also studied. In general, animal toxicity studies areintended to alert the clinical investigators of the potential toxic effects asso-ciated with the investigational drugs so that those effects may be watchedfor during the clinical investigations. In most circumstances, acute andsubacute toxicity studies are typically conducted in rodent and nonrodentmammalian species. It is necessary to conduct segments I, II, and III repro-ductive toxicity studies and chronic and carcinogenic studies to provide thecomplete spectrum of toxicological effects an investigational drug can elicit.In addition, it is necessary to perform absorption distribution, metabolism,and excretion (ADME) studies in animals in order to identify those phar-macokinetic parameters that are similar to those in humans and to verifythe applicability of the animal species used in the toxicological tests.

    1.1.5 Clinical Development

    Clinical development involves phases 1-4 of clinical investigation. The pri-mary objective of phase 1 is not only to determine the metabolism andpharmacologic activities of the drug in humans, the side effects associatedwith increasing doses, and the early evidence on effectiveness, but also toobtain sufficient information about the drugs pharmacokinetics and phar-macological effects to permit the design of well-controlled and scientificallyvalid phase 2 studies. Phase 2 studies are the first controlled clinical stud-ies of the drug. The primary objectives of phase 2 studies are not only tofirst evaluate the effectiveness of a drug based on clinical endpoints for aparticular indication or indications in patients with the disease or conditionunder study, but also to determine the dosing ranges and doses for phase3 studies and the common short-term side effects and risks associated withthe drug. Phase 3 studies are expanded controlled and uncontrolled tri-

    Copyright 2002 by Marcel Dekker, Inc. All Rights Reserved.

  • Chapter 1. Introduction

    als. The primary objectives are to (a) gather additional information aboutthe effectiveness and safety needed to evaluate the overall benefit-risk re-lationship of the drug, and (b) to provide an adequate basis for physicianlabeling. Phase 4 studies are usually conducted to further elucidate theincidence of adverse reactions and determine the effect of a drug on mor-bidity or mortality. Phase 4 studies are considered useful market-orientedcomparison studies against competitor products.

    According to the 1988 FDA guideline, the efficacy and safety of a studydrug can only be established through the conduct of adequate and well-controlled cl.ipical trials (FDA, 1988). Table 1.1.2 lists characteristics of adequate and well-controlled study as specified in 21 CFR 314.126.

    Table 1.1.2. Characteristics of an Adequate and Well-Controlled Study

    CriterionObjectivesMethods of analysis

    Design

    Selection of subjects

    Assignment of subjects

    Participants of studies

    Assessment of responsesAssessment of the effect

    CharacteristicsClear statement of investigation purposeSummary of proposed or actual methods ofanalysisValid comparison with a control to providea quantitative assess~nent of drug effectAdequate assurance of the disease orconditions under studyMinimization of bias and assurance ofcomparability of groupsMinimization of bias on the part of subjects,observations, and analysisWell defined and reliableRequirements of appropriate statisticalmethods

    1.1.6 Regulatory Registration

    .At the end of clinical development, the sponsor is required to submit all ofthe information collected from the studies conducted at different phases ofdrug development for regulatory review and approval. Regulatory reviewis a very intensive process, which involves reviewers from different disci-plines such as chemistry, toxicology, clinical pharmacology, medicine, andbiostatistics. The purpose of this intensive review is to make sure that thereis substantial evidence to support the safety and efficacy of the drug prod-uct under investigation. In addition, the r~view assures that the proposed

    Copyright 2002 by Marcel Dekker, Inc. All Rights Reserved.

  • 1.2. ~egul~or~ Requirements 7

    labeling for the investigational drug is appropriate for the target patientpopulation with the intended indication. At the stage of regulatory regis-tration, it is helpful to build up a communication with the FDA for adviceon deficiencies or concerns regarding the safety and efficacy of the investi-gational drug. The FDA may approve the drug product if the sponsor hascommitted to conduct additional clinical studies to address the deficienciesand/or concerns/issues raised during the regulatory review/approval pro-cess. These additional clinical studies are usually referred to as phase 3Bstudies or pending approval clinical studies.

    1.2 Regulatory Requirements

    1.2.1 Regulatory Milestones

    A century ago, the pharmaceutical industry in the United States was es-sentially unregulated. Drug companies could advertise their products astreatments for any and all diseases. The first effective legislation regardingdrug products can be traced back to the Biologics Act of 1902. This wasPrecipated by a tragedy involving diptheria antitoxin contaminated withtetanus, which resulted in the death of 12 children and subsequently led tothe passage of the Pure Food and Drugs Act of 1906. The purpose of thisact was to prevent misbranding and adulteration of food and drugs and yetthe scope is rather limited. In 1912, the Sherley Amendment to the actwas passed to prohibit the labeling of medicines with false and fraudulentclaims. The concept of testing marketed drugs in human subjects did notbecome a public issue until the Elixir Sulfanilamide disaster occurred inthe late 1930s. The disaster was regarding a liquid formulation of a sulfadrug, which had never been tested in humans before its marketing. Thisdrug caused more than i00 deaths and raised the safety concern that ledto the passing of the Federal Food, Drug, and Cosmetic Act in 1938. Thisact extended regulatory regulation to cosmetics and therapeutic devices.More importantly, it requires the pharmaceutical companies to submit fullreports of investigations regarding the safety of new drugs. In 1962, asignificant Kefauver-Harris Drug Amendment was passed, which not onlystrengthened the safety requirements for new drugs, but also establishedan efficacy requirement for new drugs for the first time. In 1984, Congresspassed the Price Competition and Patent Term Restoration Act to providean increased patent protection to compensate for patent life lost duringthe approval process. Based on this act, the FDA was authorized to ap-prove generic drugs only based on bioavailability and bioequivalence trialson healthy subjects.

    Copyright 2002 by Marcel Dekker, Inc. All Rights Reserved.

