A Meta-analysis on Molecular Gene Expression Profiles in Coronary Artery Disease

INTRODUCTION

Coronary artery disease (CAD) is one of the leading causes of death globally, accounting for approximately 17 million morbidity and mortality cases worldwide every year. The growing geriatric populations and the increase in lifestyle-related risk factors have led to the rise in the incidence of CAD which is likely to increase consistently in the future. The pathogenesis of CAD is multifactorial, including genetic susceptibility, environmental factors, and interruptions in chief molecular mechanisms.^[1] Advances in genomic technologies, consisting of high-throughput transcriptomics and next-generation sequencing, have improved our understanding of the disease at the molecular level. These mechanisms have facilitated the recognition of specific gene expression profiles, regulatory networks, and non-coding RNAs which are essential in the disease development and progression.^[2,3] CAD includes complex mechanisms such as complex lipid metabolism, vascular remodeling, inflammation, and immune regulation. Transcriptome and genome-wide association studies (GWAS) signify the functions of oxidative stress, endothelial dysfunction, and extracellular matrix (ECM) remodeling as main mechanisms in CAD pathophysiology.^[4]

Certain studies have reported the crucial role of non-coding RNAs, such as microRNAs (miRNAs) and competing endogenous RNAs (ceRNAs), in post-transcriptional gene regulation in CAD.^[5,6] miR-146a and -21 are associated with vascular smooth muscle cell (VSMC) proliferation and inflammatory signaling, leading to disease progression. ceRNA networks, including circular RNAs (circRNAs) and long non-coding RNAs (lncRNAs), regulate the main CAD-related molecular pathways.^[6,7]

There are integrative analytical methods and validated gene co-expression network analysis (WGCNA), which depict a promising interpretation of the complex transcriptomic data in CAD research work. WGCNA has recognized modules of co-expressed genes linked with the severity of CAD, especially those related to immune cell infiltration and mitochondrial dysfunction.^[8,9] Meta-analyses of gene expression data have constantly shown molecular biomarkers like matrix metalloproteinases (MMPs), which facilitate the degradation of the ECM components, thus leading to plaque instability.^[10,11] Polymorphisms in MMP-9 have also been associated with elevated risk of CAD and poor clinical consequences.^[12] However, substantial variability exists across studies owing to the different sample sizes, population characteristics, tissue sources (e.g., vascular tissue vs. peripheral blood), and analytical methods.^[13] There are studies on inflammatory cytokines such as tumor necrosis factor-alpha (TNF-α) and interleukin-6 (IL-6), which are inconsistent, thus restricting their consistency as clinical biomarkers.^[14] These differences signify the necessity for robust systematic reviews or meta-analyses to identify unswerving molecular signatures in CAD.

There are previous integrative studies that have identified prominent genes which are significant in the pathophysiology of CAD, such as LPA and PCSK9. These genes help to regulate lipid metabolism, and a few others, such as ADAMTS7 and PHACTR1, are associated with plaque formation and coronary calcification.^[15-17] However, unpredictability across populations and experimental platforms can lead to existing gaps in the literature. Hence, the present systematic review and meta-analysis was conducted which aimed to address these gaps by involving transcriptomic datasets from various sources to recognize consistent gene expression signatures, regulatory networks, and genetic variations involved in CAD. The primary objective was to discover robust DEGs, miRNA-Messenger RNA (mRNA) interactions, and dysregulated pathways that facilitate the progression of the disease. Certain molecular processes, such as thrombosis, inflammation, and vascular remodeling, were emphasized to comprehend their function in CAD development. Hence, this comprehensive analysis helps to improve the understanding of CAD’s molecular groundworks and would help to translate these understandings into reliable biomarkers and targeted therapeutic strategies.

METHODOLOGY

Study selection process, data extraction, and quality assessment

A total of 641 articles were initially retrieved from multiple databases, of which only eight met the inclusion criteria for meta-analysis, as outlined in Figure 1. The selection process adhered to PRISMA 2020 guidelines. Two independent reviewers screened the articles and resolved any disagreements through consensus. Data extraction was standardized, capturing study design, sample size, demographics, tissue source, experimental platform, and gene expression data. Quality was appraised using an adapted Newcastle-Ottawa Scale (NOS). Meta-analysis utilized standardized mean differences and random-effects models, with heterogeneity assessed through the I² statistic.

