In recent decades, several studies have thoroughly investigated circulating miRNAs as candidates for discriminating between BC patients and healthy controls, and possibly contributing to BC diagnosis in the case of incertitude.

The miRNAs most frequently cited in the literature for their potential diagnostic value are miR-21-5p and miR-155-5p (oncogene-like), and miR-let-7a-5p and miR-34a-5p (tumor suppressor-like). Strong and consistent data from multiple studies have shown that circulating miR-21-5p levels were significantly higher in the serum and plasma of BC patients compared to healthy women, regardless of subtypes [27,28,29,30,31,32]. In addition, Rodriguez-Martinez et al. reported significantly higher miR-21-5p levels in advanced BC patients compared to early BC, suggesting that circulating miR-21-5p may be a promising candidate not only for early tumor detection but also as a marker of tumor burden [32]. Similarly, the analyses of circulating miR-155-5p levels showed highly consistent results across studies, indicating a marked increase in serum miRNA levels in BC patients compared to healthy controls [30,33,34]. In contrast, for the tumor suppressor-like miRNAs miR-let-7a-5p and miR-34a-5p inconsistent results have been reported among different studies. In fact, Marques et al. reported markedly lower levels of miR-let-7a-5p in plasma and serum samples from BC patients compared to healthy controls [35]. However, in other studies significantly higher levels of miR-let-7a-5p were found in blood or serum of BC patients compared to healthy individuals [27,36]. Similarly, circulating miR-34a-5p levels were significantly higher in BC women compared to healthy volunteers in several studies [34,37,38] but, in the case-control series reported by Freres et al., the authors found opposing results [39].

As part of diagnostic evaluation, many studies then analyzed the possible correlation among circulating miRNA levels and some of the key BC clinical-pathological features. The strongest evidence concerns tumor grade, tumor size (T), lymph node involvement, clinical stage, molecular subtypes, HR, and HER2 expression.

Several miRNAs have been associated with tumor grade in different studies. In particular, serum miR-125b-5p levels were significantly increased in high pathologic grade compared to low-grade tumors in a series of Luminal B patients [40]. Similarly, serum miR-155-5p, plasma miR-21-5p, and blood miR-195-5p levels were higher in G3 than in G2 tumors [33,41].

Serum levels of miR-21-3p, miR-10b-3p, miR-145-3p, and miR-let-7a-3p were shown to correlate directly with T stage: higher levels were associated with higher tumor size [42]. In addition, serum miR-21-5p [32] and miR-34a-5p [34] levels were significantly higher in T3-T4 stages compared to T1-T2 in BC patients. Similar results were observed by Heneghan et al., who reported a significant increase in the circulating levels of miR-195-5p in blood samples of BC patients with T3-T4 tumors compared to T1-T2 ones [36]. Moreover, Stevic et al. reported that the plasma levels of six miRNAs (-185-5p, -376a-3p, -382-5p, -410-3p, -433-3p, and -628-5p) were significantly associated with a higher tumor stage (i.e., T3-T4 versus T1-T2) in a cohort of HER2-positive BC patients [43].

Across different studies, several circulating miRNAs have been investigated in early BC patients in association with nodal involvement. Increased levels of circulating miR-210-3p [29] in plasma and miR-125b-5p [30] and miR-21-5p [31] in serum directly correlated with the positivity of locoregional nodes. However, several other miRNAs in plasma (-24-3p, -92a-3p, -143-3p, -146-5p, -185-5p, -193b-3p, and -484) [44] and serum (-155-5p [30,45], -182-5p [45], and -3200-3p [46]) have been found to be inversely related to nodal status, with lower levels found in patients with nodal involvement.

Multiple circulating miRNAs were also assessed for their possible relationship with clinical stage. Among them, miR-21-5p [31,45,47], miR-155-5p [30,33,45,47], miR-182-5p [45], miR-373-3p [28,48], miR-221-3p [47], miR-125b-5p [30,33,48], and miR-10b-5p [30] had significantly higher levels in both the serum and plasma of BC patients in advanced clinical stages compared to earlier stages (i.e., II versus III, or I-II versus III-IV).

