Value of Blood-Based microRNAs in the Diagnosis of Acute Myocardial Infarction: A Systematic Review and Meta-Analysis

Background: Recent studies have shown that blood-based miRNAs are dysregulated in patients with acute myocardial infarction (AMI) and are therefore a potential tool for the diagnosis of AMI. Therefore, this study summarized and evaluated studies focused on microRNAs as novel biomarkers for the diagnosis of AMI from the last ten years. Methods: MEDLINE, the Cochrane Central database, and EMBASE were searched between January 2010 and December 2019. Studies that assessed the diagnostic accuracy of circulating microRNAs in AMI were chosen. The pooled sensitivity, specificity, positive likelihood ratio, negative likelihood ratio, diagnostic odds ratio, and area under the curve (AUC) were used to assess the test performance of miRNAs. Results: A total of 58 studies that included 8,206 participants assessed the diagnostic accuracy of circulating miRNAs in AMI. The main results of the meta-analyses are as follows: (1) Total miRNAs: the overall pooled sensitivity and specificity were 0.82 (95% CI: 0.79–0.85) and 0.87 (95% CI: 0.84–0.90), respectively. The AUC value was 0.91 (95% CI: 0.88–0.93) in the overall summary receiver operator characteristic (SROC) curve. (2) The panel of two miRNAs: sensitivity: 0.88 (95% CI: 0.77–0.94), specificity: 0.84 (95% CI: 0.72–0.91), AUC: 0.92 (95% CI: 0.90–0.94). (3) The panel of three miRNAs: sensitivity: 0.91 (95% CI: 0.85–0.94), specificity: 0.87 (95% CI: 0.77–0.92), AUC: 0.92 (95% CI: 0.89–0.94). (4) Results by types of miRNAs: miRNA-1: sensitivity: 0.78 (95% CI: 0.71–0.84), specificity: 0.86 (95% CI: 0.77–0.91), AUC: 0.88 (95% CI: 0.85–0.90); miRNA-133a: sensitivity: 0.85 (95% CI: 0.69–0.94), specificity: 0.92 (95% CI: 0.61–0.99), AUC: 0.93 (95% CI: 0.91–0.95); miRNA-208b: sensitivity: 0.80 (95% CI: 0.69–0.88), specificity: 0.96 (95% CI: 0.77–0.99), AUC: 0.91 (95% CI: 0.88–0.93); miRNA-499: sensitivity: 0.85 (95% CI: 0.77–0.91), specificity: 0.95 (95% CI: 0.89–0.98), AUC: 0.96 (95% CI: 0.94–0.97). Conclusion: miRNAs may be used as potential biomarkers for the detection of AMI. For single, stand-alone miRNAs, miRNA-499 may have better diagnostic accuracy compared to other miRNAs. We propose that a panel of multiple miRNAs with high sensitivity and specificity should be tested.


INTRODUCTION
Although advanced clinical medications have recently been developed for the diagnosis and prevention of coronary heart disease (CAD), acute myocardial infarction (AMI), which includes ST-elevation myocardial infarction (STEMI) and non-STEMI (NSTEMI), is still considered a primary public health threat, with high morbidity and mortality worldwide (Moss et al., 1996;GBD 2013 Mortality andCauses of Death Collaborators, 2015). Acute-phase reaction during ischemic damage is a crucial pathogenesis of ischemia myocardial issue (Hoffmeister et al., 2003). Dependent on accurate recognition and diagnosis, early effective revascularization treatment is an important strategy to repair ischemic myocardium and can significantly reduce the mortality of AMI patients (Hung et al., 2013). Currently, the most widely used biomarkers of myocardial injury during clinical practice are cardiac troponin and creatine kinase-MB (CK-MB), which may provide effective benefits for patients with revascularization therapy (Dohi et al., 2015;Anand et al., 2019). However, the elevation of cardiac troponin may be involved in serious, non-cardiac disease such as neuromuscular disorders, severe sepsis, and chronic renal insufficiency (Lamb et al., 2006;Finsterer et al., 2007;Vallabhajosyula et al., 2017). High levels of cardiac troponin have also been detected in patients with heart failure (Myhre et al., 2018). Therefore, early diagnostic biomarkers and improvement of the accuracy of approaches for the early prediction of AMI are still warranted.
Potential novel genetic and molecular biomarkers are currently being explored (Lorenzano et al., 2019). MicroRNAs (miRNAs/miRs) are endogenous, non-coding RNAs ∼19-25nt that play crucial post-regulatory roles in animals and plants by targeting mRNAs for translational or cleavage repression (Bartel, 2004). MiRNAs can inhibit or reduce target gene expression, subsequently affecting protein expression (Saxena et al., 2018). Thus, miRNAs play important regulatory roles in cell growth, development, and differentiation (Gabisonia et al., 2019). Further, miRNAs have been identified in extracellular fluid and can be extremely stable despite the presence of endogenous RNase (Chevillet et al., 2014). In recent years, a number of studies have reported that miRNAs are dysregulated in CAD, and that specific circulating miRNA signatures might be useful as biomarkers for the diagnosis of AMI and as therapeutic targets (Jakob et al., 2017). However, the results of previous studies were significantly different, potentially due to sample size, specimen types, and different detection technologies. Therefore, the purpose of this systematic review and meta-analysis was to summarize the diagnostic values of blood-based miRNA levels from published articles from the last 10 years and to appraise the accuracy of results to determine whether miRNAs may be used as novel biomarkers for the diagnosis of AMI.

