Identification of necrosis-related signature for predicting prognosis and immunotherapy response in gastric cancer

Introduction

Gastric cancer (GC) constitutes the fourth most prevalent malignancy globally, with an annual incidence exceeding 1.3 million cases. It represents the second leading cause of cancer-related mortality (1,2). GC typically exhibits significant phenotypic and molecular heterogeneity. The alarming mortality rate, evidenced by the 784,000 fatalities reported in 2018, underscores the urgent need for effective interventions (3). Early-stage GC often manifests with nonspecific symptoms or remains clinically silent. Unfortunately, in many regions lacking robust early diagnostic capabilities, approximately 80% of individuals are diagnosed only after the disease has progressed to advanced stages (4). The development of GC is influenced by a complex interplay of genetic and environmental factors (5). For early-stage GC, endoscopic resection is the primary treatment modality. Patients with advanced disease or those ineligible for surgery may benefit from alternative strategies, including adjuvant chemotherapy, radiation therapy, hyperthermic intraperitoneal chemotherapy (HIPEC), and pressurized intraperitoneal aerosol chemotherapy (PIPAC) (6). Despite these efforts, median survival remains below 1 year (7). In recent years, immune checkpoint inhibitors (ICIs) have significantly transformed the therapeutic landscape for chemorefractory GC. Programmed death-1 (PD-1), a pivotal inhibitory receptor expressed on activated T cells, is one such checkpoint (8).

Inhibition of PD-1 prevents its binding to its ligand [programmed death-ligand 1 (PD-L1)], thereby reestablishing anti-tumor immunity (9). The current literature indicates that overexpression of PD-L1 is prevalent in gastric GC, suggesting that targeting PD-1 inhibition holds significant promise in this disease area (10). Moreover, a growing number of ICIs have been identified, including programmed death-ligand 2 (PD-L2), cytotoxic T-lymphocyte-associated protein 4 (CTLA-4) and its ligands CD86 and CD80, T-cell immunoglobulin and mucin domain 3 (TIM3) and its ligand galectin-9 (11). However, a notable limitation of these ICIs is the lack of response in over half of patients undergoing PD-1/PD-L1 immunotherapy (12). Therefore, identifying patients who are likely to benefit from immunotherapy is imperative. Currently, carbohydrate antigen 19-9 (CA19-9), carcinoembryonic antigen (CEA), carbohydrate antigen 15-3 (CA15-3), and carbohydrate antigen 72-4 (CA72-4) are widely employed in the diagnosis and therapy monitoring of GC (13). While individual tumor biomarkers may exhibit limited sensitivity and specificity, the development of novel biomarkers or predictive models could facilitate prognostication and serve as potential therapeutic targets for GC patients.

Necrosis has traditionally been regarded as an accidental and disordered form of cell death. In contrast to apoptosis and autophagy, necrosis has been implicated in neoplasm aggressiveness and progression as a “reparative cell death” (14-16). Morphological hallmarks of necrosis include cell swelling, loss of cytoplasmic integrity, and release of cytoplasmic contents into the extracellular milieu. High mobility group box 1 (HMGB1), a multifunctional nonhistone chromatin protein, orchestrates nucleosome dynamics and modulates transcriptional regulation through DNA bending and histone interactions, acts as a tumor-promoting and proinflammatory cytokine when released by necrotic cells (17,18). These released proteins recruit immune cells, potentially leading to epigenetic alterations, proto-oncogenic mutations, angiogenesis, neoplasm cell proliferation, and invasiveness, ultimately accelerating tumor progression (19,20). Previous studies have demonstrated elevated HMGB1 expression in GC tissues compared to adjacent non-cancerous tissues. Moreover, short hairpin RNA (shRNA)-mediated knockdown of HMGB1 has been shown to significantly inhibit cell proliferation and invasive potential, suggesting its potential as a diagnostic and therapeutic target for GC (21). However, due to technical restrictions, these researches have been confined to one or two necrosis-related proteins. Given the synergistic interplay of multiple neoplasm-inhibiting factors, the broader landscape of necrosis-related proteins as potential immunotherapy biomarkers in GC treatment remains largely unexplored.

The tumor microenvironment (TME) represents a dynamically organized ecosystem comprising diverse cell types, including immune and inflammatory cells, endothelial cells, extracellular matrix (ECM) elements, fibroblasts, and cancer-secreted cytokines. These components interact to create a chronic inflammatory, pro-tumoral, and immunosuppressive environment that significantly contributes to neoplasm development and progression (22,23). ICIs are closely linked to the host immune system and TME. The intricate interplay between neoplasm cells and the TME involves the coexistence of both oncogenes and anti-oncogenes. Recent evidence highlights the close association between the TME and GC. For instance, tumor-associated neutrophils (TANs) are crucial stromal partners in carcinogenesis, enhancing the migration and invasive capacities of GC cells (24). Activated GC-associated fibroblasts (GCAFs) can also promote the migration and invasion properties of GC cells both in co-culture and through the presence of GCAFs-conditioned medium (25). Tumor-associated macrophages (TAMs) exist in two phenotypes, M1 and M2, with the M2 phenotype being essential for tumor growth and progression (24). Consequently, a comprehensive analysis of the heterogeneity and complexity within the TME, identification of distinct tumor immunophenotypes, exploration of promising new therapeutic approaches, and accurate prediction of clinical efficacy would be valuable discoveries.

Recent studies have demonstrated the significant role of necrosis-related proteins in the TME. Necrosis is intricately linked to the release of damage-associated molecular patterns (DAMPs). Many current neoplasm treatment strategies, including chemotherapy, radiotherapy, and hormonal therapy, increase DAMP release following therapy-induced neoplasm death through necroptosis and necrosis (26). Released DAMPs can critically influence the TME by affecting the immune response or impacting angiogenesis and stroma formation (27,28). Despite advancements in understanding the molecular interplay between necrosis and the TME, previous research has predominantly focused on limited necrosis regulators and has been conducted in vitro. Historically, the combination of necrosis and TME has not been incorporated into GC prognostic models. Consequently, a comprehensive analysis of necrosis-related proteins and the TME in real-world patient cohorts is urgently needed. Therefore, this study aimed to elucidate the role of necrosis in combination with the TME in GC. We present this article in accordance with the TRIPOD reporting checklist (available at https://tcr.amegroups.com/article/view/10.21037/tcr-2025-1435/rc).

