Plasma miR-502-5p and coronary artery vasospasm: An exploratory study

Abstract

Background

Coronary artery vasospasm (CAV) is a major cause of myocardial ischemia in patients without obstructive coronary disease. Its detection remains challenging due to the lack of biomarkers. We investigated circulating microRNAs (miRNAs) as potential biomarkers for CAV.

Methods

The ANFIBIO study is a prospective, multicenter cohort study that enrolled consecutive patients presenting with chest pain of presumed coronary origin (ClinicalTrials.gov identifier: NCT05374694). Patients were referred for invasive physiological evaluation. For this substudy, only those who underwent a coronary vasospasm test were included (n = 70). Circulating miRNA profiling was performed using RT-qPCR.

Results

The mean age of participants was 66.5 years and 40.0% were women. The prevalence of hypertension, dyslipidemia, and diabetes mellitus was 72.9%, 61.4% and 20.0%, respectively. Patients with CAV showed a higher use of oral nitrates. Notably, most individuals with undetectable plasma levels of miR-502-5p exhibited CAV (60%). Accordingly, detectability was used as the primary representation of miR-502-5p in further analyses. Adding miR-502-5p to a clinical model (age, sex, smoking status, nitrate use and transferrin concentration) was associated with a numerically higher AUC (0.74–0.81) and improved reclassification metrics (NRI = 0.533; IDI = 0.079).

Conclusions

In this cohort, plasma miR-502-5p was associated with CAV and showed exploratory incremental value when added to a clinical model. These findings should be considered hypothesis-generating and require further validation in larger and independent cohorts.

Resumen

Antecedentes

El vasoespasmo de arteria coronaria (CAV) es una causa importante de isquemia miocárdica en pacientes sin enfermedad coronaria obstructiva. Su detección es difícil por la ausencia de biomarcadores. En este trabajo investigamos los microRNA (miRNA) circulantes como posibles biomarcadores de CAV.

Métodos

ANFIBIO es un estudio prospectivo, multicéntrico, con pacientes consecutivos con dolor torácico y sospecha de etiología coronaria (ClinicalTrials.gov: NCT05374694). Los pacientes fueron remitidos para una evaluación fisiológica invasiva. En este subestudio se incluyeron los que realizaron prueba de vasoespasmo (n = 70). El análisis de miRNA circulantes se realizó mediante RT-qPCR.

Resultados

La edad media de los participantes fue de 66,5 años y el 40,0% eran mujeres. La prevalencia de hipertensión, dislipidemia y diabetes mellitus fue del 72,9%, 61,4% y 20,0%, respectivamente. Los pacientes con CAV presentaron un mayor uso de nitratos orales. Entre los individuos con niveles plasmáticos indetectables de miR-502-5p, el 60% presentó CAV. En consecuencia, la detectabilidad se utilizó como la principal forma de representación de miR-502-5p en los análisis posteriores. Su incorporación a un modelo clínico (edad, sexo, tabaquismo, uso de nitratos y concentración de transferrina) se asoció con una tendencia hacia una mayor discriminación (AUC 0,74→0,81) y con una mejora significativa en las métricas de reclasificación (NRI = 0,533; IDI = 0,079).

Conclusiones

En esta cohorte, miR-502-5p plasmático se asoció con CAV y aportó un valor incremental al integrarlo en un modelo clínico, a nivel exploratorio. Estos hallazgos son generadores de hipótesis y requieren validación adicional en cohortes independientes y de mayor tamaño.

Introduction

Coronary artery vasospasm (CAV), characterized by transient and partial or complete coronary artery occlusion, is one of the leading causes of myocardial ischemia and angina, particularly in patients without obstructive coronary artery disease.¹ Vasospastic disease significantly increases the risk of myocardial infarction, life-threatening arrhythmias and sudden cardiac death.^1,2 Despite its clinical importance, CAV diagnosis remains challenging. Coronary angiography with provocative tests is currently the gold standard for the diagnosis of CAV.¹ Nevertheless, the procedure, which is not widely available, carries risks and requires experienced personnel. Circulating biomarkers could aid in clinical decision-making.

Circulating microRNAs (miRNAs) have emerged as next-generation biomarkers with transformative potential in several conditions.³ These small non-coding RNA molecules, which regulate gene expression at a post-transcriptional level, are actively released into the bloodstream and might reflect pathological processes.⁴ Their stability in biological fluids, resistance to degradation and cost-effective quantification using standard laboratory techniques make them ideal candidates for clinical applications.⁵ Several miRNA-based biomarkers are already being evaluated in clinical trials and some have been approved for diagnostics.⁶

Given their potential to reflect disease-specific molecular alterations, circulating miRNAs could serve as non-invasive indicators of vasospasm. To date, no study has directly evaluated circulating miRNAs in the context of coronary artery vasospasm, underscoring a major gap in current knowledge. This absence of prior evidence highlights both the novelty and the exploratory nature of our investigation. Using a real-world clinical scenario, we therefore aimed to identify circulating miRNAs associated with CAV and to evaluate their diagnostic performance.