Several circulating miRNAs were found to have different levels depending on the BC molecular subtype. Serum circulating miR-200c-3p levels were lower in triple-negative breast cancer (TNBC) patients compared to estrogen receptor (ER)- and progesterone receptor (PgR)-positive BC patients [49]. Conversely, Luminal-like tumors showed lower plasma miR-185-5p levels compared to TNBC [44]. Finally, Rodríguez-Martínez et al. reported definite level profiles of miRNA-222-3p in serum samples according to molecular subtype: it was decreased in Luminal A compared to basal-like and Luminal B tumors [32].

A large number of circulating miRNAs were investigated in relation to ER status. Indeed, high levels of circulating miR-221-3p in serum [50] and miR-185-5p [44] and miR-34a-5p [37] in plasma were associated with the negative expression of both ER and PgR.

Correlation among circulating miRNAs and ER was also observed for miR-195-5p (in blood samples) [41], miR-let-7a-5p [47], and miR-145-5p [48] (in plasma samples), with higher levels directly related to higher ER expression. In addition to the aforementioned results, circulating miR-221-3p, miR-185-5p, and miR-34a-5p, and serum miR-21-5p were also found to be inversely correlated with ER positivity (i.e., low circulating levels were related to higher ER expression) in a study conducted by Al-Khanbashi M and colleagues [46]. Concerning PgR expression, serum miR-222-3p [32] and miR-10b-5p [48] showed an inverse association with PgR positivity, with higher levels mostly found in PgR-negative BC.

Finally, regarding HER2 status, Zhang et al. identified three circulating miRNAs that were significantly associated with HER2 expression: serum levels of miR-375-3p, miR-718, and miR-4516 were lower in patients with HER2-negative tumors than in those with HER2-positive BC [40]. Moreover, higher plasmatic levels of miR-24-3p and miR-185-5p were associated with HER2-positive tumors [44]. Currently, there are no data on the correlation between circulating miRNA levels and low HER2 status (i.e., tumors expressing HER2 protein at the 1+ or 2+ immunohistochemistry level without HER2 gene amplification).

For the prediction of clinical and pathological outcomes in an NAC setting, a plethora of circulating miRNAs have been investigated.

Among them, miR-21-5p has emerged as an independent predictor of response in several studies. In fact, low levels of circulating miR-21-5p before NAC (based on paclitaxel and doxorubicin) have been associated with a higher likelihood of favorable response to NAC in HR-positive BC patients [44]. Similarly, in a multicentric prospective study by McGuire and colleagues, which assessed a predefined panel of circulating miRNAs (selected on their reported relevance in BC) as a measure of NAC response in all BC subtypes, the whole-blood miR-21-5p levels of responders (defined as patients who had a complete response or >90% reduction in primary T) were considerably lower than those of non-responders (defined as patients with <90% reduction in primary T) in the HR-positive subtype (p = 0.048). On the contrary, in TNBC and HER2-positive BC, the response to NAC was not associated with circulating levels of miR-21-5p [41]. Likewise, similar results were reported in a Ukrainian retrospective study, which included 182 patients with stage II–III HR-positive/HER2-negative BC undergoing NAC with polychemotherapy with fluorouracil, doxorubicin, and cyclophosphamide or doxorubicin/cyclophosphamide. Patients with chemo-sensitive tumors (evaluation of NAC response performed every two cycles by mammography) showed changing in serum baseline levels of miR-21-5p lower than two-fold, while those with resistant tumors had a change above three-fold [45]. In the same study, serum miR-205-5p levels increased more than four-fold in chemotherapy-sensitive patients with the HR-positive/HER2-negative subtype and decreased lower than 2.5-fold in patients with a poorer response [45]. In contrast, another observational study with a similar cohort of patients (68 luminal A stage II-III BC patients for the discovery group, and 56 patients for the validation one) revealed that serum miR-205-5p levels in patients undergoing epirubicin- and paclitaxel-based NAC were higher in the resistant group (defined as no or minor reduction (≤30%)) compared to the sensitive group (defined as a decrease of >30% or no residual invasive cancer; p < 0.05) [53].

