Circulating microRNA signature for the diagnosis of childhood dilated cardiomyopathy

Circulating miRNAs are proposed as a biomarker of heart disease. This study evaluated whether circulating miRNAs could be used as a biomarker for childhood dilated cardiomyopathy (CDCM). A total of 28 participants were enrolled in a discovery set, including patients with CDCM (n = 16) and healthy children (n = 12). The cardiac function of patients with CDCM was characterized by echocardiography and serum miRNA profiles of all participants were assessed by miRNA sequencing. After miRNA profiling, we quantitatively confirmed 148 regulated miRNAs in patients with CDCM compared with healthy subjects, and none were downregulated. Validation of candidate miRNAs was assessed by quantitative real-time polymerase chain reaction in other patients with CDCM (n = 30) and healthy controls (n = 16). A unique signature comprising mir-142-5p, mir-143-3p, mir-27b-3p, and mir-126-3p differentiated patients with CDCM from healthy subjects. Importantly, we observed an increase in mir-126-3p or let-7g in parallel with a robust decrease in the ejection fraction in patients with CDCM, which could differentiate heart failure patients from non-heart failure patients with CDCM. Moreover, mir-126-3p and let-7g were significantly negatively associated with the left ventricular ejection fraction. This study shows that a signature of four serum miRNAs may be a potential biomarker for diagnosing CDCM and assessing heart failure.

Given the difficulties of tissue sampling in children, circulating miRNAs would be appealing as CDCM biomarkers, which would also provide biological insight into the mechanism behind this disease and its therapy targets. We thus aimed to characterize the profile of circulating miRNAs in CDCM and validate this expression pattern in additional subjects. Moreover, we assessed the sensitivity and specificity of miRNAs as a diagnostic biomarker and their correlation with the degree of heart function, as evaluated by transthoracic echocardiography. We found that eight circulating miRNAs were significantly increased in CDCM patients compared with their levels in matched subjectithout CDCM. Four miRNAs (mir-142-5p, mir-143-3p, mir-27b-3p, and mir-126-3p) served as sensitive and specific biomarkers in CDCM, which could potentially be used to diagnose this condition. Notably, only mir-126-3p was also correlated with heart failure.

Results
Clinical characteristics of patients. A total of 46 dilated cardiomyopathy patients were enrolled in this study, which included a discovery phase and a validation phase. The patients had been diagnosed by three pediatric cardiologists, based on echocardiography and clinical symptoms. 15 patients were in heart failure, as diagnosed by the Canadian Cardiovascular Society guidelines. In addition, 28 healthy age-and sex-matched subjects were enrolled in the study to serve as healthy controls, who had no family history of cardiovascular disease. Both CDCM patients and healthy children were recruited and enrolled from Beijing Anzhen Hospital, affiliated to Capital Medical University, China. The discovery phase included 12 healthy children and 16 CDCM patients, the clinical characteristics of whom are shown in Table 1. A total of 30 CDCM patients with/without heart failure and 16 healthy controls were used for subsequent validation of 11 selected miRNAs by quantitative real-time polymerase chain reaction (qRT-PCR) validation, for which the division into two subgroups (heart failure or non-heart failure) was based on the echocardiography results (shown in Table 2. The study design is shown in Supplementary Figure 1. Identification of miRNAs differentially expressed in CDCM patients using next-generation sequencing analysis. To identify miRNA that are differentially expressed in CDCM and control, we performed miRNA sequencing on RNA from the serum of 16 CDCM and 12 health controls. A total of 1675 miRNAs were identified in both groups. We performed filtering between the CDCM patients and healthy controls, using the following criteria: fold change ≥2.0 (or ≤0.5) and P-value ≤ 0.05. A hierarchical clustering of differential miRNA was shown in Fig. 1A, including 148 upregulated miRNAs and no downregulated miRNA.
Unsupervised hierarchical cluster analysis further revealed that the serum sample of CDCM patients could be distinguished clearly from that of healthy children based on their pattern of significantly higher expression of certain miRNAs. Among the 148 miRNAs analyzed, and the TOP 50 miRNAs were as shown in Fig. 1B

Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway analyses.
A total of 10,543 target genes were predicted for the 148 differentially expressed miRNAs; the number of regions of overlap among the three different databases was 73. Gene Ontology enrichment analysis was used on the target gene candidates of the top 10 differentially expressed miRNAs. The predicted target genes of the differentially expressed miRNAs were further classified to annotate the pathways in which they are involved, using KEGG pathway annotation. A total of 24 associated biological pathways were identified. The top 10 pathways are shown in Supplementary  Figure 2. Some of these pathways, such as the MAPK pathway, the FoxO signaling pathway, and the Erb signaling pathway, were shown to be involved in cardiac fibrosis and apoptosis [9][10][11] . Our data demonstrate the broad range of target genes for those differentially expressed miRNAs potentially involved in the pathological process of DCM.
Diagnostic value of miRNAs for CDCM. In the validation phase, mean age and gender were similar between the CDCM cases and controls, as shown in Table 2.
To discriminate between the patients with CDCM and the healthy controls (diagnostic value), the receiver-operating characteristic curves were plotted for every single miRNA in validation of qRTPCR that was expressed at a significantly higher level in patient serum ( Fig. 2A-K). The areas under the curves (AUCs) ranged from 0.731 to 0.992, and almost all of them were higher than 0.9 (let-7g5p, miR-142-5p, miR-126-3p, miR-143-3p, miR-26a-5p, and miR-27b-3p), suggesting that these circulating miRNAs may be useful for CDCM detection and diagnosis (Fig. 3A-H and Supplementary Table 2). miRNAs were associated with heart dysfunction in CDCM. To evaluate the role of miRNAs in the classification of CDCM patients, 30 CDCM patients were divided into two subgroups according to ejection fraction (EF) in the conventional transthoracic echocardiography (non-heart failure, EF >55%; heart failure, EF ≤55%). There was no significant difference in clinical characteristics, including gender, age, and ultrasound results, except for ejection fraction and left ventricular end diastolic diameter. The relationship between the expression level of circulating miRNAs and cardiac function was further studied. In 8 selected miRNAs, mir126-3p and let-7g-5p expression significantly increased in the CDCM patients with heart failure compared with that in the non-heart failure CDCM patients (p < 0.05) (Fig. 4A-H), suggesting that circulating miRNAs may be associated with the progression of heart failure in CDCM patients. The areas under the curves (AUCs) of miR-126-3p and let-7g-5p were 0.782 and 0.760, respectively. Interestingly, a combination of these two miRNAs increased the AUC to 0.849 (Fig. 5A-C). We also examined the relationship of the concentration of mir-126-3p with the ejection fraction as a measure of heart function. As illustrated in Fig. 6, there was a significant inverse

