The ErbB-driven autocrine growth pathway has been implicated in the development and progression of most common human epithelial malignancies; its blockade is therefore a promising therapeutic strategy, and several candidate drugs are currently undergoing clinical trials. Paradoxically, little is known of the expression pattern of these 4 genes in human tumors, and the clinical significance of the 2 most recently discovered ERBB genes, ERBB3 and ERBB4, is unclear. We used a real-time quantitative RT-PCR assay to quantify ERBB family mRNA copy numbers in a large series of breast tumors from patients with known long-term outcome. ERBB gene expression varied widely, by more than 2 orders of magnitude for ERBB1 and ERBB3, more than 3 orders for ERBB2 and more than 4 orders for ERBB4. We found a positive correlation between ERBB3 and ERBB4 mRNA levels, and a negative correlation between the expression of these 2 latter genes and that of ERBB1. Compared to normal breast tissue, ERBB1 was underexpressed (82.3% of tumors), ERBB2 (16.9%) and ERBB3 (46.2%) were overexpressed and ERBB4 was both underexpressed (24.6%) and overexpressed (29.2%). Links were also found between ERBB status on the one hand and Scarff-Bloom-Richardson (SBR) histopathological grade and estrogen receptor alpha (ERa) status on the other hand. Relapse-free survival (RFS) was shorter among patients with ERBB3-overexpressing tumors (p=0.0092) and longer among those with ERBB4-underexpressing tumors (p=0.0085) relative to patients with normal expression of the respective genes; in contrast, RFS was not significantly influenced by ERBB1 or ERBB2 mRNA status. Only ERBB4 status retained prognostic significance in Cox multivariate regression analysis (p=0.015). Our results point to the involvement of several ErbB-specific ligands (amphiregulin and neuregulin 1) and enzymes or adaptor molecules (PI3K, Src, Shc and Grb7) in the ErbB pathway dysregulation associated with breast cancer. These findings reveal a complex expression pattern of ERBB gene family members in breast tumors and suggest that it is this pattern of expression, rather than the expression of individual family members, that should be taken into account when evaluating antitumoral drugs designed to target these receptors.
Copyright 2003 Wiley-Liss, Inc.