Long non-coding RNAs (lncRNAs) have been reported to be involved in the pathogenesis of cardiovascular disease (CVD), but whether circulating lncRNAs can serve as a coronary artery disease (CAD), biomarker is not known. The present study screened lncRNAs by microarray analysis in the plasma from CAD patients and control individuals and found that 265 lncRNAs were differentially expressed. To find specific lncRNAs as possible CAD biomarker candidates, we used the following criteria for 174 up-regulated lncRNAs: signal intensity ≥8, fold change >2.5 and P<0.005. According to these criteria, five intergenic lncRNAs were identified. After validation by quantitative PCR (qPCR), one lncRNA was excluded from the candidate list. The remaining four lncRNAs were independently validated in another population of 20 CAD patients and 20 control individuals. Receiver operating characteristic (ROC) curve analysis showed that lncRNA AC100865.1 (referred to as CoroMarker) was the best of these lncRNAs. CoroMarker levels were also stable in plasma. The predictive value of CoroMarker was further assessed in a larger cohort with 221 CAD patients and 187 control individuals. Using a diagnostic model with Fisher's criteria, taking the risk factors into account, the optimal sensitivity of CoroMarker for CAD increased from 68.29% to 78.05%, whereas the specificity decreased slightly from 91.89% to 86.49%. CoroMarker was stable in plasma because it was mainly in the extracellular vesicles (EVs), probably from monocytes. We conclude that CoroMarker is a stable, sensitive and specific biomarker for CAD.

CLINICAL PERSPECTIVES

  • CAD is the major cause of death worldwide; a lot of patients with acute coronary syndrome or severe coronary stenosis have benefited from the improved therapeutic methods in clinics, including medications, percutaneous coronary intervention and coronary artery bypass surgery.

  • However, the mortality rate still remains high. The obvious reason is that CAD is not easy to be found in its earlier stage by regular examinations, such as electrocardiography (ECG), cardiac ultrasound. Therefore, it is necessary to detect CAD in its early stage.

  • Our present study showed that plasma CoroMarker is a stable, sensitive and specific biomarker for CAD, with 78.05% sensitivity and 86.49% specificity, we hope it would be helpful to get more and more CAD patients to be diagnosed and treated in the earlier stage of disease, consequently reduced the CAD-related mortality.

INTRODUCTION

Cardiovascular diseases (CVDs), especially coronary artery disease (CAD), continue to be the major cause of death worldwide, causing a major socioeconomic burden. Although therapeutic methods such as medications, percutaneous coronary intervention and coronary artery bypass surgery have improved the prog-nosis of CAD, mortality remains high. Therefore, it is necessary to detect CAD in its early stages, especially before the development of left ventricular dysfunction [1,2]. Early identification of patients with CAD at high risk of adverse cardiovascular outcomes, using circulating or imaging biomarkers, may help in this regard [3]. However, currently available CAD biomarkers have limited risk prediction [4,5].

Genome-wide analyses have identified that almost all of the human genome is transcribed, with a large number of long non-coding RHAs (lncRNAs) [6,7]. LncRNAs, ranging from 200 nucleotides to >10000 nucleotides [8], have been found to be involved in specific physiological and pathological processes through epigenetic, transcriptional or post-transcriptional regulatory mechanisms in a wide range of human diseases and disorders [9,10], e.g. cancers [11] and neurological disorders [12]. Recently, several studies have shown that some lncRNAs are involved in the development of various types of CVDs [1315], e.g. heart failure [1618], cardiac hypertrophy [19], cardiometabolic diseases [20] and myocardial infarction [21]. The circulating levels of some lncRNAs, such as ANRIL and LincP21, are markedly increased in atherosclerosis which may be important in its pathogenesis [2224].

LncRNAs can be stable in the plasma and other body fluids and could therefore serve as biomarkers for some diseases, e.g. a prostate-specific lncRNA PCA3 in urine has been identified as the most specific biomarker for the detection of prostate cancer, with a higher specificity than the widely used prostate-specific antigen (PSA) test [25]. Other lncRNA biomarkers in plasma include H19 for gastric cancer [26], lncRNA HULC for hepatocellular carcinoma [27] and lncRNA LIPCAR for heart failure after a myocardial infarction [16]. Therefore, we hypothesized that there are specific circulating lncRNAs that may serve as CAD biomarkers. In the present study, lncRNAs were screened by microarray analysis and validated in different cohorts with or without CAD. We found that one plasma lncRNA, AC100865.1 (referred to as CoroMarker), is stable in plasma and exists in extracellular vesicle (EVs), mostly from monocytes. It can be a sensitive and specific marker of CAD.

EXPERIMENTAL

Study cohorts

This study was a single-centre Third Military Medical University (MMU)-based National Institute of Health-sponsored trial (NIH.gov clinical trial NCT01629225). The initial ‘microarray cohort’ was composed of 15 male patients with or 15 male patients without CAD, who were admitted to the Department of Cardiology, Daping Hospital, because of newly diagnosed or suspected CAD. CAD was confirmed by coronary angiography. The first validation cohort was composed of 20 CAD patients and 20 control individuals. To investigate the relationship between lncRNA expression and the risk of CAD, a case–control study was conducted on a cohort with 221 CAD patients and 187 non-CAD individuals, confirmed by coronary angiography, who were consecutively enrolled from January 2013 to March 2014. The characteristics of these participants are shown in Supplementary Tables S1–S3; 180 cases of CAD and 150 control individuals were randomly selected as the training group and the remaining 78 as the test group. To investigate the specificity of the association of CoroMarker with CAD, we also quantified CoroMarker levels in plasma from patients with other CVDs. Exclusion criteria were as follows: (i) patients with malignant tumours and other severe systemic diseases (such as renal failure or hepatic disease); (ii) patients with serious acute infection within 6 weeks before admission; (iii) patients with active chronic inflammatory disease; and (iv) patients suspected of drug or alcohol abuse. This study was approved by the Ethics Committee of the Medical Faculty of Daping Hospital. Written informed consent was obtained from all patients or their families, in accordance with the Declaration of Helsinki.

Definition of CAD and preparation of blood samples

CAD was diagnosed by coronary angiography according to American College of Cardiology/American Heart Association guidelines [28], i.e. the percentage narrowing of one coronary artery segment was ≥50% diameter reduction. The percentage narrowing of coronary artery segments was estimated by visual assessment from two independent interventional cardiologists. The control individuals had no coronary artery stenosis. Due to the difficulty in diagnosing patients with small vessel disease, we used the treadmill exercise test to screen CAD patients, with only those patients with negative results from the treadmill exercise test being included as controls.

Blood samples (8 or 50 ml) were collected from radial arteries, after insertion of the arterial catheter and before the administration of any anticoagulants, and put into a test tube containing EDTA. Plasma was then carefully collected, divided into aliquots and stored at −80°C before use.

Isolation of EVs

EVs were isolated from 25 ml of plasma (obtained from approximately 50 ml of whole blood) from each patient by differential centrifugation according to previous reports [29,30]. The plasma samples were centrifuged at 500 g for 20 min, and the initial pellets discarded to remove residual cells. Then, the supernatants were re-centrifuged at 1500 g for 20 min to remove any debris, re-centrifuged at 110000 g for 70 min, and the final pellets containing the EVs were suspended in 100 μl of PBS. All steps were performed at 4°C. The isolated EVs were then subjected to RNase I (20 Kunitz units/ml, Sigma–Aldrich) digestion (30 min at 37°C) to remove any RNA outside the EVs. These EVs were used to extract RNA.

