Role of PDGFA expression in the prognosis of head and neck squamous cell carcinoma: construction of prognostic nomogram and functional mechanism exploration

Introduction

Head and neck squamous cell carcinoma (HNSCC) arises from the mucosal epithelium of the oral cavity, pharynx and larynx, representing the most prevalent malignancy in the head and neck region (1). Its development is influenced by multiple risk factors, including tobacco-derived carcinogens, excessive alcohol consumption, and human papillomavirus infection (2). Current treatment strategies mainly include surgery, radiation, immunotherapy and systemic therapy, but some patients encounter poor prognosis and shorter overall survival (OS) due to drug resistance and post-treatment recurrence (3,4).

The platelet-derived growth factor (PDGF) family consists of four ligands (PDGFA, PDGFB, PDGFC, and PDGFD) that exert their cellular effects by binding to two receptor tyrosine kinases, PDGFRα and PDGFRβ (5). The PDGF family is upregulated in various malignant tumor cells and tissues (6), including esophageal squamous cell carcinoma (7), liver cancer (8), and glioma (9). Upon binding to PDGFRs, PDGF triggers receptor dimerization and phosphorylation (5), activating several downstream signaling pathways, including PI3K/Akt and MAPK/ERK, which promote cell proliferation, migration, survival, angiogenesis, and tumor progression in the tumor microenvironment (TME) (10). As a result, new cancer treatment strategies targeting the PDGF/PDGFR pathway have emerged. For instance, PDGFB plays important roles in gastric cancer, serving as valuable biomarkers for immune response modulation (11). Yan et al. developed novel PDGFB-conjugated copper gadolinium oxide nanoclusters (PDGFB-CGO), which could actively target tumor tissues, inhibit epithelial-mesenchymal transition and tumor angiogenesis, and cancer cell migration and invasion (12). These findings prompted us to investigate the role of PDGF/PDGFR in HNSCC, which may uncover potential targets and a theoretical basis for clinical management.

While the oncogenic role of PDGF/PDGFR signaling has been established in various cancers, its specific contribution to HNSCC progression and its potential as a prognostic biomarker remain incompletely characterized. Therefore, this study analyzed PDGF/PDGFR expression in HNSCC and its association with prognosis based on The Cancer Genome Atlas (TCGA) and the Gene Expression Omnibus (GEO) databases, and constructed a nomogram to predict the OS of patients. Additionally, the biological role of selected gene in HNSCC and its potential molecular mechanisms were investigated. Importantly, this study focuses specifically on the prognostic significance of PDGF/PDGFR family members in HNSCC, rather than providing a comprehensive mechanistic dissection of their signaling pathways. Although we briefly discuss downstream effectors to establish biological context, the primary objective is to evaluate whether PDGF/PDGFR expression levels correlate with clinical outcomes and could serve as practical prognostic model. We present this article in accordance with the TRIPOD reporting checklist (available at https://tcr.amegroups.com/article/view/10.21037/tcr-2025-1-2907/rc).

Methods

Data source

Transcriptome and clinical data for 522 HNSCC samples were gathered from TCGA database (https://portal.gdc.cancer.gov/repository). To validate expression of candidate PDGF/PDGFR-related genes, the GSE65858 dataset from GEO database was employed. The analysis was performed in R software (version 4.4.0, Table 1). Due to the large number of samples with unclear clinical characteristics, this study used the American Joint Committee on Cancer (AJCC) version 7 data, removing data with “Not Available”, “T0”, “TX”, “NX” and survival time less than 0. This resulted in a final dataset of 311 HNSCC samples for subsequent analysis. The GSE9844 dataset includes 26 samples of oral tongue squamous cell carcinoma (OTSCC) and 12 control samples. The GSE41613 dataset contains 97 samples of oral squamous cell carcinoma (OSCC). The GSE65858 dataset consists of 270 HNSCC samples. The clinical feature information for TCGA, GSE9844, GSE41613, and GSE65858 datasets is provided in supplementary tables 1-4 (available at https://cdn.amegroups.cn/static/public/tcr-2025-1-2907-1.xlsx). To minimize platform-specific biases in gene expression, data from different microarray platforms were standardized prior to analysis. The study was conducted in accordance with the Declaration of Helsinki and its subsequent amendments.

