Identification of circular RNAs as a promising new class of diagnostic biomarkers for human breast cancer.

Endogenous noncoding circular RNAs (circRNAs) have gained attention for their involvement in carcinogenesis, but their expression pattern in breast cancer has remained largely unknown. In this two-stage study, we first used an Arraystar Human circRNA Array to construct a genome-wide circRNA profile. We then selected candidate circRNAs for validation using a quantitative real-time polymerase chain reaction system. CircRNA/miRNA interactions were predicted and sequence analyses were performed. Among 1155 differentially expressed circRNAs, 715 were upregulated and 440 were downregulated in breast cancer tissues. The validation study demonstrated that hsa_circ_103110, hsa_circ_104689 and hsa_circ_104821 levels were elevated in breast cancer tissues, whereas hsa_circ_006054, hsa_circ_100219 and hsa_circ_406697 were downregulated. These circRNAs targeted complementary miRNA response elements. The area under the receiver operating characteristic curve for distinguishing breast cancer was 0.82 (95% CI: 0.73-0.90) when hsa_circ_006054, hsa_circ_100219 and hsa_circ_406697 were used in combination. This study provides evidence that circRNAs are differentially expressed in breast cancer and are important in carcinogenesis because they participate in cancer-related pathways and sequester miRNAs.