Collection and purification of monocytes

Peripheral blood (1 ml) containing anticoagulant (EDTA) was incubated with a fluorochrome-labelled monoclonal anti-human mouse antibody, PerCP-CD14 (15 μl, 340585, BD Biosciences), in the dark for 15 min. Thereafter 15 ml of ammonium chloride lysing buffer was added to lyse the red blood cells. Of the staining medium (PBS with 3% heat-inactivated serum and 0.1% sodium azide), 15 ml was then added to stop the lysis reaction. After mixing gently, the samples were centrifuged at 200 g for 5 min and then the cell pellet washed with PBS five times. Thereafter, the cells were suspended in 500 μl of staining medium, mixed and sorted immediately by FACS (FC500, Beckman Coulter). At least 5 million events were acquired from the FACS. FACS data were analysed using Flowjo software (Treestar). The absolute monocyte count was obtained using a Coulter ACT/differential cell counter (Beckman Coulter).

RNA isolation and qPCR

Total RNA was extracted from plasma, EVs and cells using TRIzol (Invitrogen) and purified with an RNeasy kit (Qiagen) [27,31]. RNA was extracted from 1 ml of plasma and dissolved in 15 μl of diethylpyrocarbonate (DEPC) water. The quantity and quality of total RNA were determined with a Bioanalyzer (Agilent 2100) and NanoDrop (Gene Quant pro, GE Healthcare), and approximately 400–600 ng of RNA was obtained from 1 ml of plasma. No difference in the amount of extracted RNA in a unit of plasma was found between control and CAD samples. The quantity and quality of total RNA from monocytes were determined using a NanoDrop instrument, and samples were used only if the ratio of the absorbance at 260 and 280 nm (A260/A280) was between 1.8 and 2.1. RNA samples with concentrations >0.2 μg/μl were used for each reverse transcription reaction. Then, 11 μl of purified RNA from plasma or EVs or 2 μg of purified RNA from monocytes was used for cDNA synthesis. After reverse transcription (Superscript II, Invitrogen), quantitative PCR (qPCR) was performed using the Brilliant SYBR Green Mastermix-Kit and the MX4000 multiplex qPCR System from Stratagene.

According to the manufacturer's recommendations, 20 μl of final reaction mixture containing 10 μl of SYBR Green, 0.5 μl of sense primer, 0.5 μl of antisense primer, 7 μl of sterile deionized water and 2.0 μl of the synthesized cDNA. The CT value was defined as the cycle number at which the fluorescence (∆Rn) exceeded the threshold. A threshold of 0.20 was used as the default setting. The lncRNA expression levels were quantified in triplicate. The levels of lncRNAs in plasma were calculated using the ΔCT method because there is, as yet, no consensus about stable and suitable internal controls for lncRNA in plasma samples. The change in gene expression was calculated using the equation 2−ΔCT [32,33]. The relative expression level of lncRNA in monocytes was normalized to the internal control ACTB expression and calculated by the comparative CT (∆∆CT) method. A melt curve analysis was used to confirm the specificity of amplification and lack of primer dimers. The primers used in qPCR of the lncRNAs are listed in Supplementary Table S4.

Microarray and computational analysis

For the initial lncRNA screening, RNA was isolated from plasma obtained from 15 men with and 15 men without CAD. Equal volumes of plasma from five samples were mixed together as one sample for microarray analysis. Therefore, the microarray analysis was performed in six pooled samples in total (three pooled samples for CAD patients and three pooled samples for control individuals). RNA was pre-amplified and then microarrayed (Arraystar, Human LncRNA array, version 2.0). This procedure allowed the simultaneous detection of 33045 lncRNAs and 30215 coding transcripts. After filtering for low-intensity lncRNA, the lncRNA of at least two of six samples with flags in the present or marginal (‘all targets value’) category were chosen for quantile normalization and further data analysis. Quantile normalization was performed using Expander 6 and subsequent data processing was performed using the GeneSpring GX v11.5.1 software package (Agilent Technologies). To find a potential biomarker candidate lncRNA (CoroMarker), all lncRNA transcripts that were included in the microarrays were screened according to the following criteria: fold change >2 and P<0.01 (see Supplementary Table S5).

The microarray data analysed in this study have been deposited in the NCBI Gene Expression Omnibus database under accession number GSE68506 (http://www.ncbi.nlm.nih.gov/projects/geo/index.cgi).

Statistical analysis

Data are presented as means±S.D.s mean rank or number out of number (n/n), unless otherwise described. Horizontal lines indicated the median in the scatter plots of lncRNAs' expression. The Shapiro–Wilk and Kolomogorov–Smirnov tests were used to test for non-gaussian distribution. For continuous variables, the two-tailed Student's t test was used for normal distribution and homogeneity of variance, and the Mann–Whitney U test for abnormal distribution. Discrete variables were compared using the χ2 2×2 contingency table. The association of the selected CoroMarker with the risk of CAD was assessed for age and body mass index (BMI) using the Kruskal–Wallis H test. In addition, regression analyses were performed for the association of CoroMarker with CAD in cohorts adjusting data for BMI (cohorts 1 and 2) and sex (cohort 3). Stepwise linear regression analyses were also done to determine the contribution of CAD, BMI, age and sex to the diagnostic value of CoroMarker in CAD patients. Receiver operating characteristic (ROC) curves and the area under the curve (AUC) were used to assess the specificity and sensitivity of using plasma CoroMarker as a novel diagnostic tool for the detection of CAD. We also determined whether the predictive accuracy of CoroMarker as a marker for CAD was improved with male sex, tobacco use, alcohol use and hypertension, using Fisher's criteria. The AUC of the two models was compared using the Mann–Whitney U test. P<0.05 was considered statistically significant. Finally, Pearson's correlation analyses were employed to verify the relationship between monocyte and plasma CoroMarker. Statistical analyses were performed using IBM SPSS Statistics 20, Prism 5 (GraphPad Software) and the MATLAB (V2013a) program.

RESULTS

LncRNA expression profiles in the plasma of CAD patients

To determine whether there were specific lncRNAs expressed in CAD patients, we profiled plasma lncRNA expression in 15 male CAD patients and 15 control individuals using the Human LncRNA Array v2.0 (8660 K, Arraystar). The clinical and demographic characteristics of this population were detailed in Supplementary Table S1. The plasma levels of lncRNAs differed significantly between the two groups, as illustrated in the hierarchical clustering analysis (Figure 1). Of the 33045 lncRNAs detected on the microarray, 265 were found to be differentially expressed in CAD patients with a fold change >2 and P<0.01; 174 lncRNAs were up-regulated whereas 91 were down-regulated (see Supplementary Table S5). To make potential lncRNA markers easy to be measured in the clinic, we selected possible biomarkers in the 174 up-regulated lncRNAs, using the following strategy: an average normalized intensity >8, a fold change >2.5 and P<0.005 (see Supplementary Table S6). Only five intergenic lncRNAs met these criteria: ENST00000504882, AC100865.1, ENST00000442318, AF118081 and AK126077; these are highlighted in red in Supplementary Table S5. Next, the expression of these five lncRNAs was verified using qPCR in the same patients. The lncRNA primers are listed in Supplementary Table S4. The fold increase in CAD, which differed significantly from control, was 4.3 for ENST00000504882, 3.6 for AC100865.1, 3.1 for ENST00000442318 and 2.8 for AF118081 (Figure 2).

