Unraveling the Molecular Complexity of Myxomatous Mitral Valve Degeneration: Integrating Transcriptomic and miRNA Profiling Using Random Forests with Boruta Feature Selection

Abstract

Background

Myxomatous mitral valve disease (MMVD) is a degenerative disorder marked by excess tissue fibrosis and matrix remodeling, leading to leaflet prolapse. While recent studies linked transforming growth factor beta (TGF-β) activation to MMVD development and progression, upstream regulators of this and other modulatory/causal pathways remain largely unexplored.

Objectives

In this study, we utilized high-throughput sequencing to conduct unbiased analyses of mRNA and microRNA (miRNA) levels in myxomatous and healthy mitral valve tissues, aiming to uncover novel molecular mechanisms involved in MMVD.

Methods

We defined differentially expressed mRNAs and miRNAs transcripts displaying a fold-change > 1.5 and P < .05. Pathway analysis was performed using Ingenuity Pathway Analysis and DAVID, with key findings validated via qRT-PCR. A total of 2378 transcripts were differentially expressed between myxomatous and normal valves. Established pathways, including TGF-β signaling, were confirmed as major contributors to MMVD. Additionally, Random Forests with Boruta Feature Selection identified transcripts with a 95% likelihood of importance, and subsequent pathway analysis on this subset of genes revealed unique signaling pathways.

Results

Most notably, circadian rhythm disruption emerged as a novel, highly ranked pathway in MMVD. Key miRNAs, such as miR-1, miR-133a, and miR-217 were highlighted as highly relevant, with miRNA–mRNA interactions displaying distinct molecular signatures predictive of MMVD.

Conclusions

Collectively, this study represents the first comprehensive analysis of both miRNA and mRNA expression in MMVD, revealing both established and novel disease-associated pathways. The discovery of circadian rhythm disturbances and new regulatory miRNAs suggests promising directions for further research and potential therapeutic targets for nonsurgical treatment strategies in patients with MMVD.

Graphical abstract

This is a visual representation of the abstract.

Keywords

microRNAs molecular mechanisms mitral valve prolapse machine learning extracellular remodeling and fibrosis genetic omics and epigenetics

Key Points

What Is Known:

Myxomatous mitral valve disease (MMVD) is associated with excessive fibrosis and matrix remodeling, with TGF-β signaling playing a key role in disease progression.

Previous studies identified molecular pathways contributing to MMVD, but upstream regulators and novel modulators of disease remain largely unexplored.

While some circulating miRNAs have been implicated in MMVD, a comprehensive high-throughput analysis of miRNA and mRNA expression within mitral valve tissue has not been performed.

What the Study Adds:

This study provides the first unbiased high-throughput sequencing analysis of both mRNA and miRNA expression in human myxomatous and nonmyxomatous mitral valve tissue.

Novel miRNA–mRNA interactions were identified in MMVD tissue, including miR-1, miR-133a, and miR-217.

Random Forests with Boruta Feature Selection identified key pathways beyond TGF-β signaling, including GP6 signaling and FOXO hyperactivation. Aberrations in circadian rhythm gene expression emerged as a particularly novel molecular feature of MMVD using this approach, which has mechanistic links to extracellular matrix remodeling and cell proliferation.

The study highlights novel miRNA–mRNA interactions and unique regulatory networks that could serve as future therapeutic targets for MMVD.

Introduction

Myxomatous mitral valve degeneration (MMVD) is characterized by progressive structural changes in the mitral valve leaflets, leading to mitral valve prolapse (MVP), regurgitation, and ventricular dysfunction.^1–3 Despite being a common condition with significant clinical implications, the precise molecular mechanisms driving MMVD development and its progression remain elusive, warranting comprehensive investigations to deliver new insights into disease etiology and therapeutic strategies.

Transforming growth factor beta (TGF-β) signaling pathway stands out as a central player in MMVD contributing to tissue fibrosis and matrix remodeling, however, its upstream and downstream effectors in the context of myxomatous degeneration remain incompletely characterized.^4–7 Although preclinical studies suggested a role for the use of angiotensin II receptor blockers (ARBs) to mitigate TGF-β-induced fibrogenic signaling,^8,9 recent clinical trials produced conflicting evidence regarding their effect on disease progression in syndromic and nonsyndromic valvular and vascular diseases driven by TGF-β.^10–15 These data are largely confirmed by clinical veterinary experience, where ACE inhibition showed little to no benefit in delaying heart failure or reduction mortality in dogs with preclinical MMVD and cardiomegaly.¹⁶ This is not to say that subsets of patients may not benefit, as elderly patients with moderate-to-severe MMVD treated with ACE inhibitors (ACE-I) or ARBs showed improved event-free survival, making them a valuable option for patients requiring more conservative treatment.¹⁷ Recent studies highlight other potential therapeutic opportunities for serotonin (5HT) antagonism, supported largely by evidence of increased risk of mitral prolapse in small groups of patients with elevated 5HT signaling (eg, carcinoid disease), hyperactivation of 5HT signaling in valve tissue from patients with end-stage MMVD, mechanistic work linking hyperactivation of 5HT signaling to induction of matrix remodeling genes in canine mitral valve interstitial cells, and findings that 5HTR2B inhibition slows the progression of MVP in experimental animals.¹⁸ Beyond hyperactivation of TGFβ, 5HT signaling, and matrix remodeling genes, however, additional molecular mechanisms underlying the pathogenesis of MMVD remain poorly understood.

In our previous efforts to identify novel therapeutic targets in MMVD,⁷ we used standard, microarray analyses on valve tissue from patients with MMVD and identified TGF-β, bone morphogenetic protein signaling (a TGF superfamily member), Wnt/β-catenin signaling (which can be transactivated by TGF-β), and immune cell profiles as molecular signatures strongly associated with MMVD. Of particular interest was the substantive overlap of these signaling pathways with those observed in calcific aortic valve stenosis, suggesting that the phenotypic consequences of the sustained hyperactivation of these pathways is highly context dependent. Consistent with this hypothesis, a recent study demonstrated that inhibition of Wnt signaling (a pathway implicated in aortic valve calcification and stenosis) reduced myxomatous valve degeneration and immune cell infiltration in a Marfan mouse model, but that this benefit was only evident when treatment was started in early life (ie, prior to 2 months of age).¹⁹ Intriguingly, this observation compliments the work of Norris and colleagues, who found that genetic mutations impairing function of the primary cilia (which is only active during development) are sufficient to cause subclinical aberrations in the extracellular matrix that cause MMVD later in life.^20,21

To build on these findings—and in an attempt to expand the field beyond TGF-β signaling—we comprehensively investigated the regulatory networks governing MMVD pathogenesis, and present herein an integrative approach that combines RNA-sequencing data, microRNA (miRNA) profiling, and advanced analytical techniques to enrich analyses based on statistically defined, highly relevant genes. Our rationale for taking this approach is based on the interconnected nature of miRNA–mRNA interactions, where dysregulated miRNAs may be key orchestrators of the global transcriptomic fingerprints observed in MMVD. To handle the complexity of these extensive datasets and identify highly relevant, differentially expressed mRNAs and miRNAs, we employed Random Forests (RF) with Boruta Feature Selection (RF-B) followed by pathway analysis only on those genes with a likelihood of importance greater than 95%. Ultimately, the goal of using this novel dataset and data processing approach was to deepen our understanding of MMVD at the molecular level, elucidate the intricate molecular networks that underlie this complex valvular disorder, and uncover novel therapeutic targets to mitigate MMVD's clinical burden.

