A prognostic model for head and neck squamous cell carcinoma based on eosinophil extracellular trap related genes

Introduction

Head and neck squamous cell carcinoma (HNSCC), a globally prevalent malignancy ranking sixth in incidence, originates from the stratified squamous epithelium of the oral cavity, pharynx, and larynx (1). Epidemiological projections indicate a persistent increase in its incidence, with an estimated 30% increase from current levels by 2030 (2-4). As one of the most prevalent malignancies in the head and neck region, HNSCC exhibits marked aetiological heterogeneity. Betel quid chewing, alcohol consumption, tobacco smoking, infection with high-risk human papillomavirus (HPV), and genetic susceptibility represent well-established risk factors for HNSCC (5). Current therapeutic strategies for HNSCC primarily involve surgical resection, radiotherapy, and systemic pharmacotherapy. However, clinical outcomes remain unsatisfactory, primarily due to the high rates of tumor recurrence and metastasis. This compelling clinical challenge underscores the urgent need to develop novel prognostic models to guide and optimise treatment decision-making.

An increasing body of evidence highlights the central role of the tumor microenvironment (TME) in regulating HNSCC progression and prognosis (6). The TME comprises a complex network of infiltrating immune cells, among which eosinophils are notable for their widespread distribution and considerable functional plasticity at various stages of HNSCC progression (7,8). Previous studies have shown that eosinophils are capable of expressing cytokines, tumor necrosis factor-alpha, and macrophage inflammatory protein-alpha. Furthermore, superoxide generated by eosinophils infiltrating the TME enhances their cytotoxic activity, contributing to antitumor immunity via the superoxide-eosinophil axis. In the context of tumor-associated tissue eosinophilia (TATE), a high level of TATE is significantly associated with improved survival (9-11). Eosinophil extracellular traps (EETs) are DNA-protein mesh-like structures released by eosinophils into the extracellular space to mediate immune responses (12). Within this DNA mesh-like structure, both common histones and loaded bioactive proteins are present, including HMGB1 (13). Literature indicates that the HMGB1 protein can bind to the (RAGE) receptor, thereby inducing activation of the NF-κB pathway to promote cell proliferation, invasion, and metastasis in hepatocellular carcinoma cell lines, influencing tumor invasion through the HMGB1-RAGE axis (14). Given the significant involvement of eosinophils in tumor pathophysiology, investigating the role of EETs in tumor occurrence and development is of paramount importance. Building upon this rationale, the present study sought to identify key genes shared between HNSCC and EETs and to develop a prognostic model using machine learning algorithms, with the aim of providing novel prognostic biomarkers and therapeutic insights for patients with HNSCC. We present this article in accordance with the TRIPOD reporting checklist (available at https://tcr.amegroups.com/article/view/10.21037/tcr-2026-1-0071/rc).

Methods

Data acquisition and processing

For this study, RNA sequencing data were retrieved from The Cancer Genome Atlas (TCGA) database, comprising 44 normal head and neck tissue samples and 522 HNSCC specimens, alongside corresponding clinical data for 528 patients with HNSCC (Table S1). A systematic literature review was conducted using PubMed to identify studies related to EET, from which EET-related genes were extracted based on a summary of 17 previous studies (13,15-30) (Table S2). External validation was performed using the GSE41613 HNSCC dataset obtained from the Gene Expression Omnibus (GEO) database. This study was conducted in accordance with the Declaration of Helsinki and its subsequent amendments.

Processing of differentially expressed genes (DEGs)

Employing the Wilcoxon rank-sum test in R, we detected significant DEGs from raw transcripts per million (TPM) data, applying thresholds of absolute at |log₂ fold change (FC)| ≥1 and an adjusted P<0.05. These DEGs were visualized through volcano plots and hierarchically clustered heatmaps. Subsequently, the DEGs were intersected with EETs-related genes, and those present in both sets were extracted to generate a common gene subset for downstream analyses.

Screening of key genes via univariate Cox regression, least absolute shrinkage and selection operator (LASSO), and multivariate Cox regression with risk score calculation

To construct a prognostic model, three statistical algorithms were employed to screen for EETs-related DEGs in HNSCC. Initially, key genes were identified using univariate Cox regression analysis, with significance defined as P<0.05. LASSO regression analysis was then conducted using 10-fold cross-validation to further narrow down genes associated with patient survival outcomes, generating deviance plots, coefficient profiles, and expression patterns of the selected genes. Finally, multivariate Cox regression modeling finalized gene selection and computed integrated risk scores.