Differential expression of lncRNAs in CAD patients and control individuals

Figure 1
Differential expression of lncRNAs in CAD patients and control individuals

Hierarchical clustering analysis of 265 lncRNAs that were differentially expressed in the two groups of participants. (A) Expression values are represented in red and green, indicating expression above and below the median expression value in CAD patients (P1F, P2F, P3F) or control individuals (N1F, N2F, N3F), respectively. (B) Volcano plot of fold change and corresponding P value for each lncRNA.

Figure 1
Differential expression of lncRNAs in CAD patients and control individuals

Hierarchical clustering analysis of 265 lncRNAs that were differentially expressed in the two groups of participants. (A) Expression values are represented in red and green, indicating expression above and below the median expression value in CAD patients (P1F, P2F, P3F) or control individuals (N1F, N2F, N3F), respectively. (B) Volcano plot of fold change and corresponding P value for each lncRNA.

Expression levels of lncRNAs ENST00000504882, AC100865.1, ENST00000442318, AF118081 and AK126077 quantified by qPCR

Figure 2
Expression levels of lncRNAs ENST00000504882, AC100865.1, ENST00000442318, AF118081 and AK126077 quantified by qPCR

*P<0.05, versus control.

Figure 2
Expression levels of lncRNAs ENST00000504882, AC100865.1, ENST00000442318, AF118081 and AK126077 quantified by qPCR

*P<0.05, versus control.

Independent validation of lncRNA expression

To validate independently the increased expression of ENST00000504882, AC100865.1, ENST00000442318 and AF118081 in CAD plasma, these lncRNAs were quantified in plasma samples obtained from another set of 20 CAD patients and 20 control individuals. The clinical and demographic characteristics of the patients are shown in Supplementary Table S2. The fold increase in CAD was 2.5 for ENST00000504882, 4.2 for AC100865.1, 2.2 for ENST00000442318 and 2 for AF118081 (Figures 3A–3D).

Expression levels of lncRNAs ENST00000504882, AC100865.1, ENST00000442318 and AF118081 levels quantified by qPCR (20×20)

Figure 3
Expression levels of lncRNAs ENST00000504882, AC100865.1, ENST00000442318 and AF118081 levels quantified by qPCR (20×20)

(AD) Expression levels of lncRNAs: (A) ENST00000504882, (B) AC100865.1, (C) ENST00000442318 and (D) AF118081. *P<0.05, versus control. (EH) ROC curve analyses of lncRNAs ENST00000504882, AC100865.1, ENST00000442318 and AF118081 for diagnosis of CAD.

Figure 3
Expression levels of lncRNAs ENST00000504882, AC100865.1, ENST00000442318 and AF118081 levels quantified by qPCR (20×20)

(AD) Expression levels of lncRNAs: (A) ENST00000504882, (B) AC100865.1, (C) ENST00000442318 and (D) AF118081. *P<0.05, versus control. (EH) ROC curve analyses of lncRNAs ENST00000504882, AC100865.1, ENST00000442318 and AF118081 for diagnosis of CAD.

To determine the relationship between these lncRNA levels and CAD, ROC analysis was performed. The AUC was 0.830 for ENST00000504882, 0.898 for AC100865.1, 0.805 for ENST00000442318 and 0.788 for AF118081 (Figures 3E–3H). These results indicated that AC100865.1 may be a good candidate biomarker to predict CAD. AC100865.1 was renamed CoroMarker in the present study.

Plasma CoroMarker level sensitive for CAD

As the above-mentioned observations were done in two small populations, we further assessed CoroMarker as a marker for CAD in a larger group (the third group) of patients (CAD patients, n=221, and control individuals, n=187). The clinical and demographic characteristics of this population are shown in Supplementary Table S3. ROC curve analysis of CoroMarker showed an AUC of 0.795 and 95% confidence interval (CI) 0.753–0.838, indicating that CoroMarker may be a potential candidate marker for CAD (Figure 4A).

ROC curve analyses of CoroMarker alone and CoroMarker combined with four risk factors for the diagnosis of CAD in the original group, training group and test group

Figure 4
ROC curve analyses of CoroMarker alone and CoroMarker combined with four risk factors for the diagnosis of CAD in the original group, training group and test group

(A) The third group, (B) the training group and (C) the test group.

Figure 4
ROC curve analyses of CoroMarker alone and CoroMarker combined with four risk factors for the diagnosis of CAD in the original group, training group and test group

(A) The third group, (B) the training group and (C) the test group.

To further analyse the diagnostic accuracy of CoroMarker as a marker for CAD, we constructed a diagnostic model using Fisher's criteria. The third group was randomly divided into a training group (n=330, 180 CAD patients and 150 control individuals) (Figure 4B) and a test group (n=78, 41 CAD patients and 37 control individuals) (Figure 4C). Fisher's criteria were applied to establish discriminant function with the training group and then validated by the test group. We adopted a random sampling method with constant uniform distribution, performed the ROC curve analysis each time, and constructed discriminant function using Fisher's method with optimal sampling. This procedure was repeated 100 times, and the chosen optimal discriminant function, corresponding to the 48th training group, was the final discriminant function. The AUC was 0.796 with a 95% CI of 0.749–0.843 (Figure 4B) and, according to the diagnostic model of Fisher's method, the signature was defined as: f=−0.395528x+0.053712, where x denoted the expression level of CoroMarker. A patient was classified as ‘CAD’ if f <0 according to the patient's lncRNA expression value, and as ‘control’ if f >0. The CAD and control patients in the training groups differed significantly in age, sex, the presence of hypertension, hyperlipidaemia, smoking or drinking, and the use of angiotensin receptor blockers (ARBs), β-adrenergic receptor blockers, calcium channel blockers (CCBs) and statins (see Supplementary Table S7).

CoroMarker was then validated in the test group. The same model and criteria as those derived from the training group were used. The AUC was 0.811 (95% CI 0.710–0.912; Figure 4C). When repeated 100 times, it correctly classified 20 and 34 into the CAD patients and control individuals, respectively. The corresponding average sensitivity and specificity in the test group was 48.78% and 91.89%, respectively (see Supplementary Table S8-1). The optimal discrimination number (48th time) was 28 CAD patients and 34 control individuals, corresponding to a sensitivity and a specificity of the test group (48th time) of 68.29% and 91.89%, respectively (see Supplementary Table S8-2). In the test group, significant differences between CAD patients and control individuals were also found with the presence of hypertension and hyperlipidaemia, and the use of ARBs and CCBs (see Supplementary Table S7).

