Different neutrophil extracellular trap related Ewing sarcoma subtypes exhibit distinct prognosis, and immune microenvironment characteristics

Abstract

Ewing sarcoma (EWS) is a rare bone cancer that is most usually detected in children. Neutrophil extracellular traps (NETs) are closely related to the prognosis of cancer, but the significance of NET-related features in EWS remains uncertain. We constructed a NET signature utilizing four crucial NET-related genes in EWS to forecast prognosis and investigate the potential immunological value of this signature in EWS. EWS data were collected from the International Cancer Genome Consortium and Gene Expression Omnibus databases. We identified the subtypes mediated by NET-related genes in EWS and analyzed the function infiltration and immune signature of NET-related subtypes in EWS. The expression levels of proteins in EWS cells were detected via western blotting analysis. NET could distinguish EWS patients into two NET-related subtypes: C1 and C2. EWS patients with the C1 subtype exhibited a more unfavorable prognosis and higher levels of TIDE and T cell dysfunction when compared to individuals with the C2 subtype. C1 and C2 subtypes had different immune characteristics. A NET-related prognostic model including AKT1, MAPK3, ATG7, and SELPLG was established to predict the prognosis of EWS patients. The risk score model was an independent prognostic factor for EWS, and high-risk EWS patients exhibited significantly inferior prognosis. AKT1 and ATG7 expression was significantly increased in EWS samples. The protein levels of AKT1 and ATG7 were increased in EWS cells, while the protein levels of SELPLG was decreased. The NET-related prognostic model is a critical biomarker for predicting prognosis, defining molecular subtypes, and describing immune signatures in patients with EWS.

Introduction

Ewing sarcoma (EWS) is a primary aggressive myeloid-derived tumor that mainly occurs in children and adolescents¹. Approximately 15% of EWS occurs in the soft tissues, 25% of EWS occurs in the pelvic bones, and 20% in the femur^2,3. The primary symptoms of EWS include localized pain and palpable masses. Additionally, some patients may present with systemic symptoms such as fever, elevated white blood cell counts, and an accelerated erythrocyte sedimentation rate⁴. Some patients may experience systemic symptoms like fever, fatigue, and weight loss in extraosseous disease⁵. In recent years, the combination of multiagent chemotherapy, surgery and also radiotherapy have observably improved the prognosis of patients with EWS, with the 5-year survival rate reaching 70%⁶. However, survival for a quarter of patients with metastatic EWS remains poor, with a 5-year survival rate of less than 30%⁷. In addition, patients with local and distant disease recurrence have a worse prognosis, with a 5-year event-free survival rate of only 10%². Therefore, an urgent need persists for novel diagnoses and treatment strategies to enhance long-term survival.

The most prevalent leukocyte in in the bloodstream, neutrophils are vital to the innate immune response to invasive infections⁸. In the tumor microenvironment, neutrophil extracellular traps (NETs) released by activated tumor associated neutrophils (TANS) can promote tumor development, progression, metastasis, and tumor related thrombosis⁹. NETs are made of DNA fibers, histones, and antimicrobial proteins that protect organisms against infections and are extruded by active neutrophils^10,11. In gastric cancer (GC), NETs can promote tumor cell metastasis¹². In colorectal cancer, high systemic NET expression is associated with reduced recurrence-free survival, and NETs stimulate colorectal cancer (CRC) cell migration via neutrophil elastase (NE) produced during NETosis to promote extracellular signal-regulated kinase (ERK) activity¹³. Moreover, it has been reported that EWS patients with high neutrophil-to-lymphocyte ratio, high systemic inflammatory index, and high neutrophil and platelet counts have poorer prognosis¹⁴. A study has shown that neutrophils in the EWS tumor bed are induced to produce NETs¹⁵. NET release significantly contributes to resistance against chemotherapy, immunization, and radiation therapy¹⁶. Patients with metastatic EWS have more tumor-infiltrating neutrophils (TINs) and NETs at diagnosis than patients with localized disease, and higher NETs at diagnosis produce predictive effects of neoadjuvant chemotherapy side effects, relapse, and mortality¹⁷. Accordingly, high levels of NET may play significant role in progression of EWS.

