臨床試驗 Study design (I) 病人族群、研究設計 ( 一 ) 2015-3-4 簡國龍老師 [email protected]

臨床試驗 Study design (I) 病人族群、研究設計 ( 一 ) 2015-3-4 簡國龍老師 [email protected] 【 本 著 作 除 另 有 註 明 外 ， 採 取 創用 CC 「姓名標示－非商業性－相同方式分享」台灣 3.0 版授權釋出】 1

In a randomized trial, the investigator (a) selects a sample from population, (b) measures baseline variables, (c) randomizes the participants, (d) applies interventions (one should be a blinded placebo, if possible), (e) follows up the cohort, (f) measures outcome variables (blindly, if possible)and analyzes the results. 2

Experimental vs. observational Strength: Causality inference Weakness: Expensive Time-consuming Address a narrow clinical question Expose patients to potential harm Not every question is amenable to clinical trial design 3

Clinical trial Protocol Selecting the participants Measuring baseline variables Randomizing Applying the interventions Follow-up and adherence to the protocol Measuring the outcome Analyzing the results Monitoring clinical trials Alternatives to the randomized blinded trial 4

Principles in clinical trial design Randomly assigned Double blinding Placebo-controlled Guideline for procedures Endpoints Primary Secondary 5

Outline Phases of clinical trial Issues in clinical trial design Selection of patients, controls Randomization Blinding Types of designs 6

Stages in Testing New Therapies Preclinical Studies in Cell Culture and Animals Phase I Unblended, uncontrolled studies in a few volunteers to test safety Phase II Relatively small randomized, controlled, blinded trials to test tolerability and different intensity or dose of the intervention on surrogate outcomes Relatively large randomized, controlled, blinded trials to test the effect of the therapy on clinical outcomes Phase III Phase IV Large trials or observational studies conducted after the therapy has been approved by the FDA to assess the rate of serious side effects and evaluate additional therapeutic uses 7

Phases of clinical trial Phase I: initial introduction of new drug to humans Phase II: first controlled Determine metabolic and pharmacologic activities and side effects in human & PK and PD effects for Phase II Effectiveness based on clinical endpoints, dosing ranges IIA: dosing evaluation IIB: effectiveness Phase III: expanded Effectiveness and safety, adequate basis for labeling IIIB: after submission before approval 8

Phases of clinical trial Phase IV: after approval for marketing Adverse reactions, morbidity or mortality, children population, marked-oriented comparison studies against competitor products Phase V: new indications and/or new formulations or dosage forms 9

Issues in study design Selection of patients Selection of controls Statistical considerations Randomization Blinding 10

Principles of validity in comparison Validity of comparison between the treatment group and control group Comparability of population Comparability of information Are the baseline characteristics of the two groups comparable? Are the information obtained from the two groups comparable? Comparability of effects (e.g. placebo, sham procedures) Are the non-therapeutic effects of the treatments11

Selection of patients Representative samples Define the intended patient population Select homogeneous as possible Eligibility criteria Patient characteristics and disease severity Inclusion Exclusion Patient selection process Single vs. multicenter 12

Which site better? Individual investigator’s qualification and experience for disease Feasibility of the investigator’s site for conducting the proposal trial Dedication, education, training, and experience of the personnel at the site Availability of certain equipments Geographic location 13

Within the site center, which patients? Initial guess of how many patients will meet the eligibility criteria Screening based on diagnostic criteria Patient’s disease change Concurrent diseases/medications Psychological factors Fear of toxicity Informed consent 14

Selection of controls Well-controlled, comparative Concurrent control Patient characteristics between treatment groups are homogeneous Placebo concurrent control: placebo effect Dose-comparison concurrent control: phase II Active treatment concurrent control: ethical No treatment concurrent control: available effectiveness & a very short period of time & negligible placebo effect Historical control High and predictable mortality or self-evident effect 15

