Diagnostic and prognostic value of circulating microRNAs in patients with acute chest pain

Authors


Abstract

Objectives

To address the diagnostic value of circulating microRNAs (miRNAs) in patients presenting with acute chest pain.

Design

In a prospective, international, multicentre study, six miRNAs (miR-133a, miR-208b, miR-223, miR-320a, miR-451 and miR-499) were simultaneously measured in a blinded fashion in 1155 unselected patients presenting with acute chest pain to the emergency department. The final diagnosis was adjudicated by two independent cardiologists. The clinical follow-up period was 2 years.

Results

Acute myocardial infarction (AMI) was the adjudicated final diagnosis in 224 patients (19%). Levels of miR-208b, miR-499 and miR-320a were significantly higher in patients with AMI compared to those with other final diagnoses. MiR-208b provided the highest diagnostic accuracy for AMI (area under the receiver operating characteristic curve 0.76, 95% confidence interval 0.72–0.80). This diagnostic value was lower than that of the fourth-generation cardiac troponin T (cTnT; 0.84) or the high-sensitivity cTnT (hs-cTnT; 0.94; both < 0.001 for comparison). None of the six miRNAs provided added diagnostic value when combined with cTnT or hs-cTnT (ns for the comparison of combinations vs. cTnT or hs-cTnT alone). During follow-up, 102 (9%) patients died. Levels of MiR-208b were higher in patients who died within 30 days, but the prognostic accuracy was low to moderate. None of the miRNAs predicted long-term mortality.

Conclusion

The miRNAs investigated in this study do not seem to provide incremental diagnostic or prognostic value in patients presenting with suspected AMI.

Introduction

Acute myocardial infarction (AMI) is a major cause of death and disability worldwide. Its rapid and accurate diagnosis has been improved markedly by the recently introduced more sensitive cardiac troponin (cTn) assays [1, 2]. Nevertheless, even these sensitive cTn assays lack adequate sensitivity during the immediate phase after the onset of AMI. Serial testing of cTn is still indispensable in the majority of patients with acute chest pain [3] not least because of the increase in ‘troponin-positive’ results observed in patients with noncardiac causes of chest pain (NCCP) [4].

Recently, based on the presence of stable cardiomyocyte-enriched micro-RNAs (miRNAs) circulating in human peripheral blood, it has been suggested that certain miRNAs may serve as novel diagnostic markers of AMI [5-10]. Since the discovery of miRNAs in Caenorhabditis elegans in 1993 [11], more than 2000 human miRNAs have been cloned and sequenced. miRNAs are evolutionarily conserved, short, noncoding (approximately 22 nucleotides) RNA molecules involved in post-transcriptional gene regulation [5]. It is currently estimated that they control the expression of up to 50% of protein-coding genes [12, 13]. Although the complex biological functions of miRNAs with regard to their regulation of messenger RNA are incompletely understood [14, 15], they are known to be present in a tissue- and cell-specific manner [16] and are considered to regulate gene expression [14] mostly via messenger RNA destabilization resulting in endogenous gene silencing [15, 17]. miRNAs in plasma or serum are resistant to RNAse digestion and remain stable in the RNAse-rich environment of blood [18] as well as under extreme conditions such as during freeze–thaw cycles [19]. They are crucial for myriad cellular processes and are a prerequisite for normal cardiac function [20]. The release of miRNAs is thought to be not only a consequence of cell death and plasma membrane disruption but also an active response to ischaemia [21, 22]; miRNAs have been suggested as possible biomarkers of clinical value in the early diagnosis of AMI.

The present aim was to evaluate the diagnostic and prognostic value of cardiomyocyte-enriched miR-133a, miR-208b and miR-499, platelet-enriched miR-223, activated platelet-enriched miR-320a and red blood cell-enriched miR-451 in patients presenting with acute chest pain and suspected AMI in a large, prospective, observational, international, multicentre study. The rationale behind the choice of these miRNAs is as follows. Clinical studies have demonstrated markedly higher levels of miRNAs in patients with AMI, compared with healthy subjects, and in particular miR-133a [6, 9], miR-208b [7, 23, 24] and miR-499 [7-9]. Furthermore, platelet aggregation and activation are early events preceding thrombus formation resulting in AMI, and miR-223 and miR-320a may provide useful insights into these processes. Finally, we studied miR-451 which is abundant in red blood cells and reflects haemolysis.

