Circulating microRNAs as biomarkers of cardiovascular disease: A systematic review

Abstract

BACKGROUND

MicroRNAs (miRNAs) are small, non-coding RNAs that play essential roles in various biological processes. Numerous miRNAs have been identified as biomarkers for various pathologies, including cardiovascular diseases (CVDs). Circulating miRNAs specifically have potential as non-invasive biomarkers, reflecting pathological changes at the cellular level in CVDs. Despite substantial advances in identifying circulating miRNAs associated with CVDs, considerable uncertainty remains regarding their clinical utility, consistency across studies, and integration into standardized diagnostic and prognostic workflows.

AIM

To summarize the identified circulating miRNAs used in the diagnosis or prognosis of CVD.

METHODS

In accordance with PRISMA 2020 guidelines and a registered protocol (PROSPERO CRD420251011773), we conducted comprehensive search of PubMed and CENTRAL from inception to March 2025. Records were screened in duplicate, and we included observational studies or clinical trials evaluating blood-based miRNAs in CVD. Data extraction was performed independently by two reviewers.

RESULTS

We identified 34 studies, comprising approximately 17642 participants, spanning diverse cardiovascular conditions, including acute coronary syndromes, coronary artery disease (CAD), heart failure (HF)/cardiomyopathy, stroke, cardiogenic shock (CS), and hypertension. Diagnostic findings included the upregulation of miR-133a and miR-328 in acute myocardial infarction (AMI), with miR-208b showing an area under the curve (AUC) of 0.76; a miR-132/150/186 panel for unstable angina (AUC = 0.91); and decreased miR-423-3p (AUC = 0.80) along with female-specific elevation of miR-18a in CAD. Prognostic associations included the miR-19 family with cardiovascular mortality in CAD; miR-26b-5p, miR-320a, miR-660-5p, and miR-122-5p/miR-133b ratio, with major adverse cardiovascular events after ST-segment elevation myocardial infarction; miR-328, miR-134, and miR-192 with incident HF after AMI; and miR-423-5p with higher mortality and miR-20b-5p with better survival in CS. Reported diagnostic accuracy varied (AUC values shown where available), and results were influenced by variation in sampling matrix, timing, normalization strategies, and assay platforms.

CONCLUSION

Although circulating miRNAs show potential for CVD diagnosis and risk stratification, study heterogeneity and limited external validation constrain their clinical utility. Future research should prioritize standardized pre-analytical procedures, harmonized reporting, and prospective multicenter validation to demonstrate incremental value over established clinical risk scores and biomarkers.

Key Words: Cardiovascular disease; MicroRNAs; Biomarkers; Extracellular biomarkers; Precision medicine

Core Tip: This systematic review summarizes current evidence on circulating microRNAs (miRNAs) as diagnostic and prognostic biomarkers across cardiovascular disease, including acute coronary syndromes, coronary artery disease, heart failure, stroke, cardiogenic shock, and hypertension. Several miRNAs showed promising disease-specific signals, including miR-133a, miR-328, miR-423-5p, miR-19a, and miR-21. However, clinical translation remains limited by heterogeneity in sampling methods, assay platforms, normalization strategies, and limited external validation. Standardized multicenter prospective studies are needed before circulating miRNAs can be incorporated into routine cardiovascular risk stratification.

INTRODUCTION

Cardiovascular diseases (CVDs) are the leading cause of morbidity and mortality worldwide[1]. In this review, CVD is used as an umbrella term that includes acute presentations—acute coronary syndromes [ACS; acute myocardial infarction (AMI) and unstable angina (UA)]—and chronic conditions, such as stable coronary artery disease (CAD), heart failure (HF)/cardiomyopathy, stroke, cardiogenic shock (CS), and hypertension (HTN). Early identification, accurate risk stratification, and prognostic assessment improve outcomes across coronary, myocardial, vascular, and hypertensive disorders[2]. Moreover, reliable prognostic tools may aid in further personalization of management strategies to optimize patient outcomes. Although blood-based biomarkers and imaging have advanced CVD care, routinely used tools still have important setting-specific limitations: In ACS, early rule-in and rule-out can require serial sampling and borderline results may reduce specificity; in HF, natriuretic peptides are influenced by age, obesity, and renal function; in HTN, validated circulating markers for resistant disease and silent organ damage remain limited; and imaging, while essential, can be costly, operator-dependent, and less feasible for frequent serial assessment. This need has led most researchers and clinicians to strive to identify novel biomarkers that can be used in the identification, prevention, and even management of these CVDs[3]. However, despite the significant efforts and resources invested, identifying reliable biomarkers that can be measured routinely in plasma or serum remains a significant challenge[4]. Recent research has been increasingly focused on microRNAs (miRNAs), which are key regulators of various cellular physiological and pathological processes in the cardiovascular system[3]. Despite advances in identifying circulating miRNAs relevant to CVD, uncertainty remains regarding their consistency across studies, robustness of diagnostic and prognostic performance, and readiness for integration into routine clinical workflows, creating a critical evidence gap that this review aims to address.

The miRNAs are single-stranded, non-coding RNAs that play significant roles in various biological processes, such as cell proliferation, differentiation, apoptosis, and autophagy (Figure 1)[5]. Circulating miRNAs are secreted by multiple cell types and can be detected in various body fluids, including serum, plasma, blood, saliva, urine, and breast milk. These circulating miRNAs possess various properties that support their potential as CVD biomarkers[6]. First, miRNAs are stable in blood and circulation since endogenous RNases do not readily degrade them[7]. This stability extends to extreme conditions such as high pH, extended storage, and non-ambient temperatures[7]. Moreover, miRNAs demonstrate disease and tissue specificity and can be easily detected from serum/plasma using sequence-specific amplification techniques, enabling relatively rapid and minimally invasive disease detection across organ systems, including the cardiovascular system[8].