Increase in CoroMarker expression independent of CAD risk factors and other CVDs

To assess whether the diagnostic ability of CoroMarker was independent of risk factors of CAD, we chose patients with risk factors and other CVDs in addition to CAD. According to the European guidelines on cardiovascular disease prevention in clinical practice [34], the risk factors included mainly age, sex, BMI, tobacco use, alcohol use, hypertension, hyperlipidaemia, diabetes, medication history [statins, angiotensin-converting enzyme inhibitors (ACEIs), ARBs, CCBs, β-adrenergic receptor blockers and antiplatelet drugs] and the degree of coronary artery narrowing. The results from the third group (221 CAD and 187 control patients) and the training group (180 CAD and 150 control patients) CAD patients showed that sex (P=0.045 and 0.047), tobacco use (P=0.026 and 0.042), alcohol use (P=0.017 and 0.048) and hypertension (P=0.044 and 0.049) were significantly associated with the diagnosis of CAD, but not with the other risk factors (see Supplementary Table S9).

To investigate further whether the plasma level of CoroMarker was elevated in other CVDs besides CAD, we categorized the patients according to specific diseases (Figure 5). Although the sample size was small, the plasma level of CoroMarker was significantly increased in patients with CAD (P<0.05), relative to patients without CAD but with atrial fibrillation (n=10), valvular disease (n=6), dilated cardiomyopathy (n=7), peripheral artery disease (n=5), hyperlipidaemia (n=10), hypertension (n=10), diabetes mellitus Type 2 (n=10), abdominal aortic aneurysm (n=5) or viral myocarditis (n=7); these different populations had similar clinical features.

CoroMarker levels in the plasma of patients with a variety of CVDs

Figure 5
CoroMarker levels in the plasma of patients with a variety of CVDs

Horizontal lines indicate the median.

Figure 5
CoroMarker levels in the plasma of patients with a variety of CVDs

Horizontal lines indicate the median.

Combination with the risk factors increased diagnostic prediction of CoroMarker as a CAD signature

As indicated above, we found an association between plasma CoroMarker levels and the male sex, tobacco use, alcohol use and hypertension (see Supplementary Table S9). To determine whether these factors had an additive effect on the prediction values for plasma CoroMarker level, we performed another ROC curve analysis of CoroMarker combining these risk factors in the third group, and found that the diagnostic prediction was slightly increased, i.e. AUC was 0.799 (95% CI 0.756–0.841) (see Figure 4A). Using the combination of those risk factors, we reconstructed the diagnostic model, using the same method and criteria; the optimal AUC was 0.791 (95% CI 0.743–0.839) (see Figure 4B), the signature was: f=− 0.517468x1 − 0.094203x2 − 0.069296x3 − 0.063057x4 − 0.036825x5+0.175763, where x1 denoted the expression level of CoroMarker, and x2, x3, x4 and x5 indicated hypertension, the male sex, tobacco use and alcohol use, respectively. The value of x2, x3, x4 and x5 was 1 or 0, where 1 denoted the presence of hypertension, the male sex, smoking and alcohol use, and 0 the absence of hypertension, the female sex, and absence of smoking and alcohol use. A patient was classified as ‘CAD’ if f <0 according to the patient's CoroMarker expression value and risk factors, and as ‘control’ if f >0.

The CoroMarker was also tested for its diagnostic accuracy in the test group, using the same model and criteria as those used in the training group. The AUC was 0.854 (95% CI 0.769–0.939) (see Figure 4C). The average of 100 repeats correctly classified 25 and 31 patients of the test group into the CAD patient and control group, respectively. The corresponding average sensitivity and specificity were 60.98% and 83.78%, respectively (see Supplementary Table S10-1). The optimal discrimination number (48th time) was 32 patients into both the case and the control groups, corresponding to the sensitivity and specificity of 78.05% and 86.49%, respectively (see Supplementary Table S10-2). The clinical and pathological characteristics of the training and test groups were detailed in Supplementary Table S7.

Finally, to compare the sensitivity and specificity of diagnostic accuracy between CoroMarker alone and CoroMarker in combination with the four risk factors (male sex, tobacco use, alcohol use and hypertension), we performed ROC curve analysis again. According to the diagnostic models, in the third, training and test groups no significant difference in prediction between the CoroMarker alone and the combined model (CoroMarker plus four risk factor groups) was found (P=0.878, 0.831 and 0.321, respectively) (see Figures 4A–4C). Although the optimal sensitivity in the combined model increased significantly from 68.29% to 78.05%, the specificity actually decreased from 91.89% to 86.49% (compare Supplementary Table S8 with Supplementary Table S10), which could indicate that the diagnostic accuracy of the combined model does not enhance CAD discrimination.

CoroMarker plasma is stable and can be a CAD biomarker

For a reliable biomarker, a stable level is a basic requirement. We studied the stability of CoroMarker in plasma exposed to various periods at room temperature and different freeze–thaw cycles. As indicated in Figure 6A, exposure of the plasma to room temperature for 6 h had no effect on CoroMarker expression. Even after 48 h, CoroMarker levels were still at 80% of the unexposed levels. So CoroMarker levels were stable at room temperature. A freeze–thaw cycle is another major factor affecting RNA levels in plasma. We found that three freeze–thaw cycles had no effect on CoroMarker levels, although seven freeze–thaw cycles did reduce the CoroMarker to half its original levels (Figure 6B).

Stability of CoroMarker

Figure 6
Stability of CoroMarker

Plasma was exposed to various periods (A) at room temperature and (B) with a different number of freeze–thaw cycles. *P<0.05, versus control, n=12.

Figure 6
Stability of CoroMarker

Plasma was exposed to various periods (A) at room temperature and (B) with a different number of freeze–thaw cycles. *P<0.05, versus control, n=12.

One might wonder why the lncRNA is so stable in the plasma that contains RNase. Localization inside EVs and/or RNA decoration with proteins is reported to be a possible mechanism [35]. The present study showed that most of the CoroMarkers were inside EVs. We therefore speculated that the EVs may protect the CoroMarker from degradation. Whether there are protein decorations on those enzyme-free lncRNAs needs to be determined. CoroMarker levels were four times higher in plasma with EVs than in EV-free plasma (Figures 7A and 7B). Consistent with CoroMarker levels in whole plasma, CoroMarker was higher in EVs from CAD patients than from control individuals (Figure 7C).

CoroMarker in plasma and EVs

Figure 7
CoroMarker in plasma and EVs

(A) CoroMarker levels were determined by PCR: monocytes (lane 1), plasma (lane 2), plasma without EVs (lane 3) and plasma EVs (lane 4). (B) CoroMarker levels were determined by qPCR in EV-free plasma or EVs from control individuals (*P <0.05, versus plasma, n=6). (C) CoroMarker levels were determined in EVs from CAD patients and control individuals (*P<0.05, versus control, n=12).

Figure 7
CoroMarker in plasma and EVs

(A) CoroMarker levels were determined by PCR: monocytes (lane 1), plasma (lane 2), plasma without EVs (lane 3) and plasma EVs (lane 4). (B) CoroMarker levels were determined by qPCR in EV-free plasma or EVs from control individuals (*P <0.05, versus plasma, n=6). (C) CoroMarker levels were determined in EVs from CAD patients and control individuals (*P<0.05, versus control, n=12).