Kaplan-Meier survival analysis

To evaluate the association between the core genes (ANXA5, CCL26, CXCL8, PDIA3, and ZAP70) and overall survival (OS) in HNSCC patients, Kaplan-Meier analysis was performed on TCGA-derived transcriptomic and clinical datasets. Patients were dichotomized into high-/low-risk groups by median gene expression thresholds, with survival curves generated using R’s Survminer package. Ultimately, a clinically applicable nomogram incorporated the expression profiles of these five signature genes. Each gene’s expression value was standardised into a score, and the composite score was applied to assess the predictive efficacy for disease risk.

Prognostic risk model construction via random forest

Prognostic risk scores were computed from regression coefficients and expression profiles of hub genes. Cohort survival-risk associations were evaluated, dichotomizing patients into high-/low-risk groups. Differential expression patterns of the five signature genes were compared between these strata, along with the relationships of OS and survival status with the risk scores. Furthermore, we calculated an eosinophil infiltration score for each sample through single-sample gene set enrichment analysis (ssGSEA) and evaluated its correlation with the risk scores. Given limited confounders in external datasets, univariate/multivariate Cox regressions on shared prognostic variables from training/validation cohorts identified independent predictors. A random forest algorithm constructed the five-gene prognostic signature. Multivariable survival analysis incorporating risk scores, age, and other covariates predicted OS. External validation was performed using an independent cohort. Time-dependent receiver operating characteristic (ROC) curves assessed model accuracy, with area under the curve (AUC) comparisons across cohorts. Clinical utility was quantified via decision curve analysis (DCA), computing net benefit against “treat-all” and “treat-none” benchmarks.

Gene Ontology (GO)/Kyoto Encyclopedia of Genes and Genomes (KEGG) and Gene Set Enrichment Analysis (GSEA) enrichment analysis

Functional enrichment analysis of GO terms and KEGG pathways was conducted via R’s the “clusterProfiler” package, with significant terms filtered by dual criteria (P<0.05, q<1). Results were visualized through bar charts and enrichment dot plots. Concurrently, GSEA identified disease-relevant biological pathways and gene sets associated with HNSCC and EETs.

Immune cell infiltration analysis

The CIBERSORT algorithm was used to analyse EETs-related gene expression data from HNSCC samples and estimate the relative proportions of different immune cell types. Additionally, a correlation coefficient matrix depicting the interactions among 22 immune cell subtypes in the TME was constructed using the “corrplot” package in R. Immune subtype distributions across risk strata were visualized via Vioplot-generated violin diagrams. Intergroup significance was evaluated by Wilcoxon testing. TME indices, including stromal/immune components, ESTIMATE composites, and purity metrics, were derived through ESTIMATE algorithm implementation (R estimate package). These scores were subsequently visualised and compared according to the predefined risk stratification. We further performed correlation and clustering analyses between the five signature genes and a panel of key Th2 cell-related genes as well as eosinophil chemokines, with Pearson correlation coefficients calculated and visualized using a heatmap.

Drug sensitivity prediction

Based on the intersecting gene expression dataset, predictions of sensitivity to various commonly used targeted therapeutics were made using the “oncoPredict” R package. The prediction model was trained using data from the Genomics of Drug Sensitivity in Cancer (GDSC) database, with a fixed random seed (seed =999) employed throughout to ensure reproducibility. Subsequently, the half-maximal inhibitory concentration (IC₅₀) was estimated using the empirical Bayes method. Predicted drug sensitivities were integrated into the risk-stratified model, Wilcoxon rank-sum tests assessed intergroup risk differences (P<0.001). Finally, multigroup comparisons were illustrated using boxplots to highlight differential responses to targeted therapies between the two risk cohorts.

Statistical analysis

Results

DEGs associated with EETs in HNSCC

The workflow for analysing EETs-related gene signatures in HNSCC is shown in Figure 1. Differential gene expression analysis between tumor tissues and matched normal controls in the training cohort identified 3,573 DEGs (Figure 2A). Genes overlapping between DEGs in HNSCC and EETs-related genes were designated as EETs-related (Figure 2B,2C). In total, 57 overlapping genes were identified (Table S3).

Figure 1 Flow diagram of the study. DEG, differentially expressed gene; EET, eosinophil extracellular trap; GEO, Gene Expression Omnibus; HNSCC, head and neck squamous cell carcinoma; KM, Kaplan-Meier; LASSO, least absolute shrinkage and selection operator; ROC, receiver operating characteristic; TCGA, The Cancer Genome Atlas.

