Robust BRCA1‐like classification of copy number profiles of samples repeated across different datasets and platforms

• DOI: 10.1016/j.molonc.2015.03.002
• Publication Date: 2015-04-05T16:59:44Z
• Authors: Philip C. Schouten; Anita Grigoriadis; Thomas Kuilman; Hasan Mirza; Johnathan A. Watkins; Saskia A. Cooke; Ewald van Dyk; Tesa M. Severson; Oscar M. Rueda; Marlous Hoogstraat; Caroline V.M. Ver...; Rachael Natrajan; Suet-Feung Chin; Esther H. Lips; Janneke Kruizinga; Arno Velds; Marja Nieuwland; Ron M. Kerkhoven; Oscar Krijgsman; Conchita Vens; Daniel Peeper; Petra M. Nederlof; Carlos Caldas; Andrew N. Tutt; Lodewyk F. Wessels; Sabine C. Linn
• Keywords: Comparative genomic hybridization; Molecular Inversion Probe; Bacterial artificial chromosome; Copy number analysis; Concordance

Breast cancers with BRCA1 germline mutation have a characteristic DNA copy number (CN) pattern. We developed test that assigns CN profiles to be ‘BRCA1‐like’ or ‘non‐BRCA1‐like’, which refers resembling BRCA1‐mutated tumor without mutation, respectively. Approximately one third of the BRCA1‐like breast has hypermethylation promoter and an unknown reason for being BRCA1‐like. This classification is indicative patients' response high dose alkylating platinum containing chemotherapy regimens, targets inability deficient cells repair double strand breaks. investigated whether this can reliably obtained next generation sequencing platforms other than bacterial artificial chromosome (BAC) array Comparative Genomic Hybridization (aCGH) on it was originally developed. samples from 230 cancer patients profile had been generated two five platforms, comprising low coverage sequencing, extraction targeted panels (CopywriteR), Affymetrix SNP6.0, 135K/720K oligonucleotide aCGH, Oncoscan FFPE (MIP) technology, 3K BAC 32K aCGH. Pairwise comparison genomic position‐mapped original aCGH platform revealed concordance. For most cases, biological differences between exceeded within sample. observed same across different in over 80% kappa values at least 0.36. Differential could attributed were not strongly associated class. In conclusion, we shown regions define our classifier are robustly measured by profiling technologies, providing possibility retro‐ prospectively investigate wide range platforms.