Nomogram construction and validation

Six PDGF/PDGFR genes including PDGFA, PDGFB, PDGFC, PDGFD, PDGFRA, and PDGFRB were selected as the candidate genes to investigate their correlation with the HNSCC prognosis. Univariate Cox regression analysis was performed using R packages such as “survival”, “survminer”, “rms”, “regplot”, and “timeROC” to screen for genes linked with the prognosis of HNSCC. The R package “GEOquery” was used to download the GSE9844 data, verify the expression of the selected genes, and the R packages “ggplot2” and “ggpubr” were used to draw box plots to visualize the results. After univariate Cox regression analysis, only PDGFA was selected among the candidate genes with a P value of less than 0.05. Then the relationship between PDGFA expression and clinical features (e.g., age, tumor, and node) was evaluated, followed by Kaplan-Meier (KM) survival analysis to assess its association with OS. GSE41613 was used to validate the relationship between PDGFA expression and prognosis. The R package “autoReg”, “MASS” was used to select clinical characteristics to include in the nomogram, and the nomogram was drawn using the clinical characteristics with a univariate and multivariate Cox regression analysis result, as well as stepwise selection based on the Akaike information criterion (AIC). Additionally, calibration curves for 2-, 3-, and 5-year OS were then constructed. The performance of nomogram was evaluated using time-dependent receiver operating characteristic (ROC) curve analysis and decision curve analysis (DCA). The efficacy of the nomogram was validated using the GSE65858 dataset. The concordance index (C-index) values of the nomograms in these two cohorts were calculated.

Somatic mutation analysis based on PDGFA grouping

Somatic mutation data for HNSCC were obtained from TCGA database in mutation annotation format. The R package “maftools” was employed to analyze and process the MAF data (13). The low and high PDGFA expression groups were compared, using a P<0.01 as the threshold, to identify differentially mutated genes (DMGs).

Functional enrichment analysis

The eligible samples were stratified into low- and high- expression groups of PDGFA according to the median value. The R packages “clusterProfiler” (v4.12.6) and “org.Hs.eg.db” were used for the Gene Ontology (GO) functional enrichment analysis, Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway analysis according to the previous description (14). GO results were categorized into three categories: molecular function (MF), biological process (BP), and cellular component (CC). To study the biological pathways of PDGFA, the correlation between PDGFA and other genes in HNSCC expression data was calculated. Gene Set Enrichment Analysis (GSEA), a computational method used to determine whether a predefined set of genes shows significant differences between two biological states (15-17), was performed using the C2 and C5 datasets from the MSIGDB database. The statistical thresholds for significance were set as follows: |Normalized Enrichment Score (NES)| >1, nominal P value (NOM P)<0.05, and false discovery rate adjusted Q-value (FDR Q-value) <0.25.

Statistical analysis

Groups differences were assessed using the Student’s t-test, the nonparametric Mann-Whitney test, or one-way analysis of variance (ANOVA) followed by Bonferroni correction. OS comparison was compared by KM analysis with the log-rank test. Statistical significance was defined as P<0.05.

Results

Screening of prognosis-related PDGF/PDGFR genes

Univariate Cox regression was performed to identify prognostic genes within the PDGF and PDGFR families (including PDGFA, PDGFB, PDGFC, PDGFD, PDGFRA, and PDGFRB). Among these, PDGFA [hazard ratio (HR) =1.472, 95% confidence interval (CI): 1.106–1.959, P=0.008] and PDGFC (HR =1.273, 95% CI: 1.038–1.562, P=0.02) was significantly associated with the HNSCC prognosis (Table 2). Based on the median expression levels of PDGFA and PDGFC, KM survival curves were plotted. High PDGFA expression was correlated with a poor prognosis (P=0.01), while there was no significant difference in OS between the high and low expression groups of PDGFC (P=0.29) (Figure 1A,1B). Therefore, PDGFA was selected for further analysis. The association between PDGFA expression and prognosis was validated in the GSE41613 dataset (P=0.046, Figure 1C). Differential expression of PDGFA was further confirmed in the GSE9844 dataset (Figure 1D). To further investigate its clinical relevance, PDGFA expression was analyzed across tumor or node stages in HNSCC patients. The results revealed significant differences in PDGFA expression between several tumor stages (Figure 1E,1F). However, no significant variation was detected across node stages.

