DOI: https://doi.org/10.21203/rs.3.rs-2254466/v1
Few studies have compared the performances of those reported miRNAs as biomarkers for heart failure with reduced EF (HFrEF) in a population at high risk. The purpose of this study is to investigate comprehensively the performance of those miRNAs as biomarkers for HFrEF. By using bioinformatics methods, we also examined these miRNAs' target genes and possible signal transduction pathways. We collected serum samples from patients with HFrEF at Zhongshan Hospital. Receiver operating characteristic (ROC) curves were used to evaluate the accuracy of those miRNAs as biomarkers for HFrEF. miRWALK2.0, Gene Ontology (GO) analysis, and Kyoto Encyclopedia of Genes and Genomes (KEGG) enrichment analysis were performed to predict the target genes and pathways of selected miRNAs. The study included 48 participants, of whom 30 had HFrEF and 18 had hypertension with normal left ventricular ejection fraction (LVEF). MiR-378, miR-195-5p were significantly decreased meanwhile ten miRNAs were remarkably elevated (miR-21-3p, miR-21-5p, miR-106-5p, miR-23a-3p, miR-208a-3p, miR-1-3p, miR-126-5p, miR-133a-3p, miR-133b, miR-223-3p) in the serum of the HFrEF group. All miRNAs had an area under the curve (AUC) > 0.70, except for miR-21-5p and miR-22a-3p. The combination of miR 133a-3p, miR 106b-5p, miR 1-3p, miR 133b, and miR 378 has a good diagnostic performance for HFrEF and multitudes of possible mechanisms/pathways through which dysregulation of these miRNAs may affect the crapshoot of HFrEF.
Heart failure (HF) is a complex clinical syndrome characterized by dyspnea or fatigue resulting from impaired ventricular filling, blood ejection, or both. There are three categories of HF based on the left ventricular (LV) ejection fraction (EF): heart failure with preserved ejection fraction (HFpEF) (LVEF ≥ 50%), heart failure with mid-range ejection fraction (HFmEF) (LVEF 41%-49%), and heart failure with reduced ejection fraction (HFrEF) ( LVEF ≤ 40%)[1].
N-terminal fragment of B-type natriuretic peptide (NT-pro-BNP)measurement has proved to be an effective screening tool for patients with various heart diseases, regardless of underlying etiology and the degree of systolic dysfunction of the left ventricle, which is associated with an increased risk of producing cardiovascular events[2].
A subgroup of non-coding RNA species known as microRNAs is recognized to play critical roles in post-transcriptional regulation of the expression of most protein-coding genes[3]. Functional miRNA studies have revealed that some miRNAs play a role in pathogenic mechanisms leading to heart failure, such as remodeling, hypertrophy, and apoptosis[4, 5]. Nevertheless, extracellular circulating miRNAs were first discovered in 2008[6–8]. In response to these findings, many studies have investigated miRNAs as potential disease biomarkers, including heart failure[9–11].
In previous studies, investigators have mainly focused on single miRNAs and their functions and lacked comprehensive interactions between miRNAs in the HFrEF cohort.[12–14]. Even if there are systematic studies on miRNAs and HFrEF, they primarily select healthy who do not have hypertension or diabetes as control groups[15]. However, screening for HF patients in a population at high risk for a potential diagnostic marker may be more meaningful.
Thus, we aim to this screen out the potential biological markers of potential miRNAs for HFrEF based on the hypertensive population as control while focusing on these miRNAs and studying their target genes and possible signal transduction pathways by bioinformatics methods.
Patients registered in the China National Heart Failure Registry (CN-HF) had their serum samples taken. The Shanghai Institute of Cardiovascular Diseases, Zhongshan Hospital Fudan University (the head unit) is leading the national, multicentered, prospective, and observational registry project known as the CN-HF, which also includes 50 to 100 secondary and tertiary hospitals.
HFrEF is defined as LVEF ≤ 40%. We recruited 30 HFrEF patients from April 2012 to October 2013. 23 cases were confirmed HF after MI, along with 6 cases of dilated cardiomyopathy, and 1 case of valvular cardiomyopathy. 18 hypertension patients were enrolled in the control group (NoHF). These hypertension patients should meet the following criteria: 1. No signs or symptoms of HF; 2. LVEF ≥ 50%; 3. NT-proBNP < 100 pg./mL. ALL patients with severe renal failure or comorbid diseases that indicated a life expectancy of less than a year were excluded.
