Dysregulation of microRNAs in breast cancer and their potential role as prognostic and predictive biomarkers in patient management

MicroRNAs (miRNAs) are an emerging class of gene expression modulators with relevant roles in several biological processes, including cell differentiation, development, apoptosis, and regulation of the cell cycle. Deregulation of those tiny RNA molecules has been described frequently as a major determinant for the initiation and progression of diseases, including cancer. Not only miRNAs but also the enzymes responsible for miRNA processing could be deregulated in cancer. In this review, we address the role of miRNAs in the pathogenesis of breast cancer, since there are oncogenic, tumor-suppressive, and metastatic-influencing miRNAs. Additionally, the different detection platforms and normalization strategies for miRNAs will be discussed. The major part of this review, however, will focus on the capability of miRNAs to act as diagnostic, predictive, or prognostic biomarkers. We will give an overview of their potential to correlate with response to or benefit from a given treatment and we will consider their ability to give information on prognosis in breast cancer. We will focus on miRNAs validated by more than one study or verified in independent cohorts or where results rely on preclinical as well as clinical evidence. As such, we will discuss their potential use in the personalized management of breast cancer.


Breast cancer and breast cancer heterogeneity
Breast cancer is the most frequent carcinoma and second most common cause of cancer-related mortality in women [1]. Its heterogeneous character is reflected in the classification into four intrinsic subtypes (luminal A, luminal B, basal-like, and ErbB2 + ), a normal-like group, and a new subtype, referred to as claudin-low. Histologically, breast cancer can be divided into in situ and invasive carcinoma, both of which can be further subdivided into ductal and lobular. Stratification by integrative clustering is based on genomic and transcriptomic data [2]. Curtis and colleagues [2] revealed new biological subgroups as integrative clusters, characterized by welldefined copy number alterations, gene expression, and distinct clinical outcomes.

MicroRNAs
MicroRNAs (miRNAs) are a group of small non-coding RNAs able to regulate gene expression at the posttranscriptional level by binding to the 3′ untranslated region (UTR) of target mRNAs. This leads to cleavage of the target mRNA by the Ago2 ribonuclease in the RNAinduced silencing complex (RISC) or inhibition of translation [3]. Alternatively, miRNAs can exert their effect by modulating the relationship between effector and target mRNAs rather than acting as regulators of particular mRNAs [4]. Deregulation of miRNA expression and potential altered gene expression may contribute to the development of cancerous phenotypes [5]. Several studies showed a differential miRNA expression profile in cancer as compared with normal, and a global miRNA downregulation was a common trait of human malignancies [6,7].

Materials and methods
Relevant published articles were searched on PubMed databases from 2001 to present by using the following search criteria to retrieve articles and abstracts: (micro-RNA* or miR*) AND (breast cancer); depending on the chapter discussed in the review, we complemented this with AND (diagnostic OR prognostic OR predictive).
We focused on recent publications or findings validated by several independent studies. Studies were considered eligible on the basis of the following inclusion criteria: (i) miRNAs were evaluated in breast cancer cells or samples (blood or tissue) derived from patients with breast cancer, including publically available breast cancer cohorts; (ii) the relationship between the studied miR-NAs and breast cancer biology was investigated, or (iii) the relationship between miRNAs and outcome or therapy response in breast cancer was examined. Articles were excluded if they met one of the following criteria: (i) published articles were retracted articles or comments; (ii) lack of key information on breast cancer biology, prognosis, or prediction of therapy response; and (iii) manuscripts reported on cancer types other than breast cancer.

MicroRNAs and breast cancer
The role of microRNAs in breast cancer biology and metastasis In cancer, miRNAs play a role in oncogenesis, metastasis, and resistance to various therapies and can be classified as oncogenes (oncomirs) or tumor-suppressor genes [8][9][10]. Additionally, both pro-metastatic ('metastamiRs') and metastasis-suppressor miRNAs can be identified [11,12].
Many miRNA genes are located in genomic regions involved in chromosomal alterations. Chromosomal regions encompassing oncogenic miRNAs may be amplified, resulting in increased expression of the oncomir. Tumor-suppressive miRNAs could reside in fragile sites characterized by deletions or mutations, leading to reduced levels of these miRNAs [13,14]. An overview of the most prominent miRNAs involved in the pathogenesis of breast cancer is depicted in Tables 1 and 2.

