Identification of immunoheterogeneous key prognostic protein ZC3H12D in OSCC and analysis of prognostic models and drug resistance

Abstract

Background

We constructed a nomogram model utilizing key clinical and molecular biological variables to predict the 1-year, 3-year, and 5-year survival probabilities for patients with oral squamous cell carcinoma (OSCC). The model’s predictive efficacy was assessed through calibration curves. Additionally, a detailed Spearman correlation analysis of the ZC3H12D gene was executed, with patient data stratified into high and low expression categories based on its expression levels for differential gene analysis. Following this, Gene Ontology (GO) and Kyoto Encyclopedia of Genes and Genomes (KEGG) enrichment analyses, as well as Gene Set Enrichment Analysis (GSEA), were performed on the identified differentially expressed genes to delineate their specific contributions to biological processes and metabolic pathways.

Results

The nomogram model demonstrated a high degree of predictive accuracy, effectively predicting the survival probabilities of oral cancer. patientsThe correlation analysis of the ZC3H12D gene revealed significant correlations with various immune cell types. A total of 439 significantly differentially expressed genes were identified through differential gene analysis. GO and KEGG enrichment analyses, along with GSEA analysis, further elucidated the specific roles of these differentially expressed genes in biological processes and metabolic pathways. Moreover, immune infiltration analysis indicated significant differences between the high and low expression groups of ZC3H12D in terms of immune cell infiltration, immune scores, and ESTIMATE scores.

Conclusion

The nomogram model constructed in this study provides a practical tool for prognostic assessment in OSCC patients. Additionally, the in-depth analysis of the ZC3H12D gene reveals its significant role in the occurrence and development of oral cancer, providing data support and a theoretical basis for further research.

Introduction

Oral squamous cell carcinoma (OSCC) represents the most prevalent malignant tumor type within the oral cavity, comprising approximately 90% of all oral tumor instances reported [1–3]. Despite the availability of various treatment modalities, including but not limited to adjuvant radiotherapy, targeted therapy, and biological therapy, the lack of highly specific early diagnostic biomarkers has resulted in unsatisfactory treatment outcomes and prognosis for patients with advanced oral squamous cell carcinoma. Therefore, in-depth research and exploration of molecular markers for the early diagnosis and prognostic assessment of OSCC, as well as a deeper understanding of the molecular mechanisms of OSCC, hold significant clinical importance and value for developing more effective treatment strategies [4, 5].

Oral cancer (OC), a prevalent form of head and neck malignancy, is seeing ongoing advancements in its treatment protocols. Immunotherapy, an up-and-coming therapeutic approach, is increasingly drawing attention as a focal point of research. This method offers fresh avenues for oral cancer treatment by stimulating or intensifying the body’s immune response to detect and combat cancerous cells [6]. In oral cancer, cancer cells often evade surveillance by upregulating certain immune checkpoint molecules, such as epidermal growth factor receptor (EGFR) and programmed cell death protein ligand-1 (PD-L1) [7, 8]. The high expression of these molecules is closely related to immune evasion, providing clear targets for immunotherapy. Currently, immunotherapy drugs such as monoclonal antibodies against EGFR (such as cetuximab and panitumumab) and immune checkpoint inhibitors against the PD-1/PD-L1 pathway (such as pembrolizumab, nivolumab, and atezolizumab) have shown positive therapeutic effects in clinical trials. However, immunotherapy is not a panacea, and it has certain limitations. First, the response rate to immunotherapy is not high, and only some patients benefit from it [9, 10]. Second, immunotherapy may cause immune-related adverse events, such as immune-mediated inflammatory responses. In addition, the heterogeneity of oral cancer means that not all patients are suitable for the same immunotherapy regimen. Therefore, the precise selection of patients with specific immune characteristics and the development of more personalized treatment strategies are urgent problems to be solved in the field of immunotherapy [11].

With the continuous progress and wide application of computational biology [12, 13], it plays an increasingly important role in screening drug targets. The goal of this research was to apply sophisticated bioinformatics techniques, including weighted gene co-expression network analysis (WGCNA), to pinpoint crucial genes that influence immune diversity in oral squamous cell carcinoma (OSCC).Through thorough examination of the signaling pathways related to these crucial genes, our aim is to reveal the molecular mechanisms that contribute to the immune heterogeneity of OSCC.The ultimate goal of this study is to provide new ideas and strategies to improve the therapeutic effect of OSCC, especially to increase the response rate of immunotherapy, so as to bring more effective treatment for OSCC patients.

