Introduction

Head and neck cancer (HNC) is the seventh most common cancer worldwide, affecting about 453 000 deaths [1], of which head and neck squamous cell carcinoma (HNSCC) is the most common type. Despite rapid advances in treatment for HNC, the 5-year overall survival rate of patients remains at about 50% [2]. In addition, the classic clinical factors including lymph node metastasis and histological grade are not sufficient to predict the prognosis of patients, owing to the heterogeneity of molecular mechanisms and tumor behaviors related to HNSCC [3]. Therefore, identifying clinically relevant biomarkers for HNSCC can effectively improve the prognosis of patients. However, an accurate biomarker for the prognosis of HNSCC patients is still lacking in the clinic.

Immunotherapies eradicate cancerous cells by enhancing immune system activity [4]. Immune checkpoint inhibitors (ICIs) are an effective type of immunotherapies that block inhibitory immune checkpoint pathways to enhance anti-tumor immune activity [5]. In 2016, the US Food and Drug Administration (FDA) granted the first immunotherapeutic approval, for nivolumab and pembrolizumab, as anti-PD-1 immune checkpoint inhibitors, for the treatment of patients with recurrent HNSCC that is refractory to platinum-based regimens. The European Commission followed in 2017 with approval of nivolumab for the treatment of the same patient population, and shortly thereafter with the approval of pembrolizumab monotherapy for the treatment of recurrent or metastatic HNSCC in adults whose tumors express PD-L1 with a ≥ 50% tumor proportion score and have progressed on or after platinum-containing chemotherapy. Then in 2019, the FDA approved PD-1 inhibition as the first-line treatment for patients with metastatic or unresectable, recurrent HNSCC [6]. Despite the successful application of immunotherapy across a broad range of human cancers, the majority of patients have limited or no response to these therapies [7]. Therefore, excavating a predictive biomarker to assess the response to immunotherapies for defining patient benefit early is a desperate need. However, accurate biomarkers for predicting clinical outcomes and immunotherapy responses continue to be largely unexplored.

The Endoplasmic reticulum (ER) is an organelle widely present in eukaryotic cells that regulates protein synthesis, folding, and aggregation after synthesis. An aggregation of misfolded proteins in the lumen and an imbalance of Ca2 + in the cytoplasm caused by various factors can lead to ER dysfunction, which can induce a series of related protein expression and cell phenotype changes, a condition called endoplasmic reticulum stress (ERS) [8]. ERS has been proven to usually play an essential role in cell function and survival [9]. A previous study showed that ceramide synthase 1 could promote the aggressiveness of oral squamous cell carcinoma and chemotherapeutic drug resistance through ERS [10]. Furthermore, the expression of PERK-ATF4 has been indicated to be positively correlated with VEGF [11]. Thus, we can find that ERS may be associated with the prognosis of HNSCC and has enormous potential to predict it. However, there has been no report on ERS-related biomarkers for clinical outcomes and immunotherapy responses of HNSCC patients.

Therefore, in this study, we identify and validate an ERS-related signature for predicting the prognosis of HNSCC patients. The ERS-related signature showed a significant correlation with tumor immunity in Gene Set Enrichment Analysis (GSEA) analysis. Moreover, the infiltration of naive B cells and CD8 + T cells are significantly elevated in patients with low-risk scores compared to those with high-risk scores. By contrast, macrophages and activated mast cells were remarkably enhanced in high-risk groups compared to low-risk groups. Furthermore, we also find that the ERS-related signature can be a potential biomarker for immunotherapy response in HNSCC. This is pioneering research to identify an ERS-related signature that provides further insights for the prediction of prognosis and immunotherapy response of HNSCC patients. Moreover, a detailed analysis of the cellular immune response in HNSCC has the potential to enhance clinical and immunotherapy response prediction.

