The Expression and Prognostic Significance of Major MicroRNA Genes in Breast Cancer Based on Bioinformatics Analysis

Objective: Breast Cancer (BC) is characterized by high complexity and heterogeneity, and microRNA (miRNA) is bound up with the occurrence and development of BC. In this study, we evaluated the prognostic value of miRNA in BC. Background: Breast ductal and lobular cancers are the most common types of Breast Carcinomas (BC) and indicate the high complexity heterogeneity in this disease. Each BC patient has unique morphological and molecular features. MicroRNAs (miRNAs) play a critical role in human oncogenesis, progression, and prognosis. Our study aimed to identify potential prognostic biomarkers of breast ductal and lobular cancers to predict the overall survival outcome. Methods: All analyzed miRNA sequencing and clinical data were obtained from the Genomic Data Commons Data Porta. edgeR package in R software was used to analyze the differential miRNA expression profiles. Complete survival information and differentially expressed miRNA expression were obtained and the Caret package was used for random division of the samples along with their profiles into two groups (training group and test group). We performed univariate Cox regression analyses for miRNAs in the training group. We utilized three different web-based tools to identify the target genes of miRNAs and used the Perl language to evaluate the target genes for miRNA signature. STRING database was used to assess PPIs. Results: A total of 304 differentially expressed miRNAs were identified (213 were upregulated and 91 were downregulated). Among these, nine (hsa-miR-204-5p, by Cox regression analysis and miRNA signature risk score built. And then we performed the model of BC patients for three years survival risk, the AUCs of ROC were 0.804, 0.667, and 0.739 in the training, test, and entire groups, respectively. miRNAs were differentially expressed in tumor-related biological processes and pathways by functional enrichment and bioinformatic analysis. Conclusion: The current study provided novel insights into the mi-RNA-based mRNA network in BC. The nine miRNA and ten hub genes may be independent prognostic signatures for survival prediction in BC patients.


Introduction
At present, BC of female patients has surpassed lung cancer as the leading cause of global cancer incidence. Globally, with an estimated 2.3 million new cases being reported, which corresponds to 11.7% of all cancer cases, it is the fifth leading cause of cancer-related mortality, contributing to 685,000 deaths [1]. There are several histologic subtypes of invasive breast carcinoma, including, but not limited to the following: invasive carcinoma of no special type (commonly known as invasive ductal carcinoma); invasive lobular carcinoma; tubular carcinoma; carcinoma with medullary features (sometimes called medullary carcinoma); and metaplastic carcinoma. Among these subtypes, Invasive (or Infiltrating) Ductal Carcinoma (IDC) accounts for the majority (approximately 75%) of BCs, followed by Invasive Lobular Carcinoma (ILC) which accounts for 5% -15% of BCs [2].
BC is a heterogeneous and complex disease with unique morphological and molecular features, unlike a disease where only a few genes, proteins, and/or signaling pathways contribute to its progression in a simple, independent, and autonomous manner. Previous studies showed that patients with the same type of BC can show differential treatment responses, which further contributes to the high heterogeneity in this disease [3]. Therefore, histological features have certain limitations for diagnosis, prognosis, and prediction of clinical outcomes for individual patients [4]. Many prognosis factors are available for BC; therefore, it is necessary to develop suitable prognostic biomarkers for the diagnosis and treatment of BC.
MicroRNAs (miRNAs) are a group of small non-coding RNAs that can interrupt the expression of protein-coding genes by binding to their mRNAs and inhibiting the subsequent protein translation [5]. miRNAs can regulate the expression of genes associated with various biological phenomena including homeostasis, development, proliferation, differentiation, and apoptosis [6]. Deregulated signaling and expression of miRNAs have been well-studied in the pathogenesis of various cancers. Aberrant expression of miRNAs is vital for the initiation and and oncogenes [7]. This study aimed at identifying the potential prognostic biomarkers to predict overall survival in BC. We also evaluated the clinical potential of miRNAs as biomarkers for early as well as differential diagnosis and prognosis of BC.

