MicroRNAs as diagnostic biomarkers and predictors of antidepressant response in major depressive disorder: a systematic review

2.1. Literature search and study selection

From March 2022 through March 2023, we searched the PubMed/MEDLINE, Cochrane, PsycINFO, Embase, and LILACS databases using the following terms: (micrornas OR microrna OR mirnas OR micro rna) AND (antidepressive agents OR antidepressive drugs OR antidepressive agent OR antidepressive OR antidepressants OR antidepressant OR antidepressant drugs OR antidepressant OR antidepressant drug OR antidepressive treatment). No limits to publication years were added. The articles were required to be published in English and to involve humans, animals, or cell cultures. The titles and abstracts of the retrieved items were independently screened by four reviewers (LG, BC, DH, LS), and the two main reviewers (LG, BC) then independently assessed the full texts of the selected studies.

2.2. Eligibility criteria

The strategic question that guided the manual search and the inclusion and exclusion criteria was as follows: “What is the levels of miRNAs expression before and after AD treatment?” The inclusion criteria were (1) original studies only; (2) human subjects aged 18 or over with a clinical diagnosis of MDD and undergoing treatment with ADs; and (3) animal models and cell cultures with miRNA assessments after treatment with ADs. For each study, we prepared a standardized spreadsheet that presented the following information: i) authors and publication year, ii) publication journal, iii) population studied, iv) treatment conducted, v) instrument used for diagnosis and stratification of MDD, vi) study design, vii) miRNA extraction site, viii) follow-up time, and ix) miRNA expression before and after intervention. We excluded studies that evaluated herbal medicines, non-pharmacological therapies, or epigenetic mechanisms other than miRNAs.

2.3. Risk of bias assessment

The quality of the study was evaluated by two reviewers (DF, DH) using the strengthening the reporting of observational studies in epidemiology (STROBE) [19] and systematic review center for laboratory animal experimentation (SYRCLE) [20] tools for the human and animal studies, respectively (Tables 1 and 2). We did not assess the quality of the cell culture studies because no standardized tool exists for this purpose. Discrepancies between the two reviewers were resolved through discussions until a consensus was reached.

3.1. Human beings

We found 28 studies comparing the miRNA expressions of people undergoing AD treatment (Table 3). The miRNA levels of patients were assessed following various therapeutic regimens. In eleven studies, participants were treated exclusively with selective serotonin reuptake inhibitors (SSRIs), with escitalopram/citalopram drugs used in eight of these.

In 2013, Bocchio-Chiavetto et al. [21] determined that after 12 weeks of treatment with escitalopram, 28 miRNAs became overexpressed, with fold changes ranging from 2.07 to 4.68, whereas only two miRNAs were highly downregulated (miR-34c-5p and miR-770-5p), with fold decreases of 2.03 × 10-2 and 1.69 × 10-2, respectively [21]. Homorogan et al. [22] found that miR-494 was differentially expressed in exosome-depleted plasma after 12 weeks of treatment with escitalopram. Additionally, in a study comparing MDD patients with healthy controls, the MDD group presented lower miR-135a levels in the blood samples taken, and after 12 weeks of escitalopram treatment or cognitive behavioral therapy (CBT), patients undergoing CBT had an increase in miR-135a expression [23]. These effects were not significant in the tests with miR-16 [23]. Furthermore, the brain miR-135a and miR-16 levels of suicide victims were significantly lower than those of the controls, specifically in the dorsal raphe and raphe magnus subnuclei [23]. Enatescu et al. [24] assessed the peripheral expression changes of 222 miRNAs in 5 MDD patients treated with escitalopram for 12 weeks and found that 23 were upregulated and that many of these presented a greater than 2.5 times increases, while 17 of the miRNAs were downregulated [24].