Close

Figure 1:

Preferred Reporting Items for Systematic Reviews and Meta-Analyses 2020 flowchart for the review.

Export to PPT

Risk of bias and assessment of publication bias

The studies were assessed using a modified NOS^[18] for observational studies, evaluating participant selection, group comparability, and outcome assessment. Publication bias was examined using funnel plot [Figure 2] visualization and statistical tests, including Egger’s regression^[19] and Begg’s rank^[20] correlation. Where publication bias was suspected, sensitivity analyses were conducted to test the robustness and reliability of the meta-analytic findings and pooled estimates.

Close

Figure 2:

Funnel plot for publication bias assessment

Export to PPT

RESULTS

Table 1 summarizes 21 studies analyzing gene expression profiles in CAD patients using diverse study designs (mostly case-control), sample sizes (n = 16 to >26,000), and tissue sources, primarily peripheral blood.^[21-32] Diagnostic criteria commonly involved coronary angiography with ≥50– 70% stenosis.^[21] Participants ranged from 24 to 88 years, predominantly male. CAD subtypes included stable CAD, acute coronary syndrome, acute myocardial infarction, ST-elevation myocardial infarction, and premature CAD. Platforms included microarrays (Affymetrix, Agilent), RNASeq, and polymerase chain reaction (PCR)-based methods (quantitative reverse transcription PCR, Restriction fragment length polymorphism -PCR, and TaqMan).^[22-25,33] Some studies used Gene Expression Omnibus datasets for meta-analyses and validation.^{[26,27,29,34]} Genetic and transcriptomic association studies featured large cohorts from the UK Biobank and All of Us programs.^[30,35,36] Studies span diverse regions (India, China, UK, USA, Europe, Middle East), reflecting varied methodological strategies in CAD molecular research.

Table 2 summarizes the findings from the studies exploring the molecular mechanisms underlying CAD, with a focus on DEGs, ncRNAs, and genomic regulatory networks (GRNs). Arvind et al.^[21] identified 190 DEGs, with significant upregulation of pro-inflammatory genes such as IL-8 and early growth response 1 (EGR1), each showing over 8-fold expression, underscoring the central role of immune activation in CAD. Kazmi et al.^[14] reported 636 DEGs and four lncRNAs, revealing sex-specific gene expression patterns influencing disease progression. Similarly, Kang et al.^[8] detected 610 DEGs and 92 lncRNAs, highlighting chemokines CXCL1 and CXCL8 as key biomarkers associated with vascular inflammation. In the domain of post-transcriptional regulation, Abdallah et al.^[28] reported 14 DEGs and 14 miRNAs, with miR-182 showing a 6.12-fold upregulation, pointing to its diagnostic potential. Al Hageh et al.^[31] identified 14 CAD-linked SNPs, particularly PHACTR1 rs9349379G, associated with a 44% elevated risk (odds ratio [OR] 1.44). Liou et al.^[3] demonstrated that polygenic risk scores (PRSs) could explain up to 22% of CAD subtype variance. These findings underscore the multifactorial nature of CAD and advocate for integrated multi-omics biomarkers in disease diagnosis and risk prediction.

Table 3 summarizes the quality assessment of 17 case–control studies using the NOS, which evaluates various domains. All studies scored ≥7, indicating overall high methodological quality. Studies published after 2019, including those by Zeng et al.,^[25] Kang et al.,^[8] Miao et al.,^[27] and Liu et al.,^[5] consistently achieved the maximum score of 10, reflecting improved design rigor, comprehensive confounder adjustment, and robust exposure assessment. Earlier studies, such as Arvind et al.^[21] and Beton et al.,^[11] scored slightly lower (7–8), often due to limited adjustment for socioeconomic factors or incomplete exposure data.^[28] Despite minor limitations, 94% of studies demonstrated strong case–control definitions and high-quality exposure ascertainment.^[31] Table 4 presents the NOS-based assessment of four cross-sectional studies. Three studies, including Merzah et al.,^[29] and Mahmood et al.,^[32] scored a perfect 10, indicating excellent methodological quality, including well-defined populations and reliable control for confounders. Kazmi et al.^[14] scored eight due to a lack of clarity in one selection criterion and non-response assessment. Thus, both case–control and cross-sectional studies demonstrated high quality, supporting the reliability of the review’s synthesized findings.