Tumor-suppressive and metastasis-suppressive microRNAs
Tumor-suppressor miRNAs exhibit a lower expression in cancer cells and suppress oncogene expression, thereby controlling cellular differentiation [28]. In regard to the global downregulation of miRNAs in cancer, it is worth mentioning that the majority of miRNAs have a tumorsuppressive function, some of which exhibit anti-metastatic properties as well. (See Table 2 and list of abbreviations for miRNA targets).
MiR-205 regulates HMGB3 and its ectopic expression significantly inhibits cell proliferation and promotes apoptosis in breast cancer [32,33]. Jiang and colleagues [34] revealed how expression of the miR-17-92 cluster in triple-negative breast cancer (TNBC) was associated with significantly reduced metastasis [35]. miR-206 inhibits cell proliferation, migration, and invasion by targeting cyclin D2 and Cx43 [36,37]. Certain tumor-suppressive miRNAs have an inhibitory role in the metastasis cascade. For example, the miR-200 family regulates EMT via E-cadherin expression by targeting several EMT inducers [38][39][40]. MiR-146b inhibits nuclear factor kappa B (NF-κB)-dependent interleukin (IL)-6 expression, which is associated with impaired survival and metastasis of cancer cells [41,42]. Decreased expression of miR-126 was seen in a variety of human cancers, and restoration of expression reduced metastatic endothelial recruitment and angiogenesis [43,44]. MiR-335 is known to target several metastasis-associated genes (for example, SOX4 and TNC) [44,45]. Finally, independent studies have reported the pleiotropic actions of miR-31 on breast cancer metastasis and its ability to suppress multiple steps of the invasion-metastasis cascade, by targeting RhoA and WAVE3 [46][47][48].

Context-dependent microRNAs
The action of certain miRNAs is dependent upon cellular or environmental context and results in both tumorsuppressive and -promoting roles [17,49]. This could in part explain possible inconsistent outcomes from distinct miRNA-examining studies.
As described above, miR-146 exerts its tumor-suppressive action through negative regulation of NF-κB signaling [41]. However, an additional target is the pro-apoptotic DNA repair enzyme BRCA1 [50]. MiR-29 stimulates metastasis and EMT by suppressing the cell-adhesion molecule peroxidasin homologue but is also capable of restraining cancer progression by targeting proliferationinducing oncogenes, suppressing DNA methylation of tumor-suppressor genes, and increasing chemosensitivity [51]. Although most data support an oncogenic function for miR-373 and -520c [26,27], a recent study reported this miR-family as a tumor suppressor in estrogen receptor (ER)-negative breast cancer [52].

MicroRNA normalization
Good normalization procedures for miRNA expression are still lacking. A normalization strategy for miRNA quantitative RT-PCR (qRT-PCR) is the use of endogenous controls, like reference miRNAs (for example, miR-16) or other small non-coding RNAs (for example, small nuclear and nucleolar RNA). However, it is recommended to use reference molecules belonging to the same RNA class since they possess the same physicochemical properties [59]. This is why miR-16 would be preferable over the use of small nucleolar and nuclear RNAs and was used in our study to normalize quantitative PCR data obtained from miRNA expression analysis in blood [60]. Also, the quantification accuracy increases by use of more than one reference gene [59]. Another example of a potential reliable reference gene is miR-484, whose referencegene utility may be enhanced by averaging with other potential breast cancer housekeeping miRNAs (for example, miR-330-3p, -4327, and -1271) [4]. Other normalization strategies for miRNA expression include the mean expression value strategy, as proposed by Mestdagh and colleagues [61].

MicroRNAs as diagnostic, predictive, and prognostic biomarkers
The ideal biomarker should be easily accessible, sensitive enough to detect all tumors, and specific and therefore not detectable in healthy individuals. Because of their high tissue-specificity, great stability, and aberrant expression in different tumor types, miRNAs are thought of as specific biomarkers with diagnostic, predictive, and prognostic potential ( Figure 1).