Method

Data process of OSCC

In order to perform an in-depth bioinformatics analysis, we first need to download the tpm dataset for head and neck squamous cell carcinoma (HNSC) from the xena website (https://xenabrowser.net/). After the download, we used clinical data to carefully screen and organize these datasets, and finally obtained a set of high-quality oral cancer (OSCC) mRNA sequencing data, including a total of 314 OSCC samples and 44 adjacent tissues. The sample was included in the study, and these data will serve as the basis for our follow-up studies. In addition, to enrich our data sources and make comparative analysis, we have downloaded GSE30784 [14] datasets from the GEO database.

RNA was extracted from 167 OSCC, 17 dysplasia, and 45 normal oral tissues and hybridized to Affymetrix U133 2.0 Plus GeneChip arrays. The GEO database, a public repository for high-throughput gene expression, microarray data, and sequence data, offers a vast array of biomedical research information, facilitating more comprehensive and in-depth analyses.

The Immunoinfiltration and differential immune cells in OSCC

For this research, we employed the GSE30784 dataset, the CIBERSORT algorithm [15], and the immune cell LM22 gene set to compute and assess the relative abundance of 22 distinct lymphocyte subtypes within various tumor and normal samples. During the analysis, we excluded samples that were not statistically significant (p value greater than 0.05). We then mapped the proportion of immune cells in these samples to present the data more visually. To further investigate the disparities between tumor samples and normal samples with regard to 22 specific types of immune cells, a Wilcoxon test was employed for comparative analysis at a professional level. In the results, we use specific symbols to indicate the level of statistical significance, where p-values less than 0.001 are indicated by “***”, p-values between 0.001 and 0.01 are indicated by “*”.

WGCNA analysis for the differential immune related genes

In the study, we selected the relative abundance of eight significantly different immune cells as biological phenotypic data based on the proportion of immune cell differences associated with OSCC (oral squamous cell carcinoma). These data were used to perform a weighted gene co-expression network analysis [16] (WGCNA). In the WGCNA analysis, we set a soft threshold of 7 and performed precise identification of different modules using dynamic tree chopping. In order to ensure that each module contains sufficient gene information, we specify a minimum number of genes per module of 50. Next, we adjusted the mergeCutHeight parameter to 0.25 in order to merge modules that are similar in features. Through these steps, we further analyzed the correlation between the final module and differential immune cells.

After pooling and screening the modules, we successfully developed 12 different modules and evaluated the correlation between these modules and differential immune cells in order to screen for the most significant correlations. Among these modules, we pay particular attention to the turquoise module, which is most relevant to differential immune cells. We identified key module genes by setting screening criteria for key module genes (module membership MM greater than 0.5, gene significance GS greater than 0.3, and p-value less than 0.05). Subsequently, we extracted the intersections of key module genes such as B.cells.memory, macrophages.M0, and Mast.cells.activated cells as gene collections associated with differential immune cells for further screening and analysis.

Screening of immune-related genes in OSCC and construction of prognostic models

For this research, we employed the robust bioinformatics tool DESeq2 [17, 18] to perform a thorough examination of the OSCC (Oral Squamous Cell Carcinoma) count data retrieved from the TCGA (The Cancer Genome Atlas) database. By comparing tumor samples with normal samples, we successfully identified differentially expressed genes (DEGs) that met specific selection criteria, Specifically, we utilized professional criteria, namely an absolute log fold change (|logFC|) greater than 0 and a statistically significant p-value of less than 0.05. Among the comprehensively identified list of Differentially Expressed Genes (DEGs), we further narrowed our focus to genes specifically associated with immune cells and pinpointed 135 genes that demonstrated differential expression between tumor and normal samples. In order to attain a profounder understanding of the functions and roles of these genes, we employed Venn diagram analysis for a detailed exploration. These identified immune-related genes are considered key genotypes influencing the progression of oral squamous cell carcinoma, potentially playing significant roles in the development of cancer.

