9 Bayesian Analysis Methods in Clinical Trials 223
9.1 Bayesian Models . . . . . . . . . . . . . . . . . . . . . . . . . 224
9.1.1 Bayes' Theorem . . . . . . . . . . . . . . . . . . . . . 224
9.1.2 Posterior Distributions for Some Standard Distributions 226
9.1.2.1 Normal Distribution with Known Variance . 226
9.1.2.2 Normal Distribution with Unknown Variance 227
9.1.2.3 Normal Regression . . . . . . . . . . . . . . . 227
9.1.2.4 Binomial Distribution . . . . . . . . . . . . . 228
9.1.2.5 Multinomial Distribution . . . . . . . . . . . 228
9.1.3 Simulation from the Posterior Distribution . . . . . . 228
9.1.3.1 Direct Simulation . . . . . . . . . . . . . . . 229
9.1.3.2 Importance Sampling . . . . . . . . . . . . . 229
9.1.3.3 Gibbs Sampling . . . . . . . . . . . . . . . . 230
9.1.3.4 Metropolis{Hastings Algorithm . . . . . . . . 231
9.2 R Packages in Bayesian Modeling . . . . . . . . . . . . . . . 232
9.2.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . 232
9.2.2 R Packages Using WinBUGS . . . . . . . . . . . . . . . 233
9.2.2.1 R2WinBUGS . . . . . . . . . . . . . . . . . . . 233
9.2.2.2 BRugs . . . . . . . . . . . . . . . . . . . . . . 234
9.2.2.3 rbugs . . . . . . . . . . . . . . . . . . . . . . 234
9.2.2.4 Typical Usage . . . . . . . . . . . . . . . . . 234
9.2.3 MCMCpack . . . . . . . . . . . . . . . . . . . . . . . . . 235
xiv Contents
9.3 MCMC Simulations . . . . . . . . . . . . . . . . . . . . . . . 236
9.3.1 Normal-Normal Model . . . . . . . . . . . . . . . . . . 236
9.3.2 Beta-Binomial Model . . . . . . . . . . . . . . . . . . 238
9.4 Bayesian Data Analysis . . . . . . . . . . . . . . . . . . . . . 242
9.4.1 Blood Pressure Data: Bayesian Linear Regression . . . 243
9.4.2 Binomial Data: Bayesian Logistic Regression . . . . . 246
9.4.3 Count Data: Bayesian Poisson Regression . . . . . . . 250
9.4.4 Comparing Two Treatments . . . . . . . . . . . . . . . 251
9.5 Concluding Remarks . . . . . . . . . . . . . . . . . . . . . . 256
10 Analysis of Bioequivalence Clinical Trials 257
10.1 Data from Bioequivalence Clinical Trials . . . . . . . . . . . 258
10.1.1 Data from Chow and Liu (2009) . . . . . . . . . . . . 258
10.1.2 Bioequivalence Trial on Cimetidine Tablets . . . . . . 258
10.2 Bioequivalence Clinical Trial Endpoints . . . . . . . . . . . . 260
10.3 Statistical Methods to Analyze Bioequivalence . . . . . . . . 262
10.3.1 Decision CIs for Bioequivalence . . . . . . . . . . . . . 262
10.3.2 The Classical Asymmetric Condence Interval . . . . 263
10.3.3 Westlake's Symmetric Condence Interval . . . . . . . 264
10.3.4 Two One-Sided Tests . . . . . . . . . . . . . . . . . . 264
10.3.5 Bayesian Approaches . . . . . . . . . . . . . . . . . . . 265
10.3.6 Individual-Based Bienayme-Tchebyche (BT) Inequality
CI . . . . . . . . . . . . . . . . . . . . . . . . . . . 266
10.3.7 Individual-Based Bootstrap CIs . . . . . . . . . . . . . 267
10.4 Step-by-Step Implementation in R . . . . . . . . . . . . . . . 267
10.4.1 Analyze the Data from Chow and Liu (2009) . . . . . 267
10.4.1.1 Load the Data into R . . . . . . . . . . . . . 267
10.4.1.2 Tests for Carryover Eect . . . . . . . . . . . 269
10.4.1.3 Test for Direct Formulation Eect . . . . . . 271
10.4.1.4 Analysis of Variance . . . . . . . . . . . . . . 273
10.4.1.5 Decision CIs . . . . . . . . . . . . . . . . . . 274
10.4.1.6 Classical Shortest 90% CI . . . . . . . . . . . 274
10.4.1.7 The Westlake Symmetrical CI . . . . . . . . 275
10.4.1.8 Two One-Sided Tests . . . . . . . . . . . . . 276
10.4.1.9 Bayesian Approach . . . . . . . . . . . . . . 276
10.4.1.10 Individual-Based BT CI . . . . . . . . . . . . 276
