State of the Art of microRNAs Signatures as Biomarkers and Therapeutic Targets in Parkinson’s and Alzheimer’s Diseases: A Systematic Review and Meta-Analysis

Identifying target microRNAs (miRNAs) might serve as a basis for developing advanced therapies for Parkinson’s disease (PD) and Alzheimer’s disease. This review aims to identify the main therapeutic targets of miRNAs that can potentially act in Parkinson’s and Alzheimer’s diseases. The publication research was conducted from May 2021 to March 2022, selected from Scopus, PubMed, Embase, OVID, Science Direct, LILACS, and EBSCO. A total of 25 studies were selected from 1549 studies evaluated. The total number of miRNAs as therapeutic targets evidenced was 90 for AD and 54 for PD. An average detection accuracy of above 84% for the miRNAs was observed in the selected studies of AD and PD. The major signatures were miR-26b-5p, miR-615-3p, miR-4722-5p, miR23a-3p, and miR-27b-3p for AD and miR-374a-5p for PD. Six miRNAs of intersection were found between AD and PD. This article identified the main microRNAs as selective biomarkers for diagnosing PD and AD and therapeutic targets through a systematic review and meta-analysis. This article can act as a microRNA guideline for laboratory research and pharmaceutical industries for treating Alzheimer’s and Parkinson’s diseases and offers the opportunity to evaluate therapeutic interventions earlier in the disease process.


Introduction
In the scenario of neurodegenerative diseases, Alzheimer's disease (AD) occupies the first position of the most incident neurodegenerative disease, and in second place is Parkinson's disease (PD), impacting 1% of the population over 60 years of age [1]. Patients with PD usually have non-motor symptoms, including autonomic nervous system disorders such as constipation, bladder dysfunction, orthostatic hypotension, impaired sleep, and smell, in addition to motor symptoms such as resting tremor, postural instability, gait disturbances, rigidity, and bradykinesia [1]. Otherwise, at the molecular level, the abnormal accumulation of the α-synuclein protein (α-Syn) is related to the degeneration of dopaminergic neurons, with consequent dopamine deficiency [2]. In this sense, the accumulation of α-Syn, the formation of Lewy bodies and Lewy neurites, and their mutations and multiplication are linked to hereditary PD, according to Braak's hypothesis [3].
Institutional Review Board of PROSPERO International of systematic reviews (protocol code CRD354228, 17 August 2022). Table 1 shows the main variables that were addressed in the present study, according to the designation of the PICOS literary search strategy (Participants; Intervention; Control; Outcomes, and Study Design).

Instruments and Professionals Used for Study Eligibility
The studies were rigorously chosen following the search strategy in Table 1, presented scientific quality according to the GRADE classification [38], and did not present a risk of significant bias, that is, they did not compromise the safety of the data results, according to the Cochrane instrument [39].
For the selection and enrollment of the studies, two independent reviewers performed the research and study selection. Data extraction was performed by reviewer one and was thoroughly reviewed by reviewer two. A third investigator decided on some conflicting points for the final selection of the articles. Only studies reported in English were evaluated.

Eligibility Criteria, Study Quality, and Risk of Bias
According to the recommendations of GRADE [38], the quality of scientific evidence in the studies addressed was classified as high, moderate, low, or very low, according to the risk of evidence bias, sample size, clarity of comparisons, precision, and consistency in the effects of the analyses. High-quality evidence was assigned through seven criteria: (1) In vitro controlled randomized clinical trials (human biological samples); (2) Sample size greater than 15 biological samples; (3) Studies that showed an accuracy (%) of quantitative polymerase chain reaction (qPCR) measurements above 50%; (4) Studies that showed Alzheimer's and Parkinson's diseases with a genetic cause and not by transitory or epigenetic effects; (5) Studies that were controlled by biological samples from patients with mild cognitive impairment (MCI), frontotemporal lobar degeneration, DLB (dementia with Lewy bodies), multiple system atrophy, and Progressive Supranuclear palsy; (6) Studies with statistically well-designed results; (7) Studies that were published in indexed journals and had a significant impact factor. The Cochrane Instrument [39] was adopted to assess the risk of bias in the selected studies, using the Cohen Test to calculate the effect size (Effect Size) versus the inverse of the Standard Error (precision or sample size) to determine the Risk of Bias of the studies using the Funnel Plot.