Methods

RNA and miRNA Isolation

Intact human mitral valve tissue was pulverized and transferred to a solution of Lysis/Binding Buffer and 0.1% miRNA Homogenate Additive (Ambion, Life Technologies). Total RNA was extracted from lysates using mirVana™ miRNA Isolation spin columns (Ambion, Life Technologies). A table listing all reagents, vendors, and expanded methods is provided in the online supplemental materials.

Nonbiased High-Throughput RNA and miRNA Sequencing

We utilized whole genome MAPRseq v1.2.1.3 for mRNA sequencing and CAP-miRSeq v1.1 for miRNA sequencing to assess mRNA and miRNA expression in nonmyxomatous (n = 10) and myxomatous (n = 10) human mitral valve tissues.²² Differential RNA expression was determined using linear modeling, with significance thresholds set at P < .05 and fold-change > 1.5. Canonical pathways were identified through Ingenuity Pathway Analyses²³ and DAVID bioinformatics.^24,25 Identification of predicted miRNA targets was accomplished using Ingenuity Pathway Analyses and miRbase,²⁶ miRDB,^27,28 and TargetScanHuman 6.2 bioinformatics tools.^11,29

Confirmation of Differentially Regulated mRNAs and miRNAs in MMVD

Differential expression of genes based on highest and lowest fold change between nonmyxomatous (n = 43) and myxomatous human mitral valves (n = 43) was confirmed by quantitative real-time polymerase chain reaction on a StepOne Plus RT-PCR machine (qRT-PCR; see expanded methods and Tables SM1-2 in the Data Supplement for gene expression primers used). Confirmation of miRNAs based on highest and lowest fold change was based on unpaired t tests with an alpha of P < .05.

Random Forest With Boruta Feature Selection

The Boruta³⁰ approach of variable importance measure has been shown to be one of the most powerful³¹ variable selection procedure in RF.³² The essential idea is to create shadow random variables by permuting the values of each original feature independently, and evaluate whether the original features contains more significant information than these shadow variables. These shadow variables serve as a baseline for comparison since they are completely unrelated to the outcome. After the creation of shadow variables, the dataset is then extended by appending these shadow variables to the original features, and RFs are refit on the augmented data. For each feature, the importance score is calculated using the standard RF mechanism (eg, mean decrease in impurity). The importance of each original feature is compared against the distribution of importance scores for the shadow variables. This allows for the computation of a P-value, which quantifies the probability that a feature's importance could have arisen by chance, which is the case of the shadow variables. This comparison-based approach to variable selection is particularly powerful, as it provides an objective baseline derived from the shadow variables, a property not feasible with standard RF variable importance measures alone. As an application of this method, by systematically eliminating variables with low importance, the Boruta approach significantly reduces the dimensionality of the data without sacrificing the predictive power of the model. Furthermore, this process inherently increases the stability of the model by focusing only on truly relevant features.

The RF-B was executed on a computer system with Intel Core i7-975 CPU with 5GB RAM provided by the National Center for Supercomputing Application at UIUC using custom analytical software consisting of Python and R code.

Statistical Methods

All data are reported as mean ± SE. Significant differences in patient characteristics were detected using unpaired t tests for continuous variables, and with χ² (when group observations were >5) or Fisher exact test (when group observations were <5) for categorical variables. In our validation datasets using qRT-PCR, significant differences between nonmyxomatous and myxomatous human valve tissue were detected using unpaired t tests.

Results

Differentially Regulated Pathways Based on Fold-Change mRNA-seq Data

Analysis of our mRNA-seq dataset yielded 2378 differentially regulated mRNA transcripts in myxomatous mitral valves (see Supplemental Table SM5). Of all differentially regulated genes, 1002 were upregulated and 1376 were downregulated reflecting a wide range in fold change (see Figure 1A and B). Using qRT-PCR to validate a subset of these genes (5 most upregulated mRNAs and 5 most downregulated mRNAs based on fold-change), we found the upregulated genes with the greatest fold-change in myxomatous versus nonmyxomatous mitral valves were: cellular retinoic acid binding protein 2, phosphoinositide interacting regulator of transient receptor potential channels (PIRT), apelin (APLN), and SH3 and multiple ankyrin repeat domains 2 (SHANK2). ATPase phospholipid transporting 10B (ATP10B), however, was surprisingly decreased in myxomatous valves (P ≤ .05, see Figure 1C-G). The genes exhibiting the most significant downregulation in terms of fold change by RNA sequencing methodologies were: myosin heavy chain 6 (MYH6), matrix metallopeptidase 3 (MMP3), myosin light chain 7 (MYL7), natriuretic peptide A (NPPA), and troponin I3, cardiac type (TNNI3). All were confirmed to have significantly decreased expression in myxomatous mitral valves by qRT-PCR (P ≤ .05, see Figure 1H-L). The activation of fibrosis, matrix remodeling, and cellular proliferation pathways are hallmarks in MMVD which was reaffirmed in our Ingenuity Pathway Analysis (IPA) and DAVID analysis of differentially regulated genes based on fold-change. More specifically, and as shown in Table 1, we found differential regulation of genes related to hepatic fibrosis, extracellular matrix–receptor interaction, focal adhesion and integrin signaling cascades.

Figure 1.

Standard mRNA-seq analysis identified 2378 differentially regulated transcripts in myxomatous mitral valves, with 1002 upregulated and 1376 downregulated genes. We performed validation experiments using independent tissue samples from 43 normal and 43 myxomatous mitral valves and were able to validate 9 of the 10 most differentially regulated genes. Notably, key upregulated genes in the RNA-seq dataset that were successfully validated included CRABP2, PIRT, APLN, and SHANK2, while ATP10B was unexpectedly decreased. Reciprocally, key downregulated genes in the RNA-seq dataset that were successfully validated included MYH6, MMP3, MYL7, NPPA, and TNNI3. All changes were significant (P < .05), comparing myxomatous to non-myxomatous valves (n = 43, 43).

Table 1.

