Search for new loci and low-frequency variants influencing glioma risk by exome-array analysis

• Published in: European journal of human genetics : EJHG Vol. 24; no. 5; pp. 717 - 724
• Main Authors: Kinnersley, Ben, Kamatani, Yoichiro, Labussière, Marianne, Wang, Yufei, Galan, Pilar, Mokhtari, Karima, Delattre, Jean-Yves, Gousias, Konstantinos, Schramm, Johannes, Schoemaker, Minouk J, Swerdlow, Anthony, Fleming, Sarah J, Herms, Stefan, Heilmann, Stefanie, Nöthen, Markus M, Simon, Matthias, Sanson, Marc, Lathrop, Mark, Houlston, Richard S
• Format: Journal Article
• Language: English
• Published: England Nature Publishing Group 01.05.2016
• Subjects: Algorithms; Amino Acyl-tRNA Synthetases - genetics; Brain cancer; Brain Neoplasms - genetics; BRCA2 protein; BRCA2 Protein - genetics; Breast cancer; Case-Control Studies; Disease; Epidemiology; Exome; Female; Genes; Genetic Loci; Genetics; Genomes; Genotyping Techniques - methods; Glioblastoma; Glioma; Glioma - genetics; Health risk assessment; Humans; Isocitrate Dehydrogenase - genetics; Licenses; Lung cancer; Male; Medical prognosis; Medical research; Middle Aged; Mutation; Polymorphism, Single Nucleotide; Protein arrays; Proteins; Statistical analysis; Tumors; United Kingdom > UK; France; Germany
• ISSN: 1018-4813; 1476-5438; 1476-5438
• DOI: 10.1038/ejhg.2015.170

To identify protein-altering variants (PAVs) for glioma, we analysed Illumina HumanExome BeadChip exome-array data on 1882 glioma cases and 8079 controls from three independent European populations. In addition to single-variant tests we incorporated information on the predicted functional consequences of PAVs and analysed sets of genes with a higher likelihood of having a role in glioma on the basis of the profile of somatic mutations documented by large-scale sequencing initiatives. Globally there was a strong relationship between effect size and PAVs predicted to be damaging (P=2.29 × 10(-49)); however, these variants which are most likely to impact on risk, are rare (MAF<5%). Although no single variant showed an association which was statistically significant at the genome-wide threshold a number represented promising associations - BRCA2:c.9976A>T, p.(Lys3326Ter), which has been shown to influence breast and lung cancer risk (odds ratio (OR)=2.3, P=4.00 × 10(-4) for glioblastoma (GBM)) and IDH2:c.782G>A, p.(Arg261His) (OR=3.21, P=7.67 × 10(-3), for non-GBM). Additionally, gene burden tests revealed a statistically significant association for HARS2 and risk of GBM (P=2.20 × 10(-6)). Genome scans of low-frequency PAVs represent a complementary strategy to identify disease-causing variants compared with scans based on tagSNPs. Strategies to lessen the multiple testing burden by restricting analysis to PAVs with higher priors affords an opportunity to maximise study power.