In this study, we investigated the association between necroptosis-related gene abundance and immune resistance, aiming to uncover novel insights into their role in cancer immunotherapy. Using GSVA, we computed necroptosis scores across the TCGA pan-cancer cohort, revealing a significant positive correlation between necroptosis scores and immune-related gene expression across 30 distinct cancer types ( Figure 1A ). These findings suggest a broad impact of necroptosis on immune modulation within the tumor microenvironment (TME).To further understand this impact, we examined immune cell infiltration in tumors with high necroptosis scores, observing a notable increase in immune cell presence, underscoring a link between necroptosis and immune activation. Importantly, our analysis also revealed positive correlations between necroptosis scores and both intratumor heterogeneity (ITH) and tumor mutational burden (TMB) ( Figures 1D, E ), suggesting that necroptosis influences anti-tumor immune responses by modulating immune cell activity and regulating tumor heterogeneity. These results provide new evidence that tumors with elevated necroptosis scores display enhanced anti-tumor immune responses, positioning necroptosis as a potential predictive biomarker for cancer immunotherapy efficacy. Our findings thus contribute to the understanding of necroptosis as a mechanism influencing immune resistance, with implications for identifying patients who may benefit most from immunotherapeutic interventions.

Acknowledging the critical role of necroptosis in orchestrating anti-tumor immune responses, we developed a predictive model based on necroptosis marker genes to optimize immunotherapy strategies. RNA-Seq data and clinical records from ten immune checkpoint inhibitor (ICI) cohorts were collected. Using the ABESS algorithm, five key necroptosis marker genes (ANKRD28, CREB3L2, ISG20, SLAMF7, MEI1) were identified. The workflow for model development is depicted in Figure 2A , where six different machine learning algorithms were employed, utilizing 10-fold cross-validation and grid search for parameter optimization to generate prediction models. Among these models, the Random Forest (RF) model exhibited the best performance, with an AUC of 0.713 ( Figure 2B ). Subsequently, the model was validated on the validation and independent testing sets, yielding AUCs of 0.71 and 0.74, respectively ( Figure 2C ). To further assess the model’s predictive capacity for overall survival (OS), ICI-treated patients were stratified into high-risk and low-risk groups. Kaplan-Meier survival analysis demonstrated that the low-risk group exhibited significantly prolonged OS in both the validation and testing sets (P < 0.01) ( Figure 2C , right).

We evaluated the predictive power of the necroptosis gene signature (Necroptosis.Sig) against other immune checkpoint inhibitor (ICI)-related gene signatures. In a direct comparison with a spectrum of pan-cancer signatures, including INFG.Sig (26), T.cell.inflamed.Sig (26), PDL1.Sig (27), LRRC15.CAF.Sig (28), NLRP3.Sig (29), and Cytotoxic.Sig (30), Necroptosis.Sig emerged as the most effective predictor in the testing set, achieving an AUC of 0.74, with INFG.Sig trailing closely behind at an AUC of 0.66 ( Figure 3A ). Necroptosis.Sig demonstrated superior performance across all evaluated cohorts, spanning three distinct cancer types: melanoma (SKCM), glioblastoma (GBM), and gastric cancer (GC), underscoring its versatility as a predictive model for ICI responsiveness across various malignancies ( Figure 3B ). When compared with signatures specifically tailored for melanoma (CRMA.Sig, IMPRES.Sig, IPRES.Sig, TcellExc.Sig, ImmunCells.Sig, IMS.Sig, and TRS.Sig), Necroptosis.Sig remained a leading predictor, achieving an AUC of 0.72 in predicting ICI response in melanoma patients. However, IMPRES.Sig and CRMA.Sig outperformed Necroptosis.Sig in this subset, registering slightly higher AUCs of 0.81 and 0.77, respectively ( Figure 3C ).