Figure 1 Screening of prognosis-related genes. (A) KM survival curve of PDGFA in TCGA dataset. (B) KM survival curve of PDGFC in TCGA dataset. (C) KM survival curve of PDGFA in GSE41613 dataset. (D) Validation of differential expression of PDGFA in GSE9844 dataset. (E,F) Relationship between PDGFA gene expression and tumor and node in HNSCC patients. ns, not significant; *, P<0.05; **, P<0.01; ***, P<0.001. HNSCC, head and neck squamous cell carcinoma; KM, Kaplan-Meier; N, node; OTSCC, oropharyngeal squamous cell carcinoma; PDGFA, platelet-derived growth factor A; T, tumor; TCGA, The Cancer Genome Atlas.

Construction and validation of nomogram

In the multivariate model, PDGFA expression (HR =1.43, 95% CI: 1.07–1.92, P=0.02) and N2_N3 stage (HR =2.00, 95% CI: 1.19–3.35, P=0.008) were independent prognostic factors (Table S1). Stepwise Cox regression, based on the AIC, was used to select variables. The optimal prognostic model, which included age, PDGFA, tumor, and node, achieved the lowest AIC (666.41) and a higher C-index (0.695) (Table S2). Although some variables were not statistically significant in the multivariate Cox analysis, they contributed to the overall model fit and predictive performance. Based on the results from both univariate and multivariate Cox regression analyses, as well as AIC-based model selection, a prognostic nomogram incorporating PDGFA, age, tumor, and node was developed (Figure 2A). The prognostic index (PI) was calculated as:

PI=0.0582×tumor(T2)+0.7365×tumor(T3)+0.5590×tumor (T4)−0.2074×tumor(TX)−0.0906×node(N1)+0.6114×node(N2_N3)+0.7224×node(NX)+0.0220×age+0.2518×PDGFA[1]

Figure 2 Establishment and validation of nomogram. (A) Nomogram. (B) Calibration curves for the 2-, 3- and 5-year OS. (C) Diagnostic ROC curve of PDGFA in HNSCC. (D) Calibration curve validation in GSE65858 dataset. (E) Diagnostic ROC curve validation in GSE65858 dataset. AUC, area under the curve; CI, confidence interval; HNSCC, head and neck squamous cell carcinoma; N, node; OS, overall survival; PDGFA, platelet-derived growth factor A; Pr, probability; ROC, receiver operating characteristic; T, tumor.

Detailed data are shown in supplementary table 5 (available at https://cdn.amegroups.cn/static/public/tcr-2025-1-2907-1.xlsx). The estimated baseline survival probabilities were 0.67 at 2 years, 0.59 at 3 years, and 0.47 at 5 years. The individual survival probability at time t was calculated as:

S(t|X)=S0(t)exp(PI)[2]

In the training set, the model demonstrated a C-index of 0.695 (95% CI: 0.625–0.765). Calibration curves for 2-, 3-, and 5-year survival showed close concordance between predicted and observed outcomes (Figure 2B). ROC curve analysis (Figure 2C) yielded AUC values of 0.726, 0.709, and 0.604 for 2-, 3-, and 5-year OS, respectively, indicating that the model has moderate discriminative ability, particularly in predicting 2- and 3-year prognosis, though its performance for 5-year outcomes was limited.

In the external validation cohort (GSE65858), the model exhibited a C-index of 0.607 (95% CI: 0.561–0.654), consistent with the training set results. Both calibration curves (Figure 2D) and ROC curves (Figure 2E) for 2-, 3-, and 5-year OS suggested favorable predictive performance, with corresponding AUC values of 0.621, 0.665, and 0.589. Furthermore, DCA indicated potential clinical utility of the model (Figure S1).

Mutation analysis and functional pathway enrichment analysis

Somatic mutation analysis revealed that TP53, OBSCN, OR2T3, ZNF107, and ADGRB3 exhibited higher mutation frequencies in the PDGFA high expression group [Figure 3A, supplementary table 6 (available at https://cdn.amegroups.cn/static/public/tcr-2025-1-2907-1.xlsx)].