Plasma EVs are from blood cells, with monocytes being the major source. We found that CoroMarker levels in monocytes were higher in CAD patients than control individuals (Figure 8A). There was a correlation between CoroMarker levels in plasma and monocytes (R2=0.276) (Figure 8B), indicating that plasma EVs could be from monocytes. Stepwise linear regression analyses taking CAD, BMI, age and sex as independent variables showed that, for plasma CoroMarker, only CAD emerged as a potential predictor [regression coefficient (B)=−0.322; P=0.000] compared with BMI (P=0.147), age (P=0.222) and sex (P=0.378). The regression equation was plasma CoroMarker=0.678 − 0.322 CAD. Compared with BMI (P=0.287), age (P=0.375) and sex (P=0.178), for monocyte CoroMarker, only CAD was a potential predictor (B=−2.154, P=0.000). The regression equation was monocyte CoroMarker=−2.154 − 4.910 CAD.

Relationship of CoroMarker levels in plasma and monocytes

Figure 8
Relationship of CoroMarker levels in plasma and monocytes

(A) CoroMarker level was determined by qPCR in CAD patients and control individuals (*P<0.05, versus control, n=15). (B) Pearson's correlation scatter plot of CoroMarker levels between monocytes and plasma (R2=0.276, CAD=41, control=33).

Figure 8
Relationship of CoroMarker levels in plasma and monocytes

(A) CoroMarker level was determined by qPCR in CAD patients and control individuals (*P<0.05, versus control, n=15). (B) Pearson's correlation scatter plot of CoroMarker levels between monocytes and plasma (R2=0.276, CAD=41, control=33).

DISCUSSION

CVDs continue to be the number one cause of morbidity and mortality worldwide, and CAD is the leading cause of death among CVDs [36,37]. Prompt diagnosis of CAD could improve its prognosis with aspirin, statin and other medical or surgical interventional treatments. The diagnosis of CAD involves non-invasive and invasive methods. Non-invasive methods include the electrocardiogram, Holter monitor, treadmill exercise test, ultrasonic cardiogram, and measurement of intima–media thickness in brachial and carotid arteries [38]. However, the poor sensitivity and specificity of these tests lower their diagnostic values [39]. Invasive methods, including spiral computed tomography (CT) angiography and coronary angiography, are alternative choices. Coronary angiography is a ‘gold standard’ in the diagnosis of CAD diagnosis and CT angiography has gained popularity over the last decade, with the rapid development of imaging technology. However, both spiral CT angiography and coronary angiography are expensive and invasive, and are used only in the diagnosis of patients with high-risk CAD. Therefore, there is a need for new non-invasive methods to aid in the diagnosis of CAD.

LncRNAs belong to a novel class of non-coding RNAs, which are conventionally defined as transcripts longer than 200 nucleotides with no protein-coding capability [6,7]. Previous studies have shown that lncRNAs are involved in fundamental cellular processes, such as RNA processing, gene regulation, chromatin modification, gene transcription and post-transcriptional gene regulation [710,40]. Some lncRNAs have been reported to be involved in a number of human diseases, including CVD and cancers [1114,1619,21,41], such as lncRNA-P21, which regulates neointima formation, vascular smooth muscle cell proliferation, apoptosis and atherosclerosis [42], and HOTAIR, which reprograms chromatin to promote cancer metastasis [43]. Some lncRNAs have been reported to be biomarkers for the diagnosis and poor prognosis of cancers, e.g. the lncRNA MALAT-1 has been reported as a plasma-based biomarker for the diagnosis of prostate cancer [32], increased levels of the lncRNA HOTAIR denote poor prognosis in colorectal cancers [44], whereas decreased expression of the lncRNA GAS5 is indicative of a poor prognosis in gastric cancer [45].

miRNAs, another type of non-coding RNAs, are important in the development of CVDs; these miRNAs have also been reported as biomarkers for the diagnosis and prognosis of CAD. Although miRNAs represent a minority of the non-coding transcriptome, the tangle of lncRNAs is likely to contain as yet unidentified classes of molecules, so lncRNAs as diagnostic tools have properties that are advantageous relative to miRNAs. LncRNAs are functional molecules involved in epigenetics, alternative splicing, nuclear import processes and cell microstructure, and serve as small RNA precursors. The above-mentioned characteristics make lncRNAs possible candidate biomarkers for the diagnosis of CAD. The present study compared the lncRNA profiles between CAD and non-CAD patients, and found that the levels of the lncRNA CoroMarker are increased in the plasma of CAD patients and have a 92% specificity for CAD. Other CVDs have normal plasma levels of CoroMarker.

Stability is a basic requirement for any biomarker. Previous studies have shown that the stability of miRNAs in plasma is due to their being packaged in microvesicles or formation of protein complexes with miRNA-binding proteins and lipoproteins [46,47]. Whether lncRNAs have properties similar to these described for miRNAs are not known. Some recent studies have shown that several lncRNAs can be detected stably in the body fluids of patients, such as plasma lncRNA LIPCAR and urinary lncRNA PCA3, which can be biomarkers for severe LV remodelling after a myocardial infarction [16] and prostate cancer [25], respectively. We studied the stability of CoroMarker in plasma under severe conditions, including exposure to room temperature and freeze–thaw cycles. Our data clearly showed that CoroMarker was very stable in plasma samples, consistent with the report of others [16]. Our finding that CoroMarker was mainly contained in the plasma EVs may explain the stability of CoroMarker in plasma.

Monocytes are an important group of cells in the innate immune system. In inflamed tissue, monocytes transform into macrophages that clear pathogens and modulate tissue repair and healing. Previous studies have shown a pivotal role for monocytes in the pathogenesis of CAD and atherosclerotic plaque progression [48]. In the early stages of CAD, a large number of monocytes become attached to the luminal endothelium in atherosclerosis-predisposed areas of the arteries [49]. The continuous and increasing adhesion of monocytes to the luminal endothelium is a hallmark in the development of CAD. The present study shows that there is positive correlation between CoroMarker levels in plasma and monocytes, and that monocytes are the major source of EVs in plasma from CAD patients.

We have to admit the limitation of this study, as the patients in this study are from one hospital, and whether there is a difference for patients from different areas and races is not known. Therefore, its validity should be tested further in more prospective cohorts.

In conclusion, we have shown that circulating CoroMarker is differentially expressed in plasma and monocytes of CAD patients compared with controls. These findings indicate for the first time that CoroMarker is a stable and specific plasma biomarker for CAD. Prospective clinical trials should be carried out to determine the usefulness of CoroMarker as a stable plasma biomarker for CAD.

AUTHOR CONTRIBUTION

All authors have contributed to and agreed on the content of this paper. Yujia Yang and Yue Cai performed the research. Yujia Yang and Yue Cai analyzed the data. Yujia Yang, Yue Cai and Xionwen Chen wrote the paper. Yujia Yang and Yue Cai designed the research study. Pedro A. Jose, Chunyu Zeng and Lin Zhou edited and revised manuscript. Gengze Wu, Xinjian Chen, Yukai Liu, Xinquan Wang, Junyi Yu and Chuanwei Li interpreted results of experiments. Yujia Yang approved the final version of manuscript. All authors reviewed the manuscript.

FUNDING

This work was supported by the National Natural Science Foundation of China [grant numbers 31430043 and 31471089].