MicroRNAs as diagnostic biomarkers for breast cancer heterogeneity Early diagnosis of breast cancer
Early diagnosis is of utmost importance to reduce the mortality rate in cancer. Therefore, the search for new diagnostic biomarkers remains a persisting quest. The main examined diagnostic miRNA signatures will be discussed, with a focus on those miRNAs validated by more than one study or verified in different cohorts or both.
(For a detailed overview of the individual miRNAs, see Table 4). In a recent study, a signature of nine circulating miR-NAs was capable of discriminating between early-stage breast cancer and healthy controls [62]. A meta-analysis concerning miR-155 showed highly sensitive and specific diagnostic accuracy [63]. By comparing miRNA profiles between serum samples from breast cancer patients and healthy volunteers, Chan and colleagues [64] identified four miRNAs as significant diagnostic markers. Cuk and colleagues [65] found another four upregulated miRNAs in plasma of patients with breast cancer, capable of detecting stage I or II breast cancer, making them attractive candidates for early breast cancer detection. Another study reported three upregulated miRNAs and one downregulated miRNA in patients with breast cancer compared with normal controls [66]. A prospective study identified three significantly overexpressed serum miRNAs in women who eventually developed breast cancer (cases) compared with breast cancer-free controls [67]. This study introduces a new perspective on the miRNA function by describing their potential to predict increased risk of developing breast cancer.

Breast cancer molecular subtypes and microRNAs
We have shown that microRNA expression profiles are able to recapitulate the molecular breast cancer subtypes by using mRNA profiling [60,68]. Molecular miRNA signatures which distinguish between different breast cancer subtypes were described for the first time (to the best of our knowledge) by Blenkiron and colleagues [69] and reviewed by Serpico and colleagues [70]. The commonly found molecular subtype-related miRNAs are presented in Table 5.
Blenkiron and colleagues [69] profiled 309 miRNAs in 93 breast tumors with different molecular subtypes. The differential miRNA expression resulted in a correct classification of basal versus luminal subtypes. An identified 31-miRNA signature was able to distinguish the different breast cancer subtypes. de Rinaldis and colleagues [71] found consistent results by revealing a 46-miRNA signature able to differentiate between breast cancer subtypes Table 3 The most prominent microRNA detection methods

Detection method Throughput Sensitivity Specificity Reference
Northern blotting Low Low High [53] Microarray High Low Low [54] Bead-based flow cytometry High Medium High [6] qRT-PCR High High High [55] In situ hybridization Low Low Low [56] Next-generation sequencing High High High [57,58] qRT-PCR, quantitative real-time polymerase chain reaction. (n = 173). A substantial degree of consistency was also observed by Dvigne and colleagues [4], and the common miRNA signatures for molecular breast cancer subtypes are listed in Table 5.
In another study, 453 miRNAs in 29 early-stage breast cancer tumors were profiled, identifying signatures that accurately predict ER, progesterone receptor (PR), and HER2 status (Table 6) [72]. miR-342 showed the highest expression in ER-positive and HER2/neu-positive luminal B tumors, which was verified in recent studies [73,74], and showed a decreased expression in TNBC. MiR-520 g was downregulated in ER-and PR-positive tumors.

Breast cancer histological subtypes and microRNAs
Little is known about the correlation of histological subtypes and specific miRNA expression patterns and it is definitely worthwhile to further investigate this relation in the future.
Volinia and colleagues [75] found a nine-miR invasiveness signature by profiling miRNAs in ductal carcinoma in situ and invasive ductal carcinoma (IDC). Giricz and colleagues [76] revealed six differentially expressed miR-NAs during progression from lobular carcinoma in situ to invasive lobular carcinoma. Van der Auwera and colleagues [77] showed a differential expression of 13 miR-NAs in inflammatory breast cancer (IBC) versus non-IBC.
Another study examined the miRNA expression profile in IBC and found five miRNAs able to accurately classify between IBC and non-IBC [78].

