Exploration of Potential Genetic Biomarkers for Heart Failure: A Systematic Review

Patients with heart failure (HF) often present with signs and symptoms that are often nonspecific and with a wide differential diagnosis, making diagnosis and prognosis of HF by clinical presentation alone challenging. Our knowledge on genetic diversity is rapidly evolving with high-throughput DNA sequencing technology, which makes a great potential for genetic biomarker development. The present review attempts to provide a comprehensive review on the modification of major genetic components in HF patients and to explore the potential application of these components as clinical biomarkers in the diagnosis and in monitoring the progress of HF. The literature search was conducted using six databases, resulting in the inclusion of eighteen studies in the review. The findings of these studies were summarized narratively. An appraisal of the reporting quality of the included studies was conducted using a twelve-item checklist adapted from the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) checklist. The findings showed that changes in genetic components in patients with HF compared to healthy controls could be noninvasive diagnostic or prognostic tools for HF with higher specificity and sensitivity in comparison with the traditional biomarkers. This review provided evidence for the potential of developing genetic biomarkers of HF.


Introduction
Heart failure (HF) remains a major cause of mortality, morbidity, hospitalizations, and poor quality of life. There is a global rise in the incidence of HF, with a regional variation in mortality. For example, Africa exhibits the highest case fatality rate of 34% for chronic HF, which is double the world average of 16.5% [1]. Given the increasing prevalence of HF worldwide, it is of utmost importance to ensure timely treatment and management of this potentially fatal condition among patients through various pharmacological and/or nonpharmacological means.
For the better prospect of timely management and/or treatment of HF and the optimal therapeutic effects of treatment among patients, accurate and early detection of the condition needs to be achieved. In the past, detection of HF relied largely on the presence of signs and symptoms of congestion, coupled with the examination of patients' disease history [2]. However, this form of HF detection is prone to inaccuracies, particularly so among older adults because many of the characteristic signs and symptoms of HF are nonspecific, resulting in the suspicion of HF rather than giving a definitive diagnosis. This limits the utility of signs and symptoms in diagnosing HF [3,4]. Alternative ways for HF diagnosis need to be sought. It has been demonstrated that physiological changes often precede clinical deterioration that would lead to a patient attending hospital [5]. Invasive mechanisms such as pacemaker devices with physiological monitoring mechanisms can alert the physician to the clinical deterioration [6]. However, these are invasive and not all patients with HF have a pacemaker. Biomarkers, such as genes or other genetic materials found to be associated with the susceptibility of a disease, can serve as an alternative, noninvasive way for HF diagnosis. These biomarkers have become increasingly important in current medical practice as they offer an easy way to either diagnose an illness or to monitor disease progression. Indeed, biomarkers have been used to assess the status of a patient with HF along with the possibility of monitoring changes induced by patient management, primarily through the assessment of the level of expression of certain genes [7,8]. These biomarkers may, therefore, be used to assess whether therapies against HF are effective in treating the condition among patients. Consistent with this, biomarkers were suggested to have the potential to serve as indicators to the level of response to chemotherapeutic treatments and, thus, their treatment efficacies, thereby, facilitating the determination of the optimal chemotherapeutic drugs for use in treatment [9].
In addition, identification of biomarkers of HF may also potentially facilitate a better understanding of the etiologies of the disease. Indeed, previous studies had also suggested the potential of the identification of biomarkers in providing useful clues to the pathogenesis of neurodegenerative diseases, based on the discovery that variants of certain genes with known function may modify an individual's susceptibility to these diseases [10]. Moreover, certain genes may potentially be differentially expressed in patients with various diseases including cardiovascular disease, compared to nondiseased controls [11]. Based on the known function of these genes, the molecular pathways that can contribute to the disease condition can be suggested. This not only could provide clues to further research into the disease etiologies but also provide novel molecular targets for the development of new therapies against the disease. Taken together, the identification of new biomarkers contributes to the facilitation of the diagnosis and treatment of diseases.
Given the value of biomarkers in informing us on the etiologies and diagnosis of diseases including HF, a review on the identified biomarkers of HF to date is needed. This would help provide updated, summarized data on the potential molecular pathways that can be targeted for therapeutic development against HF, as well as data on further choices of markers for assessing HF treatment efficacies. To the best of our knowledge, there are no recent systematic reviews that provide a summary of the identified biomarkers of HF. In this review, we aim to provide a comprehensive overview of the findings of previous studies examining the regulations in HF, so as to summarize the genetic biomarkers identified to date. Further, we want to explore the potential application of these findings in developing biomarkers for diagnosing and monitoring HF.