Materials and methods

Data and resources

Gene expression data and associated clinical characteristics of HNSCC patients were downloaded from the Cancer Genome Atlas (TCGA, http://cancergenome.nih.gov/). We downloaded RNA-Seq data expressed as transcripts per million (TPM) from TCGA database. R package ‘edgeR’ was utilized to normalize and process the data by using R version 4.0.4 software. This cohort has 487 HNSCC patients with the corresponding gene expression data and clinical information (Table 1).

Table 1 Clinical characteristics of HNSCC patients in TCGA database
Full size table

Microarray datasets including gene expression profiles and corresponding clinical information data of GSE65858 were downloaded from the Gene Expression Omnibus database (GEO, https://www.ncbi.nlm.nih.gov/geo/). GSE65858 was conducted by GPL10558 ( Illumina HumanHT-12 V4.0 expression beadchip), including 270 HNSCC samples which were involved in this work as a validating set.

Construction and confirmation of a prognostic signature

In view of the essential role of ERS in cell function and survival [9], we dedicated to study ERS-related genes in HNSCC. An ERS-related gene set was defined based on Molecular Signatures Database (MSigDB, https://www.gsea-msigdb.org/gsea/msigdb/index.jsp), including GO:0036500 [12], GO:0036498 [12] and R-HSA-380,994 gene sets. These gene lists were displayed in Table 2. Then, to exploit genes associated with overall survival of HNSCC patients, we analyzed these genes by using univariate Cox regression analysis. Genes with P < 0.05 were further applied to multivariate Cox regression analysis. Thus, a prognostic signature of overall survival which was correlated with ERS was developed for HNSCC.

Table 2 The list of gene sets
Full size table

Next, a risk score was calculated for HNSCC patients according to expression levels of the genes (expi) and coefficients of multivariate Cox regression analysis (bi). Then, patients were split into high-risk and low-risk groups in the TCGA and GEO cohorts by median risk score values, respectively. The formula used was as follows:

$$\text{R}\text{i}\text{s}\text{k}\text{s}\text{c}\text{o}\text{r}\text{e}=\sum _{i=1}^{n}expi*bi$$

Evaluation of endoplasmic reticulum stress score and check-point score

ERS score and check-point score were calculated based on the single-sample gene-set enrichment analysis (ssGSEA) [13] using the ERS-related gene sets and check-point gene sets (Table 2) to quantify the expression levels of these genes for HNSCC patients. We estimated the ERS score and check-point score between low- and high- risk groups of HNSCC.

Survival analysis

We compared the overall survival (OS), recurrence-free survival (RFS), disease-free survival (DSS) and progression-free survival (PFS) of HNSCC patients in low- and high- risk groups. Kaplan-Meier (K-M) curves were carried out to compare the difference in prognosis. P values from Log-rank tests were calculated, and less than 0.05 was considered statistically significant.

Gene set enrichment analysis

To explore the potential biological functions of the ERS-related genes between low- and high- risk groups in HNSCC, we conducted Gene Set Enrichment Analysis (GSEA) based on the curated gene sets c5.go.bp.v7.4.symbols. Normalized P-value < 0.05 was considered statistically significant.

Evaluation of tumor-infiltrating immune cells

Normalized gene expression data were used to estimate the relative proportions of the infiltrating immune cells using the CIBERSORT algorithm [14]. Gene expression datasets were prepared using standard annotation files and data uploaded to the CIBERSORT web portal (https://cibersort.stanford.edu), with the algorithm run using the default signature matrix at 1,000 permutations. Here, we applied the original CIBERSORT gene signature file LM22 and analyzed gene profiles of HNSCC patients from TCGA database grouped by risk score. Subsequently, we selected 410 samples (consisted of 201 low-risk samples and 209 high-risk samples) which met the criteria of CIBERSORT P value < 0.05.