10.4.1.11 Bootstrap CIs . . . . . . . . . . . . . . . . . 277
10.4.2 Analyze the Data from Cimetidine Trial . . . . . . . . 282
10.4.2.1 Bioavailability Endpoints Calculations . . . . 282
10.4.2.2 ANOVA: Tests for Carryover and Other Effects
. . . . . . . . . . . . . . . . . . . . . . . 286
10.4.2.3 Decision CIs . . . . . . . . . . . . . . . . . . 290
10.4.2.4 Classical Shortest 90% CI . . . . . . . . . . . 290
10.4.2.5 The Westlake Symmetrical CI . . . . . . . . 291
10.4.2.6 Two One-Sided CI . . . . . . . . . . . . . . . 292
Contents xv
10.4.2.7 Bayesian Approach . . . . . . . . . . . . . . 292
10.4.2.8 Individual-Based BT CI . . . . . . . . . . . . 292
10.4.2.9 Bootstrap CIs . . . . . . . . . . . . . . . . . 293
10.5 Concluding Remarks . . . . . . . . . . . . . . . . . . . . . . 297
11 Analysis of Adverse Events in Clinical Trials 299
11.1 Adverse Event Data from a Clinical Trial . . . . . . . . . . . 300
11.2 Statistical Methods . . . . . . . . . . . . . . . . . . . . . . . 302
11.2.1 Condence Interval (CI) Methods . . . . . . . . . . . 302
11.2.1.1 Comparison Using Direct CI Method . . . . 302
11.2.1.2 Comparison Using Indirect CI Methods . . . 303
11.2.1.3 Connection between Direct and Indirect CI
Methods . . . . . . . . . . . . . . . . . . . . 304
11.2.2 Signicance Level Methods (SLM) . . . . . . . . . . . 304
11.2.2.1 SLM Using Normal Approximation . . . . . 304
11.2.2.2 SLM Using Exact Binomial Distribution . . 305
11.2.2.3 SLM Using Resampling from Pooled Samples 305
11.2.2.4 SLM Using Resampling from Pooled AE Rates 306
11.3 Step-by-Step Implementation in R . . . . . . . . . . . . . . . 306
11.3.1 Clinical Trial Data Manipulation . . . . . . . . . . . . 306
11.3.2 R Implementations for CI Methods . . . . . . . . . . . 307
11.3.3 R Implementations for Indirect CI Methods . . . . . . 308
11.3.4 R for Signicant Level Methods . . . . . . . . . . . . . 313
11.3.4.1 R for SLM with Normal Approximation . . . 313
11.3.4.2 R for SLM with Exact Binomial . . . . . . . 314
11.3.4.3 R for SLM Using Sampling{Resampling . . . 316
11.4 Concluding Remarks . . . . . . . . . . . . . . . . . . . . . . 320
12 Analysis of DNA Microarrays in Clinical Trials 321
12.1 DNA Microarray . . . . . . . . . . . . . . . . . . . . . . . . . 322
12.1.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . 322
12.1.2 DNA, RNA, and Genes . . . . . . . . . . . . . . . . . 322
12.1.3 Central Dogma of Molecular Biology . . . . . . . . . . 323
12.1.4 Probes, Probesets, Mismatch, and Perfect Match . . . 324
12.1.5 Microarray and Statistical Analysis . . . . . . . . . . . 324
12.1.6 Software: R/Bioconductor . . . . . . . . . . . . . . . 324
12.2 Breast Cancer Data . . . . . . . . . . . . . . . . . . . . . . . 325
12.2.1 Data Source . . . . . . . . . . . . . . . . . . . . . . . . 326
12.2.2 Low-Level Data Analysis . . . . . . . . . . . . . . . . 328
12.2.2.1 Introduction . . . . . . . . . . . . . . . . . . 328
12.2.2.2 Library affy . . . . . . . . . . . . . . . . . . 329
12.2.2.3 Quality Control . . . . . . . . . . . . . . . . 331
12.2.2.4 Background, Normalization, and Summarization
. . . . . . . . . . . . . . . . . . . . . . . 334
12.2.3 High-Level Analysis . . . . . . . . . . . . . . . . . . . 337
xvi Contents
12.2.3.1 Statistical t-test . . . . . . . . . . . . . . . . 339
12.2.3.2 Model Fitting . . . . . . . . . . . . . . . . . 340
12.2.3.3 Number of Signicantly Expressed Genes . . 345
12.2.4 Functional Analysis of Gene Lists . . . . . . . . . . . . 345
12.3 Concluding Remarks . . . . . . . . . . . . . . . . . . . . . . 346
Bibliography 349
Index 3
|