Sage Journals: Discover world-class research

Abstract

Objective

Numerous studies indicate that immune-related genes (IRGs) are closely related to tumorigenesis and tumor progression. We aimed to establish a robust IRGs-based signature to predict risk of recurrence for patients with laryngeal squamous cell carcinoma (LSCC).

Methods

Gene expression profiles were acquired to select differentially expressed IRGs (DEIRGs) between tumor and adjacent normal tissues. Functional enrichment analysis was performed to explore the biological roles of DEIRGs in LSCC. Univariate Cox analyses and LASSO regression model were used to construct a IRGs-based signature with the ability to predict recurrence for LSCC patients.

Results

A total of 272 DEIRGs were identified, of which 20 DEIRGs were significantly associated with recurrence-free survival (RFS). Subsequently, we constructed an eleven-IRGs signature that could classify patients into high-risk or low-risk groups in TCGA-LSCC training cohort. Patients in high-risk groups suffered from shorter RFS (log-rank p = 9.69E−06). Besides, the recurrence rate of high-risk group was significantly higher than that of low-risk group (41.1% vs. 13.7%; Fisher’s exact test p < 0.001). The predictive performance was validated in an independent cohort (GSE27020, log-rank p = 1.43E-03). Person correlation analysis showed that the risk scores calculated by eleven-IRGs signature were significantly associated with filtrating immune cells. Furthermore, three immune checkpoint molecules were significantly over-expressed in the high-risk group.

Conclusion

Our findings for the first time constructed a robust IRGs-based signature to precisely predict risk of recurrence and further provided a deeper understanding of IRGs regulatory mechanism in LSCC pathogenesis.

Keywords

Laryngeal squamous cell carcinoma immune-related genes recurrence-free survival signature

Introduction

Laryngeal carcinoma (LC) is a prevalent and severe malignancy, which is representative of >20% of all head and neck cancer.^1,2 Laryngeal squamous cell carcinoma (LSCC) is currently the most common pathological classification of LC.³ Currently, a variety of treatments, such as surgery, radiation therapy, and chemotherapy, are available to improve clinical outcome for LSCC patients.^4,5 Despite advances in therapies, nearly half of LSCC patients experience disease recurrence caused by chemotherapy resistance and high invasiveness.⁶ For clinicians, it is a difficult task to predict recurrence and survival risk as tumor heterogeneity and diverse clinical behaviors. Therefore, a valid and convenient predictive model is urgently needed to improve recurrence risk prediction.

Tumor microenvironment (TME) plays a critical regulatory role in cancer progression and treatment responses. As the essential components of the tumor microenvironment, tumor-infiltrating immune cells have been validated to be involved in tumor growth, invasiveness, and metastasis. Many researches have reported that pronounced immunosuppressive tumor activity is closely related to tumorigenesis. In addition, immunodeficiency also affects recurrence and prognosis.^7,8 Recently, cancer immunotherapy has emerged as an effective way to improve clinical outcomes.^9-11 For example, targeting programmed death protein (PD-1) and programmed death ligand 1 (PD-L1) with antibodies against PD-1 rescues effector T cell function, maintaining tumor cell-killing capacity of T cell.¹² For LSCC patients with high expression of PD-L1, antibodies against PD-1 are efficacious in reducing the risk of recurrence and improving clinical outcome.^13-15 Therefore, exploring the roles of immune-related genes (IRGs) in LSCC progression is beneficial to discovering more effective immunotherapy targets.

With the development of chip sequencing technology, many studies have explored the function of IRGs using gene expression data. Gao et al. found four IRGs with a higher prognostic performance in renal clear cell carcinoma. Prognostic scores calculated by four IRGs effectively reflect the level of immune cell infiltration.¹⁶ Xu et al. comprehensively analyzed the expression data and clinical data and then developed an individualized immune signature to improve prognostic evaluation for bladder cancer patients.¹⁷ Many studies have confirmed that IRGs as key regulators were involved in the occurrence, metastasis, diagnosis, and treatment of LSCC.^18,19 Mo et al. analyzed the role of IRGs to develop an immune-related prognostic signature, which can make risk classification for LSCC patients.²⁰ IRGs signature has gradually become an effective method to explore the relationship between immune cells and tumor progression. However, few signatures based on IRGs are developed to predict the risk of recurrence of patients with LSCC. Therefore, developing a IRGs-based signature to estimate risk of recurrence is beneficial to understanding the potential predictive value of tumor-infiltrating immune cells and exploring new immunotherapy targets.

In this study, using gene expression profiles and clinical information of LSCC extracted from TCGA database, we comprehensively analyzed the expression of IRGs in LSCC and its potential predictive value of recurrence. We further constructed a IRGs-based signature to predict the risk of recurrence in LSCC patients. Besides, we analyzed the association between risk scores calculated by IRGs-based signature and LSCC immune cell infiltration, and further evaluated the correlation with three immune checkpoint molecules. Our study explored the potential clinical utility of IRGs on predicting risk of recurrence and laid a foundation for potential LSCC immunotherapy targets.