To further investigate the professional relationship between these genes and clinical prognosis in a scientific manner, we integrated clinical data sourced from the TCGA database with the expression data pertaining to these 135 common genes. Initially, we utilized the Cox proportional hazards model for the purpose of preliminary screening, with the aim of identifying genes that are correlated with patient prognosis. Subsequently, in order to more precisely select feature genes with predictive value, we implemented three distinct machine learning algorithms: LASSO (Least Absolute Shrinkage and Selection Operator), Random Forest [19], and XGBoost [20]. Through the application of these algorithms, we generated the lambda plot for LASSO regression, the 10-fold cross-validation plot, the Random Forest plot, and the plot showing the negative log-likelihood of Cox proportional hazards regression in XGBoost as a function of iteration number. These charts provided us with rich information, aiding our understanding of the performance and effectiveness of different algorithms in feature selection.

Statistics analysis

The GO and KEGG [21] functional enrichment analyses are important means to explore the roles of key genes in biological processes. By conducting GO enrichment analysis on the selected feature genes, we can reveal the biological processes, molecular functions, and cellular components in which these genes are primarily involved. KEGG enrichment analysis further aids in understanding the metabolic pathways or signaling pathways in which these genes play a role. Additionally, using the Pearson correlation coefficient, we identified co-expressed genes that are closely related to ZC3H12D. The expression levels of these genes show significant correlation with ZC3H12D, suggesting that they may jointly participate in the occurrence and development of a specific biological process or disease. Through in-depth analysis of these co-expressed genes, we can gain a more comprehensive understanding of the mechanism by which ZC3H12D contributes to tumorigenesis. Furthermore, the ssGSEA method is employed to assess the tumor proportion within the tumor microenvironment. This method quantifies the proportions of different cell types, including tumor cells and immune cells, in the tumor microenvironment by calculating the expression levels of specific gene sets in the samples. This is of significant importance for understanding the composition of the tumor microenvironment, the interactions between tumors and immune cells, and the evaluation of the effects of immunotherapy. Through the analysis using the ssGSEA method, we can gain a more accurate understanding of the state of the tumor microenvironment, providing new insights for tumor treatment.

Results

Data process

The tpm dataset for head and neck squamous cell carcinoma (HNSC) from the xena website was obtained, and obtained a high-quality set of oral cancer (OSCC) mRNA sequencing data after screening and organizing with clinical data. The dataset includes 314 OSCC samples and 44 adjacent tissues. Additionally, we downloaded GSE30784 datasets from the GEO database for comparative analysis, which contains RNA data from 167 OSCC, 17 dysplasia, and 45 normal oral tissues. The GEO database provides a large amount of biomedical research data, enabling us to conduct more comprehensive and in-depth analysis.

Screening OSCC immune-related differential genes

Based on cibersort, 22 immune cells proportions were obtained, and according to the differential analysis of OSCC normal tissues and tumor tissues, we found that B cell naive, B cells memory, In professional terms, significant subdivisional differences were observed in OSCC tissues for Employing specialized English terminology, we began by identifying ten unique immune cell types: naive T cells CD4, resting memory T cells CD4, follicular helper T cells, macrophages M0, macrophages M1, resting dendritic cells, activated mast cells, and neutrophils (Fig. 1A). The WGCNA methodology was then utilized to construct a co-expressed genes network for these aforementioned ten differentially mungermized cell types. The soft threshold parameter was established at 7 (Fig. 1B), and distinct modules were determined using a dynamic tree cutting approach, with a minimum of 50 genes per module. The mergeCutHeight was subsequently adjusted to 0.25 to promote the merging of similar modules. Further analysis was conducted to examine the correlation between the final module and the differential immune cells.

Fig. 1.

A The CIBERSORT algorithm, paired with the LM22 gene set, was applied to quantify the relative proportions of 22 immune cell types. The Wilcoxon test was then used to evaluate the differences in these proportions between tumor and normal samples, with significance levels set at p < 0.05, p < 0.01, and p < 0.001. B A soft-threshold value of 7 was determined to construct a weighted gene co-expression network that adheres to scale-free network properties. C Gene co-expression modules were constructed, and the correlation of each module with differential immune cells was calculated to screen for potential functional gene modules. D–F Analysis of the correlation between module membership (MM) and gene significance (GS) in the turquoise module, screening for potential key genes. G A Venn diagram shows the intersection of genes related to memory B cells, M0 macrophages, and activated mast cells

After the merging process, a total of 12 modules were established, and an evaluation of the correlation between these modules and the differential immune cells was conducted to identify the most significant correlations (Fig. 1C). In the end, the turquoise module, which showed the highest correlation with the differential immune cells, was pinpointed, and the key module genes were identified based on the criteria of MM > 0.5, GS > 0.3, and p < 0.05 (Fig. 1D–F) and extracted the intersections of B.cells.memory, Macrophages.M0, and Mast.cells.activated key module genes in this module as differentially immune cell-related genes for further screening (Fig. 1G).