Statistical analysis

The expression profiles of mRNAs from TCGA and GEO database were shown as raw data and each mRNA was normalized by log2 transformation for further analysis. We plotted ROC curves using the R package ‘survivalROC’, and the area under the ROC curve (AUC) was used to assess the prognostic performance of the ERS-related signature. Statistical analysis was performed by using GraphPad Prism version 7.0 or SPSS version 19.0 software package. Significant difference between two various groups was using Mann–Whitney test. Chi-squared test was applied to calculate the difference of clinicopathologic factors between low- and high- risk groups in HNSCC samples from TCGA dataset. P < 0.05 was considered statistically significant. We calculated the correlation between two variables using the Spearman method. The threshold of P < 0.05 (Spearman’s correlation test) indicates the significance of correlation. A two-tailed P < 0.05 was considered statistically significant.

Results

The activity of ERS is significantly different between tumor and normal samples in HNSCC

In this study, the datasets of 502 HNSCC samples and 44 normal samples were obtained from the TCGA database. Two samples without a pathological stage, three samples without tumor grade information, and ten samples with OS less than 30 days were excluded (Fig. 1). To estimate the ERS activity of normal and tumor samples, we employed the ssGSEA algorithm using a list of ERS-related genes. The results showed that there was a significant difference in ERS activity between tumor and control samples in HNSCC (P < 0.0001, Fig. 2A). Moreover, tumor samples displayed higher ERS activity compared to normal samples. Then, we further analyzed the expression levels of 98 ERS-related genes in normal and tumor samples and found that there were 28 genes differently expressed between normal and tumor samples (P < 0.05, Fig. 2B; Table 3). These results suggest that ERS may be associated with the prognosis of HNSCC.

Fig. 1
figure 1

The patient flow diagram

Full size image
Fig. 2
figure 2

Activity of ERS is significantly different between tumor and normal samples in HNSCC. (A) Significant difference of ERS scores between normal and tumor samples in HNSCC. P-values are based on the Mann–Whitney test. (B) Heat map showing the expression levels of the 28 ERS-related genes differently expressed between normal and tumor samples in HNSCC from the TCGA.

Full size image
Table 3 The difference of the expression levels of 98 ERS-related genes between normal and tumor samples
Full size table

Identification of an ERS-related prognostic signature for HNSCC patients

To exploit the prognosis-related genes from the 98 ERS-related genes in HNSCC, we first utilized univariate Cox regression analysis. Thus, 19 genes related to OS of HNSCC were obtained (P < 0.05, Fig. 3A; Table 4). Finally, as a result of multivariate Cox regression analysis, a six-mRNA (ASNS, EXOSC6, BAK1, TPP1, EXOSC8, and TATDN2) prognostic model was developed to predict the prognosis of HNSCC. Among these six genes, the expression of ASNS and EXOSC8 has a marked difference between normal and tumor samples in HNSCC (P < 0.05, Figure S1). Hence, based on these six genes, the HNSCC risk score system was established as follows: risk score= (0.6624 × expression value of ASNS) + (1.2499× expression value of EXOSC6) + (0.9704 × expression value of BAK1) + (1.1852 × expression value of TPP1) + (0.7885 × expression value of EXOSC8) + (-2.0666 × expression value of TATDN2).

Fig. 3
figure 3

Construction of the ERS-related prognostic signature for OS of HNSCC in the TCGA dataset. (A) 19 genes related to OS of HNSCC were obtained by utilizing univariate Cox regression. (B) Kaplan–Meier curves of OS of low- and high- risk groups stratified by the risk score in HNSCC. (C) The distributions of risk scores and patient survival status for HNSCC. (D) ROC curve for the 5-year OS prediction by the ERS-related signature. (E) Kaplan–Meier curves of RFS, DSS and PFS of low- and high- risk groups stratified by the risk score in HNSCC, respectively. (F) ROC curves for the 5-year RFS, DSS and PFS prediction by the ERS-related signature, respectively

Full size image
Table 4 The list of 19 genes related to OS of HNSCC in univariate Cox regression analysis
Full size table

According to this formula, each patient has endowed a risk score. Then, 500 patients in the TCGA dataset were classified into high-risk and low-risk groups by the cut-off value of 1.058 as the median value of the risk score. Kaplan-Meier (K-M) curves showed that patients in high-risk groups tended to have poorer clinical outcomes compared to those in low-risk groups (Log-rank P = 0.0002, Fig. 3B). The distribution of gene risk scores and patients’ survival status of 500 HNSCC patients showed that patients’ mortality rate was augmented in the high-risk group compared to those in the low-risk group (Fig. 3C). Furthermore, the AUC value of 5-year OS was 0.687, which illustrated a medium accuracy of the ERS-related signature in OS prediction in HNSCC (Fig. 3D).