Saeedi et al. [25] found that the combination of changes in miR-21-5p, miR-30d-5p, and miR-486-5p with escitalopram treatment was associated with treatment response (AUC: 0.8254, specificity 84%, sensitivity 76%, P < 0.01). This was the only study that analyzed the specificity and sensitivity of a cluster of microRNAs and highlighted them as good biomarkers of the response to AD treatment with escitalopram. Moreover, when comparing the miR-132 and miR-124 plasma changes among MDD patients that were drug naïve, those that used citalopram, and a control group of individuals without depression, those authors found that miR-132 levels were 2.4 times higher in the drug naïve group when compared to the control, while MDD patients undergoing citalopram had similar miR-132 levels to those of the control group [26]. However, both MDD groups showed higher miR-124 plasma levels than the control group (1.8-fold and 4-fold in the drug naïve and citalopram-treated groups, respectively) [26]. After two months of treatment with citalopram, a decreasing trend in plasma miR-132 and an increase in miR-124 was detected in the 14 drug naïve patients tested [26]. Another study indicated that miR-155 levels were downregulated after treatment with citalopram [27], whereas both cellular and serum levels of miR-155 were found to be significantly upregulated among MDD patients compared to those of healthy controls [27]. Kuang et al. [28] found decreased levels of miR-451, and higher miR-34a-5p and miR-221-3p levels in depressed patients pre- and post-paroxetine treatment compared to the healthy control group. After treatment with paroxetine, miR-451a levels increased, although both miR-34a-5p and miR-221-3p levels decreased significantly when compared to the basal levels [28].

Participants treated solely with SSRIs presented increased levels of miR-16, while patients treated with either SSRIs or serotonin and norepinephrine reuptake inhibitors (SNRI) exhibited increased expressions of miR-183 and miR-212 [29]. Low baseline levels of miR1202 were associated with improved therapeutic responses to SSRIs and SNRIs; MDD patients who responded to AD treatment had lower basal levels of miR-1202, and these levels increased after the drug was administered [30]. Moreover, the prediction analyses of nonresponse found a sensitivity of 91.7%, a specificity of 67.7%, and an AUC=0.812 [30]. Among patients treated with an SSRI (paroxetine or sertraline), a significant correlation was identified two weeks after treatment initiation between 228 miRNAs (especially miR-483-5p and miR-3151-5p) and depression severity among patients treated with mirtazapine, with the strongest association found with miR-483-3p [31].

Three studies assessed miRNA levels in patients treated with duloxetine, venlafaxine, or ketamine. For patients treated with duloxetine, baseline levels of hsa-miR-23a-3p were associated with a remission of depressive symptoms [32]. Belzeaux et al. [11] reported elevated baseline levels of miR-135a-5p and miR-16 and an improved treatment response in patients administered citalopram. Patients whose depressive symptoms abated had nominally significant decreases in miR-1202 and miR-425-3p levels [33]. Among the patients who received ketamine, there were no differences in miRNA expression when compared to the control group [34]. Other studies evaluated patients who received different ADs. Belzeaux et al. [35] found that after the treatment of patients experiencing a major depressive episode, miR-107, miR-342-5p, and let-7b levels were upregulated, while miR-200c expression was downregulated [35]. Depressed patients showed significantly higher expressions of miR-26b, miR-4743, miR-4498, miR-4485, and miR-1972 before treatment [36].

Various studies have investigated the basal characteristics of patients in the search for treatment response predictors and the untreated MDD miRNA profile. Peripheral blood mononuclear cells (PBMCs) showed elevated levels of miR124 in patients with depression, and a reduction in these levels was associated with improved treatment responses [37]. After adjustments for age, sex, smoking status, and body mass index (BMI), the levels of let-7e, miR-21-5p, miR-146a, and miR-155 were significantly lower in PBMCs isolated from baseline MDD patients [38], while let-7e and miR-146a were negatively correlated with the HAMD-17 score after treatment, and miR-155 was positively correlated with this value [38]. Qi et al. [39] used machine-learning analyses of miRNA to reveal high hsa-miR-5701 levels among responders in one cluster and high hsa-let-7b-3p and hsa-miR-130b-3p levels in a second cluster [39]. Reduced levels of miR-146b-5p, miR-24-3p, and miR-425 were associated with better responses to treatment in both human and animal models [12]. Hung et al. [40] showed that patients who achieved remission after treatment with the different antidepressants had lower basal levels of let-7e, miR-21-5p, miR-145, miR-146a, and exosomal miR-155 compared to those of healthy controls, which was followed by a notable increase in these levels post-treatment. Although most studies showed changes in miRNA levels after treatment, Liu et al. [41] found reduced baseline levels of miR-324b and mir-10a in depressed patients when compared to healthy controls that were unchanged after 8 weeks of treatment with classic antidepressants. The sertraline treatment reduced the high miR-4485 levels shown in MDD patients that were negatively correlated with HAMD scores [42].

