Meta-analysis of genome-wide association studies in 125,000 women identifies fourteen new breast cancer susceptibility loci. K. Michailidou1, P. Hall2, P. Kraft3, D. F. Easton1, Breast Cancer Association Consortium, DRIVE 1) Centre for Cancer Genetic Epidemiology, Department of Public Health and Primary Care, University of Cambridge, Cambridge, UK; 2) Department of Medical Epidemiology and Biostatistics, Karolinska Institutet, Stockholm, Sweden; 3) Department of Epidemiology, Harvard School of Public Health, Boston, Massachusetts, USA.
Genome Wide Association Studies (GWAS) and large scale replication studies have successfully identified 77 common variants associated with breast cancer. These variants explain ~15% of the familial risk of the disease. In an effort to identify new susceptibility loci, we performed a meta-analysis of 11 GWAS consisting of ~16K breast cancer cases and ~19K controls, and ~47K cases and ~43K controls from 41 studies genotyped on a 200K custom array (iCOGS). Analyses were restricted to women of European ancestry. Each study was imputed separately using the March 2012 release of the 1000 Genomes reference panel. Summary per-allele odds ratio estimates and standard errors were obtained by inverse variance fixed effect meta-analysis.
Approximately 11M SNPs were reliably imputed, with imputation r2>0.3 and minor allele frequency (MAF)>0.005, in at least one of the studies. We identified more than 300 variants in 28 regions reaching p<5x10-8 not previously reported as being associated with breast cancer. In 13 of the regions, the most significant SNP has an imputation r2>0.8 (in iCOGS), and supporting evidence was provided by genotyped SNPs in the same region. Three further regions lie within 1Mb of the top hit of regions previously found to be associated with the disease; these may reflect confounding with previous signals or secondary associations. In 3 of the regions the signal was provided by only one study and a further region from only three studies; in each case the MAF of the most significant SNP was <0.03. Finally, for 8 regions there was only one SNP in the region that reached the GWAS significance threshold, and/or the best SNP had imputation r2<0.8. Further evaluation will be needed to determine whether these reflect true associations or imputation artefacts.
You may contact the first author (during and after the meeting) at