Methods

Date sources

The study was conducted in accordance with the Declaration of Helsinki and its subsequent amendments. Gene expression data [fragments per kilobase million (FPKM)] and corresponding clinical and pathological data for GC were retrieved from public databases: The Cancer Genome Atlas (TCGA) (https://portal.gdc.cancer.gov/) and the Gene Expression Omnibus (GEO) (https://www.ncbi.nlm.nih.gov/geo/). Within the R statistical environment, we utilized the FPKM function from the limma package to convert the FPKM values into transcripts per kilobase million (TPM). This transformation ensured consistent units for subsequent analyses. A total of 1,109 GC patients were included for further investigation. Clinical variables encompassed age, sex, tumor-node-metastasis (TNM) stage, overall survival, and gene expression profiles of GC samples. Datasets meeting all inclusion criteria were subjected to copy number variation (CNV) analysis. Copy number changes of genes regulating necrosis-related pathways were visualized across chromosomes using the Rcircos package.

Consensus clustering analysis of necrosis-related genes (NRGs)

A total of 159 genes associated with necrosis were obtained from the Molecular Signatures Database (MSigDB) Team’s REACTOME_PYROPTOSIS gene set (http://www.broad.mit.edu/gsea/msigdb/) (detailed information provided in Table S1). To classify patients into distinct molecular subtypes based on the expression of these NRGs, unsupervised consensus clustering analysis was performed using the R package “ConsensusClusterPlus”.

Relationship between molecular subtypes with the clinical features and prognosis of GC

To investigate the clinical characteristics of the three subtypes identified through consensus clustering, we quantified the interdependencies between molecular patterns, clinicopathological features, and survival outcomes. Patient parameters encompassed age, sex, TNM stage, and survival status. Cox regression modeling was employed to assess survival disparities among the three NRGs patterns. Additionally, disparities in prognosis among the distinct subtypes were evaluated using Kaplan-Meier curves generated with the survival and survminer R packages.

Gene set variation analysis (GSVA) and gene enrichment function annotation

To elucidate the differential activities of NRGs in physiological events, GSVA was performed using gene sets from the “c2.cp.kegg.v6.2 symbol” category within the MSigDB (29). Subsequently, gene ontology (GO) function annotations and Kyoto Encyclopedia of Genes and Genomes (KEGG) pathways associated with NRGs were analyzed using the “clusterProfiler” package, with a false discovery rate (FDR) threshold of <0.01. To delve deeper into the potential functions and differences in the activities of necrosis pattern-related differentially expressed genes (DEGs), functional enrichment analyses were conducted on these DEGs using the “clusterProfiler” package in the R statistical environment.

Immune cell infiltrating difference characteristics analysis

To assess the composition of tumor immune infiltrates, we employed two complementary approaches. First, we utilized the CIBERSORT algorithm (https://cibersort.stanford.edu/) to estimate the relative abundances of various immune cell populations within the TME of each patient. Second, we leveraged the single-sample gene set enrichment analysis (ssGSEA) algorithm to quantify the enrichment of gene sets specific to different immune cell types, further providing insights into the overall immune cell infiltration landscape within the GC TME.

DEGs identification and functional annotation

To identify DEGs between the distinct necrosis subtypes, the limma package in R was employed, applying a fold-change threshold of 1.5 and an adjusted P value of <0.05. To delve deeper into the potential functions of necrosis pattern-related DEGs and uncover associated gene functions, GO and KEGG enrichment analyses of DEGs were conducted using the clusterprofiler package in R.

Establishment of a NRG signature

A scoring system termed “necrosis score” was developed to quantify distinct NRG subtypes within individual GC cases. Univariate Cox regression identified genes with significant prognostic associations between the clusters. Subsequently, a consensus clustering algorithm based on the expression of overlapping genes was utilized to classify patients into three distinct pattern groups: gene cluster A, gene cluster B, and gene cluster C. Next, the principal component analysis (PCA) was used to construct the RM_score. Principal components 1 and 2 (PC1 and PC2) were selected to constitute the signature scores. This approach leverages their ability to capture the largest blocks of highly correlated (or anti-correlated) genes while down-weighting genes that do not track with other members of the set. The RM_score was defined as follows:

RM_score=S(PC1i+PC2i)[1]

where i represents the expression level of an RNA modification phenotype-related gene. Then the patients were stratified into high and low RM_score groups using the maximally selected rank statistics method and analyzed via Kaplan-Meier survival analysis (log-rank tests, P<0.001).

Mutation and drug susceptibility analysis

To investigate somatic mutations in GC patients categorized into high- and low-risk cohorts, tumor mutational burden (TMB) scores were calculated for each patient using the “maftools” R package. These scores were then compared between the two groups to identify potential differences in therapeutic responses to immunotherapeutics. The Wilcoxon test was employed to assess the statistical significance of any observed disparities.

Statistical analysis

To assess differences between GC and normal tissues, statistical analyses were performed using t-tests. Spearman correlation coefficients were calculated to determine the relationship between TME-infiltrating immune cells and the expression of NRG regulators. To assess statistical significance among several cohorts, one-way analysis of variance (ANOVA) and the Kruskal-Wallis test were employed. Univariate regression analyses identified hazard ratios (HRs) associated with necrosis and NRGs. Prognostic analysis utilized Kaplan-Meier curves and the log-rank test to evaluate the significance of survival differences. Finally, the Chi-squared test compared the proportions of immune subtypes between high- and low-necrosis score groups. All statistical analyses were conducted using R version 4.1.0 (https://www.r-project.org/).