Data Sources, Research Strategy, and Study Publication Date
The search strategies for the present study were based on the keywords of the medical subject headings (Mesh Terms): Parkinson's disease; Alzheimer's disease; Biomarkers; Therapeutic target; Diagnosis; Exosomes; MicroRNA. Search filters designated as clinical studies and clinical studies with biological samples were used. The publication search was developed based on Scopus, PubMed, Embase, OVID, Science Direct, LILACS, and EBSCO. In addition, a combination of the keywords with the Booleans "OR" and "AND" and the "NOT" operator were used to target scientific articles of interest. The title and abstracts were examined under all conditions. Table 2 presents an example of the search structure in PubMed. The same search strategy was used in the other databases.

Statistical Analysis-Meta-Analysis
The statistical programs Minitab 18 ® (version 18, Minitab, LLC, State College, PA, USA) and OriginPro ® 9 (DPR Group, Inc., Northampton, MA, USA) were used. Descriptive statistical analysis was performed for numerical variables, with the mean values, standard deviation, confidence interval (CI), and percentage. The Anderson-Darling (AD) normality test was performed for non-binary numerical variables, adopting p > 0.10 as normal (standard). The Cohen test was performed to calculate the effect size (Effect Size). The inverse of the standard error (precision or sample size) was established to determine the risk of bias in the studies using the Funnel Plot. The Heterogeneity Test (Chi-Square Test ≥ X 2 ) of the results between the studies was also determined, with p < 0.05 and with no statistically significant difference, in the 95% CI, adopting low association codes ≤ 25%, medium association = 25% < X < 50%, and high association ≥ 50%. The One-Way test (ANOVA) was performed between the values of the means of identification accuracy of the microRNAs, adopting the α level lower than 0.05, with a statistically significant difference for the 95% CI. To know the chances of a particular microRNA being identified more than once, the Nominal Logistic Regression analysis test was carried out, adopting a referential group with the Odds Ratio (OR) calculation to know the probability ratio between the analyzed groups, with 95% CI.

