Abstract PO4-05-14: Gene expression in metastatic breast cancer – patterns and prognostic potential

Abstract

Abstract Introduction Metastatic breast cancer (mBC) is the leading cause of breast cancer mortality. However, mBC has a considerable variance in prognosis and course of disease, reflecting a substantial biological heterogeneity. Despite this, there is a lack of prognostic biomarkers validated for mBC. New prognostic markers and improved understanding of the molecular variants of mBC could aid in prognostication and personalizing treatments, and lead to the discovery of novel drug targets. The primary objective of this study was to explore the prognostic value of gene expression (GEX) in the distant metastasis (DM) of mBC. Secondly, we aimed to identify molecular subtypes of mBC, and describe their biological niches. Methods We used a cohort of mBC patients enrolled in a prospective trial (NCT01322893) that includes tissue from primary tumor (PT) and matched lymph node metastasis (LNM) and/or DM (1). The inclusion criteria were diagnosed mBC with a life expectancy of >2 months, ECOG score of 0-2, and an age of >18 years. RNA was extracted from formalin-fixed, paraffin-embedded tumor tissue. GEX data were acquired for n=123 PT, n=71 LNM, and n=74 DM using the NanoString Breast Cancer 360 assay, which comprises 758 genes and >30 GEX signatures. The relationships between GEX and the endpoints overall survival (OS) and progression-free survival (PFS) were assessed using cox proportional hazards regression with GEX as a continuous score. Multivariable cox regressions were adjusted for PAM50 of the DM, number of metastatic sites, visceral metastasis, ECOG, and age at mBC diagnosis. Visualization by heatmap and k-means clustering were used to identify GEX patterns. When clustering single genes, we included only the 100 genes with the highest standard deviation. Results In relation to PFS, we found prognostically unfavorable roles of DM GEX of the signatures for Treg cells (HR=1.3, p=0.02), cytotoxicity (HR=1.3, p=0.03), and p53 (HR=1.3, p=0.02). Further, our data indicate prognostically favorable value of DM GEX of ESR1 (HR=0.77, p=0.02), PGR (HR=0.65, p=0.002), FOXA1 (HR=0.76, p=0.02), and AR (HR=0.68, p=0.0006). The prognostic value of AR remained in multivariable analysis. In relation to OS the first 2 years after mBC prognosis, we found prognostically favorable value of DM GEX of B7H3 (HR=0.65, p=0.04), TGFB (HR=0.53, p=0.02) and the signatures for stroma (HR=0.7, p=0.05) and claudin-low (HR=0.61, p=0.02). Finally, our data suggests that the previously published PAM50MET panel performs well as a prognostic indicator, especially when considered on a continuous scale, for PFS (HR=1.3, p=0.0023) and OS (HR=1.3, p=0.003) (2). Based on the 100 most highly variable genes, we identified five stable tumor clusters with distinct GEX profiles, where some strongly associated with metastatic site. Using the same approach on the GEX signatures, we established four GEX profiles able to outperform PAM50 in identifying tumors with poor OS in this material. Conclusion In this study, we demonstrate prognostic value for GEX-based panels in DM samples. First, we find GEX signatures related to hormone responsiveness to be favorable, and genetic instability to be unfavorable for mBC prognosis. We also demonstrate AR to be an independent marker for better PFS. Further, we confirm the prognostic value of PAM50MET, a previously published panel with the advantage of including only PAM50 genes and clinical parameters. Finally, we identify new GEX profiles related to metastatic site and outcome. These results illuminate the biological differences between mBC in relation to outcome and metastatic site. Better understanding of the mBC GEX subtypes may open new venues for tailored treatment.