  • Chapter 1. Introduction

    1.2.2 International Conference on Harmonization

    For the marketing approval of pharmaceutical entities, the regulatory pro-cess and requirements may vary from country (region) to country (region).The necessity to standardize these similar and yet different regulatory re-quirements has been recognized by both regulatory authorities and thepharmaceutical industry. Hence, the International Conference on Harmo-nization (ICH), which consists of the European Community (EC), Japan,and the United States, was organized to provide an opportunity for impor-tant initiatives to be developed by regulatory authorities as well as industryassociations for the promotion of international harmonization of regulatoryrequirements. The ICH, however, is only concerned with tripartite har-monization of technical requirements for the registration of pharmaceuticalproducts among the three regions of the EC, Japan, and the United States.That is, the information generated in any of the three areas of the EC,Japan and the United States would be acceptable to the other two areas.Basically, the ICH consists of six parties that operate in these three re-gions, which include the European Commission of the European Union,the European Federation of Pharmaceutical Industries Association (EF-PIA), the Japanese Ministry of Health and Welfare (MHW), the JapanesePharmaceutical Manufacturers Association (JPMA), the Center for DrugEvaluation and Research (CDER) of FDA, and the Pharmaceutical Re-search and Manufacturers of America (PhRMA). To assist sponsors in thedrug development process, the ICH has issued a number of guidelines anddraft guidances. These guidelines and guidances are summarized in Table1.2.1.

    1.2.3 U.S. Regulations

    In this section, for illustration purposes, we focus on the regulatory processand requirements currently adopted in the United States. The current regu-lations for conducting studies, regulatory submission, review, and approvalof results for pharmaceutical entities (including drugs, biological products,and medical devices under investigation) in the United States can be foundin the Code of Federal Regulations (CFR). Table 1.2.2 lists the most rele-vant regulations with respect to drug development and approval. The Cen-ter for Drug Evaluation and Research (CDER) of the FDA has jurisdictionover the administration of regulations and the approval of pharmaceuticalproducts classified as drugs. These regulations include investigational newdrug application (IND) and new drug application (NDA) for new drugs,orphan drugs, and over-the-counter (OTC) human drugs and abbreviatednew drug application (ANDA) for generic drugs.

    Copyright 2002 by Marcel Dekker, Inc. All Rights Reserved.

  • 1.2. Regulatory Requirements 9

    Table 1.2.1. ICH Clinical Guidelines and Draft Guidelines

    GuidelinesSafety1 S1A: The need for long-term rodent carcinogenicity studies of

    pharmaceuticals2 SIB: Testing for carcinogenicity of pharmaceuticals3 S1C: Dose selection for carcinogenicity studies of pharmaceuticals4 S1C(R): Guidance on dose selection for carcinogenicity studies

    pharmaceuticals: Addendum on a limit dose and related notes5 S2A: Specific aspects of regulatory genotoxicity tests for

    pharmaceuticals6 S2B: Genotoxicity: A standard battery for genotoxicity testing of

    pharmaceuticals7 SJA: Toxicokinetics: The assessment of systemic exposure in

    toxicity studies8 SJB: Pharmacokinetics: Guideline for repeated dose tissue

    distribution studies9 S4A: Duration of chronic toxicity testing in animals (rodent and

    nonrodent toxicity testing)10 S5A: Detection of toxicity to reproduction for medicinal products11 S5B: Detection of toxicity to reproduction for medicinal products:

    Addendum on toxicity to male fertility12 $6: Preclinical safety evaluation of biotechnology -- derived

    pharmaceuticalsJoint Safety/Efficacy1 M4: Common technical document

    Efficacy1 E1A: The extent of population exposure to assess clinical safety

    for drugs intended for long-term treatment of non-life-threatening conditions

    2 E2A: Clinical safety data management: Definitions and standardsfor expedited reporting

    3 E2B:Data elements for transmission of individual case report forms4 E2C: Clinical safety data management: Periodic safety update

    reports for marketed drugs5 EJ: Structure and content of clinical studies

    Copyright 2002 by Marcel Dekker, Inc. All Rights Reserved.

  • 10 Chapter 1. Introduction

    Table 1.2.1. (Continued)

    Efficacy6 E4: Dose-response information to support drug registration7 E5: Ethnic factors in the acceptability of foreign clinical data8 E6: Good clinical practice: Consolidated guideline9 ET: Studies in support of special populations: Geriatrics

    10 E8: General considerations for clinical trials11 E9: Statistical principles for clinical trials12 Ell: Clinical investigation of medicinal products in the

    pediatric populationQuality

    56789

    Q1A:Q1B:Q1C:Q2A:Q2B:Q3A:Q3B:Q3C:

    Stability testing of new drug substances and productsPhotostability testing of new drug substances and productsStability testing for new dosage formsTest on validation of analytical proceduresValidation of analytical procedures: MethodologyImpurities in new drug substancesImpurities in new drug productsImpurities: Residual solvents

    Q5A: Viral safety evaluation of biotechnology products derivedfrom cell lines of human or animal origin

    10 Q5B: Quality of biotechnology products: analysis of theexpression construct in cells used for production of r-DNAderived protein products

    11 Q5C: Quality of biotechnology products: Stability testing ofbiotechnological/biological products

    12 Q5D: Quality of biotechnology/biological products: Derivationand characterization of cell substrates used for productionof biotechnological/biological products

    13 Q6A: Specifications: Test procedures and acceptance criteria fornew drug substances and new drug products: Chemicalsubstances

    14 Q6B: Test procedures and acceptance criteria for biotechno-logical/biological products

    Copyright 2002 by Marcel Dekker, Inc. All Rights Reserved.

  • 1.2. Regulatory Requirements 11

    Table 1.2.1. (Continued)

    Draft GuidelinesSafety1 $7: Safety pharmacology studies for human pharmaceuticalsEfficacy1 El0: Choice of control group in clinical trials2 EI2A: Principles for clinical evaluation of new antihypertensive

    drugsQuality1 Q1A(R): Stability testing of new drug substances and products2 Q3A(R): Impurities in new drug substances3 Q3B(R): Impurities in new drug products4 QTA: Good manufacturing practice for active pharmaceutical

    ingredients

    Table 1.2.2. U.S. Codes of Federal Regulation (CFR) for Clinical TrialsUsed to Approve Pharmaceutical Entities

    CFR Number Regulations21 CFR 5O21 CFR 5621 CFR 312

    Subpart E21 CFR 314

    Subpart CSubpart H

    21 CFR 601

    Subpart E21 CFR 31621 CFR 32021 CFR 33021 CFR 81221 CFR 81421 CFR 6021 CFR 20121 CFR 202

    Protection of human subjectsInstitutional review boards (IRB)Investigational new drug application (IND)Treatment INDNew drug application (NDA)Abbreviated applicationsAccelerated approvalEstablishment license and product license applications

    (ELA and PLA)Accelerated approvalOrphan drugsBioavailability and bioequivalence requirementsOver-the-counter (OTC) human drugsInvestigational device exemptions (IDE)Premarket approval of medical devices (PMA)Patent term restorationLabelingPrescription drug advertsing

    Copyright 2002 by Marcel Dekker, Inc. All Rights Reserved.