Results
A total of 25 studies (11 studies of Parkinson's disease (PD) only, 12 studies of Alzheimer's disease (AD) only, and two studies that presented both AD and PD in the same work) were selected from a total of 1549 evaluated studies (581 (PD) and 968 (AD)), comprising a total of 2160 human participants, a moderate to a high quality of scientific evidence, and an average degree of confidence and a recommendation of 80%, according to the GRADE classification (Supplementary Figure S1). In addition, it was observed that the analyzed studies showed homogeneity in the results in terms of accuracy in identifying samples of AD and PD miRNAs, showing 98.95% (X 2 ). Table 3 shows the results of the Detection Rate (Accuracy (%) or accuracy of miRNA identification by qPCR in each selected study). Through the correlation between the test and control groups in each study, the Chi-Square method (X 2 ) test showed that all correlations presented a statistically significant difference, with p > 0.05 in the 95% CI, for both AD and PD studies. Table 3 also presents the results of the effect size (Cohen's Test) and the 1/standard deviation (sample size) to determine the risk of bias in the studies addressed in this work. Figure 1 presents the results of the risk of bias of the studies through the Funnel Plot, showing the calculation of the Effect Size (magnitude of the difference) using the Cohen Test (d). This graph presented a symmetrical behavior, not suggesting a significant risk of bias, both among studies with a small sample size (lower precision, with a total of eight (8) studies), which are shown at the base of the graph (red balls), and studies with a high sample size, with a total of 17 studies, which are presented in the upper region of Figure 1.    [41]) presented results of the quantification of miRNAs for both AD and PD. As evidenced by therapeutic targets, the total of up-regulated and downregulated miRNAs was 90 for AD and 54 for PD, obtained mainly from CSF, serum, and plasma. Most studies had two types of controls: a control composed of healthy participants, and one composed of participants with mild cognitive impairment (MCI), fronto-   Table 4 summarizes the main general findings of each study addressed in this work. Of the 25 studies selected to compose the meta-analysis, only two studies (Burgos et al., 2014 [40] and Nie et al., 2020 [41]) presented results of the quantification of miRNAs for both AD and PD. As evidenced by therapeutic targets, the total of up-regulated and downregulated miRNAs was 90 for AD and 54 for PD, obtained mainly from CSF, serum, and plasma. Most studies had two types of controls: a control composed of healthy participants, and one composed of participants with mild cognitive impairment (MCI), frontotemporal lobar degeneration, DLB (dementia with Lewy bodies), multiple system atrophy, and paralysis progressive supranuclear.   Figure 2 shows the number of deregulated miRNAs that were identified in both AD and PD studies. Six miRNAs of intersection were found between AD and PD (miR-197-3p, Mir-576-5p, miR-1468-5p, miR-375, miR-let-7e-5p and miR-122-3p).  Figure 2 shows the number of deregulated miRNAs that were identified in both AD and PD studies. Six miRNAs of intersection were found between AD and PD (miR-197-3p, Mir-576-5p, miR-1468-5p, miR-375, miR-let-7e-5p and miR-122-3p). Through the Forest Plot graph presented in Table 5 and 6 relating to AD, the values distribution of each study's means and the standard deviation of accuracy (%) concerned the total mean of 84.37 ± 7.94%, in the confidence interval of 95%. Through this, eight studies were identified with accuracy values (%) equal to or above the total average. These eight studies are identified by reference numbers 2, 3, 4, 5, 8, 11, 12, and 14.
These eight studies were selected to determine their respective types of miRNAs. They presented the highest accuracies (%) in identifying and quantifying the miRNAs, with greater scientific credibility as biomarkers and therapeutic targets in identifying AD, either in the up-or downregulation.
In addition, Tukey's statistical analysis (One-Way ANOVA) showed that there was no statistically significant difference between the studies with the highest accuracy (%), with p > 0.05 in the 95% CI. The study groups presented these results with the same letter, as shown in Table 6. Through the Forest Plot graph presented in Tables 5 and 6 relating to AD, the values distribution of each study's means and the standard deviation of accuracy (%) concerned the total mean of 84.37 ± 7.94%, in the confidence interval of 95%. Through this, eight studies were identified with accuracy values (%) equal to or above the total average. These eight studies are identified by reference numbers [2][3][4][5]8,11,12,14]. Table 5. Results of the statistical analysis of the accuracy (%) of identification and quantification by qPCR of miRNAs concerning Alzheimer's disease (AD), with p > 0.05 and no statistically significant difference, at 95% CI.  These eight studies were selected to determine their respective types of miRNAs. They presented the highest accuracies (%) in identifying and quantifying the miRNAs, with greater scientific credibility as biomarkers and therapeutic targets in identifying AD, either in the up-or downregulation.