Results

Identification of the NRG clusters

To elucidate the expression characteristics of NRGs in GC, we employed a consensus clustering algorithm to classify patients based on the expression profiles of 159 NRGs. The optimal number of clusters was three, designated as cluster-1, cluster-2, and cluster-3 (Figure 1A). Survival analysis revealed that patients in cluster-2 had a more favorable prognosis than those in the other two NRGs patterns (log-rank test, P=0.02; Figure 1B). The heatmap demonstrated the differential expression of NRGs across the subtypes, with heightened expression in cluster-2 and negligible expression in cluster-1 and cluster-3 (Figure 1C).

Figure 1 NRGs patterns and clinicopathological characteristics. (A) Consensus matrix heatmap defining three clusters (k=3) and their correlation area. (B) Kaplan-Meier curves of the overall survival in the TCGA-STAD and GSE84437 cohorts based on the three NRGs patterns. (C) Heat map showing differences in clinicopathologic features and expression levels of NRGs between the three distinct patterns. N, node; NRG, necrosis-related gene; STAD, stomach adenocarcinoma; T, tumor; TCGA, The Cancer Genome Atlas.

Characteristics of TME and biological processes (BPs) under different NRG clusters

To elucidate the differences in biological function among various NRG modification patterns, we employed GSVA. As depicted in Figure 2A-2C, distinct functional pathways were enriched in each cluster. Cluster-1 was primarily concentrated in phosphatidylinositol, calcium neuroactive ligand-receptor interaction, and ABC, Taste transduction signaling pathways. Cluster-2 exhibited significant enrichment in immune fully-activated pathways, such as natural killer cell-mediated cytotoxicity, T cell receptor signaling pathway, RIG-I-like Toll-like NOD-like receptor and chemokine signaling pathways, autoimmune thyroid, type 1 diabetes mellitus, and graft-versus-host diseases. Cluster-3 was markedly activated in cell cycle, protein export, RNA polymerase, amino sugar and nucleotide sugar metabolism, and proteasome pathways. Subsequent analysis of the role of NRGs in the TME of GC revealed distinct immune microenvironment characteristics among clusters (Figure 3A). We evaluated the associations between the three clusters and 23 human immune cell subsets using the CIBERSORT algorithm. Cluster-2 was significantly enriched in activated B cells, CD4⁺ T cells, CD8⁺ T cells, dendritic cells, CD56 bright natural killer cells, eosinophils, gamma delta T cells, immature B cells, immature dendritic cells, myeloid-derived suppressor cells (MDSCs), macrophages, mast cells, monocytes, natural killer T cells, natural killer cells, plasmacytoid dendritic cells, regulatory T cells, T follicular helper cells, type 1 T helper cells, type 17 T helper cells, and type 2 T helper cells. These findings demonstrate that the three NRG modification patterns exhibit distinct infiltration characteristics of TME cells. Furthermore, PCA revealed significant differences in the transcriptome profiles of these modification patterns (Figure 3B).

Figure 2 Biological characteristics of NRGs patterns. (A) GSVA analyzed the differences between functional pathways in NRGs patterns 1 and 2. Blue represents the NRGs pattern 1, and orange represents the NRGs pattern 2. (B) GSVA analyzed the differences between functional pathways in NRGs patterns 1 and 3. Blue represents the NRGs pattern 1, and red represents the NRGs pattern 3. (C) GSVA analyzed the differences between functional pathways in NRGs pattern 2 and 3. Orange represents the NRGs pattern 2, and red represents the NRGs pattern 3. GSVA, gene set variation analysis; KEGG, Kyoto Encyclopedia of Genes and Genomes; NRG, necrosis-related gene.

Figure 3 Identification of gene subtypes based on DEGs. (A) The differential expression analysis of 23 immune cells among three NRGs patterns. (B) PCA revealed significant differences in the transcriptome profiles of these modification patterns. (C) The number of unique and shared DEGs from different comparisons is revealed in a Venn diagram. (D,E) GO and KEGG enrichment analyses of DEGs among three NRGs patterns. BP, biological process; CC, cellular component; DEGs, differentially expressed genes; GO, gene ontology; KEGG, Kyoto Encyclopedia of Genes and Genomes; MF, molecular function; NRG, necrosis-related gene; PC, principal component; PCA, principal component analysis.

Identification of gene subtypes in GC based on DEGs

To delve deeper into the potential biological characteristics of each NRG pattern, our team identified DEGs linked to necrosis-related molecular subsets using the R package “limma” (Figure 3C) and performed functional enrichment analysis (Figure 3D). GO analysis revealed that the DEGs were primarily enriched in the BP of T cell activation, the cellular components of the DEGs were predominantly enriched in the external side of the plasma membrane, and the molecular functions of DEGs were correlated with cytokine receptor binding. KEGG analysis indicated enrichment of osteoclast differentiation and cytokine-cytokine receptor interaction pathways (Figure 3E). In addition, we performed univariate Cox regression analysis to identify DEGs associated with the prognosis of GC and selected prognostic DEGs (P<0.05) for subsequent studies. Consistent with the NRG modification patterns, unsupervised clustering algorithms also identified three distinct genomic phenotypes (gene cluster-A, gene cluster-B, and gene cluster-C) based on these prognostic DEGs (Figure 4A). Survival evaluation demonstrated that patients in cluster-B had a favorable prognosis, while patients in cluster-C had the worst prognosis (log-rank test, P<0.001; Figure 4B). The heat map of genetic modification patterns, including clinicopathological features, revealed that most genes were highly expressed in gene cluster-B and lowly expressed in gene cluster-C (Figure 4C).

Figure 4 Identification of gene subtypes in GC based on the prognostic DEGs. (A) Consensus matrix heatmap defining three clusters (k=3) and their correlation area based on these prognostic DEGs. (B) Kaplan-Meier curves of the overall survival based on these prognostic DEGs. (C) Heat map showing differences in clinicopathologic features and expression levels of prognostic DEGs between the three distinct patterns. DEGs, differentially expressed genes; GC, gastric cancer; M, metastasis; N, node; T, tumor.