  • 12 Chapter 1. Introduction

    1.2.4 Institutional Review Board (IRB)

    Since 1971, the FDA has required that all proposed clinical studies bereviewed both by the FDA and an institutional review board (IRB). Theresponsibilities of an IRB are to evaluate the ethical acceptability of theproposed clinical research and to examine the scientific validity of the studyto the extent needed to be confident that the study does not expose itssubjects to an unreasonable risk. The composition and function of an IRBare subject to the FDA requirements. Section 56.107 in Part 56 of 21CFR states that each IRB should have at least five members with varyingbackgrounds to promote a con~plete review of research activities commonlyconducted by the institution. Note that in order to avoid conflict of interestand to provide an unbiased and objective evaluation of scientific merits,ethical conduct of clinical trials, and protection of human subjects, theCFR enforces a very strict requirement for the composition of members ofan IRB. These strict requirements include (i) no IRB is entirely composedof one gender, (ii) no IRB may consist entirely of members of one profession,(iii) each IRB should include at least one member whose primary concernsare in the scientific area and at least one member whose primary concernsare in nonscientific area, and (iv) no IRB should have a member participatein the IRBs initial or continuous review of any project in which the memberhas a conflicting interest, except to provide information requested by theIRB.

    1.2.5 Investigational New Drug Application (IND)

    Before a drug can be studied in humans, its sponsor must submit an INDto the FDA. Unless notified otherwise, the sponsor may begin to investi-gate the drug 30 day~ after the FDA has received the application. TheIND requirements extend throughout the period during which a drug is un-der study. By the time an IND is filed, the sponsor should have sufficientinformation about the chemistry, manufacturing, and controls of the drugsubstance and drug product to ensure the identity, strength, quality, purity,stability, and reproducibility of the investigational drug covered by the IND.In addition, the sponsorshould provide adequate information about phar-macological studies for absorption, distribution, metabolism, and excretion(ADME), and acute, subacute, and chronic toxicological studies and repro-ductive tests in various animal species to support that the investigationaldrug is reasonably safe to be evaluated in humans.

    Copyright 2002 by Marcel Dekker, Inc. All Rights Reserved.

  • 1.3. Good Pharmaceutical Practices 13

    1.2.6 New Drug Application (NDA)For the approval of a new drug, the FDA requires that at least two ad-equate well-controlled clinical studies be conducted in humans to demon-strate substantial evidence of the effectiveness and safety of the drug. Sub-stantial evidence can be obtained through adequate and well-controlledclinical investigations. Section 314.50 of 21 CFR specifies the format andcontent of an NDA, which contains the application form (365H), index,summary, technical sections, samples and labeling, and case report formsand tabulations. The technical sections include chemistry, manufactur-ing, and controls (CMC), nonclinical pharmacology and toxicology, humanpharmacology and bioavailability, microbiology (for anti-infective drugs),clinical data, and statistics. As a result, the reviewing disciplines includechemistry reviewers for the CMC section; pharmacology reviewers for non-clinical pharmacology and toxicology; medical reviewers for clinical datasection; and statistical reviewers for statistical technical section.

    1.2.7 Advisory Committee

    The FDA has also established advisory committees in designated drugclasses and subspecialities each consisting of clinical, pharmacological, andstatistical experts and one consumer advocate (not employed by the FDA).The responsibilities of the committees are to review the data presented inthe NDAs and to advise the FDA as to whether there exists substantialevidence of safety and effectiveness based on adequate and well-controlledclinical studies. As a result, the advisory committees address the followingquestions posted by the FDA followed by an intensive discussion at the endof the advisory committee meeting:

    1. Are there two or more adequate and well-controlled trials?

    2. Have the patient populations been well enough characteristized?

    3. Has the dose-response relationship been sufficiently characterized?4. Do you recommend the use of the drug for the indication sought by

    the sponsor for the intended patient population?Note that the FDA usually follows the recommendations made by the advi-sory committee for marketing approval, though they do not have to legally.

    1.3 Good Pharmaceutical Practices

    To ensure the success of drug development in compliance with regulatoryrequirements for approval, good pharmaceutical practices have to be imple-

    Copyright 2002 by Marcel Dekker, Inc. All Rights Reserved.

  • Chapter 1. Introduction

    mented. Good pharmaceutical practices are established standards to assurethe drug under investigation meets the requirements of the Federal Food,Drug, and Cosmetic Act related to safety and efficacy before marketingapproval and drug characteristics of identity, strength, quality, purity, sta-bility, and reproducibility after approval. Good pharmaceutical practicesinclude good laboratory practice (GLP), good clinical practice (GCP), current good manufacturing practice (cGMP). GLP, GCP, arid cGMP areregulations governing the conduct of preclinical safety studies and in vitrostudies, clinical studies, and the conduct of manufacturing operations, re-spectively. GLP, GCP, and cGMP are briefly described in the subsequentsections.

    1.3.1 Good Laboratory Fractice (GLP)Good laboratory practice (GLP) for nonclinical laboratory studies is cod-ified in 21 CFR 58. GLP applies to animal and in vitro studies for safetyassessment. Similar to cGMP, GLP covers regulations with respect to re-quirements for organization and personnel, facilities, equipment, testingfacilities operations, test and control articles, protocol for and conduct ofa nonclinical laboratory study, records and reports, and disqualification oftesting facilities. GLP requires the existence of a quality assurance unit(QAU). The QAUs responsibilities include that (i) sampling plan, procedure, acceptance/rejection criteria are properly documented, (ii) anyidentified deviations from standard operating procedure (SOP) or protocolare properly corrected and documented, (iii) internal audits are properlyconducted and documented, and (iv) form 483 notice of observations as theresult of an FDAs inspection are properly addressed and documented.