Figure 1 Schematic of microRNA biogenesis and gene silencing. Primary microRNA (pri-miRNA) is transcribed by RNA polymerase II from DNA. The microprocessor complex, consisting of Drosha and DiGeorge syndrome critical region 8, cleaves pri-miRNA into a hairpin precursor microRNA (miRNA). The hairpin is exported from the nucleus by exportin-5/RAN-GTP. In the cytoplasm, RNase III endoribonuclease TRBP generate a miRNA duplex. The duplex unwinds, and the mature strand associates with the RNA-induced silencing complex, which directs translational repression or mRNA target cleavage. RISC: RNA-induced silencing complex; miRNA: MicroRNA; Pri-miRNA: Primary microRNA.

The potential role of miRNAs as biomarkers for the diagnosis and prognosis of CVDs has been highlighted in various studies[9-11]. This systematic review therefore aims to summarize and synthesize the current evidence from original human observational and clinical studies on circulating blood-based miRNAs evaluated as diagnostic or prognostic biomarkers in CVDs, including diagnostic accuracy measures and clinical prognostic outcomes.

MATERIALS AND METHODS

Protocol and registration

This systematic review was conducted in accordance with the PRISMA guidelines[12]. The protocol of the systematic review has been registered in PROSPERO (ID: CRD420251011773).

Literature search

The literature search was conducted independently by two reviewers, Popat A and Sharma PP. A comprehensive search strategy was developed and applied to two electronic databases, PubMed and CENTRAL, to identify peer-reviewed articles published from database inception through March 2025 (Figure 2). The following search terms and their combinations were used: (“Circulating microRNAs” OR “blood microRNAs” OR “serum microRNAs”) AND (“cardiovascular disease” OR “acute coronary syndrome” OR “coronary artery disease” OR “myocardial infarction” OR “cardiovascular disease”). Reference lists of all eligible studies were screened to identify additional relevant articles not captured in the initial database searches. Lastly, the ClinicalTrials.gov registry was searched to identify completed trials with available results that had not been published.

Figure 2 A PRISMA diagram summarizing the search strategy. CVD: Cardiovascular disease; miRNA: MicroRNA.

Eligibility criteria

After records were retrieved from the databases and registries, two reviewers applied prespecified eligibility criteria to assess studies for inclusion. Studies were eligible if they met all of the following criteria: (1) Were published in English; (2) Evaluated the utility of circulating (blood-based) miRNAs for the diagnosis or prognosis of CVDs, including CAD, ACS, HF, stroke, angina, or CS; and (3) Used a clinical trial or primary observational design (case-control, cohort, or cross-sectional).

Studies were excluded if they met any of the following criteria: (1) Did not include patients with CVD; (2) Evaluated biomarkers other than miRNAs; (3) Measured miRNAs in immune cells or other non-blood tissues rather than in blood; or (4) Were secondary studies (systematic reviews, narrative reviews, or meta-analyses). We also excluded case reports and case series.

Data extraction

Two reviewers (Popat A and Sharma PP) independently extracted data for this review. Any discrepancies were resolved through discussion until consensus was reached. In line with the PRISMA guidelines, all references identified at each stage were screened before data extraction. Titles and abstracts were screened for relevance, and clearly irrelevant articles were excluded. Full texts of the remaining studies were then assessed against the eligibility criteria, and data were extracted from those meeting all inclusion criteria. For each included study, the following information was collected: Study setting, sample size, type of miRNA investigated, biological source of the measured miRNA, primary findings, and the CVD phenotype evaluated (Table 1).

Table 1 Characteristics of the included studies.