Construction of the prognostic necrosis score

To systematically quantify the necrosis profiles in individual GC patients, we developed a necrosis score based on the necrosis-related signature. Patients were categorized into high-necrosis score and low-necrosis score groups according to the optimal cutoff value. Survival analysis demonstrated that patients with low-necrosis scores exhibited markedly longer overall survival durations compared to those in the group with high-necrosis score (Figure 5A). The alluvial diagram comprehensively illustrated the process of necrosis score fraction construction (Figure 5B). To explore the prospective impact of the necrosis score on immune regulation within the TME, we conducted an immune correlation analysis. The results revealed a significant negative correlation between the necrosis score and activated CD4 T cells, activated CD8 T cells, activated dendritic cells, gamma delta T cells, immature B cells, MDSCs, macrophages, monocytes, natural killer T cells, natural killer cells, regulatory T cells, T follicular helper cells, type 1 T helper cells, and type 2 T helper cells (Figure 5C). The necrosis score not only differentiated necrosis patterns but also distinguished gene subtypes. Differential analysis of necrosis scores within necrosis patterns revealed the lowest score in necrosis cluster-2 compared to the remaining subgroups (Figure 5D). Similarly, the lowest score was observed in gene cluster-B (Figure 5E).

Figure 5 Identification of the gene subtypes and construction of necrosis score. (A) Kaplan-Meier curves of the overall survival based on the high-necrosis score and low-necrosis score groups. (B) Alluvial diagram of subtype distributions in groups with different necrosis and survival outcomes. (C) The correlation analysis between the necrosis and immune cells, with red representing positive correlation and blue representing negative correlation. (D) Differential expression analysis of the necrosis score in the NRGs subtypes. (E) Difference analysis of the necrosis score in the gene subtypes. *, P<0.05. NRG, necrosis-related gene.

To elucidate the relationship between the necrosis score and various clinical features, including age, gender, grade, stage, and TNM stage, we investigated the distribution features of these clinical characteristics in the two necrosis score group. Our analysis revealed significant differences in grade and T stage between the two necrosis score groups (Figure 6A). Kaplan-Meier survival curves demonstrated that patients with low scores had a significantly favorable overall survival compared to those with high scores within the T1 and T2 stages and between the T3 and T4 stages (Figure 6B,6C).

Figure 6 Identification of the clinical characteristics and prediction of GC progression. (A) Differences in clinicopathologic features of high- and low-necrosis score groups of GC from the TCGA cohort. (B) Kaplan-Meier analysis of the survival rate of T1–2 between the high-necrosis score group and the low-necrosis score group in the TCGA cohort. (C) Kaplan-Meier analysis of the survival rate of T3–4 between the high-necrosis score group and the low-necrosis score group in the TCGA cohort. *, P<0.05; and P<0.05 were considered statistically significant. GC, gastric cancer; M, metastasis; N, node; T, tumor; TCGA, The Cancer Genome Atlas.

Tumor somatic mutation analysis between the high- and low-necrosis score groups

Emerging research suggests that TMB is a promising biomarker for predicting immune checkpoint blockade (ICB) efficacy. TMB has become a valuable biomarker across various tumor types to identify patients likely to benefit from immunotherapy (30). Our analysis of mutation data from the TCGA-stomach adenocarcinoma (STAD) cohort in GC samples revealed a lower TMB in the high-score group compared to the low-score group (Figure 7A), suggesting that patients in the low-risk group might benefit from immunotherapy. Spearman correlation analysis demonstrated a negative association between the necrosis score and TMB (Figure 7B). As shown in Figure 7C, the survival curve of TMB combined with the necrosis score indicated that patients in both the high tumor mutation group and the low-necrosis score group had the best prognosis. We subsequently analyzed the distribution variations of somatic mutations between the two necrosis score groups in the TCGA-colon adenocarcinoma (COAD) cohort. The top ten mutated genes in the high- and low-necrosis score groups were TTN, TP53, MUC16, ARID1A, LRP1B, SYNE1, FLG, FAT4, CSMD3, and PCLO. Notably, patients in the low-necrosis score group exhibited relatively higher mutation frequencies than those in the high-necrosis score group (96.43% and 85.61%, respectively) (Figure 7D,7E). Patients with a low-necrosis score had markedly higher frequencies of TTN, MUC16, and ARID1A mutations compared to those with a high-necrosis score.

Figure 7 Evaluation of the necrosis score and tumor somatic mutation. (A) Stratified analysis of the necrosis score for GC patients by TMB. (B) Spearman correlation analysis of the necrosis score and TMB. (C) Survival analysis among four groups of GC patients was layered according to both necrosis scores and TMB (log-rank tests, P<0.001). (D,E) The waterfall plot of somatic mutation features established with high- and low-necrosis scores. Each column represented an individual patient. The upper barplot showed TMB, and the number on the right indicated the mutation frequency in each gene. The right barplot showed the proportion of each variant type. GC, gastric cancer; H, high; L, low; TMB, tumor mutational burden.

Relationship of necrosis score with microsatellite instability (MSI) and microsatellite stable (MSS) index

MSI-high (MSI-H), present in less than 10% of GC patients, is a relatively favorable prognostic marker. Growing evidence suggests that patients with MSI-H are more responsive to immunotherapy and may benefit from this treatment approach (31). Correlation analyses revealed a significant association between a low-necrosis score and MSI-H, while a high-necrosis score was associated with MSS status (Figure 8A,8B). Thereafter, we examined the expression of ICIs in the high- and low-necrosis score cohorts. As shown in Figure 8C, a significant difference in PD-L1 expression was observed between the high- and low-necrosis score groups, with increased expression observed in the low-necrosis score group. Conversely, PD-1 expression was significantly decreased in the high-necrosis score group and was negatively associated with the necrosis score (Figure 8D,8E). Analysis of immunotherapy scores in the high- and low-necrosis score groups revealed that CTLA-4/PD-1 inhibitor-based ICI therapy served as a significant role in played a significant role in antineoplastic efficacy. Figure 8F,8G showed that CTLA-4 negative and PD-1 positive therapy, CTLA-4 immunotherapy, and PD-1 immunotherapy differed between the high- and low-necrosis score groups (P=1.2e−07).