Table 3.3.3 Conditions for the Ethical Inclusion of a Placebo Concurrent Control (from Chou & Liu, 1998 ) 1. No Standard treatment exists. 2. Standard treatment is ineffective or unproved to be effective. 3. Standard treatment is appropriate for the particular clinical trials. 4. The placebo has been reported to be relatively effective in treating the disease or condition. 5. The disease is mild and lack of treatment is not considered to be medically important. 6. The placebo is given as an add-on treatment to an already existing regimen that is not sufficient to treat patients. 7. Allowing concomitant medicine is one measure of efficacy in these clinical trials. 8. The disease process is characterized by frequent spontaneous exacerbations and remission (e.g., peptic ulcer). 9. “Escape clauses” or points are designed into the protocol. Source: Spiller (1991). 16

Selection of participants Entry criteria defining a target population Inclusion criteria Exclusion criteria Optimize Rate of primary outcome Expected effectiveness of the active treatment Generalizability of findings Easy of recruitment Compliance with treatment and follow-up High risk group Strength Weakness 17

Statistical considerations Primary and secondary response variables Criteria for efficacy and safety assessment Sample size estimation Interim analysis and data monitoring Statistical and clinical inference 18

Endpoints: Response variables: validated Efficacy assessment: intention to treat, subgroup analysis (?) Parameters to be measured, timing and frequency, normal values for laboratory, definition of test abnormalities Safety analysis: at least one dose of treatment Incidence of adverse effect 19

Sample size estimation What design to be used? What hypotheses are to be tested? What statistic is to be performed? 20

Issues in interim analysis and data monitoring Protection of the overall type I error rate informal confirmatory clinical trials designed to establish efficacy Safeguarding of the blinding of a study Use of interim analyses for administrative or planning purposes to generate hypotheses for further studies or to assess safety 21

Who will have access to the randomization code? How the blinding will be broken? Who will have access to the interim results? Whether ongoing patients will be included in the analysis? 22

Statistical and clinical inference Statistical inference as only a part of induction process Internal validity External validity: from sample to population Age range, childbearing women Bioequivalence trials: limited validity Different investigators at different study centers in different countries Meta analysis(?) 23

Single site versus multi-sites Treatment-by-center interaction in more centers A central laboratory for appropriate and consistent measures 24

Treatment duration Typical phases: Placebo run-in phase Active treatment period Follow-up or maintenance phase 25

Compliance & missing Patient compliance: 20-80% Medication event monitor system Poor compliance (dose) vs. poor adherence (time) Missing value and dropout Assumption of missing mechanism 26

Randomization and blinding Avoid subjective assignment 1960s gastric freezing Avoid bias Random selection of a representative sample from a targeted patient population Random assignment of patients in order to study the medicines Blinding: open label, single, double and triple blinding 27

Randomizing, randomization Rationale Do a good job of random assignment Basis for testing the statistical significance of differences between these groups in the measured outcome Mal-distribution effect due to chance Table of random numbers Set up a separate randomization facility Consider special randomization techniques Blocked randomization Stratified blocked randomization 28

Randomization models Population model Invoked population model Randomization model As if a formal sampling procedure performed Permutation test: nonparametric, Wilcoxin rank sum test Stratification Variability between-strata and within-strata Matching, cross-over 29

Methods of randomization (1) Simple, complete randomization Completely binomial design Balance? Random allocation Equal allocation of sample size, restricted Randomly select the N/2 out of N without replacement Proc PLAN Accidental bias 30

Methods of randomization (2) Permuted-block randomization Imbalance in size and covariate Time-heterogeneous population Block size of 4, six possible permutations SAS procedure PLAN Stratified randomization Adaptive randomization Based on previous patient assignment Treatment, covariates, clinical response 31

Table 4.3.1 Unacceptable Methods of Assignment of Patients to Treatment 1. Assignment of patients to treatment according to the order of enrollment, (every other patient is assigned to one group) 2. Assignment of patients of treatment according to patient’s initial. 3. Assignment of patients of treatment according to patient’s birthday 4. Assignment of patients according to the dates of enrollment 32

Response adaptive randomization Randomized play-the-winner (RPW) rule Suitable conditions: There is a single outcome or hypothesis of interest Outcomes are ascertainable in a short period of time The study has important public health consequences, but the diseases are not life-threatening The study has an adequate sample size and the composition of the sample is not likely to change over time The participants in the study have the resources to logistically implement the randomization procedure Limitations: complex outcomes, multiple visits 33