Materials and methods

Data acquisition and processing

This study was retrospective in nature. All data in this study was obtained from public database. We systematically searched LSCC-related array datasets from the public databases since 2014. The LSCC microarray datasets were retrieved from Gene Expression Ominibus (GEO) database (https://www.ncbi.nlm.nih.gov/geo/) with the following criteria: (1) laryngeal squamous cell carcinoma; (2) the number of patients ≥50; (3) with more than 12,000 protein coding genes. Finally, GSE127165 (n = 114) containing 57 paired tumor and adjacent normal tissues were retrieved. As the training cohort, RNA-sequencing data and clinical data of 102 LSCC samples were downloaded from TCGA database (available at: https://portal.gdc.cancer.gov/). Besides, an external validation datasets (GSE27020), containing 107 samples with both expression data and clinical data, were extracted from GEO database to test the prognostic ability. We selected patients with LSCC according to anatomic subdivision and extracted clinical features (including age, gender, grade, stage, relapse-free survival, relapse status, and vital status). All patients underwent surgical removal of the tumor. Then, patients with the missing values in any clinical feature were removed. Clinical information is described in Table 1. For gene expression profiles from GEO database, background correction and normalization were performed using the “affy” package of R/Bioconductor. For RNA-sequencing data from TCGA database, fragments per kilobase of exon per million reads mapped (FPKM) were extracted as gene expression measurements.

Table 1.

Clinical characteristics of patients with laryngeal squamous cell carcinoma.

	TCGA-LSCC	GSE27020
Sample
Normal	—	—
Tumor	102	107
Mean age (years; range)	65 (39–83)	63 (41–88)
Gender
Male	84	—
Female	18	—
Grade
G1	8	42
G2	64	49
G3	27	16
GX	3	0
Stage
I	2	12
II	11	18
III	18	36
IV	71	43
Tumor status
With tumor	30	34
Tumor-free	72	73
Vital status
Alive	70	-
Dead	32	-
Platform	Illumina HiSeqV2	Affymetrix U133A

Differentially expressed IRGs (DEIRGs) analysis

We derived a list of immune-related genes (IRGs) from the Immunology Database and Analysis Portal (ImmPort) Database. Student t-test was applied to select the DEIRGs between the tumor and normal tissue samples. The threshold was set to false discovery rate (FDR) < 0.05 and |log2fold change (FC)|>1. Volcano plots and heatmaps were generated using the “ggplot2” and “pheatmap” packages in R software, respectively.

Functional enrichment analysis and protein–protein interaction network

In order to investigate the biological processes that the DEIRGs may be involved in, functional enrichment analysis were performed using Database for Annotation, Visualization and Integrated Discovery (DAVID, available at: http://www.david.abcc.ncifcrf.gov/). Besides, to illustrate the key cellular activities in carcinogenesis, a protein–protein interaction (PPI) network was built by The Search Tool for the Retrieval of Interacting Genes/Proteins (STRING, available at: https://string-db.org/). A strict criterion (combined score >0.9) was set to identify reliable interactions among the DEIRGs. The PPI network was visualized using Cytoscape (available at: www.cytoscape.org). Then, hub genes with high degrees of connectivity were selected by cytoHubba application.

Establishment of the IRGs-based signature for LSCC

DEIRGs, which were significantly associated with relapse-free survival (RFS) using univariate Cox regression analysis, were selected to establish the prognostic signature by the least absolute shrinkage and selection operator (LASSO) method in TCGA-LSCC training cohort. The Lasso-penalized Cox model is a popular method to achieve the feature selection and obtain an optimal model subsequently. Regression coefficients of each DEIRG in the model were determined by Lasso-penalized Cox model. The risk scoring model is:

Risk score = ∑ ⁿ _i=0G_iβ_iG_i is the normalized expression value of gene i in risk scoring model. β_i is the regression coefficient of gene i, which represents the contribution of gene i to risk score. Based on the risk scoring model, patients whose risk scores were larger than the median risk score were classified into high-risk group, otherwise classified into low-risk group.

Evaluation of relationship between IRGs-based signature and immune cell infiltration, and immune checkpoint molecules

Tumor Immune Estimation Resource (TIMER, available at: https://cistrome.shinyapps.io/timer/) is an online tool to analyze and visualize the abundances of tumor-infiltrating immune cells. Based on gene expression data, the algorithm estimated the abundance of six subtypes of tumor-infiltrating immune cells (including B cells, CD4 T cells, CD8 T cells, macrophages, neutrophils, and dendritic cells) for each sample from TCGA public database. Estimation results of immune cell infiltration abundance for LSCC patients were retrieved from TIMER to analyze the correlation between immune cell infiltration and risk scores. Meanwhile, Estimation of Stromal and Immune cells in Malignant Tumor tissues using Expression data (ESTIMATE) was a method to evaluate the infiltration level of stromal and immune cells for samples from TCGA.²¹ We extracted the immune scores and stromal scores of LSCC samples to calculate the difference between high-risk group and low-risk group, respectively. To investigate the correlation between risk scores and three important immune checkpoint molecules (including PD-1, PD-L1, and CTLA4), we performed T-test to observe the expression difference of three immune checkpoint molecules between high-risk and low-risk groups, respectively.