Figure 3 Mutation analysis and functional pathway enrichment analysis. (A) Somatic mutation analysis. (B) Differentially expressed genes between low and high PDGFA expression groups. (C) GO analysis. (D) GSEA analysis of the top ten enrichment results of C5. (E) GSEA analysis. (F) GSEA analysis of the top ten enrichment results of C2. **, P<0.01; ***, P<0.001. BP, biological process; CC, cellular component; GO, Gene Ontology; GSEA, Gene Set Enrichment Analysis; MF, molecular function; OR, odds ratio; PDGFA, platelet-derived growth factor A.

A number of differentially expressed genes (DEGs) were identified when comparing the low and high PDGFA expression groups (Figure 3B). These DEGs were further examined through GO and KEGG analysis (Figure 3C). GO analysis indicated that BPs were enriched in external encapsulation structural tissue, extracellular matrix (ECM) tissue, extracellular structural tissue, epidermal development, and skin development. CC were observed in the collagen-containing ECM, endoplasmic reticulum cavity, basement membrane, cell-matrix junction, and lesion adhesion. MF were enriched in ECM structural components, ECM binding, growth factor binding, ECM structural components that impart tensile strength, and collagen binding (supplementary table 7 available at https://cdn.amegroups.cn/static/public/tcr-2025-1-2907-1.xlsx). KEGG pathway analysis visualization highlighted the top ten enriched pathways (Figure 3E), including cytoskeleton in muscle cells, cornified envelope formation, lesion adhesion, ECM receptor interaction, proteoglycans in cancer, protein digestion and absorption, human papillomavirus infection, PI3K-Akt signaling pathway, amebiasis, and small cell lung cancer in diabetic complications (supplementary table 8 available at https://cdn.amegroups.cn/static/public/tcr-2025-1-2907-1.xlsx).

GSEA revealed that the top ten enriched BPs in C5 [Figure 3D, supplementary table 9 (available at https://cdn.amegroups.cn/static/public/tcr-2025-1-2907-1.xlsx)] included positive regulation of natural killer (NK) cell mediated cytotoxicity, positive regulation of NK cell-mediated immunity, assembly of peptide antigen and major histocompatibility complex (MHC) class II protein complex, NK cell-mediated immune regulation, cornified mantle, T cell receptor complex in cell components, structural components of the skin epidermis, and CC chemokine binding. The top ten enriched results from C2 [Figure 3F, supplementary table 10 (available at https://cdn.amegroups.cn/static/public/tcr-2025-1-2907-1.xlsx)] included the anti-tumor necrosis factor (TNF) therapy non-responder post treatment up, breast cancer cluster 1, graft versus host disease, β-defensins, NK cells, response to oncocytic virus and cyclic RGD, defensins, tumorigenesis downregulation, antigen processing and presentation by MHC class II molecules, and primary immunodeficiency. It should be noted that these enriched pathways represent potential biological associations inferred from transcriptomic data, rather than experimentally validated mechanisms.

Discussion

PDGFA overexpression have been reported to correlate with poorer survival and other adverse clinical outcomes across multiple malignancies. However, its specific roles in the prognosis of HNSCC remains incompletely defined. In this study, we demonstrated that PDGFA is associated with the poor prognosis of HNSCC and explored its underlying mechanisms.

Importantly, we constructed and externally validated a nomogram incorporating PDGFA expression, age, tumor, and node. While Wang et al. (18) also constructed a nomogram for HNSC, incorporating multiple clinical variables and PDGFA, but without external validation. In contrast, our model was validated using the independent cohort GSE65858, confirming its robust predictive performance and potential clinical application. Compared with traditional staging systems, the model in this study integrates clinical features and molecular markers, improving the accuracy and individualization of prognostic prediction. Traditional staging systems [such as the tumor-node-metastasis (TNM) classification] often rely solely on anatomical information and fail to reflect the molecular heterogeneity and multifactorial influences of patients. By incorporating the PDGFA and multidimensional clinical variables, the nomogram not only improves discrimination and calibration consistency but also demonstrates a higher net benefit in DCA, highlighting its potential for application in clinical risk stratification and personalized treatment decision-making.

