Regional complexity in breast tumors reveals patterns of prognostic value

Genomic complexity in breast tumors has been associated with aggressive disease and less favorable outcome. Recently, we developed an algorithm to calculate complexity scores per chromosome arm in order to sub-classify breast tumors with respect to progression paths and prognosis. We have further developed this method to calculate probe-focused complexity scores per sample, enabling identification of regions with recurrent complexities and grouping of tumors according to genome-wide complexity patterns. Probe-focused complexity scores were calculated based on data derived from aCGH analyses of 394 invasive ductal breast carcinomas. These continuous complexity scores (CCI) were used to identify regions with recurrent high level complexity, and the regional complexity scores were correlated to clinicopathological variables and survival data. A total of 25 recurrent regions with high level complexity were identified. Regional complexities on chromosome arms 8p, 11q and 17q were each associated with clinical parameters associated with aggressive disease. Complexity patterns were different between tumors of different gene expression subtypes. Multivariate Cox analysis revealed that regional complexity on 17q21.32-q21.33 was significantly associated with shorter survival, independent of established clinical variables. 394 invasive ductal breast carcinomas were analyzed using whole genome CGH arrays from Agilent. In addition 20 normal breast biopsies were included as controls, a total of 414 samples. All samples were hybridized to the arrays together with a commercial female reference sample from Promega.