Statistical analysis

Statistical analysis for all data in the current study was implemented by R software. T-test was used to observe the gene expression difference between different groups. Multivariable Cox regression analysis was used to assess whether the risk score was independent of other clinical features, such as stage, age, and gender. Hazard ratios (HRs) and 95% confidence intervals (CIs) were computed based on the Cox regression analysis. The Kaplan–Meier method was used to assess the differences in survival time of low- and high-risk LSCC patients, and the log-rank test was used to determine the statistical significance of observed differences between groups. The receiver operating characteristic (ROC) curve was generated using the “pROC” package. The difference in mortality rate between different risk groups was tested by Fisher’s exact test. p < 0.05 was regarded as statistically significant.

Results

Identification of differentially expressed IRGs

Using a threshold of |log2FC| >1 and FDR <0.05, 3267 differentially expressed genes were identified between LSCC and adjacent normal tissues in GSE127165 (Figure 1(a)). Then, a list of IRGs containing 1811 genes were downloaded from ImmPort database, of which 272 IRGs were significantly differentially expressed (including 143 up-regulated IRGs and 129 down-regulated IRGs) and filtrated for further analysis (Figure 1(b)).

Figure 1.

Volcano plots of differentially expressed genes (A) and heat maps of differentially expressed immune-related genes (B).

Functional enrichment of differentially expressed IRGs

Using the DAVID database, we performed functional enrichment analysis (including GO annotation and KEGG pathway enrichment) to explore the underlying biological processes related to 272 DEIRGs. Functional enrichment results are shown in Figure 2. GO functional annotation revealed that DEIRGs were significantly involved in many tumor-related terms (Figure 2(a)). As expected, inflammatory pathways were most frequently enriched. “immune response,” “extracellular region,” and “cytokine activity” were the most frequent biological terms among biological processes, cellular components, and molecular functions, respectively. KEGG pathway enrichment indicated that 272 DEIRGs were mainly enriched in “Cytokine-cytokine receptor interaction,” “Chemokine signaling pathway,” and “Jak-STAT signaling pathway” (Figure 2(b)).

Figure 2.

Enrichment of Gene Ontology (GO) terms (A) and Kyoto Encyclopedia of Genes and Genomes (KEGG) pathways (B) of differentially expressed immune-related genes.

Protein–protein interaction network

With a strict criterion (combined score >0.9), a PPI network consisting of 94 nodes and 166 edges was constructed using STRING online tool to explore the interactions between 272 DEIRGs. The PPI network was visualized using Cytoscape software (Figure 3(a)). The top 10 IRGs were generated by cytoHubba, including IL1B, PIK3R1, CSF2, STAT1, IL12A,IL1A, CCL22, CXCL10, OASL, and RSAD2 (Figure 3(b)). We found that most genes were closely involved in tumorigenesis and progression.

Figure 3.

The protein-protein interaction network constructed by differentially expressed immune-related genes (A) and top 10 immune-related genes generated by cytoHubba (B).

Development and validation of the IRGs-based signature

First, 20 DEIRGs that were significantly related with RFS were identified by performing the univariate Cox regression in TCGA-LSCC cohort (Table 2). Based on predictive ability of these genes for LSCC, we obtained the optimal model containing eleven significant DEIRGs through multiple cross-validation in TCGA-LSCC training cohort using LASSO method. Therefore, an eleven-IRGs signature was constructed to guide the prediction of recurrence for LSCC patients (Table 3). The risk score formula was created as follows: Risk Score = (0.5365*expression value of PSMD2) + (0.1994*expression value of ULBP1) + (−0.1374*expression value of LCN2) + (−0.0963*expression value of PTGDS) + (0.3473*expression value of NEDD4) + (0.0020*expression value of SEMA3C) + (0.2101*expression value of DKK1) + (0.2292*expression value of GAL) + (−1.0828*expression value of IL12 A) +(-0.9937*expression value of NRTN) + (0.1432*expression value of IL1RAP). Based on the eleven-IRGs signature, risk scores of patients in TCGA-LSCC cohort were calculated. Then, we divided patients into a high-risk (n = 51) or a low-risk (n = 50) group according to median risk score and observed the RFS difference between different risk groups. The results showed that patients in high-risk group suffered significantly shorter RFS (HR: 6.14, 95% CI: 2.47–15.26, log-rank p = 9.69E−06; Figure 4(a)). The recurrence rate of high-risk group was significantly higher than that of low-risk group (41.1% vs. 13.7%; Fisher’s exact test p < 0.001; Figure 4(b)). However, there was no difference in mortality between high-risk and low-risk groups (Fisher’s exact test p = 0.286; Figure 4(b)). Furthermore, the multivariate Cox regression analysis showed that the eleven-IRGs signature could serve as an independent prognostic factor for LSCC patients after adjusting for age, gender, grade, and stage in the TCGA-LSCC training cohort (HR: 3.96, 95% CI: 2.43–6.46, p < 0.001, Figure 4(c)). The risk score distribution, survival status, and expression profiles of eleven IRGs are shown in Figure 4(d). The predictive performance of eleven-IRGs signature was tested using an independent cohort extracted from GEO database. In GSE27020 cohort, Patients with high-risk scores had shorter RFS than those with low-risk scores (HR: 2.92, 95% CI: 1.35–6.30, log-rank p = 1.43E-03; Figure 5(a)). We observed that the recurrence rate was also significantly higher in the high-risk group (44.4% vs. 18.2%; Fisher’s exact test p = 0.004; Figure 5(b)). Multivariate Cox regression results confirmed the independent prognostic factor for LSCC (HR = 2.21, 95% CI = 1.35–3.57, p = 0.002, Figure 5(c)). The risk score distribution, survival status, and expression profiles of eleven IRGs are shown in Figure 5(d). Besides, we calculated the AUC values of ROC curve in TCGA-LSCC and GSE27020 datasets. The AUC values were 0.813 and 0.704 in TCGA-LSCC and GSE27020 datasets, respectively.