    1.3.2 Good Clinical Practice (GCP)Good clinical practice (GCP) is usually referred to as a set of standards forclinical studies to achieve and maintain high-quality clinical research in asensible and responsible manner. The FDA, the Committee for ProprietaryMedicinal Products (CPMP) for the European Community, the Ministryof Health and Welfare of Japan, and agencies in other countries have eachissued guidelines on good clinical practices. For example, the FDA pro-mulgated a number of regulations and guidelines governing the conduct ofclinical studies from which data was to be used to support applications formarketing approval of drug products. The FDA regulations referring toGCP are specified in CFR parts 50 (protection of human subjects), 56 (in-stitutional review boards, IRB), 312 (investigational new drug application,IND), and 314 (new drug application, NDA). Note that in 1992, Dr. B. Lisook at the FDA assembled a GCP packet to assist sponsors in the

    Copyright 2002 by Marcel Dekker, Inc. All Rights Reserved.

  • 1.3. Good Pharmaceutical Practices 15

    planning, execution, data analysis, and submission of results to the FDA(Lisook, 1992). On the other hand, the European Community establishedprinciples for their own GCP standards in all four phases of clinical in-vestigation of medicinal products in July 1990. The ICH also issued theGuideline on Good Clinical Practice: Consolidated Guideline. This guide-line defines the responsibilities of sponsors, monitors, and investigators inthe initiation, conduct, documentation, and verification of clinical studiesto establish the credibility of data and to protect the rights and integrityof study participants.

    In essence, GCP deals with patients protection and the equality of dataused to prove the efficacy and safety of a drug product. GCP engures thatall data, information, and documents .related to a clinical study can beconfirmed as being properly generated, recorded, and reported through theinstitution by independent audits.

    1.3.3 Current Good Manufacturing Practice (cGMP)cGMP was promulgated in 1962 as part of the Federal Food, Drug and Cos-metic Act to afford greater consumer protection during the manufacture ofpharmaceutical products, cGMP is now codified in 21 CFR 211, whichprovides minimum requirements for the manufacture, processing, packing,and holding of a drug product. Its purpose is to assure that the drug prod-uct meets the requirements of safety, identity, strength, quality, and puritycharacteristics that it purports to possess, cGMP covers regulations withrespect to requirements for organization and personnel, buildings and facil-ities, equipment, control of components and drug product containers andclosures, production and process control, packing and process control, hold-ing and distribution, laboratory controls, records and reports, and returnedand salvaged drug products. Failure to comply with cGMP renders the drugadulterated and the person responsible subject to regulatory action.

    Record retention is an important part of cGMP. The FDA suggests thatrecord retention should follow the principles of (i) two years after FDAapproval of research or marketing permit, (ii) five years after applicationfor IND, and (iii) two years if no application filed.

    1.3.4 FDA Inspection

    One of the FDAs major responsibilities is to enforce the Federal Food,Drug, and Cosmetic Act for the protection of public health. This is oftendone through the FDA Facility Inspection Program, which is to assure thatthe manufacturer is in compliance with all applicable FDA regulations. Thepurpose of an FDA inspection is multifold. First, it is to determine and to

    Copyright 2002 by Marcel Dekker, Inc. All Rights Reserved.

  • 16 Chapter 1. Introduction

    evaluate a firms adherence to the concepts of cGMP. Second, it is to ensurethat production and control procedures include all reasonable precautionsto ensure the identity, strength, quality, purity, stability, and reproducibil-ity of the finished products. In addition, it is to identify deficiencies thatcould lead to the manufacturing and distribution of products in violationof the Federal Food, Drug, and Cosmetic Act. Furthermore, it is to obtaincorrection of those deficiencies and to ensure that new drugs are manufac-tured by essentially the same procedures and the same formulations as theproducts used as the basis for approval.

    Basically, the FDA may initiate different types of inspections dependingupon the circumstances at different stages of drug research and develop-ment. These types of inspections include routine scheduled/unscheduled,survey, compliant, recall, bioresearch monitoring, government contract, andpreapproval inspections. The process of a FDA inspection usually beginswith a notice of inspection (form 482). The sponsor should beprepared foran FDA inspection upon the receipt of 482. After the inspection, the FDAinspector(s) will prepare an establishment inspection report (EIR) includ-ing 483 notice of observations. The sponsor should address 483 issues fullyin response to the FDA Local Field Office. If no action is required as theresult of the sponsors response to 483 and EIR, the sponsor is considered tohave met the requirement of the FDAs inspection. However, if an actionis required, samples will be sent to the FDA District Office ComplianceOffice and a warning letter will be issued. Failure to comply with cGMP,GLP, and/or GCP could subject the sponsor to a regulatory action, suchas seizure, injunction, prosecution, citation (483 notice of observations),detention (hold of shipment), fine, affection of license or permits, or importdetention. For a better understanding, Figure 1.3.1 provides a flowchartfor the FDA inspection process.

    In the FDAs inspection for GLP, cGMP, and GCP, some commonlyobserved errors should be avoided. These errors inlcude (i) document notsigned, (ii) incomplete records, e.g., missing date, time, specification num-ber, batch number, and anything else left blank, etc., (iii) correction notsigned, (iv) out of specific condition, e.g., not performed at proper time specified in the study protocol, (v) incorrect sampling plan, e.g., sampleswere not uniformly and randomly selected from the target population toconstitute a representative sample, (vi) incorrect information, e.g., wrongcalculation and typos in date and transposed number, etc., and (vii) incor-rect techniques, e.g., inappropriate use of pencil, blue ink, white out/liquidpaper, and crossed out mistakes. The FDA inspection for GLP, cGMP, andGCP assures not only that the sponsor is in compliance with all applicableFDA regulations during the drug development process, but also that thedrug product meets regulatory requirements for the good drug characteris-tics of identity, strength, quality, purity, Stability, and reproducibility.

    Copyright 2002 by Marcel Dekker, Inc. All Rights Reserved.

  • 1.3. Good Pharmaceutical Practices 17

    482Notice of Inspection

    ERIEstablishment

    Inspection Report

    [Notice of Observations[~

    (gz product samples)/

    FDA Local Field Office

    Supervisor

    Action Required483 & EIR

    & Samples sent to

    FDADistrict Office

    Compliance Officer

    No Action Requiredin response to

    483 & EIR

    Discussed inExit Interview

    copy sent toCompany

    Company Responseto 483

    FDADistrict Director

    Warning Letterissued

    Figure 1.3.1. FDA Inspection Process

    Copyright 2002 by Marcel Dekker, Inc. All Rights Reserved.