Figure 2 Identification of EETs related genes in HNSCC patients. (A) A volcano plot displays differential expression of genes between normal and tumor tissues, with upregulated genes in tumor tissue marked in red and downregulated genes in green, in black. (B) Heatmap analysis of DEGs in HNSCC. (C) Fifty-seven differentially expressed genes with prognostic significance. DEG, differentially expressed gene; EET, eosinophil extracellular trap; FC, fold change; HNSCC, head and neck squamous cell carcinoma.

Identification of pivotal genes

Univariate Cox regression identified 10 prognostically significant candidate genes. This gene set was further refined using LASSO regression and multivariate Cox analysis, resulting in a five-gene signature comprising ANXA5, CCL26, CXCL8, PDIA3, and ZAP70 (Figure 3A,3B). Kaplan-Meier survival analysis stratified patients into high- and low-expression groups based on the median expression level of each gene. Log-rank tests revealed statistically significant differences in OS for three of the five genes (PDIA3, ANXA5, and ZAP70; P<0.05) (Figure 3C-3E), whereas CXCL8 and CCL26 showed no significant prognostic value (P>0.05) (Figure 3F,3G), suggesting their limited utility as individual biomarkers. A total risk score was generated by summing the prognostic scores of the five key genes. As shown in Figure 3H, a significant positive correlation was observed between increasing total scores and elevated HNSCC risk, indicating that the combined gene signature offers strong predictive performance for disease risk.

Figure 3 Identification of pivotal genes. (A) LASSO-Cox regression analysis. (B) Tenfold cross-validation. (C-G) Kaplan-Meier survival analysis of HNSCC patients stratified by EET-related gene expression levels. Functional analysis of the 5 EETs-related gene model. (H) The scoring system based on expression levels of five hub genes. EET, eosinophil extracellular trap; HNSCC, head and neck squamous cell carcinoma; LASSO, least absolute shrinkage and selection operator.

Construction and validation of EETs-related prognostic model for HNSCC patients

Risk stratification using five-gene-derived scores (calculated from regression coefficients and expression levels) categorized patients into high-/low-risk cohorts (Figure 4A). A gene expression heatmap demonstrated upregulated expression of four genes in high-risk patients versus downregulated ZAP70 (Figure 4B). Mortality was significantly elevated in the high-risk cohort (41.6%) compared to low-risk counterparts (24.7%) (Figure 4C). The correlation plot between the eosinophil infiltration scores and the risk scores revealed that increased eosinophil infiltration correlates with a reduced risk score, implying a role for eosinophils in antitumor immune responses and indicating a favorable prognosis (Figure 4D). Multivariate analysis established age and risk scores as independent prognostic indicators (Table S4). A prognostic model was constructed using a random forest algorithm incorporating the five-gene signature. Time-dependent ROC analysis demonstrated robust predictive performance in the training cohort (1-year, AUC =0.815; 3-year, 0.831; 5-year, 0.832). Validation cohort results showed concordant accuracy (0.755, 0.723, 0.727 at matched intervals) (Figure 4E). The AUC values for the training set all exceeded 0.8, demonstrating that the five-gene model can predict patient prognosis with a high degree of accuracy for 1-, 3-, and 5-year survival rates. Furthermore, external validation maintained an AUC above 0.7, reflecting the model’s excellent generalizability and its ability to provide reliable prognostic guidance in independent populations. DCA revealed that the five-gene scoring model offered a greater net clinical benefit than the intervention for all patients or intervention for none strategies, particularly at threshold probabilities >0.6 (Figure 4F).

Figure 4 Prognostic model of HNSCC based on key EETs-related DEGs. (A) Risk scores of two risk groups by analyzing the training set. (B) Expression levels of five prognostic genes validated by the prognosis model based on the risk scores obtained by an analysis of the training set. (C) Survival of the training set. The abscissa represents risk scores and the ordinate represents survival time. Red dots indicate death samples and blue dots indicate survival samples. (D) Correlation between eosinophil infiltration and risk score. (E) Accuracy of the prognostic model in predicting 1-, 3-, and 5-year survival rates across training and validation cohorts. (F) Decision curve analysis. AUC, area under the curve; DEG, differentially expressed gene; EET, eosinophil extracellular trap; HNSCC, head and neck squamous cell carcinoma; ROC, receiver operating characteristic.