MATERIALS AND METHODS
This study was conducted according to the Preferred Reporting Items for Systematic Reviews and Meta-analysis guidelines (Hutton et al., 2015;Moher et al., 2015). Two reviewers (CN Zhai, R Li) were independently involved with study selection, data extraction, and quality assessment.

Study Selection
An electronic search of MEDLINE (including PubMed), the Cochrane Central database, and EMBASE was performed to identify relevant articles published between January 2010 and December 2019. The following medical subject heading terms were used: ("plasma" OR "serum" OR "circulating") AND ("microRNA" OR "miRNAs" OR "miR * ") AND ("myocardial infarction" OR "AMI" OR "coronary heart disease" OR "coronary artery disease" OR "coronary syndrome" OR "ischemic heart disease"). No language restrictions were imposed. All relevant review articles were retrieved, and duplicates were removed by manually searching. Based on the title and abstract, manuscripts of interest were obtained for full-text review. Only full-text references were included.

Inclusion and Exclusion Criteria
Inclusion and exclusion criteria were developed by the investigative team. The inclusion criteria were: (1) human studies, (2) studies related to circulating miRNAs levels and AMI, and (3) studies that contained enough data to evaluate the diagnostic value of miRNAs in AMI. Exclusion criteria were based on the following: (1) studies evaluating tissue miRNA or miRNA in other body fluids; (2) case reports, conference abstracts, and reviews; and (3) non-human studies.

Data Extraction, Meta-Analysis, and Quality Assessment
Each manuscript was assessed independently by two researchers (CN Zhai and K Hou). Disagreements among reviewers were resolved by consensus. Data extracted included the following: authors, publication year, country, type of blood-based fluid (serum or plasma), characteristics of the study population (both case and control), study design (qRT-PCR detection method), whether miRNA screening was performed, number of miRNAs assessed, listing of the specific dysregulated miRNAs in AMI patients compared with controls, and outcome of statistical analyses including details of miRNA analysis, such as type of reference miRNA utilized. Studies reporting on single miRNA were included in the meta-analysis and were evaluated according to the Quality Assessment of Diagnostic Accuracy Studies 2 (QUADAS-2) checklist (Whiting et al., 2011), which is designed to assess the risk of bias and the applicability of studies of diagnostic accuracy. The following four key domains were included: patient selection, the index test, the reference standard, and flow and timing. Each was assessed with respect to the risk of bias, and the first three domains were assessed with respect to applicability.