Methods

Study design and population

Advantageous Predictors of Acute Coronary Syndrome Evaluation (APACE) is an ongoing prospective, international, multicentre study coordinated and designed by the University Hospital Basel (ClinicalTrials.gov number, NCT00470587). From April 2006 to June 2009, 1267 unselected patients presenting to the emergency department (ED) with symptoms suggestive of AMI with an onset or peak within the previous 12 h were recruited [1]. To attain a high rate of comparability of the results, patients were included in the analysis if simultaneous measurements of all six miRNAs, as described below, were performed at presentation. Due to interindividual difficulties in collecting blood samples and varying degrees of willingness amongst patients to undergo repetitive blood sampling, measurements of all six miRNAs were achieved simultaneously in 1155 patients; this group constitutes the study population. Patients with terminal kidney failure requiring dialysis were excluded. Also patients were excluded if the final diagnosis remained unclear after adjudication in combination with at least one measurement for which high-sensitivity cardiac troponin T (hs-cTnT) was ≥14 ng L−1 during serial sampling (= 38). The study was carried out according to the principles of the Declaration of Helsinki and approved by the local ethics committees at each institution. Written informed consent was obtained from all patients. The authors designed the study, collected and analysed the data, wrote the paper and are responsible for the integrity of the data and analyses and the decision to publish. The sponsors had no role in conducting the study or analysing the data.

Routine clinical assessment

The initial clinical assessment included clinical history, physical examination, 12-lead electrocardiography (ECG), continuous ECG monitoring, pulse oximetry, standard blood tests and chest radiography.

Measurements of cTn levels were performed at presentation and 6–9 h later or as long as clinically indicated [3]. Timing of measurements and treatment of patients were left to the discretion of the attending physician who was unaware of the centrally measured miRNA and hs-cTnT values and only aware of the locally available conventional cTn levels.

Adjudication of final diagnosis

Adjudication of the final diagnosis was performed centrally in the core laboratory (University Hospital Basel) for all patients twice: once according to conventional cTn levels used onsite (this method was used in the initial analyses to examine the performance of hs-cTn assays [25-27]) and once including levels of Roche hs-cTnT (Roche Diagnostics, Basel, Switzerland) to also take advantage of the higher sensitivity and higher overall diagnostic accuracy offered by hs-cTn assays [4] (this allows the additional detection of small AMIs that were missed by the adjudication based on conventional cTn assays). Two independent cardiologists reviewed all available medical records – patient history, physical examination, results of laboratory tests (including hs-cTnT levels), radiology, ECG, echocardiography, cardiac exercise test, lesion severity and morphology in coronary angiography – for each patient from the time of ED presentation to 90 days of follow-up. In cases of disagreement about the diagnosis, cases were reviewed and adjudicated in conjunction with a third cardiologist.

AMI was defined and cTn levels interpreted as recommended in current guidelines [28, 29]. In brief, AMI was diagnosed if there was evidence of myocardial necrosis in association with a clinical setting consistent with myocardial ischaemia. Myocardial necrosis was diagnosed by at least one cTn value above the 99th percentile (or for the conventional cTn assays above the 10% imprecision value if not fulfilled at the 99th percentile) together with a significant rise and/or fall [25, 29]. The criteria used to define rise and/or fall in conventional cTn and hs-cTnT are described in detail in the online Supplementary Material. Unstable angina was diagnosed in patients with normal cTn levels and typical angina at rest, those with a deterioration of a previously stable angina, and in cases of positive cardiac exercise testing or cardiac catheterization with coronary arteries with ≥70% stenosis. As we adjudicated the cause of the presentation to the ED (i.e. acute chest pain) and not the cause of increases in hs-cTnT, ‘stable coronary artery disease (CAD)’ was not considered a potential diagnosis; a patient with ‘stable CAD’ who deteriorated (acute chest pain) would therefore be classified as having ‘unstable angina’ or another suitable final diagnosis as a result of the adjudication process. The category of cardiac but noncoronary causes (CNCD) included myocarditis, pericarditis, heart failure, cardiac dysrhythmia and hypertensive emergency. A further category, noncardiac chest pain (NCCP), included musculoskeletal pain and gastro-oesophageal disorders. If all diagnostic procedures and tests were inconclusive then symptoms were classified as ‘of unknown origin’.

