A comprehensive modular map of Diffuse Large B-cell Lymphoma

Diffuse large B-cell lymphoma (DLBCL) is characterised by pronounced genetic and biological heterogeneity. Several partially overlapping classification systems exist - developed from mutation, rearrangement or gene expression data. We apply a customised network analysis to nearly five thousand DLBCL cases to identify and quantify modules indicative of tumour biology. We demonstrate that network-level patterns of gene co-expression can enhance the separation of DLBCL cases. This allows the resolution of communities of related cases which correlate with genetic mutation and rearrangement status, supporting and extending existing concepts of disease biology and delivering insight into relationships between differentiation state, genetic subtypes, rearrangement status and response to therapeutic intervention. We demonstrate how the resulting fine-grained resolution of expression states is critical to accurately identify potential responses to treatment. Significance statementWe demonstrate how exploiting data integration and network analysis of gene expression can enhance the segregation of diffuse large B-cell lymphoma, resolving pattens of disease biology and demonstrating how the resolution of heterogeneity can enhance the understanding of treatment response.