Previous studies have predominantly focused on analyzing the level of NET in relation to cancers and its significance in the prognosis of patients. However, the potential impact of NET-related genes on the prognosis of EWS patients remains largely unstudied. Therefore, using bioinformatics methods, we aimed to identify distinct subtypes of EWS based on NET-related gene expression. Additionally, we constructed a NET signature utilizing four crucial NET-related genes in EWS to forecast prognosis and investigate the potential immunological value of this signature in EWS.

Material and method

Human EWS samples and expression profile datasets

Transcriptional and clinical information of EWS patients in three public datasets were extracted from the International Cancer Genome Consortium (ICGC; https://icgc.org) and Gene Expression Omnibus (GEO, https://www.ncbi.nlm.nih.gov/geo/, ID: GSE63155, GSE68776) database. ICGC-EWS and GSE63155 contained 57 patients (Table 1) and 46 patients with complete follow-up information, respectively. GSE68776 included 32 EWS samples and 33 normal skeletal muscle samples (control). The “ComBat” function of the “sva “R package was used to remove batch effects.

Table 1.

Clinicopathological characteristics of Ewing Sarcoma patients from ICGC database.

Characteristics		Patients(N = 57)
		NO	%
Sex	Female	26	45.61%
	Male	31	54.39%
Age	≤ 21(Median)	30	52.63%
	> 21(Median)	27	47.37%
Survival time	Long(> 5 years)	29	50.88%
	Short(< 5 years)	28	49.12%
OS status	Dead	29	50.88%
	Alive	28	49.12%

Differential genes and functional enrichment analyses

Differential genetic analysis was performed on the two groups using the R language “limma” package. Differentially expressed genes (DEGs) were screened with |Log₂FC|> 2 and False Discovery Rate (FDR) < 0.05. False discovery rate (FDR) was estimated using the Benjamini–Hochberg method. Degree enrichment analysis of Kyoto Encyclopedia of Genes and Genomes (KEGG) and Gene Ontology (GO) was performed using the “clusterprofiler” function in R package¹⁸. The p < 0.05 was used to screen the significantly enriched pathways.

Survival analysis

Overall patient survival was estimated using the R language packages “survival” and “survminer” (http://CRAN.R-project.org/package=survival). The significance of survival differences between different groups was determined using the log-rank test.

Immune cell infiltration

In previous studies, 28 genes related to immune cells were identified¹⁹. Based on the expression of these genes, single-sample gene set enrichment analysis (ssGSEA) was used to predict immune cell infiltration in the samples. The TIDE algorithm (http://tide.dfci.harvard.edu/login/) was applied to evaluate the potential efficacy of immune checkpoint inhibition (ICI) treatment in different groups.

Protein–protein interaction (PPI) network

STRING (https://string-db.org/) database was used to construct PPI network. The hub genes were selected in PPI network using MCODE plug-in algorithm in Cytoscape and subjected to ClueGO enrichment analysis and visualization. The Interactions between proteins were analyzed using GeneMANIA (http://www.genemania.org). GeneMANIA is an advanced tool that aids in generating hypotheses about gene function through proficient analysis of gene lists and effective prioritization of genes for functional assays. By utilizing a vast array of genomic and proteomic data, GeneMANIA intelligently identifies genes with similar functions based on a given query gene list. GeneMANIA provides valuable insights by presenting weights that indicate the predictive significance of each selected data set for the query at hand.

Least absolute shrinkage selection operator (LASSO) Cox regression analysis

To identify genes associated with prognosis of EWS patients, univariate Cox regression analysis was performed. Then, to further evaluate hub prognosis-related genes, LASSO Cox regression analysis was done using in R language “glmnet” package²⁰. The Risk score of samples was computed using the following algorithm based on the hub prognosis-related genes:

Coefi: Calculate the hazard ratio for each factor using the LASSO Cox model. Xi: gene expression level. Second, risk score values were analyzed using survival test, survminer, and two-sided log-rank test in the R package. EWS patients were then divided into low-risk and high-risk groups based on the average risk score (0.07885563).

Cell culture

Human bone marrow mesenchymal stem cells (hBMSC) (CP-H166) and EWS cell line (A673 rhabdomyosarcoma) (CL-0017) were purchased from Procell (Wuhan, China). hBMSC was cultured in human bone marrow mesenchymal stem cell complete culture medium (CM-H166, Procell, Wuhan, China) at 37 °C in 5% CO₂ cell culture incubator. A673 was cultured in DMEM medium (PM150270, Procell, Wuhan, China) supplemented with 10% FBS (PM164210, Wuhan, China) and 1% P/S (P1400, Wuhan, China) at 37 °C in 5% CO₂.