Construction of prognostic model of immune-related genes

Based on the intersection of OSCC (oral squamous cell carcinoma) differential genes and WGCNA (weighted gene co-expression network analysis) immune cell-related co-expressed genes, we successfully screened 135 key immune-related genes (Fig. 2A). This step provides a solid foundation for our subsequent research, allowing us to focus more on those genes that may play an important role in the immune response to OSCC. Subsequently, we performed a preliminary prognostic correlation screen for these immune-related genes using the cox regression model (Fig. 2B). Through this step, we expect to be able to identify those genes that are closely related to the prognosis of OSCC patients, thereby providing potential targets for personalized treatment of patients. To further refine our trait gene set, we employ three advanced machine learning algorithms: LASSO, randomForest, and xgboost. First, we used the LASSO regression model to select features, and plotted the lambda selection plot of LASSO regression and the 10-fold cross-validation plot (Figs. 2C and D). These graphs helped us to identify the optimal lambda values to screen out the most predictive gene traits. Subsequently, we used the random forest algorithm to evaluate the importance of the genes and plot the VIP (Variable Importance Projection) plot (Fig. 2E). The VIP plot visually shows how much each gene contributes to the random forest model, making it easier for us to identify those genes that have the greatest impact on the model’s predictions. In addition, we used the xgboost algorithm to perform feature screening and plot the negative bias logarithm of COX risk proportional regression as a function of the number of iterations (Fig. 2F). This graph not only shows the convergence process of the XGboost model, but also helps us to further identify those genes that are important for prognostic prediction. Finally, we used three machine learning methods to solidify an OSCC prognostic model composed of 6 genes (Fig. 2G).

Fig. 2.

Feature Gene Selection. A  A Venn diagram shows the intersection of differentially expressed genes (DEGs) and genes selected by weighted gene co-expression network analysis (WGCNA). B COX regression analysis was used to calculate the hazard ratio (HR) and p-value of genes to assess their correlation with prognosis. C LASSO regression was used for variable selection, with the x-axis representing log(λ) and the y-axis representing the partial likelihood deviation. D LASSO regression coefficient path diagram, showing the trend of gene regression coefficients under different λ values. E The importance of genes in the model was evaluated using the Random Forest (RF) algorithm. F The trend of the negative log-likelihood of the XGBoost-Cox model with the number of iterations. G A Venn diagram shows the intersection of genes selected by COX regression, Random Forest, and XGBoost methods

ZC3H12D gene was identified as the key prognostic gene for OSCC

In the study, we conducted a multifactorial Cox regression analysis on six intersecting genes with the aim of identifying key genes associated with oral cancer. Through this method, we ultimately identified the ZC3H12D gene as a key gene due to its minimum p-value, as illustrated in (Fig. 3A). To further validate the role of the ZC3H12D gene in oral cancer, we plotted the expression difference of this gene between unpaired and paired samples in oral cancer tissues. The results indicated that the ZC3H12D gene exhibited significantly high expression levels in oral cancer tissues, a finding reflected in (Fig. 3B and C). To explore the expression patterns of the ZC3H12D gene across different cancer types, we downloaded STAR pipeline RNAseq data from 33 different tumor projects from the TCGA database (accessible via the link https://portal.gdc.cancer.gov). We organized this data and extracted it in TPM format. Subsequently, we analyzed the expression differences of the single gene ZC3H12D across pan-cancer. The analysis results demonstrated that the ZC3H12D gene exhibited high expression characteristics in various cancer types, a finding visually represented in (Fig. 3D). The ROC curve reveals an area under the curve (AUC) of 0.756 for the ZC3H12D molecule (Fig. 4A). Subsequently, oral cancer patients were divided into two groups based on the median expression level of ZC3H12D, and a thorough analysis was conducted to examine the correlation between the expression level of ZC3H12D and the overall survival (OS) as well as disease-specific survival (DDS) rates among these patients (Fig. 4B–C). The analysis results revealed that patients exhibiting low expression levels of ZC3H12D demonstrated significantly higher survival rates compared to those with high expression levels, hinting at the potential significance of ZC3H12D as a tumor suppressor gene. Furthermore, the results underscored a notable association between low expression of ZC3H12D and poorer survival outcomes across various pathological T-stages and primary treatment outcome groups (Fig. 4D). These findings further imply that ZC3H12D may play a pivotal role in inhibiting tumor invasion and metastasis. A detailed exploration of the clinical phenotypic associations of ZC3H12D in oral cancer is presented (Fig. 5A–F).