Implementation of randomization Standard operating procedure, SOP Appropriate method of randomization and related logistic issues for implementation Quality assurance (QA) procedures Laboratory evaluation Central administrative telephone-based assignment system 34

Generalization of controlled randomized trials Internally vs. externally Efficacy or safety Population vs. individual Average efficacy versus. Variability of efficacy Individual efficacy: N-of-1 randomized trial Theophylline efficacy, N pairs Patient-by-treatment interaction Limitation: short time, hard endpoint or irreversible Guyatt G, et al. Determining optimal therapy—Randomized trials in individuals patients. New Engl J Med 1986;314:889-92 35

Blinding Various groups of individuals involved with the trial are withheld from the knowledge of treatments Patient, study center or investigator (two teams), sponsor Masking 4 types 36

Types of blinding Open-label Single-blind All of health care personnel Triple-blind Patient unaware of his/her assignment Double-blind Ethical: phase I dose-escalating studies in oncology, pre-marketing, post-marketing surveillance Highest degree for the validity Maintain the blindness throughout the entire course Matched placebo, identical container Multiple-placebo, double dummy To ask both to guess 37

Design types of clinical trial Parallel designs Parallel-group designs Matched pairs parallel designs Cross-over designs Titration designs Enriched designs 38

Parallel designs Simple, universally accepted, applicable to acute conditions Matched Run-in period Establish comparability 39

R Patients U N I N R A N D O MI Z A TI O N Test Control A Control B Figure 5.2.1 Parallel-group design 40

Crossover designs p x q: p sequences of treatment in q different time periods Benefits: Within-patient comparison between treatment Remove inter-patient variability Best unbiased estimates Washout period: carryover effects 41

Cross-over design (vs. parallel design) treatment treatment Group I A (wash-out period ) B Group II B (wash-out period ) A (Three or more groups are possible) Advantages of cross-over design Minimize confounding, -Smaller number of patients required Disadvantages of cross-over design Potential carry-over effects (depends on the pharmocokinetics of drugs) Underlying disease process has to be stable (i.e. not applicable to all diseases) 42

Suitable situations for cross-over designs Objective measures and interpretable data Chronic and relatively stable disease Prophylactic drugs with short half-life Short treatment periods Baseline and washout periods are feasible Adequate number of patients to allow carryover effect and expected dropouts 43

Patients RA N D O MI ZA TI O N Period I Sequence 1 Test Sequence 2 Control II W AS H O UT Control Test Figure 5.3.1 Standard two-sequence, two-period crossover design. 44

Higher-order crossover designs Handle carryover effects and intra-subject variability Extended-period design Replicated design: doubled standard 2x2 Williams designs More than two treatments Variance-balanced designs Balanced incomplete block designs 45

Patients R A N D O M IZ A TI O N Period II I Sequence 1 Test Sequence 2 Control W AS H O UT III W AS H O UT Control Test Control Test Figure 5.3.2 Two-sequence dual crossover design. 46

Table 5.3.1 Higher-Order Crossover Designs I. Balaam’s design AA BB AB BA II. Two-sequence dual design ABB BAA III. Doubled (replicated) design AABB BBAA IV. Four-sequence design AABB BBAA ABBA BAAB 47

Table 5.3.2 Williams Designs I. Williams’s design with three treatments ACB BAC CBA BCA CAB ABC II. Williams’s design with four treatments ADBC BACD CBDA DCAB 48

Table 5.3.3 Balance Incomplete Block Designs I. Four treatments with two periods AB BC CD AC BD DB CA AD DC CB BA II. Four treatments with three periods BCD CDA DAB ABC 49

Examples of crossover design Study Purpose Sample Size Chan et al. (1993) Effects on lipids 12 Nagi and Yudkin (1993) Effects on lipids and risk factors for cardiovascular disease 27 Elkeles (1991) Effects of lipids 35 Duration (Period 1 Washout Period 2) 4 wk 0 wk 4 wk 12 wk 2 wk 12 wk 3 mo 6 wk 3 mo Primary Endpoint Metabolic and Hemodynamic index Insulin resistance Glycemic control Cardiovascular risk Serum lipids Lipoproteins Blood glucose Glycosylated hemoglobin 50

