Prognostic role of microRNAs in breast cancer: A systematic review

MicroRNAs (miRNAs) have been found to play an important role in breast cancer, functioning either as potential oncogenes or tumor suppressor genes, but their role in the prognosis of patients remains unclear. The aim of the present review study is to highlight recent preclinical and clinical studies performed on both circulating and tissue-specific miRNAs and their potential role as prognostic markers in breast cancer. We systematically searched the PubMed database to explore the prognostic value of miRNAs in breast cancer. After performing the literature search and review, 117 eligible studies were identified. We found that 110 aberrantly expressed miRNAs have been associated with prognosis in breast cancer. In conclusion, the collective data presented in this review indicate that miRNAs could serve as novel prognostic tools in breast cancer, while the clinical application of these findings has yet to be verified.


INTRODUCTION
Breast carcinoma is the leading cause of cancer death in women worldwide [1]. According to the GLOBOCAN 2018 worldwide estimates of cancer incidence and mortality, in 2018, about 2,088,849 new cases were diagnosed and approximately 626,679 women were predicted to die from the disease [2]. These data support the need to develop more efficient strategies for preventive intervention, evaluation of therapy, and prediction of prognosis [3].
Undoubtedly, TNM staging is of great prognostic value; however, considering all the limitations of the currently available prognostic strategies, it is overall recognized that new affordable more accurate methods indicative of molecular characteristics of tumors are needed to achieve personalized treatment [4]. Still, it remains difficult to achieve these goals, because of the absence of refined (sensitive and specific) biomarkers for disease monitoring and for addressing breast cancer on an individual basis.
MicroRNAs are a small class of endogenous, evolutionarily conserved, single-stranded noncoding RNAs, with a length of approximately 19-24 nucleotides [5]. Interaction between miRNAs and mRNAs, within the 3′untranslated region of the target genes, leads to the degradation or inhibition of mRNA translation [6]. In the past few years, miRNAs have attracted considerable attention in the cancer research field, due to their regulatory actions in multiple levels [7,8]. Depending on the target gene that they regulate, miRNAs can either serve as "tumor suppressor miRs" by repressing oncogenes or as "onco-miRs" by targeting tumor suppressor genes. However, a number of miRNAs play both tumor suppressor and onco-miR roles depending on the cellular context and tumor type [9].
Particularly in breast cancer, microRNAs (miRNAs or miRs) have been proposed as promising biomarkers because they can be readily detected in tumor biopsies (noncirculating miRNAs) and can also be identified in blood, plasma, serum, and saliva (circulating miRNAs) [10]. Review Furthermore, circulating miRNAs are bound to lipoproteins such as HDL, are associated with Argonaute 2 (Ago2) protein, or are packaged into exosome-like microparticles, micro-vesicles, and apoptotic bodies [11]. Therefore, they are protected from endogenous RNAase activity, and hence they are reliable.
Several lines of evidence have proven that in breast cancer, the expression levels of miRNAs are altered due to key mechanisms, such as epigenetic control, transcription factors, or the effect of mutated proteins [10]. According to previous publications [12], miRNAs are considered as tumor suppressive or protective when they are down-regulated in cancer compared to their normal counterpart, or else, they are termed oncogenic miRNAs or onco-miRs. In this context, miRNAs are increasingly recognized as promising biomarkers, given the fact that they are easy to isolate, and they maintain their structural stability under different conditions of sample processing and isolation. A prognostic biomarker should indicate a patient's outcome, for example disease recurrence or disease progression, independent of the treatment regimen that was followed, and they are highly desirable for personalized or precise patient treatment [13].
The aim of the present review is to highlight recent preclinical and clinical studies performed on both circulating and tissue-specific miRNAs and therefore to identify their potential role as prognostic markers in breast cancer. We will particularly focus on the potential role of miRNAs in breast cancer prognosis, and on how miRNAs have the potential to answer actual clinical needs, such as identification of biomarkers for prognosis, in order to achieve the goal of individualized breast cancer treatment.