All study patients got a thorough history and physical examination, NYHA functional class assessment, and phlebotomy. Peripheral venous blood samples were tested for full blood count, liver function, creatinine, HbA1C, Α2-macroglobulin, β2-microglobulin, NT-proBNP, and lipid profile.
In all cases, Doppler echocardiography was performed. In accordance with the recommendations of the European Society of Echocardiography, one of two blinded operators performed the Doppler echocardiographic assessment. All data represent the mean of three measurements on consecutive cardiac cycles. Left atrium diameter (LAD), left ventricular end-diastolic diameter (LVDd), left ventricular end-systolic diameter (LVDs), interventricular septum diameter (IVSd), left ventricular posterior wall diameter (LVPWd), Pulmonary artery pressure (PAP) and left ventricular ejection fraction (LVEF) were measured.
Donor blood samples were collected and placed at room temperature for 2 h. Separation of serum was accomplished by centrifugation at 3000rpm for 10 min. The supernatant was centrifuged to completely remove the cell debris. Total RNA was extracted from serum using miRcutes serum/plasma miRNA isolation kit (Tiangen, Beijing, China) according to the manufacturer’s instructions.
The miRNA isolated from blood sample was polyadenylated and reverse transcribed to cDNA in a final volume of 20 µl using miRcute miRNA First-Strand cDNA Synthesis Kit (Tiangen, Beijing, China). Real-time PCR was performed in duplicate using the miRcute Plus miRNA qPCR Detection Kit (SYBR Green) (Tiangen, Beijing, China). The miRNA-specific primer sequences were designed by a biologics company (Tiangen, Beijing, China). Each amplifying reaction was conducted in a final volume of 20 µl containing 1 µl of the cDNA, 0.2 mM of each primer and 1× miRcute Plus miRNA Premix (with SYBR ROX). Cat number of the primers for microRNA(Tiangen, Beijing, China) are listed in Extended data Table 1. MiR-16 was used as endogenous control[16]. Then qRT-PCR was performed in triplicate using an ABI Prism 7500 sequence detection system (Applied Biosystems). The amplification reactions were incubated at 95°C for 30 min followed by 40 cycles at 94°C for 15 s, 55°C for 30 s, and 70°C for 30 s. At the end of the PCR cycles, melting curve analyses as well as electrophoresis of the products on 3.0% agarose gels were performed. This was done to validate the specific generation of the expected PCR product. Analyses were conducted in duplicate on each sample. The expression level of the miR-19b was quantified in accordance with the cycle threshold (Ct) method. The relative expression level was calculated as follows: relative microRNA expression = 2− (ΔCt sample − ΔCt miR−16) .
MiRWALK2.0 (http://mirwalk.umm.uni-heidelberg.de) was used for the prediction of miRNA targets[17, 18]. Venn’ s diagram was plotted using Jvenn[19]. Kyoto Encyclopedia of Genes and Genomes (KEGG) enrichment analysis (mirPath V.3) was applied to identify molecular pathways that were potentially altered. Gene Ontology (GO) analysis (mirPath V.3), which included cellular component (CC), molecular function (MF), and biological process (BP), was performed to analyze the primary function of putative target genes[19]. Bar charts and bubble diagrams, respectively, were used to visualize the results of the KEGG pathway and GO terms enrichment analyses.
Data were expressed in terms of mean ± standard error or standard deviation. For relative gene expression, the mean value of the control group is defined as 1 or 100%. Comparisons of continuous variables amongst groups were performed by the student’s t test. For comparison of categorical variable, chi-square test was used. The correlation ship between miRNAs and biochemical indicators in HFrEF patients and NoHF group was evaluated by Pearson correlation coefficient. Diagnostic potential of microRNA in differentiating HFrEF from the controls was conducted by sensitivity, specificity, and area under the curve analysis (AUC). Statistical analysis was performed with SPSS 26.0 (SPSS Inc., Chicago, Illinois, U.S.A.) and GraphPad Prism 9.20 software (GraphPad Software Inc., La Jolla, CA, USA). A p value < 0.05 was considered significant.