Fig. 3.

Expression Analysis of Key Gene ZC3H12D. A Multivariate COX regression analysis of 6 cross-featured genes, showing the hazard ratio (HR), 95% confidence interval (CI), and p-value of ZC3H12D and other genes. B Comparison of ZC3H12D expression levels in tumors (Tumor) and normal tissues (Normal), with data represented as TPM (log₂(TPM + 1)), and Wilcoxon test used to assess statistical differences. C Paired analysis of matched tumor and normal tissue samples, showing the expression changes of ZC3H12D. D Expression differences of ZC3H12D in tumors (red) and normal tissues (blue) across different cancer types. Statistical significance is marked as follows: *p < 0.05; **p < 0.01; ***p < 0.001

Fig. 4.

Survival Analysis of ZC3H12D. A Receiver operating characteristic (ROC) curve assessing the predictive ability of ZC3H12D for a certain classification task. AUC = 0.756, 95% confidence interval (CI): 0.675–0.838. B Kaplan-Meier survival analysis. C Kaplan-Meier survival analysis showing disease-specific survival (DSS) probabilities for patients with high (red) and low (blue) ZC3H12D expression. D Clinical pathological characteristics table comparing basic information of patients with high and low ZC3H12D expression. Chi-square test was used to assess group differences, with statistical significance set at P < 0.05 (bolded)

Fig. 5.

Differences in ZC3H12D Expression Levels. A Heatmap of ZC3H12D expression, grouped by age, gender, Grade, Stage, T, M, N staging. B–F Comparison of ZC3H12D expression in different clinical feature groups

Through the application of univariate and multivariate Cox regression analysis methods, we conducted an in-depth assessment of the expression levels of the ZC3H12D gene and the impact of multiple clinical variable factors on patient survival duration. The research findings revealed an important discovery: there is a significant correlation between high expression levels of the ZC3H12D gene and longer survival duration in patients. This finding indicates that the expression of the ZC3H12D gene can serve as an independent prognostic protective factor, which holds significant clinical implications for the prognostic evaluation of oral cancer patients (Figs. 6A, B). To further enhance the practicality and accuracy of prognostic assessment, we constructed a nomogram based on these significant clinical and molecular biological factors, which can predict the survival probabilities of patients with oral squamous cell carcinoma (OSCC) at 1 year, 3 years, and 5 years (Fig. 6C). Additionally, to validate the predictive capability of this nomogram model, we also plotted calibration curves, which demonstrated the accuracy of the model’s predictions, indicating that the model has a high degree of accuracy in predicting the survival probabilities of oral cancer patients (Fig. 6D).

Fig. 6.

Prognostic Model of ZC3H12D and Survival Analysis. A Univariate Cox regression analysis showing the impact of age, pathological T stage, pathological N stage, pathological stage, and ZC3H12D on survival prognosis. Data are hazard ratios (HR) and their 95% confidence intervals (CI). B Multivariate Cox regression analysis incorporating significant variables to assess the independent prognostic value of ZC3H12D. C Nomogram model predicting patient survival probabilities at 1, 3, and 5 years. D Nomogram calibration curve comparing the fit of model-predicted survival probabilities with actual survival probabilities, with the gray solid line representing ideal prediction

