The Comparative Toxicogenomics Database (CTD) (23–26) is a valuable resource for enhancing our knowledge of the impact of environmental exposures on human health. An information source that offers details on the connections among molecules, genes, and proteins, along with the associations between chemicals, diseases, and genes. To identify genes closely associated with the disease, we obtained the genes related to periodontal disease from the CTD database. The Tumor Genome Atlas (TCGA) was launched in 2006 through a collaboration between the National Cancer Institute (NCI) and the National Human Genome Research Institute (NHGRI), with funding provided by the National Cancer Institute. Extensive collaboration was employed to enhance comprehension of the molecular foundation of cancer by utilizing advanced genome analysis technology on a large scale. To further examine the genes associated with this type of cancer, we obtained transcript data and clinical information pertaining to HNSC. This study included a cohort of 504 individuals diagnosed with HNSC and 44 healthy controls. Moreover, the ImmPort repository held a grand total of 3178 genes associated with the immune system.

In the first step, TCGA expression data for HNSC were downloaded. Using standardization, normalization, analysis, and processing, gene expression data were divided into groups representing HNSC and groups representing normal tissue. Furthermore, we acquired genes that have a strong correlation with periodontal disease and the immune system. Additionally, we classified genes as differentially expressed if their |log₂FC| was greater than 1 and their p-value was less than 0.05.

Using the ClusterProfiler package, we annotated key genes and explored their functions. The relationships between the genes were explored using Gene Ontology (GO) and the Kyoto Encyclopedia of Genes and Genomes (KEGG). Statistical significance was determined when the p values of GO and KEGG enrichment pathways were both less than 0.05.

The PPI network of interactive genes was generated using STRING (https://www.string-db.org/). Composite Ratings with values above 0.4 were deemed to be statistically significant. Furthermore, the PPI networks were examined and displayed using Cytoscape version 3.7.2.

In order to build a model that predicts prognosis, we utilized univariate and multivariate cox regressions, along with lasso regressions. For every patient, a risk score formula was developed by considering values from individual genes and assigning weights according to their estimated regression coefficients in the lasso regression analysis. A classification was made by using the risk score formula to divide the patients into two groups: one with low risk and another with high risk. The baseline was established using the median risk score value. The log-rank statistical approach was employed to compare the survival rates of the two groups using the Kaplan-Meier survival analysis. To assess the impact of risk score on the patient’s prognosis, a study employed Lasso regression and stratified analysis. ROC curves were utilized to assess the precision of model predictions. Additionally, the analysis of univariate and multivariate independent cox regression was employed to ascertain if IRPDGPI served as an autonomous prognostic factor in HNSC.

CIBERSORT, a widely utilized technique for assessing the cellular makeup through gene expression profiling, is frequently employed in the estimation and examination of immune cell infiltration. The CIBERSORT algorithm was utilized to analyze the RNA-seq data for determining the ratio of immune-infiltrating cells in both tumor and normal groups. Each sample had an identical score of 1 for all estimated immune cell types. Statistical significance was determined by performing a spearman correlation analysis between gene expression and immune responses in cell content, with a significance level of P<0.05.

Clinical data including age, sex, grade, T, N, and M stages were obtained from the TCGA repository. Additionally, the DCA curve and nomogram were constructed using R studio. The evaluation of the correlation between IRPDGPI and clinical value was conducted using the ‘RColorBrewer’ package in R.

Gene sets were obtained using MSigDB, available at http://www.gsea-msigdb.org/gsea/downloads.jsp. Enriched GO terms and KEGG pathways were identified by conducting GSEA on the gene sets. We chose the most significant 50 terms from each subtype.

The SCC-25 cell line, originating from human oral squamous carcinoma, was successfully re-established in our laboratory. These cells were sustained in a 1:1 blend of Dulbecco’s minimum essential medium and Ham’s F-12, enriched with 10% fetal bovine serum along with antibiotics, specifically 100 units/mL of penicillin and 100 μg/mL of streptomycin. The culturing conditions involved a humid environment with a controlled atmosphere of 5% CO2, held at a steady temperature of 37°C. Our laboratory successfully revitalized the NOK-SI cell line, comprising spontaneously immortalized human oral keratinocytes, sourced from normal tissue. These cells were nurtured in Dulbecco’s modified Eagle’s medium, enhanced with 10% fetal bovine serum (FBS) and 1% antibiotic/antimycotic combination. The culturing process was conducted at a constant temperature of 37°C within an atmosphere containing 5% CO2.

