Introduction
Breast cancer (BC) has become a modern global epidemic among women as its morbidity has markedly increased over the past few decades (1). According to the Cancer Statistics 2016, BC is the most frequently diagnosed malignancy and the second leading cause of tumor-related death in females in the US, with 246,660 new diagnoses and 40,450 deaths projected to have occurred in 2016 (2). Recently, accumulating evidence indicates that early diagnosis and treatment result in favorable prognoses; however, current clinical tumor biomarkers lack sensitivity and specificity for early detection of BC (3). Consequently, identification of novel diagnostic biomarkers with high sensitivity and specificity for differentiating breast cancer from benign breast disease and normal samples is urgently needed. Previous studies have reported that microRNAs (miRNAs) play vital roles in the initiation and progression of BC and that certain miRNAs can serve as potential diagnostic biomarkers (4,5).
miRNAs are short non-coding RNA molecules of 18–25 nucleotides that play essential roles in regulating gene expression by binding to target mRNAs and directing RNA-induced silencing complexes (RISCs) to degrade target mRNAs or repress mRNA translation (6,7). Notably, miRNAs are reportedly linked to the occurrence and development of a variety of malignant tumors. Numerous studies have indicated that miRNAs could serve as oncogenes or tumor suppressors by regulating cellular biological processes, such as cell differentiation, proliferation and apoptosis (4,8,9). Moreover, miRNAs are also involved in multiple cancer-related cell signaling pathways in various malignancies, including breast cancer (10,11). Previous studies have reported that many miRNAs are aberrantly expressed in BC and thus potentially act as biomarkers for cancer diagnosis. Fu et al (12) showed that miR-141, miR-145, miR-183, miR-21 and miR-638, which are differentially expressed in BC and normal tissues, function as potential diagnostic indicators to distinguish tumor cases from normal cases. However, most of the previous studies focused on the diagnostic value of a single miRNA. Additionally, the small sample size of current studies and the lack of sensitivity and specificity of candidate miRNAs limit the clinical application of these biomarkers. In recent years, the development of high-throughput technologies and bioinformatics has promoted the development of such diagnostic markers.
The Cancer Genome Atlas (TCGA) provides open access to many comprehensive miRNA-sequencing datasets. In the present study, we analyzed miRNA expression data published by TCGA to identify miRNAs differentially expressed between breast tumor and normal tissues. Ultimately, we determined a nine-miRNA signature for diagnosing breast cancer. Furthermore, we predicted the targets of these miRNAs through in silico algorithms and conducted gene oncology annotation and pathway enrichment analyses to determine the potential biological functions of the nine miRNAs in the signature.
Discussion
Increasing evidence indicates that dysregulation of miRNAs may be closely related to many malignances, including breast carcinoma (BC) (15–17). In the present study, we comprehensively analyzed high-throughput miRNA expression profiles of breast cancer specimens from TCGA and selected nine miRNAs with high diagnostic accuracy, including seven upregulated miRNAs (hsa-miR-21, hsa-miR-96, hsa-miR-183, hsa-miR-182, hsa-miR-141, hsa-miR-200a and hsa-miR-429) and two downregulated miRNAs (hsa-miR-139 and hsa-miR-145).
Among the seven upregulated miRNAs, miR-21 showed the most obvious overexpression in BC specimens. As one of the first mammalian onco-miRNAs identified, miR-21 has been experimentally confirmed to participate in the initiation and progression of various carcinomas. In BC, miR-21 also plays a vital role in carcinogenesis. miR-21 is located on chromosome 17q, a region containing several oncogenes, including HER2, TOP2A and PPM1D (18–20). miR-21 has been reported to function as an onco-miRNA by directly regulating several tumor suppressors, such as tropomyosin 1 (TPM1), programmed cell death 4 (PDCD4), maspin and phosphatase and tensin homolog (PTEN) (21,22). Additionally, it has been reported that miR-21 is involved in certain cancer-related signaling pathways, such as the PTEN/PI3K/AKT, the ERK1/2/MAPK and the miR-21/smad7/ERK signaling pathways, thus, promoting tumor growth and metastasis in vivo and in vitro (21,23–26).
Previous studies have shown that the miR-183/miR-96/miR-182 cluster is overexpressed in various malignant tumors, including BC; thus, this cluster may be an onco-miRNA cluster and could potentially act as a multi-marker for diagnosing malignancies (27). The upregulated miR-183/miR-96/miR-182 cluster inhibits transcriptional factors, such as FOXO1 and FOXO3, and thus, promotes BC cell proliferation, invasion, and migration and reduces cell apoptosis in vivo and in vitro (28,29).