Given the predictive value of the signature for OS, we are curious whether the signature can also predict RFS, DFS, and PFS in HNSCC. Surprisingly, K-M curves illustrated that patients in high-risk groups had significantly shorter RFS, DFS, and PFS when compared with those in low-risk groups (RFS: Log-rank P = 0.0160, DFS: Log-rank P = 0.0001, PFS: Log-rank P = 0.0007, Fig. 3E). In addition, AUC values of 5-year RFS, DFS, and PFS were 0.623, 0.711, and 0.678, respectively, implying medium prediction performance of our signature for RFS, DFS, and PFS of HNSCC patients (Fig. 3F).

The correlation of the ERS-related signature with clinical characteristics of HNSCC patients

Correlation analysis indicated that survival status, T, cancer status, perineural invasion present, and treatment response were significantly associated with the risk score based on the ERS-related signature (P < 0.05, Fig. 4A; Table 5). The strip chart showed that the risk of patient mortality rate elevated, while the T staging gradually rose and the treatment response rate markedly decreased as the risk score ascended (Fig. 3A). The chi-squared test demonstrated that the high-risk group significantly tended to high probability of death (P < 0.0001, Fig. 4B), older (P < 0.0001, Fig. 3B), survival with tumor (P = 0.0404, Fig. 4B) and no response to treatment (P = 0.0014, Fig. 4B) compared to the low-risk group of HNSCC. These results suggested that highly malignant HNSCC were associated with high-risk scores, and our risk score system based on the ERS-related signature had tremendous potential to predict prognosis for HNSCC patients.

Fig. 4
figure 4

The relationship of the ERS-related signature with clinical characteristics of HNSCC patients. (A) The strip chart of risk score and clinical characteristics for patients with HNSCC in TCGA datatset. (B) Pie charts showing the Chi-squared test of clinicopathologic factors for low- and high- risk groups in HNSCC samples from TCGA dataset

Full size image
Table 5 The correlation of the ERS-related signature with clinical characteristics of HNSCC.
Full size table

The ERS-related signature is robust for predicting OS of HNSCC patients

To validate the predictive ability of the ERS-related signature, we applied the signature to the GSE65858 dataset. In this set, we used the same risk score model to calculate each patient’s risk score. Then, 270 patients were divided into high- and low-risk groups using the median of the risk scores value of 3.053. K-M curves implied that a higher risk score was associated with poorer prognosis in HNSCC, which was consistent with our previous findings (P = 0.0266, Fig. 5A). Moreover, the distributions of risk scores and patients’ survival status of 270 HNSCC patients showed that patients’ mortality rate increased as the risk score was elevated in HNSCC patients (Fig. 5B). The AUC value of 5-year OS was 0.684, implying a medium power of the signature in OS prediction of HNSCC (Fig. 5C).

Fig. 5
figure 5

Validation of the ERS-related prognostic signature in GEO dataset. (A) Kaplan–Meier curves of OS of low- and high- risk groups stratified by the risk scores in HNSCC in GEO dataset. (C) The distributions of risk scores and patient survival status for HNSCC in TCGA dataset. (D) ROC curve for the 5-year OS prediction by the ERS-related signature in GEO dataset

Full size image

The ERS-related signature is an independently prognostic biomarker for HNSCC patients

To investigate the independence of the prognostic signature, Cox regression analysis was carried out combined clinical characteristics with our ERS-related signature in the TCGA dataset. Firstly, univariate Cox regression analysis indicated that the risk score, age, N, M, stage, grade, cancer status, and perineural invasion present were prognostic factors for OS of HNSCC patients (P < 0.05, Fig. 6A). Moreover, multivariate Cox regression analysis showed that the ERS-related signature, cancer status, and perineural invasion present were independent predictors for OS, after adjusting to other four clinical factors (P < 0.05, Fig. 6B). Taken together, the above results suggest that the ERS-related signature is an independent adverse prognostic factor for OS of HNSCC patients, which relates to poor clinical outcomes.