Necroptosis, a unique form of programmed cell death distinct from classical apoptosis, is driven by the activation of key molecular pathways involving RIPK1, RIPK3, and MLKL. Unlike apoptosis, necroptosis causes cell membrane rupture, releasing various immune-stimulating factors that trigger an inflammatory response. This distinct role of necroptosis in the TME has drawn considerable attention in recent cancer research. To further elucidate the impact of the Necroptosis Gene Signature on tumor immune evasion and anti-tumor immune responses, we conducted an in-depth analysis. Our findings, visualized in a heatmap ( Figure 4A ), highlight the association between necroptosis-related gene sets and necroptosis risk across diverse cancer types. These insights underscore the potential of necroptosis as a regulatory mechanism within the TME, advancing our understanding of how necroptosis may influence immune resistance and support novel strategies for enhancing cancer immunotherapy.

The gene modules are classified into different functional groups (C1 to C8), and patients are divided into high-risk and low-risk groups based on necroptosis risk scores. Notably, these gene sets are primarily involved in key biological pathways such as cell cycle regulation, chromosome segregation, and metabolic processes, indicating a close relationship between necroptosis and cellular proliferation and metabolism. Next, t-SNE analysis ( Figure 4B ) further demonstrates the distribution of necroptosis risk scores across various cancer types, including acute myeloid leukemia, renal cell carcinoma, and antigen processing pathways. Necroptosis enhances the immune system’s ability to recognize tumor cells by promoting antigen presentation, thereby influencing anti-tumor immune responses. Through color coding (ranging from low to high risk), the relationship between necroptosis and key biological functions, such as protein modification, immune response, and cellular stress adaptation, is clearly displayed. Additionally, the network diagram ( Figure 4C ) highlights the complex interactions between gene modules involved in cell cycle control and chromosome segregation, suggesting that necroptosis not only affects cell death but may also regulate cancer progression by altering cellular proliferation and differentiation. At the same time, GO enrichment analysis ( Figure 4D ) reveals a strong association between necroptosis genes and immune responses, particularly in terms of leukocyte activation, cytokine signaling, and inflammation regulation. This further demonstrates the critical role of necroptosis in triggering anti-tumor immune activity. Finally, the GSEA bar graph ( Figure 4E ) shows the enrichment of necroptosis-related pathways, including cytokine-cytokine receptor interaction, neutrophil extracellular trap formation, and Th1/Th2/Th17 cell differentiation, all of which are immune-related processes. These results collectively emphasize the central role of necroptosis in immune cell activation, differentiation, and tumor immunity, further validating its potential as a therapeutic target in cancer immunotherapy.

The relationship between necroptosis-related gene signatures and immune infiltration across various cancer types is analyzed to highlight novel findings and their implications in cancer immunotherapy. We explored how Necroptosis.Sig correlates with immune infiltration across diverse cancer types, aiming to understand its impact on the immune microenvironment. Figure 5A presents a bar graph illustrating the distribution of high-risk and low-risk groups based on Necroptosis.Sig across multiple cancer types. This analysis reveals that certain cancers, such as uveal melanoma and uterine carcinosarcoma, exhibit a higher proportion of high-risk patients, whereas others, like low-grade glioma, predominantly feature low-risk patients, suggesting necroptosis plays distinct roles across different cancers. In Figure 5B , violin plots compare immune-related metrics between high-risk and low-risk groups, showing significantly elevated scores in the low-risk group for leukocyte fraction, lymphocyte fraction, tumor-infiltrating lymphocyte (TIL) fractions, and CD8+ T cell infiltration. These findings indicate a more active immune system in the low-risk group. Figure 5C expands on this with boxplots showing that high-risk groups display markedly lower immune signature scores across multiple immune cell types, particularly antigen-presenting cells (APCs), B cells, and CD8+ T cells, implying a weaker immune response in the high-risk group. Further validation through Danaher et al.’s scoring analysis in Figure 5D reinforces these results, as the low-risk group shows significantly higher immune activity in cell populations such as B cells, dendritic cells, macrophages, and CD8+ T cells, emphasizing a robust immune presence. Additionally, the heatmap in Figure 5E clusters immune cells by risk group, revealing strong associations between the low-risk group and immune cell activity, particularly in Th1 cells, dendritic cells, and NK cells, thus supporting the view that the low-risk group has a more active immune microenvironment. Finally, Figure 5F summarizes immune infiltration differences between the two groups, illustrating that 59.66% of high-risk patients have low immune infiltration, while 65.2% of low-risk patients demonstrate high immune infiltration. This comprehensive analysis highlights that Necroptosis.Sig is inversely related to immune infiltration, with higher necroptosis risk correlating with reduced immune presence and potentially influencing tumor immune evasion and patient prognosis. This reorganization underscores our key finding that Necroptosis.Sig’s relationship with immune infiltration varies by cancer type and risk group, contributing novel insights to the field of cancer immunotherapy.