Mechanistically, we integrate somatic mutation analysis and functional enrichment of DEGs, suggest that PDGFA may contribute to HNSCC progression through a multi-level regulatory network encompassing somatic mutation, TME remodeling, immune modulation, and oncogenic signaling. Patients with high PDGFA expression exhibited increased mutation frequencies in the TP53, OBSCN, OR2T3, ZNF107 and ADGRB3. TP53 mutations serve as an independent predictor of poor disease-specific survival in HNSCC (19), OBSCN and ADGRB3 may influence cell survival and differentiation (20,21). However, the direct association between increased mutation frequencies of OR2T3 and ZNF107 genes and HNSCC has not been previously reported in the literature. Our study demonstrated that these genes are crucial for the development of HNSCC, suggesting that high PDGFA expression may be linked to the mutation status of these genes, thereby affecting tumor biological behavior and patient prognosis.

Functional enrichment analysis of DEGs between low and high PDGFA expression groups revealed significant involvement in ECM, immune-related processes, cancer-related BPs, cell structure, and tissue development. In cancer, ECM dysregulation stimulates malignant cell transformation (22), thereby promoting tumorigenesis and proliferation. Previous research has demonstrated that the ECM plays an important role in the development of HNSCC. Collagen in ECM proteins affects cell adhesion and proliferation, thereby promoting cancer progression (23,24). Based on this, we speculate that PDGFA may provide a more favorable microenvironment for tumor cell migration by altering the composition and structure of the ECM.

Immune response is pivotal in determining cancer treatment outcomes. In our study, the DEGs also enriched in pathways related to the positive regulation of NK cell-mediated immunity and MHC-II molecules. NK cells are innate immune effector cells that can initiate adaptive anti-tumor immune responses and directly eliminate tumor cells (25). It has been reported that HNSCC-derived cytokines can inhibit NK cell function by suppressing agonist receptor signaling or enhancing inhibitor receptor signaling, allowing HNSCC to evade immune surveillance (26). MHC-II is an important mediator in the presentation of tumor-associated antigens (27), which can triggered a Th1 response and activated CD4⁺ T-cell expansion, suppressing HNSCC growth in a CD4⁺ T-cell-dependent manner (28). Specifically, high PDGFA expression could potentially reduce NK cell cytotoxic activity and interfere with antigen presentation via MHC-II, thereby limiting tumor recognition by CD4⁺ T cells. These findings support a role for PDGFA in establishing an immunosuppressive TME, which may further promote tumor progression and impact patient survival.

The PI3K/AKT pathway plays a crucial role in cancer development (29). Xu et al. demonstrated that PDGFA/PDGFRα-regulated Golgi membrane protein 1 might promote glioma progression by activating the key signaling kinase AKT (30). Numerous studies have shown that dysregulation of the PI3K/AKT pathway influences the pathogenesis of squamous cell carcinoma and patient survival (31). Although no studies have yet shown that PDGFA affects the development of HNSCC by regulating the PI3K/AKT signaling pathway, our study provides preliminary evidence for this potential mechanism.

Collectively, these results support a conceptual model in which high PDGFA expression regulates somatic mutation, TME remodeling, immune response, and PI3K/AKT signaling activation to promote HNSCC progression and poor prognosis (Figure 4). However, the causal relationships between PDGFA expression and these BPs have not yet been fully established, representing a significant gap in current knowledge. Therefore, further in vitro and in vivo studies are required to validate these associations and to elucidate the underlying mechanisms by which PDGFA shapes tumor progression and the immune microenvironment in HNSCC.

Figure 4 Schematic of the hypothesized mechanism. ECM, extracellular matrix; HNSCC, head and neck squamous cell carcinoma; MHC, major histocompatibility complex; PDGFA, platelet-derived growth factor A.

There are several limitations in this study. First, TCGA-HNSCC cohort encompasses multiple subsites, while validation cohort differ in subsite composition and experimental platforms, which may affect generalizability. Second, the availability and missingness of some clinical variables varied across cohorts (e.g., M stage was frequently missing), which may affect the comparability and stability of the combined model. Finally, this study relied entirely on publicly available databases for analysis, lacking experimental validation. Further confirmation of our findings through in vitro and in vivo experiments is required.