ZC3H12D biological and pathway analysis

A comprehensive Spearman correlation analysis of the ZC3H12D gene was carried out to delve deeper into the genes closely related to its expression levels, visually demonstrating the complex interactions among the ten most significantly correlated genes through a meticulously crafted chord diagram (Fig. 7A). Based on the expression levels of ZC3H12D, the integrated patient data were meticulously divided into high expression and low expression groups, and a systematic differential gene analysis was performed. By rigorously applying the dual criteria of logFC > 1.5 and p < 0.05, we ultimately identified 439 genes with significant differences. To further explore the biological functions and potential signaling pathways of these differentially expressed genes, comprehensive GO (Gene Ontology) and KEGG (Kyoto Encyclopedia of Genes and Genomes) enrichment analyses were conducted (Fig. 7B), aiming to reveal their specific roles in biological processes and the metabolic pathways they participate in. Finally, to gain a deeper understanding of the specific roles of these differentially expressed genes in the occurrence and development of diseases, GSEA (Gene Set Enrichment Analysis) was performed on the differentially expressed genes with p < 0.05, with detailed presentations in (Fig. 7C and D) further validating the enrichment of these genes in specific biological processes, providing robust data support and theoretical basis for subsequent research.

Fig. 7.

ZC3H12D-Related Gene Analysis and Functional Enrichment. A Interaction network of ZC3H12D and its related genes. B GO and KEGG pathway enrichment analysis. C Gene set enrichment analysis (GSEA) showing significantly enriched pathways. D Pathway interaction network, where node size indicates gene set size, color indicates NES value, and connecting lines represent pathway similarity

ZC3H12D affects the tumor immune microenvironment

This study employed three immune infiltration analysis methods: CIBERSORT, ssGSEA, and ESTIMATE, to conduct an in-depth investigation of the performance of ZC3H12D in immune cell infiltration and correlation analysis, where P-values less than 0.001, 0.01, and 0.05 were represented by “***”, “**”, and “*” respectively to indicate significance levels. Firstly, the CIBERSORT method was utilized to explore the correlation between ZC3H12D expression and various immune cell types. The results indicated a significant correlation between ZC3H12D and multiple immune cell types (Fig. 8A). Furthermore, the proportions of immune cell types significantly correlated with ZC3H12D were compared between high and low expression groups, revealing a significant difference in these proportions (Fig. 8B). Secondly, the ssGSEA method was employed to analyze the correlation between ZC3H12D expression and immune cell types, further validating the correlation of ZC3H12D with various immune cell types (Fig. 8C). Simultaneously, the enrichment scores of different immune cell types in high and low expression groups of ZC3H12D were compared, Indicating a marked disparity in enrichment scores among diverse immune cell types, this analysis underscores the significance of ZC3H12D expression levels. Furthermore, a comparative assessment was conducted on stromal score, immune score, and ESTIMATE score between the high and low expression groups of ZC3H12D. The findings revealed that the high expression group exhibited notably elevated immune scores and ESTIMATE scores in comparison to the low expression group, thereby emphasizing the potential biological implications of ZC3H12D in immune regulation and tumorigenesis (Fig. 8DE). A further analysis of the correlation between ZC3H12D expression levels and immune scores and stromal scores revealed a significant positive correlation with both immune scores and stromal scores (Fig. 8F).

Fig. 8.

Correlation Analysis of ZC3H12D Expression and Immune Infiltration. A CIBERSORT method analyzing the correlation between ZC3H12D and the proportion of different immune cells. B Stacked bar chart of immune cell composition in different ZC3H12D expression groups (low expression vs. high expression). C ssGSEA method analyzing the correlation between ZC3H12D and immune cell infiltration, marking significantly related cell types. D Comparison of stromal score (StromalScore), immune score (ImmuneScore), and ESTIMATE score (ESTIMATEScore) between high and low ZC3H12D expression groups. E ssGSEA assessing the enrichment of different immune cell subsets in high and low ZC3H12D expression groups. F Spearman correlation analysis of ZC3H12D expression levels with immune score and stromal score. Statistical significance is marked as follows: *p < 0.05; **p < 0.01; ***p < 0.001

Discussion

This study delved into the clinical and molecular biology of oral squamous cell carcinoma (OSCC) patients and created a predictive model to improve the utility and accuracy of prognostic assessment. Through the analysis of clinical data and molecular biological indicators, we identified a series of factors that significantly affect the prognosis of patients, and constructed a nomogram (Fig. 6C) that predicts the 1-, 3-, and 5-year survival of patients with OSCC. In order to test the predictive performance of the nomogram model, we plotted a calibration curve, and the results showed that the model had high prediction accuracy, demonstrating the efficiency of the model in predicting the survival rate of oral cancer patients. To deeply understand ZC3H12D genes’ role in oral cancer, we conducted a thorough Spearman correlation analysis. This aimed to uncover the complex relationships between these genes and their expression levels. We also created a detailed chord plot to visually represent the interactions among ten genes closely related to ZC3H12D. The plot offers clear insights into their interplay and potential roles in oral cancer development. Based on the expression levels of the ZC3H12D, we divided the comprehensive patient data into high expression and low expression groups, and performed systematic differential gene analysis. By rigorously applying the double standard of logFC > 1.5 and p < 0.05, we finally identified 439 significantly differentially expressed genes. To further investigate the biological functions and potential signaling pathways of these differentially expressed genes, we performed comprehensive GO and KEGG enrichment analyses with the aim of revealing their specific roles in biological processes and the metabolic pathways they are involved in.

