Clinical and molecular analysis of long-term HER2 positive metastatic breast cancer survivors.

ackground: Several multigene tests have been developed in Metastatic Breast Cancer (MBC) disease, in order to identify predictive factors correlated to clinical outcomes. The purpose of this study is to investigate the clinico-pathological and molecular characteristics that could differentiate long term responders from patients experiencing early progression during anti-HER2 treatments. Methods: A total of 34 HER2 positive MBC patients were included: 20 patients with a time to progression longer than 3 years in Long Responders group (LR) and 14 patients with a progression disease within one year of anti-HER2 therapy in Poor Responders group (PR). Tumor characteristics and treatment information were collected. The expression of 770 genes and 13 molecular pathways were evaluated using Nanostring PanCancer pathway panel performed on BC formalin-fixed paraffin-embedded tissues from diagnostic core biopsy or surgical resection. Results: Baseline patients and tumor characteristics were similar between the two groups, although PR patients were more likely to have CNS spread and more metastatic burden of disease compared to LR (29% vs. 0, p = 0.02 and 57% vs. 20%, p = 0.04, respectively). Gene expression analysis identified 30 genes with significantly different expression in the two cohorts; five of these were driver genes (BRCA1, PDGFRA, AR, PHF6 and MSH2). The majority of these genes were over-expressed, mainly in LR patients, and encoded growth factors, pro- or anti-inflammatory interleukins and DNA repair factors. Only four genes were down-regulated, all in PR group (TNFSF10, CACNG1, IL20RB and BRCA1). Most of these genes were involved in MAPK and PI3K pathways (9 and 8, respectively). MAPK pathway was differently expressed between LR and PR (p = 0.05). Even if not statistically significant but clinically relevant, PI3K was the only pathway overexpressed in PR patients (median expression LR: 1441 ± 485 vs 1759 ± 762 in PR group; p= 0.1). Conclusions: Whole genome expression analysis comparing LR vs. PR identified a group of genes that may predict more favourable long-term outcomes. Up-regulation of MAPK and down-regulation of PI3K pathways could be a positive predictive factors. Further clinical implications are warranted.