1. Introduction

The most common cell source in oral cancer is squamous cells.[1] Oral squamous cell carcinoma (OSCC) is an aggressive tumor with different degrees of differentiation, which tends to early and extensive lymph node metastasis. It has become an increasingly serious worldwide problem, mainly occurring in smokers and alcoholics aged 40 to 70.[2,3] OSCC refers to the malignant transformation of oral squamous epithelium and the formation of malignant tumor. It is a common and harmful malignant tumor in oral cavity, which needs to be paid attention to.[4] Symptoms of OSCC may include a localized mass, lesions of the white mucous membrane, and localized ulcers. It can cause the loss of oral function, such as long influence the language on the tongue, swallowing function, the function of the oral cavity of the affected, when further development may cause the stench or distant metastasis, and lead to local fester, affect the quality of life, in the end may also harm to the patient’s life, Once detected, OSCC must be treated early.[5,6] Early OSCC after treatment, the cure rate is high. If metastasis occurs, a certain degree of cure can be achieved, usually in combination with radiotherapy and chemotherapy.[7]

CDSN (Corneodesmosin) encodes keratindesmosin, an adhesion glycoprotein secreted by keratinocytes granulosa cells and a key protein for human epidermal barrier function. Combined with the existing desmosomes, it exists in the connective structure between cuticle - cuticle desmosomes, also plays a key role in the adhesion of cuticle, is a part of the desmosomes plaque modified at the top of the epidermis.[8] CDSN is a protein-coding gene located in the major histocompatibility complex Class I region of chromosome 6.[9] The CDSN gene encodes a protein found in intercuticles, which locate in the human epidermis and other keratinized squamous epithelium. The coding protein undergoes a series of cleavages during keratinocyte maturation. This gene is highly polymorphic in the population, with variations associated with skin diseases such as psoriasis, oligotrichosis, and skin peeling syndrome through pathways including keratosis and developmental biology.[10] CDSN -associated genebody annotations include protein homologous dimer activity. Related studies have shown that CDSN is associated with ankylosing spondylitis.[11] It has also been suggested that dermatitis is a widespread inflammatory form of dermatitis syndrome caused by autosomal recessive nonsense mutations of CDSN.[12] Thus, CDSN may be associated with various types of inflammation. However, the relationship between CDSN and OSCC remains unclear.

Bioinformatics studies biological problems using the methods of applied mathematics, informatics, statistics, and computer science. Bioinformatics research materials and results are a variety of biological data, its research tools are computers, research methods include biological data search (collection and screening), processing (editing, sorting, management, and display) and utilization (calculation, simulation). The research interests include sequence alignment, gene recognition, gene recombination, protein structure prediction, gene expression, protein response prediction, and development of evolutionary models.

We hypothesized that CDSN expression plays an important role in the development of OSCC. In this study, bioinformatics analysis was used to verify the potential role of CDSN in OSCC, and 200 OSCC patients were recruited to study the impact of CDSN expression mutation on the prognosis and survival time of OSCC patients, so as to provide new ideas for its molecular mechanism.

2. Methods 2.1. The description and verification of the dataset

At the April 20, 2022, the keywords were set as “oral squamous cell carcinoma,” “human,” “Expression profiling by array,” and “gene symbol,” and the datasites were screened. After reviewing the abstract and design of per dataset, GSE41613 was downloaded from the gene expression omnibus (GEO) database (https://www.ncbi.nlm.nih.gov/gds/?term=). In the GSE41613 dataset, there were 97 OSCC samples, including 41 low grade cases and 56 high grade OSCC tissues. Principal component analysis (PCA) is a statistical method. A set of potentially correlated variables is transformed into a set of linearly uncorrelated variables through orthogonal transformation, and the transformed set of variables is called principal components. In order to verify the dataset GSE41613, the PCA was performed.