Next, this study delves into the relationship between high-risk and low-risk groups, as defined by Necroptosis.Sig, in terms of genomic and immunological characteristics. Figure 6A compares the mutation rate, neoantigen burden, T-cell receptor (TCR) and B-cell receptor (BCR) diversity, aneuploidy score, and intratumor heterogeneity between the two groups. The results show that the low-risk group has higher TCR and BCR diversity, indicating a more diverse immune repertoire, while the high-risk group exhibits greater genomic instability, with higher aneuploidy scores and intratumor heterogeneity. Figure 6B presents a heatmap of the differential expression of MHC-I and MHC-II molecules, as well as immune checkpoint inhibitors, with the low-risk group showing higher MHC expression, and the high-risk group displaying increased immune checkpoint expression, suggesting potential immune evasion mechanisms in the high-risk cohort. Figure 6C illustrates mutation signatures related to DNA repair defects, which are more pronounced in the high-risk group. Figure 6D confirms these findings by showing that mutation signatures associated with polymerase epsilon mutations, UV exposure, DNA mismatch repair defects, and homologous recombination defects are significantly higher in the high-risk group. Figures 6E, F compare the activity of several oncogenic pathways, such as Cell Cycle, MYC, HIPPO, TGF-beta, and WNT, with the high-risk group showing higher activity in these pathways, suggesting more aggressive tumor behavior and potential therapeutic vulnerabilities. Overall, this figure indicates that high-risk patients with elevated Necroptosis.Sig exhibit greater genomic instability, more active oncogenic pathways, and increased expression of immune checkpoint molecules, contributing to a more aggressive and immunosuppressive tumor microenvironment.

To further investigate the role of HMGB1, one of the modeling genes of Necroptosis.Sig, in A549 lung cancer cells and animal experiments, we conducted a series of analyses. The results showed that knocking down HMGB1 significantly reduced its relative expression levels ( Figure 7A ) and inhibited cell proliferation. In the CCK-8 assay, cells with HMGB1 knockdown exhibited a significantly slower proliferation rate at different time points ( Figures 7B, C ). The EdU staining assay further demonstrated that the percentage of EdU-positive cells was significantly reduced in the HMGB1 knockdown group ( Figures 7D, E ), confirming the suppression of cell proliferation. Meanwhile, flow cytometry analysis indicated that the apoptosis rate was significantly increased in HMGB1 knockdown cells ( Figures 7F, G ). In the mouse xenograft model, tumors in the HMGB1 knockdown group were noticeably smaller, and tumor growth was significantly slower ( Figures 7H, I ). Furthermore, Kaplan-Meier survival curves showed that mice in the HMGB1 knockdown group had significantly improved survival rates ( Figure 7J ). Overall, these results suggest that HMGB1 plays a critical role in tumor cell proliferation, apoptosis, and tumor growth, further supporting its importance as a Necroptosis.Sig modeling gene in tumor development and immune evasion.

We retrieved multi-omics datasets from the TCGA Pan-cancer collection via UCSC Xena (https://xenabrowser.net/datapages/) to explore the relationship between Necroptosis.Sig and immunosuppression across cancers. DLBC, LAML, and THYM were excluded due to their high immune cell content, which could introduce bias. The necroptosis gene sets were obtained from the GSEA database (https://www.gsea-msigdb.org/gsea/index.jsp). To address the potential batch effects arising from different tumor cohorts within the TCGA dataset, we applied the ComBat algorithm from the sva package (31), which effectively adjusts for batch effects while preserving biological variation across cancer types.