Materials and methods

This is a substudy of the ANFIBIO project (ClinicalTrials.gov identifier: NCT05374694), a multicenter, prospective investigation conducted across four centers in Spain between May 2022 and November 2023. The study adhered to the ethical principles outlined in the Declaration of Helsinki and was approved by the “Comité de Ética de Investigación con Medicamentos del Hospital Arnau de Vilanova deLleida”. All participants provided written informed consent before study inclusion.

Consecutive patients with suspected angina referred for coronary angiography were enrolled according to the recommendations of the European Society of Cardiology (ESC).⁷ The inclusion and exclusion criteria have been previously described⁸ and defined in Supplemental Table S1. Invasive physiological evaluation was performed, including fractional flow reserve and index of microcirculatory resistance. In patients with evidence of isolated microvascular involvement or normal coronary indices, vasospasm test was performed with acetylcholine. Briefly, CAV was defined as the presence of epicardial spasm (arterial diameter constriction ≥90%, chest pain, and electrocardiographic changes suggestive of myocardial ischemia) or microvascular spasm (arterial diameter constriction < 90%, chest pain, and electrocardiographic changes suggestive of myocardial ischemia).

Data collection methods, sample collection procedures and laboratory analyses have been previously described.⁸

A panel of 19 circulating miRNAs was selected for analysis based on a comprehensive literature review conducted in the context of the ANFIBIO project.⁸ These candidates have been described as biomarkers of chronic coronary syndrome and have been linked to molecular pathways potentially altered in this condition across in vitro, in vivo and patient-based studies. The complete list of analyzed miRNAs is provided in Supplemental Table S2.

To ensure unbiased processing, miRNA quantification was performed in a fully blinded manner, with researchers unaware of the study group assignments. RNA isolation and RT-qPCR were performed as previously described.^9,10 Briefly, total RNA was isolated from 180 μL of plasma using a miRNeasy Serum/Plasma Advanced Kit (Qiagen). miRNA quantification was carried out using the miRCURY LNA Universal RT microRNA PCR System (Qiagen). Complementary DNA was synthesized with the miRCURY LNA RT Kit in 10 μL reaction volumes on a Veriti™ 96-Well Thermal Cycler under the following thermal profile: 42 °C for 60 min, followed by 95 °C for 5 min and cooling at 4 °C. A synthetic RNA spike-in (UniSp6) was added to monitor reverse transcription efficiency. Synthesized cDNA was stored at −20 °C until analysis. Quantitative PCR was performed using 384-well miRCURY LNA Custom Panels (Qiagen) containing primers for the selected miRNAs and spike-in controls. Reactions (10 μL) were run on a QuantStudio™ 7 Flex Real-Time PCR System using the following cycling conditions: 95 °C for 2 min, followed by 40 cycles of 95 °C for 10 s and 56 °C for 1 min. Specificity was confirmed through melting curve analysis. Only single-product amplification curves without primer-dimer artifacts were considered valid. Fluorescence data were analyzed using QuantStudio Software v1.3.¹¹ Relative expression was calculated using the 2^−ΔCq method, normalized to cel-miR-39-3p and log-transformed for downstream analysis. Cq values above 35 cycles were classified as undetectable. For statistical analyses, undetectable values were handled using a conservative imputation (minimum observed value for each miRNA). This approach allows inclusion of low-expression samples.¹¹ Technical controls included non-template reactions and RNA spike-ins to verify extraction quality, reverse transcription consistency and absence of inhibitors. Samples failing these quality control criteria were excluded.