Search Strategy
The present systematic review was reported according to the criteria outlined in the Preferred Reporting Items for Systematic reviews and Meta-Analyses (PRISMA) guideline [12]. Online databases of PubMed, Ovid Medline, Ovid Nursing, Embase, CINAHL, and Cochrane were searched from inception to March 2021 to find articles that described the study of genetic modification in HF. We also performed manual searches of the bibliographies of relevant articles. The detailed search strategy used for the literature search is shown in Table 1. "Heart Failure" AND "transcriptome" OR "microRNA" OR "miRNA" OR "circular RNA" OR "epigenetic" OR "long noncoding RNA" OR "long non-coding RNA" OR "single nucleotide polymorphisms" OR "genome wide association studies" OR "GWAS" OR "polygenic risk score" OR "variants" AND "population" OR "cohort" OR "human" OR "patients" OR "cases" OR "subjects" OR "participants" AND "correlated" OR "correlation" OR "correlations" OR "associated" OR "association" OR "associations" OR "allele frequencies" OR "genotype frequencies" OR "differentially expressed" OR "upregulated" OR "downregulated" AND "genes" OR "gene" OR "biomarkers" OR "transcripts" OR "RNA"

Eligibility Criteria
A study was included in the review process if it (a) was written in English, (b) reported a case-control cohort study, (c) compared the differential expression of genetic markers in healthy subjects and patients with HF. Studies were excluded if cardiac diseases other than HF were analyzed. Moreover, articles that reported animal or cellular studies only, reviews, study protocols or conference abstracts were excluded. A full-text examination was performed to confirm their eligibility for inclusion or exclusion.

Data Extraction
After a literature search by a combination of keywords, the duplicates of records between different databases were screened and eliminated manually. Then, the selection of studies to be included in this review was performed by screening the titles and abstracts of the studies. Articles that did not meet the proposed selection criteria were excluded. The remaining articles were evaluated by full-text examination to confirm their eligibility for inclusion.
Further data extraction was performed by one author and verified by a second author after the inclusion of studies was confirmed. Extracted data included study design, study setting, applied country, cohort size, characteristics of participants, method of measurement of genetic change, and major findings of the studies. Any disagreements about data extraction from studies that occurred between the two authors were resolved by discussion with the third reviewer.
Owing to the diversity of outcome measures involved in the included studies, a metaanalysis of the extracted data was not feasible. The findings of the included studies have been presented in a narrative and tubular manner.

Quality Assessment
The methodological quality assessment was conducted using twelve selected items adopted from the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) checklist for cohort studies [13]. Selected items in the assessment include whether the included studies have provided an adequate description of the following: (1) the basic characteristics of the studies, including study design and settings, the rationale of the sample size used, and eligibility criteria used in subject recruitment; (2) the methodological details, including the variables, sources of measures used, statistical methods used, and how the authors address biases; (3) the outcomes of the studies, including the number of participants involved at each stage of the study, the demographic and clinical characteristics of participants, and the unadjusted and/or adjusted estimates obtained from these data.
One mark was awarded for each item in the checklist if the item was achieved in the studies. Any items that were not achieved by the studies were given a zero score. Studies were categorized as low, moderate, or high quality if they scored <7, 7-9, or >9, respectively (arbitrary thresholds defined by authors).

Search Results
The search strategy identified 2864 potentially relevant studies. Of these 2864 publications, 1315 were duplicates and were therefore excluded. The remaining 1549 records were subjected to title and abstract screening, among which, 15 articles were excluded as they were not published in English, and 841 were further excluded as they were not original articles. For the remaining 693 records, the full texts were retrieved, 632 of which were excluded for reasons including (1) outcomes assessed not fitting the review objectives or (2) articles not involving human subjects. Finally, 61 studies were eligible and included in this review. The PRISMA flow diagram showing the different stages of the study selection process is presented in Figure 1.
achieved in the studies. Any items that were not achieved by the studies were given a zero score. Studies were categorized as low, moderate, or high quality if they scored <7, 7-9, or >9, respectively (arbitrary thresholds defined by authors).

Search Results
The search strategy identified 2864 potentially relevant studies. Of these 2864 publications, 1315 were duplicates and were therefore excluded. The remaining 1549 records were subjected to title and abstract screening, among which, 15 articles were excluded as they were not published in English, and 841 were further excluded as they were not original articles. For the remaining 693 records, the full texts were retrieved, 632 of which were excluded for reasons including (1) outcomes assessed not fitting the review objectives or (2) articles not involving human subjects. Finally, 61 studies were eligible and included in this review. The PRISMA flow diagram showing the different stages of the study selection process is presented in Figure 1.

