Meta-Analysis of Gene Expression Profiling Datasets To Uncover Biological Pathways and Candidate Biomarkers Associated with Progression in DCIS
Judith N Robens, Stuart J Schnitt, Andrew H Beck. Beth Israel Deaconess Medical Center, Boston
Background: The frequency with which DCIS is encountered in clinical practice has dramatically increased in recent years. Factors associated with progression of DCIS to invasive cancer remain poorly understood. Prior genome-wide studies to identify markers associated with risk of progression in DCIS have been limited by small sample size. We undertook a meta-analysis of breast cancer progression-associated gene expression profiling data sets to discover biological pathways and candidate biomarkers of progression in DCIS.
Design: We searched the NCBI's Gene Expression Omnibus to identify gene expression profiling datasets containing samples from DCIS and/or normal breast and/or invasive ductal carcinoma (IDC) with at least 5 samples in each category. We identified a total of 5 data-sets. Within each dataset, we performed 2-class Significance Analysis of Microarrays (SAM) to identify genes differentially expressed between normal and DCIS, normal and IDC, and DCIS and IDC at a false discovery rate (FDR) ≤ 10%. We performed a total of 13 SAM analyses over 12,755 genes.
Results: Overall, the largest changes in gene expression in both the epithelium and stroma occur during progression from normal to DCIS with fewer altered genes during progression from DCIS to IDC. When analyzing all comparisons together, we identified a core set of 44 genes differentially expressed in ≥ 60% of the analyses. This gene list is highly enriched for genes related to the extra-cellular matrix (Bonferroni p = 3.5e-7) and genes regulating angiogenesis (Bonferroni p = 0.002). Top genes identified in our analysis as most consistently showing altered expression during progression (including progression from DCIS to IDC) include: CALD1 (caldesmon), POSTN (periostin), LHFP (lipoma HMGIC [high mobility group protein isoform I-C] fusion partner, LMO2 (LIM only domain 2), COL1A2 (collagen alpha 2-1), INHBA (inhibin beta-A), COL10A1 (collagen, type X, alpha-1), KRT14 (keratin 14).
Conclusions: In this meta-analysis of gene expression profiling data sets, we identified biological pathways and a core set of genes consistently showing altered expression during breast cancer progression. The most dramatic changes in gene expression occur in the transition from normal to DCIS. The top genes associated with the transition from DCIS to IDC are stromal-related rather than epithelial-related. These data provide insight into the biology of breast cancer progression, and core genes identified in our analysis represent candidate prognostic biomarkers for DCIS.
Tuesday, March 20, 2012 9:30 AM
Poster Session III # 36, Tuesday Morning