Baseline characteristics were summarized using descriptive statistics. Continuous variables are reported as means with standard deviations (SD) or medians with percentiles [P25–P75], as appropriate. Categorical variables are presented as frequencies and percentages. Comparisons between groups for continuous variables were performed using either the Student's t test or the Mann–Whitney U test, depending on the distribution of the data. Categorical variables were compared using the Chi-squared test or Fisher's exact test, as appropriate. Differential miRNA expression among groups was assessed using linear models with empirical Bayes statistics (limma). Given the targeted nature of the panel, results were interpreted as exploratory; therefore, no formal multiple-testing correction was applied. Given the high proportion of undetectable miR-502-5p measurements (Cq>35), this miRNA was evaluated in terms of detectability (detectable vs non-detectable). The association between miR-502-5p detectability and CAV was assessed using a contingency table and quantified using univariable logistic regression, reporting OR (95% CI) and p-values. We used multivariable logistic regression to construct the base clinical model for CAV, including age and sex a priori and additional covariates selected according to the prespecified univariable screening criterion. Candidate clinical covariates were screened using an permissive threshold in univariable comparisons with CAV status. The p-value < 0.10 threshold was used only for preliminary screening and was not interpreted as statistical significance. Smoking status was modeled as a binary variable (ever smoker [former/current] vs never smoker). Model performance was assessed using Cox & Snell pseudo-R². Discrimination was evaluated using receiver operating characteristic (ROC) curves and the area under the curve (AUC, with 95% CI). AUCs were compared using DeLong's test. Reclassification performance was assessed with the Integrated Discrimination Improvement (IDI) and Net Reclassification Improvement (NRI) indices. Statistical significance was set at a two-tailed p-value < 0.05. All analyses were performed using R statistical software (version 4.4.2, available at www.r-project.org).

Results and discussion

The characteristics of the study population according to the presence of CAV are summarized in Table 1. The mean age of participants was 66.5 years and 40.0% were female. The prevalence of hypertension, dyslipidemia and diabetes mellitus was 72.9%, 61.4% and 20.0%, respectively. Patients with CAV showed a higher prescription of oral nitrates. Regarding laboratory parameters, the CAV group showed a trend toward lower transferrin levels. No differences were observed in cardiac biomarkers including hs-Troponin I, NT-proBNP or CRP.

Table 1.

Demographic and clinical characteristics of the study population.

Patient characteristics	ALLN = 70	Non-CAVN = 37	CAVN = 33	p-Value
Age, mean (SD), years	66.5 (9.67)	67.2 (9.56)	65.8 (9.88)	0.542
Female, n (%)	28 (40.0)	17 (45.9)	11 (33.3)	0.406
Cardiovascular risk factors
Smoking, n (%)				0.085
Never smoker	31 (44.3)	21 (56.8)	10 (30.3)
Ex-smoker	31 (44.3)	13 (35.1)	18 (54.5)
Current smoker	8 (11.4)	3 (8.1)	5 (15.2)
Diabetes, n (%)	14 (20.0)	9 (24.3)	5 (15.2)	0.510
Hypertension, n (%)	51 (72.9)	30 (81.1)	21 (63.6)	0.171
Dyslipidemia, n (%)	43 (61.4)	25 (67.6)	18 (54.5)	0.384
Main cardiovascular comorbidities
Heart failure, n (%)	4 (5.71)	2 (5.41)	2 (6.06)	1.000
Previous acute coronary syndrome, n (%)	15 (21.4)	6 (16.2)	9 (27.3)	0.404
Pharmacological treatment
Beta-blockers, n (%)	38 (55.9)	23 (65.7)	15 (45.5)	0.151
Calcium-channel blocker, n (%)	23 (33.8)	15 (42.9)	8 (24.2)	0.172
Nitrates, n (%)	18 (26.5)	5 (14.3)	13 (39.4)	0.038
Statins, n (%)	56 (82.4)	31 (88.6)	25 (75.8)	0.286
Clinical characteristics
Exertional angina, n (%)	54 (79.4)	28 (80.0)	26 (78.8)	1.000
Angina at rest, n (%)	14 (20.6)	7 (20.0)	7 (21.2)	1.000
Analytical parameters
Estimated glomerular filtration rate, median [P25–P75], mL/min/1.73 m^2	80.6 [70.9–89.0]	83.5 [72.7–90.0]	75.0 [67.0–86.3]	0.129
Total cholesterol, mean (SD), mg/dL	166 (45.9)	170 (45.0)	162 (47.4)	0.518
LDL-C, mean (SD), mg/dL	92.2 (37.1)	96.5 (34.3)	87.3 (40.1)	0.313
HDL-C, mean (SD), mg/dL	54.1 (13.6)	53.3 (13.5)	55.1 (13.8)	0.594
Triglycerides, median [P25–P75], mg/dL	92.0 [69.0–124]	90.0 [68.0–111]	98.0 [74.0–129]	0.410
Hs-troponin I, median [P25–P75], pg/mL	4.65 [3.0–8.1]	4.80 [2.9–8.1]	4.20 [3.1–7.8]	0.524
Ferritin, median [P25–P75], ng/mL	106 [62.2–212]	87.3 [45.9–202]	123 [85.8–218]	0.118
Transferrin, mean (SD), mg/dL	260 (46.9)	270 (52.7)	248 (36.9)	0.057
C-Reactive Protein, median [P25–P75], mg/L	1.50 [0.8–4.5]	1.50 [1.1–4.1]	1.50 [0.8–4.5]	0.520
NT-proBNP, median [P25–P75], pg/mL	119 [65.5–371]	143 [71.9–414]	117 [55.0–235]	0.377
Leukocytes, median [P25–P75], ×10^9/L	6.61 [5.7–7.7]	6.88 [6.1–7.6]	6.41 [5.3–7.8]	0.251
Hemoglobin, mean (SD), g/dL	14.4 (1.4)	14.4 (1.4)	14.5 (1.5)	0.583
Hematocrit, mean (SD), %	42.5 (3.8)	42.5 (3.7)	42.6 (3.9)	0.917
Platelets, mean (SD), ×10^9/L	215 (56.3)	223 (64.2)	207 (45.5)	0.217
Angiographical and physiological characteristics
Vasospasm				<0.001
None, n (%)	37 (52.9)	37 (100)	0 (0.0)
Microvascular vasospasm, n (%)	6 (8.6)	0 (0.0)	6 (18.2)
Macrovascular vasospasm, n (%)	27 (38.6)	0 (0.0)	27 (81.8)

