Gene expression profiling can predict pathological complete response (pCR) to anthracycline-based therapy in estrogen-receptor (ER) negative breast cancer (BC) patients.

Desmedt C, André F, Azambuja E, Haibe-Kains B, Larsimont D, D’Hondt V, Di Leo A, Piccart MJ, Pusztai L and Sotiriou C

Background: Breast cancers show variable sensitivity to anthracycline (A)-based therapy. Here we aimed to identify gene expression profiles associated with pCR to this treatment. As it has repeatedly and consistently been shown that ER is the most dominant factor influencing the molecular composition of breast cancer, defining different types of BC disease and because we wanted to eliminate the confounding effect of indirect ovarian suppression in ER+ BC, we focused in this study on ER-negative patients only.
 
Methods: We analyzed Affymetrix gene expression profiles generated from 132 ER- pre-treatment samples, constituting the largest series of ER- preoperatively A-treated BC (n=132/35 pCR). Sixty-two samples derived from the prospective multicentric TOP trial (epirubicin single-agent), 41 from Institut G. Roussy (retrospective selection/ FEC) and 27 from MD Anderson (prospective study/ FAC).
 
Results: A student t-test analysis on the combined population of A-treated pts was performed identifying 102 genes that were significantly associated with pCR (p<.01). These genes were mainly involved in cell death, DNA replication and recombination, molecular transport, cell cycle and morphology. Interestingly, 14 of these genes were located on the topoIIα amplicon. Of interest, none of these 14 genes seem to carry any prognostic value in untreated ER- pts (N=161). When we considered gene expression indices for specific A-targets such as topoIIα and helicase, we found that both were associated with pCR. However, subgroup analysis revealed that topoIIα index was predictive in ERBB2+ but not in ERBB2- subgroup. None of the genes from the adriamycin predictor (Potti et al.) or the p53 signature (Miller et al.) were significantly associated with pCR.
 
Conclusions: This study suggests that a group of genes associated with topoIIα can identify ER-negative BC pts likely to respond to A-based therapy. These promising results are currently being validated in a larger series.