Ingenuity pathway analysis						DAVID
Rank	Pathway	P value	# of total implicated miRNA and mRNAs	# of miRNAs that target a differentially regulated mRNA	# of unique miRNA target genes that are differentially regulated	Rank	Pathway	P value	# of total implicated miRNA and mRNAs	# of miRNAs that target a differentially regulated mRNA	# of unique miRNA target genes that are differentially regulated
1	Hepatic fibrosis / hepatic stellate cell activation	2.00E-18	80	14	16	1	Focal adhesion	1.50E-09	75	10	7
2	Hepatic fibrosis signaling pathway	1.26E-16	117	19	19	2	Hypertrophic cardiomyopathy (HCM)	3.50E-09	38	4	2
3	Axonal guidance signaling	3.98E-16	138	21	23	3	ECM-receptor interaction	2.30E-08	36	1	1
4	Cardiac hypertrophy signaling (enhanced)	2.51E-10	118	16	19	4	Dilated cardiomyopathy	3.20E-08	39	5	3
5	Agranulocyte adhesion and diapedesis	8.51E-10	60	8	9	5	Protein digestion and absorption	1.20E-07	36	2	2
6	GP6 signaling pathway	1.07E-09	43	5	4	6	Cardiac muscle contraction	1.20E-06	32	3	1
7	STAT3 pathway	1.29E-09	48	8	8	7	cGMP-PKG signaling pathway	6.00E-06	50	4	4
8	Granulocyte adhesion and diapedesis	1.78E-09	58	9	11	8	Arrhythmogenic right ventricular cardiomyopathy (ARVC)	1.60E-05	28	3	1
9	Phagosome formation	1.74E-08	44	7	6	9	Calcium signaling pathway	1.80E-05	54	5	6
10	Signaling by Rho family GTPases	4.79E-08	63	6	7	10	Axon guidance	2.30E-05	48	10	10
11	Epithelial adherens junction signaling	5.62E-08	46	5	5	11	Adrenergic signaling in cardiomyocytes	3.00E-05	47	7	6
12	Gap junction signaling	6.17E-08	57	8	7	12	Malaria	3.40E-05	22	2	2
13	Cellular effects of sildenafil (Viagra)	2.19E-07	38	2	2	13	Platelet activation	4.10E-05	41	3	3
14	Th1 and Th2 activation pathway	2.40E-07	48	5	7	14	Aldosterone-regulated sodium reabsorption	6.30E-05	21	4	3
15	Neuroinflammation signaling pathway	2.63E-07	77	13	12	15	FoxO signaling pathway	8.30E-05	49	11	8
16	Leukocyte extravasation signaling	3.63E-07	56	9	8	16	PI3K-Akt signaling pathway	1.60E-04	89	12	11
17	RhoA signaling	3.98E-07	40	6	6	17	Amoebiasis	2.40E-04	36	5	4
18	Calcium signaling	5.75E-07	55	7	4	18	Regulation of lipolysis in adipocytes	2.70E-04	22	2	2
19	IL-6 signaling	6.03E-07	43	9	8	19	Regulation of actin cytoskeleton	6.00E-04	54	4	6
20	Atherosclerosis signaling	7.41E-07	36	2	2	20	Complement and coagulation cascades	7.00E-04	26	4	4

Differentially Regulated Pathways Based on a Subset of Highly Relevant mRNA Species From RF-B

After RF-B of 2378 differentially regulated mRNA transcripts, 588 mRNAs (394 upregulated, 194 downregulated) were identified by the Boruta Feature Selection procedure as highly relevant. Among these, 459 mRNAs (comprising 325 upregulated and 134 downregulated) had a likelihood of importance of at least 95% (P ≤ .05). Examining the differentially expressed genes using RF-B, we noticed a significant predominance of upregulated genes in myxomatous valves (Figure 2A). However, when sorted by likelihood of importance, we observed a more balanced distribution of upregulated and downregulated genes, suggesting that gene significance is not solely determined by fold-change magnitude (Figure 2B and C). We validated the highly relevant mRNAs identified through our RF-B analysis using RT-qPCR. In myxomatous valves, we observed significantly elevated gene expression levels of the most highly relevant upregulated mRNAs: synaptotagmin 17 (SYT17), plexin domain containing 1 (PLXDC1), WW domain containing E3 ubiquitin protein ligase 2 (WWP2), TGFB3, and collagen type V alpha 1 chain (COL5A1) (P ≤ .05, see Figure 2D-H). Conversely, we noted marked reductions in mRNA levels of FOS-like 2, AP-1 transcription factor subunit (FOSL2), Rho GTPase activating protein 10 (ARHGAP10), interleukin 10, salt inducible kinase 1 (SIK1), and homeodomain interacting protein kinase 2 (P ≤ .05, Figure 2I-M). Ingenuity Pathway Analysis revealed differentially regulated genes involved in signaling cascades similar to the analysis based solely on fold-change such as hepatic fibrosis and ECM–receptor interaction. However, there was robust evidence of highly relevant differentially regulated genes associated with circadian rhythm and axonal guidance signaling pathways (Table 2).

Figure 2.

Random forests with Boruta feature selection of the 2378 differentially regulated mRNA transcripts revealed 588 highly relevant mRNAs (394 upregulated, 194 downregulated). Of these 588 highly relevant mRNAs, 459 mRNAs (325 upregulated, 134 downregulated) showed a likelihood of importance of at least 95% (P ≤ .05). While upregulated genes were predominant in myxomatous valves (A), sorting by likelihood of importance (B) suggested a nuanced distribution, with upregulated genes outweighing downregulated genes when plotted by importance/ranking. Further evidence suggesting consideration of biological significance beyond fold-change can be inferred from Panel C, where there is clear unlinking of the association between fold change, likelihood of importance, and rank. We performed validation experiments using independent tissue samples from 43 normal and 43 myxomatous mitral valves and were able to validate 10 most important differentially regulated genes. Notably, key upregulated genes in the RNA-seq dataset that were successfully validated included SYT17, PLXDC1, WWP2, TGFB3, and COL5A1. Reciprocally, key downregulated genes in the RNA-seq dataset that were successfully validated included FOSL2, ARHGAP10, IL10, SIK1, and HIPK2. All changes were significant (P < .05), comparing myxomatous to non-myxomatous valves (n = 43, 43).

Table 2.