2.2. Screening of differently expressed genes (DEGs) related with the OSCC

GEO2R (https://www.ncbi.nlm.nih.gov/geo/geo2r/) is a software for differential analysis of expression profile chips based on the GEO database. Using this tool we can compare 2 sets of samples (low grade OSCC and high grade OSCC) in the GSE41613 dataset and obtain DEGs. After the analysis by GEO2R, the DEGs were identified by the criteria of P value ≤.05 and Log2 (fold change) ≥1 or ≤ -1.

2.3. Protein-protein interaction (PPI) analysis and the identification of hub genes

The Search Tool for the Retrieval of Interacting Genes (STRING) database (https://string-db.org/) is an online database for searching known protein interactions. Currently, a total of 2031 species and 9643763 protein interactions are stored. The above DEGs related with the OSCC were added into the STRING database, and the PPI was constructed.

Cytoscape (version 3.5.1) is a software focused on open source network visualization and analysis. Its core is to provide basic functional layout and query network, and to combine basic data into a visual network. Cytoscape is originated from systems biology and is used to integrate biomolecular interaction networks with high-throughput gene expression data and other molecular state information. Its most powerful function is for large-scale PPIs, and genetic interactions. And the software was used for visualization for the PPI network. Furthermore, the cytoHubba, one plug-in of the Cytoscape, was implied for the identification of the hub gens related with the OSCC. In order to explore the most significant hub genes, the EcCentricity, EPC, and MNC were used, and the VENN diagram was performed. After that, the BottleNeck was used for the verification of the most significant hub genes.

2.4. Functional annotation of the DEGs related with the OSCC by DAVID

Database for Annotation, Visualization and Integrated Discovery (DAVID, (https://david.ncifcrf.gov) is not only a biological database, but also an online analysis software. It can be used for differential gene analysis and pathway enrichment. DAVID could associate genes in the input list with biological annotations. The tool implements functional annotation of genes, presenting in tabular form the annotation of each gene in the input list in the selected database. Enrichment analysis of the DEGs related with the OSCC was made by the DAVID (version 6.7).

2.5. Analysis of the metascape

Metascape integrates multiple authoritative data resources such as genebody, Kyoto Encyclopedia of Genes and Genomes, UniProt and DrugBank, so that it cannot only complete pathway enrichment and biological process (BP) annotation, but also perform gene-related protein network analysis and related drug analysis. In terms of data processing scale, Metascape can process lists containing thousands of genes at one time, and supports simultaneous upload of multiple different gene lists. Therefore, the DEGs related with the OSCC were input in the Metascape to explore the enrichment terms.

2.6. The analysis for the relationship between OSCC and hub genes

The comparative toxicogenomics database (CTD) The CTD database integrates a large number of interaction data between chemical substances, genes, functional phenotypes and diseases, providing great convenience for the study of disease-related environmental exposure factors and the potential mechanism of action of drugs. The inference score and references of CDSN for the OSCC were calculated by the CTD. In addition, the expression level of hub genes was analyzed by the heatmap. And the relationships among the hub genes were presented. The overall survival analysis of CDSN for the OSCC was made. Finally, the role of hub genes on the pathological stage of OSCC was explored.

2.7. RT-PCR assay

Primer Premier 3.0 was used to design specific primers for CDSN. Primers and probes were synthesized by Fuzhou Boshang Biotechnology Co., Ltd., diluted with RNase-free water to 10 μmol/L, and stored at -20°C for later use. Total RNA was extracted from tissue samples according to the instructions of the RNA extraction kit (Hunan Aikerui Bioengineering Co., Ltd.), and the concentration and purity of RNA were measured with a nucleic acid protein analyzer, and stored at -80 °C for later use. Perform conventional RT-PCR amplification according to the instructions of the one-step RT-PCR kit. The reaction system: One Step Enzyme Mix 1 μL, 2×One-Step Reaction Solution 12.5 μL, upstream and downstream primers CBPV-F/R (10 μmol/L)) 1 μL each, template RNA 2 μL, supplemented with RNase-free water to 25 μL. The reaction conditions were: reverse transcription at 50°C for 30 minute; pre-denaturation at 95°C for 3 minute; denaturation at 95°C for 30 second, annealing at 56°C for 30 second, extension at 72°C for 30 second, 35 cycles; extension at 72°C for 5 minute. The relative expression levels of genes were calculated based on CT values. Take GAPGH was set as a reference.

