Efficient Study Designs and Semiparametric Inference Methods for Developing Genomic Biomarkers in Cancer Clinical Resear
In the development of genomic biomarkers and molecular diagnostics, clinical studies using high-throughput assays such as DNA microarrays generally require enormous costs and efforts. Several efficient study designs for reducing the costs of such expensiv
- PDF / 365,430 Bytes
- 20 Pages / 439.37 x 666.142 pts Page_size
- 77 Downloads / 189 Views
Abstract In the development of genomic biomarkers and molecular diagnostics, clinical studies using high-throughput assays such as DNA microarrays generally require enormous costs and efforts. Several efficient study designs for reducing the costs of such expensive measurements have been developed, mainly in the field of epidemiology. Under these efficient designs, expensive measurements are collected only on selected subsamples based on adequate response-selective sampling schemes, and total measurement costs are effectively reduced. In this study, we discuss the application of these effective designs to genomic analyses in cancer clinical studies, and provide relevant statistical methods such as gene selection (e.g., multiple testing based on the false discovery rate). Efficient semiparametric inference methods using auxiliary clinical information are also discussed.
Keywords Nested case-control study Case-cohort study Two-phase designs Genomic biomarker Semiparametric inference Weighted estimating equation Calibration estimator
1 Introduction The establishment of high-throughput technologies such as DNA microarrays has enabled the genome-wide investigation of cancer tumor samples to characterize diseases at a molecular level, namely, that of genes. Such genomic studies are potentially useful for elucidating disease biology and aggressiveness, identifying new therapeutic targets, and developing new molecular diagnostics for optimized
H. Noma (&) Department of Data Science, The Institute of Statistical Mathematics, 10-3 Midori-cho, Tachikawa, Tokyo 190-8562, Japan e-mail: [email protected] © Springer Nature Singapore Pte Ltd. 2017 S. Matsui and J. Crowley (eds.), Frontiers of Biostatistical Methods and Applications in Clinical Oncology, DOI 10.1007/978-981-10-0126-0_23
381
382
H. Noma
medicine for individual patients [1–3]. One of the primary objectives of these genome-wide studies is the screening of differentially expressed genes among different phenotypes, such as clinical subtypes and prognostic classes of disease, for further investigation. Because of the large scale of these data, false findings are a serious issue, and thus many researchers are concerned about controlling false positives in the framework of multiple testing, particularly controlling the false discovery rate (FDR) [4, 5]. However, distinguishing relevant genes from thousands of non-interesting genes with null associations generally requires large sample sizes to achieve sufficient statistical power [2, 3]. Furthermore, these genomic studies usually require enormous financial and other resources to collect and/or process the large-scale measurements involved. In particular, it is still expensive to implement biological experiments using high-throughput assays such as microarrays, and clinical researchers often find it burdensome to plan and conduct such studies. Also, in most previous cancer clinical studies using such high-throughput assays, these expensive experiments have been conducted for all samples in the corresponding coho
Data Loading...