Ingenuity pathway analysis						DAVID
Rank	Pathway	P value	# of total implicated miRNA and mRNAs	# of miRNAs that target a differentially regulated mRNA	# of unique miRNA target genes that are differentially regulated	Rank	Pathway	P value	# of total implicated miRNA and mRNAs	# of miRNAs that target a differentially regulated mRNA	# of unique miRNA target genes that are differentially regulated
1	Hepatic fibrosis / hepatic stellate cell activation	5.12E-07	20	4	4	1	Circadian rhythm	1.20E-04	8	1	1
2	GP6 signaling pathway	2.63E-06	13	1	1	2	Protein digestion and absorption	4.40E-04	11	1	1
3	Axonal guidance signaling	1.86E-05	26	2	2	3	ECM-receptor interaction	1.80E-03	9	-	-
4	Hepatic fibrosis signaling pathway	2.51E-05	25	5	4	4	PI3K-Akt signaling pathway	3.60E-03	21	2	2
5	Circadian rhythm signaling	3.09E-05	7	1	1	5	Amoebiasis	6.20E-03	10	1	1
6	Factors promoting cardiogenesis in vertebrates	6.45E-05	11	2	2	6	FoxO signaling pathway	8.00E-03	12	2	2
7	Aryl hydrocarbon receptor signaling	8.70E-05	14	3	2	7	Pathways in cancer	1.30E-02	22	3	4
8	Apelin cardiac fibroblast signaling pathway	8.32 E-05	5	1	1	8	Axon guidance	1.70E-02	9	-	-
9	Adipogenesis PATHWAY	0.001	12	3	3	9	Focal adhesion	1.80E-02	13	1	1
10	Iron homeostasis signaling pathway	0.0011749	13	4	2	10	p53 signaling pathway	3.00E-02	8	2	2
11	Apelin liver signaling pathway	0.00131826	4	-	-	11	Nicotinate and nicotinamide metabolism	3.90E-02	4	-	-
12	Basal cell carcinoma signaling	0.00234423	7	1	1	12	Hippo signaling pathway	4.30E-02	11	2	2
13	VDR/RXR activation	0.00354813	7	1	1	13	African trypanosomiasis	5.30E-02	5	1	1
14	1D-myo-inositol Hexakisphosphate biosynthesis II (Mammalian)	0.00436516	3	-	-	14	Glycosaminoglycan biosynthesis - keratan sulfate	5.70E-02	4	1	1
15	Osteoarthritis pathway	0.00691831	11	1	1	15	Aldosterone synthesis and secretion	5.90E-02	7	1	1
16	Epithelial adherens junction signaling	0.00831764	8	-	-	16	Tyrosine metabolism	6.20E-02	4	-	-
17	Serotonin receptor signaling	0.00851138	5	1	1	17	Proteoglycans in cancer	7.70E-02	12	3	3
18	Role of osteoblasts, osteoclasts and chondrocytes in rheumatoid arthritis	0.00912011	13	3	3	18	Colorectal cancer	7.70E-02	6	1	1
19	NRF2-mediated oxidative stress response	0.00977237	10	1	1	19	Aldosterone-regulated sodium reabsorption	8.00E-02	5	1	1
20	LPS/IL-1 mediated inhibition of RXR function	0.01023293	11	1	1	20	Hepatitis B	8.30E-02	9	1	1

Analysis of miRNA-seq Data Based on Fold-Change

From the miRNA-seq dataset we observed 67 miRNAs differentially regulated between nonmyxomatous and myxomatous mitral valves with 24 miRNAs upregulated and 43 downregulated (see heat map in Figure 3A). For data validation, we examined the miRNA molecules with the greatest fold change using quantitative real-time PCR (5 most upregulated mRNAs and 5 most downregulated mRNAs based on fold-change). The upregulated miRNAs with the greatest fold change were hsa-miR-210 and hsa-miR-34c-5p, and both were confirmed to be significantly increased in myxomatous mitral valves (n = 43) compared to nonmyxomatous (n = 43, P ≤ .05, Figure 3F and G). Conversely from our validation data, we found no difference or slight increases in hsa-miR-216a-5p, hsa-miR-216a-3p, and hsa-miR-217 (Figure 3C-E). The most downregulated miRNAs with the greatest fold-change included hsa-miR-1, hsa-miR-133a, hsa-miR-499a-5p, and hsa-miR-143-5p, and all were significantly decreased in myxomatous mitral valves compared to nonmyxomatous mitral valves when validated with qRT-PCR (Figure 3H-J). miR490-3p failed to meet the threshold for statistical significance when validated by qRT-PCR (Figure 3L). MicroRNA-seq analysis of differentially regulated miRNA molecules using IPA identified 60 significant diseases and biological functions including cancer, organismal injury, and abnormalities, as well as cellular proliferation, fibrosis, and cellular senescence.

Figure 3.

Standard miRNA-seq analysis revealed 67 differentially regulated miRNAs in myxomatous mitral valves, with 24 upregulated and 43 downregulated. Validation experiments on independent tissue samples (n = 43 Each) confirmed 6 of the top 10 differentially regulated miRNAs. Among validated upregulated miRNAs were HsamiR-210 and hsa-miR-34c-5p, While hsa-miR-217, hsa-miR216a-5p and hsa-miR216a-3p tended to increase but lacked significance. Conversely, validated downregulated miRNAs included hsa-miR-1, hsa-miR-133a, hsa-miR-499a-5p, and hsa-miR-143-5p, while hsa-miR-490-3p showed a decreasing trend but did not reach statistical significance. All changes significant (p < 0.05), comparing myxomatous to non-myxomatous valves (n = 43, 43).

Analysis of miRNA-seq Data Based on RFs With Boruta Features Selection

Random Forests with Boruta Feature Selection identified 43 miRNAs (17 upregulated, 26 downregulated) as highly relevant, of which 34 miRNAs (15 upregulated, 19 downregulated) had a likelihood of importance of at least 95% (P ≤ .05, Figure 4A-C). Using qRT-PCR we confirmed that levels of the miRNAs hsa-miR-363-3p, hsa-miR-223-5p, hsa-miR-337-3p, hsa-miR-133a, hsa-miR-1, hsa-miR-654-3p, and hsa-miR-411-5p were significantly different between myxomatous and nonmyxomatous tissues, whereas hsa-miR-758-3p, hsa-miR-7-5p, and hsa-miR-296-3p did not reach statistical significance in this validation step (P ≤ .05, see Figure 4D-M). Two miRNAs, hsa-miR-133a and hsa-miR-1, were featured as highly significant in both analytical approaches (ie, pure fold-change and RF-B). Of the 55 disease and biological functions found, IPA largely categorized these miRNAs into psychological disorders, cancer, organismal injury and abnormalities, cellular proliferation, and cardiovascular diseases.

Figure 4.

Random forests with Boruta feature selection of the 67 differentially regulated miRNA transcripts identified 43 highly relevant miRNAs (17 upregulated, 26 downregulated), with 34 miRNAs (15 upregulated, 19 downregulated) showing at least 95% likelihood of importance (P ≤ 0.05). While downregulated miRNAs outweighed in myxomatous valves (A), sorting by importance (B) revealed a nuanced distribution with greater importance. Further evidence suggesting consideration of biological significance beyond fold-change can be inferred from Panel C, where there is clear unlinking of the association between fold change, likelihood of importance, and rank. Validation experiments on independent tissue samples (n = 43 each) confirmed 7 of the 10 most important differentially regulated miRNAs. Validated miRNAs included hsa-miR- 363-3p, hsa-miR-223-5p, hsa-miR-337-3p, hsa-miR-133a, hsa-miR-1, hsa-miR-654-3p, and hsa-miR-411- 5p, while hsa-miR-758-3p, hsa-miR-7-5p, hsa-miR-296-3p did not reach significance (P < .05), comparing myxomatous to non-myxomatous valves (n = 43, 43).