Besides miR-21-5p and miR-205-5p, miR-375-3p has also been shown to be a promising predictive biomarker whose basal levels are associated with response. Zhang et al. prospectively investigated the role of several miRNAs in 37 luminal B BC patients undergoing NAC with taxane- and/or anthracycline-based regimens, plus trastuzumab for HER2-positive cases. In luminal B HER2-negative patients, relatively low baseline serum levels of miR-375-3p were found to be associated with pathological complete response (pCR) (p = 0.043) and “comprehensive response”, defined as partial or complete response in clinical evaluation and loss > 30% or no residual invasive cancer in pathological evaluation (p = 0.023) [40]. Similar results were observed in the luminal A subtype in the abovementioned Ukrainian study, where decreased baseline levels predicted sensitivity to NAC with fluorouracil, doxorubicin, and cyclophosphamide, or doxorubicin and cyclophosphamide [45]. On the contrary, in the study by Wu and colleagues, when applying de novo sequencing to identify circulating miRNAs associated with BC clinical outcome, lower levels of miR-375-3p significantly correlated with not achieving pCR in HER2-positive BC patients receiving doxorubicin/cyclophosphamide treatment followed by carboplatin and nab-paclitaxel plus trastuzumab [54].

Multiple studies are concordant on the possible association between circulating miR-125b-5p and NAC response. In fact, circulating baseline levels of miR-125b-5p were found to be higher in non-responders (defined as stable or progressive disease) compared to responders (defined as partial or complete response; p = 0.008) in serum samples of 56 BC patients with invasive ductal carcinoma undergoing four to six cycles of NAC with 5-fluorouracil, epirubucin, and cyclophosphamide [33]. Similarly, a Chinese study involving 118 patients diagnosed with stage II/III BC and undergoing four to six cycles of NAC with docetaxel, epirubicine, and cyclophosphamide showed that significantly higher levels of serum miR125b-5p were present in patients who had stable or progressive disease compared to those that had reached partial or complete response [30].

The identification of biomarkers that can be repeatedly tested through non-invasive approaches could provide the possibility of analyzing real-time information on disease evolution during treatment [55]. In this context, many studies have aimed to identify the predictive biomarkers of NAC response by monitoring miRNA dynamic changes on multiple blood samples collected during therapy.

Again, one of the most investigated circulating miRNAs is miR-21-5p. A prospective clinical trial by Davey MG et al. evaluated the possible role of circulating miRNAs in decision making for NAC [56]. Blood miRNA levels were measured at diagnosis (Timepoint 1, or T1), and after two cycles of NAC (T2) in a total of 120 patients (59 luminal A, 21 luminal B, 15 HER2-positive, 25 TNBC). In the overall cohort, no circulating miRNAs were associated with response to NAC, but decreased or increased miR-21-5p levels trended to significance as associated with treatment response in specific BC subtypes. A second analysis evaluated levels of serum miRNAs during NAC in 83 HER2-positive early BC patients treated with four to six cycles of taxane-carboplatin plus trastuzumab [48]. Serum samples were collected before treatment, at the end of the second cycle, and at the end of therapy. The results showed that serum miR-21-5p levels in clinical responders were significantly lower at the end of the second cycle and at the end of therapy compared to baseline level (p < 0.001 for both), while there was no significant difference in non-responders. Dynamic circulating miR-21-5p levels were also investigated by Liu B. et al. in 118 patients affected by early HER2-negative BC [30] receiving four to six cycles of NAC with an association of docetaxel, epirubicin, and cyclophosphamide (TEC regimen). Blood samples were collected before treatment, at the end of the second cycle, and at the end of NAC. In line with the abovementioned results, the mean miR-21-5p level was lower during and after NAC than at baseline in responders (p = 0.016 for both), but not in non-responders. These results suggested that a decreased level of serum miR-21-5p detected after the second NAC cycle (compared to baseline) could be able to predict responder patients.