Abbreviations

     
  • ARB

    angiotensin receptor blocker

  •  
  • AUC

    area under the ROC curve

  •  
  • BMI

    body mass index

  •  
  • CAD

    coronary artery disease

  •  
  • CCB

    calcium channel blocker

  •  
  • CI

    confidence interval

  •  
  • CT

    computed tomography

  •  
  • CVD

    cardiovascular disease

  •  
  • EV

    extracellular vesicle

  •  
  • lncRNA

    long non-coding RNA

  •  
  • qPCR

    quantitative PCR

  •  
  • ROC

    receiver operating characteristic

References

References
1
Daneault
 
B.
Généreux
 
P.
Kirtane
 
A.J.
Witzenbichler
 
B.
Guagliumi
 
G.
Paradis
 
J.M.
Fahy
 
M.P.
Mehran
 
R.
Stone
 
G.W.
 
Comparison of three-year outcomes after primary percutaneous coronary intervention in patients with left ventricular ejection fraction <40% versus ≥ 40% (from the Horizons-AMI trial)
Am. J. Cardiol.
2013
, vol. 
111
 (pg. 
12
-
20
)
[PubMed]
2
Waldo
 
S.W.
Secemsky
 
E.A.
O'Brien
 
C.
Kennedy
 
K.F.
Pomerantsev
 
E.
Sundt
 
T.M.
McNulty
 
E.J.
Scirica
 
B.M.
Yeh
 
R.W.
 
Surgical ineligibility and mortality among patients with unprotected left main or multivessel coronary artery disease undergoing percutaneous coronary intervention
Circulation
2014
, vol. 
130
 (pg. 
2295
-
2301
)
[PubMed]
3
Malaud
 
E.
Merle
 
D.
Piquer
 
D.
Molina
 
L.
Salvetat
 
N.
Rubrecht
 
L.
Dupaty
 
E.
Galea
 
P.
Cobo
 
S.
Blanc
 
A.
, et al 
Local carotid atherosclerotic plaque proteins for the identification of circulating biomarkers in coronary patients
Atherosclerosis
2014
, vol. 
233
 (pg. 
551
-
558
)
[PubMed]
4
Wykrzykowska
 
J.J.
Garcia-Garcia
 
H.M.
Goedhart
 
D.
Zalewski
 
A.
Serruys
 
P.W.
 
Differential protein biomarker expression and their time-course in patients with a spectrum of stable and unstable coronary syndromes in the integrated biomarker and imaging study-1 (IBIS-1)
Int. J. Cardiol.
2011
, vol. 
149
 (pg. 
10
-
16
)
[PubMed]
5
Eitel
 
I.
Blase
 
P.
Adams
 
V.
Hildebrand
 
L.
Desch
 
S.
Schuler
 
G.
Thiele
 
H.
 
Growth-differentiation factor 15 as predictor of mortality in acute reperfused ST-elevation myocardial infarction: insights from cardiovascular magnetic resonance
Heart
2011
, vol. 
97
 (pg. 
632
-
640
)
[PubMed]
6
Guttman
 
M.
Amit
 
I.
Garber
 
M.
French
 
C.
Lin
 
M.F.
Feldser
 
D.
Huarte
 
M.
Zuk
 
O.
Carey
 
B.W.
Cassady
 
J.P.
, et al 
Chromatin signature reveals over a thousand highly conserved large non-coding RNAs in mammals
Nature
2009
, vol. 
458
 (pg. 
223
-
227
)
[PubMed]
7
Ponting
 
C.P.
Oliver
 
P.L.
Reik
 
W.
 
Evolution and functions of long noncoding RNAs
Cell
2009
, vol. 
136
 (pg. 
629
-
641
)
[PubMed]
8
Bertone
 
P.
Stolc
 
V.
Royce
 
T.E.
Rozowsky
 
J.S.
Urban
 
A.E.
Zhu
 
X.
Rinn
 
J.L.
Tongprasit
 
W.
Samanta
 
M.
Weissman
 
S.
, et al 
Global identification of human transcribed sequences with genome tiling arrays
Science
2004
, vol. 
306
 (pg. 
2242
-
2246
)
[PubMed]
9
Hung
 
T.
Wang
 
Y.
Lin
 
F.
Koegel
 
A.K.
Kotake
 
Y.
Grant
 
G.D.
Horlings
 
H.M.
Shah
 
N.
Umbricht
 
C.
Wang
 
P.
, et al 
Extensive and coordinated transcription of noncoding rnas within cell-cycle promoters
Nat. Genet.
2011
, vol. 
43
 (pg. 
621
-
629
)
[PubMed]
10
Wang
 
K.C.
Chang
 
H.Y.
 
Molecular mechanisms of long noncoding rnas
Mol. Cell
2011
, vol. 
43
 (pg. 
904
-
914
)
[PubMed]
11
Spizzo
 
R.
Almeida
 
M.I.
Colombatti
 
A.
Calin
 
G.A.
 
Long non-coding RNAs and cancer: a new frontier of translational research?
Oncogene
2012
, vol. 
31
 (pg. 
4577
-
4587
)
[PubMed]
12
Knauss
 
J.L.
Sun
 
T.
 
Regulatory mechanisms of long noncoding rnas in vertebrate central nervous system development and function
Neuroscience
2013
, vol. 
235
 (pg. 
200
-
214
)
[PubMed]
13
Schonrock
 
N.
Harvey
 
R.P.
Mattick
 
J.S.
 
Long noncoding RNAs in cardiac development and pathophysiology
Circ. Res.
2012
, vol. 
111
 (pg. 
1349
-
1362
)
[PubMed]
14
Mathiyalagan
 
P.
Keating
 
S.T.
Du
 
X.J.
El-Osta
 
A.
 
Interplay of chromatin modifications and non-coding RNAs in the heart
Epigenetics
2014
, vol. 
9
 (pg. 
101
-
112
)
[PubMed]
15
Hu
 
Y.W.
Zhao
 
J.Y.
Li
 
S.F.
Huang
 
J.L.
Qiu
 
Y.R.
Ma
 
X.
Wu
 
S.G.
Chen
 
Z.P.
Hu
 
Y.R.
Yang
 
J.Y.
, et al 
Rp5–833a20.1/mir-382–5p/nfia-dependent signal transduction pathway contributes to the regulation of cholesterol homeostasis and inflammatory reaction
Arterioscler. Thromb. Vasc. Biol.
2015
, vol. 
35
 (pg. 
87
-
101
)
[PubMed]
16
Kumarswamy
 
R.
Bauters
 
C.
Volkmann
 
I.
Maury
 
F.
Fetisch
 
J.
Holzmann
 
A.
Lemesle
 
G.
deGroote
 
P.
Pinet
 
F.
Thum
 
T.
 
Circulating long noncoding RNA, lipcar, predicts survival in patients with heart failure
Circ. Res.
2014
, vol. 
114
 (pg. 
1569
-
1575
)
[PubMed]
17
Papait
 
R.
Kunderfranco
 
P.
Stirparo
 
G.G.
Latronico
 
M.V.
Conderelli
 
G.
 