A total of 30 HFrEF patients and 18 NoHF patients were included in this study according to the eligibility criteria. As was shown in Table 1, all populations had a similar gender split and a comparable prevalence of hypertension. The HFrEF cohort exhibits numerous traits specific to this HF subtype, including being significantly older than the NoHF cohort and demonstrating a dominant ischemic etiology. In contrast to the NoHF group, HFrEF patients had significantly higher NT-proBNP and β2-microglobulin and considerably lower glomerular filtration rate levels. Once more, these groups' echocardiographic findings are characterized by noticeably larger LAD, LVDd, LVDs, PAP and remarkable reduced EF in the HFrEF cohort.
| Baseline characteristics | NoHF patient (n = 18) | HFrEF patient (n = 30) | P-value |
|---|---|---|---|
| Age | 59.78 ± 10.15 | 66.80 ± 10.32 | < 0.05 |
| Male (%) | 10(55.6%) | 16(53.3%) | 0.884 |
| Hypertension (%) | 18(100%) | 27(90%) | 0.083 |
| Ischemic heart disease (%) | 0(0%) | 23(76.7%) | < 0.01 |
| Clinical assessment | |||
| CTnT (ng/ml) | 0.01 ± 0.01 | 0.11 ± 0.27 | 0.053 |
| Α2-macroglobulin (mg/L) | 1.76 ± 0.79 | 1.79 ± 0.43 | 0.914 |
| β2-microglobulin (mg/L) | 2.00 ± 0.54 | 3.79 ± 2.55 | < 0.01 |
| NT-proBNP (pg./ml) | 56.18 ± 28.78 | 4933.59 ± 6179.34 | < 0.01 |
| Hb (g/L) | 131.72 ± 10.64 | 130.43 ± 17.45 | 0.752 |
| WBC (X10^9/L) | 5.83 ± 1.33 | 6.81 ± 2.00 | 0.07 |
| ALT (U/L) | 24.33 ± 11.84 | 20.33 ± 12.94 | 0.290 |
| AST (U/L) | 21.06 ± 5.14 | 25.60 ± 1.65 | 0.058 |
| eGFR (mL/min/1.73m2) | 88.09 ± 25.40 | 67.52 ± 25.52 | < 0.01 |
| HbA1C (%) | 6.05 ± 1.13 | 6.15 ± 1.06 | 0.799 |
| TC (mmol/L) | 4.02 ± 0.87 | 3.80 ± 0.81 | 0.393 |
| TG (mmol/L) | 2.01 ± 1.77 | 1.34 ± 0.45 | 0.138 |
| HDL (mmol/L) | 1.20 ± 0.31 | 1.24 ± 0.80 | 0.842 |
| LDL (mmol/L) | 2.66 ± 2.62 | 2.09 ± 0.73 | 0.277 |
| Echocardiography | |||
| LAD (mm) | 38.614.88 | 46.47 ± 8.48 | < 0.01 |
| LVDd (mm) | 46.5 ± 4.11 | 65.97 ± 8.35 | < 0.01 |
| LVDs (mm) | 28.11 ± 3.01 | 54.23 ± 10.53 | < 0.01 |
| IVSd (mm) | 8.94 ± 1.70 | 8.90 ± 1.71 | 0.931 |
| LVPWd (mm) | 8.67 ± 1.24 | 8.97 ± 1.43 | 0.463 |
| PAP (mmHg) | 28.17 ± 6.63 | 42.57 ± 15.53 | < 0.01 |
| LVEF (%) | 69.00 ± 5.29 | 32.60 ± 5.99 | < 0.01 |
We performed miRNA profiling from the plasma RNA pool between NoHF patients and HFrEF patients. Among the 28 candidate miRNAs, two miRNA expressions were significantly decreased (miR-378, miR-195-5p) (Fig. 1A-B) meanwhile ten miRNAs were remarkably elevated (miR-21-3p, miR-21-5p, miR-106-5p, miR-23a-3p, miR-208a-3p, miR-1-3p, miR-126-5p, miR-133a-3p, miR-133b, miR-223-3p) in the serum of HFrEF group compared to the NoHF cohort(Figure 1C-I). The expression of other 16 miRNAs was shown in Extended data figure S1.
The optimum cut-off values for these miRNA biomarker candidates were identified by drawing ROC curves. The cut-off values are shown in Table 2.