Follow-up and clinical end-points

Patients were contacted 3, 12 and 24 months after hospital discharge by telephone or by post/by email. All patients received long-term follow-up irrespective of AMI versus non-AMI status at initial presentation. Information regarding death was also obtained from the national registry on mortality. The primary end-point was all-cause mortality and AMI rate during follow-up.

Determination of plasma miRNAs

Circulating levels of six miRNAs (miR-133a, miR-208b, miR-223, miR-320a, miR-451 and miR-499) were measured at presentation. Total RNA was extracted from plasma samples using the mirVana PARIS kit (Ambion, Applied Biosystem, Lennik, Belgium) without enrichment for small RNAs. A mixture of three supplemented synthetic C. elegans miRNAs (Qiagen, Venlo, the Netherlands), which lacked sequence homology to human miRNAs, was added to plasma samples for correction of extraction efficiency. Potential genomic DNA contamination was eliminated by use of DNase (Qiagen). Reverse transcription of RNA was performed with the miScript reverse transcription kit (Qiagen). The resulting cDNA was diluted 10-fold before quantitative polymerase chain reaction (PCR) was performed with the miScript SYBR-green PCR kit (Qiagen). miRNA-specific miScript primer sets were obtained from Qiagen. Expression values were normalized using the mean threshold cycle (Ct) obtained from the spiked-in controls [calculation formula: 2exp(mean Ct spiked-in controls − Ct target miRNA)] and log-transformed. The detection limit of the PCR assay was defined as log transformation of the lowest detected miRNA level, divided by 10 (see Devaux et al. [7] for further details).

Biochemical analysis

Blood samples for determination of cardiac troponin T (cTnT, fourth-generation assay Roche), and hs-cTnT were collected at presentation to the ED and serially thereafter at 1, 2, 3 and 6 h. Serial sampling was discontinued when the diagnosis of AMI was certain and the patient needed to be transferred to the catheter laboratory for treatment. After centrifugation, samples were frozen at −80 °C until assayed in a blinded fashion in a dedicated core laboratory. hs-cTnT was assayed in serum in a blinded fashion using the Modular® Analytics E170 analyser (Roche Diagnostics). It has been determined that the limit of blank and limit of detection of hs-cTnT are 3 and 5 ng L−1, respectively, with an imprecision corresponding to the 10% coefficient of variation at 13 ng L−1 and the 99th percentile of a healthy reference population at 14 ng L−1 [26]. cTnT was measured using the Elecsys 2010 analyser with a limit of detection of 0.01 μg L−1, a 99th percentile cut-off value of <0.01 μg L−1 and a coefficient of variation of <10% at 0.035 μg L−1.

Statistical analysis

Categorical variables are presented as number and percentage, and continuous variables as median and interquartile range (IQR). Comparisons between groups were made using the chi-squared method, Mann–Whitney U-test or Kruskal–Wallis test. Receiver operating characteristic (ROC) curves were constructed to assess the sensitivity and specificity of miRNAs, hs-cTnT and cTnT for both diagnostic and prognostic purposes, and compared as recommended by DeLong et al. [30]. Likelihood ratios were used for comparisons of nested models [31], and a stepwise forward approach was used for the calculation of multivariate Cox proportional hazard regression analysis. Kaplan–Meier survival analysis was performed with log-rank values to assess statistical significance. We used integrated discrimination improvement (IDI) analysis, which is not dependent on certain risk groups because probability differences are used instead of categories. The IDI can be calculated as the difference between improvement in average sensitivity and changes in the average of ‘one minus specificity’ [32]. All hypothesis testing was two-tailed, and a P-value of <0.05 was considered statistically significant. Statistical analyses were performed using spss for windows 19.0 (SPSS Inc., Chicago, IL, USA) and medcalc 9.6.4.0 (MedCalc Software, Mariakerke, Belgium).

Results

Characteristics of patients

Baseline characteristics of the 1155 patients are shown in Table 1. The adjudicated final diagnosis was AMI in 19% of patients [= 224; 20% with ST-segment elevation MI (STEMI), 80% non-ST-segment elevation MI (NSTEMI)], unstable angina in 12%, CNCD in 14%, NCCP in 48% and symptoms of unknown origin in 6%.