Primers of CDSN and sequences for PCR analysis were followed:

CDSN-F: GCCGATGACTGAGATAAGG

CDSN-R: GCTGCTGCTGAACTGAAA.

2.8. Immunofluorescence assay

Add 0.01 mol/L PBS with pH 7.4 dropwise to the specimen to be tested, and discard it after 10 min to keep the specimen at a certain humidity. Add appropriately diluted fluorescently labeled antibody solution dropwise to completely cover the specimen, place it in an enamel box with a lid, and keep it for a certain period of time (30 minute). Take out the slides, put them on the slide rack, rinse them with 0.01 mol/L PBS, pH 7.4, and then soak them in 3 cylinders of 0.01 mol/L, pH 7.4 PBS in sequence, 3 to 5 minute in each cylinder, Oscillate from time to time. Remove the slide, absorb excess water with filter paper, but not dry the specimen, add a drop of buffered glycerol, and cover with a coverslip. Immediately observe with a fluorescence microscope.

2.9. Patients and ethics

A total of 200 patients diagnosed with OSCC at Ethics Committee of the Fourth Hospital of Hebei Medical University from March 2015 to June 2020 were selected.

Determine the sample size:

n=Z2d2σ2

Z is the confidence interval, n is the sample size, d is the sampling error range, and σ is the standard deviation, which is generally 0.5.

Inclusion criteria: 18 to 80 years old, diagnosed with OSCC; Cardiopulmonary function was normal; Normal coagulation.

Exclusion criteria: age < 18 or age > 80 requiring emergency surgery; The patients and their families did not agree to participate in the trial.

The study was approved by Ethics Committee of the Fourth Hospital of Hebei Medical University, and all written informed consent was obtained from the patients.

2.10. The parameter

According to clinical data, patients were classified by gender (male/female), age (≤65/>65), tumor size (≤1 cm/>1 cm), family history of OSCC (no/yes), tumor grade (low/high), CDSN (low/high), tumor stage (early/late), and prognosis (good/poor). Patients’ survival time was recorded during follow-up. The tumor grade was classified as grade I, II, III, and IV, with grade I and II as low, grade III and IV as high. The level of CDSN expression was defined according to the mean CDSN expression level of the included patients. Lower than the average level is low expression, vice versa is high expression. The general stage of the tumor is stage I, II, III, and IV, with stage I and II as low, stage III and IV as high. The prognostic classification was based on the recurrence rate and 5-year survival rate.

2.11. Statistical method

The data is expressed as a percentage of the total. Pearson chi-square test and Spearman correlation coefficient were used to analyze the clinical parameters and prognosis of OSCC. Univariate and multivariate Logistic regression analysis were used to calculate the OR values of prognostic variables in OSCC patients. Forward conditional selection was used to build the multivariable model. Any test results reaching a liberal statistical threshold of P < .2 for each comparison were then entered multivariable regression model to identify independently predictive factors for OSCC. The risk factors were forced to enter into the multivariate regression model to identify independently predictive factors for OSCC. Then, the overall survival time of OSCC was converted into natural logarithmic equivalent value for statistical analysis. Cox proportional risk regression analysis was used to explore the correlation between survival time and related factors in OSCC patients. The receiver operating characteristic curves were obtained using MedCalc software. Receiver operating characteristic analysis was made to explore the sensitivity and specificity of the expression level of CDSN on the overall survival time of OSCC. And the Kaplan-Meier analysis was performed to explored the effect of CDSN on the overall survival. BP neural network is not only a universal model, but also an error correction function. According to the results obtained by training and the expected results, error analysis is carried out each time, then the weight and threshold value is modified, and the model whose output is consistent with the expected results is obtained step by step. The model finally trained by the support vector machine (SVM) algorithm is determined by some support vectors, which can determine the vector of the final model. Therefore, the BP-neural network and the SVM were constructed by the MATLAB (version: 2017a), and in the model, the expression of CDSN was set as the input value, and the survival time of OSCC was set as the output value. Statistical analyses were performed using SPSS software, version 21.0 (IBM, Armonk, NY). P < .05 was considered statistically significant.