Predictive microRNAs
Predictors help to individualize therapy and diagnosis of breast cancer, correlate with response to a given treatment, and determine the treatment benefit. Recently, several miRNAs have been described to serve as putative therapeutic targets. (See Table 7 for described miRNAs and Additional file 1: Table SI 1 for a more extensive overview of predictive miRNAs).

MicroRNAs associated with hormone therapies
MiRNAs appear to be involved in the process of endocrine resistance [79], and research has been conducted to identify miRNAs associated with clinical benefit of hormone therapies.
miR-375 miR-375 demonstrated its sensitizing effect on tamoxifen response via direct targeting of metadherin (MTDH). Loss of MTDH restored sensitivity to tamoxifen and correlated with disease-free survival (DFS) in tamoxifen-treated patients [80].
miR-342 He and colleagues [74] found that miR-342 expression positively correlates with ERα expression and that introducing miR-342 into estrogen-dependent breast cancer cell lines enhanced sensitivity to tamoxifen-induced apoptosis. Cittelly and colleagues [81] verified these findings and reported miR-342 downregulation to be associated with tamoxifen resistance.

miR-221/222
The miR-221/222 cluster is an inhibitor of ERα and is being associated with tamoxifen resistance in breast cancer cells [82][83][84]. This miRNA cluster has also been involved in resistance to fulvestrant, a selective ER downregulator [85].
Let-7f Aromatase inhibitors are used in endocrine therapy since they decrease estrogen production by blocking the aromatase gene CYP19A1, a direct target of let-7 f. MiRNA expression profiling before and after treatment with letrozole showed an increase in let-7f in preclinical as well as clinical settings [86].
MicroRNAs associated with targeted therapies miR-210 Jung and colleagues [87] examined the plasma miR-210 expression level in patients with HER2 + breast cancer before (that is, baseline expression) and after Common microRNAs (miRNAs) are indicated in bold. a miRs only detected in plasma, not in tissue. b Cases, women who developed breast cancer (BC); controls, women who stayed BC-free. neoadjuvant chemotherapy (NCT) that included trastuzumab. MiR-210 was the only miRNA with a significant upregulated baseline expression in the residual disease group before treatment. Therefore, high miR-210 baseline expression was associated with resistance to trastuzumab-included chemotherapy. Results were validated by comparing miR-210 expression in trastuzumabsensitive and -resistant breast cancer cells and in an independent set of pre-and post-operative plasma samples. A significantly higher miR-210 level was found in the trastuzumab-resistant cells and in the independent patient cohort before surgery.

MicroRNAs associated with response to chemotherapeutic agents
Drug sensitivity varies with each patient, making predictors of benefit or resistance toward proposed chemotherapeutic agents essential. In this way, the proportion of patients with a beneficial treatment would increase and toxicity of ineffective treatments would be avoided.
miR-125b Zhou and colleagues [88] reported resistance to various chemotherapies induced by miR-125b targeted repression of Bak1 (BCL2 antagonist killer 1), which was verified by the clinical findings of Wang and colleagues [89], who reported higher miR-125b expression in nonresponsive patients after admission of 5-fluorouracil. Climent and colleagues [90] reported miR-125b deletion  on chromosome 11q to be correlated with benefit of anthracycline-based chemotherapy and low recurrence rate in patients with lymph node-negative breast cancer.
miR-30c Bockhorn and colleagues [91] described how increased levels of miR-30c sensitized the drug response of breast cancer cell lines to paclitaxel and doxorubicin, and this was also seen in the preclinical model by Fang and colleagues [92].

miR-21
Mei and colleagues [93] reported miR-21 upregulation to be associated with taxol resistance in breast cancer cells. These results were validated in a recent study, in which miR-21 upregulation resulted in an increase of the anti-apoptosis protein BCL-2 and chemoresistance in breast cancer cells [9]. The increase of BCL-2 expression was induced by direct targeting by miR-21, which was able to unconventionally upregulate the expression of its direct target by binding to its 3′ UTR [94].
MicroRNAs associated with radiotherapy miR-34a Low levels of miR-34a rendered breast cancer cells more resistant to radiotherapy [95]. These findings were verified by Stankevicins and colleagues [96], who postulated that miR-34a was involved in breast cell responses to low-dose X radiation. Moreover, miR-34a was found to be upregulated by p53 in response to radiation in normal and breast cancer cell lines [96].

