Multi-omics analysis of pyroptosis-related genes for prognosis and immune landscape in head and neck cancer

Abstract

Dear Editor,

Despite the demonstrated efficacy of immunotherapy in various cancers, treating head and neck squamous cell carcinoma (HNSCC) continues to pose significant challenges.^{1, 2} Pyroptosis, a distinct form of programmed cell death, is intricately associated with tumour progression and immune response modulation.^{3, 4} This study undertakes a comprehensive multi-omics analysis to elucidate the complex role of pyroptosis-related genes (PRGs) in the context of HNSCC, with the objective of developing a robust prognostic signature that could substantially advance the understanding of the prognosis of HNSCC and its associated immune landscape.

Figure 1A provides a comprehensive overview of the study's workflow, delineating the principal steps and methodologies employed in our investigation. The study encompasses 528 cancer samples and 44 normal controls from the TCGA database, along with 270 cancer samples from the GEO database. We identified 64 PRGs, of which 51 were differentially expressed in HNSCC tissues (Figure S1A). Survival analysis showed that 33 of these genes were linked to patient outcomes (Figure S2). A prognostic network was developed to elucidate the interrelationships among these genes (Figure 1B). Analysis revealed that 409 of 510 samples had PRG mutations, an 80.2% mutation rate (Figure S1B). Additionally, PRGs often showed copy number variations (CNVs), with gains or losses illustrated in Figure S1C, and their chromosomal distribution was shown in Figure S1D.

Hierarchical clustering analysis identified two clusters in HNSCC (Figure 1C,D), with cluster B showing a significantly better prognosis than cluster A (Figure 1E). The clinical characteristics and PRGs expression profiles associated with these subtypes are presented in Figure 1F. Gene Set Variation Analysis (GSVA) and Gene Set Enrichment Analysis (GSEA) analyses demonstrated that cluster B was associated with immune-related pathways, while cluster A was enriched in metabolic pathways (Figure 1G,H). This observation is further corroborated by ssGSEA, which revealed a higher degree of immune cell infiltration within cluster B (Figure 1I).

We further identified 717 differentially expressed genes (DEGs) related to pyroptosis subtypes (Figure 2A), with 169 DEGs significantly affecting prognosis (Figure S3A). The results of the enrichment analysis for the DEGs were presented in Figure S3B. We further performed a clustering analysis and found that k = 3 was optimal (Figure 2B,C). Notably, patients in group C had a better prognosis than those in other groups (Figure 2D). Moreover, there is a notable overlap in clinical traits and DEG expression between geneCluster group C and PRGCluster cluster B (Figure 2E). To develop a novel prognostic signature for HNSCC, randomly selected patients were assigned to a training cohort for signature development and a validation cohort for evaluation. Through the application of Least Absolute Shrinkage and Selection Operator (LASSO) regression and multivariate Cox regression analyses, seven key DEGs were identified as essential for the construction of the prognostic signature (Figure 2F,G and Table S1). The signature's gene expression, risk score differentiation, prognoses for high- and low-risk groups and prediction accuracy were consistent across the training group (Figure 2H–K), validation group (Figure 2L–O) and the overall cohort (Figure 2P–S). Both univariate and multivariate analyses, along with the concordance index (C-index) curves, substantiated that the prognostic signature provided superior predictive effect for the survival of HNSCC patients compared to other clinical characteristics (Figure 2T,U). A nomogram was developed to estimate survival rates across different follow-up periods, utilising clinical features and risk scores (Figure 2V). The calibration curve revealed an excellent agreement between the survival probabilities forecasted by the nomogram and the actual patient outcomes, signifying a high level of predictive precision (Figure 2W). Additionally, cluster B in PRGCluster and group C in geneCluster were linked to better prognoses and lower risk scores (Figure S4A–C), which also exhibited higher expression of PRGs (Figure S4D,E).

Immune cell infiltration analysis identified significant associations between immune cells, risk scores and signature genes (Figures 3A–C and S5). The high-risk group exhibited reduced immune function and lower expression of immune checkpoint genes, suggesting greater immune evasion and poor response to immunotherapy (Figure 3D,E). Tumour microenvironment (TME) scores,⁵ tumour immune dysfunction and exclusion (TIDE) scores⁶ and immunophenoscore (IPS)⁷ collectively indicated that the low-risk group had better immune infiltration and responses, while the high-risk group showed greater immune evasion, potentially reducing its response to immune checkpoint blockade (ICB; Figure 3F–H). As well as gene mutation frequency, we assessed tumour mutational burden (TMB) between high- and low-risk groups. Our results indicated that the TP53 gene showed a high mutation frequency in both groups (Figure 3I). Furthermore, the high-risk cohort exhibited an increased TMB level, indicating a potential link between elevated TMB and heightened genomic instability, which may be associated with a less favourable prognosis (Figure 3J,K).

Among the candidate genes, Transglutaminase (TGM2) was selected for experimental validation due to its uncharacterised role in HNSCC. Our observations revealed a significant upregulation of TGM2 in HNSCC tissues, which was associated with adverse prognostic outcomes (Figure 4A,B). Additionally, single-cell RNA sequencing data⁸ indicated that TGM2 is predominantly expressed in mast cells and monocytes/macrophages within HNSCC (Figure S7). To substantiate these findings, we employed TGM2-siRNAs to achieve TGM2 knockdown, and the efficiency of the siRNAs was verified (Figure 4C). The suppression of TGM2 expression markedly reduced the proliferation (Figure 4D–F), migration and invasion abilities of HNSCC cells (Figure 4G–J), while promoting cell death (Figure 4K,L) and inhibiting epithelial–mesenchymal transition (EMT; Figure 4M), thereby elucidating the oncogenic function of TGM2. Furthermore, utilising the GSCA platform,⁹ we conducted an analysis to explore the correlation between Genomics of Drug Sensitivity in Cancer (GDSC) pharmacological agents and TGM2 mRNA levels (Figure 4N). Subsequently, a Venn analysis was conducted, integrating these data with drug sensitivity information derived from the risk signature (Figure S6) and TGM (Figure S8), which led to the identification of two potential therapeutic agents: Dasatinib and WH-4-023 (Figure 4O). Finally, AutoDocktools was employed to examine the docking interactions between TGM2 and these two compounds (Figure 4P).

To conclude, the current study clarifies the significance of PRGs in the prognosis of HNSCC by developing a prognostic signature that may improve the prediction of patient survival and identifying TGM2 as a potential therapeutic target, thereby providing insights into the immune landscape of HNSCC. We believe that these findings have significant practical implications for enhancing patient management and informing the development of novel therapeutic strategies for HNSCC.

Shikang Zheng conducted the bioinformatic analysis and drafted the original manuscript. Qinghua Liu and Cheng Wang made significant contributions to experimental operation and data acquisition. Rongqi Zhang and Xin Peng were responsible for resources and supervision. Junda Fan and Haiming Xu handled software and visualisation. Xiazhi Pan and Nanxiang Chen were in charge of data validation and performed quality control checks. Mingbo Liu and Kai Zhao conducted the conception, funding acquisition and manuscript revision. All the authors made substantial contributions to the article and approved the final version for publication.

The authors declare no conflicts of interest.