Titration designs Phase I safety and tolerance studies Maximum tolerable dose, MTD Pre-specified proportion: 1/3 Up-and-down design 51

Enrichment designs To identify the patients is likely to be beneficial, specific population Two phases Enrichment phase: open-label with a titration design to class patients into groups Randomize and double-blind phase Examples: best dose Screening process further to a small selective group 52

Further topics in clinical trial designs Multi-center trial Active control trial Primary objectives Combination trial Treatment-by-center interaction Fixed-combination prescription drugs Multi-level factorial design Global superiority of combination drug Method of response surface Equivalence trial 53

(a) No Interaction treatment A treatment B (b) Quantitative Interaction 1 Center 2 treatment A treatment A treatment B Center treatment B Center (c) Quantitative Interaction treatment A treatment B Figure 6.2.1 Treatment-by-center interaction. Center 54

Figure 6.2.3 Mean seated diastolic blood pressures versus study site. 55

Summary: randomized, double-blind, controlled trials Treatment group and control group Randomization Concurrent controls, Self controls, Historical controls, and Literature controls Assignment of treatment through a random process Blinding/ masked Both the patient and outcomes assessor do not know the treatment assignment 56

57

58

版權聲明 頁碼 作品 Design and Analysis of Clinical Trials: Concepts and Methodologies 3rd. Edition, p.111 / Shein-Chung Chow & Jen-Pei Liu Wiley, October 2013http://as.wiley.com/WileyCDA/WileyTitle/productCd-0470887656.html 本作品依據著作權法第 46 、 52 、 65 條合理使用。 2 In a randomized trial, the investigator (a) selects a Design and Analysis of Clinical Trials: Concepts and Methodologies 3rd. Edition, p.15-16 / Shein-Chung Chow & Jen-Pei Liu Wiley, October 2013http://as.wiley.com/WileyCDA/WileyTitle/productCd-0470887656.html 本作品依據著作權法第 46 、 52 、 65 條合理使用。 Design and Analysis of Clinical Trials: Concepts and Methodologies 3rd. Edition, p.90 / Shein-Chung Chow & Jen-Pei Liu Wiley, October 2013http://as.wiley.com/WileyCDA/WileyTitle/productCd-0470887656.html 本作品依據著作權法第 46 、 52 、 65 條合理使用。 7 11-15 來源 / 作者 本作品轉載自 Microsoft Office 2010 PowerPoint 設計主題範本 -Pixel ，依據 Microsoft 服務合約及著作權法第 46 、 52 、 65 條合理使用。 1-63 2 版權圖示 Validity of comparison between the treatment group and control group .Wellcontrolled, comparative Design and Analysis of Clinical Trials: Concepts and Methodologies 3rd. Edition, p.94-97 / Shein-Chung Chow & Jen-Pei Liu Wiley, October 2013http://as.wiley.com/WileyCDA/WileyTitle/productCd-0470887656.html 本作品依據著作權法第 46 、 52 、 65 條合理使用。 59

版權聲明 頁碼 作品 32 33 來源 / 作者 Design and Analysis of Clinical Trials: Concepts and Methodologies 3rd. Edition, p.103 / Shein-Chung Chow & Jen-Pei Liu Wiley, October 2013http://as.wiley.com/WileyCDA/WileyTitle/productCd-0470887656.html 本作品依據著作權法第 46 、 52 、 65 條合理使用。 16 30-31 版權圖示 Simple, complete randomization Permuted-block randomization Design and Analysis of Clinical Trials: Concepts and Methodologies 3rd. Edition, p.124126 / Shein-Chung Chow & Jen-Pei Liu Wiley, October 2013http://as.wiley.com/WileyCDA/WileyTitle/productCd-0470887656.html 本作品依據著作權法第 46 、 52 、 65 條合理使用。 Design and Analysis of Clinical Trials: Concepts and Methodologies 3rd. Edition, p.124 / Shein-Chung Chow & Jen-Pei Liu Wiley, October 2013http://as.wiley.com/WileyCDA/WileyTitle/productCd-0470887656.html 本作品依據著作權法第 46 、 52 、 65 條合理使用。 Design and Analysis of Clinical Trials: Concepts and Methodologies 3rd. Edition, p.140 / Shein-Chung Chow & Jen-Pei Liu Wiley, October 2013http://as.wiley.com/WileyCDA/WileyTitle/productCd-0470887656.html 本作品依據著作權法第 46 、 52 、 65 條合理使用。 60