MicroRNA–mRNA Interactions Within Each Analytic Subset (Fold Change/RFs)

To integrate these two datasets, we first took highly differentially regulated miRNA species and determined whether they were strongly associated with genes clustered in highly differentially regulated signaling pathways from our pathway analyses (IPA, DAVID). As one miRNA can have multiple mRNA targets and likewise one mRNA can be targeted by various miRNAs we examined the total number of implicated miRNA and mRNAs species and also noted the number of unique miRNAs and mRNA in each differentially regulated pathway. In our highest ranked differentially regulated pathway, Hepatic Fibrosis / Hepatic Stellate Cell Activation, we identified a total of 80 miRNA–mRNA interactions between 14 distinct miRNAs and 16 mRNAs, and four such miRNA–mRNA interactions included miRNA and mRNA species that were classified as highly relevant based on RF-B modeling. The noteworthy pathways and interactions, prominently ranked through each analytical method, are showcased in Tables 1 and 2.

Validation of Changes in Circadian Rhythm and Integration of miRNA–mRNA Interactions in MMVD

A particularly unique finding was dysregulated expression of genes implicated in the circadian rhythm pathway in myxomatous mitral valve tissues. Upon further investigation of this pathway, we discovered a total of 13 differentially regulated mRNAs (6 upregulated, 7 downregulated, Figure 5A). mRNA levels of a subset of these differentially regulated genes were confirmed using RT-qPCR with the downregulated mRNAs, basic helix-loop-helix ARNT like 1 (BMAL1), neuronal PAS domain protein 2 (NPAS2), nuclear factor, interleukin 3 regulated (NFIL3 [E4BP4]), and RAR-related orphan receptor B (RORB) were significantly decreased in myxomatous valves compared to nonmyxomatous (n = 43, 43) (P < .05, see Figure 5B-E). Upregulated genes included period circadian regulator 2 (PER2), period circadian regulator 3 (PER3), and basic helix-loop-helix family member e41 (BHLHE41 [DEC2]), which were confirmed to be significantly increased in MMVD with a modest increase in cryptochrome circadian regulator 2 (CRY2) and nuclear receptor subfamily 1 group D member 2 (NR1D2 expression) (see Figure 5F-J).

Figure 5.

In the circadian rhythm pathway, we found 13 differentially regulated mRNAs (6 upregulated, 7 downregulated). RT-qPCR confirmed downregulation of BMAL1, NPAS2, NFIL3, and RORB, and upregulation of PER2, PER3, and BHLHE41 (DEC2) in myxomatous valves compared to non-myxomatous (n = 43, 43, p < 0.05).

From our integration of the highly differentially regulated circadian rhythm pathway and miRNAs via IPA, we have identified the highly relevant miRNA–mRNA interaction of hsa-miR-133a and PER2. As stated previously, hsa-miR-133a was identified as both a highly ranked based on fold-change and highly relevant miRNA based on our RF-B modeling. We further confirmed by qRT-PCR that miR-133a was significantly decreased and mRNA levels of PER2 were significantly increased in myxomatous mitral valve tissue (P < .05, see Figures 3I and 5F). Figure 6 delineates a model representation of these highly relevant transcriptional changes in relation to circadian rhythm.

Figure 6.

This model outlines the significant transcriptional changes in the circadian rhythm pathway occurring in MMVD, with red denoting upregulated and green indicating downregulated molecules. Integration of mRNA and miRNA data revealed hsa-miR-133a's interaction with PER2, validated by qRT-PCR, showing decreased hsa-miR-133a and increased PER2 expression in myxomatous valves (n = 43, 43, P < .05). Created in BioRender. Hagler (2025) https://BioRender.com/g01f275.

Discussion

To our knowledge, this is the first report of high-throughput sequencing of both miRNA and mRNA species in human myxomatous and nonmyxomatous mitral valve tissue. Our study yielded five key findings: (1) several of the highly differentially regulated mRNAs map on to highly differentially regulated pathways previously implicated in MMVD but also identify several other novel molecular mechanisms not characteristically associated in MMVD, (2) numerous miRNA species are differentially regulated in MMVD, (3) many highly differentially regulated transcripts—for both miRNA and mRNA species—do not predict presence of MMVD using RF-B, (4) miRNA–mRNA interactions and pathway analyses from “highly relevant” transcripts result in significantly different molecular fingerprints predicting presence of MMVD, and (5) our data specifically implicate aberrations of circadian clock gene expression as a novel molecular change strongly associated with presence of MMVD. Collectively, using this exploratory method our findings lay a novel foundation for subsequent research efforts targeting these highlighted molecular pathways and highly relevant regulatory miRNA mechanisms that have previously gone unrecognized in MMVD.

Established and Novel Molecular Contributors to MMVD

Using standard analyses for RNA-seq data, our dataset strongly implicated pathways associated with hyperactivation of TGF-beta signaling and cellular proliferation, in MMVD. This finding is not particularly novel given the extensive body of previous reports implicating TGF-beta in the extensive matrix production and remodeling that are hallmarks of MMVD, including previous reports from our research group.^5,7 There were, however, several novel genes identified from our analyses that significantly modulate TGF-beta signaling, including SIK1, WWP2, and LOXL2. Given the challenges associated with the direct, systemic targeting of TGF-beta signaling, exploring therapies that would alternatively target modulators of this pathway may prove to be a more fruitful approach toward attenuating this major biological pathway that is so centrally implicated in the evolution of MMVD.

While standard IPA analyses of a subset of highly relevant genes (identified through RFs) yielded similar findings of TGF-beta hyperactivation, cellular proliferation, and cellular hypertrophy, several other novel pathways associated with MMVD also emerged, including GP6 signaling, molecular signals associated with chemotaxis, hyperactivation of adipogenic signaling, FOXO hyperactivation, and suppression of circadian rhythm. Importantly, these findings extend upon previous reports from our laboratory implicating TGF-beta signaling, BMP signaling, Wnt/beta-catenin signaling, and immune cell infiltration in the presence and progression of MMVD, but also draw interesting parallels to a growing body of literature implicating these pathways in extracellular matrix remodeling and cellular proliferation in various cancers.^33–35 It is also interesting to note that the upregulation of adipogenic signaling may be a key modulator of classically “osteogenic” signaling pathways activated in MMVD, thereby shifting the phenotypic consequence of these pathways toward cellular proliferation and matrix remodeling in the myxomatous valve. Ultimately, we feel the continued molecular characterization of diseased valvular tissue using such nonbiased, high-throughput methods in a variety of patient cohorts will provide key insights into molecular drivers and modulators of disease progression.