Table 1. Baseline characteristics
 All patientsAcute myocardial infarctionP-value
(n = 1155)Yes (n = 224)No (n = 931)(AMI/non AMI)
  1. All values are reported as median (interquartile range) or numbers (%).

  2. aeGFR, estimated glomerular filtration rate. bACB, aortocoronary bypass. cPTCA, percutaneous transluminal coronary angioplasty.

Characteristics
Age, years63 (50–76)72 (61–80)61 (49–74)<0.001
Male sex – n (%)768 (67%)158 (71%)610 (66%)0.153
BMI26 (24–30)25.9 (23.8–29)27 (24–30)0.215
eGFR (mL min−1 1.73 m−2)a85 (68–101)73.6 (59–95)87 (70–102)<0.001
STEMI 45 (20%)  
NSTEMI 179 (80%)  
Risk factors – n (%)
Hypertension762 (66%)179 (80%)583 (63%)<0.001
Hypercholesterolemia515 (45%)117 (52%)398 (43%)0.01
Diabetes208 (18%)57 (26%)151 (16%)0.001
Current smoking281 (24%)56 (25%)225 (24%)0.794
History of smoking402 (35%)79 (35%)323 (35%)0.871
History – n (%)
Coronary artery disease436 (38%)106 (47%)309 (33%)<0.001
Previous myocardial infarction281 (24%)72 (32%)209 (23%)0.002
Previous ACBb110 (10%)33 (15%)77 (8%)0.003
Previous PTCAc276 (24%)55 (25%)221 (24%)0.797
Peripheral artery disease78 (7%)30 (13%)48 (5%)<0.001
Previous stroke63 (6%)24 (11%)39 (4%)<0.001
Vital status (median, interquartile range)
Systolic blood pressure, mmHg142 (127–160)142 (126–163)143 (127–160)0.994
Diastolic blood pressure, mmHg84 (75–93)83 (73–93)84 (75–93)0.261
Heart rate – beats per minute76 (66–89)79 (68–90)75 (65–89)0.055
Symptoms
Beginning before x h5 (3–12)6 (3–14)5 (3–12)0.082
Maximum before x h4 (2–7)4 (2–8)4 (2–7)0.123
Electrocardiographic findings – n (%)
Left bundle branch block42 (4%)18 (8%)24 (3%)<0.001
ST-segment elevation60 (5%)41 (18%)19 (2%)<0.001
ST-segment depression130 (11%)72 (32%)58 (6%)<0.001
T-wave inversion149 (13%)53 (24%)96 (10%)<0.001
No significant changes774 (67%)40 (18%)734 (79%)<0.001

Levels of miRNAs according to final diagnosis

Of the six miRNAs studied, miR-208b (< 0.001), miR-499 (< 0.001) and miR-320a (= 0.031) levels were significantly higher in patients with AMI than in those without AMI (Fig. 1a). Levels of miR-451 were comparable between groups, suggesting that haemolysis of plasma samples did not represent a bias in the study. In patients with AMI, levels of miR-133a, miR-208b, miR-499 and miR-451 were significantly higher in those with STEMI compared to NSTEMI (Fig. 1b). In general, the levels of miR-223, miR-320a and miR-451 were higher than the levels of cardiomyocyte-enriched miR-133a, miR-208b and miR-499. Also, there were large overlaps of miRNAs both between patients with and without AMI, and between the various subgroups studied (Fig. 1a+b; Figure S1a+b).

Figure 1.

Levels of six miRNAs at presentation in patients with (= 224) and without (= 931) AMI (a) and in patients with STEMI (= 45) and NSTEMI (= 179) (b). Log-transformed miRNA values were used for analyses. P-values describe the significance level of differences for each miRNA between patients with and without AMI. AMI, acute myocardial infarction; STEMI, ST-segment elevation myocardial infarction; and NSTEMI, non-ST-segment elevation myocardial infarction.

miRNAs for the early diagnosis of AMI

The diagnostic accuracy of miRNA levels at presentation for the diagnosis of AMI was highest for miR-208b [area under the ROC curve (AUC) 0.76, 95% confidence interval (CI) of 0.72–0.80] and miR-499 (AUC 0.65, 95% CI 0.61–0.70) (Fig. 2a). ROC curves yielded an optimal cut-off value of −11.1 for miR-208b with a sensitivity of 64.7% and a specificity of 80.2%, and an optimal cut-off value of −10.9 for miR-499 with a sensitivity of 35.7% and a specificity of 90.3%. In very ‘early presenters’ (presentation to the ED within 3 h of the onset of chest pain; = 409), the AUC of miR-208b was 0.70 (95% CI 0.62–0.78); AUC values did not reach statistical significance for the other miRNAs studied in this analysis (Fig. 2d).