GO/KEGG and GSEA enrichment analysis

To explore the functional implications of risk-related genes, enrichment analyses were conducted using GO and KEGG pathways. GO biological process (BP) terms revealed enrichment in leukocyte migration, whereas focal adhesions and cell-substrate junctions were significantly represented in the cellular component (CC) category. For molecular function (MF), chemokine activity, chemokine receptor binding, cytokine activity, cytokine receptor binding, and G protein-coupled receptor binding were prominent (Figure 5A,5B). KEGG pathway analysis indicated enrichment in viral protein interactions with cytokines and cytokine receptors, nuclear factor (NF)-kappa B signalling, Yersinia infection, chemokine signalling, and human cytomegalovirus infection. GSEA results further identified enrichment in pathways associated with the cell cycle, cytokine receptor interaction, extracellular matrix (ECM)-receptor interaction, focal adhesion, and cancer-related signalling (Figure 5C).

Figure 5 Functional analysis of the EET-related gene model. (A) GO enrichment analysis. (B) KEGG enrichment analysis. (C) GSEA enrichment analysis. BP, biological process; CC, cellular component; EET, eosinophil extracellular trap; GO, Gene Ontology; GSEA, Gene Set Enrichment Analysis; KEGG, Kyoto Encyclopedia of Genes and Genomes; MF, molecular function.

Immune cell infiltration analysis

Using the CBERSORT algorithm, a correlation coefficient matrix was generated for 22 immune cell subtypes within the HNSCC TME (Figure 6A). Notably, eosinophils exhibited a strong positive correlation with activated mast cells, whereas CD4⁺ memory-activated T cells were positively correlated with CD8⁺ T cells, suggesting potential immune synergy in the TME. Conversely, M0 macrophages showed a significant negative correlation with CD8⁺ T cells, indicating that M0 macrophages may suppress CD8⁺ T cell function. The correlation heatmap illustrated the relative proportions of the 22 immune cell subtypes across patient samples, with each column representing a cell type and each row a patient sample (Figure 6B). Violin plots demonstrated statistically significant intergroup differences for 10 immune cell types (Figure 6C). Plasma cells and T effector cells [including CD4⁺ memory-activated T cells, CD8⁺ T cells, follicular helper T cells, and regulatory T cells (Tregs)], as well as resting mast cells, were more prevalent in the low-risk group than in the high-risk group. In contrast, the high-risk group showed significantly higher proportions of M0 macrophages, natural killer (NK) cells, CD4⁺ memory resting T cells, and activated dendritic cells. Furthermore, ESTIMATE analysis showed significantly higher stromal, and ESTIMATE scores in the high-risk group than in the low-risk group (P<0.05), indicating a more complex TME with increased stromal cell infiltration in these tumors (Figure 6D).

Figure 6 Analysis of the tumor immune microenvironment and drug sensitivity between high- and low-risk groups. (A) The proportional composition of 22 immune cell subsets within the HNSCC TME. (B) The correlation heatmap. (C) Violin plots. (D) TME score plots. (E) Target genes correlations with Th2 and chemokine genes. (F) Drug sensitivity prediction. *, P<0.05; ***, P<0.001. HNSCC, head and neck squamous cell carcinoma; TME, tumor microenvironment.

The resulting correlation heatmap revealed that, among the five target genes, ZAP70 exhibited positive correlations with most Th2-associated genes on the horizontal axis. Notably, the strongest positive correlations were observed with IL13 (r=0.40) and GATA3 (r=0.35), suggesting that ZAP70 expression is closely associated with the activation of Th2-type immune responses. Meanwhile, ANXA5 showed a relatively strong positive correlation with CCR3 (r=0.33), whereas its correlations with other genes were weaker, implying a specific role for ANXA5 in chemokine receptor-mediated cell migration. CCL26 exhibited a weak negative correlation with GATA3. In the clustering analysis, ZAP70 was clustered into a distinct group, indicating that its expression pattern differed significantly from those of the other four genes (CCL26, CXCL8, ANXA5, and PDIA3), which aligns consistently with our previous findings on the expression profiles of these five genes between the high- and low-risk groups (Figure 6E).

Drug sensitivity prediction

The risk assessment model demonstrates greater drug sensitivity in the low-risk group than in the high-risk group (Figure 6F). Pharmacological sensitivity analysis in HNSCC, aligned with clinical practice, indicated that certain widely used chemotherapeutic agents—particularly platinum-based drugs (cisplatin, oxaliplatin)—were more effective in the high-risk group than in the low-risk group. Taxanes (docetaxel, paclitaxel) showed efficacy in both risk groups, whereas topoisomerase inhibitors (mitoxantrone) and targeted agents (JQ1, Entinostat, VE-822) demonstrated greater efficacy in the low-risk group than in the high-risk group. These agents exert their antitumor effects through distinct molecular mechanisms involving multiple HNSCC-associated biological pathways.