Statistical Analysis
Analysis was based on the accuracy of the identified miRNAs for diagnosing the presence of AMI, as determined using Receiver Operator Characteristic (ROC) curves via the Area Under the Curve (AUC) value, and sensitivity and specificity where available (Carter et al., 2016). We calculated the pooled sensitivity, specificity, positive likelihood ratio (PLR), negative likelihood ratio (NLR), and diagnostic odds ratio (DOR), generated the bivariate summary receiver operator characteristic (SROC) curve, and calculated the area under the curve (AUC) to assess the overall diagnostic accuracy of miRNAs in distinguishing AMI patients from controls. Forest plots were constructed using STATA (15.0 StataCorp LP, College, Station, TX, USA). Due to the presumed heterogeneity of studies, a random-effects model (DerSimonian-Laird method) was used (Mahid et al., 2006). The heterogeneity of included studies was assessed using I 2 , and the P-value was considered significant if I 2 was >50% or P < 0.05. Subgroup analyses were performed to explore the potential source of heterogeneity as follows: (1) based on the type of blood sample (plasma or serum), (2) the method of qRT-PCR detection (SYBR Green or TaqMan), (3) the type of reference control used for normalization (RNU or Cel-miRNA), (4) sample size (Sample size ≥ 100 or Sample size < 100), and (5) different populations (Caucasian or East Asian). To assess the publication bias of the included studies, we performed Deeks' test of funnel plot asymmetry (Deeks et al., 2005).

Identification of Dysregulated miRNAs in the Included Studies
All included studies used Quantitative reverse transcription polymerase chain reaction (qRT-PCR) to detect the expression levels of miRNAs. A summary of all study methods is provided in Table 2. Seven studies performed miRNA screening to compare blood-based miRNAs between AMI patients and control groups Meder et al., Zhang R. et al., 2015;Shalaby et al., 2016;Wang et al., 2019), six studies identified a panel of three miRNAs (Long et al., 2012b;Wang et al., 2014Wang et al., , 2016Li H. et al., 2019;Xue et al., 2019a,b), and one study identified a panel of more than four miRNAs that were elevated in AMI .

Sensitivity, Specificity, and Area Under the Curve
Among the included studies, the most common methods of assessing the diagnostic accuracy of dysregulated miRNAs were AUC and sensitivity and specificity, as determined from ROC curves. In the identified dysregulated miRNAs, AUC values ranged from 0.510 to 1.0, sensitivity ranged from 48.6% to 100%, and specificity from 33.1% to 100%. In the single identified miRNAs, the highest AUC sensitivity and specificity combination was reported for miRNA-208b (AUC 1.0, sensitivity 100%, specificity 100%) . In the panel of two miRNAs, the highest value combination was reposted for miRNA-499 and miRNA-210 (AUC 0.98, sensitivity 97.9%, specificity 100%) . In the panel of three miRNAs, the highest value combination was reported for miRNA-21-5p, miRNA-361-5p, and miRNA-519-5p (AUC 0.989, sensitivity 93.9%, specificity 100%) .
In the panel of ≥ 4 miRNAs, there was one study about AUC 0.811, which had sensitivity 55.3% and specificity 90.1% .

Panels of Multiple miRNAs
As shown in Table 3, four studies Zhang R. et al., 2015;Shalaby et al., 2016;Wang et al., 2019) focused on the diagnostic value of a panel of two types of miRNAs, and the pooled sensitivity ( Figure 3A) and specificity ( Figure 3B) estimates were 0.88 (95% CI: 0.77-0.94) and 0.84 (95% CI: 0.72-0.91), respectively. The area under the SROC curve ( Figure 3C) was 0.92 (95% CI: 0.90-0.94). The Deeks' test ( Figure 3D) was performed to evaluate publication bias, and results suggested a low probability of publication bias. From the analysis of a panel of three types of miRNAs in six studies (Long et al., 2012b;Wang et al., 2014Wang et al., , 2016Ke-Gang et al., 2016;Li H. et al., 2019;Xue et al., 2019a,b) (Table 3), the pooled sensitivity ( Figure 4A) and specificity ( Figure 4B) estimates were 0.91 (95% CI: 0.85-0.94) and 0.87 (95% CI: 0.77-0.92), respectively. The area under the SROC curve ( Figure 4C) was 0.92 (95% CI: 0.89-0.94). The Deeks' test was performed and suggested that publication bias likely had a low effect on the summary estimates ( Figure 4D).   Table 4. According to compared diagnostic values from each of the above groups, no major improvement or differences were found in the diagnostic accuracy values.

