Abstract
Based on the assumption that molecular mechanisms involved in cancerogenesis are characterized by groups of coordinately expressed genes, we developed and validated a novel method for analyzing transcriptional data called Correlated Gene Set Analysis (CGSA). Using 50 extracted gene sets we identified three different profiles of tumors in a cohort of 364 Diffuse large B-cell (DLBCL) and related mature aggressive B-cell lymphomas other than Burkitt lymphoma. The first profile had high level of expression of genes related to proliferation whereas the second profile exhibited a stromal and immune response phenotype. These two profiles were characterized by a large scale gene activation affecting genes which were recently shown to be epigenetically regulated, and which were enriched in oxidative phosphorylation, energy metabolism and nucleoside biosynthesis. The third and novel profile showed only low global gene activation similar to that found in normal B cells but not cell lines. Our study indicates novel levels of complexity of DLBCL with low or high large scale gene activation related to metabolism and biosynthesis and, within the group of highly activated DLBCLs, differential behavior leading to either a proliferative or a stromal and immune response phenotype.