Various datasets were collected from the GEO database (https://www.ncbi.nih.gov/geo/). GSE52093 includes ascending aorta gene expression data in patients with acute Stanford type A aortic dissection (n = 7) and healthy controls (n = 5); GSE153434 includes ascending aorta gene expression data in patients with Stanford type A aortic dissection (n = 10) and healthy controls (n = 10); GSE102397 includes vascular mRNA expression data from 3 to 4-month-old mice (n = 7) and 12-month-old mice (n = 6); GSE71226 includes patients with coronary heart disease (n = 3) and healthy people (n = 3); GSE9128 includes three patient groups:control subjects (n = 12) were divided into C1, C2, and C3; ischemic cardiomyopathy (ICM) patients (n = 12) were divided into ICM1, 2, 3, and 4; and nonischemic dilated cardiomyopathy (NIDCM) patients (n = 12) were divided into NIDCM1, 2, 3, and 4; and GSE12480 includes old mice (n = 10) and young mice (n = 10). GSE137869 is the single-cell RNA sequencing data of nine different rat tissue cells, and includes young (age 5 months) rats (n = 6, 3 male and 3 female rates) and old (age 27 months) rats (n = 6, 2 male and 4 female rates); the single-cell RNA sequencing data of the rat aorta, including the young group and the old group, were selected from this dataset. GSE9106 is a genome-wide gene expression profile of peripheral blood samples (collected from 59 patients with confirmed TAA and 34 healthy controls), including a training set (36 patients with TAA and 25 healthy controls) and a test set (23 patients with TAA and 9 healthy controls). Three of the datasets were missing grouping information and were removed from the subsequent analysis. Information on the mentioned datasets were shown in Table 1. Human aging genes were compiled from the official website of the Aging Atlas (12) (https://ngdc.cncb.ac.cn/aging/index).

GSE52093 was screened for DEGs using the data analysis tool GEO2R (https://www.ncbi.nlm.nih.gov/geo/geo2r/) in the GEO database. According to P < 0.05, |log₂FC| > 1, DEGs were screened, and 1,558 genes were obtained, including 794 up-regulated genes and 764 down-regulated genes. A total of 500 human aging-related genes were downloaded from the Aging Atlas (https://ngdc.cncb.ac.cn/aging/index). Finally, 70 DEGs related to aging were obtained from the intersection.

The gene expression differences between patients with acute Stanford type A aortic dissection and healthy controls in the GSE52093 dataset were analyzed with principal component analysis (PCA) in R software, and the results were visualized as a volcano plot, heatmap, and boxplots drawn with the “ggplot2” package.

GO and KEGG enrichment analysis of DEGs was performed using the “clusterProfiler” package in R software (13). GO analysis consisted of biological process (BP), cellular component (CC), and molecular function (MF). The up and downregulated genes of GSE52093 were subjected to GSEA and the results were visualized using the “ggplot2” package.

PPI analysis of aging-related DEGs was performed using the String (https://string-db.org/cgi/) database and Cytoscape software (version 3.8.2). Correlation analysis of aging-related DEGs was performed using Spearman correlation in the “ggplot2” package.

Five ranking methods of the cytoHubba plugin (Degree, Closeness, EPC, MNC, Radiality) in Cytoscape software were used to rank DEGs, then the top 10 DEGs were screened for overlapping genes and the top five overlapping genes were considered as hubgenes. NetworkAnalyst3.0 (https://www.networkanalyst.ca/) is a comprehensive network visualization analysis platform for gene expression analysis (14). This platform was used to analyze interactions between hubgenes. Hub gene–disease association information was shown using the DisGeNet database (version 7.0) (https://www.disgenet.org), which is only applicable to human data. Hub gene–drug interactions were shown using the DrugBank database (version 5.0) (http://go.drugbank.com), which is also only applicable to human data.

Using the GSE137869 single-cell sequencing dataset, four aortic samples—F-O (n = 4), F-Y (n = 3), M-O (n = 2), and M-Y (n = 3) (female old mice, female young mice, male old mice, and male young mice respectively)—were selected for single-cell sequencing analysis. Data processing and methods for clustering and identification of cell types are detailed in the original literature (15).

Gene expression was verified in the aging-related datasets GSE102397 and the TAAD-related dataset GSE153434. In addition, gene expression was verified in the coronary heart disease-related dataset GSE71226, the heart failure-related dataset GSE9128, and the aging cardiac tissue dataset GSE12480. Gene expression maps were drawn with the “ggplot2” package.

Using the dataset GSE9106 (testing set and training set), the ROC curve analysis and result visualization of the hubgenes were performed with the “pROC” package and “ggplot2”, respectively.

Differences between normally distributed variables were assessed using the t-test and differences between non-normally distributed variables were assessed using the Wilcoxon rank sum test. For all statistical tests, P < 0.05 was statistically significant.

The GSE52093 dataset is a genome-wide analysis of gene expression in the ascending aorta of patients with acute Stanford type A aortic dissection (n = 7) and normal controls (n = 5). To assess the reproducibility of the data within groups, PCA was performed. The data in GSE52093 was confirmed to have good reproducibility; PC1 of the GSE52093 dataset reflects 19.9% and PC2 reflects 11.7% of the variation in data characteristics (Figure 1A). Next, GEO2R was used to analyze the GSE52093 dataset according to P < 0.05, |log₂FC| > 1, which identified 1,558 genes, including 794 upregulated genes and 764 downregulated genes. A volcano map was drawn using the “ggplot2” package in R software (Figure 1B). A total of 500 human aging-related genes were downloaded from the Aging Atlas and took the intersection with TAAD-related genes was taken to obtain 70 aging-related DEGs, including 45 upregulated genes and 25 downregulated genes. A Venn diagram was produced using the “ggplot2” package (Figure 1C). A heatmap of the 70 aging-related DEGs in TAAD was generated using the “ggplot2” package (Figure 1D). In addition, boxplots were produced showing the expression of the 70 aging-related up- and downregulated DEGs in TAAD (Supplementary Figure S1).