Figure 2.

Receiver operating characteristic curves showing diagnostic accuracy for the six miRNAs (a), miR-208b and miR-499 in comparison and combination with a conventional (cTnT4, Roche Diagnostics) and high-sensitivity cardiac troponin assay (hs-cTnT, Roche Diagnostics) (b and c) and the six miRNAs in early presenters (presenting within 3 h from symptom onset) (d).

The diagnostic accuracy of cTnT4 and hs-cTnT for AMI at presentation was superior to all miRNAs, with AUC values of 0.84 and 0.94, respectively (both < 0.001). Also, neither miR-208b nor miR-499 (the two miRNAs with highest AUC values for the detection of AMI) added incremental diagnostic benefit on top of cTnT4 or hs-cTnT (Fig. 2b,c). Similar results were obtained in a multivariate binary regression analysis: none of the miRNAs that were significant in univariate analysis (miR-133a, miR-208b and miR-499) remained significant in multivariate analysis with hs-cTnT or cTnT4 (Table 2).

Table 2. Univariate and multivariate binary regression analysis for the diagnosis of acute myocardial infarction using miRNAs and troponins
 Univariate analysisMultivariate analysis (hs-cTnT)Multivariate analysis (cTnT)
Hazard ratio (95% CI)P-valueHazard ratio (95% CI)P-valueHazard ratio (95% CI)P-value
  1. Calculated with all miRNA values and cTnT4 × 103, except miR499 calculated with 106. hs-cTnT per increase of 1 ng L−1 and cTnT4 per increase of 1 mg L−1. miRNA, microRNA; CI, confidence interval; cTnT4, fourth-generation cardiac troponin T; and hs-cTnT, high-sensitivity cTnT.

miR-133a1.116 (1.035–1.204)0.0050.960 (0.801–1.150)0.6550.946 (0.794–1.126)0.532
miR-208b6.073 (2.506–14.717)<0.0010.907 (0.335–2.457)0.8481.009 (0.541–1.881)0.977
miR-2231.000 (0.999–1.001)0.691    
miR-320a1.000 (0.999–1.001)0.779    
miR-4511.000 (1.000–1.000)0.712    
miR-4991.022 (1.013–1.030)<0.0010.999 (0.996–1.002)0.4350.999 (0.994–1.004)0.748
hs-cTnT1.042 (1.034–1.049)<0.0011.042 (1.034–1.049)<0.001 
cTnT41.040 (1.031–1.048)<0.001 1.040 (1.031–1.049)<0.001

Furthermore, the IDI calculated for miR-208b (−0.00001, = 0.964) and miR-499 (0.00015, = 0.486) in addition to hs-cTnT did not lead to an improvement in prediction of AMI diagnosis.

miRNAs for the prediction of death and future AMI

During a median follow-up of 27 months (IQR 25–31 months), there were 102 (9%) deaths with a median time to death of 293 days (IQR 55–618). Only median miR-208b levels were significantly higher in patients who died (−11.8, IQR −16.9 to −10.2) than in survivors (−12.5, IQR −16.9 to −11, = 0.018). All other miRNAs did not differ significantly between survivors and deceased patients.

The prognostic accuracy for the prediction of short-term (30 days) and long-term mortality (730 days) by ROC curve analysis is shown in Figure S2a for all miRNAs analysed and (hs-)cTnT. Only in the case of miR-208b was there a borderline significant AUC for short-term mortality (0.67; 95% CI 0.52–0.81) whereas all other miRNAs were not significant predictors of short-term mortality. In the long-term, none of the miRNAs significantly predicted mortality (Figure S2b). Equally, univariate Cox proportional hazard analysis showed that none of the miRNAs analysed predicted mortality (Table 3A). According to Kaplan–Meier analysis (only shown for miR-208b, which displayed the best performance in ROC curve analysis), the outcome of patients with chest pain in terms of survival was significantly better for those with lower miR-208 concentrations (≤−10.38), as shown in Fig. 3a,b.