Total RNA was extracted from cells using the RNeasy Mini Kit following the manufacturer’s instructions. The concentration and purity of the RNA samples were determined by NanoDrop spectrophotometer. Subsequently, complementary DNA (cDNA) was synthesized from 1 µg of total RNA using the High Capacity cDNA Reverse Transcription Kit, according to the manufacturer’s guidelines. Real-time PCR analysis was performed on a QuantStudio 5 Real-Time PCR System using PowerUp SYBR Green Master Mix. Each 20 µl reaction mixture contained 10 µl of SYBR Green Master Mix, 1 µl of forward primer, 1 µl of reverse primer, 1 µl of cDNA, and 7 µl of nuclease-free water. The thermal cycling conditions were set as follows: initial denaturation at 95°C for 10 minutes, followed by 40 cycles of 95°C for 15 seconds and 60°C for 1 minute. Relative gene expression was calculated using the 2^−ΔΔCt method, with GAPDH used as the endogenous reference gene.

Survival curves were created and compared using the Kaplan-Meier technique, with log rank as the basis for comparison. The Cox proportional hazard model was utilized to perform the multivariate analysis. R (version 3.6) was utilized for conducting all statistical analyses. Obtaining a p-value of 0.05 in all statistical tests was deemed to be statistically significant.

The online tool database reported that periodontal disease had close associations with 3178 genes, whereas immunity had close associations with 10218 genes. Figure 1A displayed a Venn diagram indicating that 1325 genes had a strong correlation with both periodontal disease and immune function. The analysis included a total of 548 samples, with 44 being normal samples and 504 being tumor samples obtained from patients diagnosed with HNSC (as shown in Figure 1B ). There were 54492 genes whose expression level has been detected. Based on the findings, a total of 382 genes exhibited up-regulation in HNSC samples. In comparison to the normal sample, there was a decrease in the expression of 471 genes in the HNSC sample.

Various pathways showed significant enrichment of genes associated with immune response and periodontal disease, as indicated by the GO and KEGG enrichment analysis. For instance, the GO enrichment analysis indicated that several genes were linked to processes related to the muscular system, such as muscle contraction, sliding of muscle filaments, sarcomere, myofibril, contractile fiber, structural component of muscle, binding to actin, and binding to alpha-actinin ( Figure 1C ). Moreover, the KEGG enrichment analysis revealed the enrichment of numerous genes in pathways related to the contraction of cardiac muscle, hypertrophic cardiomyopathy, and dilated cardiomyopathy. Our research utilized the CTD and immune-related databases to detect interconnected genes associated with periodontal disease and immune response, forming a network of protein-protein interactions (PPI) among these genes. The findings indicated that several genes exhibited over 30 interactive counts, such as B2M, HSPA5, PSMB8, HLA-B, HLA-C, HLA-A, CANX, CALR, HLA-DRA, and CXCL10 ( Figure 1D ).

To further investigate the genes associated with periodontal disease and HNSC, the clinical data of patients with HNSC was acquired. Next, we performed a univariate Cox regression analysis and a lasso regression analysis to discover the distinctive genes in HNSC. Based on the data, univariate Cox regression ( Figure 2A ) identified 13 genes associated with prognosis (p value<0.01). In the subsequent stage, we executed the multivariate cox regression analysis ( Figures 2B, C ). Following the multivariate cox regression, we acquired the optimal risk score (Risk Score = CXCL5 * 0.00341351711657313 + ADM * 0.000848728116009434 + FGF9 * 0.339265398865143 + AIMP1 * 0.0168592756657467 + STC1 * 0.00269346117976538 + CDKN2A * -0.00662950138257665 + TRIB3 * 0.00830904795288386) values for further examination. Based on the median risk score, patients were categorized into groups with low and high risk ( Figure 3A ). Furthermore, the prognosis of HNSC patients (P<0.001) ( Figures 2D–J ) was found to be strongly associated with ADM, AIMP1, STC1, CDKN2A, and TRIB3 according to the Kaplan-Meier curves. Furthermore, the results from the ROC curve demonstrate that the AUC for 1, 2, and 3 years surpasses 0.6, signifying a commendable rate of success in validating the model ( Figure 3C ). In order to evaluate the independent prognostic factors in HNSC, we conducted both univariate and multivariate independent prognostic analyses. Through the utilization of both univariate and multivariate independent prognosis analysis, it was discovered that age, stage, and risk score played significant roles as independent determinants in HNSC (as depicted in Figures 3D, E ).