Fig. 6
figure 6

The ERS-related signature is an independent prognostic factor in HNSCC patients. Univariate (A) and multivariate (B) Cox regression of prognosis factor for OS of HNSCC patients. (C) Kaplan–Meier analysis of OS for HNSCC patients stratified by sex, age, ethnicity, T, stage, grade, cancer status and lymphnode neck dissection status

Full size image

Furthermore, stratified analyses based on these clinical characteristics were carried out to identify suitable patient groups for the prediction of the risk score system. The results showed that the risk score remained an independent prognostic factor for the subgroups of female, younger than 61, not Hispanic or Latino, T3 and T4, stage III and IV, grade 1 and 2, tumor-free and lymphnode neck dissection (P < 0.05, Fig. 6C). Moreover, the results of stratification analysis showed that high-risk patients in each stratum of those clinical parameters had significantly shorter OS than low-risk patients.

The ERS-related signature is associated with tumor immunity in HNSCC

To illustrate the potential biological functions of the ERS-related signature in HNSCC, we conducted a GSEA analysis based on the curated gene sets c5.go.bp.v7.4.symbols. We found that negative regulation of ubiquitin-dependent protein catabolic process was upregulated in low-risk score groups, which validated that our signature was ERS-associated (Fig. 7A). Moreover, some immune-associated pathways that upregulated in low-risk score groups in HNSCC, such as response to interleukin 4 and B cell receptor signaling pathway, imply our ERS-associated signature might be associated with immune surveillance against cancer cells.

Fig. 7
figure 7

The ERS-related signature shows strongly association with immune surveillance against cancer cells in HNSCC. (A) The biological functions of the ERS-related signature in HNSCC using GSEA analysis based on the curated gene sets c5.go.bp.v7.4.symbols. (B) The difference of immune cell infiltration between low- and high- risk groups based on the ERS-related signature in HNSCC using CIBERSORT algorithm. P-values are based on the Mann–Whitney test. * P < 0.05, **P < 0.01, ***P < 0.001

Full size image

A large body of research has proved that ERS usually affected many physiological processes, including immune function [15]. Our results also show that the ERS-related signature is associated with immune surveillance against cancer cells in HNSCC. Therefore, to further illustrate the potential mechanism of the correlation between the ERS-related signature and immune surveillance against cancer cells, we applied the CIBERSORT algorithm to the TCGA dataset. Excluding samples which CIBERSORT P-value > 0.05, 410 samples from the TCGA dataset were enrolled, including 201 low-risk samples and 209 high-risk samples. The analysis of immune cells infiltrations between low-risk and high-risk groups showed that the infiltrations of naive B cells, CD8 + T cells, follicular helper T cells, and plasma cells were significantly higher in low-risk groups than in high-risk groups (P < 0.05, Fig. 7B). However, macrophages and activated mast cells were remarkably enhanced in high-risk groups compared to those in low-risk groups (P < 0.05, Fig. 7B).

The ERS-related signature has the potential to predict immunotherapy response

Given the association between the ERS-related signature and immune surveillance against cancer cells, we investigated whether the ERS-related signature also correlated with immunotherapy response. Therefore, we first calculated immune check-point scores using the expression levels of immune checkpoint genes (the list of genes shown in Table 2) for each patient by ssGSEA algorithm and measured the association between the risk score and immune check-point score. Strikingly, we observed that the risk score was significantly negatively correlated with the immune check-point score in HNSCC (P < 0.0001, R=-0.2034, Fig. 8A). The expression patterns of immune check-point genes were further compared between the high- and low-risk groups stratified by the ERS-related signature. And the results revealed that patients with high-risk scores exhibited significantly lower expression levels of immune checkpoint genes than those with low-risk score exhibited (P = 0.0002, Fig. 8B).