Table 3. Cox proportional hazard analysis for prediction of all-cause death (A) and future acute myocardial infarction (AMI; B) during follow-up (730 days)
 (A)(B)
All-cause death (univariate analysis)Future AMI (univariate analysis)
Hazard ratio (95% CI)P-valueHazard ratio (95% CI)P-value
  1. Calculated with all miRNA values and cTnT4 × 103, except miR499 calculated with 106. hs-cTnT per increase of 1 ng L−1 and cTnT4 per increase of 1 mg L−1. cTnT4, fourth-generation cardiac troponin T; hs-cTnT, high-sensitivity cTnT.

miR-133a1.020 (0.954–1.089)0.5650.998 (0.845–1.179)0.982
miR-208b1.084 (1.001–1.174)0.0471.077 (0.925–1.254)0.336
miR-2231.000 (0.998–1.001)0.4440.999 (0.997–1.001)0.318
miR-320a1.000 (0.999–1.001)0.8710.996 (0.989–1.004)0.347
miR-4511.000 (1.000–1.000)0.9681.000 (1.000–1.000)0.711

miR-499

hs-cTnT

1.000 (1.000–1.001)

1.001 (1.000–1.001)

0.305

<0.001

1.001 (0.999–1.003)

1.001 (1.000–1.001)

0.383

0.058

cTnT41.001 (1.000–1.001)<0.0011.001 (1.000–1.001)0.075
Figure 3.

Kaplan–Meier survival curves for all patients with acute chest pain showing cumulative survival during short-term (30 days) and long-term follow-up (730 days). Patients were subdivided into two groups according to optimal cut-off value (−10.38) calculated using the Youden index (sensitivity 31.8%, specificity 86.2%). Log-rank values were employed to assess statistical significance. miRNA values are log-transformed.

With regard to the prediction of both short-term (30 days) and long-term (730 days) future AMI, AUC values were not statistically significant for any of the miRNAs investigated (Figure S3a,b). Similarly, Cox proportional hazard analysis demonstrated that none of the miRNAs analysed predicted future AMI (Table 3B).

Discussion

In this prospective, multicentre, international study of 1155 patients, we assessed the diagnostic and prognostic value of circulating levels of six miRNAs (miR-133a, miR-208b, miR-223, miR-320a, miR-451 and miR-499) in patients with acute chest pain. In particular, their incremental diagnostic value in combination with both conventional and hs-cTnT assays was investigated. To our knowledge, this is the largest analysis of miRNAs in patients presenting with suspected AMI reported so far. We report five major findings of this study. First, circulating levels of miR-208b, miR-320a and miR-499 were significantly higher in patients with AMI than in those with other diagnoses. Secondly, the diagnostic accuracy for AMI was moderate for miR-208b (AUC 0.76) and miR-499 (AUC 0.65), whereas the other miRNAs provided no diagnostic value. Thirdly, the diagnostic accuracy of all miRNAs examined was significantly lower than that of cTnT or hs-cTnT. Fourthly, there was no incremental benefit of combining any of the miRNAs with cTnT or hs-cTnT with respect to diagnostic accuracy for AMI. Finally, the six miRNAs studied have no or very low predictive value for the occurrence of future AMI or death during long-term follow-up (730 days).