Studies
In addition, Tukey's statistical analysis (One-Way ANOVA) showed that there was no statistically significant difference between the studies with the highest accuracy (%), with p > 0.05 in the 95% CI. The study groups presented these results with the same letter, as shown in Table 6. Table 5 represents the statistical analysis results of the accuracy (%) of identification and quantification by qPCR of miRNAs concerning AD. Fourteen studies were listed, showing each study's mean and standard deviation of accuracy (%), with a total mean of 84.37 ± 7.94%.
Through the Forest Plot graph presented in Table 7 relating to PD, each study's mean values and standard deviation of accuracy (%) concerning the total mean value of 84.32 ± 7.15% (CI 95%) were distributed. Thus, seven studies were identified with accuracy values (%) above the total average. These seven results are demonstrated by the studies with reference numbers [2,16,18,[21][22][23]25].
These seven studies were selected to determine their respective types of miRNAs. They present the highest accuracy (%) in identifying and quantifying the most scientifically credible miRNAs as biomarkers and therapeutic targets in identifying PD in up-and downregulation. Table 7 represents the statistical analysis results of the accuracy (%) of identification and quantification by qPCR of miRNAs concerning PD. A total of thirteen studies were listed, showing each study's mean and standard deviation of accuracy (%), with a total mean of 84.32 ± 7.15%. Table 7. Results of the statistical analysis of the accuracy (%) of identification and quantification by qPCR of miRNAs concerning PD, with p > 0.05 and no statistically significant difference, at CI 95%. These seven studies were selected to determine their respective types of miRNAs. They present the highest accuracy (%) in identifying and quantifying the most scientifically credible miRNAs as biomarkers and therapeutic targets in identifying PD in up-and downregulation. Table 7. Results of the statistical analysis of the accuracy (%) of identification and quantification by qPCR of miRNAs concerning PD, with p > 0.05 and no statistically significant difference, at CI 95%.  Table 7 represents the statistical analysis results of the accuracy (%) of identification and quantification by qPCR of miRNAs concerning PD. A total of thirteen studies were listed, showing each study's mean and standard deviation of accuracy (%), with a total mean of 84.32 ± 7.15%.

Studies
After identifying the most scientifically reliable miRNAs of the selected studies through the accuracy (%) or precision analysis, as shown in Tables 5 and 7, the main miR-NAs for AD and PD were listed in up-and downregulation, respectively, as shown in Figure 3.
Additionally, there was no significant difference between studies with higher accuracy (%), according to Tukey's analysis, with p > 0.05 in the 95% CI. The study groups presented these results with the same letter, as shown in Table 8. After identifying the most scientifically reliable miRNAs of the selected studies through the accuracy (%) or precision analysis, as shown in Tables 5 and 7, the main miRNAs for AD and PD were listed in up-and downregulation, respectively, as shown in Figure 3.

AD/PD miRNAs Odds Ratio (OR)/ p-Value (95% CI)
AD miR-26b-5p miR-615-3p miR-4722-5p miR23a-3p miR-27b-3p OR = 2.55 (1.023-3.432); p = 0.004 < 0.05 PD miR-374a-5p OR = 2.16 (0.087-3.567); p = 0.0035 < 0.05 Based on the results presented in Table 9, a search was conducted to determine which of these miRNAs are present in the groups of the main AD and PD miRNAs selected by the accuracy criterion (%) shown in Figure 3. The results showed that all miRNAs (AD and PD) that had the highest chances of being identified by qPCR (Table 9) were included in the groups of the main miRNAs of high accuracy (%), except for miR-27b-3p, belonging to the AD group in Table 9, as shown in Figure 4. and PD) that had the highest chances of being identified by qPCR (Table 9) were included in the groups of the main miRNAs of high accuracy (%), except for miR-27b-3p, belonging to the AD group in Table 9, as shown in Figure 4.  Table 7 in the respective groups of Figure 3. Alzheimer's disease (AD); Parkinson's disease (PD).