3. Results: 3.1. Verification of the GSE41613 and DEGs related with the OSCC

After analysis via the PCA, the samples in the GSE41613 were classified in the PC1 and the value is 68.1% (Fig. 1A). Volcano plot manifested that there were plenty of DEGs between the low grade OSCC and high grade OSCC, and the green presented the down-regulated DEGs and the red showed the up-regulated DEGs in the high grade OSCC (Fig. 1B). The PPI network was successfully constructed and there was a maze of connections among the DEGs (Fig. 1C). After calculated by the cytoHubba, 9 hub genes were identified, which including SPRR2G, SPRR2B, SPRR1A, LCE3D, LOR, SPRR3, LCE2B, PI3, and CDSN (Fig. 1D).

Figure 1.:

Verification of the GSE41613 and DEGs related with the OSCC. (A) Principal component analysis. (B) Volcano plot manifested that there were plenty of DEGs between the low grade OSCC and high grade OSCC. The green presented the down-regulated DEGs and the red showed the up-regulated DEGs in the high grade OSCC (C) The PPI network of the DEGs. (D) Hub genes by the cytoHubba. DEGs = differently expressed genes, OSCC = oral squamous cell carcinoma, PPI = protein-protein interaction.

3.2. CDSN as the most significant hub gene of the OSCC

After calculated by the EcCentricity, 10 hub genes were identified, which including ENO3, PYGM, MYBPC2, CALML5, TCHH, LOR, KRT1, KRTDAP, CDSN, KLK1 (Fig. 2A). Through the EPC analysis, the LCE2B, SPRR2B, LOR, KRTDAP, CRCT1, LCE3D, SPRR1A, SPRR2G, SPRR3, CDSN were identified (Fig. 2B). By the MNC analysis, SPRR3, SPRR2G, SPRR2B, SPRR1A, PI3, LOR, LCE3D, LCE2B, CDSN were screened (Fig. 2C). Through the VENN diagram, the CDSN was identified as the most significant hub gene of the OSCC (Fig. 2D). Finally, after the verification by the BottleNeck, the CDSN was also the most significant hub genes of the OSCC (Fig. 2E).

Figure 2.:

CDSN as the most significant hub gene of the OSCC. (A) Hub genes by the EcCentricity. (B) Hub genes by EPC analysis. (C) Hub genes by MNC analysis. (D) Through the VENN diagram, the CDSN was identified as the most significant hub gene of the OSCC. (E) After the verification by the BottleNeck, the CDSN was also the most significant hub genes of the OSCC. CDSN = corneodesmosin, MNC = major histocompatibility complex, OSCC = oral squamous cell carcinoma.

3.3. Enrichment analysis via the DAVID

In the aspect of BP, the DEGs of the OSCC mainly enriched in the cornification, epidermis development, keratinocyte differentiation, sarcomere organization, cell adhesion, extracellular matrix organization, epithelial cell differentiation (Fig. 3A). In the aspect of cell component, the DEGs of the OSCC mainly enriched in cornified envelope, extracellular region, extracellular space, myofibril, secretory granule, sarcomere, myosin filament, cytosol, sarcoplasmic reticulum, muscle thin filament tropomyosin, platelet alpha granule lumen, stress fiber, extracellular exosome (Fig. 3B). In the aspect of molecular function, the DEGs of the OSCC mainly enriched in the actin binding, actin filament binding, structural constituent of muscle, structural constituent of epidermis, extracellular matrix binding, serine-type endopeptidase activity, NADP-retinol dehydrogenase activity, serine-type endopeptidase inhibitor activity, extracellular matrix structural constituent (Fig. 3C). In the aspect of Kyoto encyclopedia of genes and genomes, the DEGs of the OSCC mainly enriched in the apelin signaling pathway, cell adhesion molecules (Fig. 3D).