  • 18 Chapter 1. Introduction

    1.4 Good Statistics Practice

    Good statistics practice (GSP) is defined as a set of statistical principlesfor the best pharmaceutical practices in design and analysis of studies con-ducted at various stages of drug research and development (Chow, 1997a).The purpose of GSP is not only to minimize bias but also to minimizevariability that may occur before, during, and after the conduct of thestudies. More importantly, GSP provides a valid and fair assessment ofthe drug product under study. The concept of GSP can be seen in manyguidelines and guidances that were issued by the FDA at various stagesof drug research and development. These guidelines and guidances includegood laboratory practice (GLP), good clinical practice (GCP), current manufacturing practice (cGMP), and good regulatory practice (GRP). 6ther example of GSP is the guideline on Statistical Principles in ClinicalTrials recently issued by the International Conference on Harmonization(ICH, 1998b). As a result, GSP can not only provide accuracy and reliabil-ity of the results derived from the studies, but also assure the vMidity andintegrity of the studies.

    The implementation of GSP in pharmaceutical research and develop-ment is a team project that requires mutual communication, confidence,respect, and cooperation among statisticians, pharmaceutical scientists inthe related areas, and regulatory agents. The implementation of GSP in-volves some key factors that have an impact on the success of GSP. Thesefactors include (i) regulatory requirements for statistics, (ii) the dissemi-nation of the concept of statistics, (iii) appropriate use of statistics, (iv)effective communication and flexibility, and (v) statistical training. Thesefactors are briefly described below.

    In the pharmaceutical development and approval process, regulatory re-quirements for statistics are the key to the implementation of GSP. Theynot only enforce the use of statistics but also establish standards for sta-tistical evaluation of the drug products under investigation. An unbiasedstatistical evaluation helps pharmaceutical scientists and regulatory agentsin determining (i) whether the drug product has the claimed effectivenessand safety for the intended disease, and (ii) whether the drug productpossesses good drug characteristics such as the proper identity, strength,quality, purity, and stability.

    In addition to regulatory requirements, it is always helpful to dissemi-nate the concept of statistical principles described above whenever possible.It is important for pharmaceutical scientists and regulatory agents to rec-ognize that (i) a valid statistical inference is necessary to provide a fairassessment with certain assurance regarding the uncertainty of the drugproduct under investigation, (ii) an invMid design and analysis may resultin a misleading or wrong conclusion about the drug product, and (iii)

    Copyright 2002 by Marcel Dekker, Inc. All Rights Reserved.

  • 1.4. Good Statistics Practice 19

    larger sample size is often required to increase statistical power and preci-sion of the studies. The dissemination of the concept of statistics is criticalto establish the pharmaceutical scientists and regulatory agents brief instatistics for scientific excellence.

    One of the commonly encountered problems in drug research and devel-opment is the misuse or sometimes the abuse of Statistics in some studies.The misuse or abuse of statistics is critical, which may result in either hay-ing the right question with the wrong answer or having the right answer forthe wrong question. For example, for a given study, suppose that a rightset of hypotheses (the right question) is established to reflect the study

    objective. A misused statistical test may provide a misleading or wronganswer to the right question. On the other hand, in many clinical trials,point hypotheses for equality (the wrong question) are often wrongly usedfor establishment of equivalence. In this case, we have right answer (forequality) for the wrong question. As a result, it is recommended that ap-propriate statistical methods be chosen to reflect the design which shouldbe able to address the scientific or medical questions regarding the intendedstudy objectives for implementation of GSP.

    Communication and flexibility are important factors to the success ofGSP. Inefficient communication between statisticians and pharmaceuticalscientists or regulatory agents may result in a misunderstanding of theintended study objectives and consequently an invalid design and/or in-appropriate statistical methods. Thus, effective communications amongstatisticians, pharmaceutical scientists and regulatory agents is essentialfor the implementation of GSP. In addition, in many studies, the assump-tion of a statistical design or model may not be met due to the nature ofthe drug product under investigation, experimental environment, and/orother causes related/unrelated to the studies. In this case, the traditionalapproach of doing everything by the book does not help. In practice, sinceconcerns from a pharmaceutical scientist or the regulatory agent may trans-late into a constraint for a valid statistical design and appropriate statisticalanalysis, it is suggested that a flexible and yet innovative solution be de-veloped under the constraints for the implementation of GSP.

    Since regulatory requirements for the drug development and approvalprocess vary from drug to drug and country to country, various designsand/or statistical methods are often required for a valid assessment of adrug product. Therefore, it is suggested that statistical continued/advancededucation and training programs be routinely held for both statisticians andnonstatisticians including pharmaceutical scientists and regulatory agents.The purpose of such a continued/advanced education and/or training pro-gram is threefold. First, it enhances communications within the statis-tical community. Statisticians can certainly benefit from such a trainingand/or educational program by acquiring more practical experience and

    Copyright 2002 by Marcel Dekker, Inc. All Rights Reserved.

  • 2O Chapter 1. Introduction

    knowledge. In addition, it provides the opportunity to share/exchange in-formation, ideas and/or concepts regarding drug development between pro-fessional societies. Finally, it identifies critical practical and/or regulatoryissues that are commonly encountered in the drug development and regu-latory approval process. A panel discussion from different disciplines mayresult in some consensus to resolve the issues

    , which helps in establishing

    standards of statistical principles for implementation of GSP.

    1.5 Aim of the Book and Outline of PracticalIssues

    1.5.1 The Aim and Scope of the Book

    As indicated earlier, drug development is a lengthy and costly process. Thepurpose of this lengthy and costly process is not only to ensure the safetyand efficacy of the drug under investigation before approval, but also toassure the drug possesses some good drug characteristics such as identity,strength, quality, purity, and stability after approval. Statistics plays an im-por.tant role in the process of drug research and development. In the processof drug development, misusing, abusing, or not using statistics often resultsin misleading results/conclusions and inflating the false positive/negativerates. Appropriate use of statistics provides a fair and unbiased assessmentof the drug under investigation with a certain degree of accuracy and re-liability. The concept of good statistics practice in design, analysis, andinterpretation of studies conducted during the drug development process isthe key to the success of drug development.

    This book is intended to provide a well-balanced summarization of cur-rent and emerging practical issues and the corresponding statistical method-ologies in various stages of drug research and development. Our emphasis ison recent development in regulatory requirement and statistical methodol-ogy for these practical issues. It is our goal to fill the gap between pharma-Ceutical and statistical disciplines and to provide a comprehensive referencebook for pharmaceutical scientists and biostatisticians in the area of drugresearch and development.

