Background: Inflammatory breast cancer (IBC) is the most aggressive manifestation of primary epithelial breast cancer and differs from non-IBC both from a clinical and a biological perspective. The high metastatic potential of IBC is highlighted by the low 5-year survival rate of 50%. In the past, we demonstrated that IBC is characterized by a specific molecular profile, partly mediated by increased NFkB-activation. In the current study we aim to identify additional molecular alterations responsible for the altered gene expression profile.
Methods: We profiled 46 IBC samples and 56 non-stage matched nIBC samples using the Affymetrix HGU133plus2 platform. Using a set of recently published gene signatures (Gatza et al, PNAS, 2010) we calculated the activation probability scores for 18 distinct pathways (ER, PR, HER2, P53, P63, EGFR, RAS, SRC, TNFa, INFa, INFg, TGFb, STAT3, E2F1, PI3K, AKT, CTNNB and MYC). Unsupervised cluster analysis was performed on the IBC samples only to identify IBC-specific transcriptional heterogeneity. The silhouette-algorithm was used to test the robustness of clustering. Supervised analysis was performed to compare IBC to nIBC. All P-values were FDR-corrected.
Results: Using only IBC samples, unsupervised hierarchical cluster analysis identified four robust sample clusters (Average silhouette width = 0.345; P<0.0001). Interestingly, the primary bifurcation of the resulting dendrogram is determined by INFa-, INFg-and TNFa-signalling. In addition to ER-, PR-, HER2- and P53-signalling, the transcriptional heterogeneity in IBC is also determined by AKT-, RAS-, STAT3- and TGFb-signalling. When comparing IBC to nIBC we identified 12 pathways that are differentially regulated (P<0.05). Out of these 12 pathways, 4 (CTNNB, PI3K, E2F1 and P63) did not reveal any IBC-specific heterogeneity and all of these were hyperactivated in IBC. Interestingly, the PI3K-, E2F1- and P63-pathways form a tightly correlated signaling nodule (Spearman correlation coefficients > 0.65, P<0.001). Conclusions: In this study we demonstrated that IBC is transcriptionally heterogeneous and this heterogeneity is mainly determined INFa-, INFg-and TNFa-signalling. Also, we identified four pathways (CTNNB, PI3K, E2F1 and P63) which are differentially regulated in IBC as compared to nIBC but which do not contribute to the establishment of transcriptional heterogeneity within IBC. Therefore, we reason that these pathways can be added to the molecular repertoire of IBC.
Citation Information: Cancer Res 2010;70(24 Suppl):Abstract nr P6-04-07.