CAV: coronary artery vasospasm; FFR: fractional flow reserve; HDL-C: high-density lipoprotein cholesterol; hs: high-sensitivity; IMR: index of microcirculatory resistance; LDL-C: low-density lipoprotein cholesterol; NT-proBNP: N-terminal pro-B-type natriuretic peptide; SD: standard deviation.

Next, we assessed the levels of the selected miRNA panel. Among the candidates, miRNA expression analysis revealed a significant reduction in plasma miR-502-5p levels in patients with CAV (Fig. 1A). Because miR-502-5p showed a high rate of undetectable values (Cq>35), we evaluated this marker primarily in terms of detectability. Fig. 1B shows the proportion of patients with CAV within each detectability stratum. The corresponding contingency table with absolute counts and effect estimates is provided in Supplemental Table S3, together with the univariable logistic regression results for miR-502-5p detectability. Consistently, CAV was more frequent among patients with non-detectable miR-502-5p (25/42; 59.5%) than among those with detectable miR-502-5p (8/28; 28.6%). miR-502-5p detectability was less common in CAV cases than in non-CAV participants (8/33; 24.2% vs 20/37; 54.1%).

Fig. 1.

Role of miR-502-5p as biomarker of coronary vasospasm (CAV). (A) Expression levels of miR-502-5p between CAV and no CAV patients. A violin plot illustrates the distribution of expression levels between the two groups, with individual data points and quartiles (25th, 50th and 75th percentiles). Expression levels are represented as log₂(2^−ΔCq) and multiplied by 10⁶ for visualization. (B) Percentage of patients with CAV within each miR-502-5p detectability stratum. The bar plot shows the percentage of patients with CAV in the detectable and non-detectable miR-502-5p categories (the percentage is indicated on each bar). (C) ROC curves comparing the clinical model (red curve) with the clinical model incorporating miR-502-5p as a binary variable: detectable vs non-detectable (blue curve). The discriminative performance of both models is quantified by the AUC. A DeLong's test was performed to compare the AUCs of both models. Reclassification analyses, IDI and NRI indexes, were implemented to evaluate the added value of miR-502-5p.

Demographic, clinical and biochemical data were stratified according miR-502-5p detectability. No consistent differences were observed across most baseline characteristics between groups (data not shown), except for cancer prevalence, which differed according to miR-502-5p detectability (25.0% vs 2.3%, p-value = 0.006), a finding previously reported in the literature.¹²

Then, we evaluated the potential of miR-502-5p as a biomarker of CAV. In ROC analysis, miR-502-5p alone showed limited discriminative ability (AUC = 0.63), in line with previous studies of miRNA-based diagnostics.¹³ This finding suggests its evaluation as a complementary marker rather than as a standalone diagnostic tool. Consistent with this view, previous work from our group has shown that miRNAs tend to provide greater value when incorporated into models that include clinical variables rather than when assessed in isolation.¹⁴ Therefore, we proceeded to construct a clinical prediction model incorporating relevant covariates: age, sex, smoking status, nitrate use, transferrin levels and miR-502-5p detectability (Supplemental Fig. S1). Although cancer was associated with miR-502-5p detectability, it was not associated with CAV in this cohort and therefore was not included as an adjustment covariate in the model. Nevertheless, future studies should consider its potential impact on specificity.