Although perhaps disappointing to some extent, these results have important clinical implications and extend previous findings on the possible clinical use of miRNAs [6-9, 23, 24]. Current hs-cTn assays [1, 2] detect troponins at concentrations approximately 10-fold lower than conventional troponin assays and have recently been introduced in AMI diagnosis guidelines [3]. Yet, the markedly increased number of patients with positive hs-cTn values due to noncoronary cardiac disease [4], various other diseases [33] or age [27] and the remaining insensitivity of hs-cTn in the initial hour(s) after AMI cause considerable problems for physicians in terms of diagnosis. miRNAs have been shown to be early biomarkers of cardiac necrosis. Recently, several studies comparing circulating miRNA levels in healthy control subjects and AMI patients [6, 7] and a small study of miRNA levels in 66 patients with acute chest pain [9] have shown very promising results, especially for cardiomyocyte-enriched miRNAs, with regard to the diagnosis of AMI, demonstrating the potential utility of miRNAs as markers of cardiac damage [8]. Similarly, we have been able to confirm that levels of cardiomyocyte-enriched miR-208b and miR-499 are increased in patients with AMI compared to those with other types of acute chest pain. Yet, their diagnostic accuracy for AMI in the heterogeneous (and clinically highly relevant) group of patients with acute chest pain is only moderate and is clearly outperformed by both conventional and hs-cTn. Furthermore, in comparison with reference biomarkers, the miRNAs studied in this analysis do not seem to add incremental benefit when used in a dual-marker approach with either conventional or hs-cTn. In agreement with our findings, Widera et al. [10] analysed six circulating miRNAs in acute coronary syndrome (including miR-133a, miR-208b and miR-499) and showed a high degree of overlap between patients with AMI and unstable angina. The large overlap of miRNA levels observed in our study between patients with and without AMI, and between patients with STEMI and NSTEMI, rules out the robust diagnostic value of these miRNAs.

Amongst the miRNAs that were significantly increased in AMI patients (miR-133a, miR-208b and miR-499), only miR-208b has been shown to be exclusively released by myocardial injury (and not by muscle injury, as is the case for miR-133a and miR-499) [34]. Therefore, it is not surprising that, of all miRNAs studied in this analysis, miR-208b demonstrated the highest diagnostic accuracy for AMI.

Comparison of the various miRNA studies that have been conducted is complicated by several issues. First, in addition to the limited sample size of most existing miRNA studies [6, 8, 9], most have mainly performed analyses between AMI patients and healthy control subjects [6, 7], Secondly, due to the relative immaturity of the field of miRNAs in cardiovascular medicine, varying RNA extraction protocols [21] methods of nucleic acid detection and normalization procedures [21] are used, further complicating a direct comparison of the various miRNA studies [35]. The protocol used for miRNA measurements in the present study has been widely used, and its reliability has been previously demonstrated [7]. However, the exact time course of miRNA release into the bloodstream is still poorly defined. From a clinical perspective, a rapid and accurate diagnosis of AMI remains crucial for patients with acute chest pain. Current methods for measuring miRNAs quantitatively are still relatively time-consuming (taking several hours), and the low levels of most circulating miRNAs make accurate quantification challenging with existing technology [35]. By contrast, the protein troponin can be reliably and accurately quantified within 60 min in most circumstances. Thus, novel biomarkers such as miRNAs have to compete with the current reference biomarker troponin not only with regard to diagnostic accuracy but also in terms of the time it takes for reliable and accurate measurement. Nevertheless, in contrast to the classic antibody-based assays used for measurement of the myocardial protein troponin, which are subject to assay interference, miRNAs have the advantage that they can be quantified using PCR, enabling highly specific and simultaneous measurement of various miRNAs.

Even though miRNAs appear to be very early markers of cardiac damage (peaking within hours) [5, 7, 9, 23], including in the group of very early presenters (i.e. presentation to the ED within 3 h of onset of acute chest pain), there seemed to be no further benefit of the six miRNAs studied. It is interesting that neither the diagnostic accuracy of hs-cTnT nor that of conventional cTnT, before 3–6 h after onset of symptoms, could be improved by adding miR-208b or miR-499, as also shown by IDI analysis. The limited benefit of miRNAs even in early presenting patients and the fact that miRNA levels did not differ between patients with STEMI and NSTEMI might be supported in part by the findings of De Rosa et al. [36]. In their analysis, the authors compared circulating levels of miR-499 and miR-133 with hs-cTn levels in the coronary sinus blood of patients with acute coronary syndrome and revealed that there appears to be a threshold for hs-cTn below which circulating miRNAs are no longer measurable despite small increases in the level of hs-cTn [36].

The possibility remains that other miRNAs add benefit in addition to the high diagnostic accuracy of existing conventional and hs-cTn assays. Novel self-learning pattern recognition algorithms as described by Meder et al. [37], who proposed a signature of 20 miRNAs to predict AMI, might detect other promising miRNAs for single or combined use in the early evaluation of patients with acute chest pain.