Ref.	Design/setting (country)	Sample size	miRNAs	Source	Sample timing	Cases	Controls	Function of miRNA	Main findings	Diagnostic/prognostic values
Cardiomyopathy
Klenke et al[15], 2018	PCS (Germany)	91	1 miRNA (hsa-miR-192-5p)	Serum	During routine heart failure clinic check-ups	Patients diagnosed with HF due to ICM (59.3%)	Patients with HF due to non-ICM (40.7%)	Prognosis (p53-responsive prognostic marker involved in cardiomyocyte apoptosis)	High expression of miR-192 was associated with worse survival in ICM (P = 0.003)	Independent prognostic marker (P = 0.014) for survival in ICM; high expression resulted in a 17-year lower median age at death (67 years vs 84 years)
Heart failure
Shah et al[16], 2018	PCS (United States)	2763	3 miRNAs (miR-106b-5p, miR-17-5p, miR-20a-5p)	Plasma	Baseline plasma from the eighth Framingham Offspring Cohort examination	Patients without HF	No controls	Prediction of incident HF	Higher circulating levels of miR-17, miR-20a, and miR-106b were associated with a lower risk of incident HF	Among 2681 participants, 116 developed HF during a median follow-up of 7.7 years. MiR-20a: HR = 0.86, P = 0.047; miR-17: HR = 0.84, P = 0.03; miR-106b: HR = 0.85, P = 0.04
Acute coronary syndrome
Shen et al[18], 2021	Prospective nested case-control study (China)	1190	14 miRNAs (miR-4286 significant for ACS)	Plasma	Baseline blood samples were collected in 2013, with a median follow-up of 1.8 years	Incident ACS cases during follow-up	Matched controls free of CVD and cancer at the time of the case event	Prediction of incident ACS risk	Upregulation of miR-4286 was associated with increased risk of incident ACS	Highest tertile vs lowest tertile: OR = 1.80 (95%CI: 1.28-2.53); per IQR increase: OR = 1.26 (95%CI: 1.07-1.48)
Pilbrow et al[17], 2014	PCS (New Zealand)	300	5 miRNAs (miR-323-3p, miR-652, miR-27b, miR-103, and miR-208a)	Plasma	Baseline (approximately 30 days post-ACS), 4 months, and 12 months post-ACS	Patients with recent ACS	Healthy controls matched according to age and gender, where possible	Diagnosis and prognosis	MiR-323-3p, miR-652, and miR-27b were elevated in ACS patients. miR-652 in the lowest tertile was associated with heart failure readmission	MiR-652: AUC = 0.861; miR-323-3p: AUC = 0.806; miR-27b: AUC = 0.799 for ACS discrimination; low baseline miR-652 was associated with HF readmission (log-rank P < 0.001)
Biener et al[19], 2022	PCS (Germany)	1042	11 miRNAs (miR-134-5p, miR-21-5p, miR-191-3p, miR-29a-3p, miR-22-3p, miR-92a-3p, miR-126-3p, miR-122-5p, miR-132-3p, miR-133a-3p, and miR-423-5p)	Plasma	At presentation to the emergency department	Patients presenting with suspected ACS	NR	Prognosis	Higher expression of miR-133a-3p was observed in patients with higher troponin levels at baseline. Higher miR-21-5p and lower miR-122-5p were observed in patients who died during follow-up; higher miR-126-3p was observed in patients with the composite endpoint of all-cause mortality, AMI, and stroke	MiR-21-5p: AUC = 0.659 (95%CI: 0.596-0.723); miR-122-5p: AUC = 0.605 (95%CI: 0.526-0.685) for all-cause mortality; ROC-optimized miR-21-5p cutoff predicted all-cause mortality with HR = 3.3 (P25-P75: 1.2-9.4)
Acute myocardial infarction (STEMI and NSTEMI)
Eyyupkoca et al[20], 2022	PCS (Turkey)	10 (all male)	8 miRNAs (miR-26b-5p, miR-301a-3p, miR-23b-3p, miR-374a-5p, miR-199a-5p, miR-483-5p, miR-423-5p, and miR-652-3p)	Plasma	Baseline (1 day), 2 weeks, and 6 weeks after AMI	STEMI patients with ALVR	STEMI patients with reverse left ventricular remodeling	Prediction of ALVR	At day 1 after AMI, miR-199a-5p, miR-23b-3p, and miR-483-5p were downregulated, whereas miR-26b-5p and miR-652-3p were upregulated in ALVR; at 2 weeks, miR-374a-5p was downregulated; at 6 weeks, miR-301a-3p and miR-374a-5p were downregulated, whereas miR-423-5p was upregulated in ALVR	ALVR was defined as a ≥ 13% increase in LVEDV and LVESV at 6 months after AMI
Cai et al[21], 2024	Observational study (China)	100	1 miRNA (miR-133)	Plasma	At diagnosis and 24 hours after diagnosis	Patients with AMI	Healthy subjects and unstable angina patients	Diagnosis	MiR-133 levels were upregulated in AMI patients	MiR-133 showed high diagnostic accuracy for AMI (AUC = 0.956; sensitivity = 87.6%; specificity = 94.7%)
Grabmaier et al[22], 2017	CCS (Germany)	62	4 miRNAs (miR-1, miR-21, miR-29b, and miR-92a)	Plasma	Day 4, day 9, and 6 months after AMI	Patients with AMI in the SITAGRAMI trial	Age-, sex- and risk factor-matched controls	Diagnosis and prognosis	MiR-1, miR-21, and miR-29b were increased after AMI; miR-1 and miR-29b were inversely correlated with infarct volume changes, and miR-29b at day 9 inversely correlated with changes in LVEDV at 6 months	MiR-1 and miR-29b inversely correlated with change in infarct volume (P < 0.05), and miR-29b inversely correlated with change in LVEDV (P < 0.05)
Wang et al[23], 2011	CCS (China)	79	2 miRNAs (miR-133 and miR-328)	Plasma	Within 24 hours and at 7 days after AMI	Patients with AMI	Control subjects (CAD excluded by coronary angiography)	Diagnosis	Plasma miR-133 and miR-328 levels were increased in AMI patients (10.9-fold in plasma and 16.1-fold in whole blood) and returned to control levels by 7 days	MiR-133 plasma AUC = 0.890 (95%CI: 0.772-0.965); miR-328 plasma AUC = 0.810 (95%CI: 0.705-0.916)
Devaux et al[24], 2015	Prospective observational study (multicenter)	1155	6 miRNAs (miR-133a, miR-208b, miR-223, miR-320a, miR-451, and miR-499)	Plasma	At presentation with acute chest pain	Patients with AMI	Patients with final diagnoses other than AMI	Diagnosis and prognosis	Levels of miR-208b, miR-499, and miR-320a were higher in AMI; miR-208b levels were higher in patients who died within 30 days	MiR-208b showed the highest diagnostic accuracy for AMI (AUC = 0.76, 95%CI: 0.72-0.80); miR-208b for predicting 30-day mortality (AUC = 0.67, 95%CI: 0.52-0.81)
Jakob et al[25], 2017	Nested case-control study (multicenter)	189	14 miRNAs (miR-26b-5p, miR-320a, and miR-660-5p)	Plasma	At presentation with STEMI	Patients with STEMI who experienced MACE	Patients without MACE during 1-year follow-up	Prognosis	MiR-26b-5p was decreased, whereas miR-320a and miR-660-5p were increased in patients with MACE	MiR-26b-5p AUC = 0.707; miR-660-5p AUC = 0.683; miR-320a AUC = 0.672; combined AUC = 0.718
Matsumoto et al[26], 2013	CCS (Japan)	86	3 p53-responsive miRNAs (miR-192, miR-194, and miR-34a)	Serum	Median 18 days after AMI onset	Post-AMI patients who developed HF within 1 year	Matched controls without subsequent cardiovascular events	Prognosis	Serum miR-192, miR-194, and miR-34a were upregulated in patients who developed ischemic HF after AMI	MiR-194 and miR-34a correlated with larger left ventricular diastolic dimension at 1 year (r = 0.33, P = 0.01; r = 0.38, P = 0.003)