DISCUSSION
This systematic review and meta-analysis of 58 manuscripts that utilize blood circulating miRNAs (including plasma-or serum-based) in the diagnosis of AMI identified 51 significantly dysregulated miRNAs between AMI cases and controls. Additionally, this review assessed the feasibility of using these miRNAs as novel biomarkers for the diagnosis of AMI patients. Sixteen of the abnormally expressed miRNAs were investigated by more than one study, including thirteen upregulated miRNAs: miRNA-1, , and miRNA-663b, and three downregulated miRNAs: miRNA-26a, miRNA-191, and miRNA-375. A further 34 dysregulated miRNAs were only reported by one study. The overall pooled diagnostic data of total miRNAs expression were as follows: SROC curve with AUC: 0.91, sensitivity: 0.82, specificity: 0.87, showing that circulating miRNAs might be suitable for use as potential biomarkers of AMI. Furthermore, the present meta-analysis was conducted via subgroup analyses based on type of miRNAs, including miRNA-1, miRNA-133a/b, miRNA-208a/b, and miRNA-499. MiRNA-499 had the highest diagnostic value (sensitivity: 0.85, specificity: 0.95, SROC curve with AUC: 0.96), followed by miRNA-133a (sensitivity: 0.85, specificity: 0.92, SROC curve with AUC: 0.93), and miRNA-208b had better specificity (0.96) than sensitivity (0.80). These results indicate a relatively high diagnostic accuracy for AMI based on significantly dysregulated miRNAs.
It is well-known that an early, accurate diagnosis and effective revascularization therapy play vital roles in reducing morbidity and mortality in patients with AMI (Yeh et al., 2010). At present, cardiac troponin, creatine kinase-MB (CK-MB), and myoglobin are the most widely used biomarkers in the diagnosis of AMI (de Winter et al., 1995). A good standard for the early diagnosis of AMI is an increase of cTn level (Celik et al., 2011). However, these markers are also likely to be elevated in patients with other diseases, whether or not CAD is also present (French and White, 2004). The detection of cTn has time constraints, as significant levels are reached 4-8 h following the onset of ischemia symptoms. Thus, novel genetic and molecular biomarkers of myocardial damage that have high sensitivity and specificity are still urgently needed.
Significant advances have occurred in the field of cardiovascular disease and miRNAs since their first discovery in the blood (Karakas et al., 2017). A growing number of studies indicate that the abnormal expression of miRNAs plays a critical role due to their various pathological functions in the presence of myocardial infarction (Gurha, 2016;Moghaddam et al., 2019). MiRNAs are steadily present in bodily fluids (including plasma, serum, urine, and saliva) due to protection from RNase via binding to argonaute proteins and the ability to be released from cells via microvesicles, exosomes, or bound to proteins (Meister, 2013). Moreover, recent studies have showed that cTn is more difficult to detect than miRNAs in patients with MI during the earlier acute stage, as it is usually below the cut-off value . This suggests a difference between these two types of biomarkers in the physiological process of myocardial infarction. Cardiac troponin releases into the blood during necrosis and during the pathological process of myocardial hypoxia and ischemia (Wu and Ford, 1999). However, miRNAs can be released in response to several forms of cellular stress occurring earlier then cell necrosis such as anoxia, lactic acidosis, and cellular edema (Edeleva and Shcherbata, 2013). Thus, experts may consider the expression levels of dysregulated miRNAs at an earlier stage of AMI, as they might be reliable candidate biomarkers for the diagnosis of AMI .
In this study, the summary of single miRNAs showed that miRNA-1, miRNA-133, miRNA-208, and miRNA-499 are potential candidates for the detection of AMI, as they were most frequently detected in the previous studies. Among these individual miRNAs, circulating miRNA-499 might be an effective candidate biomarker of AMI. Some studies have demonstrated that miRNA-499 was specifically expressed in the myocardium and skeletal muscle of mammals  and played a critical role in the recovery process following cardiac injury (Hosoda et al., 2011). Based on the present study, miRNA-499 has a higher sensitivity and specificity in identifying patients with AMI, and these results were similar to or better than previous studies Zhao et al., 2019). Previous studies were based on relatively small sample sizes and did not include studies published in recent years. Thus, the results of the present study are more reliable and convincing. Furthermore, we strictly considered the precise setting of specific miRNAs in the diagnosis of AMI and that assessing their diagnostic performance in combination with other biomarkers, especially highly sensitive troponin immunoassays, may be warranted. A study by Olivieri et al. showed a significant correlation between miRNA-499 and cTnT, and that the diagnosis value of this combination was superior to either one alone . However, another study reported that the diagnostic value of miRNA-499 and hs-cTnT combined was not better than either them alone . Thus, due to the limited sample sizes of these studies, further research is required.
There are only a small number of high-quality studies with large sample sizes focused on the diagnostic value of miRNAs in AMI. To the best of our knowledge, this study has included the largest sample size of dysregulated miRNAs as novel biomarkers for AMI for summary and evaluation. Further high-quality studies should be performed to acquire more reliable data for use in formulating standard diagnostic criterion and to determine optimal cut-off values. Moreover, due to the different diagnostic values of miRNAs, our study demonstrated that a panel of 2 or 3 miRNAs might be superior for diagnostic accuracy. The combination of 3 miRNAs trended toward a higher sensitivity than others for use as biomarkers in the diagnosis of AMI. Therefore, to further improve the feasibility of clinical diagnosis, future research should explore the most effective combination of multiple miRNAs, especially those confirmed to have a higher utilization value in the single miRNA groups.
Aside from the many studies investigating miRNA profiles in the detection of AMI, researchers should pay closer attention to the search for the technology to detect miRNAs quickly and accurately. Methods of RNA detection tend to be timeconsuming, expensive, require sophisticated techniques, and are difficult to implement for urgent testing, especially in some developing countries (Lippi et al., 2013). However, novel detection technologies developed in recent years may provide a solution to these tissues, which would also support the clinical application of miRNAs in the future. Examples of new technologies include isothermal reactions based on cleavage with DNAzyme and signal amplification, which can simultaneously amplify and detect RNA . This method is thought to be immune to genomic DNA pollution and has a relatively reliable sensitivity and specificity. We believed that new, precise methods of diagnosis that can detect miRNAs very rapidly and inexpensively need to be continuously improved.
In addition, the heterogeneity of the results of this study were substantial and could not be completely resolved. We found that the sources of heterogeneity included the following: quality of included studies, age, gender proportion, regional and environmental factors, and sampling criteria. Despite the heterogeneity, we believe the results of this study are worthwhile and valuable. Results of the overall and stratified analysis of different subgroups trended toward satisfactory values of miRNAs as novel biomarkers in the diagnosis of AMI. Furthermore, it was remarkable that some publication bias was found in the present study and might imply that the potential negative results were less likely to be published.
Although this study had a satisfactory result regarding the use of miRNAs for AMI detection, conclusions should be drawn cautiously due to several limitations: (1) there was a lack of standardization due to different normalization procedures across the included studies; (2) the exclusion of non-English articles may have caused important studies to be overlooked and publication bias due to significant results being more easily published; (3) the combined analysis of multiple miRNAs for a panel were insufficient, and we were unable to conduct a meta-analysis based on the data from limited studies; and (4) the results may have been affected by the impact of inevitable clinical heterogeneity, including the general condition of the included individuals, effects of medication, and medical history.