RESULTS
The search strategy retrieved 192 articles. Of these articles, 42 were irrelevant, 11 were reviews, eight (8) were meta-analyses, six (6) were retracted articles, three (3) were not in English, three (3) were duplicates, two (2) were comments and 117 were eligible. The aforementioned steps concerning the selection of studies are illustrated in detail in Figure 1. Therefore, a total of 117 articles were eligible for this systematic review and the prognostic role of 110 miRNA molecules is described (Table 1). Furthermore, we retrieved five studies, in which authors have identified six distinct microRNA signatures with prognostic value in breast cancer ( Table 2).
According to our results, presented in Table 1, the majority of publications have not taken into account the distinct breast cancer subtypes during the development of their research protocol, since in 60.8% of studies breast cancer samples were not classified. The remaining 25.8% focused on Triple Negative Breast Cancer (TNBC) samples or involved Luminal A (5.0%), Luminal B (1.7%) and HER2-positive (1.7%) breast cancer samples. Of note, 5.0% of the selected studies accessed the prognostic value of miRNAs through experiments performed on breast cancer cell lines. Different detection methods, as well as different sample types were used for the detection of the prognostic miRNA expression levels (i.e., paraffin-fixed, formalin-fixed, freshly frozen tumors, plasma or serum). Concerning the detection methods, quantitative reverse transcriptase real-time polymerase chain reaction (qRT-PCR) was used in 35,8% of the eligible studies, while in 21,7% of the studies qRT-PCR was performed along with Microarray analysis. Additionally, next generation Sequencing technologies (9,2%), in situ hybridization techniques (9,2%), luciferase report assays (6,7%) or a combination of various techniques (10,8%) were employed.