Fig. 8
figure 8

The ERS-related signature is a potential predictor for immunotherapy response in HNSCC. (A) The correlation of the ERS-related signature with the expression levels of check-point genes in HNSC from the TCGA. R: Spearman’s correlation coefficient. (B) The expression levels of check-point genes among HNSCC samples grouped by risk score in TCGA dataset. Figure S1. Different expression levels of ASNS (A) and EXOSC8 (B) between normal and tumor samples in HNSCC

Full size image

Discussion

HNSCC is a multisite malignancy characterized by a low cure rate and easy recurrence [16]. Recent studies have indicated that HNSCC is one of the few malignant tumors that respond well to immunotherapy [6]. Despite the initial enthusiasm, the clinical benefits of immunotherapy in HNSCC patients are overall limited [17]. Therefore, it is important to exploit a biomarker that can predict the response and durability of clinical benefits of immunotherapy to realize better tailor treatments for individual patients. However, biomarkers that can accurately predict prognosis and immunotherapy in HNSCC are still lacking. ERS has been indicated to play a key role in the process of anti-tumor immune response mediation [15]. Targeting ERS-related components may be a promising way of improving the efficacy of cancer immunotherapies [18], suggests potential of ERS for predicting response to cancer immunotherapies. However, there has been no research about ERS as a biomarker for prognosis and cancer immunotherapies in HNSCC.

Therefore, in this work, we first measured the ERS activity in normal and tumor samples using the ssGSEA algorithm and found that the activity of ERS was significantly different between tumor and normal samples of HNSCC. These data hint to us that the ERS activity may be associated with HNSCC. Subsequently, Cox regression analysis was carried out and we mined six ERS-related genes (ASNS, EXOSC6, BAK1, TPP1, EXOSC8, and TATDN2) that exhibited significant prognostic value for OS in HNSCC patients. K-M curves (Log-rank P = 0.0002) and ROC curves (AUC of 0.687) also indicated that these genes had a great value for predicting 5-year OS for HNSCC. Surprisingly, we found that the ERS-related signature could also predict RFS, DSS, and PFS of HNSCC patients, and ROC curves indicated high prediction performance for RFS, DSS, and PFS in HNSCC.

Further correlation analysis of the ERS-related signature with clinical characteristics of HNSCC displayed that the risk of patient mortality rate increased, while the T staging gradually rose and the treatment response rate markedly decreased as the risk score increased, which were in consistence with our previous results that the ERS-related signature is a risk factor for HNSCC prognosis and is associated with grim clinical outcomes.

In this study, we developed and validated an ERS-related prognostic signature (ASNS, EXOSC6, BAK1, TPP1, EXOSC8, and TATDN2) for HNSCC. Asparagine synthetase (ASNS) has been proven to promote cell proliferation and tumor growth in multiple cancers [19, 20]. Moreover, ASNS is also associated with cancer metastasis [21]. The researchers indicated that circulating tumor cells could survive insults such as hypoxia and nutrient deprivation as the result of upregulation of factors such as ATF4, ATF3, and ASNS which were beneficial for cancer cell survival once they detach from the primary tumor and enter the bloodstream. Exosome Component 6 (EXOSC6) derived from intracellular proteins has either well-defined or putative roles in breast cancer development and progression [22]. BAK1 is reported as a prognosis-related gene for HNSCC [23, 24]. Tripeptidyl-peptidase 1 (TPP1) is overexpressed in hepatocellular carcinoma (HCC) tissues and significantly correlated with poor prognosis [25]. Moreover, in larynx squamous cell carcinoma, the expression of TPP1 is elevated, and TPP1 may be a prognostic marker for response to radiotherapy for patients [26]. The oncogenic roles of Exosome Component 8 (EXOSC8) have been confirmed in colorectal carcinoma [27] and prostate cancer [28]. Bin Yu et al. find that the TatD DNase domain containing 2 (TATDN2) is related to early recurrence and has the potential for predicting the prognosis of hepatocellular carcinoma [29]. These previous data are consistence with our results that the six ERS-related genes are related to the prognosis of HNSCC, and have predicted potential for the survival of patients with HNSCC.