    The scope of this book covers practical issues from nonclinical, pre-clinical, and clinical areas. Previous sections of this chapter provide anintroduction to the drug development process and regulatory requirementfor approving a drug product and an overview of good pharmaceutical prac-tices and good statistics practice. Nonclinical and preclinical applicationsare presented in Chapter 2 through Chapter 5. Practical issues that arecommonly seen in clinical development are discussed in Chapter 6 through

    Copyright 2002 by Marcel Dekker, Inc. All Rights Reserved.

  • 1.5. Aim of the Book and Outline of Practical Issues 21

    Chapter 12.The following is an outline of the practical issues covered in this book

    and what each of Chapters 2-12 covers.

    1.5.2 Pharmaceutical Validation

    As indicated by Bergum and Utter (2000), pharmaceutical validationstarted in the early 1970s with assay verification. However, most of theFDAs attention was directed toward the validation of sterile processes ofinjectable products. In the early 1980s, the FDA began to focus on the vali-dation of nonsterile processes such as solid dosage, semisolid dosage, liquids,suspensions, and aerosols. Basically, pharmaceutical validation includes thevalidation of laboratory instruments, such as gas chromatography (GC),and high-performance liquid chromatography (HPLC), or analytical meth-ods developed based on these instruments and manufacturing processes forspecific compounds. The cGMP requires that the sponsors establish thereliability of test results through the appropriate validation of the test re-sults at appropriate intervals (21 CFR 210 and 211). More specifically,the cGMP requires that the accuracy and reliability of the test results beestablished and validated. The purpose of analytical method validation isto ensure that the assay result meets the proper standards of accuracy andreliability, while the purpose of the manufacturing process is to ensure thatthe rnanufacturing process does what it purports to do (Chow, 1997b).

    The USP/NF defines the validation of analytical methods as the processby which it is established, in laboratory studies, that performance charac-teristics of the methods meet the requirements for the intended analyticalapplication (USP/NF, 2000). The analytical application could be a drugpotency assay for potency and stability studies, immunoassay for the invitro activity of an antibody or antigen, or a biological assay for the invivo activity. The performance characteristics include accuracy, precision,selectivity (or specificity), linearity, range, limit of detection, limit of quan-titation, and ruggedness which are useful measures for the assessment ofthe accuracy and reliability of the assay results. In practice, the selectionof a model and/or weights for the standard curve in calibration is criticalfor assurance of the accuracy and reliability of the assay results. Statisticalmethods for evaluation of the analytical method based on each performancecharacteristic under the selected model with appropriate weight are neces-sarily developed and justified in compliance with regulatory requirementsfor good validation practice.

    The FDA defines process validation as establishing documented evi-dence which provides a high degree of assurance that a specific process willconsistently produce a product meeting its predetermined specifications and

    Copyright 2002 by Marcel Dekker, Inc. All Rights Reserved.

  • 22 Chapter 1. Introduction

    quality characteristics (FDA, 1987a). A manufacturing process is necessar-ily validated to ensure that it does what it purports to do. A valid processassures that the final product has a high probability of meeting the stan-dards for identity, strength, quality, purity, stability, and reproducibilityof the drug product. A manufacturing process is a continuous process in-volving a number of critical stages. A manufacturing process is consideredvalidated if at least three batches (or lots) pass the required USP/NF tests.A batch is considered to pass the USP/NF tests if each critical stage andthe final product meet the USP/NF specifications for identity, strength,quality, purity, stability, and reproducibility. USP/NF tests are referredto as tests for potency, content uniformity (weight variation), disintegra-tion, dissolution, and stability, which are usually conducted according tothe testing plan, sampling plan, and acceptance criteria as specified in theUSP/NF. In practice, it is of interest to establish in-house specificationlimits for each USP/NF tests at each critical stage of the manufacturingprocess, so that if the test results meet the in-house specification limits,there is a high probability that future batches (lots) will also meet theUSP/NF specifications prior to the expiration dating period.

    More details regarding regulatory requirements and statistical issues forselection of an appropriate standard curve in calibration for assay develop-ment and validation are given in Chapter 2. Also included in Chapter 2are the concept of in-process controls and process validation and statisticalmethods for establishment of in-house specification limits.

    1.5.3 Dissolution Testing

    For oral solid dosage forms of drug products, dissolution testing is usuallyconducted to assess the rate and extent of drug release. The purpose ofdissolution testing is multifold. First, it is to ensure that a certain amountof the drug product will be released at a specific time point after adminis-tration in order for the drug to be efficiently delivered to the site of actionfor optimal therapeutic effect. Second, it is to ensure that the dissolution ofthe drug product meets the acceptance limits for quality assurance beforethe drug product is released to the marketplace. Finally, it is to monitorwhether the dissolution of the drug product meets the product specificationlimits prior to the expiration dating period (or shelf-life) of the drug prod-uct. For a valid and fair assessment of the dissolution of the drug product,specific sampling plan, acceptance criteria, and testing procedure are nec-essarily conducted in order to meet regulatory requirements for accuracyand reliability (USP/NF, 2000). The USP/NF suggests that a three-stagesampling plan be adopted. At each stage, a set of criteria must be metin order to pass the test. A set of in-hou~se specification limits is usuallyconsidered to ensure that there is a high probability of passing the USP/NF

    Copyright 2002 by Marcel Dekker, Inc. All Rights Reserved.

  • 1.5. Aim of the Book and Outline of Practical Issues 23

    dissolution test if the test results meet the in-house specification limits.Recently, the in vitro dissolution testing has often been considered a

    surrogate for in vivo bioequivalence testing, which in turn serves as a sur-rogate for clinical outcomes. When two drug products are bioequivalentto each other, it is assumed that they reach the same therapeutic effect orthat they are therapeutically equivalent. Under the assumption that thereis a correlation between the in vitro dissolution testing and the in vivobioequivalence testing, the FDA requires that dissolution profiles betweenthe two .drug products be compared, in addition to the usual USP test fordissolution. The FDA recommends a similarity factor, which is known asthe f2 similarity factor, be evaluated to determine whether two drug prod-ucts have similar dissolution profiles. The use of the f2 similarity factorhas received much criticism since it was introduced by the FDA.