Discussion
In the present study, a set of miRNA biomarkers were discovered and validated by the evaluation of serum miRNA profiles from two experimental sample sets. The first set was profiled using miRNA sequencing to identify differentially expressed miRNAs specific to DCM in children. A quantitative RT-PCR assay was then performed to validate the candidate differentially expressed miRNAs in the second sample set. The main findings of this study can be summarized as follows: (i) a substantial difference in miRNA expression between CDCM and healthy controls; (ii) a novel signature of four circulating miRNAs (mir-142-5p, mir-143-3p, mir-27b-3p, and mir-126-3p) allowing CDCM patients to be distinguished from healthy controls; and (iii) only mir-126-3p negatively correlated with heart systolic dysfunction.
An increasing amount of evidence indicates that miRNAs play a crucial role in idiopathic dilated cardiomyopathy. First, miRNAs may be involved in the mechanism regulating the development of DCM. Bioinformatic analysis revealed that the downregulation of key pro-survival miRNAs promoted apoptotic signaling and heart decompensation 7 . Using mdx mice as a model of Duchenne muscular dystrophy-induced cardiomyopathy, another previous study found that the downregulation of mir448 promoted cellular oxidative stress, which subsequently triggered cardiomyocyte death and cardiomyopathy progression 12 . Cardiomyocyte-specific miR-30C overexpression causes DCM by inhibiting mitochondrial oxidative phosphorylation in complexes III and IV 13 . Biomarkers have taken on an important role in the diagnosis and prognosis of heart failure in adults 14 . Second, serum miRNAs are remarkably stable. A previous study reported that serum miRNAs (mir-3135b, mir-3908, and mir-5571-5p) could be used as diagnostic biomarkers for adult DCM 15 . The levels of serum mir5571-5p significantly correlated with the New York Heart Association classification. However, little is known about the profile of circulating miRNAs and their diagnostic value in pediatric DCM. So far, Brian's group found four miRNAs (mir-155, mir-636, mir-646 and mir-639) are specific to CDCM who have the potential to recover 16 . There is thus a need for involvement of miRNAs in pediatric DCM patients to be more clearly defined, for their potential application in assisted diagnosis and risk stratification. Using miRNA sequencing, we found 148 upregulated  miRNAs in CDCM, but no downregulated ones (Fig. 2). qRT-PCR assay showed that miRNA alterations were well correlated with the findings from miRNA sequencing in independent sets. GO enrichment analysis based on the target genes of the differentially expressed miRNAs showed that the top 10 biological processes involved metabolic processes. Adult DCM patients exhibited decreased fatty acid metabolism and increased myocardial glucose metabolism 17 . The specific miRNA profile in CDCM patients may reflect abnormalities in heart metabolism. At present, there are no biomarkers to differentiate DCM from healthy subjects. In our study, we identified for the first time a unique signature of differentially expressed miRNAs that can distinguish CDCM patients from healthy children.
Left systolic function is associated with clinical outcomes in dilated cardiomyopathy patients 22 . To explore the correlation between miRNA expression and heart function, CDCM was divided into two subgroups based on EF (non-heart failure, EF > 55%; heart failure, EF < 55%). Mir-126-3p and let-7g exhibited a significant difference in expression between the non-heart failure and heart failure groups. However, we found an increase of mir-126-3p and let-7g in parallel with a robust decrease of EF in CDCM patients (p < 0.05). mir-126, as an important regulator of endothelial cell homeostasis, holds promise as a prognostic biomarker of vascular damage and endothelial dysfunction. Recent basic and clinical research has demonstrated its important role in heart failure. In contrast with our results, in atrial fibrillation patients, mir-126 expression was positively correlated with LVEF and negatively correlated with the N-terminal pro-brain natriuretic peptide levels 23 . However, in coronary heart disease patients, a positive association between circulating mir-126 and fatal myocardial infarction was identified 24 . Genome-wide profiling of microRNAs revealed that the expression of let-7 family was upregulated in heart failure tissue from dilated cardiomyopathy and ischemic cardiomyopathy patients compared with normal heart tissue 25 . Cardiomyocyte death/regeneration and metabolic disorder promoted development of heart failure. Let-7 family regulated cardiomyocyte regeneration, cell survival, metabolic control and glucose utilization 26 . The clinical study found the let-7g combined with other 4 miRNAs predict future fatal myocardial infarction in health participants 27 . In the present study, a negative correlation between mir126 or let-7g and LVEF was found, suggesting mir-126 and let-7g as a potential biomarker for the assessment of heart failure in CDCM. But further study is needed to determine whether mir126 and let-7g may participate in the pathogenesis of DCM in children.
However, the present study has certain limitations. First, the current lack of knowledge about the biology of miRNAs hinders our ability to assess the level of circulating miRNAs accurately. In addition, the sample size in this single-center study was small. Fewer controls were enrolled in the validation phase, because recruiting the healthy children is challenging. It is believed that the diagnostic value of miRNAs should be further validated in a multi-center study. Moreover, this study involved a cross-sectional study and there were no long-term follow-up data on the prognosis of CDCM. Future studies are needed to overcome these limitations.
In summary, the profile of four circulating miRNAs was found to distinguish CDCM patients from healthy children. Mir-126-3p was also shown to be negatively correlated with LVEF.