Tao Qu et al. developed a prognostic signature consisting of 10 ERS-related LncRNAs for HNSCC [30], which contained huge numbers of RNA, resulting in higher costs and great expenses for clinical applications. In contrast, our signature has a lower cost and is more applicable. Moreover, Tao Qu`s signature was proven to predict OS for HNSCC patients, while our prognostic signature not only can predict OS but also showed predictive power for RFS, DSS, and PFS.

To further unveil the potential mechanism of the correlation of the ERS-related signature with the prognosis of HNSCC, GSEA analysis was employed in this work. The results showed that negative regulation of ubiquitin-dependent protein catabolic process was upregulated in low-risk score groups, suggesting that our signature associated with ERS was robust. Surprisingly, we also found that some immune-associated pathways upregulated in low-risk score groups in HNSCC, such as response to interleukin 4 and B cell receptor signaling pathway, implying our ERS-associated signature might be associated with immune surveillance against cancer cells. Previous research has proved that ERS usually plays a vital role in mediating immune surveillance against cancer cells [15]. Results of GSEA analysis also show that the ERS-related signature is associated with tumor immunity in HNSCC.

Therefore, to further illustrate the potential mechanism of the correlation between the ERS-related signature and immune surveillance against cancer cells, we applied the CIBERSORT algorithm to TCGA dataset. The results indicated that the infiltration of naive B cells, CD8 + T cells, follicular helper T cells, and plasma cells were significantly higher in low-risk groups than in high-risk groups. However, macrophages and activated mast cells were remarkably elevated in high-risk groups than in low-risk groups. The functions of, and interactions between, the innate and adaptive immune systems are vital for anti-tumor immunity. Naive B cells undergo robust proliferation and differentiation that can result in the production of anti-tumor factors such as memory B cells through appropriate stimulation [31, 32]. CD8 + T cells are the most powerful effectors in the anti-tumor immune response and form the backbone of current successful cancer immunotherapies [33]. However, mast cells are proven to support pro-cancerous immune cell infiltrations which induced tumor progression [34, 35]. This research is consistent with our results that the high-risk score is related to low anti-tumor immunity and results in a poor outcome clinical in HNSCC.

Given the association between the ERS-related signature and immune surveillance against cancer cells, we further investigate the correlation between our ERS-related signature and immunotherapy response. Surprisingly, we found that the risk score was significantly negatively correlated with the immune check-point score in HNSCC. Moreover, patients with high-risk scores exhibited significantly lower expression levels of immune checkpoint genes compared with those in the low groups. These results suggest that our ERS-related signature is correlated with immunotherapy response and can be a potential biomarker for immunotherapy response. Of course, the association between the ERS-related signature and immunotherapy response revealed in this study needs to be validated in clinical.

Conclusion

In conclusion, in this study, we mine an independent and accurate ERS-related signature that can predict OS, RFS, DFS, and PFS of HNSCC patients. GSEA analysis unveils that the ERS-related signature is strongly associated with tumor immunity. Moreover, the infiltration of naive B cells and CD8 + T cells are significantly elevated in patients with low-risk scores compared to those with the high-risk score. By contrast, macrophages and activated mast cells are remarkably enhanced in high-risk groups compared to those in low-risk groups. Furthermore, we find that the ERS-related signature can also be a potential biomarker for immunotherapy response in HNSCC. To the best of our knowledge, this is the first study to analyze the predictive potential of ERS-related genes for prognosis and immunotherapy response in HNSCC, highlighting the impact of ERS-related genes on immune response, potentially enabling more precise and personalized immunotherapy response in the future.