Figure 3 presents the ROBINS-I-based risk of bias assessment across seven domains for all included studies. Most studies exhibited low risk (green) in key areas such as confounding, participant selection, and outcome measurement, reflecting strong methodological quality. However, moderate (yellow) and unclear (blue) risks were identified primarily in domains D2 (participant selection), D3 (intervention classification), and D7 (reporting bias), especially in earlier studies such as Rodríguez-Pérez et al.^[22] and Daraei et al.^[24] Studies published post-2021 demonstrated uniformly low risk, indicating improved study design, clearer intervention protocols, and enhanced reporting transparency. Figure 4 summarizes the overall distribution of risk of bias across all domains and studies. Most domains, including bias from confounding, missing data, and outcome measurement, consistently scored low, affirming strong internal validity. However, moderate risk and unclear areas persisted in intervention classification and reporting, highlighting occasional gaps in methodological clarity. Figure 5 depicts a forest plot from a meta-analysis of eight studies evaluating molecular gene expression in CAD. The pooled OR was 1.74 (95% confidence interval [CI]: 0.85–3.56; P = 0.13), indicating non-significant results with high heterogeneity (I² = 97%). Divergent findings, such as upregulation in Arvind et al.^[21] versus downregulation in Merzah et al.,^[29] suggest methodological and biological variability, warranting further subgroup analyses.

Close

Figure 3:

Risk of bias assessment done using the individual studies – ROBINS I.

Export to PPT

Close

Figure 4:

Risk of bias assessment done using the overall studies (ROBINS I).

Export to PPT

Quality assessment was performed using the NOS for observational studies, which evaluates studies across three domains:

• Selection (maximum four stars):

  • S1: Representativeness of the exposed cohort

  • S2: Selection of the non-exposed cohort

  • S3: Ascertainment of exposure

  • S4: Demonstration that the outcome of interest was not present at the start of the study

• Comparability (maximum two stars):

  • C1–C2: Comparability of cohorts on the basis of design or analysis

• Exposure/Outcome (maximum three stars):

  • E1: Assessment of outcome/exposure

  • E2: Adequacy of follow-up duration

  • E3: Adequacy of follow-up completeness

A maximum score of 10 stars indicates high methodological quality, while studies scoring ≥7 stars were considered high quality.

In Figure 5, forest plot showing the pooled OR for upregulated versus downregulated expression across included studies using a random-effects Mantel–Haenszel model. Individual study estimates are presented with 95% CIs, and the size of the squares reflects study weight. The diamond represents the pooled effect estimate. Substantial heterogeneity was observed (I² = 97%). The overall effect was not statistically significant (Z = 1.51, P = 0.13).

Close

Figure 5:

Forest plot analysis done for the molecular gene expression profiles in coronary artery disease.