Quality of Included Studies
The ratings of the quality of the included studies are presented in Supplementary  Table S1. Overall, the quality of the reporting of observational studies by the included studies was moderate to high. The scores of the included studies ranged from seven to eleven, and 83.6% of the included studies scored nine or above. All of the included studies described their outcomes and variables, the methods of measurement, and the number of participants at each stage well. Most of the included studies provided the characteristics of participants to show the clinical conditions of the participants. However, none of the studies described their rationale for sample size.

Quality of Included Studies
The ratings of the quality of the included studies are presented in Supplementary  Table S1. Overall, the quality of the reporting of observational studies by the included studies was moderate to high. The scores of the included studies ranged from seven to eleven, and 83.6% of the included studies scored nine or above. All of the included studies described their outcomes and variables, the methods of measurement, and the number of participants at each stage well. Most of the included studies provided the characteristics of participants to show the clinical conditions of the participants. However, none of the studies described their rationale for sample size.

Transcriptome
Eight of the included studies reported the transcriptome analysis in HF [23,29,36,41,46,54,61,71]. Two studies involved the use of peripheral blood [29,71]. Five studies involved the use of cardiac tissues [23,41,46,54,61], and one study involved the use of EAT [36]. In one study, the result of the transcriptome analysis was validated in human peripheral blood [23]. These studies reported that the differential expressed genes were associated with various pathways including (1) fibrosis [23]; (2) cell adhesion, focal adhesion, cardiac muscle contraction [54]; (3) T cell receptor signaling pathway, primary immunodeficiency, endometrial cancer, drug metabolism cytochrome P450, tyrosine metabolism, complement and coagulation cascades, and Jak-STAT signaling pathway [36]; (4) mitochondrial oxidative signaling and noncanonical autophagy [46]. Hua et al. reported that all three RNA types, i.e., mRNAs, lncRNAs, and miRNAs, could largely distinguish HF patients from healthy controls. The results suggested that HF samples had distinct transcriptomic changes at multiple molecular levels when compared to control samples, implicating the potential use of transcriptomes as biomarkers for differentiating HF and normal controls [23]. This study identified that the genes ASPN, COL1A1, COLQ, and IGFBP3 were associated with fibrosis. By investigating the survival time before heart transplantation, COL1A1 was found as a potential biomarker for HF progression. A plasma level of COL1A1 ≥ 256.5 ng/mL was associated with poor survival within 1 year of heart transplantation from HF (HR 7.4, 95% CI 3.5-15.8, Log-rank p-value <1.0 × 10 −4 ) [23]. In the other two studies, right ventricle (RV)-specific myocardial biomarkers, which were related to HF, were reported [41,46]. Di Salvo et al. reported that the top 10 differential expressed genes (DEGs) exhibited high sensitivities, specificities and predictive values included SERPINA3, SERPINA5, LCN6, LCN10, STEAP4, AKR1C1, STAC2, SPARCL1, VSIG4, and F8. Among them, STEAP4, SPARCL1, VSIG4 were differentially expressed in HF involving the RV versus HF involving the left ventricle (LV). The results supported that these three genes are potential RV-specific biomarkers [41]. Moreover, Tzimas et al. uncovered that a shift toward noncanonical autophagy in the failing RV was correlated with RV-specific WIPI1 upregulation [46].

LncRNA
Nine of the included studies reported the regulation of lncRNA in HF patients [23,29,31,34,36,41,53,54,59]. Four studies involved the examination of lncRNA in cardiac tissue, one study in EAT and one in PBMCs. In six studies, the modulation of lncRNA in HF patients was validated in peripheral blood [23,29,31,34,53,59]. In a study conducted by Greco et al. [53], 13 lncRNAs were significantly modulated (10 up-and 3 downregulated) in HF patients when compared to control subjects. In order to validate these identified HF lncRNA signatures, RNAs derived from more severe ischemic end-stage HF patients were analyzed. It was found that nine lncRNAs (CDKN2B-AS1, EGOT, H19, HOTAIR, LOC285194, RMRP, RNY5, SOX2-OT, and SRA1) were significantly modulated in a concordant manner in both end-and non-end stage HF patients. Further investigation using PBMCs from 25 HF patients and 18 healthy individuals revealed that CDKN2B-AS1, HOTAIR, and LOC285194 showed similar modulation in PBMCs and heart tissue, suggesting the potential roles of these three genes as HF biomarkers [53]. In another study, the modulation of lncRNA heart-disease-associated transcript 2 (Heat2) was characterized in different HF cohorts, indicating its role as a potential biomarker [59]. Heat2 expression was elevated in HF patients with reduced ejection fraction (HFrEF) diagnosed with dilated cardiomyopathy (DCM) and ischemic cardiomyopathy (ICM), with p = 0.01 and p = 0.02, respectively. Furthermore, in a larger cohort study with HFrEF (n = 107), Heat2 was found to have a significant discriminatory power to predict the presence of HFrEF using wholeblood samples with an AUROC of 0.705 (95% CI: 0.604-0.805, p = 0.0004) [59]. Zhang et al. demonstrated that the expression of lncRNA taurine upregulated gene 1 (TUG1) is a potentially useful plasma biomarker for the diagnosis of heart failure with preserved ejection fraction (HFpEF). TUG1 was also confirmed to have a positive correlation with major serum marker N-terminal pro-brain natriuretic peptide (NT-proBNP) for HF [34].