To validate the predictive capability of Necroptosis.Sig, we systematically collected transcriptomic data and clinical information from 10 pretreated ICI RNA-Seq cohorts. These cohorts include 5 melanoma (SKCM) cohorts (32–36), 2 urothelial carcinoma (UC) cohorts (37, 38), 1 glioblastoma multiforme (GBM) cohort (39), 1 gastric cancer (GC) cohort (40), and 1 renal cell carcinoma (RCC) cohort (41). Anti-PD-1, anti-PD-L1, anti-CTLA-4, and anti-PD-(L)1 + anti-CTLA-4 combination therapies were applied to 6, 2, 1, and 1 cohort, respectively. The Hugo 2016 cohort contains 27 pretreated tumor samples from 26 patients (32), while the Zhao 2019 cohort includes 34 pretreated tumor samples from 17 patients (39).

Spearman correlation analysis was used to evaluate the relationship between immune checkpoint gene expression and different immune functions. GSVA (Gene Set Variation Analysis) (42) was employed to assess the activity of immune-related signaling pathways (e.g., interferon gamma response, IL6-JAK-STAT3 signaling) across various cancer types. Additionally, the relationship between the median necroptosis GSVA score and tumor mutation burden (TMB), as well as intratumor heterogeneity (ITH), was analyzed, and linear regression trend lines were plotted on scatter plots.

We compiled RNA-Seq and clinical data from 10 Immune Checkpoint Inhibitor (ICI) RNA-Seq datasets, covering five cutaneous melanoma datasets, two urothelial carcinoma datasets, one glioblastoma dataset, one gastric cancer dataset, and one renal cell carcinoma dataset. Different immunotherapies targeting PD-1, CTLA-4, and PD-L1 were administered across cohorts. To address batch effects, we used the ComBat-seq method.

The primary outcomes were Objective Response Rate (ORR) and Overall Survival (OS), with ORR evaluated using RECIST v1.1 for most cohorts, except one, which used irRECIST guidelines. Patients were categorized as responders (Complete/Partial Response) or non-responders (Stable/Progressive Disease).

We developed a predictive signature for the ICI dataset by creating a combined cohort of 772 samples from five ICI RNA-Seq datasets. This cohort was split into a training set (80%, n=618) and a validation set (20%, n=154), with the remaining datasets (n=149) used as an independent test set. After training various models, the top-performing one was selected from the training set and subsequently tested on both the validation and independent datasets to evaluate its predictive performance. The Necroptosis.Sig signature was then compared to six pan-cancer and seven melanoma-specific signatures for its accuracy in predicting ICI response within the test set. Additionally, melanoma-specific signatures were further assessed using two separate melanoma cohorts.

Using the abess algorithm, we identified key necroptosis marker genes. Six machine learning models (SVM, Naive Bayes, Random Forest, KNN, AdaBoost, XGBoost) were built with 10-fold cross-validation and grid search for tuning (43).

We performed gene set enrichment analysis (GSEA) and ssGSEA using MSigDB, GSVA, and clusterprofiler R packages to analyze functional and immune characteristics (44, 45). To assess immune infiltration, CIBERSORT was used to estimate the abundance of 22 immune cell subsets in the tumor microenvironment from normalized transcriptomic data (46). Tumor-infiltrating lymphocytes (TILs) were evaluated using both genomic and histopathological data, while lymphocyte fractions were estimated based on CIBERSORT results and DNA methylation profiles. Immune infiltration scores for 14 immune cell types were calculated using a 60-marker gene signature and validated through immunohistochemistry and flow cytometry. We also assessed 29 immune signatures, quantifying enrichment levels across individual samples with the ssGSEA method. Additionally, immunogenomic indicators such as intratumor heterogeneity (ITH), TCR, and BCR diversity were derived from previous studies, utilizing the ABSOLUTE algorithm for copy number aberrations and Shannon entropy for receptor diversity.

Protocol for A549 lung cancer mouse cell culture: Begin by preparing necessary materials and sterilizing the clean bench. For cell revival, quickly thaw the cryogenic tube, mix it with preheated medium, centrifuge, and replace with fresh medium. Cultivate the cells in a CO₂ incubator. When cell density reaches 80%-90%, digest with trypsin, centrifuge, resuspend the cells, and passage them at a 1:2 ratio, continuing cultivation in the CO₂ incubator. For cryopreservation, wash and digest the cells, centrifuge, add cryopreservation solution, and gradually freeze before storage.