The prognostic benefit of miRNAs in cardiovascular medicine has been considerably less well studied than the potential diagnostic benefit. Matsumoto et al. [38] performed a microarray analysis in patients who survived AMI and identified 11 miRNAs differentially expressed in the serum of patients at high risk of cardiac death and eventually only two miRNAs (miR-155 and miR-380) with significantly higher levels in patients who subsequently died due to cardiac causes. Widera et al. [10] found that miR-133a and miR-208b were significantly associated with the risk of death in patients with acute coronary syndromes. In a large-scale prospective study, Zampetaki et al. [39] identified a group of three miRNAs, including miR-223, consistently associated with the risk of MI. The miRNAs investigated in the present study, including miR-223, were not robustly associated with mortality. However, using a systems-based approach, we recently identified miR-150 as a significant biomarker of left ventricular remodelling in two independent cohorts of AMI patients [40]. Therefore, circulating miRNAs still hold some potential to aid in prognostication of patients with cardiac diseases.

Several limitations of the current study should be considered. First, we cannot draw any conclusions about patients with terminal kidney failure requiring dialysis because these patients were excluded from our study. Of note, we observed that patients with terminal kidney failure have increased levels in miR-499 and cTnT without any sign of myocardial injury [41]. Secondly, we cannot exclude the possibility that miRNAs other than those tested in this study might be clinically useful. As such, the present study is limited by the selection of miRNAs for investigation. Thirdly, the adjudication of AMI was performed by applying the criteria defined in the universal definition of MI (3, 28, 29). As cardiac troponins have an important role in the universal definition of MI, any state-of-the-art analysis following current ESC/ACC/AHA/IFCC/WHF guidelines to apply this definition puts cardiac troponins as a comparator in a very strong position. Therefore, it is very difficult for new biomarkers to demonstrate significant added value on top of cardiac troponins. Fourthly, P-values are not usually adjusted in observational studies comparing several biomarkers against an independent diagnostic or prognostic gold standard (1, 2, 4, 41). Therefore, we have chosen not to do so in this analysis of the potential value of six miRNAs. We wish to acknowledge, however, that with a higher number of potential analytes for evaluation, the chances of finding a positive association between an analyte and the end-point increases. Finally, because circulating levels of miRNAs fluctuate very rapidly after MI, it is possible that admission levels of miRNAs may not correspond to peak values. However, this represents the clinical scenario of patients presenting to the ED several hours after symptom onset.

Conclusion

The findings of the present study temper speculation about the potential clinical usefulness of circulating miRNAs in patients with acute chest pain. There does not appear to be an increase in diagnostic or prognostic value provided by the miRNAs investigated in this study.

Conflict of interest statement

Professor Mueller has received research grants from the Swiss National Science Foundation (PP00B-102853) and the Swiss Heart Foundation, the Cardiovascular Research Foundation Basel, 8sense, Abbott, ALERE, Brahms, Critical Diagnostics, Nanosphere, Roche, Siemens and the University Hospital Basel, as well as speaker honoraria from Abbott, ALERE, Brahms, Novartis, Roche and Siemens. TR has received research grants from the Swiss National Science Foundation (PASMP3-136995), the Swiss Heart Foundation, the University of Basel, the Professor Max Cloetta Foundation and the Department of Internal Medicine, University Hospital Basel as well as speaker honoraria from Brahms and Roche. None of the other authors has any conflict of interests relevant to this study to declare. The sponsors had no role in the design of the study, the analysis of the data, the preparation of the manuscript or the decision to submit the manuscript for publication.

Acknowledgments

The authors thank Christelle Nicolas, Laurent Quennery, Bernadette Leners and Justine Gofinet for expert technical assistance. We also thank the patients who participated in the study, the staff of the emergency department, the research coordinators and the laboratory technicians (particularly Michael Freese, Claudia Stelzig, Esther Garrido, Irina Klimmeck, Melanie Wieland, Janine Voegele, Beate Hartmann and Fausta Chiaverio) for their most valuable contributions. This work was supported by the Ministry of Culture, Higher Education and Research of Luxembourg, and the National Research Fund of Luxembourg. EG and JZ received fellowships from the National Research Fund of Luxembourg.

Ancillary