Abstract
Background: Abdominal aortic aneurysm (AAA) represents a permanent and localized widening of the abdominal aorta, posing a potentially lethal risk of aortic rupture. Several recent studies have highlighted the role of pyroptosis, a pro-inflammatory programed cell death, as critical molecular regulators in AAA occurrence, progression, and rupture. However, the potential effects of pyroptosis in AAA and its upstream microRNA (miRNA) have not been comprehensively clarified. Methods: Through a search of the gene expression omnibus (GEO) database, the expression profiles of mRNAs (GSE7084, GSE57691, and GSE98278) and miRNAs (GSE62179) and corresponding clinical features were downloaded, respectively. Expression profiles of 15 AAA and 10 normal vascular samples were consecutively collected for in vitro experimentation and subsequent analysis. Various machine learning techniques were employed to identify hub pyroptosis-related genes (PRGs), leading to the development of a predictive model termed the PRG classifier. Quantitative real-time-polymerase chain reaction (qRT-PCR), western blot (WB), and enzyme-linked immunosorbent assay (ELISA) were used to confirm the expression of the hub PRGs. The diagnostic and predictive capabilities of the model were comprehensively evaluated in GEO and hospital cohorts. Then, the crucial immune cell infiltration and molecular pathways implicated in the initiation and rupture of AAA and their association with pyroptosis were explored. Lastly, a miRNA/hub pyroptosis-related molecular regulatory axis was constructed using the TargetScan dataset, which was further explored through loss-of-function assays. Results: Differential analysis, enrichment score analysis, and principal component analysis (PCA) revealed that pyroptosis-related molecules were significantly involved in the occurrence of AAA. Utilizing multiple machine learning algorithms, eight key PRGs (cysteinyl aspartate specific proteinase [CASP]1, infiltrating lymphocyte [IL]1B, IL18, IL6, NOD-, LRR- and pyrin domain-containing protein [NLRP]1, NLRP2, NLRP3, and tumor necrosis factor [TNF]) were integrated to establish a PRG classifier. Demonstrating robust diagnostic capabilities (area under curve [AUC] > 0.90), the PRG classifier provided clinical insights across two GEO datasets and effectively differentiated small AAA from large AAA, elective stable AAA (eAAA), and ruptured AAA (rAAA), respectively. qRT-PCR, WB, and ELISA verified the mRNA and protein expression of the hub PRGs. Notably, in hospital cohorts, a substantial positive link was unveiled between the PRG classifier and AAA risk factors (hypertension history, diastolic pressure, triglyceride levels, and aneurysm diameter). Furthermore, immune cell infiltration and functional enrichment analysis revealed significant associations of the PRG classifier/PRGs with M2 macrophage infiltration, activated dendritic cells, and enrichment scores of the cytosolic deoxyribonucleic acid (DNA) sensing pathway and tryptophan metabolism, potentially mediating AAA onset and rupture. Finally, based on 90 differentially expressed miRNAs (DEmiRNAs) and eight hub PRGs through TargetScan dataset, a hsa-miR-331-3p/TNF regulatory axis was constructed, wherein upregulation of hsa-miR-331-3p expression significantly reduced TNF and CASP1 protein levels. Conclusion: A predictive model (PRG classifier) incorporating eight PRGs through multiple machine learning algorithms was developed and validated. This model may stand as a potent tool for diagnosing AAA and assessing disease severity. The identification of the cytosolic DNA sensing pathway and the hsa-miR-331-3p/TNF interaction axis may represent crucial targets for AAA treatment, offering deeper insights into its potential pathogenesis.