Discussion

HNSCC ranks among the most prevalent cancers globally, with approximately 650,000 new cases diagnosed annually (31). Despite advances in diagnosis and treatment, this malignancy continues to pose significant clinical challenges due to its propensity for recurrence and metastasis, thereby underscoring the urgent need to identify key genes involved in its pathogenesis. Emerging evidence suggests that eosinophils play a complex yet pivotal role in HNSCC development (32), with EETs potentially promoting tumor progression through immunomodulatory mechanisms. In the present study, we integrated univariate Cox regression, LASSO, multivariate Cox regression, and random forest algorithms to develop a prognostic risk model based on EETs-related genes to predict survival outcomes in patients with HNSCC. Our findings demonstrated that five key genes—ANXA5, CCL26, CXCL8, PDIA3, and ZAP70—exhibited strong predictive value for HNSCC patient survival. Time-dependent ROC analysis revealed that the model achieved high predictive accuracy for 1-, 3-, and 5-year survival. Specifically, the AUC exceeded 83% in the training cohort and 72% in the validation cohort, indicating robust predictive capability.

Based on previous literature, we identified a set of EETs-related genes through an extensive review and found all five core genes to be associated with HNSCC in various contexts. ANXA5 has been reported as a significant prognostic biomarker in oral squamous cell carcinoma (OSCC), with high expression in peritumoral fibroblast-like cells correlating with markedly reduced patient survival (33). CCL26 and CXCL8 are also strongly associated with OSCC (34). In particular, CXCL8 plays a crucial role in promoting angiogenesis, maintaining tumor vasculature, and modulating immune responses to facilitate immune evasion (35). PDIA3, initially characterised as a housekeeping gene involved in protein folding, has been identified as hyperactive in OSCC and is now recognised as a key hub gene in the disease (36). Although direct evidence linking ZAP70 to HNSCC is limited—its primary role being associated with B-cell malignancies—its inclusion in our prognostic model suggests potential relevance (37). ZAP70 may influence the tumor immune microenvironment of HNSCC by regulating or interacting with immune cells, thereby affecting patient outcomes. Further investigation is warranted to elucidate its precise role in HNSCC, which may offer novel diagnostic and therapeutic targets.

Previous studies have demonstrated that ZAP70 deficiency can activate Syk kinase and enhance STAT3 phosphorylation, leading to the polarisation of M0 macrophages into the M2 phenotype (38). Concurrently, CXCL8 suppresses CD8⁺ T cell infiltration, contributing to an immunosuppressive microenvironment (39). In our analysis, immune cell infiltration data revealed that M0 macrophage levels were significantly lower in the high-risk group, whereas CD8⁺ T cell infiltration was higher in the low-risk cohort. Furthermore, low ZAP70 expression correlated with high-risk status, and elevated CXCL8 levels were simultaneously observed in the same group, supporting the hypothesis of a potential synergistic effect between these genes. This dual mechanism—centred on the STAT3/CXCL8 signalling axis—may drive tumor progression in HNSCC. Additionally, previous studies have demonstrated that the Mincle/Syk/NF-κB pathway is essential for sustaining the tumor-promoting activity of tumor-associated macrophages (40). In line with this, our KEGG enrichment analysis showed that both Syk and NF-κB pathways were significantly enriched in the high-risk group, providing further biological validation of our findings.

In studies on neutrophil extracellular traps (NETs), circulating tumor cells have been shown to be entrapped within NETs under both static and dynamic conditions in vitro, facilitating the capture and elimination of invasive tumor cells (41). Based on the role of NETs in the TME, we propose a hypothesis regarding the function of EETs in HNSCC tumors: our key genes may facilitate the formation of a dense physical barrier around the tumor periphery by modulating the release of DNA scaffolds structures from EETs, along with their associated granule proteins (e.g., MBP, ECP), exerting a tumoricidal effect. In the low-risk group, abundant EETs encapsulated the tumor nests, acting in concert with highly infiltrated CD8⁺ T cells to exert tumoricidal effects.

Drug sensitivity prediction analysis revealed distinct therapeutic responses between the risk groups. High-risk patients were more likely to respond to platinum-based chemotherapeutic agents, whereas low-risk patients demonstrated greater sensitivity to novel targeted therapies. Notably, the BET inhibitor JQ1 has shown potent antitumor activity in preclinical models of various cancers, including oral, breast, and pancreatic malignancies (42-44). These findings suggest that JQ1 may hold promotion for clinical use in HNSCC, particularly in patients classified as low-risk, although further clinical investigations are warranted to validate its therapeutic potential and offer novel treatment paradigms for clinical practice.