Two recently published studies have identified decreased levels of miR-16 in patients diagnosed with MDD, with one study demonstrating an increase in these levels following treatment with sertraline or escitalopram (Ahmadimanesh et al., 2023 [43]; Wang et al., 2023 [44]). Sertraline, but not escitalopram, was found to significantly reduce the levels of miR-132 and miR-124, both of which were elevated in MDD patients (Ahmadimanesh et al., 2023 [43]). Furthermore, according to Wang et al., 2023 [44], it was observed that miR-16-2 exhibited a negative correlation with both HAMD-17 and HAMA-14 scores (AUC = 0.806, 95% CI: 0.721–0.891) (Wang et al., 2023[44]). Patients displaying higher expression levels of miR-16-2 were more inclined to experience a greater reduction in HAMD-17 scores during short-term follow-up compared to the low-expression group (Wang et al., 2023[44]).

The remaining two studies discovered decreased levels of miR-1237-5p in patients who experienced symptom remission with mirtazapine (AUC = 0.85, 95% confidence interval = 0.72–0.97), along with elevated basal expression of hsa-miR-6068 (Funatsuki et al., 2023 [45]; Ogata et al., 2023[46]). Moreover, one of these studies noted a decrease in basal levels of hsa-miR-4707-3p in remitters to SSRIs (Ogata et al., 2023[46]).

3.2. Animal models

We found fourteen studies comparing the miRNA expressions of rodents undergoing AD treatment (Table 4). Three SSRI studies reported diverse miRNA profiles and variable treatment response patterns among the animal samples. Patricio et al. [47] investigated changes in miR-409-5p and miR-411-5p levels in brain regions linked to depression in rats undergoing chronic mild stress (CMS). The rats received treatment with either fluoxetine or a saline solution [47]. The stress-exposed animals presented significantly increased levels of miR-409-5p in the nucleus accumbens that were not reversed with fluoxetine treatment [47]. In the dorsal dentate gyrus, chronic fluoxetine use resulted in increased levels of miR-411-5p; therefore, miR-411-5p levels were significantly decreased in the blood of fluoxetine-treated rats [47]. The genetic deletion of miR-34s impairs the effect of chronic fluoxetine on the behavioral responses of rats to an acute threat [48]. Chronic fluoxetine treatment increased the levels of miR-34a in WT mice, but this was not observed with acute administration of this compound [48].

Another study investigated the changes in the expression of miR-326 and miR-9 [49]. Adult rats exposed to chronic unpredictable stress (CUS) and maternal deprivation had lower levels of miR-9 in the nucleus accumbens and striatum, and these levels did not rise after escitalopram treatment [49]. In the nucleus accumbens, the altered expression of miR-9 was negatively correlated with an overexpression of the dopamine receptor D2 (DRD2) protein, which appears to be involved in chronic stress and early-life adversity [49]. The expression of miR-326 was increased in the nucleus accumbens and decreased in the striatum of rats exposed to CUS [49]. In addition, animals with maternal deprivation showed increased levels of miR-326 in the striatum [49]. In contrast to the miR-9 findings, miR-326 levels returned to normal after escitalopram treatment [49]. In another study, escitalopram treatment reversed miRNA expression alterations in rats after chronic unpredictable mild stress (CUMS) [50]. Although those authors found 18 miRNA alterations, they suggested that miR-10b-5p and miR-214-3p had important contributions to the miRNA network of the nucleus accumbens since the expression of these miRNAs differed significantly between the stressed group and the control, and between the treated group and the control [50].

The miRNAs related to serotoninergic networks were influenced by AD treatments in some studies. A potential regulator of serotoninergic activity, miR-135, was increased after AD treatment [23]. High levels of miR-135a appear to diminish depression-like behaviors after stress exposure, whereas its absence reduces the response to ADs [23]. A study using transgenic mice suggested that miR-26a-2 levels are involved in the serotoninergic network, given that ADs upregulate the expression of this miRNA in the dorsal raphe nucleus [51].