Western blotting analysis

Proteins were extracted from cells using RIPA buffer (R0010, Solarbio, Beijing, China). The lysate was centrifuged (12,000 rpm, 10 min, 4 °C) and the supernatant was collected. Proteins were separated by SDS-PAGE and transferred to PVDF membrane. The membrane was blocked with 5% skim milk and incubated with primary antibodies overnight. The membrane was then washed with 1 × TBST and incubated with secondary antibody for 1 h. The primary antibodies used in this study was GADPH (1:10000, 60,004–1-Ig, Proteintech), AKT1 (1: 2000, 10176–2-AP, Proteintech), MAPK3 (1: 2000, 11257–1-AP, Proteintech), ATG7 (1: 500, 10088–2-AP, Proteintech), and SELPLG (1: 500, Ab68143, Abcam). The secondary antibodies were). Horseradish peroxidase-labeled goat anti-mouse IgG (H + L) (1: 10000, ZB-2305, Beijing Zhongshan Jinqiao Biotechnology Co., Ltd., China) and goat anti-rabbit IgG (H + L) (1: 10000, ZB-2301, Beijing Zhongshan Jinqiao Biotechnology Co., Ltd., China). Finally, the protein bands were detected using a fully automated chemiluminescence image analysis system (Chemi6000, Clinx, Shanghai, China).

Statistical analysis

The Wilcoxon test was used to compare differences in gene expression between tumor and normal samples. Multivariable Cox proportional hazards regression models were used to determine the impact of risk scores and clinicopathological characteristics on survival of patients with EWS. All statistical analyzes were performed using R v4.10 software. p < 0.05 indicated that the difference was statistically significant.

Results

Identification of NET-related subtypes in EWS patients

According to previous studies, we obtained sixty-nine genes with NET-initial biomarkers²¹. Sixty-nine NET-initial biomarker genes were used the univariate Cox regression analysis after incorporating time-dependent covariates in the ICGC-EWS cohort to identify NET-relevant subtypes for EWS, and we discovered that 8 NET-initial biomarker genes had prognostic potential for EWS (Hazard Ratio > 1, p < 0.05), including AKT1, DYSF, SELPLG, ATG7, MAPK3, IL17A, BST1 and CPPED1 (Fig. 1A).

Fig. 1.

Identification of neutrophil extracellular trap (NET)-related subtypes in Ewing sarcoma (EWS) patients. (A). The univariate Cox regression analysis for 69 NET-related genes in ICGC-EWS cohort. HR Hazard ratio, 95% CI 95% confidence interval. (B). Determined the optimal number of clusters using “elbow method”, This method computes the total within-cluster sum of squares error (SSE) for each candidate number of clusters. The SSE is plotted against the number of clusters, and the number of clusters is determined by a “elbow” in the curve (k = 2). (C). The survival rate of EWS patients in C1 and C2 subtypes. D. TIDE and dysfunction in C1 and C2 subtypes. TIDE Tumor immune dysfunction and exclusion; ** p < 0.05.

In the ICGC-EWS cohort, we performed a K-mean clustering analysis using 8 NET-initial biomarker genes. According to the sum of the squared errors, we chose the number of clusters k = 2, suggesting that 57 EWS patients could be clustered into two subtypes: C1 and C2 (Fig. 1B). Patients with the C1 subtype exhibited a more unfavorable prognosis when compared to individuals with the C2 subtype (Fig. 1C). We also chose k = 3 and k = 4 to group the samples into three and four subtypes, respectively. Kaplan–Meier survival analysis showed that Cluster2 had the lowest overall survival rate, and Cluster1 had the highest overall survival rate (Fig. S1A,B, Table S1). Because clustering patients into three categories and clustering them into four categories are both subdivisions of clustering into two categories. Thus, the C1 and C2 were utilized for subsequent analysis. In addition, as shown in Fig. 1D, the EWS patients with C1 subtype exhibited higher level of TIDE and T cell dysfunction scores compared to patients with C2 subtypes. A higher TIDE score indicates a greater probability of immune escape and a lower rate of immunotherapy effectiveness²². These results implied that EWS patients with C2 subtype might be demonstrated a more favorable response to immunotherapy.

