Comparison of RNA-seq and microarray-based models for clinical endpoint prediction

• Type of Material: Article
• Authors: Wenqian Zhang, BGI-Shenzhen; Ying Yu, Fudan University; Falk Hertwig, University of Cologne; Jean Thierry-Mieg, NIH/NCBI; Wenwei Zhang, BGI-Shenzhen; Danielle Thierry-Mieg, NIH/NCBI; Jian Wang, Emory University; Cesare Furlanello, Fondazione Bruno Kessler; Viswanath Devanarayan, AbbVie, Inc.; Jie Cheng, GlaxoSmithKline; Youping Deng, Rush University; Barbara Hero, University of Cologne; Huixiao Hong, U.S. Food and Drug Administration; Meiwen Jia, Fudan University; Li Li, SAS Institute, Inc.; Simon M. Lin, Marshfield Clinic Research Foundation; Yuri Nikolsky, Thomson Reuters IP & Science; André Oberthuer, University of Cologne; Tao Qing, Fudan University; Zhenqiang Su, U.S. Food and Drug Administration; Ruth Volland, University of Cologne; Charles Wang, Loma Linda University; May D. Wang, Emory University; Junmei Ai, Rush University; Davide Albanese, Fondazione Edmund Mach; Shahab Asgharzadeh, Children’s Hospital Los Angeles; Smadar Avigad, Schneider Children’s Medical Center of Israel; Wenjun Bao, SAS Institute, Inc.; Marina Bessarabova, Thomson Reuters IP & Science; Murray H. Brilliant, Marshfield Clinic Research Foundation; Benedikt Brors, German Cancer Research Center; Marco Chierici, Fondazione Bruno Kessler; Tzu-Ming Chu, SAS Institute, Inc.; Jibin Zhang, BGI-Shenzhen; Richard G. Grundy, University of Nottingham; Min Max He, Marshfield Clinic Research Foundation; Scott Hebbring, Marshfield Clinic Research Foundation; Howard L. Kaufman, Rush University; Samir Lababidi, U.S. Food and Drug Administration; Lee J. Lancashire, Thomson Reuters IP & Science; Yan Li, Rush University; Xin X. Lu, AbbVie, Inc.; Heng Luo, U.S. Food and Drug Administration; Xiwen Ma, Eli Lilly and Company; Baitang Ning, U.S. Food and Drug Administration; Rosa Noguera, University of Valencia; Martin Peifer, University of Cologne; John H. Phan, Emory University; Frederik Roels, University of Cologne; Carolina Rosswog, University Children’s Hospital of Cologne; Susan Shao, SAS Institute, Inc.; Jie Shen, U.S. Food and Drug Administration; Jessica Theissen, University of Cologne; Gian Paolo Tonini, University of Padua; Jo Vandesompele, Ghent University; Po-Yen Wu, Georgia Institute of Technology; Wenzhong Xiao, Harvard University; Joshua Xu, U.S. Food and Drug Administration; Weihong Xu, Stanford University; Jiekun Xuan, U.S. Food and Drug Administration; Yong Yang, Eli Lilly & Company; Zhan Ye, Marshfield Clinic Research Foundation; Zirui Dong, BGI-Shenzhen; Ke K. Zhang, University of North Dakota; Ye Yin, BGI-Shenzhen; Chen Zhao, Fudan University; Yuanting Zheng, Fudan University; Russell D. Wolfinger, SAS Institute, Inc.; Tieliu Shi, East China Normal University; Linda H. Malkas, Beckman Research Institute; Frank Berthold, University of Cologne; Jun Wang, BGI-Shenzhen; Weida Tong, U.S. Food and Drug Administration; Leming Shi, Fudan University; Zhiyu Peng, BGI-Guangzhou; Matthias Fischer, University of Cologne
• Language: English
• Date: 2015-06-25
• Publisher: BioMed Central
• Publication Version: Final Published Version
• Copyright Statement: © 2015 Zhang et al.
• License: Creative Commons Attribution 4.0 International
• Final Published Version (URL): http://dx.doi.org/10.1186/s13059-015-0694-1
• Title of Journal or Parent Work: Genome Biology
• ISSN: 1474-7596
• Volume: 16
• Issue: 1
• Grant/Funding Information: This work was supported by grants from the National High Technology Research and Development Program of China - 863 Program (2015AA020104 to LS and 2012AA02A201 to TS), the National Science Foundation of China (31471239 to LS), the Intramural Research Program of the NIH, National Library of Medicine (USA), the Cologne Center for Molecular Medicine (to MF), the Bundesministerium für Bildung und Forschung (BMBF) through the National Genome Research Network plus (NGFNplus, grant 01GS0895 to MF), the German Cancer Aid (Deutsche Krebshilfe, grant 110122 to MF), the Fördergesellschaft Kinderkrebs-Neuroblastom-Forschung e.V. (to FB and MF), and the National Institutes of Health (U54CA119338, 1RC2CA148265, and R01CA163256) and Georgia Cancer Coalition (to MDW).; Tumor sample collection was supported by the BMBF through the Competence Network Pediatric Oncology and Hematology (KPOH) and by the Italian Neuroblastoma Foundation.
• Keywords: Genetics & Heredity; GENE-EXPRESSION; PROSTATE-CANCER; Biotechnology & Applied Microbiology; Life Sciences & Biomedicine; BREAST-CANCER; SIGNATURE; EXPRESSION-BASED CLASSIFICATION; DIFFERENTIATION; RISK STRATIFICATION; TRANSCRIPTOME; NEUROBLASTOMA PATIENTS; REPRODUCIBILITY; Science & Technology
• Research Categories: Biology, Genetics; Health Sciences, Oncology

Abstract

Background: Gene expression profiling is being widely applied in cancer research to identify biomarkers for clinical endpoint prediction. Since RNA-seq provides a powerful tool for transcriptome-based applications beyond the limitations of microarrays, we sought to systematically evaluate the performance of RNA-seq-based and microarray-based classifiers in this MAQC-III/SEQC study for clinical endpoint prediction using neuroblastoma as a model. Results: We generate gene expression profiles from 498 primary neuroblastomas using both RNA-seq and 44 k microarrays. Characterization of the neuroblastoma transcriptome by RNA-seq reveals that more than 48,000 genes and 200,000 transcripts are being expressed in this malignancy. We also find that RNA-seq provides much more detailed information on specific transcript expression patterns in clinico-genetic neuroblastoma subgroups than microarrays. To systematically compare the power of RNA-seq and microarray-based models in predicting clinical endpoints, we divide the cohort randomly into training and validation sets and develop 360 predictive models on six clinical endpoints of varying predictability. Evaluation of factors potentially affecting model performances reveals that prediction accuracies are most strongly influenced by the nature of the clinical endpoint, whereas technological platforms (RNA-seq vs. microarrays), RNA-seq data analysis pipelines, and feature levels (gene vs. transcript vs. exon-junction level) do not significantly affect performances of the models. Conclusions: We demonstrate that RNA-seq outperforms microarrays in determining the transcriptomic characteristics of cancer, while RNA-seq and microarray-based models perform similarly in clinical endpoint prediction. Our findings may be valuable to guide future studies on the development of gene expression-based predictive models and their implementation in clinical practice.