Large-scale copy number analysis reveals variations in genes not previously associated with malignant pleural mesothelioma

// Marieke Hylebos 1, 2 , Guy Van Camp 1, 2 , Geert Vandeweyer 1 , Erik Fransen 1, 3 , Matthias Beyens 1, 2 , Robin Cornelissen 4 , Arvid Suls 1 , Patrick Pauwels 2, 5 , Jan P. van Meerbeeck 2, 6 and Ken Op de Beeck 1, 2 1 Center of Medical Genetics, University of Antwerp and Antwerp University Hospital, 2650 Antwerp, Belgium 2 Center for Oncological Research, University of Antwerp, 2610 Antwerp, Belgium 3 StatUa Center for Statistics, University of Antwerp, 2610 Antwerp, Belgium 4 Department of Pulmonary Medicine, Erasmus Medical Center Cancer Institute, 3015 Rotterdam, The Netherlands 5 Laboratory of Pathology, Antwerp University Hospital, 2650 Antwerp, Belgium 6 Thoracic Oncology, Antwerp University Hospital, 2650 Antwerp, Belgium Correspondence to: Marieke Hylebos, email: Marieke.Hylebos@uantwerpen.be Keywords: malignant pleural mesothelioma; copy number variations; low-pass whole genome sequencing; the cancer genome atlas; cancer census genes Received: August 01, 2017      Accepted: November 15, 2017      Published: December 01, 2017 ABSTRACT Malignant pleural mesothelioma (MPM) is an aggressive tumor that is often causally associated with asbestos exposure. Comparative genomic hybridization techniques and arrays demonstrated a complex set of copy number variations (CNVs) in the MPM-genome. These techniques however have a limited resolution, throughput and flexibility compared to next-generation sequencing platforms. In this study, the presence of CNVs in the MPM-genome was investigated using an MPM-cohort ( N = 85) for which genomic microarray data are available through ‘The Cancer Genome Atlas’ (TCGA). To validate these results, the genomes of MPMs and matched normal samples ( N = 21) were analyzed using low-pass whole genome sequencing on an ‘Illumina HiSeq’ platform. CNVs were detected using in-house developed analysis pipelines and frequencies of copy number loss and gain were calculated. In both datasets, losses on chromosomes 1, 3, 4, 6, 9, 13 and 22 and gains on chromosomes 1, 5, 7 and 17 were found in at least 25% and 15% of MPMs, respectively. Besides the well-known MPM-associated genes, CDKN2A, NF2 and BAP1 , other interesting cancer-associated genes were listed as frequently involved in a copy number loss (e.g. EP300, SETD2 and PBRM1 ). Moreover, four cancer-associated genes showed a high frequency of copy number gain in both datasets (i.e. TERT , FCGR2B , CD79B and PRKAR1A ). A statistically significant association between overall survival and the presence of copy number loss in the CDKN2A -containing region was observed in the TCGA-set. In conclusion, recurrent CNVs were detected in both datasets, occurring in regions harboring known MPM-associated genes and genes not previously linked to MPM.