Materials and Methods
Patients. A total of 46 patients were enrolled in the study at Anzhen Hospital from November 2015 to May 2016. DCM patients, whose age from 9 months to 13 years old, were all diagnosed by clinical symptoms and echocardiography. 28 healthy children were also enrolled at the hospital as health control, with matched age and gender. Characteristics of the study including discovery phase and validation phase are given in Tables 1 and 2. All ultrasound examinations were performed using a commercially available echocardiographic machine with an S3 transducer (Vivid Seven System, GE Healthcare, Horten, Norway) when DCM and healthy subjects(CON) finished the collection of blood sample. The LV end-diastolic volume (EDV), end-systolic volume (ESV), and EF were calculated using biplane Simpson's methods from apical four and two-chamber views. The diagnosis of heart failure was based on clinical presentation and echocardiographic results according to 2013 Canadian Cardiovascular Society guideline 28 . The study protocol was approved by the Medical Ethics Committee of Beijing Anzhen Hospital affiliated to Capital Medical University and carried out according to the principles of the Declaration of Helsinki. All informed consents were obtained from children's parents or guardians. The general characteristics of participants were shown in Supplementary Table 3 RNA preparation and sequencing. Total serum RNA were isolated from the discovery set of serum samples was extracted using mirVana miRNA isolation kit following the manufacture's protocol (Ambion, Thermo Fisher Scientific Inc, Waltham, MA USA). Each RNA sample was quantified with a spectrophotometer (NanoDrop 1000, Thermo Scientific, Wilmington, Delaware). Total RNA was used to prepare the miRNA sequencing library, which included the following steps: 1) 3′-adaptor ligation; 2) 5′adaptor ligation; 3) cDNA synthesis; 4) PCR amplification; 5) size selection of ~135-155 bp PCR amplified fragments (corresponding to ~15-35 nt small RNAs).The libraries were denatured as single-stranded DNA molecules, captured on Illumina flow cells (Illumina, Inc, San Diego, CA, USA), amplified in situ as clusters and finally sequenced for 36 cycles on Illumina NextSeq. 500 per the manufacturer's instructions.
To improve prediction accuracy, we choose the putative target gene identified in all three database. To gain insight into the biological functions of the differential miRNAs., the pathways of target genes were analyzed, using KEGG (http://www.genome.jp/) Validation of candidate miRNAs expression with quantitative real time RT-PCR (qRT-PCR). Total serum RNA was isolated from the validation set of serum samplesWe spiked 100 pM of synthetic cel-mir-40-3p into serum RNA.Isolated RNA wasconverted to complementary DNA (cDNA) with TaqMan miRNA Reverse Transcription Kit (Life Technologies, Foster City, CA). TaqMan (miR-26a-5p, let-7g-5p, miR-143-3p, miR-126-3p, miR-24-3p, let-7f-5p, let-7i-5p, miR 142-5p, miR-27a-3p, miR-27b-3p and miR-98-5p), RT primers, 2ng of total RNA from each sample was reverse-transcribles following the manufacture's protocol (Applied Biosystems, Foster City, California). qRT-PCR reaction was conducted in 96-well plates with RT product with TaqMan PCR master mix and TaqMan probes for each miRNA by ABI Prism Model 7900 HT instrument. Quantitative real-time polymerase chain reactions were performed in duplicate for all samples. The data analyzed using the comparative Ct method. miRNA expression levels were normalized to a nonendogenous synthetic miRNA. The PCR primers for miRNA were shown in supplementary Table 4.

Statistical analysis.
Values are expressed as mean + SEM or median. Shapiro-Wilk and Kolomogorov-Smirnov tests were used to test for non-Gaussian distribution. Student's two-sided t-test or one-way ANOVA, followed by Bonferroni's multiple comparison test as a post hoc analysis, was used for normally distributed values. For variables w.ithout normal distribution, Mann-Whitney U test or Kruskal-Wallis test was performed. Receiver operating characteristic curves were analysed to assess specificity and sensitivity of single-plasma miR-NAs and their combination using multiple logistic regression analysis. The optimal diagnostic point of the signature was assessed at cut-off values with the largest Youden' s index (sensitivity + specificity −1). To investigate the clinical impact of our signature miRNAs, we have chosen the likelihood ratio (LR) in addition to classical test parameters-sensitivity and specificity. Odds ratio was calculated by logistic regression analysis. Linear regression analysis was performed to determine the relationships between Ejection Fractions (EF) and miR-126-3p the using spearman rank correlation. Statistical analyses were performed using SAS version 9.3 (SAS Institute Inc., Cary, North Carolina), IBM SPSS Statistics 20 (SPSS,Inc., Chicago, IL, USA), and GraphPad Prism 5. The following values were considered significant: p < 0.0001(****); p < 0.001(***); p < 0.01(**); 0.01 < p < 0.05(*).