Transcriptomic analysis of HER2 expression in metastatic breast cancer: Insights from a UAE patient cohort.

Abstract

1024 Background: HER2-low metastatic breast cancer (mBC) has emerged as a clinically significant subgroup since approval of trastuzumab deruxtecan. However, its relevance as a distinct subtype remains debated. We aimed to identify HER2-related gene signatures, examine the molecular characteristics of HER2-low mBC, and investigate the potential for classifying HER2-low mBC into subgroups with profiles resembling either HER2-positive (HER2+) or HER2-negative (HER2-) mBC. Methods: We performed differential gene expression (DGE) analysis using the TCGA-BRCA dataset, comparing HER2+ and HER2- samples to identify a 17-gene HER2 expression signature. The findings were validated on archival samples from a UAE clinical center, where DGE analysis was repeated to compare HER2+ and HER2-zero mBC. HER2-low samples were then classified into two subsets; positive-like and zero-like based on their HER2 signature gene expression level. Lastly, Gene Set Enrichment Analysis (GSEA) was conducted to evaluate the molecular characteristics of these subsets. Results: In the TCGA-BRCA dataset (n = 409; median age 59 years, range 26–90), 20.5% of samples were HER2+, and 79.5% were HER2-. DGE analysis identified a 17-gene HER2 signature that effectively separated HER2+ and HER2- mBC. These 17 genes including GSDMB, ERBB2, and MED1, are associated with HER2 expression. In the UAE cohort (n = 69; median age 52 years, range 24–84), comprising 7.25% HER2+, 60.87% HER2-low, and 31.88% HER2-zero mBC, 7 genes (GSDMB, GRB7, ERBB2, STARD3, PGAP3, MIEN1, TCAP) from the 17-gene HER2 signature were similarly upregulated in the HER2+ samples. We observed an increasing trend in HER2 signature expression across the groups, from HER2-zero to HER2-low and then HER2+ mBC showing the highest expression. This trend was supported by separation across HER2 status with gene-level expression, Gene Set Variation Analysis scores and Principal Component Analysis (PCA). Next, HER2-low mBC were classified into two distinct subgroups based on distance-to-centroid approach with PCA. Based on distance to HER2+ and HER2-zero centroids, HER2-low mBC were classified into positive-like and zero-like subgroups respectively. Separation of the positive-like and zero-like samples were observed with DGE analysis and expression of the HER2 signature genes. GSEA of HER2-low positive-like samples indicate increased activation of ERBB2 oncogenic pathway and suppression of gene sets involved in immune-mediated pathways compared to HER2-low zero-like mBC. Conclusions: This study reveals the potential utility of transcriptomic HER2 signature to characterize HER2-related molecular features in mBC, revealing a gradient of HER2 signature expression across HER2+, HER2-low, and HER2-zero. Identification of suppressed immune-mediated pathways in HER2-low positive-like mBC suggests combination with immune mediators as potential treatment strategy.