| miRNA | Cut-off value (change fold) | Sensitivity%(95%CI) | Specificity%(95%CI) |
|---|---|---|---|
| miR 133a-3p | 2.876 | 83.33(64.44–92.66) | 94.12(73.02–99.70) |
| miR 106b-5p | 1.849 | 82.14(64.41–92.12) | 94.12(73.02–99.70) |
| miR 1-3p | 5.951 | 65.22(44.89–81.19) | 94.12(73.02–99.70) |
| miR 133b | 3.092 | 75(56.64–87.32) | 94.12(73.02–99.70) |
| miR 378 | 0.3905 | 79.17(59.53–90.76) | 90(59.58–99.49) |
| miR 195-5p | 0.7397 | 69.23(50.01–83.50) | 100(81.57–100.0) |
| miR 126-5p | 1.570 | 67.86(49.34–82.07) | 88.24(65.66–97.91) |
| miR 208a-3p | 2.299 | 66.67(47.82–81.36) | 94.12(73.02–99.70) |
| miR 21-3p | 1.963 | 61.54(42.53–77.57) | 88.24(65.66–97.9) |
| miR 223-3p | 2.419 | 70.37(51.52–84.15) | 76.47(52.74–90.44) |
Notably, five miRNAs (miR 133a-3p, miR 106b-5p, miR 1-3p, miR 133b and miR 378) exhibited the greatest estimated AUCs (from 0.8375 to 0.9353). The ROC curves of these five miRNAs were integrated to test whether this improves diagnostic accuracy. As shown in Fig. 3B, the combined analysis showed an AUC of 0.996 (P < 0.01). By contrast, NT-proBNP used alone to predict HFrEF in this study yielded an AUC of 1 (Fig. 3A).
A linear correlation analysis was performed between several Echocardiographic Parameters and circulating miRNA candidates in HFrEF and NoHF patients.
Briefly, there was a negative correlation between NT-proBNP and LVEF (Fig. 4A). Interestingly, a similar trend was also found in the relationship between LVEF and the circulating miRNA candidates (miR 133a-3p, miR 1-3p, miR 106b-5p, and miR 126-5p) (Fig. 4B-E). Also, LVEF positively correlated with miR 195-5p expression (Fig. 4F). LAD showed statistically significant correlations with NT-proBNP, miR 133a-3p and miR 378 (Fig. 5); NT-proBNP, miR 126-5p, miR 21-3p and miR 195-5p were all related with LVDd (Fig. 6).
Although IVds showed no correlation with NT-pro-BNP, there was a positive correlation with miR 195-5p (Fig. 8). Similarly, a marked and relevant negative correlation was also noted between PAP and miR 378 (Fig. 9).
The target genes of the five potential biomarker candidates (miR 133a-3p, miR 106b-5p, miR 1-3p, miR 133b, and miR 378) were predicted using the miRWALK2.0 software. Among them, 130 overlapping genes (Extended data Table 2.) were extracted and plotted in the Venn diagram (Fig. 10A).
A KEGG analysis was conducted to identify the main pathways in which the candidate target genes may be engaged, which will help us better understand the biological functions of the predicted target genes. The top five pathways were endocytosis, axon guidance, TGF-beta signaling, Ubiquitin mediated proteolysis pathway, and gap junction after cancer-related pathways were eliminated (Fig. 10B). The main biological processes, molecular functions, and pathway analyses determined by GO analysis of the target genes of the five analyzed miRNAs are summarized in Fig. 10B. It demonstrated that the overlapping differentially expressed genes were dramatically concentrated in the organelle, ion binding, cellular nitrogen compound metabolic process, biosynthetic process, and cellular component (Fig. 10C).
There is a growing burden of cardiovascular disease associated with HF worldwide. Patients with chronic HF have a 1-year mortality rate of 7.2% and a 1-year hospitalization rate of 31.9%, while patients with acute HF have 17.4% and 43.9%, respectively[20]. Although the N-terminal fragment of NT-pro-BNP is a potential marker of heart failure as outlined in the European guidelines from 2008[21], circulating diagnostic biomarkers-microRNAs have their advantages. Unlike mRNAs, microRNAs are stable at room temperature and remain so after repeated freeze-thawing[8]. The present study analyzed the circulating miRNA signature of patients with HFrEF. The results of a genome-wide microarray followed by an independent qRT-PCR analysis demonstrated that two plasma microRNAs (miR 378 and miR 195-5p) were significantly downregulated while ten circulating miRNAs (miR 21-3p, miR 21-5p, miR 106b-5p, miR 23a-3p, miR 208a-3p, miR 1-3p, miR 126-5p, miR 133a-3p, miR 133b and miR 223-3p) were remarkably upregulated in HFrEF patients compared with their hypertensive controls. The AUC of all differentially expressed miRNAs was more significant than 0.70, except for miR-21-5p and miR-22a-3p. ROC analysis revealed that the combination of miR 133a-3p, miR 106b-5p, miR 1-3p, miR 133b, and miR 378 had similar discriminatory abilities in identifying HFrEF as NT-proBNP, which is an acknowledged biomarker. Drawing ROC curves for these miRNA biomarker candidates were used to determine the cut-off points for these biomarker candidates. It should be noted that the cut-off values here were change-fold, which was a relative ratio. The absolute quantification of miRNA is challenging to define and is greatly influenced by each batch's reagents, operation, and other factors. Therefore, the diagnostic value of miRNA may be somewhat affected in practical operation. The absolute RT-qPCR method, on the other hand, is capable of determining the exact number of copies of a miRNA.The signal in an unknown sample is compared to a standard curve to achieve this[22]. Recently, digital RT-PCR has been used to quantify miRNA absolute levels. Digital PCR has the inherent advantage over conventional PCR in that it does not require external calibration (standard curves) or normalization in order to estimate the concentration of an unknown target[22].