Identification of DEGs and functional information between NET-related subtypes

To further study the impact of NET subtypes in patients with EWS, we determined the DEGs between C1 and C2 subtypes. A total of 67 DEGs was obtained between subtypes C2 and C1 (Fig. 2A, Table S2). To study interactions among DEGs, we then used the STRING database to construct a PPI network of these DEGs. The PPI network was visualized using Cytoscape (Fig. S1C). and performed enrichment analysis using ClueGO. We extracted three main modules and visualized all enrichment results (Fig. 2B–D). In addition, these 67 DEGs were highly enriched 667 GO pathways and 8 KEGG pathways (Table S3). The top 10 highly enriched GO term were presented in Fig. 2E, and the significantly enriched 8 KEGG pathways were shown in Fig. 2F. Moreover, REACTOME pathway enrichment analysis showed that these DEGs were significantly enriched in 76 pathways, such as Neutrophil degranulation (Table S4).

Fig. 2.

Identification of differentially expressed genes (DEGs) and functional information between neutrophil extracellular trap (NET)-related subtypes. (A). The DEGs between C1 and C2 subtypes. The horizontal axis was fold differentially expressed (log₂FC). (B–D). ClueGo enrichment results of DEGs. E. The top 10 significantly enriched pathways in GO term. (F). The results of KEGG enrichment analysis. The details of each KEGG pathway^54–56 are available at https://www.kegg.jp/kegg/kegg1.html. DOWN Down-regulated genes, NOT Not differentially expressed genes, UP Up-regulated genes, GO Gene Ontology, KEGG Kyoto Encyclopedia of Genes and Genome, BP Biological Process, MF Molecular Function, CC Cellular Component.

Identification of immune signatures between NET-related subtypes

To further investigate the correlation of NET-related subtypes with immunity, we conducted GSVA analysis and enriched the different pathways between C1 and C2 subtype. Various signaling pathways were significantly differentially enriched between C1 and C2 subtypes (Fig. 3A). The infiltration of TME between C1 and C2 subtypes was investigated via ssGSEA. The infiltration of Activated CD4 T cell was significantly decreased, and the infiltrations of Activated dendritic cell, CD56dim natural killer cell, Effector memeory CD8 T cell, Macrophage, myeloid-derived suppressor cells (MDSC), Monocyte, Regulatory T cell (Treg) and Type 1 T helper cell were observably increased in C1 subtype compared to C2 subtype (Fig. 3B). Furthermore, we discovered that the C1 subtype was associated with fewer immune cells than the C2 subtype (Fig. 3C). Further studies on the scores of 16 immune cells and 13 immune related functions in C1 and C2 subtypes showed that pDCs, T-helper-cells, TIL and CCR, check-point, MHC-CLASS-I were significantly increased in C1 subtype compared to C2 (Fig. 3D). These results suggested that the NET-related subtypes of EWS had different immune characteristic.

Fig. 3.

Identification of immune signatures between neutrophil extracellular trap (NET)-related subtypes. (A). The results of Gene set variation analysis. (B). The immune cell infiltration in C1 and C2 subtypes was calculated using SSGSEA. (C). The correlation of C1 and C2 subtypes with immune cells. (D). The scores of 16 immune cells and 13 immune related functions in C1 and C2 subtypes. * p < 0.05, ** p < 0.01.

Construction of NET related prognostic model in EWS

Among these 8 NET-related genes with prognostic potential of EWS patients, AKT1, MAPK3, ATG7 and SELPLG were selected (Fig. 4A, the lambda value was the smallest) as the hub NET-related genes in ES via LASSO Cox regression analysis. Following that, the expression of these four genes was weighted with the LASSO Cox regression coefficient to develop a NET prognostic model: Risk Score = (AKT1 × 0.036378755) + (ATG7 × 0.036437482) + (MAPK3 × 0.009915527) + (SELPLG × 0.006536454). Interestingly, there was no significant correlation among these four genes (Fig. 4B). The GeneMANIA network showed that there were potential protein interactions between these 4 and 20 proteins (Fig. 4C).

Fig. 4.