Figure 8 Comprehensive analysis of the prognostic value according to necrosis scores. (A,B) Relationships between necrosis score, MSI, and MSS. (C) Correlations between necrosis score and PD-L1 expression. (D) Stratified analysis of the necrosis scores for GC patients by PD-1. (E) Correlations between necrosis score and PD-1 expression. (F) Differential analysis for low- and high-necrosis score in anti-PD-L1 combined with CTLA-4 immunotherapy. (G) Differential analysis for the low-necrosis score group and the high-necrosis score group in anti-PD-1 immunotherapy. CTLA-4, cytotoxic T-lymphocyte-associated protein 4; GC, gastric cancer; H, high; L, low; MSI, microsatellite instability; MSS, microsatellite stable; PD-1, programmed death-1; PD-L1, programmed death-ligand 1.

Immune analysis of necrosis score

As depicted in Figure 9A, patients with high-necrosis scores exhibited greater enrichment of B cells (naive and plasma), T cells (CD4 memory resting, regulatory, and activated), natural killer cells, monocytes, and activated mast cells compared to those with low-necrosis scores. Conversely, the low-necrosis group displayed higher levels of T cells (CD8, CD4 memory activated, follicular helper), natural killer cells (resting), and M1 macrophages. Additionally, the necrosis score was strongly correlated with the immune function score, with lower necrosis scores associated with a diverse range of immune markers (Figure 9B). The TME is now understood to be shaped by the interplay between tumor cells and various host immune cells, including T cells, B cells, infiltrating immune cells, and neutrophils (32). To delve deeper into the relationship between GC and its immune states, we categorized patients into four “immune subtypes” (C1, C2, C3, and C4) based on a previously described taxonomic sequence. As illustrated in Figure 9C, both low- and high-necrosis score groups were predominantly clustered within subtype C2. Research suggests that subtype C2, characterized by elevated IFN-γ levels, exhibits a high proliferation rate and is often associated with highly mutated GCs (10). The immunochemistry images showed that the protein expression levels of B-cell lymphoma 2 (BCL2), Janus kinase 2 (JAK2), and signal transducer and activator of transcription 3 (STAT3) were higher in GC tissues than in normal paracancerous tissues (Figure 10).

Figure 9 Comprehensive immune-related functions between the two necrosis groups. (A) The differential expression analysis of 22 immune cells among high- and low-necrosis score groups. (B) Immune function analysis of high- and low-necrosis score groups. (C) The difference in four immune subtypes between high- and low-necrosis score groups. ***, P<0.001; **, P<0.01; *, P<0.05. aDCs, activated dendritic cells; APC, antigen-presenting cell; CCR, C-C chemokine receptor; DCs, dendritic cells; HLA, human leukocyte antigen; iDCs, immature dendritic cells; IFN, interferon; IRGPI, immune-related gene prognostic index; MHC, major histocompatibility complex; NK, natural killer; pDCs, plasmacytoid dendritic cells; TCGA, The Cancer Genome Atlas; Tfh, T follicular helper; Th, T helper; TIL, tumor-infiltrating lymphocyte; Treg, T regulatory.

Figure 10 Protein expression of BCL2, JAK2, and STAT3 (hematoxylin and eosin staining; scale bars =10 µm). BCL2, B-cell lymphoma 2; JAK2, Janus kinase 2; STAT3, signal transducer and activator of transcription 3.

Discussion

Atypical early symptoms of GC often lead to delayed diagnoses, resulting in a 5-year survival rate of only 5–20% for patients with advanced disease (33,34). This heterogeneity in response to therapy highlights the need for improved prognostic markers and targeted treatment strategies (3). Next-generation sequencing represents a promising approach for classifying GC patients, enabling rapid identification of cancer characteristics and informing optimal treatment decisions. While PD-L1 holds potential as an immune marker, further validation is required (35). Similarly, the prognostic value of human epidermal growth factor receptor 2 (HER2) in GC remains controversial (36). The lack of well-defined molecular subtypes limits the ability to tailor clinical practice. Consequently, there is an urgent need to develop a robust gene signature that can accurately predict prognosis and guide the selection of appropriate treatments, particularly targeted therapies and immunotherapies.

Numerous studies have highlighted the critical role of necrosis in anti-tumor and innate immune responses. However, most research has been limited to examining individual NRGs or specific TME cell types. Consequently, the comprehensive effects and TME infiltration patterns resulting from the combined influence of multiple NRGs remain incompletely understood.

Herein, we identified three distinct necrosis patterns based on the expression levels of necrosis driver and suppressor genes in the B cohort. Compared to patients with clusters A and C, cluster B, characterized by high expression of most necrosis regulators, exhibited the most favorable prognosis. Furthermore, the GC subtypes were differentiated by the activation of significant immune pathways, including natural killer cell-mediated cytotoxicity, T cell receptor signaling, RIG-I-like Toll-like NOD-like receptor, and chemokine signaling. Furthermore, profiling of the TME revealed that the three NRG clusters diverged markedly in their immune cell infiltration characteristics. An analysis of infiltrating immune cells revealed that cluster B was enriched in B cells, CD4⁺ T cells, CD8⁺ T cells, natural killer cells, and regulatory T cells, corresponding to an immunologic inflammatory phenotype. Tumors are often categorized as “hot” or “cold” to reflect the degree and effectiveness of immune infiltration. The “immune-inflamed” phenotype, characterized by abundant T cells within the tumor parenchyma, defines “hot” tumors, although these T cells may be functionally suppressed. In contrast, “cold” tumors encompass both the “immune-desert” phenotype, which lacks TILs entirely, and the “immune-excluded” phenotype, where T cells are present at the tumor margin but fail to infiltrate the core. These “cold” phenotypes facilitate immune evasion and T cell escape (37,38). Consistent with the established role of stromal activation in fostering an immune-suppressive niche, NRG cluster-C was characterized by a prominent stroma-rich phenotype, evidenced by the enrichment of ECM-receptor interaction pathways. This environment is known to preclude effective T-cell infiltration and function. Consequently, this cluster displayed a compensatory activation of innate immunity and was associated with significantly worse survival outcomes. Our findings demonstrate that NRGs can serve as valuable predictors of clinical outcome and immunotherapy response in GC. Notably, we developed a robust prognostic score model that assesses the necrosis modification pattern and exhibits strong predictive power. We found that the necrosis-related pattern characterized by an immune-excluded phenotype conferred a high-necrosis score, which was associated with poor survival. Conversely, the pattern exhibiting an immune-inflamed phenotype was associated with a low-necrosis score and more favorable outcomes. The strong negative correlation between the necrosis score and immune cells (including CD4⁺ T cells, CD8⁺ T cells, and dendritic cells) helps to elucidate the infiltration patterns within the TME. Patients with low NRG scores demonstrated better survival, higher TMB, mutation rate, PD-L1 expression, and MSI-H status, while those with high scores exhibited the opposite characteristics.