Prognostic microRNAs
Since patients at higher risk might acquire differential therapeutic interventions, the search for prognostic biomarkers remains a continuous work in progress. Several gene-expression studies have identified new or improved miRNA prognostic markers, giving information on the course and outcome of disease in different subgroups of patients. To narrow this extensive area, we will emphasize those prognostic miRNAs that are being validated (that is, examined in more than one study or tested in the same study) but verified in more than one cohort or dataset. An overview of the miRNAs described in this review can be found in Table 8 and Table 9; for a more extensive list, see Additional file 1: Tables S2A and S2B.

MicroRNAs associated with positive prognosis
Let-7b and miR-205 Quesne and colleagues [97] revealed association of let-7b and miR-205 with prognosis in breast cancer. They applied in situ hybridization to study miRNA expression in a population-based breast tumor cohort and validated their findings by use of qRT-PCR. Deregulation of let-7b, which targets the oncogenes H-RAS and HMGA2, often occurs early in breast cancer progression and its expression is known to be downregulated during EMT and associated with less aggressive tumors [69,[98][99][100]. Within luminal breast cancer, increased let-7b expression was positively associated with survival. Additionally, the authors reported miR-205 as another miRNA with positive prognostic value [97]. MiR-205 regulates EMT by inhibiting E-cadherin suppression. MiR-205 expression is associated with tumors of ductal morphology and independently predicts survival within this tumor subtype.
Two independent studies verified these results. Ma and colleagues [101] showed how low let-7b expression levels in a heterogeneous breast cancer cohort associated with poor prognosis reflected in lower overall survival (OS) and relapse-free survival (RFS) times. Markou and colleagues [102] detected miRNA-205 levels in a heterogeneous breast cancer cohort and demonstrated how downregulation was associated with longer DFS. miR-375 Wu and colleagues [57] applied deep sequencing methods on blood of patients with primary stage II or III breast cancer to indicate the miR-375 prognostic value and validated these findings in an independent cohort by using qRT-PCR. miR-375 is known to inhibit migration and invasion which is partially carried out by targeting JAK2 [103]. A differential expression exists between patients with and without metastatic relapse.
Higher circulating levels reflect more favorable clinical outcome in terms of pathologic complete response to NCT and absence of relapse [57].