Discussion
Based on the objective of the present study, it was evidenced that the majority of the twenty-five studies of AD and PD presented a mean accuracy in identifying miRNAs by qPCR above 84%, with moderate to strong scientific evidence. These showed greater scientific credibility in the findings of each study, contributing in a tangible way to identifying the main miRNAs as selective biomarkers for the diagnosis of these diseases, as well as therapeutic targets in gene, cellular, and pharmacological treatment.
The present study's results do not show a risk of bias, both in studies with large and small sample sizes. In addition, two studies (Burgos et al., 2014 [40] and Nie et al., 2020 [41]) presented the results of the quantification of miRNAs for both AD and PD. The total up-regulated and down-regulated miRNAs as biomarkers and therapeutic targets were obtained mainly from CSF, serum, and plasma. Most miRNAs were obtained from serum and plasma, facilitating laboratories worldwide' work for rapid sampling identification and quantification sampling.
The published studies selected in the present analysis presented mainly two types of controls, one composed of healthy participants and the other composed of participants with mild cognitive impairment (MCI), frontotemporal lobar degeneration, DLB (dementia with Lewy bodies), multiple system atrophy, and Progressive Supranuclear palsy. In addition, studies that presented Alzheimer's and Parkinson's diseases with genetic causes and not by transient or epigenetic effects were selected to eliminate the main confounders in accurately identifying miRNAs for AD and PD.
Furthermore, in the studies with better accuracy rates in the identification by qPCR of AD and PD miRNAs, the distribution of the values of the means and standard deviation   Table 7 in the respective groups of Figure 3. Alzheimer's disease (AD); Parkinson's disease (PD).