Figure 3.:

Enrichment analysis via the DAVID, bar graph of enriched terms across input gene lists and network of enriched terms by the Metascape. (A) Biological process (BP). (B) Cell component (CC). (C) Molecular function (MF). (D) KEGG. (E) Bar graph of enriched terms across input gene lists, colored by P values. (F) Network of enriched terms: colored by cluster ID, where nodes that share the same cluster ID are typically close to each other. (G) Network of enriched terms: colored by P value, where terms containing more genes tend to have a more significant P value. DAVID = database for annotation, visualization and integrated discovery, KEGG = Kyoto encyclopedia of genes and genomes.

3.4. The analysis by the metascape

Through the metascape, the DEGs were mainly enriched in the formation of the cornified envelope, positive regulation of epithelial cell differentiation, TGF-beta receptor signaling, positive regulation of cation channel activity, cell-cell junction organization, retina homeostasis, regulation of cell adhesion, sensory organ morphogenesis (P < .05) (Fig. 3E–G). The significant module in the DEGs related with the OSCC was identified by the Metascape, and the CDSN was also included in the significant module, which verified CDSN as the most significant hub genes (Fig. 4A). Summary of enrichment analysis in COVID showed that the DEGs were mainly enriched in the vanderheiden, lamers intestinal-organoid expansion, proteome stukalov (Fig. 4B). Summary of enrichment analysis in cell type signatures showed that the DEGs were mainly enriched in “BUSSLINGER ESOPHAGEAL LATE SUPRABASAL CELLS,” “DESCARTES MAIN FETAL SQUAMOUS EPITHELIAL CELLS” (Fig. 4C). Summary of enrichment analysis in DisGeNET showed that the DEGs were mainly enriched in recurrent tumor (Fig. 4D). Summary of enrichment analysis in PaGenBase showed that the DEGs were mainly enriched in tongue, tonsil (Fig. 5A). Summary of enrichment analysis in Transcription Factor Targets was presented in the Figure 5B and Supplemental Digital Content (Table S1, https://links.lww.com/MD/H438-S6, https://links.lww.com/MD/H438).

Figure 4.:

The analysis by the Metascape. (A) Protein-protein interaction network and MCODE components identified in the gene lists. (B) Summary of enrichment analysis in COVID. (C) Summary of enrichment analysis in cell type signatures. (D) Summary of enrichment analysis in DisGeNET.

Figure 5.:

Summary of enrichment analysis in PaGenBase and Transcription Factor Targets. (A) PaGenBase. (B) Transcription factor targets.

3.5. The role of CDSN on the OSCC

By the CTD analysis, there was strong relationship between the CDSN and mouth neoplasms, head and neck neoplasms, squamous cell carcinoma of head and neck (Fig. 6A). Compared with low grade OSCC, the CDSN was down-regulated in the high grade OSCC (Fig. 6B). The heatmap manifested that there were positive correlations among the hub genes (P < .05) (Fig. 6C). The OSCC patients with low expression level of CDSN have poor overall survival compared with the high expression level of CDSN (HR = 0.75, 95%CI: 0.57-0.98, P = .036) (Fig. 6D). The lower the expression of CDSN was, the higher the clinicopathological stage of OSCC was (P = .0238, F value = 3.18) (Fig. 6E).