    In practice, it is of interest to evaluate the probability of passing dis-solution testing according to the sampling plan and acceptance criteria asspecified by the USP/NF. The information is useful for the construct ofin-house specification limits for the quality control and assurance of futurebatches of the drug product. In addition, it is of particular interest to studythe statistical properties of the f2 similarity factor as suggested by the FDAfor dissolution profile comparison.

    Chapter 3 provides an extensive discussion regarding the evaluation ofthe probability of passing the USP/NF dissolution test and statistical as-sessment of similarity between dissolution profiles, including a review of thef2 similarity factor and a discussion of some recently developed methods.

    1.5.4 Stability Analysis

    For each drug product on the market, the FDA requires that an expira-tion dating period (or shelf-life) be indicated on the immediate containerof the drug product. The expiration dating period (or shelf-life) is definedas the interval at which the drug characteristics remain unchanged. Asindicated in the FDA Stability Guideline (FDA, 1987b) and in the stabil-ity guideline published by the International Conference on Harmonization(ICH, 1993), the shelf-life can be determined as the time interval at whichthe lower product specification intersects the 95% confidence lower boundof the average degradation curve of the drug product. In practice, stabilitystudies are usually conducted to characterize the degradation curve of thedrug product under appropriate storage conditions. The FDA and ICHstability guidelines require that at least three batches (lots) and prefer-ably more batches be tested for the establishment of a single shelf-life forthe drug product. It is suggested that stability testing be performed at athree-month interval for the first year, a six-month interval for the second

    Copyright 2002 by Marcel Dekker, Inc. All Rights Reserved.

  • 24 Chapter 1. Introduction

    year, and yearly after that. Stability test results can be combined for theestablishment of a single shelf-life if there is no batch-to-batch variation. Apreliminary test for batch similarity should be performed at the 25% levelof significance before the data can be combined for analysis.

    In recent years, the study of the effectiveness and safety of combinationdrug products, which may or may not have been approved by regulatoryagencies, has attracted much attention. The relative proportions of indi-vidual drug products in the combination drug product and the potentialdrug-to-drug interaction (over time) may have an impact on the stabilityof the combined drug product. A typical approach is to consider the mini-mum of the shelf-lives observed from the individual drug products based onthe percent of label claim. This method, however, is not only too conser-vative to be of practical interest, but it also lacks statistical justification.Therefore, how to establish the shelf-life of a combination drug producthas become an interesting scientific question. This question also applies toChinese herbal medicines, which often contain a number of active ingredi-ents with different ratios. In practice, it is, therefore, of interest to explorestatistical methods for the shelf-life estimation of drug products with mul-tiple components. Other statistical issues related to stability include thedevelopment of statistical methods for shelf-life estimation of frozen drugproducts, practical issues and considerations in stability design and anal-ysis (such as the use of matrixing) and bracketing designs and shelf-lifeestimation with discrete responses.

    Chapter 4 provides a comprehensive review of statistical designs andmethods for stability analysis. In addition, statistical methods and recentdevelopment for two phase shelf-life estimation of frozen drug products,stability analysis with discrete responses, and shelf-life estimation of drugproducts with multiple components are discussed.

    1.5.5 Bioavailability and Bioequivalence

    When a brand-name drug is going off patent, the sponsors can file an ab-breviated new drug application (ANDA) for generic approval. For approvalof generic copies of a brand-name drug, the FDA requires that a bioequiv-alence trial be conducted to provide evidence of bioequivalence in drugabsorption (FDA, 1992; 2000a). The FDA indicates that an approvedgeneric copy of a brand-name drug can serve as the substitute of the brand-name drug. However, the FDA does not indicate that two approved genericcopies of the same brand-name drugs can be used interchangeably. As moregeneric drugs become available, the safety of drug interchangeability amonggeneric drugs of the same brand-name drug is of great concern to consumersand the regulatory agencies as well.

    Copyright 2002 by Marcel Dekker, Inc. All Rights Reserved.

  • 1.5. Aim of the Book and Outline of Practical Issues 25

    The concept of drug interchangeability can be divided into prescribabil-ity and switchability. Prescribability is referred to as a physicians choicefor prescribing an appropriate drug between the brand-name drug and itsgeneric copies, while switchability is the switc.h from a brand-name drugor its generic copies to its generic copies within the same patient. TheFDA suggests that the population bioequivalence be assessed to addressprescribability. For switchability, the concept of individual bioequivalenceis recommended. In practice, it is of interest to explore the statistical prop-erties of the bioequivalence criteria for both population bioequivalence andindividual bioequivalence, as proposed by the FDA.

    In a recent FDA guidance on population bioequivalence and individ-ual bioequivalence, the FDA recommends the method proposed by Hyslop,Hsuan, and Holder (2000) for assessment of population bioequivalence andindividual bioequivalence (FDA, 2001). In addition, the FDA recommendsthat a two-sequence, four-period (2 x 4) crossover design be used and thata two-sequence, three-period (2 x 3) crossover design may be used as alternative to the 2 x 4 crossover design if necessary. Although a detailedstatistical procedure for assessment of individual bioequivalence under a2 x 4 crossover design is provided in the FDA draft guidance, little or noinformation regarding statistical procedures for (i) assessment of individualbioequivalence under a 2 x 3 crossover design and (ii) assessment of popu-lation bioequivalence under either a 2 x 2, 2 x 3, or 2 x 4 crossover designis given.

    Chapter 5 provides details of recent development on criteria and sta-tistical methods for assessment of population bioequivalence and individ-ual bioequivalence under various crossover designs. In addition, statisticalmethods for assessment of in vitro bioequivalence are also discussed.

    1.5.6 Randomization and Blinding

    The ultimate goal of most clinical trials is to demonstrate the safety andefficacy of the study drug products. A typical approach is to first showthat there is a significant difference between the study drug product withthe control (e.g., a placebo or an active control agent) with some statis-tical assurance. Power for detection of a clinically meaningful differenceis then obtained to determine the chance of correctly detecting the differ-ence when such a difference truly exists. Statistical inference or assuranceon the uncertainties regarding the drug products under investigation canonly be obtained under a probability structure of the uncertainties. Theprobability structure cannot be established without randomization. As aresult, randomization plays an important role in clinical trials. If there is norandomization, then there is no probability structure, and hence, there isno statistical inference on which the uncertainties can be drawn. Random-

    Copyright 2002 by Marcel Dekker, Inc. All Rights Reserved.