Cortez-Dias et al[27], 2016	Prospective observational study (Portugal)	160	6 miRNAs (miR-1-3p, miR-122-5p, miR-133a-3p, miR-133b, miR-208b-3p, and miR-499a-5p)	Serum	At the time of cardiac catheterization; additional sampling at 8 hours, 16 hours, 24 hours, 48 hours, and 72 hours in a subset	Patients with STEMI undergoing primary PCI	Healthy controls and non-STEMI controls	Prognosis	A higher miR-122-5p/miR-133b ratio was associated with increased risk of death or recurrent MI and with adverse cardiovascular events after STEMI	HR = 1.49 (95%CI: 1.09-2.03), P = 0.012; patients in the highest tertile of the miR-122-5p/133b ratio had an almost 9-fold higher risk of death or recurrent MI and a 4-fold higher risk of MACE. Multivariable model (miR ratio and LVEF) for all causes mortality: AUC = 0.81 (95%CI: 0.64-0.99)
He et al[28], 2014	CCS (China)	389	2 miRNAs (miR-328, and miR-134)	Plasma	Median 6 hours after symptom onset (range 2-10 hours)	Patients with AMI	Healthy controls	Diagnosis and prognosis	Plasma miR-328 (12.04-fold) and miR-134 (6.32-fold) were elevated in AMI patients; higher levels were associated with increased risk of cardiogenic death or HF within 6 months	Diagnostic AUC: MiR-328, 0.887; miR-134, 0.818. Prognostic OR for cardiogenic death or heart failure within 6 months: MiR-328, 7.35, P = 0.004; miR-134, 2.28, P = 0.013
Bauters et al[29], 2013	PCS (France)	246	2 miRNAs (miR-423-5p and miR-133a)	Plasma	Days 3-7 after AMI, then at 1 month, 3 month, and 12 months	Patients with first anterior wall Q-wave AMI	NR	Prognosis	Circulating miR-133a increased at 12 months, and miR-423-5p increased at 1 month, 3 months, and 12 months after MI	Neither miRNA was associated with LV function, LV remodeling, or BNP during 1-year follow-up
Hromádka et al[30], 2019	Retrospective cohort study (Czech Republic)	122	3 miRNAs (miR-1, miR-133a, and miR-499)	Plasma	24 hours after admission	AMI patients who died in 1 year	AMI patients who survived for 1 year	Prognosis	MiR-1, miR-133a, and miR-499 positively correlated with hsTnT, NT-proBNP, and negatively with LVEF; levels were higher in nonsurvivors, but differences were not statistically significant	All nonsurvivors had miR-499 > 0.088 and NT-proBNP > 891.5 ng/L (vs 28.4% of survivors, P = 0.001)
Angina
Zeller et al[31], 2014	CCS (Germany and Switzerland)	246	8 miRNAs (miR-19a, miR-19b, miR-132, miR-140-3p, miR-142-5p, miR-150, miR-186, miR-210)	Serum	At admission and 6 hours later	Patients with UA	Non-coronary chest pain patients and healthy controls	Diagnosis	Lower levels of 8 circulating miRNAs were significantly associated with UA; a 3-miRNA panel of miR-132, miR-150, and miR-186 showed the best discriminatory performance	3-miRNA panel (miR-132, miR-150, miR-186): AUC = 0.91 (95%CI: 0.84-0.98)
Simionescu et al[32], 2016	Cross sectional observational (Romania)	137	6 miRNAs (miR-146a, miR-125a, miR-92a, miR-486, miR-223, and miR-122)	Serum and HDL	Fasting serum samples at enrollment	Patients with stable angina and ACS	Healthy controls	Diagnosis	MiR-223, miR-92a, miR-486, miR-122, miR-125a, and miR-146a were increased in CAD patients; in hyperglycemic ACS, levels were higher than in normoglycemic ACS, especially for miR-223, miR-92a, and miR-486 in HDL	MiR-223, miR-486, and miR-92a discriminated ACS from stable angina
Wang et al[33], 2017	Observational study (various countries)	199	5 miRNAs (hsa-miR-126-5p, hsa-miR-3135b, hsa-miR-28-3p, hsa-miR-142-5p, hsa-miR-144-5p)	Whole blood	Baseline	Patients with non-STEMI and angina	Matched controls without any CVD event	Prognosis	Lower miR-3135b, miR-126-5p, miR-142-5p, and miR-144-5p were associated with HF; higher GRACE risk score was associated with lower miR-3135b and higher miR-28-3p	MiR-126-5p: Fold change 0718, P = 0.003; miR-142-5p: Fold change 0762, P = 0.044; miR-144-5p: Fold change 0816, P = 0.044; miR-3135b: Fold change 0769 with chronic HF (P = 0.044) and 0.994 per 1-SD increase in GRACE score (P = 0.013); miR-28-3p: Fold change 1004 per 1-SD increase in GRACE score (P = 0.019)
Zhong et al[34], 2018	Observational study (China)	78	8 miRNAs with ROC curve for 4 miRNAs (miR-126-5p, miR-17-5p, miR-142-3p, and miR-340-3p)	Plasma	At enrollment	Patients with UA and STEMI	Controls with a normal coronary artery	Diagnosis	MiR-126-5p, miR-142-3p, miR-17-5p, and miR-340-3p were upregulated in UA and STEMI compared with controls	In UA patients, only miR-142-3p achieved a high ability to distinguish from the normal coronary artery patients (AUC = 0.80). In STEMI patients, miR-17-5p and miR-142-3p both had good discriminative ability compared with normal coronary artery patients, with AUCs of 0.84 and 0.85, respectively
Coronary artery disease
Polyakova et al[35], 2021	Cross-sectional study (Russia)	100	3 miRNAs (miR-27a, miR-133a, and miR-203)	Serum and atrial myocardium	Before elective cardiac surgery after 12-hour fasting	CAD patients undergoing elective CABG	Patients without CAD undergoing isolated valve surgery	Diagnosis/disease severity	MiR-27a, miR-133a, and miR-203 were higher in CAD than in controls (in serum and atrial myocardium); serum miR-203 correlated with CAD extent and was highest in multivessel disease	Serum miR-203 correlated with SYNTAX score I (r = 0.693, P < 0.001); serum miR-203 > 101 relative expression units predicted multivessel disease (OR = 5.90, 95%CI: 2.34-9.46, P < 0.001)
Karakas et al[36], 2017	PCS (Germany)	1112	8 miRNAs (miR-19a, miR-19b, miR-132, miR-140-3p, miR-142-5p, miR-150, miR-186, and miR-210)	Serum	Baseline, before angiography	Patients with documented CAD (ACS and stable angina)	None	Prognosis	Higher circulating levels of most candidate miRNAs were associated with cardiovascular death, particularly in ACS patients	In ACS patients, miR-132: HR = 2.85 per 1-SD increase, P = 0.022, AUC = 0.737; miR-140-3p: HR = 2.88, P = 0.022, AUC = 0.756; miR-210: HR = 3.10, P = 0.039, AUC = 0.754
de Gonzalo-Calvo et al[37], 2019	PCS (Spain)	237	10 miRNAs (let-7 g-5p, miR-15b-5p, miR-21-5p, miR-24-3p, miR-29b-3p, miR-130a-3p, miR-143-3p, miR-146a-5p, miR-222-3p, and miR-663a)	Plasma	At nonurgent CCTA evaluation	Patients with suspected stable CAD	None	Diagnosis/disease severity	Lower circulating levels of several miRNAs were associated with greater coronary atherosclerosis extension and severity; miR-143-3p was inversely associated with segment involvement score, and let-7 g-5p, miR-15b-5p, miR-21-5p, miR-24-3p, miR-130a-3p, miR-143-3p, miR-146a-5p, and miR-222-3p were inversely associated with segment stenosis score	Circulating miRNAs had poor discriminative ability for CAD indexes (AUC = 0.539-0.644)
