Unravelling the association between proliferation, immune response, and prognosis in colorectal, gastric, and breast cancers

B172 Recently we reported that high proliferation in colorectal and gastric cancers is associated with improved clinical outcome (in terms of increased disease-free survival, relative to less-proliferative tumours), based on an expression-based gene proliferation signature [1]. The reverse relationship (i.e., high proliferation associated with poor outcome) was observed in multiple microarray data sets from a number of non-gastrointestinal cancers, including breast, glioma and lymphoma, using the same gene-based assessment of proliferation. Earlier work in our laboratory also identified a gene signature for colorectal cancer prognosis prediction in which higher levels of expression of immune response related genes were associated with increased disease free survival [2].
 Here we use newly developed bioinformatic methodology to investigate the biology underlying this somewhat counterintuitive relationship between proliferation and prognosis in colorectal and gastric cancer, in the context of an expression-based assessment of both proliferation level and immune response for each tumor. This approach is based on identifying sets of correlated genes which undergo statistically significant alterations in expression between prognosis, proliferation, and/or immune response classes in colorectal and gastric cancer, and contrasting these relationships with those observed in breast cancer, for which there is both a large amount of high quality publicly available microarray data, as well as a widely accepted association between high proliferation and poor prognosis.
 This approach has revealed that while highly proliferative tumors express immune response related genes in a similar pattern across colorectal, gastric and breast cancers, certain genes appear to be down-regulated in highly proliferative breast cancers (relative to low proliferation tumors) that remain unchanged in colorectal and gastric tumors. Our conclusion is that subtle differences in the host immune response underly the differences in the relationship between proliferation and prognosis in colorectal, gastric and breast cancers.
 1. Lin YH, et al. A proliferation gene signature can predict clinical outcome of multiple cancers. Abstract LB-264, 2007 AACR Annual Meeting, Los Angeles, CA.
 2. Lin, YH, et al. Multiple gene expression classifiers from different array platforms predict poor prognosis of colorectal cancer., Clin Cancer Res, 13, 2 Pt 1, 2007.