Export to PPT

DISCUSSION

CAD is a multifactorial disorder arising from complex interactions between genetic, epigenetic, transcriptomic, and environmental factors. This integrative gene expression analysis identified several DEGs associated with CAD, including MMP9, IL1β, ADAMTS7, and TAGLN (SM22-α), out of which a few coincide with those previously implicated in GWAS.^[1,15,36] The findings of this meta-analysis strengthen the polygenic nature of CAD, which was in accordance with the observations of Dai et al.^[1] They highlighted the contributions from both common and rare genetic variants and showed that among the most notable DEGs, MMP9 was significantly upregulated (logFC > 1.5, adj. P < 0.001). MMP9 mediates ECM degradation, providing plaque destabilization and rupture. Its genetic variant rs3918242 (−1562 C>T) has been constantly associated with increased CAD and myocardial infarction risk across multiple studies.^[10,11,22] MMP9 polymorphisms also exhibit pharmacogenomic relevance, especially in influencing responses to lipid-lowering therapies such as simvastatin.^[31] The upregulation of key inflammatory mediators, including IL1β, TNF-α, IL-8, and NFKB1, highlights the main function of chronic inflammation in the pathogenesis of CAD.^[5,8,14,21] These cytokines contribute to immune cell recruitment, endothelial dysfunction, and plaque destabilization, as supported by prior studies utilizing gene co-expression analyses and immune profiling.^[8,14] Arvind et al. reported over 8-fold upregulation of IL-8 and EGR1, both involved in amplifying vascular inflammation,^[21] while Kang S et al. identified the chemokines CXCL1 and CXCL8 as central players in inflammatory signaling cascades within atherosclerotic lesions.^[5]

Immune cell deconvolution revealed increased macrophage and neutrophil infiltration, reflecting pro-thrombotic and pro-inflammatory activity in CAD and aligning with findings of elevated systemic inflammation in CAD patients.^[32] ADAMTS7 was significantly overexpressed (logFC >2.0), consistent with its role in VSMC migration and ECM remodeling.^[15] Genetic variants in ADAMTS7 are linked to increased CAD risk and poor cardiovascular outcomes. Dysregulation of TAGLN (SM22-α), a key VSMC contractile gene, was also noted, in line with murine models where TAGLN deletion caused mitochondrial dysfunction and worsened ischemic injury.^[37] Polymorphisms in inflammatory genes such as IL-1β −3953 C>T and MMP9 −1562 C>T may confer protection against myocardial infarction, suggesting genotype-dependent effects.^[33] These findings highlight immune and inflammatory pathways as central to CAD pathogenesis and support the rationale for targeted anti-inflammatory therapy. MMP9 and IL1β were among the most dysregulated genes in CAD, involved in ECM breakdown and immune modulation.^[38] Upregulation of IL1β and related genes in blood also points to systemic inflammation beyond vascular tissues.^[39]

This study highlights the pivotal role of ncRNAs, including miRNAs, lncRNAs, and circRNAs, in post-transcriptional gene regulation in CAD. Altered miR-146a expression was noted, consistent with its prognostic role in acute coronary syndromes. The rs2910164 polymorphism in miR-146a affects miRNA biogenesis and modulates inflammatory pathways, indicating a genetic basis for immune dysregulation in CAD.^[7] Identified ceRNA networks support previous evidence of disrupted post-transcriptional regulatory circuits in CAD.^[5] miR-182 showed a marked 6.12-fold upregulation, affirming its potential as a diagnostic biomarker,^[29] consistent with reports linking miRNA imbalance to endothelial dysfunction and atherosclerosis.^[6] LncRNAs also emerged as regulatory elements and biomarkers. Sex-specific lncRNAs, such as CH507-513H4.3 and CH507-513H4.6, suggest gender-based differences in CAD susceptibility.^[24] NEAT1_3 downregulation points to roles in oxidative stress and vascular stability.^[35] TGF-β1 polymorphisms have been linked to fibrosis, inflammation, and vascular remodeling in CAD. METTL16 variants further associate RNA methylation with sudden cardiac death risk in CAD.^[40,41]