Shengping Huang [22] and his colleagues revealed that ZC3H12D serves as a novel negative regulatory factor in TLR signaling and macrophage activation, identified as a potential tumor suppressor gene. This gene is enriched in the spleen, lungs, and lymph nodes, and is induced by TLR ligands through the JNK and NF-κB pathways. Overexpression of Zc3h12d can inhibit the activation of TLR2 and TLR4, thereby suppressing JNK, ERK, and NF-κB signaling, and reducing the inflammatory response of macrophages. Furthermore, it can lower the levels of intracellular protein ubiquitination, indicating its role in host immunity and inflammatory diseases. In this study, we first discovered the association of ZC3H12D with TLR signaling and its mediation of the immune heterogeneity of oral tumors. Hong Zhang [23] et al. found that ZC3H12D was able to inhibit the initial inflammatory storm and chronic inflammation by targeting the cytokines mRNA, NF-κB, and c-FOS.

Furthermore, this study also explores the role of ZC3H12D in the tumor immune microenvironment. We employed three immune infiltration analysis methods: CIBERSORT, ssGSEA, and ESTIMATE, to conduct an in-depth investigation of ZC3H12D’s performance in immune cell infiltration and correlation analysis. The correlation between the expression of the gene ZC3H12D and various immune cell types was subjected to further validation using the Single-Sample Gene Set Enrichment Analysis (ssGSEA) approach. This analysis served to reinforce and confirm the previously observed association of ZC3H12D with a multitude of immune cell types, thereby providing additional evidence for the gene’s role in immune system regulation. Further analysis of the correlation between ZC3H12D expression levels and immune scores and stromal scores revealed significant positive correlations with both immune scores and stromal scores. In summary, this study not only reveals the potential mechanism of action of the ZC3H12D gene in oral cancer but also enhances the accuracy of prognostic assessment by constructing a nomogram model, providing new perspectives and tools for clinical treatment. Future research should further explore the specific role of ZC3H12D in the tumor immune microenvironment and its potential as a therapeutic target.

Limitation

Despite the significant findings of this study, several limitations should be noted. Firstly, the sample size used in this research may limit the generalizability of the results to a larger population. Additionally, the study was conducted in a single institution, which may introduce bias and affect the external validity. Furthermore, while we have identified ZC3H12D as a key gene associated with oral cancer prognosis, the underlying mechanisms and interactions with other genes and pathways require further elucidation. Finally, the predictive model developed in this study needs to be validated in independent cohorts to ensure its robustness and reliability in clinical practice.

Conclusion

In this study, we analyzed data from OSCC public disease cohorts and found that ZC3H12D plays a significant role in the OSCC tumor microenvironment. At the same time, patients with high expression of ZC3H12D in OSCC had a better prognosis, while patients with high OSCC stage had low expression level of ZC3H12D. These conclusions suggest that ZC3H12D promotes T cell-related cellular immunity in the OSCC tumor microenvironment and thus plays a role in inhibiting OSCC.

Author contributions

T.D. conceived and designed the study. Q.W. performed bioinformatics analyses and data visualization. S.Q. assisted with clinical data collection and interpretation. J.H. drafted the manuscript and prepared the figures. D.L. revised the manuscript critically for important intellectual content. All authors read and approved the final manuscript.

Funding

No Funding.

Data availability

GSE30784 dataset used in this study are publicly available from open-access databases https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE30784.).No special permission is required to access these datasets.

Declarations

Ethical approval and consent to participate

Ethical statement is not applicable in this article. All the authors have no possible financial or non-financial conflicts of interest related to the work described.

Consent for publication

All authors approved to publish the manuscript.

Competing interests

 The authors declare no competing interests.