Table 2.

Univariate cox analysis of DEIRGs.

DEIRGs	HR	95% CI: Low	95% CI: High	p-value	DEdir
PSMD2	4.98	2.11	11.74	0.0002	1
ULBP3	1.65	1.02	2.69	0.0427	1
ULBP1	1.45	1.01	2.09	0.0442	1
LCN2	0.81	0.68	0.97	0.0214	−1
PTGDS	0.67	0.48	0.95	0.0248	−1
NEDD4	3.07	1.38	6.81	0.0058	1
CCL19	0.76	0.60	0.97	0.0286	−1
CCR3	4.50	1.03	19.66	0.0454	1
SEMA3C	1.41	1.02	1.94	0.0364	1
BMP2	1.57	1.12	2.19	0.0084	1
CGB7	4.69	1.01	21.87	0.0491	1
DKK1	1.30	1.03	1.63	0.0240	1
GAL	1.32	1.04	1.67	0.0236	1
IL12 A	0.12	0.02	0.87	0.0357	−1
MSTN	8.86E-06	1.09E-10	0.72	0.0437	−1
NRG1	1.52	1.00	2.32	0.0492	1
NRTN	0.31	0.15	0.63	0.0012	−1
STC2	1.53	1.10	2.13	0.0112	1
IL1RAP	2.35	1.47	3.75	0.0003	1
IL31RA	2.38	1.29	4.39	0.0054	1

Note: DEIRGs, differentially expressed immune-related genes; HR, hazard ratio; CI, confidence interval; DEdir, differentially expressed direction.

Table 3.

Gene information entering into the signature.

Gene-symbol	Entrez ID	Affy probe ID	Coefficients
PSMD2	5708	200830_at	0.5365
ULBP1	80,329	221323_at	0.1994
LCN2	3934	212531_at	−0.1374
PTGDS	5730	211663_x_at/211748_x_at/212187_x_at	−0.0963
NEDD4	4734	213012_at	0.3473
SEMA3C	10,512	203788_s_at/203789_s_at	0.0020
DKK1	22,943	204602_at	0.2101
GAL	51,083	214240_at	0.2292
IL12 A	3592	207160_at	−1.0828
NRTN	4902	210683_at	−0.9937
IL1RAP	3556	205227_at/210233_at	0.1432

Figure 4.

Construction of the eleven-IRGs signature in TCGA-LSCC training cohort. (A) Kaplan–Meier curve of the recurrence-free survival between low-risk and high-risk groups. (B) The ratio of death and recurrence in high-risk and low-risk groups, respectively. (C) Forest plot of the multivariate Cox regression analysis. (D) Risk score distribution, survival status and expression heat map of eleven genes corresponding to each sample above.

Figure 5.

Predictive performance of the eleven-IRGs signature in validation cohort (GSE27020). (A) Kaplan-Meier curve of the recurrence-free survival between low-risk and high-risk groups in GSE27020. (B) The ratio of recurrence in high-risk and low-risk groups. (C) Forest plot of the multivariate Cox regression analysis. (D) Risk score distribution, survival status and expression heat map of eleven genes corresponding to each sample above.

Relationship of the IRGs-based signature with immune cell infiltration and immune checkpoint molecules

We utilized the TIMER database to reveal the association between eleven-IRGs signature and filtrating immune cells. As shown in Figure 6, eleven-IRGs risk scores were significantly negatively correlated with levels of infiltrating B cell (r = −0.41, p < 0.01), CD4 + T cell (r = −0.26, p = 0.04) and CD8 + T cell (r = −0.33, p = 0.01). Meanwhile, Using ESTIMATE algorithm to evaluate the infiltration level of stromal and immune cells, we observed that patients with higher risk scores tended to be lower immune scores but higher stromal scores (Figures 7(a)–(b)). Besides, the expression values of three immune checkpoint molecules (including PD-1, PD-L1 and CTLA4) were compared between high-risk group and low-risk group. We found that patients within high-risk group presented with a lower expression level for all three immune checkpoint molecules (Figures 7(c)-(e)).

