Generation and application of a colorectal cancer microarray library to identify a novel prognostic biomarker, adipose differentiation related protein (adipophilin).

508 Background: Substantial volumes of genetic expression data are archived within public gene expression repositories (PGER). Although the data contained is reported according to particular standards, it is not feasible to search the data using clinically relevant search terms (i.e., oncologic outcome). This study aimed to generate an archive of microarray expression data that could be searched using oncologically-relevant terminology, and then to test this facility. Methods: A UL Colorectal Cancer Archive was established based on data derived from the Gene Expression Omnibus (GEO). A software (Rover) was developed to permit clinically relevant searches using terminology such as stage and disease-free survival. As a first test, experiments were identified that compared early and late stage colorectal cancer. From these, consensus profiles were developed and adipocyte differentiation related protein identified as the top-most frequently dysregulated gene. As a second test, the cancer archive was again challenged to identify data sets annotated with outcome data. The association between ADFP and outcome was assessed using a regression-tree (CRT) based approach and Kaplan-Meier estimates. Results: The UL Colorectal cancer archive was constructed so as to permit the novice-user to search for gene expression data associated with particular clinical parameters. Rover was a graphic user interface that permitted these searches. ADFP was identified as dysregulated across the majority of experiments comparing early and late stage colorectal cancer. CRT-analyses identified levels of ADFP above which adverse outcomes were identified. In general, increasing ADFP was independently associated with adverse disease-free outcomes in stage II and III cancer. Conclusions: A novel archive of gene expression data was generated that (a) related to colorectal cancer and (b) could be searched using clinically-relevant terms. Using this archive ADFP was identified as dysregulated between early and late stage colorectal cancer. Increasing ADFP expression levels were associated with adverse disease-free survival in stages II and III colorectal cancer.