DNA Methylation
Five of the included studies reported the change of DNA methylation in HF [25,35,60,64,70]. All of them examined the DNA methylation pattern using peripheral blood. In one study, the results were confirmed in both cardiac tissue and peripheral blood [60]. Meder et al. reported the results when the methylation pattern of myocardial tissue and peripheral blood was compared. It was found that three epigenetic loci significantly overlapped between the tissue and blood (OR: 28; p < 0.001). The resolved genes were B9 protein domain 1 (B9D1) (hypomethylated), doublecortin-like kinase 2 (hypomethylated) and neurotrimin (hypermethylated). Among these three genes, B9D1 has the greatest potential to serve as a biomarker for HF because it was found to have an area under the curve of >87% in the peripheral blood discovery cohort with robust replication in myocardial tissue as well as in peripheral blood verification cohorts [60]. In the other four studies, various potential epigenetic biomarkers were reported. They include SLC2A1, MPV17L and PLEC [25]; METTL3, METTL4, KIAA1429, FTO and YTHDF2 [35]; HDAC9 [64]; HEY2, MSR1, MYOM3, COX17, MMP2, CTGF, miR-24-1, and miR-155 [70].
Complex diseases are mediated via genomic dysregulation in multiple tissues. However, invasive procedures are required to obtain most tissues. On the other side, whole blood is a fast, simple, and safe source for genomic studies. In this review, both tissuederived and blood-derived biomarkers were included. It has been reported that an individual's whole-blood transcriptome can significantly predict tissue-specific expression levels for~60% of the genes on average across 32 tissues, with up to 81% of the genes in skeletal muscle [78]. In the future, blood transcriptome can help detect biomarkers in some cases and to be more accurate and effective model to predict tissue-specific expression. In six of the included studies, they were included in both the screening phase and the validation phase [14,17,[22][23][24]27,30,37,53,55,56,59,66,73]. The genetic modulation profile was firstly screened in cardiac tissue [23,24,37,53] or in peripheral blood [14,17,22,27,30,55,56,59,66,73]. Then, specific genetic changes were validated in peripheral blood (with the same or different cohorts). It should be noted that these studies with screening and validation phases involved the assessment of miRNA [14,[22][23][24]30,37,55,56,66,73], lncRNA [23,53,59], mRNA [23], and circRNA [17,27] but not SNP. Since the SNP of a single gene may include many allelic variants, the complex variations in SNP make it complicated as a screening tool. However, from the findings of this review, it was observed that SNP detection of specific genes could be done in large cohorts. The sample size of two of the included studies assessing SNP involved >1000 subjects [26,28].
Novel biomarkers may supplement traditional clinical and laboratory testing to improve understanding of the complex disease process of HF and possibly serve to personalize care for those affected through better individual phenotyping. In one of the included studies, RNA-seq was used to generate the transcriptional profile of end-stage human HF right ventricular myocardium and unused donor RV myocardium to identify candidate RV myocardial biomarkers. The transcriptomes of HF RVs were reviewed with echocardiographic dysfunction versus echocardiographic normal function to identify candidate biomarkers of RV dysfunction in end-stage HF. Finally, RV-specific biomarkers (STEAP4, SPARCL1, and VSIG4) were identified by comparing the findings between RV and LV transcriptomes [31]. The other two included studies, respectively, found Ex-miR-92b-5p and Heat2 to be potential biomarkers for diagnosing HFrEF [30,59]. Studies also identified potential biomarkers for diagnosis or prognosis of left ventricular dysfunction [54] and right ventricular dysfunction [41,46]. Apart from diagnostic biomarkers, the findings in four included studies demonstrated the potential of biomarkers for prognosis [21,26,28,33,36,37,41,58,61,62]. The association of IL17RA polymorphism with mortality [26] and association of ErbB4 with cardiovascular death [28] supported that these gene polymorphisms may have the potential to be prognostic biomarkers. Heart transplantation has been the most efficient treatment for end-stage HF. However, there is very limited finding regarding the biomarkers for survival from HF onset to heart transplantation. In one of the included studies, the findings showed that COL1A1 may act as a biomarker for HF progression [23]. Plasma concentration of COL1A1 was significantly negatively correlated with HF survival time. Moreover, a plasma COL1A1 content threshold could be defined for the patients being transplanted within 1 year versus longer than 1 year. Three included studies showed that potential biomarkers were related to HF severity [37,55,68]. Akat et al. reported that circulating myomiRs may be useful for predicting the severity of HF [37]. The findings showed that myomiR changes in advanced HF reversed 3 and 6 months after LVAD implantation. The levels of the myomiRs mir-208a(1), mir-208b(1), mir-499(1), and mir-1-1(4) dropped as early as 3 months after the initiation of LVAD support, approaching normal levels. At LVAD explantation, the myomiR levels rose again with alterations comparable in magnitude to those observed at implantation. Two included studies reported that the plasma levels of miR-150-5p and miR-126 are associated with New York Heart Association (NYHA) class [55,68].