Long noncoding RNA: a new player of heart failure?
J. Cardiovasc. Transl. Res.
2013
, vol. 
6
 (pg. 
876
-
883
)
[PubMed]
18
Yang
 
K.C.
Yamada
 
K.A.
Patel
 
A.Y.
Topkara
 
V.K.
George
 
I.
Cheema
 
F.H.
Ewald
 
G.A.
Mann
 
D.L.
Nerbonne
 
J.M.
 
Deep RNA sequencing reveals dynamic regulation of myocardial noncoding rnas in failing human heart and remodeling with mechanical circulatory support
Circulation
2014
, vol. 
129
 (pg. 
1009
-
1021
)
[PubMed]
19
Wang
 
K.
Liu
 
F.
Zhou
 
L.Y.
Long
 
B.
Yuan
 
S.M.
Wang
 
Y.
Liu
 
C.Y.
Sun
 
T.
Zhang
 
X.J.
Li
 
P.F.
 
The long noncoding RNA CHRF regulates cardiac hypertrophy by targeting miR-489
Circ. Res.
2014
, vol. 
114
 (pg. 
1377
-
1388
)
[PubMed]
20
Liu
 
Y.
Ferguson
 
J.F.
Xue
 
C.
Ballantyne
 
R.L.
Silverman
 
I.M.
Gosai
 
S.J.
Serfecz
 
J.
Morley
 
M.P.
Gregory
 
B.D.
Li
 
M.
, et al 
Tissue-specific RNA-seq in human evoked inflammation identifies blood and adipose lincrna signatures of cardiometabolic diseases
Arterioscler. Thromb. Vasc. Biol.
2014
, vol. 
34
 (pg. 
902
-
912
)
[PubMed]
21
Ishii
 
N.
Ozaki
 
K.
Sato
 
H.
Mizuno
 
H.
Saito
 
S.
Takahashi
 
A.
Miyamoto
 
Y.
Ikegawa
 
S.
Kamatani
 
N.
Hori
 
M.
, et al 
Identification of a novel non-coding RNA, MIAT, that confers risk of myocardial infarction
J. Hum. Genet.
2006
, vol. 
51
 (pg. 
1087
-
1099
)
[PubMed]
22
Bai
 
Y.
Nie
 
S.
Jiang
 
G.
Zhou
 
Y.
Zhou
 
M.
Zhao
 
Y.
Li
 
S.
Wang
 
F.
Lv
 
Q.
Huang
 
Y.
, et al 
Regulation of CARD8 expression by ANRIL and association of CARD8 single nucleotide polymorphism rs2043211 (p.C10x) with ischemic stroke
Stroke
2014
, vol. 
45
 (pg. 
383
-
388
)
[PubMed]
23
Holdt
 
L.M.
Hoffmann
 
S.
Sass
 
K.
Langenberger
 
D.
Scholz
 
M.
Krohn
 
K.
Finstermeier
 
K.
Stahringer
 
A.
Wilfert
 
W.
Beutner
 
F.
, et al 
Alu elements in anril non-coding RNA at chromosome 9p21 modulate atherogenic cell functions through trans-regulation of gene networks
PLoS Genet.
2013
, vol. 
9
 pg. 
e1003588
 
[PubMed]
24
Harismendy
 
O.
Notani
 
D.
Song
 
X.
Rahim
 
N.G.
Tanasa
 
B.
Heintzman
 
N.
Ren
 
B.
Fu
 
X.D.
Topol
 
E.J.
Rosenfeld
 
M.G.
, et al 
9p21 DNA variants associated with coronary artery disease impair interferon-gamma signalling response
Nature
2011
, vol. 
470
 (pg. 
264
-
268
)
[PubMed]
25
deKok
 
J.B.
Verhaegh
 
G.W.
Roelofs
 
R.W.
Hessels
 
D.
Kiemeney
 
L.A.
Aalders
 
T.W.
Swinkels
 
D.W.
Schalken
 
J.A.
 
Dd3(pca3), a very sensitive and specific marker to detect prostate tumors
Cancer Res.
2002
, vol. 
62
 (pg. 
2695
-
2698
)
[PubMed]
26
Arita
 
T.
Ichikawa
 
D.
Konishi
 
H.
Komatsu
 
S.
Shiozaki
 
A.
Shoda
 
K.
Kawaguchi
 
T.
Hirajima
 
S.
Nagata
 
H.
Kubota
 
T.
, et al 
Circulating long non-coding RNAs in plasma of patients with gastric cancer
Anticancer Res.
2013
, vol. 
33
 (pg. 
3185
-
3193
)
[PubMed]
27
Xie
 
H.
Ma
 
H.
Zhou
 
D.
 
Plasma HULC as a promising novel biomarker for the detection of hepatocellular carcinoma
Biomed. Res. Int.
2013
, vol. 
2013
 pg. 
136106
 
[PubMed]
28
Scanlon
 
P.J.
Faxon
 
D.P.
Audet
 
A.M.
Carabello
 
B.
Dehmer
 
G.J.
Eagle
 
K.A.
Legako
 
R.D.
Leon
 
D.F.
Murray
 
J.A.
Nissen
 
S.E.
, et al 
ACC/AHA guidelines for coronary angiography: executive summary and recommendations. A report of the American College of Cardiology/American Heart Association Task force on practice guidelines (committee on coronary angiography) developed in collaboration with the society for cardiac angiography and interventions
Circulation
1999
, vol. 
99
 (pg. 
2345
-
2357
)
[PubMed]
29
Cai
 
J.
Han
 
Y.
Ren
 
H.
Chen
 
C.
He
 
D.
Zhou
 
L.
Eisner
 
G.M.
Asico
 
L.D.
Jose
 
P.A.
Zeng
 
C.
 
Extracellular vesicle-mediated transfer of donor genomic DNA to recipient cells is a novel mechanism for genetic influence between cells
J. Mol. Cell. Biol.
2013
, vol. 
5
 (pg. 
227
-
238
)
[PubMed]
30
Skog
 
J.
Würdinger
 
T.
van Rijn
 
S.
Meijer
 
D.H.
Gainche
 
L.
Sena-Esteves
 
M.
Curry
 
W.T.
Carter
 
B.S.
Krichevsky
 
A.M.
Breakefield
 
X.O.
 
Glioblastoma microvesicles transport RNA and proteins that promote tumour growth and provide diagnostic biomarkers
Nat. Cell Biol.
2008
, vol. 
10
 (pg. 
1470
-
1476
)
[PubMed]
31
Li
 
D.
Chen
 
G.
Yang
 
J.
Fan
 
X.
Gong
 
Y.
Xu
 
G.
Cui
 
Q.
Geng
 
B.
 
Transcriptome analysis reveals distinct patterns of long noncoding RNAs in heart and plasma of mice with heart failure
PLoS One
2013
, vol. 
8
 pg. 
e77938
 
[PubMed]
32
Ren
 
S.
Wang
 
F.
Shen
 
J.
Sun
 
Y.
Xu
 
W.
Lu
 
J.
Wei
 
M.
Xu
 
C.
Wu
 
C.
Zhang
 
Z.
, et al 
Long non-coding RNA metastasis associated in lung adenocarcinoma transcript 1 derived miniRNA as a novel plasma-based biomarker for diagnosing prostate cancer
Eur. J. Cancer
2013
, vol. 
49
 (pg. 
2949
-
2959
)
[PubMed]
33
Li
 
Y.
Elashoff
 
D.
Oh
 
M.
Sinha
 
U.
St John
 
M.A.
Zhou
 
X.
Abemayor
 
E.
Wong
 
D.T.
 