Limitations

While our study provides a validated prognostic model, several limitations should be considered. Firstly, the scale and clinical depth of publicly available data constrained our validation efforts. The lack of extensive clinical records across databases meant that external validation had to rely on a limited set of shared prognostic factors, and we could not perform a validation using a large, independent cohort with complete clinical matching. Secondly, the generalizability of our findings may be influenced by the datasets employed. Although we utilized reputable sources (TCGA and GEO), these cohorts may not capture the full spectrum of HNSCC heterogeneity. Additionally, the study relied solely on bioinformatic analyses without experimental validation in cellular or animal models, which limits the mechanistic interpretation of the roles played by the EETs-related genes. Finally, the current biological understanding and methodological scope present a constraint. The limited number of established EETs-related genes and our use of standard analytical methods suggest that both the gene set and the computational techniques warrant further refinement in subsequent research.

Conclusions

In summary, this study utilized integrative bioinformatics approaches to identify five key EETs-related genes associated with prognosis in patients with HNSCC. The developed prognostic model demonstrated strong predictive capability and was validated through pathway enrichment analysis, immune infiltration profiling, and drug sensitivity prediction. Future research should incorporate more comprehensive EETs-related gene datasets and employ advanced analytical frameworks such as multi-omics integration and machine learning. Experimental validation of these key genes in clinical samples and prospective clinical studies will be essential to confirm their diagnostic and therapeutic relevance. The findings offer valuable insights into the tumor biology of HNSCC and may help guide future research directions.

Acknowledgments

We would like to thank TCGA and GEO for providing the data.

Footnote

Reporting Checklist: The authors have completed the TRIPOD reporting checklist. Available at https://tcr.amegroups.com/article/view/10.21037/tcr-2026-1-0071/rc

Peer Review File: Available at https://tcr.amegroups.com/article/view/10.21037/tcr-2026-1-0071/prf

Funding: This work was supported by National College Students Innovation and Entrepreneurship Training Program (No. 20240333), Natural Science Foundation of Shanxi Province (Nos. 202403021221256 and 202403021211122) and the scientific research project of Health Commission of Shanxi Province (No. 2024100).

Conflicts of Interest: All authors have completed the ICMJE uniform disclosure form (available at https://tcr.amegroups.com/article/view/10.21037/tcr-2026-1-0071/coif). The authors have no conflicts of interest to declare.

Ethical Statement: The authors are accountable for all aspects of the work in ensuring that questions related to the accuracy or integrity of any part of the work are appropriately investigated and resolved. This study was conducted in accordance with the Declaration of Helsinki and its subsequent amendments.

Open Access Statement: This is an Open Access article distributed in accordance with the Creative Commons Attribution-NonCommercial-NoDerivs 4.0 International License (CC BY-NC-ND 4.0), which permits the non-commercial replication and distribution of the article with the strict proviso that no changes or edits are made and the original work is properly cited (including links to both the formal publication through the relevant DOI and the license). See: https://creativecommons.org/licenses/by-nc-nd/4.0/.