Besides miR-21-5p, other circulating miRNAs have been investigated for their dynamic changes during NAC. The abovementioned prospective trial performed by Zhang Z. et al. evaluated circulating miRNAs in 37 luminal B early BC patients [40]. Blood samples were collected at baseline, and after two/four cycles of NAC. The circulating miR-210-3p levels during NAC were increased in non-responders, while the authors did not find a significant change in responders. In addition, significantly higher plasma miR-210-3p levels were observed in the non-pCR group than in the pCR group. Circulating levels of miR-210-3p associated with sensitivity to trastuzumab were evaluated in another trial involving 29 patients with HER2-positive early BC [29]. Patients received four cycles of taxanes followed by four cycles of anthracycline-based chemotherapy plus trastuzumab. Plasma samples were collected preoperatively and in the second postoperative week. The mean baseline level of plasma miR-210-3p was higher in samples from patients with residual disease than in the pCR group (p = 0.0359).

In the above-mentioned study by Liu B. et al. analyzing the dynamic predictive role of circulating miR-125b-5p [30], a significant association was found between miRNA level and NAC response; miR-125b-5p level was higher at all timepoints in non-responders than in responders; however, treatment did not induce statistically significant changes in miRNA levels in either group. Moreover, the analysis by Zhang Z. et al. reported that, in the luminal B/HER2-positive cohort, the levels of circulating miR-125b-5p remained relatively stable from the baseline through the first two/four cycles of NAC, in patients with both complete and partial response [40].

Recently, Todorova K. et al. performed RNA sequencing on plasma samples collected from 20 BC patients before and after the first cycle of NAC (combination of doxorubicin with cyclophosphamide). The authors showed an increased level of circulating miR-34a-5p after the first NAC dose in patients that did not achieve pCR [57]. Similarly, another analysis reported that the miR-34a-5p levels were significantly increased after two/four cycles of NAC (compared to baseline level) in luminal B responder patients, regardless of HER2 status [40]. A further study investigated serum miR-34a-5p levels during NAC in 86 HER2-negative BC patients [38]. All patients received six cycles of chemotherapy (anthracycline plus taxane-based regimen). Serum samples were collected at baseline, at the end of the second cycle, and at the end of NAC. The authors found that changes in miR-34a-5p levels during NAC were significantly associated with the chemotherapeutic responses. At the end of the second cycle and at the end of NAC, almost all responders had decreased serum miR-34a-5p levels compared to baseline (p < 0.001 for both). Finally, Zhu W. et al. evaluated the dynamics of circulating miR-34a-5p during NAC (i.e., epirubicin-paclitaxel regimen) [26]. In the HER2-positive and TNBC cohorts, plasma miR-34a-5p levels were significantly decreased in chemo-insensitive patients after the first two cycles of NAC (p = 0.027 and p = 0.006, respectively), while they remained stable throughout the course of treatment in chemo-sensitive patients (no statistically significant changes).

In an analysis by Davey G.M. et al. [56], the authors found that increased miR-let-7a-5p levels (from baseline to after the second cycle of NAC) identified patients who achieved partial or complete response in the luminal HER2-positive cohort (p = 0.049), whereas in the luminal cohort reduced miR-let-7a-5p levels predicted achieving pCR (p = 0.037). Circulating miR-let-7a-5p was also analyzed in the NEOCENT trial, a phase III translational study in which 63 patients (ER-rich) were randomized 1:1 to receive chemotherapy (anthracycline-based regimen, followed by docetaxel in the case of poor response after the first three cycles) or endocrine therapy (letrozole) [58]. Blood samples were collected at baseline, after 8 weeks, shortly before surgery, and 6-monthly for 2 years; miRNA markers were assessed from baseline to the end of treatment for both arms. An increase in circulating miR-let-7a-5p level was associated with objective radiological response in both arms, but it was statistically significant only in the chemotherapy arm (p = 0.008).

MiRNAs have been extensively investigated for their promising role as prognostic biomarkers. However, despite several analyses suggesting them as potential prognostic tools in BC, the clinical application of these findings has yet to be verified [60]. A recent meta-analysis by Zhang et al. [61] analyzed 39 miRNAs with a prognostic value from 23 studies and found that 26 miRNAs were associated with survival outcomes. Although the study did not distinguish between tissue and circulating miRNAs, they identified miR-125b-5p, miR-21-5p, and miR-7-5p as the most frequently investigated ones with significant results.