Wang et al[38], 2014	CCS (China)	181	5 miRNAs (miR-487a, miR-29b, miR-502, miR-208, and miR-215)	Serum	After 12-hour overnight fast	Patients with atypical coronary artery disease	Age-matched healthy controls	Diagnosis	Serum miR-487a, miR-502, miR-208, and miR-215 were increased, whereas miR-29b was decreased in atypical coronary artery disease patients compared with controls	The 5 miRNAs were able to distinguish CAD patients from the controls with high discriminative ability. AUC = 0.850 (95%CI: 0.734-0.966) in the training set and 0.909 (95%CI: 0.858-0.960) in the validation set
Wang et al[39], 2020	PCS (China)	2812	5 miRNAs screened; miR-423-3p showed the best predictive performance	Serum	Baseline	General population participants without pre-existing CVD	None	Prediction of primary CAD events	Higher circulating miR-423-3p levels were associated with a lower risk of incident CAD events during 6-year follow-up	MiR-423-3p tertile 2 vs tertile 1 HR = 0.48 (95%CI: 0.27-0.84); tertile 3 vs tertile 1 HR = 0.28 (95%CI: 0.14-0.57); AUC improved from 0.782-0.806 after adding miR-423-3p to traditional risk factors; net reclassification improvement 19.18%
Yuan et al[40], 2018	CCS (China)	120	1 miRNA (miR-18a)	Serum and circulating endothelial cells	At sample collection	Patients with CAD	Healthy controls	Diagnosis/mechanistic association	Levels of miR-18a were upregulated in female CAD patients than in the controls. No significant difference was observed in the levels among male CAD patients. Higher miR-18a was associated with lower HDL-C and estradiol and higher LDL-C, total cholesterol, and triglycerides	In female patients, miR-18a was overexpressed vs controls (P < 0.01); correlations: HDL-C r = -0.772, P < 0.001; LDL-C r = 0.570, P = 0.0075; estradiol r = -0.481, P = 0.001; total cholesterol r = 0.526, P = 0.003; triglycerides r = 0.513, P = 0.004
Opstad et al[41], 2017	Cross sectional observational (Norway)	200	1 miRNA (miR-21)	Plasma	At study entry in fasting state	Patients with stable CAD who smoke	Patients with stable CAD without a history of smoking	Association	Current smoking was associated with lower circulating miR-21 expression and lower IL-12p35 mRNA expression in stable CAD patients	Adjusted P < 0.05 for lower miR-21 expression in current smokers
Iusupova et al[42], 2024	Cross-sectional observational study (Russia)	157	3 miRNAs (miR-145, miR-34a, and miR-222)	Plasma	At study evaluation	Patients with CAD (obstructive or non-obstructive) presenting with complaints of chest pain and dyspnea	Healthy controls	Diagnosis	MiR-34a and miR-222 were significantly upregulated in CAD patients than in the healthy controls	MiR-145 was an independent predictor of ischemia/angina with non-obstructive coronary artery disease: OR = 2.512 (95%CI: 1.294-4.875, P = 0.006)
Stroke
Mayer et al[43], 2019	Prospective cohort study (Czech Republic)	826	5 miRNAs (miR-1, miR-19a, miR-126, miR-133a, and miR-223)	Plasma	Baseline visit, 6-36 months after qualifying vascular event	Patients with stable vascular disease (CAD or post-ischemic stroke)	None	Prognosis	Low expression of miR-1, miR-19a, miR-126, miR-133a, and miR-223 was associated with increased 5-year mortality; low miR-19a remained the strongest independent predictor, particularly in CAD patients	Low miR-19a: HR = 2.27 (95%CI: 1.59-3.23) for 5-year all-cause mortality; in CAD patients, HR = 3.00 (95%CI: 1.77-5.08)
Mick et al[44], 2017	PCS (United States)	2763	6 miRNAs (hsa-miR-877-5p, hsa-miR-124-3p, hsa-miR-320d, hsa-miR-656-3p, hsa-miR-3615, and hsa-miR-941)	Plasma	At exam 8 baseline	Patients with prevalent or incident stroke	None	Diagnosis/prediction	Prevalent stroke was associated with lower expression of hsa-miR-877-5p, hsa-miR-124-3p, and hsa-miR-320d. For incident stroke, hsa-miR-656-3p and hsa-miR-3615 were associated with reduced risk, whereas hsa-miR-941 was associated with increased risk	Prevalent stroke: MiR-877-5p OR = 0.10 (95%CI: 0.05-0.21), miR-124-3p OR = 0.14 (95%CI: 0.05-0.35), miR-320d OR = 0.14 (95%CI: 0.05-0.39); Incident stroke: MiR-656-3p HR = 0.26 (95%CI: 0.13-0.52), miR-3615 HR = 0.36 (95%CI: 0.21-0.62), miR-941 HR = 3.06 (95%CI: 1.65-5.67)
Hypertension
Yildirim et al[46], 2019	Cross sectional study (Turkey)	64	1 miRNA (miR-21)	Plasma	At study evaluation after 12-hour fasting	Patients diagnosed with stage 1 HTN	Healthy controls	Diagnosis/association with asymptomatic organ damage	Plasma miR-21 was increased in hypertensive patients and was higher in those with asymptomatic organ damage; miR-21 correlated positively with blood pressure, microalbuminuria, carotid intima-media thickness, and CRP	For asymptomatic organ damage: AUC = 0.898; sensitivity 81.3%; specificity 81.2%, cut-off 44.6 (P < 0.001)
Kara et al[47], 2021	CCS (Turkey)	82	2 miRNAs (miR-21, and miR-155)	Serum	8-12 hours (at study evaluation)	Patients with resistant hypertension or newly diagnosed hypertension	Healthy controls	Diagnosis/association with resistant hypertension	MiR-21 and aldosterone levels were significantly higher in resistant hypertension, whereas miR-155 did not differ between groups	MiR-21 predicted resistant hypertension at a cut-off of 9.6 copies/μL with 95% sensitivity and 71% specificity (AUC = 0.823, 95%CI: 0.72-0.92)
Cardiogenic shock
Jäntti et al[10], 2019	Prospective multinational observational study	179	1 miRNA (miR-423-5p)	Plasma	Baseline	Patients diagnosed with CS	None	Prognosis	Higher baseline miR-423-5p levels were associated with 90-day non-survival and with markers of hypoperfusion, including higher lactate and ALT and lower cardiac index and eGFR	MiR-423-5p above median independently predicted 90-day all-cause mortality: Adjusted HR = 1.9 (95%CI: 1.2-3.2), P = 0.01
Mäntylä et al[45], 2025	Prospective multicenter observational study	165	10 miRNAs studied; miR-20b-5p showed prognostic significance	Plasma	Baseline	Patients with cardiogenic shock	None	Prognosis	Higher baseline miR-20b-5p levels were associated with in-hospital and 90-day survival; patients with higher miR-20b-5p also had lower lactate, NT-proBNP, and growth differentiation factor-15	Highest quartile of miR-20b-5p independently predicted 90-day survival: Adjusted HR = 2.47 (95%CI: 1.16-5.28), P = 0.019