References

1. Johnson DE, Burtness B, Leemans CR, et al. Head and neck squamous cell carcinoma. Nat Rev Dis Primers 2020;6:92. [Crossref] [PubMed]
2. Ferlay J, Colombet M, Soerjomataram I, et al. Estimating the global cancer incidence and mortality in 2018: GLOBOCAN sources and methods. Int J Cancer 2019;144:1941-53. [Crossref] [PubMed]
3. Bray F, Laversanne M, Sung H, et al. Global cancer statistics 2022: GLOBOCAN estimates of incidence and mortality worldwide for 36 cancers in 185 countries. CA Cancer J Clin 2024;74:229-63. [Crossref] [PubMed]
4. Global Cancer Observatory: Cancer Today. Lyon: IARC. 2024. [Accessed April 2, 2026]. Available online: https://gco.iarc.fr/today
5. List of Classifications by cancer sites with sufficient or limited evidence in humans, Volumes 1 to 127. IARC Monographs On The Identification Of Carcinogenic Hazards To Humans. 2019. [Accessed December 9, 2025]. Available online: https://monographs.iarc.fr/wp-content/uploads/2019/07/Classification_by_cancer_site_127.pdf
6. Grisaru-Tal S, Itan M, Klion AD, et al. A new dawn for eosinophils in the tumour microenvironment. Nat Rev Cancer 2020;20:594-607. [Crossref] [PubMed]
7. Zhao C, Zhu H, Tian Y, et al. SPINK5 is a key regulator of eosinophil extracellular traps in head and neck squamous cell carcinoma. Discov Oncol 2024;15:627. [Crossref] [PubMed]
8. Tostes Oliveira D, Tjioe KC, Assao A, et al. Tissue eosinophilia and its association with tumoral invasion of oral cancer. Int J Surg Pathol 2009;17:244-9. [Crossref] [PubMed]
9. Costa JJ, Matossian K, Resnick MB, et al. Human eosinophils can express the cytokines tumor necrosis factor-alpha and macrophage inflammatory protein-1 alpha. J Clin Invest 1993;91:2673-84. [Crossref] [PubMed]
10. Hu G, Wang S, Zhong K, et al. Tumor-associated tissue eosinophilia predicts favorable clinical outcome in solid tumors: a meta-analysis. BMC Cancer 2020;20:454. [Crossref] [PubMed]
11. Ito T, Hirahara K, Onodera A, et al. Anti-tumor immunity via the superoxide-eosinophil axis induced by a lipophilic component of Mycobacterium lipomannan. Int Immunol 2017;29:411-21. [Crossref] [PubMed]
12. Shen K, Zhang M, Zhao R, et al. Eosinophil extracellular traps in asthma: implications for pathogenesis and therapy. Respir Res 2023;24:231. [Crossref] [PubMed]
13. Ueki S, Melo RC, Ghiran I, et al. Eosinophil extracellular DNA trap cell death mediates lytic release of free secretion-competent eosinophil granules in humans. Blood 2013;121:2074-83. [Crossref] [PubMed]
14. Chen RC, Yi PP, Zhou RR, et al. The role of HMGB1-RAGE axis in migration and invasion of hepatocellular carcinoma cell lines. Mol Cell Biochem 2014;390:271-80. [Crossref] [PubMed]
15. Ueki S, Konno Y, Takeda M, et al. Eosinophil extracellular trap cell death-derived DNA traps: Their presence in secretions and functional attributes. J Allergy Clin Immunol 2016;137:258-67. [Crossref] [PubMed]
16. Lu Y, Huang Y, Li J, et al. Eosinophil extracellular traps drive asthma progression through neuro-immune signals. Nat Cell Biol 2021;23:1060-72. [Crossref] [PubMed]
17. Ueki S, Tokunaga T, Melo RCN, et al. Charcot-Leyden crystal formation is closely associated with eosinophil extracellular trap cell death. Blood 2018;132:2183-7. [Crossref] [PubMed]
18. Morshed M, Yousefi S, Stöckle C, et al. Thymic stromal lymphopoietin stimulates the formation of eosinophil extracellular traps. Allergy 2012;67:1127-37. [Crossref] [PubMed]
19. Yousefi S, Gold JA, Andina N, et al. Catapult-like release of mitochondrial DNA by eosinophils contributes to antibacterial defense. Nat Med 2008;14:949-53. [Crossref] [PubMed]
20. Yousefi S, Simon D, Simon HU. Eosinophil extracellular DNA traps: molecular mechanisms and potential roles in disease. Curr Opin Immunol 2012;24:736-9. [Crossref] [PubMed]
21. Ehrens A, Lenz B, Neumann AL, et al. Microfilariae Trigger Eosinophil Extracellular DNA Traps in a Dectin-1-Dependent Manner. Cell Rep 2021;34:108621. [Crossref] [PubMed]
22. Simon D, Radonjic-Hösli S, Straumann A, et al. Active eosinophilic esophagitis is characterized by epithelial barrier defects and eosinophil extracellular trap formation. Allergy 2015;70:443-52. [Crossref] [PubMed]