Novel Characterization of miRNAs in MMVD

To our knowledge, this is the first report to characterize differentially regulated miRNAs in myxomatous mitral valve tissue using any type of high-throughput, nonbiased analytical techniques. Namely, we were successful in identifying 67 differentially regulated miRNAs in our cohorts. Several of these miRNAs—such as miR1 and miR133a—have been implicated and exhaustively studied in a broad range of cardiovascular phenotypes and other pathologies (eg, tissue fibrosis,³⁶ matrix remodeling, valvular calcification,³⁷ cellular proliferation,³⁸ cellular hypertrophy³⁹). Interestingly, the majority of validated miRNA transcripts identified in our study did not have substantive overlap with recent reports of circulating miRNAs that predict presence of MMVD, suggesting that circulating miRNAs may be more reflective of systemic stressors and adaptations than the valve microenvironment itself.⁴⁰

In an effort to more precisely identify miRNAs that are highly likely to influence disease progression in MMVD, we again performed RF-B to identify a subset of transcripts of greatest relevance. Using this approach, only 34 of the 67 transcripts reached our threshold of 95% likelihood of importance in MMVD, and only miR1 and miR133a remained in the top 10 of both analytical approaches. Importantly, our analyses—combined with proof-of-concept validation in other clinical conditions^41–45—suggest that the outputs of this work are likely to have significant clinical and mechanistic relevance. Future work focused on in vitro and in vivo mechanistic validation of the relevance of these putative targets will nonetheless be critical in confirming their therapeutic utility and advancing our understanding of MMVD.

Aberrations in Circadian Clock Gene Expression are a Novel Molecular Characteristic of MMVD

While standard IPA analyses of a subset of highly relevant genes (identified through RF-B analysis) yielded similar findings of TGF-beta hyperactivation, we were surprised to find aberrations in circadian rhythm as the lead candidate pathway following DAVID analysis of RF-B-enriched genes. There are numerous reports that disturbance of the circadian rhythm—either from clinical or epidemiological factors—is strongly associated with an increased prevalence of adverse cardiovascular outcomes.^46–50 Given the strong emergence of aberrations in circadian rhythm in our analyses, we were surprised by the paucity of epidemiological evidence suggesting a link between disturbances in the circadian rhythm and prevalence of MMVD. Systemic alterations in circadian rhythm (eg, losses in variability of HR and blood pressure⁵¹) can frequently be a consequence of sympathetic activation and/or parasympathetic withdrawal and increased blood pressure, which has direct implications for the mechanical stresses placed on the mitral valve.⁵² Interestingly—and perhaps most directly related to molecular mechanisms influencing progression of MVP, loss of circadian gene patterns in local tissues has been shown to directly regulate extracellular matrix deposition and turnover through a broad range of mechanisms.^53,54 Thus, future work focused on the mechanistic interactions between both systemic and local changes in circadian rhythm will be critical to advancing our understanding of MMVD.

Limitations

A primary limitation of this study was our inability to control for differences in age, left ventricular function, or medications between our nonmyxomatous and myxomatous cohorts. Given that the nonmyxomatous tissue typically originated from explanted hearts not suitable for transplant, ejection fraction was significantly lower, which can place strain on the mitral valve apparatus and/or induce secondary regurgitation. While a number of patients in the nonmyxomatous group did have echocardiographic evidence of mild or moderate regurgitation secondary to left ventricular dilatation, exploratory assessments of our data suggest that these points were evenly distributed throughout the data range for patients without mitral regurgitation (see Supplemental Figures SM11). Furthermore, patient medical records included reports from anatomic pathology stating an absence of mitral valve thickening and/or myxomatous degeneration. Similar qualitative subset analyses suggested negligible effects of age or pharmacological management on the directionality or distribution of data within a gene set. Notably, the proportion of patients receiving angiotensin II modulating pharmacotherapies (angiotensin converting enzyme inhibitors or angiotensin receptor blockers)—which have been implicated in the modulation of TGF-beta signaling^14–17—did not differ significantly between groups.

We also note that a small subset of transcripts detected in our sequencing data were not confirmed in our larger validation cohort. This is not uncommon when performed appropriately in developmental and/or molecular screening studies and can be due to increased variability in a separate cohort, population heterogeneity, or a reduced effect size.^55–59 Regardless of the factor driving the failure to validate a gene, this regression toward a “true mean”/”true difference” suggests reduced relevance at a general population scale. Collectively, however, we feel our approach is robust and minimizes the risk of a Type I Error for the most novel molecular signals identified and emphasized within this report (ie, circadian rhythm).

Conclusion

This study is the first to report nonbiased screening of mRNA and miRNA signatures in MMVD and implicates several pathways that are novel but importantly harbor mechanistic links to known molecular drivers of mitral valve disease. Future efforts aimed at investigating the putative targets that we highlight herein will be pivotal to advancing our understanding of MMVD and guiding the development of novel nonsurgical/noninterventional treatment options for patients with this disease.

Supplemental Material

sj-docx-1-hvs-10.1177_30494826251362814 - Supplemental material for Unraveling the Molecular Complexity of Myxomatous Mitral Valve Degeneration: Integrating Transcriptomic and miRNA Profiling Using Random Forests with Boruta Feature Selection

Supplemental material, sj-docx-1-hvs-10.1177_30494826251362814 for Unraveling the Molecular Complexity of Myxomatous Mitral Valve Degeneration: Integrating Transcriptomic and miRNA Profiling Using Random Forests with Boruta Feature Selection by Michael A Hagler, Nassir Thalji, Nate T Russell, Michael E Welge, Ruoqing Zhu, Colleen Bushell, Matthew Berry, Bryan A White, Aleksey V Matveyenko, Rakesh M Suri and Jordan D Miller in Journal of the Heart Valve Society

Footnotes

Abbreviations

Funding

The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This work was supported by the National Heart, Lung, and Blood Institute, (grant number R01HL141819).

Declaration of Conflicting Interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Supplemental Material

Supplemental Material for this article is available online.

ORCID iDs

Michael A Hagler

Nassir Thalji

Michael E Welge

References

Freed

Benjamin

Levy

, et al. Mitral valve prolapse in the general population: the benign nature of echocardiographic features in the Framingham Heart Study. J Am Coll Cardiol. 2002;40(7):1298–1304.

Freed

Levy

Levine

, et al. Prevalence and clinical outcome of mitral-valve prolapse. N Engl J Med. 1999;341(1):1–7. doi:https://doi.org/10.1056/NEJM199907013410101

Guthrie

Edwards

. Pathology of the myxomatous mitral value. Nature, secondary changes and complications. Minn Med. 1976;59(9):637–647.