DISCUSSION
We conducted a comprehensive systematic literature review to unfold the utility of miRNA biomarkers that can be evaluated for predicting prognosis in breast cancer patients. We have identified 117 studies that investigate the potential correlation between miRNA profile expression in breast cancer tissue and in the circulation and their possible use as prognostic factors. Interestingly, most of the miRNAs found to be associated with prognosis in breast cancer, were assessed in only a single study. Six miRs (miR-10b, miR-200b, miR-21, miR-203, miR-375, and miR-210) were evaluated in at least four studies and the discussion will be mainly focused on these molecules, based on an effort to merely provide some important information on the most commonly researched molecules in accordance with our systematic literature review.
Mir-210 has multiple functions in cancer cells and is involved in angiogenesis, cell cycle regulation, DNA damage repair, mitochondrial metabolism, and immune response [32]. According to our search results, including seven studies [33][34][35][36][37][38][39], high expression of miR-210 has been significantly associated with poor survival in patients with breast cancer. Notably, single miR-210 assay has been proposed as an independent prognostic factor in this disease.
Concerning miR-10b, it has been presented as a potential biomarker that could play a predictive role in lymph node metastases occurrence across TNBC and www.oncotarget.com in the incidence of high-grade tumors in non-TNBC cases [17]. Elevated expression of miR-10b in breast tumor tissue samples has been associated with adverse outcome, which is further supported from data derived from in vitro studies [40]. Finally, a survival analysis of 230 breast tissue samples has shown that high levels of miR-10b result to a short relapse free survival (RFS) of breast cancer, acting as an independent prognostic factor of RFS [41]. Our results, emphasize the oncogenic role of miR-10b and indicate that its high expression may be correlated with poor survival in breast cancer, while a recent metanalysis further strengthens our findings [30].  MiR-200 family members function as regulators of the epithelial to mesenchymal transition (EMT), which is one of the initial steps in tumor metastasis [42]. Specifically, miR-200b and miR-203 have both been characterized as tumor suppressors in multiple breast tumor types [43]. However, there seems to be an inconsistency in the existing literature, since we retrieved two studies that have found that higher expression of circulating miR-200b and miR-203 are associated with worse outcome [33,34], further substantiated by a study on breast cancer cell lines [44]. However, other studies on breast cancer tissue samples and cell lines presented inverse results [45][46][47]. These discrepancies exhibit the diverse regulatory roles of miR-200 family members, depending on the cellular context and type of biological sample (blood VS tissue), and highlight the potential prognostic impact of these EMT regulating miRNA molecules in breast cancer.
Furthermore, our search retrieved five studies that have found six miRNA signatures to be useful for predicting the outcome of breast cancer [48][49][50][51][52]. Coordinated regulation of multiple miRNAs of potential prognostic value, has helped researchers identify panels of prognostic microRNAs for breast cancer. The discovery of microRNA expression signatures shows considerable promise for determining the prognosis of individuals with breast cancer. Similar miRNA signatures have been identified in a variety of other cancers, including acute myeloid leukemia, chronic lymphocytic leukemia, colon cancer, pancreatic cancer, and non-small cell lung cancer [53]. These reports highlight that this class of RNA molecules is showing substantial potential to be used as prognostic biomarkers for cancer.
Among the limitations of this effort, it should be stressed that this process was essentially driven by the search algorithm, which focused mainly on titles of the published literature, in an effort to provide more relevant results. Furthermore, clear heterogeneity was observed in our results, due to differences in patient characteristics (ethnicity, age, tumor stage, grade and subtype) and the use of different isolation and detection methods, cut-off values for miRNA expression levels, sample preparation methods and sample types (i.e., paraffin-fixed, formalinfixed, freshly frozen tumors, plasma or serum).
Based on the results of this systematic review, we consider that miRNA detection may address the need for independent, easily accessible, prognostic molecular markers for breast cancer management in clinical practice, by assessing the impact of aberrant miRNA expression on patients' survival. Our work sums up all the available data on prognostic miRNAs and can also act as a valuable reference point for future studies. Furthermore, while prognostic studies can assist in answering important questions concerning specific patient outcomes, their vigorous and careful design is a necessary condition for ensuring the reliability of results [54]. It should be stressed out that the thorough validation of prognostic factors is a necessary and unavoidable process in order to maximize

Methods of search strategy and study eligibility
This systematic review was conducted in accordance with the PRISMA guidelines [55] and in line with the a priori protocol agreed on and signed by EZ and FZ. Eligible studies were sought in PubMed without any restriction of publication language; the end-of-search date was January 28, 2019. In instances where multiple (overlapping) publications stemming from the same study were identified, the larger size study and the one with longer follow-up were included, unless the reported outcomes were mutually exclusive. Authors working independently and blindly to each other in pairs (E.Z., F.Z.) performed the selection of eligible studies; in case of disagreement, consensus with the whole team was reached.

Data extraction
The extraction of data comprised general information, including the name of the miRNA molecule, the breast cancer type in which its expression was determined, method of detection, the sample type that was used, its prognostic value in breast cancer, its function in cancer (onco-miR or tumor suppressor-miR) and the author-year of publication. Data were independently extracted and analyzed by a pair of reviewers (E.Z. and F.Z.), with one reviewer being blinded to the other; if needed, the final decision was reached by a team consensus.
Eligible literature met the following criteria: (1) measured miR expression levels in tumor or blood samples or human cell lines and (2) only articles in English. Publications were excluded if they had one or more of the following criteria: (1) studies referring to the prognostic role of single nucleotide polymorphisms (SNPs) in miRNA genes affecting their function; (2) studies that refer to the prognostic role of target miRNA molecules (molecules regulated by miRs); (3) studies based solely on a bioinformatics approach or a computational algorithm, with survival data originated from databases without subsequent biological validation and (4) review papers, meta-analyses, comments, letters or duplicate publications.

CONFLICTS OF INTEREST
The authors declare that they have no potential conflicts of interest.