Cancer biologists are exploring the possibility of inducing cell death through necrosis as an effective antineoplastic strategy. Unlike apoptosis, necrosis often triggers a robust inflammatory response, which may contribute to tumor regression during cancer treatment. To elucidate the potential mechanisms of necrosis-related proteins in the development and progression of GC, we examined 159 NRGs in GC tissues and identified 24 genes that were differentially expressed compared to normal tissues.

Chronic gastric mucosal infection can lead to the gradual development of intestinal metaplasia and atrophic gastritis, ultimately promoting the progression of GC (1). In various solid tumors, molecular signatures associated with distinct clinical outcomes have been identified, paving the way for personalized medicine (39-41). To improve therapeutic strategies, it is essential to investigate the changes in GC cell status and mechanisms influenced by the TME. The TME comprises a diverse array of cellular components, including immune cells (granulocytes, lymphocytes, macrophages) (42), blood vessels, tumor-infiltrating immune cells (TIICs), ECM, fibroblasts, and bone marrow-derived inflammatory cells (43). Extensive research has highlighted the significant impact of the TME on tumor development, progression, and therapeutic resistance (44). To comprehensively explore the relationship between distinct NRGs patterns and TME cell infiltration characteristics in GC, we conducted immune correlation analyses. Previous studies have demonstrated a direct association between the depth of tumor invasion, nodal status, clinical stage in GC, and the degree of TAM infiltration (45). In the present study, we observed marked differences in TME immune cell infiltration characteristics among the three gene patterns, primarily involving activated B cells, CD4⁺ T cells, CD8⁺ T cells, natural killer cells, and macrophage cells.

To further dissect the intratumoral immune landscape of GC, we compared four immune signature sets between the high- and low-necrosis score groups. These sets were derived from a comprehensive integrated genome-wide analysis of DNA copy number alterations, mutations, messenger RNA (mRNA), microRNA (miRNA), and protein patterns conducted by the TCGA research network (10). These four categories (C1, C2, C3, and C4) correlate molecular features with histological phenotypes and clinical features, suggesting that tailored treatment approaches may be applicable regardless of specific cancer subtypes.

ICB therapy has reportedly improved survival in various tumor types, including GC (35,46,47). A growing body of evidence suggests that a high TMB is a promising predictive biomarker for response to ICIs (48). PD-1, an immune checkpoint receptor on T cells, enables tumor cells to evade immune surveillance and become sensitive to immunotherapy (49). Recent studies indicate that a high TMB may correlate with a durable clinical response to anti-PD-1/PD-L1 immunotherapy. In our study, we observed a strong association between TMB and NRG scores. Furthermore, patients with low NRG scores exhibited higher TMB and favorable prognosis. In addition to TMB, MSI status and PD-L1 expression have been identified as promising biomarkers for predicting response to PD-1 blockade-based immunotherapy (50). We found that the low NRG score group exhibited higher PD-L1 expression levels, suggesting potential benefit from anti-PD-L1/PD-1 treatment. Moreover, MSI-H GC has been linked to longer overall survival and therapeutic efficacy with ICIs while showing limited benefit from perioperative or adjuvant chemotherapy. Notably, the proportion of patients with MSI-H was higher in the low-necrosis score group compared to the high-necrosis score group, further supporting a favorable prognosis. Collectively, these findings highlight the potential of NRGs as promising predictive biomarkers for GC immunotherapy.

In summary, patients with low NRG scores, exhibiting higher PD-L1 expression, TMB, mutation frequency, and MSI-H status, may derive the greatest benefit from immunotherapy. In clinical practice, the NRG score can be employed to comprehensively analyze necrosis patterns and associated immune cell infiltration characteristics within individual patients, facilitating more personalized treatment decisions. Additionally, the score can be used to assess patients’ immune status, including MSI status, TMB, PD-L1 expression, and immunotherapy response prediction. The necrosis score emerged as a valuable prognostic indicator in our cohort of GC patients. Furthermore, its association with key immunotherapeutic biomarkers suggests its potential relevance in predicting immunotherapy response. These findings warrant prospective clinical cohorts and experimental studies to assess their utility for advancing personalized immunotherapy in GC.

Nevertheless, there are certain limitations in this study that should be acknowledged. Firstly, the analyses were based on publicly available datasets, and all samples were obtained retrospectively. Consequently, prospective studies and additional in vivo and in vitro experiments are warranted to further validate our discoveries. Secondly, some crucial clinical variables, such as surgery, neoadjuvant chemotherapy, and chemoradiotherapy, were not available for analysis in the datasets, potentially influencing the prognosis and immune response.

Conclusions

In conclusion, our comprehensive integrated analysis of the effects of NRG on the infiltration of individual TMEs and immunotherapy in GC. The differences in NRG score may be tightly tied to the clinicopathological characteristics, immunotherapy, and prognosis of GC patients.

Acknowledgments

None.