  • 26 Chapter 1. Introduction

    ization not only generates comparable groups of patients who constituterepresentative samples from the intended patient population, but also en-able valid statistical tests for clinical evaluation of the study drug product.

    In clinical trials, bias often occurs due to the knowledge of the identityof treatment. To avoid such a bias, blinding is commonly employed toblock the indentity of treatments. In practice, it is of interest to determinethe integrity of blinding based on the probability of correctly guessing thetreatment codes. When the integrity of blinding is questionable, adjustmentshould be made in statistical inference on the efficacy of the study drug.

    In Chapter 6, the concept of randomization, randomization models andmethods, and blinding are introduced, followed by discussions of practi-cal issues related to randomization and blinding, such as the selection ofthe number of study centers in a multicenter trial, the effect of mixed-uprandomization schedule, assessment of integrity of blinding, and inferenceunder breached blindness.

    1.5.7 Substantial Evidence in Clinical Development

    For the approval of a new drug product, the FDA requires that substantialevidence regarding the safety and efficacy of the drug product be providedthrough the conduct of at least two adequate and well-controlled clinicaltrials. The purpose for having at least two adequate and well-controlledclinical trials is not only to ensure that the clinical results observed from thetwo clinical trials are reproducible, but also to provide valuable informationregarding generalizability of the results to a similar but different patientpopulation.

    In recent years, regulatory agencies are constantly challenged for scien-tific justification of the requirement that at least two adequate and well-controlled clinical trials are necessary for generalizability and reproducibil-ity. In some cases, the variability associated with a powered clinical trialmay be relatively small, and/or the observed p-value is relatively small.These facts are often used to argue against the requirement of at least twoadequate and well-controlled trials. Note that, under certain circumstance,the FDA Modernization Act of 1997 includes a provision to allow datafrom one adequate and well-controlled clinical trial investigation and con-firmatory evidence to establish effectiveness for risk/benefit assessment ofdrug and biological candidates for approval. Thus, if one can show a highprobability of observing a significant result in future studies given that asignificant result has been observed in a clinical trial that is similar to futurestudies, then the observed result from one clinical trial may be sufficient tofulfill the regulatory requirement.

    The concept of reproducibility probability and generalizability probabil-

    Copyright 2002 by Marcel Dekker, Inc. All Rights Reserved.

  • 1.5. Aim of the Book and Outline of Practical Issues 27

    ity of a positive clinical result is introduced in Chapter 7. Several statisticalmethods for evaluation of these probabilities based on observed clinical re-sults are proposed. Applications such as bridging studies are also included.

    1.5.8 Therapeutic Equivalence and Noninferiority

    In drug research and development, therapeutic equivalence and noninferi-ority trials have become increasingly popular. Therapeutic equivalence/noninferiority trials usually involve the comparison of a test drug with acontrol. The assessment of therapeutic equivalence/noninferiority dependsupon the choice of a clinically meaningful difference which may vary fromindication to indication and from therapy to therapy.

    For the establishment of the safety and efficacy of a new drug prod-uct, the FDA prefers that placebo-control clinical trials be considered ifpossible. However, in many cases, placebo-control clinical trials may notbe feasible, such as with anti-infective agents. In addition, it may not beethical to conduct placebo-control clinical trials for patients with severe orlife-threatening diseases, such a cancer or AIDS. As a result, active con-trol trials are preferred in these situations as alternative clinical trials overplacebo-control clinical trials.

    Basically, the primary objective of an active control trial is either toshow noninferiority or therapeutic equivalence. Without the inclusion of aplacebo, the active control trial could lead to the conclusion that the testdrug and the active control agent are either both effective or both ineffec-tive. In practice, it is then of particular interest to develop an appropriatestatistical methodology for providing direct evidence of the safety and effi-cacy of the study drug provided that the safety and efficacy information ofthe active control agent as compared to a placebo is available.

    Chapter 8 introduces the concepts of therapeutic equivalence and nonin-feriority tests in clinical research, including the selection of controls, intervalhypotheses for equivalence/noninferiority, and equivalence lirnit or nonin-feriority margin. Two commonly used statistical methods for assessment oftherapeutic equivalence/noninferiority, the two one-sided tests procedureand the confidence interval approach, are introduced and some related sta-tistical issues are discussed. Also included in this chapter, are discussions onactive control trials and active control equivalence trials in clinical develop-ment. Statistical methods for determination of direct evidence of efficacyof a test drug product relative to a placebo in an active control trial arestudied.

    Copyright 2002 by Marcel Dekker, Inc. All Rights Reserved.

  • 28 Chapter 1. Introduction

    1.5.9 Analysis of Incomplete Data

    In clinical trials, dropouts and missing values may occur due to medicalevaluations, administrative considerations, or other reasons unrelated tothe conduct of the trials. As a result of dropouts or missing values, theobserved clinical data are incomplete. When dropouts or missing valuesare not related to the main response variable used to evaluate treatmenteffects, i.e., a dropout or missing value is at random, statistical inferencecan be made by conditioning on the observed incomplete data. Chapter 9studies the effect of the rate of missing values on the statistical power ofthe tests for treatment effects. The result is useful for the determination ofsample size in planning a clinical trial to account for possible missing valuesin order to achieve the desired power. In some cases, information providedby auxiliary variables can be used to improve the statistical power, whichis also discussed in Chapter 9.

    When dropouts or missing values are related to the main response vari-able used to evaluate treatment effects, the evaluation of treatment effectsbetween patients who stay and patients who drop out is necessary to providea fair and unbiased assessment of the treatment effect. As indicated in theICH guideline, the primary analysis of clinical trials should be the one basedon the intention-to-treat population. The analysis of this kind is referredto as the intention-to-treat analysis. For efficacy evaluation, the intention-to-treat population is usually defined as all randomized patients who haveat least one post-treatment assessment regardless of noncompliance anddropouts. For safety assessment, the intention-to-treat population includesall randomized patients who take any amount of study medications. Whendropouts occur during the conduct of clinical trials with repeated measure-ments, the approach using the concept of last observation carried forward(LOCF) is recommended. The validity of this approach has been challengedby many researchers. Chapter 9 provides a comprehensive evaluation of theLOCF analysis. Some other methods providing valid statistical inferencein analysis of last observation and analysis of longitudinal data, which areuseful when the LOCF analysis fails, are also introduced in Chapter 9.

    1.5.10 Meta-Analysis

    Meta-analysis is a systematic reviewing strate