PCS: Prospective cohort study; miRNA: MicroRNA; ICM: Ischemic cardiomyopathy; HF: Heart failure; ACS: Acute coronary syndrome; CVD: Cardiovascular disease; NR: Not reported; ROC: Receiver operating characteristic; AUC: Area under the curve; STEMI: ST-segment elevation myocardial infarction; NSTEMI: Non-ST-segment elevation myocardial infarction; LVEDV: Left ventricular end-diastolic volume; LVESV: Left ventricular end-systolic volume; CCS: Case-control study; CAD: Coronary artery disease; MACE: Major adverse cardiovascular events; PCI: Percutaneous coronary intervention; LV: Left ventricle; BNP: B-type natriuretic peptide; LVEF: Left ventricular ejection fraction; UA: Unstable angina; HDL: High-density lipoprotein; CABG: Coronary artery bypass grafting; CCTA: Coronary computed tomography angiography; HDL-C: High density lipoprotein cholesterol; LDL-C: Low-density lipoprotein cholesterol; HTN: Hypertension; CRP: C-reactive protein; CS: Cardiogenic shock; ALT: Alanine aminotransferase; eGFR: Estimated glomerular filtration rate; ALVR: Adverse left ventricular remodeling.

Quality assessment

The methodological quality of the included studies was assessed using the Newcastle-Ottawa Scale (NOS)[13]. The scale evaluates three domains: Selection, comparability, and outcome/exposure. Each domain was assigned using a star-based system adapted from Shamsrizi et al[14], and the total number of stars for each study was mapped to the corresponding Agency for Healthcare Research and Quality rating, as presented in Table 2.

Table 2 A Newcastle Ottawa scale summarizing the methodological quality of the included studies.

Ref.	Selection	Comparability	Reporting	AHRQ standard
Klenke et al[15], 2018	3	2	3	Good
Shah et al[16], 2018	3	2	3	Good
Pilbrow et al[17], 2014	3	2	3	Good
Shen et al[18], 2021	3	2	3	Good
Biener et al[19], 2022	3	2	3	Good
Eyyupkoca et al[20], 2022	2	1	2	Fair
Cai et al[21], 2024	3	2	3	Good
Grabmaier et al[22], 2017	3	2	3	Good
Wang et al[23], 2011	3	2	3	Good
Devaux et al[24], 2015	3	2	3	Good
Jakob et al[25], 2017	3	2	3	Good
Matsumoto et al[26], 2013	3	2	3	Good
Cortez-Dias et al[27], 2016	2	2	3	Fair
He et al[28], 2014	3	2	3	Good
Bauters et al[29], 2013	3	2	3	Good
Hromádka et al[30], 2019	3	2	3	Good
Zeller et al[31], 2014	3	2	3	Good
Simionescu et al[32], 2016	3	2	3	Good
Wang et al[33], 2017	3	2	3	Good
Zhong et al[34], 2018	3	2	3	Good
Polyakova et al[35], 2021	3	2	3	Good
Karakas et al[36], 2017	3	2	3	Good
de Gonzalo-Calvo et al[37], 2019	3	2	3	Good
Wang et al[38], 2014	3	2	3	Good
Wang et al[39], 2020	3	2	3	Good
Yuan et al[40], 2018	3	2	3	Good
Opstad et al[41], 2017	3	2	3	Good
Iusupova et al[42], 2024	3	2	3	Good
Mayer et al[43], 2019	3	2	3	Good
Mick et al[44], 2017	3	2	3	Good
Mäntylä et al[45], 2025	3	2	3	Good
Yildirim et al[46], 2019	3	2	3	Good
Kara et al[47], 2021	3	2	3	Good
Jäntti et al[10], 2019	3	2	3	Good

AHRQ: Agency for Healthcare Research and Quality.

RESULTS

The characteristics of the included studies

This review included 34 studies summarizing data from 14893 CVD patients and healthy controls. The studies were conducted in various settings worldwide. Plasma and serum were the primary sources of the samples. Table 1 summarizes the findings and characteristics of the included studies.