As members of the miR-200 family, miR-141, miR-200a and miR-429 exert multiple complex effects on tumor occurrence and development in different carcinomas. Among the three miRNAs, miR-429 and miR-200a are located on chromosome 1p36.33, while miR-141 is located on chromosome 12p13.31. Several studies suggest that miR-200a, miR-429 and miR-141 may be tumor suppressors since they are downregulated in cancer samples (30–32). However, other studies showed that these three miRNAs may function as oncogenes since they are upregulated in tumor tissues (33–36). In the present study, the expression levels of the three miRNAs were significantly elevated in BC specimens compared with normal breast tissues. These discrepancies may be partially explained by the various sources of the cases studied, different detection techniques and different platforms used. It is reported that members of the miR-200 family can reverse epithelial-mesenchymal transition (EMT) by increasing E-cadherin expression and decreasing ZEB1, ZEB2 and β-catenin expression, further inhibiting cancer cell invasion and migration (31,37–40). Notably, EMT mechanisms are distinct in various malignancies, and the expression of EMT markers (E-cadherin and ZEB) is inconsistent in diverse types of cancers and different histological types of the same cancer (41). This could partially explain the opposing actions of miR-200a, miR-429 and miR-141 on tumorigenesis.
Real-time PCR and miRNA microarray confirm that miR-139 is downregulated in breast cancer (42–44), which is consistent with our results. In vitro, forced expression of miR-139 in the BC cell line MDA-MD-231 was found to suppress cell migration and invasion by targeting genes involved in metastasis-related signaling pathways (42). Functional experiments conducted by Hua et al (45) revealed that increased expression of miR-139 reduced cell proliferation in luminal type BC cells by directly targeting topoisomerase IIα (TOP2α), an oncogene that can promote cell proliferation in vitro.
Both previous studies, and our present study, revealed that miR-145 is downregulated in BC samples (46–50). Acting as an oncogene, miR-145 promoted cell apoptosis (46), and suppressed cell growth, proliferation, migration and invasion (47,51) in vitro. More importantly, the decreased expression of miR-145 was found to be negatively correlated with several clinicopathological parameters, such as ER and HER2 status, tumor size and lymph node metastasis (52), which indicated its potential diagnostic or prognostic value in the clinic.
Current studies have focused on the diagnostic value of dysregulated miRNAs in BC (12,53). However, this is the first identification of a nine-miRNA signature as a potential multi-marker for diagnosing BC found by analyzing high-throughput miRNA expression profiles from TCGA data. In the present study, the ROC curve of the combined nine differentially expressed miRNAs showed an extremely high diagnostic accuracy with an AUC of 0.995 (95% CI=0.988–0.999), and diagnostic sensitivity and specificity of 98.7 and 98.9%, respectively. The diagnostic value of these nine miRNAs combined was higher than that of the individual miRNAs, indicating that the nine-miRNA signature could serve as a potential multi-marker for diagnosing BC.
Previous studies have demonstrated that miRNAs play vital roles in carcinogenesis through participation in cancer-related pathways. Therefore, we performed target prediction of the nine selected miRNAs and subsequently carried out GO and KEGG pathway enrichment analyses to explore the potential effects of these nine miRNAs on carcinogenesis. GO and KEGG pathways revealed that the targets of these dysregulated miRNAs were involved in many critical cancer-related biological processes and pathways, such as positive regulation of macromolecule biosynthetic process, regulation of transcription, regulation of gene expression, the enzyme-linked receptor protein signaling pathway, pathways in cancer, the TGF-β signaling pathway, and the MAPK signaling pathway. The results of enrichment analyses suggested that these nine miRNAs could regulate important pathways correlated with breast cancer by targeting oncogenes or tumor-suppressor genes.
To the best of our knowledge, this is the first study identifying a nine-miRNA signature as a multi-biomarker for diagnosing BC based on TCGA datasets. Nevertheless, the present study has several limitations. Firstly, our data were retrieved from TCGA and had not been verified in experiments. Therefore, further multicenter clinical trials are needed in the future to confirm our findings. Secondly, the expression profiling of these miRNAs was only based on tissue. It may be of more clinical value to investigate noninvasive diagnostic markers.
In conclusion, in the present study, we identified a tumor-specific miRNA signature in BC containing seven upregulated miRNAs (hsa-miR-183, hsa-miR-96, hsa-miR-21, hsa-miR-182, hsa-miR-141, hsa-miR-200a and hsa-miR-429) and two downregulated miRNAs (hsa-miR-139 and hsa-miR-145) by analyzing high-throughput data from a TCGA dataset. We discovered that these nine aberrantly expressed miRNAs may play important roles in the initiation and progression of BC. More importantly, we assessed the combination of these nine miRNAs, and found that an integrated analysis of these nine miRNAs improved the accuracy of breast cancer diagnosis. However, multicenter studies with large sample sizes are needed to validate our findings before the clinical application of this nine-miRNA signature.