Limitations
This review explores the potential use of biomarkers in assisting in HF diagnosis and predicting prognosis. However, there are some limitations that need to be acknowledged: (1) Due to the heterogeneity of genetic components examined in the studies, pooled effects using a meta-analysis is not feasible. (2) Different ethnic groups involved in different studies may limit the generalizability of the findings. Therefore, further studies in different cohorts with different ethnicity are required. (3) Limited sample size especially of studies for the cardiac tissues may limit the detection of true effects. (4) The number of studies on biomarkers of HF is still limited when compared to traditional biomarkers, suggesting the need of more studies with a vigorous methodology to verify the findings and explore other biomarkers associated with HF.

Future Perspectives on Biomarker Research
As genome technologies evolve, individual genome sequencing can be made available to all patients using a simple saliva test. Understanding how genetic differences in individuals contribute to their susceptibility to cardiovascular diseases can help guide practitioners to give patients useful and pertinent advice for them to achieve a favorable outcome. Nowadays, large-scale genome-wide association studies and meta-analyses have provided powerful insights into polymorphisms that will be the key to efficiently diagnose diseases and adverse drug responses and direct the patients toward the drugs more likely to be of benefit based on their particular profile.
Genetic testing has been widely used to diagnose monogenic diseases. However, common diseases, such as type 2 diabetes and many neurodegenerative diseases, tend to be polygenic. Therefore, the concept of a polygenic risk score (PRS) is shown to have the potential for broad-scale clinical use. The study of specific biomarkers for specific heart diseases would feed into a growing body of evidence supporting a wider use of PRS. There are studies on the PRS and coronary artery disease (CAD), which is the most common form of heart disease and the most common cause of heart failure. Marston et al. and Damask et al. reported two studies of CAD PRS, both found evidence for the value of CAD PRS in predicting recurrent cardiovascular events; furthermore, they found that some subgroups might benefit from PCSK9 inhibitors. Another study using CAD patients from the UK Biobank, reported that using PRS could potentially help in preventing 7% more cardiovascular events when a targeted approach for screening (including PRS and traditional risk factors) is used. While these studies are very promising, there are limitations on the use of PRS in the management of patients with high PRS scores for CAD, and some have opined that further work is needed before PRS can be implemented clinically while others seem more optimistic and published a guide for the general cardiologist with the implication that PRS might soon be useful for clinical practice.
This review summarizes the first step of the use of biomarkers in HF diagnosisthe discovery of potential biomarkers. In the future, efforts are needed for qualification, verification, research assay optimization, biomarker validation, and commercialization because of the successful development of this area of research will represent a step forward in the individualization of diagnosis, therapy, and monitoring of HF.

Conclusions
Genetic variation may contribute to disease largely through the misregulation of genetic components. Examination of gene regulation thus becomes an important tool for disease diagnosis. Overall, this review demonstrated the potential use of transcriptome, miRNA, SNP, lncRNA, DNA methylation, and circRNA in HF with respect to diagnosis, prognosis, safety, specificity, and sensitivity. Given the complex physiology in HF, it is reasonable to expect that the future of biomarker testing lies in the application of multimarker testing panels, precision medicine to improve HF care delivery, and the use of biomarkers in proteomics and metabolomics to further improve HF care.
Supplementary Materials: The following are available online at https://www.mdpi.com/article/10 .3390/ijerph18115904/s1, Table S1: The ratings of the reporting quality of the included studies based on the STROBE checklist, Table S2: The characteristics of the included studies.