The potential biological functions of the aging-related DEGs were analyzed by performing GO, KEGG, and GSEA enrichment analysis using R software. GO analysis showed that BPs of the aging-related DEGs were mainly focused on the response of cells to abiotic and environmental stimuli, as well as regulation of DNA metabolic processes; CCs involved the chromosomes, nuclear chromosomes, telomeric regions, transferase complexes, and transfer phosphorus-containing groups; and MF was focused on impaired DNA binding and hormone receptor binding (Figures 2A,C; Supplementary Table S1). KEGG pathway analysis showed that the aging-related DEGs were mainly enriched in the longevity regulating pathway, cellular senescence, tumor necrosis factor (TNF) signaling pathway, and hypoxia inducible factor-1 (HIF-1) signaling pathway (Figures 2B,D,E; Supplementary Table S2). GSEA of the upregulated genes in the GSE52093 dataset revealed that the aging-related genes, such as those involved in the p53 signaling pathway, were significantly enriched, while the enrichment of down-regulated genes was not significant (Supplementary Figure S2; Supplementary Table S3).

PPI network analysis indicated an interaction of senescence-related genes (Figure 3A); the number of interactions for each gene is shown in Figure 3B. Correlation analysis showed relationships between 45 upregulated and 25 downregulated aging-related DEGs (Figures 3C,D). Overlapping genes were screened with five sequencing methods of the cytoHubba plugin in Cytoscape software; the top five overlapping genes—MYC, IL6, HIF1A, ESR1, and PTGS2—were taken as hub genes (Table 2).

Hub gene–disease association information and hub gene–drug interactions were constructed using the GSE52093 dataset. Gene–disease association information were as follows (Figure 4A): HIF1A is associated with heart failure and cardiac hypertrophy; IL6 is associated with heart failure and cardiomyopathy; PTGS2 is associated with aortic aneurysms, heart failure, cardiac arrhythmias, cardiomyopathy, and CVD. Gene–drug interactions were as follows: PTGS2 may have interactions with 59 drugs, including aminosalicylic acid, acetaminophen, nabumetone, ketorolac, and tenoxicam (Figure 4B); HIF1A may have interactions with carvedilol, 2-methoxyestradiol, and N-[(1-chloro-4-hydroxyisoquinolin-3-YL) carbonyl] glycine (Figure 4C). This may provide new targets for the drug treatment of TAAD.

Using the single-cell sequencing dataset GSE137869, four aortic samples (F-O, F-Y, M-O, and M-Y) were selected for single-cell sequencing analysis. The cells in the aortic tissue were clustered in descending order, which yielded 18 cell clusters (Figure 5A). t-SNE descending (Figure 5B) and violin plots (Supplementary Figure S3) were generated based on the expression of known cellular markers in different morphological cells. The cell clusters were divided into seven cell groups (Figure 5C): SMC (Tagln, MYH11), fibroblasts (Fib) (Dcn, Lum), EC (Pecam1), B cells (BC) (CD19, CD79b), T cells (TC) (CD3e, CD3g, CD3d), M1-type macrophages (CD68, CD14, CD163), and M2-type macrophages (CD68, CD14, CD163). The distribution and expression of the hub genes were compared in the aorta of young and old rats (Figure 5D; Supplementary Tables S4, S5), and this showed that MYC expression was upregulated in SMC, Fib, EC, and TC (P < 0.001) and downregulated in M2-type macrophages (P < 0.01) in the aorta of aging rats; IL6 expression was upregulated in SMC, Fib (P < 0.001), and EC (P < 0.01) and downregulated in M1-type macrophages (P < 0.05); HIF1A expression was upregulated in M1-type macrophages (P < 0.05) and TC (P < 0.001), and downregulated in Fib (P < 0.001) and BC (P < 0.01); PTGS2 expression was up-regulated in Fib (P < 0.001) and downregulated in SMC, BC, M2 macrophages (P < 0.001), EC, TC (P < 0.01), and M1-type macrophages (P < 0.05); ESR1 expression was downregulated in Fib (P < 0.001).

Expression levels of the top five hubgenes screened by Cytoscape software were verified in the aging-related datasets GSE102397 and the results were visualized using “ggplot2”. HIF1A and PTGS2 expression was upregulated (P < 0.05) in GSE102397 (Figure 6A).

To verify the expression level of the hubgenes in TAAD-related datasets, the validation results were visualized using “ggplot2”. In GSE153434 (Figure 6B), the expression of MYC (P < 0.01) and HIF1A (P < 0.001) was upregulated, and ESR1 (P < 0.001) was downregulated in the case group. The expression levels of hub genes were also examined in the coronary heart disease-related dataset GSE71226, the heart failure-related dataset GSE9128, and the aging cardiac tissue-related dataset GSE12480, and the results are shown in Supplementary Figure S4.

Hub gene expression data in the GSE9106 dataset (testing set and training set) were selected and the “pROC” and “ggplot2” packages were used to draw a diagnostic ROC curve (Figures 6C,D). The area under the diagnostic ROC curve (AUC) showed that in the testing set the combined AUC of these five hubgenes was 0.812, while in the training set the combined AUC value was 0.731. This suggested that the combination of these five genes has a strong diagnostic power in the diagnosis of TAAD. A combined ROC with MYC and ESR1 that had individual AUC > 0.7 was then performed and the results were equal to the AUC value of the combined ROC of all five hub genes (Figure 6E). Thus, these two genes may play a more important role in the diagnosis of TAAD.