Figure 6.:

The role of CDSN on the OSCC. (A) By the CTD analysis, there was strong relationship between the CDSN and mouth neoplasms. (B) Compared with low grade OSCC, the CDSN was down-regulated in the high grade OSCC. (C) The heatmap manifested that there were positive correlations among the hub genes (P < .05). (D) The OSCC patients with low expression level of CDSN have poor overall survival compared with the high expression level of CDSN (HR = 0.75, 95%CI: 0.57-0.98, P = .036). (E) The lower the expression of CDSN was, the higher the clinicopathological stage of OSCC was (P = .0238, F value = 3.18). CDSN = corneodesmosin, OSCC = oral squamous cell carcinoma.

3.6. Verification of the effect of CDSN on the OSCC by the clinical samples

The expression of CDSN in the low grade OSCC was higher than the high grade OSCC via the RT-PCR (P < .05, Fig. 7A). The OSCC patients with low expression level of CDSN have poor overall survival compared with the high expression level of CDSN (HR = 0.096, P < .001) (Fig. 7B). Through the immunofluorescence assay, compared with the high grade OSCC, the expression of CDSN was up-regulated in the low grade OSCC (P < .05) (Fig. 7C). The expression of CDSN might be molecular target for the diagnosis of the OSCC, and the sensitivity and specificity were high (AUC = 0.764, P < .05) (Fig. 7D). The joint effect of all factors (CDSN, age, family history, sex, tumor grade, tumor size, tumor stage) on the diagnosing the OSCC was strong (AUC = 0.814, P < .05) (Fig. 7E). The respective role of factors on the diagnosis of OSCC was presented in the Figure 7F.

Figure 7.:

Verification of the effect of CDSN on the OSCC by the clinical samples. (A) The expression of CDSN in the low grade OSCC was higher than the high grade OSCC via the RT-PCR. (B) The OSCC patients with low expression level of CDSN have poor overall survival compared with the high expression level of CDSN (HR = 0.096, P < .001). (C) Through the immunofluorescence assay, compared with the high grade OSCC, the expression of CDSN was up-regulated in the low grade OSCC (P < .05). (D) The expression of CDSN might be molecular target for the diagnosis of the OSCC, and the sensitivity and specificity were high (AUC = 0.764, P < .05). (E) The joint effect of all factors (CDSN, age, family history, sex, tumor grade, tumor size, tumor stage) on the diagnosing the OSCC was strong (AUC = 0.814, P < .05). (F) The respective role of factors on the diagnosis of OSCC. CDSN = corneodesmosin, OSCC = oral squamous cell carcinoma.

3.7. Strong predictive value of CDSN for the OSCC survival time

Through the BP-neural network, best training performance is 0.0026893 at epoch 3000, and the training relevance (R) is 0.99413 between CDSN and the OSCC survival time. Furthermore, the percentage errors of the BP-neural network were small (Fig. 8A). In addition, the strong predictive value of CDSN for the OSCC survival time was also verified by the SVM model (y = 0.6041× + 20.5610, R = 0.8559) (Fig. 8B).

Figure 8.:

Strong predictive value of CDSN for the OSCC survival time. (A) BP-neural network, best training performance is 0.0026893 at epoch 3000, and the training relevance (R) is 0.99413 between CDSN and the OSCC survival time, the percentage errors of the BP-neural network were small (B) The strong predictive value of CDSN for the OSCC survival time was also verified by the SVM model (y = 0.6041× + 20.5610, R = 0.8559). CDSN = corneodesmosin, OSCC = oral squamous cell carcinoma.

3.8. Pearson’s Chi-square test was used to analyze the correlation between CDSN expression and related factors of OSCC

Pearson’s Chi-square test was used to summarize the relationship between CDSN expression level and related factors of OSCC. Prognosis (P < .001) was significantly correlated with CDSN expression level. However, Sex (P = .571), Age (P = .121), Tumor size (P = .261), Family History (P = .325), Tumor Grade (P = .199), and Tumor Stage (P = .524) were not significantly correlated with CDSN expression (Table 1).