miR-30a
The prognostic features of miR-30a were investigated in 221 patients with IDC [104], showing that miR-30a, which has been implicated in regulation of several biological processes [105][106][107], negatively regulates vimentin expression and overexpression suppresses migration and invasiveness [104]. Reduced expression is associated with unfavorable outcome (decreased RFS and DFS). The prognostic value of miR-30a was also examined by Zhang and colleagues [108] on a heterogeneous set of 96 patients with breast cancer. The authors attributed the tumor-suppressive nature of this miRNA to its ability to target MTDH, thereby suppressing breast tumor growth and metastasis. Decreased expression of miR-30a was associated with an unfavorable outcome in terms of metastasis development [108].
miR-342-5p miR-342-5p has a role in cell cycle progression and cell growth regulation. The latest research by Leivonen and colleagues [109] demonstrated how higher expression of miR-342-5p, which is an efficient negative regulator of the HER2 pathway, was significantly associated with better survival in two heterogeneous breast cancer cohorts.
miR-497 miR-497 is a tumor-suppressive miRNA and its biological role is found in the regulation of the nervous system. Several studies examined miR-497 and found its expression to be significantly decreased in breast cancer compared with normal breast and negatively correlated with adverse clinicopathological characteristics. Shen and colleagues [110] reported how elevated miR-497 expression rendered IDC patients with better prognosis. An independent study verified its prognostic role in a population-based breast cancer cohort, in which higher expression was correlated with a better 5-year DFS and OS [111].
MicroRNAs associated with negative prognosis miR-122 Wu and colleagues [57] identified circulating miR-122 in a heterogeneous breast cancer exploration and validation cohort, by sequencing and RT-PCR, respectively. They found miR-122 to be associated with outcome in terms of relapse and identified a distinct expression in patients undergoing metastatic relapse and those who remained relapse-free.
miR-27b-3p The miR-27 family is known to regulate cell cycle progression and survival by targeting the tumor-suppressive FOXO1 gene and is highly expressed in breast cancer [112]. In a recent study, Shen and colleagues [113] aimed to identify and validate a prognostic signature for patients with TNBC (n = 58) and found that lymph node status and miR-27b-3p were independent predictors of poor prognosis in terms of distant metastasis-free survival. This result was validated in a TNBC cohort (n = 41) [113].
miR-21 miR-21, which has its biological role in development, morphogenesis, and differentiation, is known to be overexpressed in breast cancer [114]. The significance of this miRNA as a prognostic factor was examined in two independent studies. Lee and colleagues [115] found significantly higher expressions in IDC compared with normal breast tissue, which positively associated with tumor size, stage, grade, and Ki-67 expression. A higher miR-21 expression also correlated with ER negativity and HER2 positivity. A lower OS could be noticed in patients with higher miR-21 expression levels [115]. In a more recent study, miR-21 was found to inversely correlate with DFS [102].
miR-210 Volinia and colleagues [75] used deep sequencing to determine miR-210 in patients with IDC. miR-210 showed association with time to metastasis and OS by uni-and multivariate analyses. Various breast cancer genes-such as BRCA1, E-cadherin, PARP1, and RB1with an antagonistic behavior to miR-210 were identified. More recently, systematic reviews and meta-analyses of previous clinical research were performed to recapitulate the significance of increased miR-210 in the prognosis of cancer. Summarizing nine published studies, the authors found that miR-210 could predict outcome, especially in patients with breast cancer, and that miR-210 overexpression predicts DFS and RFS [116,117].

miR-9
The functional significance of miR-9 is evidenced by its regulative role in neurogenesis, development, and apoptosis. Dysregulation of miR-9 influences proliferation or metastasis formation and has been reported in cancer [118]. In breast cancer, upregulation of miR-9 suppresses E-cadherin, leading to increased cell motility and invasiveness [119]. Zhou and colleagues [120] found miR-9 to discriminate between cases with and without local recurrence (LR), in which the latter group showed a significantly lower expression. In ER + cases, a lower 10-year LR-free survival was seen in patients with high miR-9 expression by using miRNA arrays [120].
miR-187 Using an in silico method in two independent breast cancer cohorts, Mulrane and colleagues [121] identified and validated miRNA-187 as being involved in breast cancer progression. miR-187 is independently associated with poor outcome in breast cancer (particularly, lymph node-positive samples) in terms of reduced breast cancer-specific survival.
miR-155 Several studies demonstrated the involvement of miR-155 in biological processes, such as cell survival, growth, migration, and invasion [23,25]. Song and colleagues [122] found a significantly higher miR-155 expression in formalin-fixed paraffin-embedded tumor compared with normal tissues. Multivariate analysis showed an inverse (yet insignificant) correlation with breast cancer outcome in terms of OS. Recently, Kong and colleagues [123] verified these findings and reported miR-155 to be frequently upregulated in various types of cancer where it plays pro-angiogenic, proliferative, and migratory roles. Moreover, they noticed miR-155 expression levels to be associated with poor prognosis in terms of OS [123].

MicroRNA biogenesis
MiRNA biogenesis (see Figure 2 for the canonical pathway of miRNA biogenesis) affects both the quantity and quality of miRNAs, providing a link between biogenesis and the diagnostic, predictive, and prognostic power of miRNAs.
MicroRNA biogenesis and the diagnostic potential of microRNAs A differential expression of miRNA-processing enzymes may result in an altered turnover of pri-and pre-miRNAs, which may influence the biology and etiology of cancer. We [68] and others [69] have seen that the expression of miRNA biogenesis genes varies between different molecular subtypes. Therefore, the hypothesis exists that molecular changes in miRNA-processing enzymes might be responsible for the subtype-specific variation of miRNA profiles. Blenkiron and colleagues [69] demonstrated how DICER1 and AGO2 showed a significantly lower and respectively higher expression in the more aggressive subtypes. Higher AGO2 and DROSHA levels and lower DICER1 levels were detected in ER-negative breast cancers.