The influence of ketamine on miRNA expression was evaluated, and one study reported variable changes in the expression of miRNA serotonergic (5HT)-2C-receptor (5HTR2C) clusters in mouse hippocampi [52]. Following 24 hours of ketamine infusion, miR-764-5p, miR-1912-3p, miR-1264-3p, miR-1298-5p, and miR-448-3p were upregulated [52], with miR-448-3p showing the greatest increase and poorest AD response when it was inhibited. [52]. These changes were not replicated in the prefrontal cortex [52]. The same miRNA changes were not observed in mice receiving fluoxetine, which suggests that these are part of a ketamine response and not just a typical AD response [52]. In addition, miR-29b-3b appears to be involved in the AD effect of ketamine. A study found that this miRNA level was reduced in the prefrontal cortex of rats after CUMS; however, after ketamine treatment, the miR-29b-3b levels were reestablished [53].

Another study found that the administration of a single dose of ketamine in CMS rats restored the original miR-9-5p levels, and this increase was correlated with higher sucrose preference/less anhedonic phenotype [54]. In addition, Huang et al. [55] explored the antidepressant effect of ketamine in an animal model of depression and observed a reduction in miR-98-5p levels in the prefrontal cortex and hippocampus after treatment.

In a study with mice undergoing CUMS, duloxetine was found to enhance the expression of miR-18a in the frontal lobe, as well as upregulate miR-132 and miR-18a and downregulate miR-134 and miR-124a in the hippocampus [56].

Higuchi et al. [57] found lower levels of miR-124 in the hippocampus of mice subjected to chronic stress. The upregulation of miR-124 in the hippocampus increased resilience in mice, while its downregulation enhanced depression-like behaviors [57]. In other hand, Zeng et al., 2023 [58] found increased level of miR-124 in the hippocampus of mice with symptoms like depressive and the fluoxetine administration reduced these levels. Guan et al., 2023 [59] investigou os níveis do miR-miR-204-5p em ratos com fenótipo tipo depressivo e possíveis alterações causadas pelo tratamento com a velanfaxina. The group found higher levels of miR-204-5p in depression and the treatment.reduced this levels.

3.3. Cell cultures

We included four cell culture studies targeting different miRNAs and a variety of sample cultures (Table 5). Three of these used paroxetine and one used duloxetine.

The exposure of human lymphoblastoid cell lines to paroxetine for three days increased the expression of miR-212, miR-132, miR-30b, let-7b, and let-7c [60], and its incubation for 21 days decreased the expression of miR-221 and miR-22.55. The first study identified a correlation that mainly concerned the miR-151-3p basal levels, which were higher in cells that were sensitive to the paroxetine treatment [60]. Another study by the same group primarily showed a downregulation of miR-221 and miR-222 after treatment [61]. The third study using paroxetine analyzed glioblastoma-astrocytoma cell line U87 and found a significant upregulation of miR-30a levels beginning 6 hours after treatment, which lasted for 12 hours after treatment [62]. The final study used duloxetine as the treatment, and this was previously discussed in the human studies section of this article. Those authors confirmed their previous findings on human and rodent studies in an in vitro analysis using human neural progenitor cells, with reduced levels of miR-146a-5p, miR-146b-5p, miR-24-3p, and miR-425-3p in the cell culture [12].

4.1. Limitations

The limitations of our study include the diversity of the populations studied (humans, animal models, and cell cultures), wide range of ADs used, and large number of miRNAs identified. Although a wide variety of miRNAs exist, there is little replication of studies that investigate the same molecules from different perspectives. Most studies are limited to evaluating the change of miRNAs in AD treatments, without assessing or describing the related action mechanisms. These aspects complicate meta-analyses; therefore, a discussion and understanding of the basic aspects of the pathophysiology of depression are limited. Although the literature has established a relationship between circulating miRNAs and MDD, it is important to understand the targets of these miRNAs using in vitro, in vivo, and in silico models.

The studies included in this review assessed specific miRNAs that were associated with MDD and AD response. They provide evidence that specific miRNAs can act as biomarkers; however, the clinical utility of miRNAs as biomarkers of MDD and responses to ADs needs to be further investigated.