Table 1 - Relevant characteristics of patients with oral squamous cell carcinoma. Characteristics CDSN P Low High Sex Male 100 51 (25.5%) 49 (24.5%) .571 Female 100 55 (27.5%) 45 (22.5%) Age ≤65 78 36 (18.0%) 42 (21.0%) .121 >65 122 70 (35.0%) 52 (26.0%) Tumor size ≤ 1cm 99 51 (25.5%) 48 (24.0%) .261 >1cm 101 55 (27.5%) 46 (23.0%) Family history No 80 39 (19.5%) 41 (20.5%) .325 Yes 120 67 (33.5%) 53 (26.5%) Tumor grade Low 101 49 (24.5%) 52 (26.0%) .199 High 99 57 (28.5%) 42 (21.0%) Tumor stage Early 108 55 (27.5%) 53 (26.5%) .524 Late 92 51(25.5%) 41 (20.5%) Prognosis Good 93 23 (11.5%) 70 (35.0%) <.001* Poor 107 83 (41.5%) 24 (12.0%)

CDSN = corneodesmosin.

Pearson chi-square test

*P < .05.

3.9. Spearman correlation coefficient was used to analyze the correlation between CDSN expression level and related factors of OSCC

Further analysis of Spearman correlation coefficient showed that CDSN expression level was significantly correlated with prognosis (ρ = -0.528, P < .001). However, Gender (rho = 0.040, P = .573), Age (rho = 0.110, P = .122) and Tumor Size (rho = 0.029, P = .679), and Family History (rho = 0.070, P = .328), Tumor Grade (ρ = -0.091, P = .201), Tumor Stage (ρ = -0.045, P = .527) had no significant correlation with CDSN expression (Table 2).

Table 2 - Relationship between characteristics of oral squamous cell carcinoma and CDSN expression level. Characteristics CDSN ρ P Sex -0.040 .573 Age -0.110 .122 Tumor size -0.029 .679 Family history -0.070 .328 Tumor grade -0.091 .201 Tumor stage -0.045 .527 Prognosis -0.528 <.001*

CDSN = corneodesmosin.

Spearman correlation analysis

*P < .05

3.10. Univariate logistic regression analysis of prognosis and related factors of OSCC

Binary Logistic regression was used to determine the relationship between OSCC related parameters and CDSN expression level, odds ratio (OR), and 95% confidence interval (95% CI). Table 3 describes the OR and 95%CI of the subjects at the univariate Logistic regression level, and the results show that the prognosis of subjects with low CDSN expression level is significantly worse than that of subjects with high CDSN expression level (OR = 0.095, 95%CI: 0.049-0.183, P < .001). However, Gender (OR = 0.852,95%CI: 0.488-1.485, P = .571), Age (OR = 0.637,95%CI: 0.359-1.128, P = .122), Tumor Size (OR = 0.889,95%CI: 0.510-1.549, P = .677), Family History (OR = 0.752,95%CI: 0.427-1.327, P = .326), Tumor Grade (OR = 0.694,95%CI: 0.397-1.213, P = .200), Tumor Stage (OR = 0.834,95%CI: 0.477-1.458, P = .524) had no significant correlation with CDSN expression level (Table 3).

Table 3 - Association of parameters associated with oral squamous cell carcinoma and CDSN expression level based on univariate Logistic regression analysis. Characteristics CDSN OR 95% CI P Sex Male 100 1 .571 Female 100 0.852 0.488-1.485 Age ≤65 78 1 .122 >65 122 0.637 0.359-1.128 Tumor size ≤ 1cm 99 1 .677 >1cm 101 0.889 0.510-1.549 Family history No 80 1 .326 Yes 120 0.752 0.427-1.327 Tumor grade Low 101 1 .200 High 99 0.694 0.397-1.213 Tumor stage Early 108 1 .524 Late 92 0.834 0.477-1.458 Prognosis Good 93 1 <.001