Figure 6.

The association between risk scores based on eleven-IRGs signature and immune cell infiltration in TCGA-LSCC patients.

Figure 7.

Comparison of immune score, stromal score, and expression of three immune checkpoint molecules between high-risk and low-risk groups. (A) Immune score. (B) Stromal score. (C) PD-1. (D) PD-L1. (E) CTLA4.

Discussion

Growing evidences have confirmed the contributions of tumor microenvironment to human diseases.^22-24 Recently, many researches have revealed that IRGs could act as a key regulator in LSCC progression.^25-27 Many efforts have been made to elaborate the clinical utility of IRGs on prognostic stratification for LSCC patients,^20,28 but few researches explored the predictive ability of IRGs for the risk of recurrence. In this study, we identified 272 DEIRGs between LSCC and adjacent normal tissues. Functional enrichment analysis results showed that DEIRGs were most frequently involved in inflammatory biological processes. Besides, other cancer-associated processes, such as “positive regulation of cell proliferation,” “G protein-coupled receptor signaling pathway,” and “signal transduction” were enriched. For example, Thomas et al. found that two G protein-coupled receptors ligands, including prostaglandin E2 and bradykinin, could activate the epidermal growth factor receptor pathway by increasing selective autocrine release of transforming growth factor-α, which plays key roles in LSCC carcinogenesis.²⁹ KEGG enrichment analysis also showed that DEIRGs were significantly enriched in “Cytokine-cytokine receptor interaction,” “Chemokine signaling pathway,” and “Jak-STAT signaling pathway.” Furthermore, ten hub IRGs were selected from PPI network. All genes have been reported to be closely related to the progression and prognosis of various tumors.^30-33 For example, PIK3R1, known as the regulatory subunit of class 1A PI3Ks, has been identified to be involved in cancer progression. Wang et al.³⁴ observed that PIK3R1 was down-regulated in LSCC tissues and cell lines. By directly binding to 3′-untranslated region (UTR), MiR-17-5p could positively regulate PIK3R1 expression to modulate the activation of PI3K/AKT signaling pathway. These researches suggested that IRGs could play critical roles in LSCC development, including proliferation, apoptosis, migration, and invasion.

Several risk-prediction models have been constructed in recent years.^35,36 For example, Cui et al.³⁷ constructed a risk-prediction nomogram for predicting recurrence probability based on eight clinicopathologic variables. Nevertheless, the risk-prediction nomogram on the basis of clinical and pathologic variables exists limitations due to tumour heterogeneity of LSCC. With the development of molecular biology and high-throughput sequencing, molecular factors may provide in-depth understanding of the mechanism of LSCC progression and improve the predict recurrence probability of LSCC. In our study, we focused on IRGs closely related to patient RFS and then developed a risk-prediction model consisting of eleven IRGs to precisely predict recurrence probability. Among the eleven IRGs, Dickkopf-1 (DKK1) is a secreted protein and Wnt signaling pathway inhibitor. Researches have revealed that DDK1 is over-expressed in LSCC tissues and can be served as an unfavorable survival predictor, which may be a promising therapeutic target for LSCC patients.^38,39 Another IRG, namely, interleukin 1 receptor accessory protein (IL1RAP), acts as an IL1R co-receptor to mediate the activation of IL-1 responsive genes.⁴⁰ In recent years, many researchers have found that IL1RAP as a key tumor-associated antigen might be used as an immunotherapy target. Pasvenskaite et al.⁴¹ found significant IL1RAP polymorphisms in LSCC carcinogenesis, which could be act as predictors for the clinical course of LSCC. Although other IRGs, such as PSMD2, LCN2, NEDD4, and IL12A, have not clearly reported related to the progression of LSCC, a large number of research studies have found that these IRGs play critical roles in cancer progression.^42-44 Treatment of larynx cancer has changed dramatically over the past several years. New targeted therapies have appeared to increasing local control rates and overall survival. In the TREMPLIN trial, researchers found that target therapies such as cetuximab showed evidence of a benefit in resectable LSCC.⁴⁵ However, adequate evidence for patients with unresectable LSCC is not available currently. Our study could lay a foundation of new specific therapeutic targets within signature to limit the development of LSCC. Targeted therapies play a role in chemoradiotherapy, but further comparative trials must be performed. Great advances have been made in larynx cancer treatment,⁴⁶ but we must continue improving our knowledge and quality standards to benefit as many patients as possible.

Although our eleven-IRGs signature showed impressive performance in LSCC recurrence prediction, there were several limitations in our study. First, the selection of predictors was limited to IRGs. Clinicopathologic factors, such as age, grade, smoking, alcohol, and TNM stage, were also important indicators in progression of LSCC. A combination of clinicopathologic factors and molecular factors may be a more powerful pattern for disease risk classification. Second, although this study was validated using an independent GEO cohort, prospective studies and multicenter prospective cohorts with large sample sizes are still needed to make a further clinical validation for the eleven-IRGs signature. Third, there are differences based on anatomical subsites in laryngeal carcinoma with distinct presentations and prognosis. However, as the details of sublocations in laryngeal carcinoma were not publicly available, we could not really explore the association of IRG signature and sublocations in depth. Finally, lack of mechanism studies hindered better application of the eleven-IRGs signature. Experimental research of each IRG within the eleven-IRGs signature could be performed to provide helpful information for understanding of their functional roles.