Serum circulating human mRNA profiling and its utility for oral cancer detection
J. Clin. Oncol.
2006
, vol. 
24
 (pg. 
1754
-
60
)
[PubMed]
34
Perk
 
J.
De Backer
 
G.
Gohlke
 
H.
Graham
 
I.
Reiner
 
Z.
Verschuren
 
W.M.
Albus
 
C.
Benlian
 
P.
Boysen
 
G.
Cifkova
 
R.
, et al 
Comitato per Linee Guida Pratiche (CPG) dell'ESC
European guidelines on cardiovascular disease prevention in clinical practice (version 2012). The Fifth Joint Task Force of the European Society of Cardiology and other societies on cardiovascular disease prevention in clinical practice [constituted by representatives of nine societies and by invited experts]
Eur. Heart J.
2012
, vol. 
33
 (pg. 
1635
-
1701
)
[PubMed]
35
Huang
 
X.
Yuan
 
T.
Tschannen
 
M.
Sun
 
Z.
Jacob
 
H.
Du
 
M.
Liang
 
M.
Dittmar
 
R.L.
Liu
 
Y.
Liang
 
M.
, et al 
Characterization of human plasma-derived exosomal RNAs by deep sequencing
BMC Genomics
2013
, vol. 
14
 pg. 
319
 
[PubMed]
36
Hoyert
 
D.L.
Xu
 
J.
 
Deaths: preliminary data for 2011
Natl. Vital Stat. Rep.
2012
, vol. 
61
 (pg. 
1
-
51
)
[PubMed]
37
(2012)
Circulation research thematic synopsis: vascular biology and disease
Circ. Res.
, vol. 
111
 (pg. 
e255
-
e273
)
[PubMed]
38
Iwamoto
 
Y.
Maruhashi
 
T.
Fujii
 
Y.
Idei
 
N.
Fujimura
 
N.
Mikami
 
S.
Kajikawa
 
M.
Matsumoto
 
T.
Kihara
 
Y.
Chayama
 
K.
, et al 
Intima–media thickness of brachial artery, vascular function, and cardiovascular risk factors
Arterioscler. Thromb. Vasc. Biol.
2012
, vol. 
32
 (pg. 
2295
-
2303
)
[PubMed]
39
Wang
 
T.J.
Gona
 
P.
Larson
 
M.G.
Tofler
 
G.H.
Levy
 
D.
Newton-Cheh
 
C.
Jacques
 
P.F.
Rifai
 
N.
Selhub
 
J.
Robins
 
S.J.
, et al 
Multiple biomarkers for the prediction of first major cardiovascular events and death
N. Engl. J. Med.
2006
, vol. 
355
 (pg. 
2631
-
2639
)
[PubMed]
40
Orom
 
U.A.
Derrien
 
T.
Beringer
 
M.
Gumireddy
 
K.
Gardini
 
A.
Bussotti
 
G.
Lai
 
F.
Zytnicki
 
M.
Notredame
 
C.
Huang
 
Q.
, et al 
Long noncoding RNAs with enhancer-like function in human cells
Cell
2010
, vol. 
143
 (pg. 
46
-
58
)
[PubMed]
41
Jaipersad
 
A.S.
Shantsila
 
A.
Lip
 
G.Y.
Shantsila
 
E.
 
Expression of monocyte subsets and angiogenic markers in relation to carotid plaque neovascularization in patients with pre-existing coronary artery disease and carotid stenosis
Ann. Med.
2014
, vol. 
46
 (pg. 
530
-
538
)
[PubMed]
42
Wu
 
G.
Cai
 
J.
Han
 
Y.
Chen
 
J.
Huang
 
Z.P.
Chen
 
C.
Cai
 
Y.
Huang
 
H.
Yang
 
Y.
Liu
 
Y.
, et al 
LincRNA-p21 regulates neointima formation, vascular smooth muscle cell proliferation, apoptosis, and atherosclerosis by enhancing p53 activity
Circulation
2014
, vol. 
130
 (pg. 
1452
-
1465
)
[PubMed]
43
Gupta
 
R.A.
Shah
 
N.
Wang
 
K.C.
Kim
 
J.
Horlings
 
H.M.
Wong
 
D.J.
Tsai
 
M.C.
Hung
 
T.
Argani
 
P.
Rinn
 
J.L.
, et al 
Long non-coding RNA HOTAIR reprograms chromatin state to promote cancer metastasis
Nature
2010
, vol. 
464
 (pg. 
1071
-
1076
)
[PubMed]
44
Kogo
 
R.
Shimamura
 
T.
Mimori
 
K.
Kawahara
 
K.
Imoto
 
S.
Sudo
 
T.
Tanaka
 
F.
Shibata
 
K.
Suzuki
 
A.
Komune
 
S.
, et al 
Long noncoding RNA HOTAIR regulates polycomb-dependent chromatin modification and is associated with poor prognosis in colorectal cancers
Cancer Res.
2011
, vol. 
71
 (pg. 
6320
-
6326
)
[PubMed]
45
Sun
 
M.
Jin
 
F.Y.
Xia
 
R.
Kong
 
R.
Li
 
J.H.
Xu
 
T.P.
Liu
 
Y.W.
Zhang
 
E.B.
Liu
 
X.H.
 
De
 
W.
 
Decreased expression of long noncoding RNA gas5 indicates a poor prognosis and promotes cell proliferation in gastric cancer
BMC Cancer
2014
, vol. 
14
 pg. 
319
 
[PubMed]
46
Vickers
 
K.C.
Palmisano
 
B.T.
Shoucri
 
B.M.
Shamburek
 
R.D.
Remaley
 
A.T.
 
MicroRNAs are transported in plasma and delivered to recipient cells by high-density lipoproteins
Nat. Cell Biol.
2011
, vol. 
13
 (pg. 
423
-
433
)
[PubMed]
47
Zernecke
 
A.
Bidzhekov
 
K.
Noels
 
H.
Shagdarsuren
 
E.
Gan
 
L.
Denecke
 
B.
Hristov
 
M.
Köppel
 
T.
Jahantigh
 
M.N.
Lutgens
 
E.
, et al 
Delivery of microRNA-126 by apoptotic bodies induces CXCL12-dependent vascular protection
Sci. Signal.
2009
, vol. 
2
 pg. 
ra81
 
[PubMed]
48
Olivares
 
R.
Ducimetiere
 
P.
Claude
 
J.R.
 
Monocyte count: a risk factor for coronary heart disease?
Am. J. Epidemiol.
1993
, vol. 
137
 (pg. 
49
-
53
)
[PubMed]
49
Knorr
 
M.
Munzel
 
T.
Wenzel
 
P.
 
Interplay of NK cells and monocytes in vascular inflammation and myocardial infarction
Front. Physiol.
2014
, vol. 
5
 pg. 
295
 
[PubMed]

Author notes

Clinical trial registration: https://register.clinicaltrials.gov; ClinicalTrials.gov Identifier: NCT01629225

1

These authors contributed equally to this work.

Supplementary data