Gene co-expression network analysis revealed dysregulation of transcriptional regulators, including GATA2, KLF4, and SP1, which influence endothelial integrity, inflammation, and smooth muscle cell phenotypes. GATA2 maintains endothelial barrier function and affects CAD susceptibility,^[42] while SP1 regulates inflammatory gene transcription.^[38] GRN analysis identified key genes such as PHACTR1 and APOC1. The PHACTR1 rs9349379G allele was linked to a 44% higher CAD risk, while the APOC1/APOE rs445925T variant was protective.^[17] Unsupervised clustering revealed CAD subtypes with distinct inflammatory and metabolic signatures.^[34,36] Liou et al. used PRSs to stratify CAD and explain 22% of variance.^[36] TRIM72 was implicated in fibrosis through TGF-β signaling,^[43] and KALRN rs9289231 was associated with early-onset CAD and cytoskeletal regulation.^[44] Mediterranean diet adherence downregulated oxidative stress genes, as supported by the PREDIMED trial.^[45] These findings support integrating genetic profiling and lifestyle interventions in CAD precision medicine. This study demonstrated significant upregulation of ADAMTS7 (logFC > 2.0), a gene involved in VSMC migration and ECM remodeling, consistent with studies linking its variants to higher CAD risk and poor cardiovascular outcomes.^[15] Dysregulation of SM22-α (TAGLN), a VSMC contractile marker, was also noted. Its deletion impairs mitochondrial function and worsens ischemic injury, highlighting its role in vascular homeostasis.^[37] CAD subtypes based on gene expression revealed divergent inflammatory and metabolic profiles, similar to previous clustering analyses.^[34] PRSs explained up to 22% of phenotypic variance, aiding personalized risk stratification.^[36] These findings underscore the importance of integrating genomic and lifestyle data for effective CAD management. MMP9 gene variants, key in ECM turnover, are linked to increased myocardial infarction risk and plaque instability, underscoring their role in vascular remodeling.^[46,47] Lipidomics profiling revealed lipid species tied to plaque features and clinical risk, supporting metabolomics in CAD subtyping.^[48] LDLR gene polymorphisms, particularly rs5925 and rs688, have been associated with increased susceptibility to CAD, with rs5925 also indicating potential pleiotropic effects in lipid metabolism pathways linked to both CAD and type 2 diabetes mellitus. These findings, along with evidence from other loci such as KALRN rs9289231 influencing vascular architecture, underscore the multifactorial genetic basis of early-onset CAD.^[49,50] PLA2G7 polymorphisms further implicate inflammatory and oxidative pathways in CAD.^[51] Multiple susceptibility loci also confirm the polygenic nature of CAD.^[52]

Several genes, such as MMP9, ADAMTS7, GATA2, and miR-146a, have emerged as promising biomarkers and therapeutic targets. The CANTOS trial validated IL-1β blockade in reducing cardiovascular events, reinforcing immune-targeted therapies.^[14] Functional polymorphisms in MMP9 (rs3918242), GATA2, and miR-146a (rs2910164) support the feasibility of genetic screening for personalized risk assessment.^[1,7,12] Our gene expression data align with earlier studies, where LRRC18 and SLC25A37 showed strong diagnostic accuracy (AUC 0.83),^[8] and NEAT1_3 emerged as a key biomarker of oxidative stress and immune modulation.^[35] Kontou et al. emphasized the need for rigorous validation pipelines to ensure clinical utility.^[13] Limitations of bulk transcriptomics may be addressed by single-cell RNA sequencing.^[24] Validation across diverse ethnic groups remains critical.^[7,22,33] Muiya et al.^[44] and Park et al. highlighted associations between CAD and GATA2, CUBN, HNF1A, and LIPC variants, linking immune and lipid metabolism pathways. Integrating gene expression with immunogenetic profiling lays a strong foundation for precision medicine. These findings collectively advance our understanding of CAD and support the translation of molecular insights into personalized care strategies.^[53]

CONCLUSION

This study offers a detailed multi-omics perspective on the molecular basis of CAD, highlighting the upregulation of MMP9, ADAMTS7, and IL-1β, and dysregulation of regulators such as GATA2 and miR-146a. These findings reinforce the central roles of inflammation, immune activation, and vascular remodeling in CAD. By validating results against 21 prior studies, this work confirms key pathogenic pathways while identifying novel gene regulatory axes. The study emphasizes the clinical potential of integrating gene expression profiles with risk stratification tools to guide precision therapeutics. It also supports the role of lifestyle and immunomodulatory strategies in modulating disease pathways. Future research should focus on single-cell transcriptomics, functional assays, and validation in diverse populations to improve clinical translation. These insights pave the way for more personalized, molecularly informed approaches to CAD prevention and treatment.