Conclusion

To summarize, we comprehensively utilized expression profiles and clinical data derived from TCGA database to identify DEIRGs, and for the first time constructed a novel eleven-IRGs signature for predicting recurrence in LSCC patients, which presents better predictive capacity and may aid the clinical decision-making process. These IRGs may act as molecular biomarkers to provide helpful information for selection of therapeutic strategies.

Footnotes

Acknowledgements

We thank the Gene Expression Omnibus and The Cancer Genome Atlas Database for sharing a large amount of data. We also thank all the R programming package developers.

Author contributions

Xiu Hong, Bing Liu, Yang Li and Yuqiang Hu contributed to data analysis, interpretation, and drafting. Qian Lv, Ying Liu, and Jun An contributed to data collection. Xiuli Wang and Rui Li involved in study design, study supervision, and final approval of the article. All authors read and approved the final article.

Declaration of Conflicting Interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This work was supported by the Health Commission of Jiangsu Province [grant number: Z2021016], Xuzhou Health Commission [grant number: XWKYHT20210590] and Xuzhou Science and Technology Bureau [grant number: KC20135].

Data availability

The source data of this study were derived from the public repositories, as indicated in Materials and Methods. All data that support the findings of this study are available from the corresponding author upon reasonable request.

Ethical approval

All information from TCGA and GEO databases is freely available to the public, so the approval of the medical ethics committee board was not applicable.

ORCID iD

Bing Liu

References

Sung

Ferlay

Siegel

, et al. (2021) Global Cancer Statistics 2020: GLOBOCAN Estimates of Incidence and Mortality Worldwide for 36 Cancers in 185 Countries. CA A Cancer J Clin 71: 209–249. DOI: 10.3322/caac.21660.

Cramer

Burtness

, et al. (2019) The changing therapeutic landscape of head and neck cancer. Nat Rev Clin Oncol 16: 669–683. DOI: 10.1038/s41571-019-0227-z.

Siegel

Miller

Jemal

Cancer statistics, 2020. CA A Cancer J Clin 2020, 70, 7-30, doi:10.3322/caac.21590.

Lo Nigro

Denaro

Merlotti

, et al. (2017) Head and neck cancer: improving outcomes with a multidisciplinary approach. Cancer Manag Res 9: 363–371. DOI: 10.2147/CMAR.S115761.

Marur

Forastiere

(2016) Head and Neck Squamous Cell Carcinoma: Update on Epidemiology, Diagnosis, and Treatment. Mayo Clin Proc 91: 386–396. DOI: 10.1016/j.mayocp.2015.12.017.

Brandstorp-Boesen

Sorum Falk

Folkvard Evensen

Boysen

Brondbo

(2016) Risk of Recurrence in Laryngeal Cancer. PLoS One 11: e0164068. DOI: 10.1371/journal.pone.0164068.

Ferris

(2015) Immunology and Immunotherapy of Head and Neck Cancer. J Clin Oncol : Official Journal of the American Society of Clinical Oncology 33: 3293–3304. DOI: 10.1200/JCO.2015.61.1509.

Ogino

Shigyo

Ishii

, et al. (2006) HLA class I antigen down-regulation in primary laryngeal squamous cell carcinoma lesions as a poor prognostic marker. Cancer Res 66: 9281–9289. DOI: 10.1158/0008-5472.CAN-06-0488.

Carter

Halpenny

Ginsberg

, et al. (2017) Immunotherapy in Non-Small Cell Lung Cancer Treatment: Current Status and the Role of Imaging. J Thorac Imag 32: 300–312. DOI: 10.1097/RTI.0000000000000291.

10.

Qin

Zheng

Gao

, et al. (2021) Tumor microenvironment and immune-related therapies of head and neck squamous cell carcinoma. Molecular therapy oncolytics 20: 342–351. DOI: 10.1016/j.omto.2021.01.011.

11.

Sangro

Sarobe

Hervas-Stubbs

, et al. (2021) Advances in immunotherapy for hepatocellular carcinoma. Nat Rev Gastroenterol Hepatol 18: 525–543. DOI: 10.1038/s41575-021-00438-0.

12.

Yang

Liu

, et al. (2018) Progresses and Perspectives of Anti-PD-1/PD-L1 Antibody Therapy in Head and Neck Cancers. Frontiers in oncology 8: 563. DOI: 10.3389/fonc.2018.00563.

13.

Ferris

Blumenschein

Jr. Fayette

, et al. (2016) Nivolumab for Recurrent Squamous-Cell Carcinoma of the Head and Neck. N Engl J Med 375: 1856–1867. DOI: 10.1056/NEJMoa1602252.