To investigate the gene expression level related to IRPDGPI, we conducted an analysis of ADM, CXCL5, ADM, FGF9, AIMP1, STC1, and CDKN2A expression in HNSC samples compared to normal samples. The findings indicated that there was a significant difference in the analysis of all 7 genes between the HNSC group and the normal group. ADM, AIMP1, CDKN2A, CXCL5, STC1, and TRIB3 exhibited increased expression in the HNSC group when compared to the normal group. In the HNSC group, the expression of FGF9 was decreased compared to the normal group as shown in Figures 4A–G .

To classify the expression data, we considered the low and high levels of gene expression associated with IRPDGPI.ADM showed a positive correlation with NK CD56 bright cells and Th2 cells, while exhibiting a negative correlation with B cells, mast cells, pDC, TFH, iDC, T cells, and NK cells. AIMP1 showed a positive correlation with NK CD56 bright cells and T cells. Moreover, there was a favorable association observed between CDKN2A and B cells, T cells, CD56 bright cells, NK CD56 dim cells, as well as pDCs. Besides neutrophils, macrophages, Tgd, DS, eosinophils, T2 cells, mast cells, and pDCs, CXCL5 exhibited a favorable association with neutrophil activity. Furthermore, it has been demonstrated that FGF9 exhibits a positive correlation with Th2, Treg, B, T, NK, NKCD56 dim, and NKCD56 bright cells. In addition, we found that STC1 has a strong association with NK CD56 bright cells, Th2 cells, Tgd and T helper cells, whereas TRIB3 is closely linked to Tgd, CD8 T cells, NK CD56 bright cells, Tcm, neutrophils, and TGCs (as shown in Figures 5A–G ).

By conducting clinical correlation analysis, we assessed the predictive significance of IRPDGPI and clinical value in determining the prognosis of patients with HNSC. Compared to the clinical data, which includes factors like age, sex, grade, T, M, and N stages, the ROC curve illustrated that IRPDGPI is a superior prognostic prediction model ( Figure 3B ). Furthermore, the DCA plot was utilized to assess the predictive significance of IRPDGPI ( Figure 5H ). To explore a better prognostic model for HNSC, an HNSC nomogram was constructed to predict survival rates ( Figure 5I ). Additionally, we investigated the association between IRPDGPI and clinical data in individuals diagnosed with HNSC. A significant correlation was observed between the stage and T stage and the IRPDGPI, with a P value of 0.05. There was no strong correlation between the IRPDGPI and factors such as age, gender, grade, M stage, and N stage. (P>0.05) ( Figures 6A–H ).

Based on the analysis of differential expression and the predictive model for prognosis, we have identified ADM as a potential crucial gene in the HNSC cohort. Hence, we conducted an analysis of GSEA and GSVA with respect to ADM. Specifically, the GSEA enrichment analysis revealed that ADM was highly enriched in various pathways including complement activation, regulation of complement activation, blood microparticle, RNA binding involved in posttranscriptional gene silencing, immunoglobulin complex, regulation of humoral immune response, B cell-mediated immunity, humoral immune response mediated by circulating immunoglobulin, and humoral immune response. Furthermore, an examination was conducted on the particular signaling pathways associated with the ADM, investigating the potential molecular mechanisms that impact the development and advancement of lung cancer ( Figure 7A ). According to the GSVA results, the distinct expression groups of ADM exhibited significant enrichment in various signaling pathways including adaptive immune response, apoptosis, biological adhesion, carbohydrate metabolism, cell cycle, cell population growth, cellular response to DNA damage, central nervous system development, and cytoskeleton organization ( Figure 7B ).

The expression levels of Adrenomedullin (ADM) in SCC-25 and NOK-SI cell lines were quantitatively evaluated using real-time PCR. Our results showed a distinct difference in ADM expression between these two cell lines. In the SCC-25 cell line, which is derived from oral squamous cell carcinoma, the ADM expression level was significantly elevated. Relative quantification using the 2^-ΔΔCt method demonstrated an increase in ADM mRNA expression in SCC-25 cells, in comparison to the NOK-SI cell line. Conversely, the NOK-SI cell line, comprising spontaneously immortalized human oral keratinocytes derived from normal tissue, showed relatively lower levels of ADM expression ( Figure 8 ).