As for diagnosis and prediction, among prognostic miRNAs, circulating miR-21-5p is one of the most extensively studied with the most robust data [28,30,31,48,62]. There is consistent evidence that low circulating miR-21-5p levels are associated with better outcomes, while higher levels are associated with worse outcomes. A study by Muller et al. was the first to investigate the effects of NAC with trastuzumab and lapatinib on serum levels of circulating miR-21-5p, miR-210-3p, and miR-373-3p in 129 HER2-positive BC patients compared with a cohort of 19 healthy controls. One of the aims of the study was to evaluate whether specific miRNA levels were associated with prognosis. Of the miRNAs investigated, only increased levels of circulating miR-21-5p before (p = 0.0091) and after (p = 0.037) NAC were associated with a statistically significantly worse overall survival (OS) [28]. In a similar cohort of HER2-positive BC patients treated with NAC combined with trastuzumab, Liu et al. evaluated the association of circulating miR-21 levels with survival. The authors analyzed blood and serum samples from 83 HER2-positive BC patients during different phases of NAC (at baseline, after two cycles, and at the end of treatment). They demonstrated that changes in serum miR-21-5p levels were significantly associated with survival outcomes. In particular, patients in whom circulating miR-21-5p levels decreased from baseline to the end of the second cycle and to the end of NAC showed better OS and disease-free survival (DFS) than patients with increased levels of this miRNA [48]. Similarly, an earlier study on HER2-negative BC confirmed that among the miRNAs investigated, a decrease in serum miR-21-5p and miR-125b-5p levels during NAC were associated with better DFS [30]. MiR-21-5p has also been proven to be associated with survival outcomes in two other studies conducted on BC patients (any subtype) [31,62]. In a study by Wang et al., conducted on more than 300 patients, high serum miR-21-5p levels were found to be an independent poor prognostic factor for both recurrence (HR 2.9; 95% CI 1.420–8.325; p = 0.008) and DFS (HR 2.7; CI 1.038–7.273; p = 0.003). In addition, patients with high miRNA levels had shorter recurrence-free survival (RFS) and disease relapse-free survival (DRFS) than patients with lower levels [62]. Accordingly, in another study, elevated miR-21-5p levels in blood samples collected before and after NAC from 75 BC patients was found to be significantly associated with poor survival (p = 0.002) [31].

It has been shown that circulating miR-34a-5p levels were associated with better survival outcomes in two studies conducted in HER2-negative BC. Liu et al. analyzed the serum miRNA levels of 86 patients during different phases of NAC and showed that changes in miR-34a-5p expression during treatment were significantly associated with treatment response and DFS. Furthermore, decreased miR-34a-5p levels between the end of the second cycle and the end of NAC compared to baseline levels were associated with improved DFS (p < 0.001) [38]. In a more recent study, blood samples from 20 patients collected before and after the first cycle of NAC were evaluated to investigate whether circulating exosomal miRNAs could predict pCR. The authors showed that decreased levels of circulating exosomal miR-34a-5p were associated with better OS [57].

Checkhun et al. analyzed the expression levels of circulating miRNAs in serum samples from 182 patients with luminal A and B BC undergoing NAC. The authors found that low serum miR-375-3p levels were associated with a lower rate of 3-year DRFS in luminal B patients. In contrast, higher levels of circulating miR-182-5p correlated with a lower 3-year DRFS rate in luminal A BC [45].

Regarding the circulating miR-200 family in BC, a study evaluating serum miR-222-3p levels in a cohort of 65 HER2-positive patients receiving anti-HER2 NAC showed that low serum miRNA levels were associated with better DFS (p = 0.029) and OS (p = 0.0037). Moreover, the study aimed to assess the association between circulating miRNA levels and trastuzumab-related adverse events. For the first time, serum miR-222-3p levels were found to be an independent protective factor for cardiotoxicity (p < 0.05) and anemia (p = 0.013), although the mechanism of action remains to be elucidated [59].

Finally, in a study conducted by Al-Khanbashi and colleagues, tissue and serum samples were collected from 27 BC patients undergoing NAC at four different timepoints (baseline, after the first and fourth cycle of doxorubicin/cyclophosphamide treatment, after the fourth cycle of docetaxel treatment) to assess the correlation between miRNA expression and different endpoints, including survival outcomes. The authors demonstrated that patients with high serum miR-451-5p levels at diagnosis were associated with better DFS (p = 0.046) [46].