14.

Saleh

Eid

Haddad

Khalife-Saleh

, et al. (2018) New developments in the management of head and neck cancer - impact of pembrolizumab. Therapeut Clin Risk Manag 14: 295–303. DOI: 10.2147/TCRM.S125059.

15.

Seiwert

Burtness

Mehra

, et al. (2016) Safety and clinical activity of pembrolizumab for treatment of recurrent or metastatic squamous cell carcinoma of the head and neck (KEYNOTE-012): an open-label, multicentre, phase 1b trial. Lancet Oncol 17: 956–965. DOI: 10.1016/S1470-2045(16)30066-3.

16.

Gao

Yang

Chen

(2020) Identification of a four immune-related genes signature based on an immunogenomic landscape analysis of clear cell renal cell carcinoma. J Cell Physiol 235: 9834–9850. DOI: 10.1002/jcp.29796.

17.

Lin

, et al. (2020) Development and validation of a molecular prognostic index of bladder cancer based on immunogenomic landscape analysis. Cancer Cell Int 20: 302. DOI: 10.1186/s12935-020-01343-3.

18.

Birtalan

Danos

Gurbi

, et al. (2018) Expression of PD-L1 on Immune Cells Shows Better Prognosis in Laryngeal, Oropharygeal, and Hypopharyngeal Cancer. Applied immunohistochemistry & molecular morphology : AIMM 26: e79–e85. DOI: 10.1097/PAI.0000000000000590.

19.

Zhang

Guan

Peng

(2015) [The correlation between stanniocalcin 2 expression and prognosis in laryngeal squamous cell cancer]. Lin Chung Er Bi Yan Hou Tou Jing Wai Ke Za Zhi 29: 102–107.

20.

Huang

, et al. (2020) Laryngeal Squamous Cell Carcinoma: Potential Molecular Mechanism and Prognostic Signature Based on Immune-Related Genes. Medical science monitor: International Medical Journal of Experimental and Clinical Research 26: e928185. DOI: 10.12659/MSM.928185.

21.

Yoshihara

Shahmoradgoli

Martinez

, et al. (2013) Inferring tumour purity and stromal and immune cell admixture from expression data. Nat Commun 4: 2612. DOI: 10.1038/ncomms3612.

22.

Fouad

Salem

Ellakwa

, et al. (2019) MMP-2 and MMP-9 as prognostic markers for the early detection of urinary bladder cancer. J Biochem Mol Toxicol 33: e22275. DOI: 10.1002/jbt.22275.

23.

Hinton

Avraham

(2010) Role of the CXCR4/CXCL12 signaling axis in breast cancer metastasis to the brain. Clin Exp Metastasis 27: 97–105. DOI: 10.1007/s10585-008-9210-2.

24.

Zheng

Yang

, et al. (2020) Insulin-like growth factor 1-induced enolase 2 deacetylation by HDAC3 promotes metastasis of pancreatic cancer. Signal Transduct Targeted Ther 5: 53. DOI: 10.1038/s41392-020-0146-6.

25.

Lovato

Franz

Carraro

, et al. (2020) Maspin expression and anti-apoptotic pathway regulation by bcl2 in laryngeal cancer. Ann Diagn Pathol 45: 151471. DOI: 10.1016/j.anndiagpath.2020.151471.

26.

Nicolau-Neto

de Souza-Santos

Severo Ramundo

, et al. (2020) Transcriptome Analysis Identifies ALCAM Overexpression as a Prognosis Biomarker in Laryngeal Squamous Cell Carcinoma. Cancers 12: 470. DOI: 10.3390/cancers12020470.

27.

Zhou

Zhang

, et al. (2020) BCL11A Promotes the Progression of Laryngeal Squamous Cell Carcinoma. Frontiers in oncology 10: 375. DOI: 10.3389/fonc.2020.00375.

28.

Yao

Yan

Lian

, et al. (2020) Prognostic value of novel immune-related genomic biomarkers identified in head and neck squamous cell carcinoma. Journal for immunotherapy of cancer 8: e000444. DOI: 10.1136/jitc-2019-000444.

29.

Thomas

Bhola

Zhang

, et al. (2006) Cross-talk between G protein-coupled receptor and epidermal growth factor receptor signaling pathways contributes to growth and invasion of head and neck squamous cell carcinoma. Cancer Res 66: 11831–11839. DOI: 10.1158/0008-5472.CAN-06-2876.

30.

Wang

, et al. (2021) The Dual Role of STAT1 in Ovarian Cancer: Insight Into Molecular Mechanisms and Application Potentials. Front Cell Dev Biol 9: 636595. DOI: 10.3389/fcell.2021.636595.

31.

Liao

Chen

Cao

, et al. (2021) HIST1H1B Promotes Basal-Like Breast Cancer Progression by Modulating CSF2 Expression. Frontiers in oncology 11: 780094. DOI: 10.3389/fonc.2021.780094.

32.