Discussion
Based on the objective of the present study, it was evidenced that the majority of the twenty-five studies of AD and PD presented a mean accuracy in identifying miRNAs by qPCR above 84%, with moderate to strong scientific evidence. These showed greater scientific credibility in the findings of each study, contributing in a tangible way to identifying the main miRNAs as selective biomarkers for the diagnosis of these diseases, as well as therapeutic targets in gene, cellular, and pharmacological treatment.
The present study's results do not show a risk of bias, both in studies with large and small sample sizes. In addition, two studies (Burgos et al., 2014 [40] and Nie et al., 2020 [41]) presented the results of the quantification of miRNAs for both AD and PD. The total up-regulated and down-regulated miRNAs as biomarkers and therapeutic targets were obtained mainly from CSF, serum, and plasma. Most miRNAs were obtained from serum and plasma, facilitating laboratories worldwide' work for rapid sampling identification and quantification sampling.
The published studies selected in the present analysis presented mainly two types of controls, one composed of healthy participants and the other composed of participants with mild cognitive impairment (MCI), frontotemporal lobar degeneration, DLB (dementia with Lewy bodies), multiple system atrophy, and Progressive Supranuclear palsy. In addition, studies that presented Alzheimer's and Parkinson's diseases with genetic causes and not by transient or epigenetic effects were selected to eliminate the main confounders in accurately identifying miRNAs for AD and PD.
Furthermore, in the studies with better accuracy rates in the identification by qPCR of AD and PD miRNAs, the distribution of the values of the means and standard deviation of the accuracy (%) of each study concerning the values of the total mean of AD and PD was, respectively, 84.37 ± 7.94% and 84.32 ± 7.15%. Among these, eight studies were identified with accuracy values (%) equal to or above the total average for AD, and seven studies were identified for PD in the identification and quantification of miRNAs (up-and downregulated) of greater scientific credibility as biomarkers and therapeutic targets in the identification of these diseases.
Additionally, from the total number of miRNAs identified (90 AD and 54 PD) which have the highest chances (Odds Ratio) of being identified by qPCR, a regression analysis was performed, which indicated five (5) miRNAs-miR-26b-5p (up-regulated), miR-615-3p (up-regulated), miR-4722-5p (up-regulated), miR23a-3p (up-regulated), and miR-27b-3p for AD, with OR = 2.55 (1023-3432) and p = 0.004 < 0.05, and only one (1) miRNA related to PD, miR-374a-5p (down-regulated), with OR = 2.16 (0.087-3.567) and p = 0.0035 < 0.05. After crossing the information, the results showed that all miRNAs (AD and PD) that presented the highest chances of being identified by qPCR (Table 9) are included in the groups of the prominent miRNAs with high accuracy (%), except for miR-27b-3p, belonging to the AD group of Table 9, as shown in Figure 4 of this study. These findings strongly highlighted the main miRNAs as biomarkers and therapeutic targets for AD and PD, thus contributing to future studies of advanced therapy with anti-miRNAs or antigenic modulation through vectors such as mesenchymal stem cell exosomes [30][31][32], as well as for pharmacological therapies [2,[27][28][29].
In this context, exosomes present a potential mechanism for the modulation of pathological α-Syn in the brain, as they can transport proteins and genetic material between cells, including mRNA and miRNA, contributing to the relief of AD and PD symptoms. Furthermore, because of their small size, exosomes can be used as vectors for the delivery of therapeutics [45][46][47][48].
Considering the critical role of α-Syn in PD, it is clear to understand the mechanisms that regulate its expression for therapeutic purposes since the reduced expression of these specific miRNAs can result in high levels of α-Syn in patients with PD. As a corollary of this, miR-7 and miR-153 have been shown to accelerate the degradation of performed α-Syn fibrils [65][66][67][68][69].
Additionally, MiR-205 levels are reduced in the frontal and striate cortex of PD patients, and LRRK2 expression is increased [70]. Genome-wide association studies have also identified variations in miR-4519 and miR-548at-5p related to PD [71]. However, the present study did not recruit this miRNA because it did not present significant accuracy in serum or plasma, given that the purpose of this study was to elect the main miRNAs of rapid and high identification and quantification for the diagnosis and monitoring of diseases and indications of biological and pharmacological direct relevance.
Based on these findings, it is essential to better understand the common genetic variants associated with AD and PD since most of the genetic risk remains uncharacterized. It is imperative to understand the role of regulatory elements such as miRNAs. The miRNAs relevant to neurodegenerative diseases are related to axonal guidance, apoptosis, and inflammation, so AD and PD likely arise from defects in underlying biological pathways. Furthermore, pathways regulated by APP, L1CAM, and genes from the caspase family may represent promising therapeutic targets of miRNAs in AD and PD, being therapeutic targets of deregulated miRNAs in both disorders [72].
As a corollary, targeting miRNAs offers a potential therapeutic opportunity for AD and PD, highlighting two strategies. The aim of this approach is based on the hypothesis that the downregulation of the specific protein level is a protective therapeutic strategy [73]. In this sense, the miRNA mimics that are used to inhibit the expression of target proteins stand out.
Moreover, miRNA-based therapy involves using anti-miRNA molecules to cause the loss of specific miRNA function [73]. As an example, miRNA-7 targets the 3 -UTR of α-Syn mRNA and facilitates the clearance of α-synuclein aggregates [74], and it exhibits protective effects against MPP+/1-methyl-4-induced toxicity -phenyl-1,2,3,6-tetrahydropyridine (MPTP) [75][76][77][78][79]. Therefore, miRNAs circulating in the blood and other biofluids can be characterized and used as non-invasive diagnostic biomarkers that facilitate early disease detection and the continuous monitoring of AD and PD disease progression. Such screen-ing is essential for understanding which types of miRNAs change in the progression of these diseases and when these changes happen [80].
In this way, the results of the present study can act as a guideline for miRNAs for research laboratories and pharmaceutical industries of interest in the possible treatments of AD and PD diseases. Soon, these results may support the diagnosis of these diseases and offer therapeutic interventions earlier in the disease process.

Conclusions
Based on the findings of this study, it was evident that most Alzheimer's and Parkinson's diseases studies showed accuracy in the qPCR identification of miRNAs above the total average, demonstrating greater scientific credibility and solidly contributing to the identification of the main microRNAs as selective biomarkers for the diagnosis of these diseases, as well as therapeutic targets.
This article can act as a microRNA guideline for research laboratories and pharmaceutical industries for treating Alzheimer's and Parkinson's diseases and offer the opportunity to evaluate therapeutic interventions earlier in the disease process.