Subtype-associated microRNA expression patterns in Taiwanese breast cancer: an exploratory microarray analysis

Background Breast cancer is a heterogeneous disease, and immunohistochemistry (IHC) remains the foundation of clinical subtyping. MicroRNAs (miRNAs) regulate gene expression post-transcriptionally and have been linked to subtype-specific tumour biology, but the patterns observed in East Asian cohorts profiled on contemporary transcriptome arrays are still limited. We profiled a Taiwanese hospital-based cohort to describe estrogen (ER)- and human epidermal growth factor receptor 2 (HER2)-associated miRNA expression patterns. Methods A total of 104 primary breast cancers were profiled on the Affymetrix Clariom D human transcriptome array. After quality filtering, 1,629 miRNA probe sets were retained. Differential expression was assessed using limma with empirical-Bayes variance moderation, and false-discovery rate was controlled. Functional context was provided by Tool for Annotations of miRNAs (TAM) 2.0 enrichment. Top candidates were replicated in TCGA-BRCA miRNA-seq cohort. Results At a nominal discovery threshold of P < 0.01, 11 miRNAs were ER-associated and 14 unique mature miRNAs (17 probe sets) were HER2-associated. The strongest signal was an approximately five-fold upregulation of miR-4728 in HER2 + tumors (P = 5.2 x 10 ^− 4 ), consistent with its intronic co-expression from the ERBB2 locus. miR-7-2 was the strongest ER-associated probe set. Replication in TCGA-BRCA recovered the HER2/miR-4728 association at genome-wide FDR significance, supporting the discovery cohort. Conclusions In this Taiwanese cohort, microarray profiling identified candidate ER- and HER2-associated miRNAs, with miR-4728 emerging as the most robust HER2-associated signal in both the discovery and the TCGA replication cohorts. These results require formal biomarker validation before being considered for diagnostic or predictive use.