Methodological quality of the included studies

Based on the NOS, most of the included studies demonstrated good methodological quality. This rating primarily reflected appropriate selection of cohorts, adequate comparability between participants with and without CVD, and clear outcome reporting. Studies rated as fair quality were mainly limited by less rigorous cohort selection, which reduced comparability between groups. A detailed summary of the quality assessment for each study is provided in Table 2.

Outcomes

Numerous circulating miRNAs were identified with various diagnostic and prognostic roles across different CVD phenotypes.

Cardiomyopathy and HF: Two studies (pooled n = 2854) investigated the prognostic utility of circulating miRNAs in HF and cardiomyopathy[15,16]. Four miRNAs (miR-192, miR-106b-5p, miR-17-5p, and miR-20a-5p) were investigated. In the community-based Framingham cohort, higher circulating levels of miR-17, miR-20a, and miR-106b were associated with a lower risk of incident HF during follow-up (miR-17: HR = 0.84, 95%CI: 0.72-0.99; miR-20a: HR = 0.86, 95%CI: 0.73-1.00; miR-106b: HR = 0.85, 95%CI: 0.73-0.99)[16]. Lower plasma levels of these three miRNAs are predictive of long-term incident HF and are linked to increased left ventricular mass, likely due to their shared regulation of transforming growth factor-β signaling, apoptosis, and cell cycle pathways[16]. In addition, miR-192-5p serves as a potent prognostic marker in established ischemic cardiomyopathy (ICM), where its high expression is significantly associated with p53-induced cardiomyocyte apoptosis[15]. ICM patients with high miR-192 levels reached a median age at death of 67 years, 17 years earlier than those with low expression (84 years), whereas no such association was found for non-ischemic patients (log-rank P = 0.003; Cox regression P = 0.014)[15].

ACS: Eighteen studies, with a pooled sample size of 5781 ACS patients and controls, evaluated miRNAs for diagnosis or prognosis in ACS[17-34]. Across these studies, numerous candidate miRNAs were identified in general ACS, AMI/ST-segment elevation myocardial infarction (STEMI)/non-ST-segment elevation myocardial infarction, and UA. In general ACS, higher baseline miR-4286 was associated with an increased risk of incident ACS[18], while lower baseline miR-652 was associated with HF readmission after ACS[17]. In patients with suspected ACS, higher miR-21-5p and lower miR-122-5p were associated with all-cause mortality, whereas higher miR-126-3p was associated with the composite endpoint of all-cause mortality, AMI, and stroke[19]. In AMI, higher levels of miR-133 and miR-328 were consistently observed and showed diagnostic utility[21,23], while miR-208b, miR-499, and miR-320a were also increased in AMI and supported diagnosis and short-term risk stratification[24]. Several miRNAs were associated with adverse remodeling and post-AMI outcomes: MiR-199a-5p, miR-23b-3p, miR-483-5p, miR-26b-5p, miR-652-3p, miR-301a-3p, miR-374a-5p, and miR-423-5p were linked to adverse left ventricular remodeling[20]; miR-26b-5p, miR-320a, and miR-660-5p were associated with major adverse cardiovascular events (MACE) after STEMI and improved discrimination when added to clinical models[25]; and a higher miR-122-5p/miR-133b ratio predicted death, recurrent myocardial infarction, and adverse cardiovascular events after primary percutaneous coronary intervention[27]. In addition, higher levels of miR-328 and miR-134 were associated with cardiogenic death or HF within 6 months[28], while miR-192, miR-194, and miR-34a were upregulated in patients who developed ischemic HF after AMI[26].

The miR-1 and miR-29b correlated with infarct volume changes[22], whereas miR-133a and miR-423-5p were not associated with left ventricular remodeling despite temporal changes after MI[29]. Higher miR-1, miR-133a, and miR-499 levels were observed in nonsurvivors, although these differences were not statistically significant[30]. Among UA studies, an 8-miRNA signature was identified, and a 3-miRNA panel consisting of miR-132, miR-150, and miR-186 showed strong discriminatory performance for UA[31]. Other studies showed that miR-223, miR-92a, and miR-486 were increased in hyperglycemic ACS and discriminated ACS from stable angina[32], whereas miR-126-5p, miR-142-3p, miR-17-5p, and miR-340-3p were upregulated in UA and STEMI, with miR-142-3p and miR-17-5p showing the best diagnostic performance[34]. Finally, in non-ST-elevation ACS, lower miR-3135b, miR-126-5p, miR-142-5p, and miR-144-5p were associated with chronic HF, while higher GRACE risk scores were associated with lower miR-3135b and higher miR-28-3p[33].

CAD: Across eight studies with a pooled sample size of 4919, 32 unique miRNAs were investigated in CAD[35-42]. Diagnostic and prognostic signals were heterogeneous, but several patterns emerged. miR-203 was associated with greater angiographic disease burden and multivessel CAD, while miR-27a and miR-133a were also elevated in CAD compared with controls[35]. In patients with documented CAD, higher circulating levels of miR-132, miR-140-3p, and miR-210 were associated with cardiovascular death, supporting a prognostic role for selected miRNAs in secondary prevention[36]. In suspected stable CAD, lower circulating levels of several vascular miRNAs, particularly miR-143-3p, were associated with greater atherosclerotic extent and severity, although overall discrimination remained modest[37]. For atypical CAD, a 5-miRNA panel consisting of miR-487a, miR-29b, miR-502, miR-208, and miR-215 showed strong diagnostic performance[38]. In a general population cohort, higher miR-423-3p levels were associated with a lower risk of incident CAD events and improved prediction beyond traditional risk factors[39]. Sex-specific findings were also reported, with miR-18a elevated in female CAD patients but not in male patients[40]. In stable CAD, current smoking was associated with lower circulating miR-21 expression[41]. Finally, in patients with obstructive and non-obstructive CAD phenotypes, miR-34a and miR-222 were elevated, whereas miR-145 emerged as an independent predictor of ischemia/angina with non-obstructive coronary arteries[42].