Construction of neutrophil extracellular trap (NET) related prognostic model in Ewing sarcoma (EWS). (A) The lambdas quality control plots of LASSO Cox regression analysis. (B). The correlation among four candidate genes. (C). The co-expression relationship of four genes explored by Genemania. (D). Sanggi diagram showed the relationship between subtype, survival status, and risk score. (E). The TIDE score in high- and low-risk groups. TIDE Tumor immune dysfunction and exclusion, IFNG Interferon gamma, MDSC Myeloid-derived suppressor cells, CAF Cancer-associated fibroblasts, CTL Cytotoxic T lymphocyte. * p < 0.05, ** p < 0.01.

In the ICGC-EWS cohort, the 67 patients were separated into high- and low-risk groups based on the best intercept value (0.07885563). The association between risk score and survival status with NET subtypes was investigated, and the findings confirmed our assumptions. The prognosis of patients with subtype C1 was significantly worse than that of patients with subtype C2, with most patients classified as high risk (Fig. 4D). This proves the effectiveness of the NET prediction model. Furthermore, despite lower MDSC, TIDE, IFNG, dysfunction, rejection, CAF, and CTL were significantly increased in the high-risk group (Fig. 4E, high vs. low), indicating that the risk assessment model was relevant to the TME.

Verification of NET prognostic model in EWS

In the ICGC-EWS cohort, the high-risk group was correlated with inferior prognosis of ES patients (Fig. 5A). The areas under the ROC curve (AUC) in years 2, 3, and 4 were 0.67, 0.71, and 0.75, respectively (Fig. 5B). At the same time, in the GSE63155 validation set, the prognosis of high-risk EWS patients also exhibited inferior prognosis (Fig. 5C). The AUCs at years 2, 3, and 4 were 0.65, 0.74, and 0.74, respectively (Fig. 5D). The model had a relatively high distinguishing ability of prognosis and could identify the high-risk group patients with worse survival results in ICGC-EWS and validation cohorts. Furthermore, we conducted a multivariable Cox regression analysis after incorporating time-dependent covariates, revealing that the risk score model serves as an independent prognostic factor (Fig. 5E). According to the findings, the NET prognostic model was an independent prognostic factor for EWS and had a good predictive potential for EWS patients. Clinicians may utilize this model to assess patient prognosis and develop personalized treatment plans, particularly for identifying high-risk patients who could benefit from more aggressive treatment or closer monitoring.

Fig. 5.

Verification of neutrophil extracellular trap (NET) prognostic model in Ewing sarcoma (EWS). (A). The survival rate of EWS patients in high- and low-risk groups in the ICGC cohort. (B). Time-dependent receiver operating characteristic (ROC) analysis of the ICGC-EWS cohort. (C). The survival rate of EWS patients in high- and low-risk groups in the GSE63155 dataset. (D). Time-dependent ROC analysis of the GSE63155 dataset. (E). The association between the risk score and clinical characteristics was determined using multivariate Cox regression analysis. AUC Area under the curve. * p < 0.05, *** p < 0.001.

Expression of four genes for constructing prognostic model in EWS patients

Next, we analyzed expression of genes for constructing prognostic model in EWS and normal samples in the GSE68776 dataset. We found that the expression of AKT1 and ATG7 was significantly increased and SELPLG expression was reduced in EWS samples compared to normal bone samples (Fig. 6A). In addition, we analyzed the expression levels of SELPLG, ATG7, MAPK3, and AKT1 proteins in hBMSC and A673 cells via western blotting analysis, and found that the protein levels of AKT1 and ATG7 were increased in A673 compared to hBMSC, while the protein levels of SELPLG was decreased (Fig. 6B).

Fig. 6.

Expression of four genes for constructing prognostic model in Ewing sarcoma (EWS) patients. (A). The expression of AKT1, SELPLG, MAPK3 and ATG7 in EWS samples and normal bone samples in the GSE68776 dataset. (B). The expression levels of SELPLG, ATG7, MAPK3, and AKT1 proteins in hBMSC and A673 cells were detected via western blotting analysis. * p < 0.05, ** p < 0.01, *** p < 0.001, ns indicates no significance.