Geirsson

Singh

Ali

, et al. Modulation of transforming growth factor-β signaling and extracellular matrix production in myxomatous mitral valves by angiotensin II receptor blockers. Circulation. 2012;126(11 Suppl 1):S189–S197. doi:https://doi.org/10.1161/Circulationaha.111.082610

Hagler

Hadley

Zhang

, et al. TGF-beta signalling and reactive oxygen species drive fibrosis and matrix remodelling in myxomatous mitral valves. Cardiovasc Res. 2013;99(1):175–184. doi:https://doi.org/10.1093/cvr/cvt083

Cheng

Myers

, et al. TGF-β-dependent pathogenesis of mitral valve prolapse in a mouse model of marfan syndrome. J Clin Invest. 2004;114(11):1586–1592. doi:https://doi.org/10.1172/Jci200422715

Thalji

Hagler

Zhang

, et al. Nonbiased molecular screening identifies novel molecular regulators of fibrogenic and proliferative signaling in myxomatous mitral valve disease. Circ Cardiovasc Genet. 2015;8(3):516–528. doi:https://doi.org/10.1161/CIRCGENETICS.114.000921

Brooke

Habashi

Judge

Patel

Loeys

Dietz

. Angiotensin II blockade and aortic-root dilation in Marfan's syndrome. New Engl J Med. 2008;358(26):2787–2795. doi:https://doi.org/10.1056/NEJMoa0706585

Habashi

Judge

Holm

, et al. Losartan, an AT1 antagonist, prevents aortic aneurysm in a mouse model of Marfan syndrome. Science. 2006;312(5770):117–121. doi:https://doi.org/10.1126/science.1124287

10.

Bowman

MAH

Eagle

Milewicz

. Update on clinical trials of losartan with and without β-blockers to block aneurysm growth in patients with Marfan syndrome: a review. Jama Cardiol. 2019;4(7):702–707. doi:https://doi.org/10.1001/jamacardio.2019.1176

11.

Mallat

Daugherty

. AT1 receptor antagonism to reduce aortic expansion in Marfan syndrome lost in translation or in need of different interpretation? Arterioscl Thromb Vasc Biol. 2015;35(2):E10–E12. doi:https://doi.org/10.1161/Atvbaha.114.305173

12.

Milleron

Arnoult

Ropers

, et al. Marfan Sartan: a randomized, double-blind, placebo-controlled trial. Eur Heart J. 2015;36(32):2160–2166. doi:https://doi.org/10.1093/eurheartj/ehv151

13.

Mullen

Jin

Child

, et al. Irbesartan in Marfan syndrome (AIMS): a double-blind, placebo-controlled randomised trial. Lancet. 2019;394(10216):2263–2270. doi:https://doi.org/10.1016/S0140-6736(19)32518-8

14.

Peng

, et al. Losartan attenuates myocardial endothelial-to-mesenchymal transition in spontaneous hypertensive rats via inhibiting TGF-beta/smad signaling. PLoS One. 2016;11(5):e0155730. doi:https://doi.org/10.1371/journal.pone.0155730

15.

Zong

Zhao

Zhang

. Irbesartan ameliorates myocardial fibrosis in diabetic cardiomyopathy rats by inhibiting the TGFbeta1/Smad2/3 pathway. Exp Ther Med. 2020;20(5):117. doi:https://doi.org/10.3892/etm.2020.9245

16.

Donati

Tarducci

Zanatta

, et al. Angiotensin-converting enzyme inhibitors in preclinical myxomatous mitral valve disease in dogs: systematic review and meta-analysis. J Small Anim Pract. 2022;63(5):362–371. doi:https://doi.org/10.1111/jsap.13461

17.

Uzel

Bruno

Jung

, et al. Clinical impact of ACE-I/ARB for conservatively treated patients with moderate to severe mitral regurgitation: a single center observational study. J Cardiovasc Dev Dis. 2023:10(4):177. doi:https://doi.org/10.3390/jcdd10040177

18.

Driesbaugh

Branchetti

Grau

, et al. Serotonin receptor 2B signaling with interstitial cell activation and leaflet remodeling in degenerative mitral regurgitation. J Mol Cell Cardiol. 2018;115:94–103. doi:https://doi.org/10.1016/j.yjmcc.2017.12.014

19.

Alfieri

Guo

Yutzey

. Wnt signaling inhibition prevents postnatal inflammation and disease progression in mouse congenital myxomatous valve disease. Arterioscler Thromb Vasc Biol. 2024;44(7):1540–1554. doi:https://doi.org/10.1161/ATVBAHA.123.320388

20.

Durst

Sauls

Peal

, et al. Mutations in DCHS1 cause mitral valve prolapse. Nature. 2015;525(7567):109–113. doi:https://doi.org/10.1038/nature14670

21.

Toomer

Fulmer

, et al. Primary cilia defects causing mitral valve prolapse. Sci Transl Med. 2019;11(493):eaax0290. doi:https://doi.org/10.1126/scitranslmed.aax0290

22.

Kalari

Nair

Bhavsar

, et al. MAP-RSeq: Mayo analysis pipeline for RNA sequencing. BMC Bioinform. 2014;15:224. doi:https://doi.org/10.1186/1471-2105-15-224

23.

Kramer

Green

Pollard J

Tugendreich

. Causal analysis approaches in ingenuity pathway analysis. Bioinformatics. 2014;30(4):523–530. doi:https://doi.org/10.1093/bioinformatics/btt703

24.

Sherman

Hao

Qiu

, et al. DAVID: a web server for functional enrichment analysis and functional annotation of gene lists (2021 update). Nucleic Acids Res. 2022;50(W1):W216–W221. doi:https://doi.org/10.1093/nar/gkac194

25.

Huang da

Sherman

Lempicki

. Systematic and integrative analysis of large gene lists using DAVID bioinformatics resources. Nat Protoc. 2009;4(1):44–57. doi:https://doi.org/10.1038/nprot.2008.211

26.

Kozomara

Birgaoanu

Griffiths-Jones

. miRBase: from microRNA sequences to function. Nucleic Acids Res. 2019;47(D1):D155–D162. doi:https://doi.org/10.1093/nar/gky1141

27.

Chen

Wang

. miRDB: an online database for prediction of functional microRNA targets. Nucleic Acids Res. 2020;48(D1):D127–D131. doi:https://doi.org/10.1093/nar/gkz757

28.

Liu

Wang

. Prediction of functional microRNA targets by integrative modeling of microRNA binding and target expression data. Genome Biol. 2019;20(1):18. doi:ARTN 18 10.1186/s13059-019-1629-z

29.

McGeary

Lin

Shi

, et al. The biochemical basis of microRNA targeting efficacy. Science. 2019;366(6472):1470. doi:ARTN eaav1741 10.1126/science.aav1741

30.

Kursa

Rudnicki

. Feature selection with the Boruta package. J Stat Softw. 2010;36(11):1–13. doi:https://doi.org/10.18637/jss.v036.i11

31.

Degenhardt

Seifert

Szymczak

. Evaluation of variable selection methods for random forests and omics data sets. Brief Bioinform. 2019;20(2):492–503. doi:https://doi.org/10.1093/bib/bbx124

32.