Footnote

Reporting Checklist: The authors have completed the TRIPOD reporting checklist. Available at https://tcr.amegroups.com/article/view/10.21037/tcr-2025-1435/rc

Peer Review File: Available at https://tcr.amegroups.com/article/view/10.21037/tcr-2025-1435/prf

Funding: This work was supported by the Taizhou Science and Technology Plan Project (No. 24ywb14) and the Project of Shaanxi Provincial Administration of Traditional Chinese Medicine (No. SZY-KJCYC-2025-JC-033).

Conflicts of Interest: All authors have completed the ICMJE uniform disclosure form (available at https://tcr.amegroups.com/article/view/10.21037/tcr-2025-1435/coif). The authors have no conflicts of interest to declare.

Ethical Statement: The authors are accountable for all aspects of the work in ensuring that questions related to the accuracy or integrity of any part of the work are appropriately investigated and resolved. The study was conducted in accordance with the Declaration of Helsinki and its subsequent amendments.

Open Access Statement: This is an Open Access article distributed in accordance with the Creative Commons Attribution-NonCommercial-NoDerivs 4.0 International License (CC BY-NC-ND 4.0), which permits the non-commercial replication and distribution of the article with the strict proviso that no changes or edits are made and the original work is properly cited (including links to both the formal publication through the relevant DOI and the license). See: https://creativecommons.org/licenses/by-nc-nd/4.0/.