Identification of DEMs and DEGs
Evaluation of Overall Survival According to median value grouping based on the risk score in the training, test, and entire groups, Kaplan-Meier curves showed that indeed the high-risk group had poorer overall survival as compared to the low-risk group (p = 1.474E−08, p = 8.708E−03, and p = 7.825E−09, respectively; Figures 3(a)-(c)). The 5-year    overall survival for training-group patients was 69.8% and 92.7% for the high and low-risk groups, respectively. The 5-year overall survival for test group patients was 76.1% and 86.9% in the high and low-risk groups, respectively. The entire group showed a 5-year overall survival of 73.8% and 90.5% in the high and low-risk groups, respectively. The AUC of ROC based on the nine-miRNA signature for three years were 0.804, 0.667, 0.739 in the training, test, and entire groups, respectively ( Figures  3(d)-(f)), which showed that the model had better performance for prediction of BC patient survival risk. In addition, in all the three groups, the high-risk score patients had higher mortality rates as compared to their low-risk score counterparts (Figures 3(g)-(i)).

GO Annotation and KEGG Enrichment Analyses
We performed GO and KEGG analyses for a detailed understanding of the selected 522 target genes associated with BC. The top thirty terms from the GO annotations for Biological Process (BP), Cellular Component (CC), and Molecular

PPI Network and Module Selection
Using the STRING database and Cytoscape software, a total of 546 target genes were incorporated into the PPI network consisting of 178 nodes and 327

Discussion
Among women, the three most frequent malignancies are breast, lung, and colorectal cancers. These account for 50% of all new diagnoses; breast cancer alone accounts for 30% of female tumors [10]. Recently, with the advancement in early diagnoses and comprehensive treatment strategies, the mortality rate of BC has gradually decreased. However, distant metastasis is a major cause of death among these patients [11] [12]. Therefore, the identification of highly specific and sensitive prognostic markers for BC is important. Recently, miRNAs have been widely demonstrated to be associated with a variety of diseases including breast cancer [13]. A vast and growing body of evidence firmly supports the involvement of miRNAs in tumors including breast cancer [14]. However, most of these have studied signal genes. To capture the genes associated with DEMs in breast cancer, we screened the DEGs and DEMs and identified the key prognostic DEMs, all of which may be potential therapeutic targets. In the present study, from the TCGA database, the miRNA and mRNA expression profile data on invasive carcinoma of the breast were analyzed. DEMs and DEGs were analyzed using the edgeR package, and their targets were predicted. All the patients were randomly   in tumor tissues were significantly lower than those in adjacent tissues. These data may provide new parameters for clinical staging and prediction of lymph node metastasis and these two miRNAs could effectively inhibit the proliferation and invasion of cancer cells [18]. Previous studies show that miR-1247 is aberrantly expressed in several cancers [9] [25]. miR-605-5p promotes invasion and proliferation by targeting TNFAIP3 in NSCLC [26]. It also inhibits melanoma progression and glutamine catabolism by targeting GLS [27]. In a recent study, Lei et al. found that ectopic miR-615-3p expression in breast cancer cells promotes EMT-like traits. The tumor-promoting role of miR-615-3p is directly linked to its ability to target 3'-UTR of PICK1 [28]. MiR-4652, miR-9-5p, miR-2115 are expressed in several cancers. However, the current research mechanism of these three miRNA in tumors has not been reported yet. Thus, experiments in the future need to focus on the potential roles of these miRNA in BC.
To further understand the regulatory mechanism of the nine miRNA-based signatures in breast cancer, the target genes of nine miRNAs were predicted us-   [43]. Heterotrimeric G proteins are crucial in multiple cellular signal transduction pathways [44]. The guanine nucleotide-binding protein G(i), α-1 subunit (GNAI1), belongs to the Gαi family [45].

Conclusion
In summary, we constructed a new predictive model based on miRNA signature through miRNA mature expression profiling for the prognosis of BC. By grouping, we also verified and evaluated the predictive ability of the model. The most important finding was its utility as an independent prognostic factor in BC. Additionally, the target genes of the model provide a new direction for investigations on the pathogenesis and diagnosis, and development of targeted inhibitors for the breast cancer. However, this is a bioinformatic study and extensive clinical validations are required to confirm our findings of a reliable predictor and therapeutic target signature for BC. Advances in Breast Cancer Research

Conflicts of Interest
The authors declare no conflicts of interest.