Discussion

As the second most common primary bone tumor in adolescents and young adults, EWS is a an extremely aggressive malignancy known for its rapid metastasis²³. It has been demonstrated that EWS cells display a distinct degree of heterogeneity, which impacts both their proliferative and migratory capabilities within the tumor²⁴. In our study, we identified two distinct molecular subtypes of EWS based on eight NET-associated gene expression, and found that EWS patients with C1 subtype exhibited a more unfavorable prognosis and had higher level of TIDE and T cell dysfunction when compared to individuals with C2 subtype. T cell dysfunction and T cell exclusion are the two basic methods for modeling tumor immune evasion. TIDE combines two important immune mechanisms: T cell dysfunction and T cell rejection expression profile, which could predict ICI therapeutic response of cancer patients²⁵. The higher the tumor TIDE score, the worse the ICI treatment effect, and the lower the survival time of patients receiving anti-PD-1 and anti-CTLA-4 treatment²⁵. EWS tumors generally show low PD-L1 expression and limited responses to ICIs, as reported by Spurny et al.²⁶. A clinical trial investigating the use of the PD-1 checkpoint inhibitor pembrolizumab to treat adult EWS patients did not demonstrate significant clinical activity²⁷. Thus, it is not surprising that the C1 subtype, with its higher TIDE scores and T cell dysfunction, would have a worse prognosis. Research reports indicate that polymorphisms in CTLA-4 are risk factors that affect the prognosis of EWS²⁸. Furthermore, in EWS, CTLA-4 may function as an immune checkpoint molecule²⁹. CTLA-4 is primarily expressed on the surface of T cells and plays a crucial role in inhibiting T cell activation during immune responses³⁰. In the tumor microenvironment of EWS, there may be an upregulation of CTLA-4 expression, which can suppress T cell attacks on tumor cells, thereby promoting tumor immune escape²⁹. Moreover, Jiang et al. reported a significant difference in the expression of CTLA-4 between EWS patients categorized into low- and high-hypoxia-related risk groups³¹. Given that EWS patients with the C2 subtype have lower TIDE scores, we hypothesize that individuals with the C2 subtype may benefit from anti-CTLA-4 immunotherapy. However, further experiments are necessary to validate this hypothesis.

It has been reported that in EWS, the presence of neutrophils and NETs is associated with pathological and clinical features of the tumor and prognosis³². Patients with metastatic EWS have higher tumor-infiltrating neutrophils and NETs at diagnosis compared with patients with localized disease³². High NET formation at diagnosis indicates adverse response to neoadjuvant chemotherapy, disease recurrence, and death³². We found that the proportion of infiltrating neutrophils showed no significant difference between the C1 and C2 EWS subtypes, which may be related to the heterogeneity of the tumor microenvironment and the characteristics of tumor cells. The microenvironment of EWS exhibits a certain degree of heterogeneity; however, its capacity to recruit neutrophils does not demonstrate significant variations across different subtypes. In a comparative study between non-malignant osteoblastoma and EWS, although there was no notable difference in the number of infiltrating neutrophils, the formation of NETs was significantly elevated in EWS samples³². This finding indicates that while the tumor microenvironment can recruit neutrophils, not all recruited neutrophils are activated to form NETs. In the C1 and C2 EWS subtypes, distinct signaling pathways or molecular mechanisms may regulate the functional state of neutrophils, resulting in no significant difference in neutrophil infiltration between the two subtypes. Moreover, different subtypes of EWS cells may secrete various cytokines and chemokines, influencing the infiltration and extravasation of neutrophils. However, neutrophil recruitment mechanisms may be present in both subtypes, characterized by high and low degrees of neutrophil extravasation, differing primarily in the expression or release of factors associated with NET formation. For instance, study has indicated that levels of multiple cytokines and chemokines in the plasma of EWS patients are significantly elevated compared to those in healthy controls³². These factors may collectively contribute to the activation of neutrophils and the formation of NETs.

Previous studies have demonstrated that NETs possess the capability to modulate immune cells, thereby exerting an influential effect on the progression of tumors³³. Therefore, we assessed the immune cell infiltration in two NET-related subtypes. EWS patients with C1 subtype had lower infiltration of activated CD4 + T cells and higher infiltration of immunosuppressive cells (MDSC, macrophage and regulatory T cell) compared to patients with C2 subtype. CD4 + T cells hold vital importance in monitoring cancer immunity. They impact the microenvironment of tumors and eliminate tumor cells. NETs can impair the immune system by inhibiting CD4 + T cells³⁴. MDSC can aid tumor growth by boosting tumor cell survival, angiogenesis, tumor cell invasion of healthy tissue, and metastasis³⁵. In most solid tumors, high macrophage infiltration corresponds with poor outcomes. It was associated with modifications in inflammation related to cancer, angiogenesis, remodeling of the extracellular matrix (ECM), and transitioning of cancer cells from epithelial to mesenchymal states (EMT)^36–38. Tregs can decrease anticancer immunity, impairing tumor immunosurveillance and effective antitumor immune responses in tumor-bearing individuals³⁹. Moreover, we discovered that the C1 subtype was associated with fewer immune cells than the C2 subtype. This finding suggested that EWS patients with C1 subtype showed higher immunosuppression compared to patients with C2 subtype.