23. Kim HJ, Sim MS, Lee DH, et al. Lysophosphatidylserine induces eosinophil extracellular trap formation and degranulation: Implications in severe asthma. Allergy 2020;75:3159-70. [Crossref] [PubMed]
24. Yousefi S, Sharma SK, Stojkov D, et al. Oxidative damage of SP-D abolishes control of eosinophil extracellular DNA trap formation. J Leukoc Biol 2018;104:205-14. [Crossref] [PubMed]
25. Muniz VS, Silva JC, Braga YAV, et al. Eosinophils release extracellular DNA traps in response to Aspergillus fumigatus. J Allergy Clin Immunol 2018;141:571-585.e7. [Crossref] [PubMed]
26. da Cunha AA, Silveira JS, Antunes GL, et al. Cysteinyl leukotriene induces eosinophil extracellular trap formation via cysteinyl leukotriene 1 receptor in a murine model of asthma. Exp Lung Res 2021;47:355-67. [Crossref] [PubMed]
27. Barroso MV, Gropillo I, Detoni MAA, et al. Structural and Signaling Events Driving Aspergillus fumigatus-Induced Human Eosinophil Extracellular Trap Release. Front Microbiol 2021;12:633696. [Crossref] [PubMed]
28. Yang L, Liu Q, Zhang X, et al. DNA of neutrophil extracellular traps promotes cancer metastasis via CCDC25. Nature 2020;583:133-8. [Crossref] [PubMed]
29. Shen S, Fang H, Li X, et al. Eosinophil extracellular traps drive T follicular helper cell differentiation via VIRMA-dependent MAF stabilization in bullous pemphigoid. J Allergy Clin Immunol 2025;155:1357-70. [Crossref] [PubMed]
30. Francischetti IMB, Alejo JC, Sivanandham R, et al. Neutrophil and Eosinophil Extracellular Traps in Hodgkin Lymphoma. Hemasphere 2021;5:e633. [Crossref] [PubMed]
31. Raj S, Kesari KK, Kumar A, et al. Molecular mechanism(s) of regulation(s) of c-MET/HGF signaling in head and neck cancer. Mol Cancer 2022;21:31. [Crossref] [PubMed]
32. Helm TN, Bhele S, Fanburg-Smith JC. Squamous Cell Carcinoma with Prominent Eosinophils. Head Neck Pathol 2024;18:115. [Crossref] [PubMed]
33. Zhou T, Zhang X, Song Y, et al. Annexin A5 is a novel prognostic biomarker in oral squamous cell carcinoma. J Oral Pathol Med 2024;53:538-43. [Crossref] [PubMed]
34. Zhao XT, Zhu Y, Zhou JF, et al. Development of a novel 7 immune-related genes prognostic model for oral cancer: A study based on TCGA database. Oral Oncol 2021;112:105088. [Crossref] [PubMed]
35. Chen X, Lei H, Cheng Y, et al. CXCL8, MMP12, and MMP13 are common biomarkers of periodontitis and oral squamous cell carcinoma. Oral Dis 2024;30:390-407. [Crossref] [PubMed]
36. He Y, Shao F, Pi W, et al. Largescale Transcriptomics Analysis Suggests Over-Expression of BGH3, MMP9 and PDIA3 in Oral Squamous Cell Carcinoma. PLoS One 2016;11:e0146530. [Crossref] [PubMed]
37. Leveille E, Chan LN, Mirza AS, et al. SYK and ZAP70 kinases in autoimmunity and lymphoid malignancies. Cell Signal 2022;94:110331. [Crossref] [PubMed]
38. Sadras T, Martin M, Kume K, et al. Developmental partitioning of SYK and ZAP70 prevents autoimmunity and cancer. Mol Cell 2021;81:2094-2111.e9. [Crossref] [PubMed]
39. Shao Y, Lan Y, Chai X, et al. CXCL8 induces M2 macrophage polarization and inhibits CD8(+) T cell infiltration to generate an immunosuppressive microenvironment in colorectal cancer. FASEB J 2023;37:e23173. [Crossref] [PubMed]
40. Li C, Xue VW, Wang QM, et al. The Mincle/Syk/NF-κB Signaling Circuit Is Essential for Maintaining the Protumoral Activities of Tumor-Associated Macrophages. Cancer Immunol Res 2020;8:1004-17. [Crossref] [PubMed]
41. Cools-Lartigue J, Spicer J, McDonald B, et al. Neutrophil extracellular traps sequester circulating tumor cells and promote metastasis. J Clin Invest 2013;123:3446-58. [Crossref] [PubMed]
42. Li Z, Duan J, Liu Z, et al. A triple-mode strategy on JQ1-loaded nanoplatform for superior antitumor therapy in pancreatic cancer. Mater Today Bio 2025;32:101696. [Crossref] [PubMed]
43. Jaksic Karisik M, Lazarevic M, Mitic D, et al. JQ1 Treatment and miR-21 Silencing Activate Apoptosis of CD44+ Oral Cancer Cells. Int J Mol Sci 2025;26:1241. Erratum in: Int J Mol Sci 2026;27:854. [Crossref] [PubMed]
44. Zhang H, Lu L, Yi C, et al. BRD4 regulates m(6)A of ESPL1 mRNA via interaction with ALKBH5 to modulate breast cancer progression. Acta Pharm Sin B 2025;15:1552-70. [Crossref] [PubMed]