References

1. Smyth EC, Nilsson M, Grabsch HI, et al. Gastric cancer. Lancet 2020;396:635-48. [Crossref] [PubMed]
2. Ilson DH. Advances in the treatment of gastric cancer: 2020-2021. Curr Opin Gastroenterol 2021;37:615-8. [Crossref] [PubMed]
3. Bray F, Ferlay J, Soerjomataram I, et al. Global cancer statistics 2018: GLOBOCAN estimates of incidence and mortality worldwide for 36 cancers in 185 countries. CA Cancer J Clin 2018;68:394-424. [Crossref] [PubMed]
4. Correa P. Gastric cancer: overview. Gastroenterol Clin North Am 2013;42:211-7. [Crossref] [PubMed]
5. Yusefi AR, Bagheri Lankarani K, Bastani P, et al. Risk Factors for Gastric Cancer: A Systematic Review. Asian Pac J Cancer Prev 2018;19:591-603. [Crossref] [PubMed]
6. Machlowska J, Baj J, Sitarz M, et al. Gastric Cancer: Epidemiology, Risk Factors, Classification, Genomic Characteristics and Treatment Strategies. Int J Mol Sci 2020;21:4012. [Crossref] [PubMed]
7. Zhang XY, Zhang PY. Gastric cancer: somatic genetics as a guide to therapy. J Med Genet 2017;54:305-12. [Crossref] [PubMed]
8. Ribas A. Tumor immunotherapy directed at PD-1. N Engl J Med 2012;366:2517-9. [Crossref] [PubMed]
9. Freeman GJ, Long AJ, Iwai Y, et al. Engagement of the PD-1 immunoinhibitory receptor by a novel B7 family member leads to negative regulation of lymphocyte activation. J Exp Med 2000;192:1027-34. [Crossref] [PubMed]
10. Comprehensive molecular characterization of gastric adenocarcinoma. Nature 2014;513:202-9. [Crossref] [PubMed]
11. Schoenfeld AJ, Hellmann MD. Acquired Resistance to Immune Checkpoint Inhibitors. Cancer Cell 2020;37:443-55. [Crossref] [PubMed]
12. Sharma P, Hu-Lieskovan S, Wargo JA, et al. Primary, Adaptive, and Acquired Resistance to Cancer Immunotherapy. Cell 2017;168:707-23. [Crossref] [PubMed]
13. Yu J, Zheng W. An Alternative Method for Screening Gastric Cancer Based on Serum Levels of CEA, CA19-9, and CA72-4. J Gastrointest Cancer 2018;49:57-62. [Crossref] [PubMed]
14. Jin S, White E. Role of autophagy in cancer: management of metabolic stress. Autophagy 2007;3:28-31. [Crossref] [PubMed]
15. Degenhardt K, Mathew R, Beaudoin B, et al. Autophagy promotes tumor cell survival and restricts necrosis, inflammation, and tumorigenesis. Cancer Cell 2006;10:51-64. [Crossref] [PubMed]
16. Lotze MT, Tracey KJ. High-mobility group box 1 protein (HMGB1): nuclear weapon in the immune arsenal. Nat Rev Immunol 2005;5:331-42. [Crossref] [PubMed]
17. Kang R, Zhang Q, Zeh HJ 3rd, et al. HMGB1 in cancer: good, bad, or both? Clin Cancer Res 2013;19:4046-57. [Crossref] [PubMed]
18. Janko C, Filipović M, Munoz LE, et al. Redox modulation of HMGB1-related signaling. Antioxid Redox Signal 2014;20:1075-85. [Crossref] [PubMed]
19. Taguchi A, Blood DC, del Toro G, et al. Blockade of RAGE-amphoterin signalling suppresses tumour growth and metastases. Nature 2000;405:354-60. [Crossref] [PubMed]
20. Schlueter C, Weber H, Meyer B, et al. Angiogenetic signaling through hypoxia: HMGB1: an angiogenetic switch molecule. Am J Pathol 2005;166:1259-63. [Crossref] [PubMed]
21. Zhang J, Kou YB, Zhu JS, et al. Knockdown of HMGB1 inhibits growth and invasion of gastric cancer cells through the NF-κB pathway in vitro and in vivo. Int J Oncol 2014;44:1268-76. [Crossref] [PubMed]
22. Gajewski TF, Schreiber H, Fu YX. Innate and adaptive immune cells in the tumor microenvironment. Nat Immunol 2013;14:1014-22. [Crossref] [PubMed]
23. Hanahan D, Coussens LM. Accessories to the crime: functions of cells recruited to the tumor microenvironment. Cancer Cell 2012;21:309-22. [Crossref] [PubMed]
24. Baj J, Brzozowska K, Forma A, et al. Immunological Aspects of the Tumor Microenvironment and Epithelial-Mesenchymal Transition in Gastric Carcinogenesis. Int J Mol Sci 2020;21:2544. [Crossref] [PubMed]
25. Seeneevassen L, Bessède E, Mégraud F, et al. Gastric Cancer: Advances in Carcinogenesis Research and New Therapeutic Strategies. Int J Mol Sci 2021;22:3418. [Crossref] [PubMed]
26. Lotfi R, Kaltenmeier C, Lotze MT, et al. Until Death Do Us Part: Necrosis and Oxidation Promote the Tumor Microenvironment. Transfus Med Hemother 2016;43:120-32. [Crossref] [PubMed]
27. Lotfi R, Lee JJ, Lotze MT. Eosinophilic granulocytes and damage-associated molecular pattern molecules (DAMPs): role in the inflammatory response within tumors. J Immunother 2007;30:16-28. [Crossref] [PubMed]
28. Lotfi R, Schrezenmeier H, Lotze MT. Immunotherapy for cancer: promoting innate immunity. Front Biosci (Landmark Ed) 2009;14:818-32. [Crossref] [PubMed]
29. Liberzon A, Birger C, Thorvaldsdóttir H, et al. The Molecular Signatures Database (MSigDB) hallmark gene set collection. Cell Syst 2015;1:417-25. [Crossref] [PubMed]
30. Chan TA, Yarchoan M, Jaffee E, et al. Development of tumor mutation burden as an immunotherapy biomarker: utility for the oncology clinic. Ann Oncol 2019;30:44-56. [Crossref] [PubMed]
31. Ganesh K, Stadler ZK, Cercek A, et al. Immunotherapy in colorectal cancer: rationale, challenges and potential. Nat Rev Gastroenterol Hepatol 2019;16:361-75. [Crossref] [PubMed]
32. Ogino S, Galon J, Fuchs CS, et al. Cancer immunology--analysis of host and tumor factors for personalized medicine. Nat Rev Clin Oncol 2011;8:711-9. [Crossref] [PubMed]
33. Mizuguchi A, Takai A, Shimizu T, et al. Genetic features of multicentric/multifocal intramucosal gastric carcinoma. Int J Cancer 2018;143:1923-34. [Crossref] [PubMed]
34. Schlansky B, Sonnenberg A. Epidemiology of noncardia gastric adenocarcinoma in the United States. Am J Gastroenterol 2011;106:1978-85. [Crossref] [PubMed]
35. Shitara K, Özgüroğlu M, Bang YJ, et al. Pembrolizumab versus paclitaxel for previously treated, advanced gastric or gastro-oesophageal junction cancer (KEYNOTE-061): a randomised, open-label, controlled, phase 3 trial. Lancet 2018;392:123-33. [Crossref] [PubMed]
36. Yoon HH, Shi Q, Sukov WR, et al. Association of HER2/ErbB2 expression and gene amplification with pathologic features and prognosis in esophageal adenocarcinomas. Clin Cancer Res 2012;18:546-54. [Crossref] [PubMed]
37. Chen DS, Mellman I. Elements of cancer immunity and the cancer-immune set point. Nature 2017;541:321-30. [Crossref] [PubMed]
38. Gajewski TF, Woo SR, Zha Y, et al. Cancer immunotherapy strategies based on overcoming barriers within the tumor microenvironment. Curr Opin Immunol 2013;25:268-76. [Crossref] [PubMed]
39. Li T, Kung HJ, Mack PC, et al. Genotyping and genomic profiling of non-small-cell lung cancer: implications for current and future therapies. J Clin Oncol 2013;31:1039-49. [Crossref] [PubMed]
40. Roepman P, Schlicker A, Tabernero J, et al. Colorectal cancer intrinsic subtypes predict chemotherapy benefit, deficient mismatch repair and epithelial-to-mesenchymal transition. Int J Cancer 2014;134:552-62. [Crossref] [PubMed]
41. Mohamed A, Krajewski K, Cakar B, et al. Targeted therapy for breast cancer. Am J Pathol 2013;183:1096-112. [Crossref] [PubMed]
42. Song W, Ren J, Xiang R, et al. Identification of pyroptosis-related subtypes, the development of a prognosis model, and characterization of tumor microenvironment infiltration in colorectal cancer. Oncoimmunology 2021;10:1987636. [Crossref] [PubMed]
43. Turley SJ, Cremasco V, Astarita JL. Immunological hallmarks of stromal cells in the tumour microenvironment. Nat Rev Immunol 2015;15:669-82. [Crossref] [PubMed]
44. Hinshaw DC, Shevde LA. The Tumor Microenvironment Innately Modulates Cancer Progression. Cancer Res 2019;79:4557-66. [Crossref] [PubMed]
45. Ishigami S, Natsugoe S, Tokuda K, et al. Tumor-associated macrophage (TAM) infiltration in gastric cancer. Anticancer Res 2003;23:4079-83.
46. Hugo W, Zaretsky JM, Sun L, et al. Genomic and Transcriptomic Features of Response to Anti-PD-1 Therapy in Metastatic Melanoma. Cell 2016;165:35-44. [Crossref] [PubMed]
47. Ferris RL, Blumenschein G Jr, Fayette J, et al. Nivolumab for Recurrent Squamous-Cell Carcinoma of the Head and Neck. N Engl J Med 2016;375:1856-67. [Crossref] [PubMed]
48. McGrail DJ, Pilié PG, Rashid NU, et al. High tumor mutation burden fails to predict immune checkpoint blockade response across all cancer types. Ann Oncol 2021;32:661-72. [Crossref] [PubMed]
49. Wu X, Gu Z, Chen Y, et al. Application of PD-1 Blockade in Cancer Immunotherapy. Comput Struct Biotechnol J 2019;17:661-74. [Crossref] [PubMed]
50. Wang F, Wei XL, Wang FH, et al. Safety, efficacy and tumor mutational burden as a biomarker of overall survival benefit in chemo-refractory gastric cancer treated with toripalimab, a PD-1 antibody in phase Ib/II clinical trial NCT02915432. Ann Oncol 2019;30:1479-86. [Crossref] [PubMed]