CONCLUSION
The current systematic review identifies numerous miRNAs associated with AMI and suggested that miRNAs may be used as a potential biomarker for the detection of AMI. For single, stand-alone miRNAs, miRNA-499 had better diagnostic accuracy than other miRNAs. A panel of two or three miRNAs might be superior for diagnostic accuracy. To develop a diagnostic test for AMI diagnosis, we suggest that a panel of miRNAs with high sensitivity and specificity should be tested. For this purpose, large scale, high-quality studies are still required to validate the clinical application of miRNAs for AMI diagnostics, as well as to identify the precise setting of dysregulated miRNAs in patients with AMI.

DATA AVAILABILITY STATEMENT
Publicly available datasets were analyzed in this study. Datasets are available through the corresponding author upon reasonable request.

ETHICS STATEMENT
All analyses were based on previous published studies; thus no ethical approval or patient consent were required. All previous published studies were approved by Ethics Committee, respectively.

AUTHOR CONTRIBUTIONS
HC and CZ designed the study. CZ carried out the statistical analysis and participated in most of the study steps. CZ, RL, JC, KH, and MA prepared the manuscript and assisted in the study processes. YH, JZ, YZ, LW, and RZ assisted in the data collection and helped in the interpretation of the study. All authors read and approved the final manuscript.