Considering the effects of NETs in the prognosis of EWS, we built a NET prognostic model using four NET-related genes, AKT1, MAPK3, ATG7 and SELPLG. AKT1 is a major hub of the PI3K/Akt signaling system that is most frequently activated in human malignancies⁴⁰. Activated AKT1 signaling can promote survival and proliferation of cell and inhibit apoptosis⁴¹, and overactive AKT1 is a hallmark of multiple malignant tumor^42,43. The mediation of survival signals by PI3K/AKT aids in rescue EWS cells from cell death triggered by fibroblast growth factor 2⁴⁴. In osteosarcoma, AKT1 knockdown clearly inhibited MAT1-mediated enhancements of cell motility and invasion in vitro, suppressed tumor development, and decreased the quantity of lung tumors that metastasized in xenografted nude mice⁴⁵. MAPK3, also known as ERK1, it plays a crucial role in the MAPK signaling pathway that is responsible for regulating a diverse range of cellular processes, such as differentiation and stress responses, and reduced ERK1/2 signaling may decrease tumorigenicity of EWS⁴⁶. ERK1/2 pathway mediates the role of c-Met inhibitor (PHA-665752) in proliferation, migration and invasion of osteosarcoma cells⁴⁷. ATG7 is a crucial protein for autophagy activation and appears to be involved in the scenario-dependent interaction between autophagic apoptosis and autophagy-dependent resistance⁴⁸. ATG7-dependent autophagy promoted erlotinib resistance, whereas ATG7 knockdown overcome the resistance of erlotinib-treated breast cancer cells⁴⁹. The PCAF-H3S28 axis induces autophagy in osteosarcoma cells by targeting ATG5 and ATG7⁵⁰. SELPLG is a selectin that is expressed in immunological and inflammatory cells and is involved in immune cell trafficking as well as the regulation of myeloid cell immune responses⁵¹. In multiple cancers, SELPLG has been reported as a potential biomarker for prognosis and diagnose, including CRC and osteosarcoma^52,53. The expression of SELPLG was found to be lower in metastatic osteosarcoma samples when compared to non-metastatic osteosarcoma samples. Furthermore, it was observed that low SELPLG expression served as an independent and unfavorable prognostic factor for patients with osteosarcoma⁵³. In the present study, we observed that the protein levels of AKT1 and ATG7 were elevated, while the protein levels of SELPLG were diminished in EWS cells compared to normal bone cells. These findings suggest that AKT1 and ATG7 may contribute to cancer progression in EWS, whereas SELPLG may have a suppressive role in this context. Nevertheless, the functional roles of these model genes in tumors have not been thoroughly investigated; therefore, the functions of the hub genes require validation and further exploration through in vitro and in vivo experiments.

Conclusion

In conclusion, we identified two different molecular subtypes based on NET, and constructed a NET-related signature consisting of AKT1, MAPK3, ATG7 and SELPLG in EWS. NET-related prognostic model is a critical biomarker for predicting prognosis, defining molecular subtypes, and describing immune signatures in patients with EWS. However, the correlation between NET and immunity in EWS needs to be further researched at the cellular and molecular levels.

Author contributions

SL participated in the design of this study, and YH performed statistical analysis. LZ carried out the study and collected background information. SL drafted this manuscript. HZ and CW reviewed and edited the manuscript. All the authors have read and approved the final manuscript.

Data availability

The data utilized in this study are available in the International Cancer Genome Consortium (ICGC; https://icgc.org) and Gene Expression Omnibus (GEO, https://www.ncbi.nlm.nih.gov/geo/, ID: GSE63155 and GSE68776) database.

Declarations

Competing interests

The authors declare no competing interests.

Ethical approval and consent to participate

This study has been reviewed by Zibo Central Hospital Institutional Review Board (IRB) and has determined this study is exempt from full IRB review as it does not constitute human subject research.

Supplementary Information

The online version contains supplementary material available at 10.1038/s41598-025-10277-7.