版權聲明 頁碼 40 44 46 47 作品 版權圖示 來源 / 作者 Design and Analysis of Clinical Trials: Concepts and Methodologies 3rd. Edition, p.167 / Shein-Chung Chow & Jen-Pei Liu Wiley, October 2013http://as.wiley.com/WileyCDA/WileyTitle/productCd-0470887656.html 本作品依據著作權法第 46 、 52 、 65 條合理使用。 Design and Analysis of Clinical Trials: Concepts and Methodologies 3rd. Edition, p.179 / Shein-Chung Chow & Jen-Pei Liu Wiley, October 2013http://as.wiley.com/WileyCDA/WileyTitle/productCd-0470887656.html 本作品依據著作權法第 46 、 52 、 65 條合理使用。 Design and Analysis of Clinical Trials: Concepts and Methodologies 3rd. Edition, p.180 / Shein-Chung Chow & Jen-Pei Liu Wiley, October 2013http://as.wiley.com/WileyCDA/WileyTitle/productCd-0470887656.html 本作品依據著作權法第 46 、 52 、 65 條合理使用。 Design and Analysis of Clinical Trials: Concepts and Methodologies 3rd. Edition, p.180 / Shein-Chung Chow & Jen-Pei Liu Wiley, October 2013http://as.wiley.com/WileyCDA/WileyTitle/productCd-0470887656.html 本作品依據著作權法第 46 、 52 、 65 條合理使用。 61

版權聲明 頁碼 48 49 50 54 作品 版權圖示 來源 / 作者 Design and Analysis of Clinical Trials: Concepts and Methodologies 3rd. Edition, p.182 / Shein-Chung Chow & Jen-Pei Liu Wiley, October 2013http://as.wiley.com/WileyCDA/WileyTitle/productCd-0470887656.html 本作品依據著作權法第 46 、 52 、 65 條合理使用。 Design and Analysis of Clinical Trials: Concepts and Methodologies 3rd. Edition p.183/ Shein-Chung Chow & Jen-Pei Liu Wiley, October 2013http://as.wiley.com/WileyCDA/WileyTitle/productCd-0470887656.html 本作品依據著作權法第 46 、 52 、 65 條合理使用。 Design and Analysis of Clinical Trials: Concepts and Methodologies 3rd. Edition, p.185 / Shein-Chung Chow & Jen-Pei Liu Wiley, October 2013http://as.wiley.com/WileyCDA/WileyTitle/productCd-0470887656.html 本作品依據著作權法第 46 、 52 、 65 條合理使用。 Design and Analysis of Clinical Trials: Concepts and Methodologies 3rd. Edition, p.240 / Shein-Chung Chow & Jen-Pei Liu Wiley, October 2013http://as.wiley.com/WileyCDA/WileyTitle/productCd-0470887656.html 本作品依據著作權法第 46 、 52 、 65 條合理使用。 62

版權聲明 頁碼 55 作品 版權圖示 來源 / 作者 Design and Analysis of Clinical Trials: Concepts and Methodologies 3rd. Edition, p.243 / Shein-Chung Chow & Jen-Pei Liu Wiley, October 2013http://as.wiley.com/WileyCDA/WileyTitle/productCd-0470887656.html 本作品依據著作權法第 46 、 52 、 65 條合理使用。 57 Controlled Clinical Trials 23 卷 The method of minimization for allocation to clinical trials. a review Scott NW1, McPherson GC, Ramsay CR, Campbell MK. Control Clin Trials. (2002), p.662-674 本作品依據著作權法第 46 、 52 、 65 條合理使用。 58 Controlled Clinical Trials 23 卷 The method of minimization for allocation to clinical trials. a review Scott NW1, McPherson GC, Ramsay CR, Campbell MK. Control Clin Trials.(2002), p.662-674 本作品依據著作權法第 46 、 52 、 65 條合理使用。 63