Conclusions

High PDGFA expression was significantly correlated with poorer prognosis in HNSCC. Notably, we constructed a nomogram incorporating PDGFA, age, tumor, and node. Furthermore, our analysis suggested that high PDGFA expression may be involved in somatic mutation, TME remodeling, immune response, and PI3K/AKT signaling activation, thereby contributing to HNSCC progression and poor prognosis. These findings support PDGFA as a potential prognostic biomarker and therapeutic target. Future studies are warranted to further elucidate its context-dependent roles within the TME and assess its potential in combination with immunotherapy.

Acknowledgments

None.

Footnote

Reporting Checklist: The authors have completed the TRIPOD reporting checklist. Available at https://tcr.amegroups.com/article/view/10.21037/tcr-2025-1-2907/rc

Peer Review File: Available at https://tcr.amegroups.com/article/view/10.21037/tcr-2025-1-2907/prf

Funding: This study was supported by the fund from Sanya Central Hospital (The Third People’s Hospital of Hainan Province) (grant No. SYZXYY202303).

Conflicts of Interest: Both authors have completed the ICMJE uniform disclosure form (available at https://tcr.amegroups.com/article/view/10.21037/tcr-2025-1-2907/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. The 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. Guo Z, Li K, Ren X, et al. The role of the tumor microenvironment in HNSCC resistance and targeted therapy. Front Immunol 2025;16:1554835. [Crossref] [PubMed]
3. Contrera KJ, Reddy PD, Helou V, et al. The Evolution of Multidisciplinary Head and Neck Cancer Treatment. Laryngoscope 2025;135:2255-60. [Crossref] [PubMed]
4. Li CX, Gong ZC, Zhang WN, et al. Radioresistance or/and radiosensitivity of head and neck squamous cell carcinoma: biological angle. Oral Maxillofac Surg 2024;28:547-55. [Crossref] [PubMed]
5. Jung SC, Kang D, Ko EA. Roles of PDGF/PDGFR signaling in various organs. Korean J Physiol Pharmacol 2025;29:139-55. [Crossref] [PubMed]
6. Zou X, Tang XY, Qu ZY, et al. Targeting the PDGF/PDGFR signaling pathway for cancer therapy: A review. Int J Biol Macromol 2022;202:539-57. [Crossref] [PubMed]
7. Han N, Zhang YY, Zhang ZM, et al. High expression of PDGFA predicts poor prognosis of esophageal squamous cell carcinoma. Medicine (Baltimore) 2021;100:e25932. [Crossref] [PubMed]
8. Wu YY, Yu LS, Zhou HY, et al. Effect of HepG2-Derived Exosome with PDGF-D Knockdown on Transformation of Normal Fibroblasts into Tumor-Associated Fibroblasts in Liver Cancer. Front Biosci (Landmark Ed) 2025;30:26045. [Crossref] [PubMed]
9. Li Y, Zhao Y, Shi M, et al. PDGF‐D Promotes Epithelial–Mesenchymal Transition of Glioma Cells Through the NF-κB/NOTCH1 Pathway. Cancer Med 2025;14:e71002.
10. Huang XL, Khan MI, Wang J, et al. Role of receptor tyrosine kinases mediated signal transduction pathways in tumor growth and angiogenesis-New insight and futuristic vision. Int J Biol Macromol 2021;180:739-52. [Crossref] [PubMed]
11. Wang S, Meng B, Liu Y, et al. Identification of PDGFRB as a prognostic immune-related biomarker in gastric cancer through bioinformatics and clinical analysis. Transl Cancer Res 2025;14:6849-63. [Crossref] [PubMed]
12. Yan M, Qi X, Xie W, et al. PDGF receptor-β-targeted copper-gadolinium-oxide self-assembled nanoclusters suppress tumor growth and metastasis via copper overload-mediated apoptosis and anti-angiogenesis. Chem Eng J 2023;457:141227.
13. Mayakonda A, Lin DC, Assenov Y, et al. Maftools: efficient and comprehensive analysis of somatic variants in cancer. Genome Res 2018;28:1747-56. [Crossref] [PubMed]