Breiman

. Random forests. Mach Learn. 2001;45:5–32. doi:10.1023/A:1010933404324.

33.

Mammadova-Bach

Gil-Pulido

Sarukhanyan

, et al. Platelet glycoprotein VI promotes metastasis through interaction with cancer cell-derived galectin-3. Blood. 2020;135(14):1146–1160. doi:https://doi.org/10.1182/blood.2019002649

34.

Yadav

Lee

Hsu

Huang

Yang

Chang

. Computational analysis for identification of the extracellular matrix molecules involved in endometrial cancer progression. PLoS One. 2020;15(4):e0231594. doi:https://doi.org/10.1371/journal.pone.0231594

35.

Foley

Yang

Graham

Howell

. Transcriptional profiling predicts running promotes cerebrovascular remodeling in young but not midlife mice. BMC Genomics. 2019;20(1):860. doi:https://doi.org/10.1186/s12864-019-6230-z

36.

Ghafouri-Fard

Gholipour

Taheri

. Role of MicroRNAs in the pathogenesis of coronary artery disease. Front Cardiovasc Med. 2021:8. doi:ARTN 632392 10.3389/fcvm.2021.632392

37.

Nappi

Iervolino

Avtaar Singh

Chello

. MicroRNAs in valvular heart diseases: biological regulators, prognostic markers and therapeutical targets. Int J Mol Sci. 2021:22(22). doi:https://doi.org/10.3390/ijms222212132

38.

Chen

Yan

Yang

Huang

You

. miRNA-150-5p promotes hepatic stellate cell proliferation and sensitizes hepatocyte apoptosis during liver fibrosis. Epigenomics. 2020;12(1):53–67. doi:https://doi.org/10.2217/epi-2019-0104

39.

Wang

Cai

. The role of microRNAs in heart failure. Biochim Biophys Acta Mol Basis Dis. 2017;1863(8):2019–2030. doi:https://doi.org/10.1016/j.bbadis.2016.11.034

40.

Songia

Chiesa

Alfieri

, et al. Putative circulating MicroRNAs are able to identify patients with mitral valve prolapse and severe regurgitation. Int J Mol Sci. 2021;20: 22(4). doi:ARTN 2102 10.3390/ijms22042102

41.

Gholaminejad

Zare

Dana

Shafie

Mani

Javanmard

. A meta-analysis of microRNA expression profiling studies in heart failure. Heart Fail Rev. 2021;26(4):997–1021. doi:https://doi.org/10.1007/s10741-020-10071-9

42.

Thavarajah

Cai

. Impact of miRNA in atherosclerosis. Arterioscler Thromb Vasc Biol. 2018;38(9):e159–e170. doi:https://doi.org/10.1161/ATVBAHA.118.310227

43.

Satrauskiene

Navickas

Laucevicius

, et al. Mir-1, miR-122, miR-132, and miR-133 are related to subclinical aortic atherosclerosis associated with metabolic syndrome. Int J Environ Res Public Health. 2021:18(4):1483. doi:https://doi.org/10.3390/ijerph18041483

44.

Wong

Wang

Liew

Richards

Chen

. MicroRNA and heart failure. Int J Mol Sci. 2016;17(4):502. doi:https://doi.org/10.3390/ijms17040502

45.

Braundmeier-Fleming

Russell

Yang

, et al. Stool-based biomarkers of interstitial cystitis/bladder pain syndrome. Sci Rep. 2016;6:26083. doi:https://doi.org/10.1038/srep26083

46.

Abu Farha

Alefishat

. Shift work and the risk of cardiovascular diseases and metabolic syndrome among Jordanian employees. Oman Med J. 2018;33(3):235–242. doi:https://doi.org/10.5001/omj.2018.43

47.

Knutsson

. Health disorders of shift workers. Occup Med (Lond). 2003;53(2):103–108. doi:https://doi.org/10.1093/occmed/kqg048

48.

Martino

Young

. Influence of the cardiomyocyte circadian clock on cardiac physiology and pathophysiology. J Biol Rhythm. 2015;30(3):183–205. doi:https://doi.org/10.1177/0748730415575246

49.

Skogstad

Mamen

Lunde

, et al. Shift work including night work and long working hours in industrial plants increases the risk of atherosclerosis. Int J Environ Res Public Health. 2019;16(3):521. doi:https://doi.org/10.3390/ijerph16030521

50.

Vyas

Garg

Iansavichus

, et al. Shift work and vascular events: systematic review and meta-analysis. Br Med J. 2012;345:e4800. doi:ARTN e4800 10.1136/bmj.e4800

51.

Ruger

Scheer

. Effects of circadian disruption on the cardiometabolic system. Rev Endocr Metab Disord. 2009;10(4):245–260. doi:https://doi.org/10.1007/s11154-009-9122-8

52.

Katsi

Georgiopoulos

Magkas

, et al. The role of arterial hypertension in mitral valve regurgitation. Curr Hypertens Rep. 2019;21(3):20. doi:https://doi.org/10.1007/s11906-019-0928-9

53.

Dudek

Swift

Meng

. The circadian clock and extracellular matrix homeostasis in aging and age-related diseases. Am J Physiol Cell Physiol. 2023;325(1):C52–C59. doi:https://doi.org/10.1152/ajpcell.00122.2023

54.

Guo

Chatterjee

, et al. The clock gene, brain and muscle Arnt-like 1, regulates adipogenesis via Wnt signaling pathway. Faseb J. 2012;26(8):3453–3463. doi:https://doi.org/10.1096/fj.12-205781

55.

Garg

Muth

Ransom

, et al. Mutations in NOTCH1 cause aortic valve disease. Nature. 2005;437(7056):270–274. doi:https://doi.org/10.1038/nature03940

56.

Kirschenlohr

Griffin

Clarke

, et al. Proton NMR analysis of plasma is a weak predictor of coronary artery disease. Nat Med. 2006;12(6):705–710. doi:https://doi.org/10.1038/nm1432

57.

Debiec

Hamby

Jones

, et al. Contribution of NOTCH1 genetic variants to bicuspid aortic valve and other congenital lesions. Heart. 2022;108(14):1114–1120. doi:https://doi.org/10.1136/heartjnl-2021-320428

58.

Ren

Fiala

Diamandis

Kulasingam

. Pitfalls in cancer biomarker discovery and validation with emphasis on circulating tumor DNA. Cancer Epidemiol Biomarkers Prev. 2020;29(12):2568–2574. doi:https://doi.org/10.1158/1055-9965.EPI-20-0074

59.

Dhingra

Vasan

. Biomarkers in cardiovascular disease: statistical assessment and section on key novel heart failure biomarkers. Trends Cardiovasc Med. 2017;27(2):123–133. doi:https://doi.org/10.1016/j.tcm.2016.07.005

Supplementary Material

Please find the following supplemental material available below.

For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.

For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.

0.00 MB

5.05 MB