It is reported that the overexpression of miR-133a significantly decreased fibrosis in rats with chronic heart failure by inhibiting the serine/threonine kinase Akt[23]. MiR-133 overexpression also suppresses the expression of multiple genes in fibroblasts, concurrently activating cardiac reprogram[24]. MiR-378 plays a dual role in suppressing cardiac hypertrophy and fibrosis through a paracrine mechanism[24]. In our study, some of these diagnostic candidate miRNAs are also associated with the LVEF (miR 133a-5p, miR 1-3p, miR 106b-5p, miR 126-5p, and miR 195-5p) and other echocardiographic parameters(LAD, LVDd, LVDs, IVSd, and PAP), which predicts that these miRNAs may be involved in myocardial hypertrophy or myocardial remodeling and deserve further investigation.
Further, we made a bioinformatic analysis of these HFrEF diagnostic candidates. It was determined that these five candidate miRNAs target 130 genes that are co-expressed using the miRWALK2.0 software. GO analysis demonstrated that the overlapping differentially expressed genes were dramatically concentrated in the organelle. As part of the KEGG analysis, we investigated whether there were any common enriched pathways related to pathophysiological processes associated with heart failure. We found support for some familiar pathway enrichment results with heart failure for the Mitogen-activated protein kinas (MAPK) signaling pathway, ErbB signaling pathway, and TGF-beta signaling pathway.
MAPK signaling cascades are critical regulators of cardiac hypertrophic response[25]. Liang reported that inhibiting p38 MAPK can reduce cardiomyocyte growth in response to hypertrophic stimuli[26]. In addition, chronic activation of the p38 MAPK pathway has been demonstrated to induce hypertrophic responses in cultured cardiomyocytes[27, 28].
Animal models have demonstrated significant changes within the cardiac NRG-1/ErbB pathway during the progression of chronic HF. In the early stages of the disease, NRG-1/ErbB expression is elevated and declines just after the pump fails[29, 30].
It has been demonstrated that the TGFβ signaling pathway plays a role in cardiac remodeling[31]. Increased TGFβ1 expression is instrumental in heart hypertrophy[32] and cardiomyocyte apoptosis[32]. It is also reported that there is a climacteric link between miR-34a, cardiovascular fibrosis, and Smad4/TGFβ1 signaling pathway[33].
New visions were also provided into Endocytosis, axon guidance, gap junction, and regulation of actin cytoskeleton.
Our study has some limitations: 1) the number of patients was relatively small, which reduced our statistical power. Therefore, more extensive studies should be conducted to confirm the diagnostic value of miRNAs. 2) it is skeptical whether these circulating miRNAs are released from cardiomyocytes, fibroblasts, macrophages, or even by non-cardiovascular tissues due to the secondary consequence of HF.
The combination of 133a-3p, 106b-5p, 1-3p, 133b, and 378 has excellent diagnostic performance for HFrEF, and there is a throng of mechanisms and pathways by which regulation of these miRNAs may affect the risk of HFrEF.
Ethical Approval
All participants provided written informed consent to participate and consent to publish in the study, which was authorized by the Zhongshan Hospital Ethical Committee of Fudan University, China, in the declaration of Helsinki (approval number: B2012-140(2)).
Competing interests
None declared.
Authors' contributions
Z.K., W.G., X.Z., and J.Y. were responsible for the conception and design. J.Y. provided administrative support. Y.M., X.Z., Y.W., Y.Y. were responsible for provision of study materials or patients. W.G., Y.M., X.Z., X.Z. were responsible for collection and assembly of data. Z.K., X.Z., and J.Y. were responsible for data analysis and interpretation. All the authors participated in manuscript writing and in final approval of manuscript.
Funding
This study was supported by the National Natural Science Foundation of China (Grant No. 81870200 and 82170279).
Availability of data and materials
Not applicable