Refinement of breast cancer molecular classification by miRNA expression profiles

Background: Accurate classification of breast cancer using gene expression profiles has contributed to a better understanding of the biological mechanisms behind the disease and has paved the way for better prognostication and treatment prediction. Results: We found that miRNA profiles largely recapitulate intrinsic subtypes. In the case of HER2-enriched tumors a small set of miRNAs including the HER2-encoded mir-4728 identifies the group with very high specificity. We also identified differential expression of the miR-99a/let-7c/miR-125b miRNA cluster as a marker for separation of the Luminal A and B subtypes. High expression of this miRNA cluster is linked to better overall survival among patients with Luminal A tumors. Correlation between the miRNA cluster and their precursor LINC00478 is highly significant suggesting that its expression could help improve the accuracy of present day’s signatures. Conclusions: We show here that miRNA expression can be translated into mRNA profiles and that the inclusion of miRNA information facilitates the molecular diagnosis of specific subtypes, in particular the clinically relevant sub-classification of luminal tumors. Illumina paired-end RNA-sequencing was performed on 1600 sequencing libraries (49 technical replicates, 1552 tumour samples) for fusion gene detection analysis. miRNA sequencing was performed on a subset of the fusion detection samples, 191 sequence libraries (5 technical replicates, 186 tumour samples), for miRNA transcript expression estimation. This series is a re-analysis of GSE100769. ------------------------------------ The authors state "due to Swedish law, the patient consent, and the risk that the sequencing data contains personally-identifiable information andhereditary mutations, we cannot deposit the short sequencing read data in a repository". -------------------------------------- This represents the miRNA sequencing component of 191 libraries only.