Malik

Kanneganti

(2018) Function and regulation of IL-1α in inflammatory diseases and cancer. Immunol Rev 281: 124–137. DOI: 10.1111/imr.12615.

33.

Roszak

Mostowska

Sowinska

, et al. (2012) Contribution of IL12A and IL12B polymorphisms to the risk of cervical cancer. Pathol Oncol Res 18: 997–1002. DOI: 10.1007/s12253-012-9532-x.

34.

Wang

Jia

Liu

, et al. (2020) Silencing of miR-17-5p suppresses cell proliferation and promotes cell apoptosis by directly targeting PIK3R1 in laryngeal squamous cell carcinoma. Cancer Cell Int 20: 14. DOI: 10.1186/s12935-020-1096-3.

35.

Cui

Wang

Zhong

, et al. (2020) Identification and validation of methylation-driven genes prognostic signature for recurrence of laryngeal squamous cell carcinoma by integrated bioinformatics analysis. Cancer Cell Int 20: 472. DOI: 10.1186/s12935-020-01567-3.

36.

Jover-Espla

Palazon-Bru

Folgado-de la Rosa

, et al. (2018) A predictive model for recurrence in patients with glottic cancer implemented in a mobile application for Android. Oral Oncol 80: 82–88. DOI: 10.1016/j.oraloncology.2018.03.021.

37.

Cui

Wang

Tan

, et al. (2020) Development and validation of nomograms to accurately predict risk of recurrence for patients with laryngeal squamous cell carcinoma: Cohort study. Int J Surg 76: 163–170. DOI: 10.1016/j.ijsu.2020.03.010.

38.

Shi

Gong

Zhou

, et al. (2014) Dickkopf-1 is a novel prognostic biomarker for laryngeal squamous cell carcinoma. Acta Otolaryngol 134: 753–759. DOI: 10.3109/00016489.2014.894251.

39.

Zhang

Wei

, et al. (2021) NK4 Regulates Laryngeal Squamous Cell Carcinoma Cell Properties and Inhibits Tumorigenicity by Modulating the DKK1/Wnt/β-Catenin Axis. Frontiers in oncology 11: 783575. DOI: 10.3389/fonc.2021.783575.

40.

Jensen

Muzio

Mantovani

, et al. (2000) IL-1 signaling cascade in liver cells and the involvement of a soluble form of the IL-1 receptor accessory protein. Journal of immunology (Baltimore, Md, 1950) 164: 5277–5286. DOI: 10.4049/jimmunol.164.10.5277.

41.

Pasvenskaite

Vilkeviciute

Liutkeviciene

, et al. (2020) Associations of IL6 rs1800795, BLK rs13277113, TIMP3 rs9621532, IL1RL1 rs1041973 and IL1RAP rs4624606 single gene polymorphisms with laryngeal squamous cell carcinoma. Gene 747: 144700. DOI: 10.1016/j.gene.2020.144700.

42.

Wang

, et al. (2020) NEDD4 E3 ligase: Functions and mechanism in human cancer. Semin Cancer Biol 67: 92–101. DOI: 10.1016/j.semcancer.2020.03.006.

43.

Zhang

Hua

, et al. (2020) An Integrative Pan-Cancer Analysis Revealing LCN2 as an Oncogenic Immune Protein in Tumor Microenvironment. Frontiers in oncology 10: 605097. DOI: 10.3389/fonc.2020.605097.

44.

Zhang

Zhao

, et al. (2019) Asporin promotes cell proliferation via interacting with PSMD2 in gastric cancer. Frontiers in bioscience (Landmark edition) 24: 1178–1189. DOI: 10.2741/4774.

45.

Lefebvre

Pointreau

Rolland

, et al. (2013) Induction chemotherapy followed by either chemoradiotherapy or bioradiotherapy for larynx preservation: the TREMPLIN randomized phase II study. J Clin Oncol : Official Journal of the American Society of Clinical Oncology 31: 853–859. DOI: 10.1200/JCO.2012.42.3988.

46.

Salvador-Coloma

Cohen

(2016) Multidisciplinary Care of Laryngeal Cancer. Journal of oncology practice 12: 717–724. DOI: 10.1200/JOP.2016.014225.

Identification of an immune-related genes signature to predict risk of recurrence for patients with laryngeal squamous cell carcinoma

Abstract

Objective

Methods

Results

Conclusion

Keywords

Introduction

Materials and methods

Data acquisition and processing

Differentially expressed IRGs (DEIRGs) analysis

Functional enrichment analysis and protein–protein interaction network

Establishment of the IRGs-based signature for LSCC

Evaluation of relationship between IRGs-based signature and immune cell infiltration, and immune checkpoint molecules

Statistical analysis

Results

Identification of differentially expressed IRGs

Functional enrichment of differentially expressed IRGs

Protein–protein interaction network

Development and validation of the IRGs-based signature

Relationship of the IRGs-based signature with immune cell infiltration and immune checkpoint molecules

Discussion

Conclusion

Footnotes

Acknowledgements

Author contributions

Declaration of Conflicting Interests

Funding

Data availability

Ethical approval

ORCID iD

References