Stroke: Across two studies with a pooled sample size of 3589, 11 unique miRNAs were investigated in relation to stroke[43,44]. In stable patients with vascular disease, lower circulating levels of miR-1, miR-19a, miR-126, miR-133a, and miR-223 were associated with increased 5-year mortality, with miR-19a emerging as the strongest independent predictor, particularly among patients with CAD[43]. In a large community-based cohort, several extracellular miRNAs were associated with stroke risk. miR-877-5p, miR-124-3p, and miR-320d were associated with prevalent stroke, whereas miR-656-3p, miR-3615, and miR-941 were associated with incident stroke during follow-up; among these, miR-656-3p was associated with lower incident stroke risk, while miR-941 was associated with higher incident stroke risk[44].

CS: Across two studies with a pooled sample size of 344, circulating miRNAs were evaluated for prognostic assessment in CS[10,45]. Increased miR-423-5p independently predicted 90-day mortality, whereas increased miR-20b-5p independently predicted 90-day in-hospital survival[10,45].

HTN: Across two studies with a pooled sample size of 146, miR-21 and miR-155 were investigated in HTN[46,47]. miR-155 showed no diagnostic or prognostic value. The miR-21, however, was a useful prognostic biomarker: Higher levels predicted resistant HTN and were associated with asymptomatic organ damage.

DISCUSSION

This systematic review synthesized evidence on circulating miRNAs for diagnosis and prognosis across the CVD spectrum, including ACS (AMI and UA), stable CAD, cardiomyopathy/HF, stroke, CS, and HTN (Figure 3). Signals were most consistent in ACS and CAD, where several miRNAs (for example, miR-133a, miR-328, the miR-19 family, and miR-423-5p) demonstrated promising diagnostic or prognostic associations, whereas fewer data were available for stroke, HTN, and CS. Collectively, these findings support the concept that circulating miRNAs are likely to complement, rather than replace, existing tools by capturing cardiomyocyte injury, endothelial activation, inflammation, and remodeling.

Figure 3 Overview of the systematic review workflow for circulating microRNAs in cardiovascular disease. A: Patients with cardiovascular disease (CVD) release circulating microRNAs (miRNAs) into the blood. Eligible studies were identified through database searches of PubMed and CENTRAL, followed by screening and final study selection; B: Data from the 34 included studies were extracted, processed, and categorized by CVD phenotype, with representative miRNAs listed for each category. The characteristics of the included studies. Created in BioRender (Supplementary material).

A quantitative synthesis was not appropriate due to heterogeneity in miRNA targets, biospecimen (plasma vs serum), sampling time points relative to symptom onset or index events, normalization approaches, assay platforms, and endpoint definitions. Single-center designs and modest sample sizes further limit generalizability and likely contribute to discordant findings. Prior work has highlighted the potential utility of circulating miRNAs in CVD[48,49], and this review extends that literature by summarizing additional evidence in cardiomyopathy, HF, HTN, and CS. Across ACS studies alone, approximately 50 unique miRNAs were investigated, and ongoing research in both human and preclinical models continues to expand the candidate pool[49,50]. However, many studies focused on highly expressed miRNAs and reported diagnostic or prognostic associations without external validation, and only a minority attempted to replicate previously reported signals; consequently, evidence for most candidates derives from single studies, which limits generalizability[10,15,43].

Reporting completeness varied across studies. Sampling time relative to symptom onset or index procedures was often not reported, and biospecimen matrices (plasma vs serum) differed between studies, reducing cross-study comparability. Control groups were variably defined (healthy volunteers vs clinical controls), and only a subset of studies reported accuracy metrics such as AUC or prespecified thresholds; where available, study-level AUCs typically fell in the moderate-to-good range (approximately 0.76-0.91). Although most studies achieved good NOS ratings, limitations in cohort selection and comparability among those rated as fair indicate a risk of selection bias and residual confounding. Furthermore, the evidence base was geographically concentrated in European and East Asian populations, with limited representation from other regions, which may constrain global generalizability.

Identifying a small and relevant set of miRNAs with diagnostic and prognostic relevance in CVD is essential because clinical adoption of miRNA biomarkers remains in an early developmental stage[51]. Translation into practice requires extensive validation of individual miRNAs and panels in large, rigorously designed studies[52]. While some miRNAs in our review were evaluated in three or more studies, most were examined in only a single investigation; further research is therefore recommended to validate promising candidates across diverse patient cohorts, clinical settings, and assay platforms.

This review used a systematic approach to summarize the current evidence on circulating miRNAs as biomarkers for CVD, but several limitations should be acknowledged. Quantitative estimates of diagnostic or prognostic performance were not feasible because of substantial heterogeneity in miRNA targets, biospecimen type (plasma vs serum), sampling time points, normalization methods, assay platforms, and outcome definitions. Most primary studies were also single-center and enrolled modest sample sizes, which limits robustness and external validity. Restriction to English-language articles may additionally have led to omission of relevant evidence published in other languages.

A formal assessment of publication bias was not feasible within disease-specific subgroups because most subgroups included few studies and marked methodological heterogeneity. Nonetheless, publication bias remains a concern because many studies assessed multiple candidate miRNAs, emphasized significant associations, and only rarely included external validation or replication of null findings.

The available evidence was derived predominantly from European, East Asian, and North American cohorts, with limited representation from other geographic regions and racially diverse populations, which may restrict generalizability. Future multicenter prospective cohorts using standardized sampling, normalization, and assay methods, prespecified miRNA panels, external validation, and broader geographic and racial/ethnic representation are needed to strengthen the evidence base.

CONCLUSION

This review identifies multiple circulating miRNAs with promising diagnostic and prognostic associations across CVD, including stroke, ACS, CAD, HTN, cardiomyopathy/HF, and CS. Nevertheless, many candidates lack external validation, and considerable heterogeneity across studies limits generalizability and hinders immediate clinical implementation. Further prospective, multicenter studies with standardized methods and transparent reporting are needed to validate these biomarkers before these miRNAs can be reliably incorporated into routine clinical practice.

ACKNOWLEDGEMENTS

We thank Yashesh Shah for designing and providing the figure illustrating microRNA biogenesis and gene silencing.