The baseline clinical model achieved an AUC of 0.74 (Fig. 1C). When miR-502-5p was incorporated as a binary variable (detectable vs non-detectable), a trend toward improved discrimination was observed, increasing the AUC to 0.81 (DeLong's test p-value = 0.103) (Fig. 1C). Additionally, miR-502-5p improved reclassification compared with the clinical model alone (IDI = 0.079, p-value = 0.016; NRI = 0.533, p-value = 0.021) (Fig. 1C). Model fit was moderate (Cox & Snell pseudo-R² = 0.295).

Since CAV includes both macrovascular and microvascular vasospasm subtypes, we performed additional analyses to assess the model's performance within each group. In cases of microvascular vasospasm, the model achieved an AUC of 0.99 (Supplemental Fig. S2). Its predictive value for macrovascular vasospasm showed also a good discrimination (AUC = 0.79) (Supplemental Fig. S2). These subgroup analyses should be interpreted with caution due to the small sample size, particularly in the microvascular CAV group, which may inflate the AUC.

Given the limited number of studies evaluating miRNAs in CAV, direct comparisons with prior research are challenging. However, miR-502-5p has been reported as a biomarker for stable coronary artery disease (CAD).¹⁵ These findings align with prior research demonstrating that miRNAs may serve as valuable adjuncts to traditional clinical models, enhancing diagnostic precision and risk stratification in cardiovascular disease.¹⁶ When integrated with clinical variables, miR-502-5p may contribute to exploratory risk enrichment among patients referred for invasive evaluation of suspected vasospastic angina. A major limitation of current CAV diagnostic approaches is their reliance on specialized centers, highly trained personnel and expensive procedures.¹ miR-502-5p is a biomarker candidate associated with CAV that requires analytical standardization and validation in larger and independent cohorts before any potential clinical application.

Mechanistically, previous investigations have suggested a potential role for miR-502-5p in the pathogenesis of coronary artery disease.¹⁷ miR-502-5p has been implicated in vascular smooth muscle cell proliferation, angiogenesis, regulation of the cell cycle and modulation of inflammatory and oxidative stress pathways.^18–20 These processes are intimately linked to endothelial dysfunction and vascular smooth muscle hyperreactivity, central mechanisms in the development of coronary vasospasm. While these observations provide biological plausibility for our findings, it is important to emphasize that the primary objective of this study was not to establish causality, but rather to evaluate circulating miRNAs as potential non-invasive biomarkers. Accordingly, further mechanistic investigations will be necessary to clarify the causal role of miR-502-5p in vasospasm.

The strengths of this study include its real-world setting. Unlike case-control studies, which may overestimate biomarker utility due to extreme case selection, our cohort represents a more clinically relevant population. However, some limitations should be acknowledged. First, given that the cohort was limited to Spanish centers, generalizability to other populations may be restricted. Second, as with any parsimonious modeling strategy, residual confounding cannot be excluded. Third, the association between baseline oral nitrate prescription and a positive spasm test may seem counterintuitive. One possible explanation is that withdrawal of chronic nitrate therapy before testing may have influenced vasomotor tone. However, this remains speculative and was not specifically tested in our study. Fourth, we did not perform a formal multiple-testing correction. While this decision was made to reduce the risk of type II error in the context of a targeted panel, it may increase the chance of false-positive results. Fifth, given the limited number of CAV events relative to the number of predictors, model estimates may be subject to overfitting. Several predictors showed considerable uncertainty in their estimated effects, reflecting limited precision due to the modest sample size and event count. In addition, no internal validation (e.g., bootstrapping) was performed.

In conclusion, plasma miR-502-5p was associated with CAV in this cohort and showed preliminay incremental value when integrated with clinical variables. Given the modest effect size and the nature of the analyses, our findings should be interpreted as hypothesis-generating associations and not as evidence of diagnostic utility.

Authors’ contributions

All authors take responsibility for all aspects of the reliability and freedom from bias of the data presented and their discussed interpretation.

Funding

Project funded by Sociedad Española de Cardiología (SEC/FEC-INV-CLI 23/04). DdGC has received financial support from Instituto de Salud Carlos III (Miguel Servet 2020: CP20/00041) co-funded by European Union. FB is supported by the ICREA Academia Program. CIBERES (CB07/06/2008) is an initiative of the Instituto de Salud Carlos III.

Conflict of interest

The authors declare no competing interests.

Data availability statement

The data supporting the findings of this study are available from the corresponding author upon reasonable request.

Acknowledgments

Human sample manipulation was performed at the Cell Culture Facility, Universitat de Lleida (Lleida, Catalonia, Spain). Work supported by IRBLleida Biobank (B.000682) and Biobank and Biomodels Platform ISCIII PT23/00032.