Using a transfection reagent, the constructed plasmid is introduced into the target cell line. The cells are then cultivated in medium containing puromycin to select for and eliminate non-transfected cells. The resulting cells, which stably express the specific shRNA, constitute the stable cell line. Lastly, these cells are expanded and characterized to confirm the gene silencing effect. The target sequences for HMGB1 were as follows: sh#1, CCGTTATGAAAGAGAAATGAA; and sh#2, CCCAGATGCTTCAGTCAACTT.

Total RNA was successfully extracted using TRIzol reagent from Sigma-Aldrich. Following this, cDNA was synthesized utilizing the TOYOBO reverse transcription kit. In the qRT-PCR procedure, we employed SYBR Green reagent from Applied Biosystems and selected GAPDH as the internal reference gene. The relative expression of genes was calculated using the 2^-ΔΔCt method. All specific primers are listed in detail in the supplementary table, ensuring the reproducibility of the experiment. The primer sequences used are as follows: GAPDH: Forward Sequence 5’-GTCTCCTCTGACTTCAACAGCG-3’, Reverse Sequence 5’-ACCACCCTGTTGCTGTAGCCAA-3’. HMGB1: Forward Sequence 5’-GCGAAGAAACTGGGAGAGATGTG-3’, Reverse Sequence 5’-GCATCAGGCTTTCCTTTAGCTCG-3’.

Collect cell samples, perform appropriate treatment, stain with fluorescently labeled apoptosis detection reagents (Annexin V combined with PI), incubate in the dark for 25 minutes, and then analyze the fluorescent signals of the cell population using flow cytometry to distinguish between live cells, early apoptotic cells, late apoptotic/necrotic cells, thereby quantitatively assessing the level of apoptosis.

The 6-8 weeks-old female Balb/c-nu mice were purchased from Beijing Vital River Laboratory Animal Technology Co., Ltd. After digestion, the cells are resuspended in sterile saline and adjusted to an appropriate concentration. Subsequently, subcutaneous injections are administered at the lateral side of the mouse’s back at a dose of 5x10⁵ cells per mouse, ensuring the cells are accurately injected into the subcutaneous tissue. Following injection, the mice are observed regularly for their growth status and tumor development, with attention also paid to their daily care. Upon reaching the observation endpoint, the mice are euthanized by cervical dislocation.

Data analysis was conducted in R 4.3.1, using the Wilcoxon rank-sum test for continuous variables and Spearman correlation for relationships. ROC curves and Kaplan-Meier survival analyses were performed using the pROC and survival packages, with p < 0.05 considered statistically significant.

All mice were kept in a specific pathogen-free environment and supplied with sterilized drinking water. All animal procedures were conducted in accordance with the ARRIVE guidelines and were approved by the Animal Ethics Committee of the Affiliated Huai’an Hospital of Xuzhou Medical University.

This study highlights the potential role of necroptosis in predicting immune checkpoint inhibitor (ICI) responses, opening new avenues for further research. Future directions will involve comprehensive in vitro and in vivo experiments to validate the regulatory roles of NecroticApop.Sig genes in necroptosis and tumor immunity, particularly within diverse tumor microenvironments. However, several limitations warrant consideration. Firstly, our findings are primarily based on bioinformatics analyses, lacking in-depth experimental validation, which may limit the biological insights derived from the study. Extensive experimental research is necessary to confirm the accuracy and applicability of NecroticApop.Sig in different tumor contexts. Secondly, although NecroticApop.Sig shows promising predictive accuracy, it has not yet been validated with independent cohorts, and current findings may be constrained by sample representativeness and dataset diversity. To ensure the model’s robustness and broader applicability, larger and more diverse independent datasets are needed for further validation. Additionally, due to the complexity of tumor immunity and necroptosis, the clinical operability and predictive power of NecroticApop.Sig may be limited by individual patient variations and tumor heterogeneity. To enhance its clinical applicability and generalizability, we plan to collect broader independent datasets, conduct prospective studies, and perform refined analyses across various tumor types and patient subgroups. These efforts will contribute to strengthening the practical utility of NecroticApop.Sig in clinical settings.