MicroRNA biogenesis and the predictive and prognostic potential of microRNAs
Aside from affecting the quantity, the miRNA biogenesis cascade may also affect the quality and function of miRNAs. MiRNAs exert their function in collaboration with various proteins in the RISC complex, such as the processing enzymes AGO2, DICER1, and TARBP2. When overexpression of a certain miRNA results in a predictive or prognostic potential, this property could be negated by inadequate expression of AGO2, the catalytic subunit of RISC. Therefore, the prognostic or predictive capability of certain miRNAs may depend on the status of particular miRNA-processing enzymes, which may in fact explain the discrepancy found between several studies.
The role of miRNA-processing enzymes in breast cancer prognosis was also examined in different studies. For example, lower DICER1 levels are significantly associated with lower metastasis and DFS rates [124]. In terms of predictive potential, a loss of Drosha was independently associated with better response to chemotherapeutic and endocrine therapy [125]. In a case-control study, Leaderer and colleagues [126] reported how higher methylation levels in the XPO5 promoter region were associated with reduced risk of breast cancer. Finally, Sung and colleagues [127] reported seven single-nucleotide polymorphisms in miRNA biogenesis genes to be significantly associated with breast cancer survival.

Conclusions
Over recent years, several research groups have explored the regulatory potential of the miRNA class of small non-coding RNAs. Differences in expression of these miRNAs were used to explain the heterogeneity and disease pathology of several types of cancer. Aside from understanding the biochemical mechanisms these molecules have in gene regulation, we are beginning to understand the carcinogenic mechanisms and pathways these small molecules could influence. Due to their obvious association with carcinogenesis, and considering their promising diagnostic and therapeutic potential, miRNAs will steadily be investigated to discover their effect on a patient's disease.
Additionally, miRNAs represent a novel class of potential biomarkers for diagnosis, prediction of treatment, or prognosis, which are essential to improve patient management. In this review, we focused on miRNAs validated by more than one study or where results rely on preclinical as well as clinical evidence. However, it needs to be emphasized that the major challenge in identifying diagnostic, predictive, and prognostic miRNAs still resides in the need to firmly validate these findings in additional independent cohorts or by additional preclinical/clinical verification studies. Also, it is important to integrate multivariate testing since many individual identified miRNAs Figure 2 The microRNA biogenesis pathway. The canonical biogenesis pathway starts in the nucleus by cleavage of the pri-miRNA transcript by the Drosha-DGCR8 microprocessor complex [128]. After nuclear processing, the resulting precursor hairpin (pre-miRNA) is transported into the cytoplasm by Exportin-5 in complex with Ran-GTP [129]. In the cytoplasm, cleavage into an approximately 22-nucleotide duplex is achieved by the RNase III Dicer [130] and its interactors TRBP and PACT [131,132]. To form the active RNA-induced silencing complex (RISC) that performs gene silencing, the functional guide strand has to be separated from the passenger strand, which is degraded subsequently [133]. miRNA, microRNA. lose their significance when correcting for multiple testing. One should also take into account that some inconsistencies can be partly attributed to differences in subtype distribution or cells of origin. In this way, sufficient experimental evidence can be obtained in order to provide more power to the currently insufficiently strong miRNA markers so that they could be employed in clinical decision-making.

Additional file
Additional file 1: Table S1. Predictive microRNAs -microRNAs involved in response (sensitivity/resistance) to conventional breast cancer therapeutic strategies. Table S2A. Prognostic microRNAs: list of major positive prognostic microRNA signatures in breast cancer. Table S2B. Prognostic microRNAs: list of major negative prognostic microRNA signatures in breast cancer.