14. Yu G, Wang LG, Han Y, et al. clusterProfiler: an R package for comparing biological themes among gene clusters. OMICS 2012;16:284-7. [Crossref] [PubMed]
15. Liberzon A, Birger C, Thorvaldsdóttir H, et al. The Molecular Signatures Database (MSigDB) hallmark gene set collection. Cell Syst 2015;1:417-25. [Crossref] [PubMed]
16. Liberzon A, Subramanian A, Pinchback R, et al. Molecular signatures database (MSigDB) 3.0. Bioinformatics 2011;27:1739-40. [Crossref] [PubMed]
17. Subramanian A, Tamayo P, Mootha VK, et al. Gene set enrichment analysis: a knowledge-based approach for interpreting genome-wide expression profiles. Proc Natl Acad Sci U S A 2005;102:15545-50. [Crossref] [PubMed]
18. Wang K, Li C, Chen H, et al. Increased Expression of PDGFA Is Associated With Poor Prognosis and Immune Infiltration in Head and Neck Squamous Cell Carcinoma. J Oral Pathol Med 2024;53:584-94. [Crossref] [PubMed]
19. Zilberg C, Lee MW, Yu B, et al. Analysis of clinically relevant somatic mutations in high-risk head and neck cutaneous squamous cell carcinoma. Mod Pathol 2018;31:275-87. [Crossref] [PubMed]
20. Rajendran BK, Deng CX. A comprehensive genomic meta-analysis identifies confirmatory role of OBSCN gene in breast tumorigenesis. Oncotarget 2017;8:102263-76. [Crossref] [PubMed]
21. Melis M, Zhang T, Scognamiglio T, et al. Mutations in long-lived epithelial stem cells and their clonal progeny in pre-malignant lesions and in oral squamous cell carcinoma. Carcinogenesis 2020;41:1553-64. [Crossref] [PubMed]
22. Zhang X, Al-Danakh A, Zhu X, et al. Insights into the mechanisms, regulation, and therapeutic implications of extracellular matrix stiffness in cancer. Bioeng Transl Med 2025;10:e10698. [Crossref] [PubMed]
23. Wang Y, Zhang F, Qian Z, et al. Targeting collagen to optimize cancer immunotherapy. Exp Hematol Oncol 2025;14:101. [Crossref] [PubMed]
24. García-Palmero I, Torres S, Bartolomé RA, et al. Twist1-induced activation of human fibroblasts promotes matrix stiffness by upregulating palladin and collagen α1(VI). Oncogene 2016;35:5224-36. [Crossref] [PubMed]
25. Charap AJ, Enokida T, Brody R, et al. Landscape of natural killer cell activity in head and neck squamous cell carcinoma. J Immunother Cancer 2020;8:e001523. [Crossref] [PubMed]
26. Dong S, Zhao M, Zhu J, et al. Natural killer cells: a future star for immunotherapy of head and neck squamous cell carcinoma. Front Immunol 2024;15:1442673. [Crossref] [PubMed]
27. Gameiro SF, Ghasemi F, Barrett JW, et al. High Level Expression of MHC-II in HPV+ Head and Neck Cancers Suggests that Tumor Epithelial Cells Serve an Important Role as Accessory Antigen Presenting Cells. Cancers (Basel) 2019;11:1129. [Crossref] [PubMed]
28. Zhang Y, Li J, Guo X, et al. Tumor-Specific MHC-II Activates CD4+ and CD8+ T Cells in Head and Neck Squamous Cell Carcinoma to Boost Immunotherapy Efficacy. Cancer Res 2025;85:3258-74. [Crossref] [PubMed]
29. Jiang N, Dai Q, Su X, et al. Role of PI3K/AKT pathway in cancer: the framework of malignant behavior. Mol Biol Rep 2020;47:4587-629. [Crossref] [PubMed]
30. Xu R, Ji J, Zhang X, et al. PDGFA/PDGFRα-regulated GOLM1 promotes human glioma progression through activation of AKT. J Exp Clin Cancer Res 2017;36:193. [Crossref] [PubMed]
31. Ghafouri-Fard S, Noie Alamdari A, Noee Alamdari Y, et al. Role of PI3K/AKT pathway in squamous cell carcinoma with an especial focus on head and neck cancers. Cancer Cell Int 2022;22:254. [Crossref] [PubMed]