GRB10 as a pan-cancer biomarker linking oncogenic signaling, immunosuppression, and immunotherapy resistance Authors: Haoran Guo, Junjie Chen Published: 2025 ABSTRACT ObjectiveGrowth Factor Receptor-Binding Protein 10 (GRB10) is an adaptor protein implicated in tyrosine kinase signaling, yet its pan-cancer role and clinical impact remain incompletely characterized. This study aims to define the pan-cancer landscape of GRB10 dysregulation and its clinical implications for prognosis and immunotherapy response.MethodsMulti-omics analysis of 33 TCGA cancers and validation in GEO cohorts assessed GRB10 expression, prognostic impact (Cox regression, Kaplan-Meier), functional enrichment (Gene Set Enrichment Analysis, Gene Set Variation Analysis), immune correlates (Spearman with immune genes, ESTIMATE/CIBERSORT/EPIC/TIMER/xCell infiltration), tumor mutational burden (TMB), and immunotherapy predictive power.ResultsIn the TCGA pan-cancer cohort, GRB10 was significantly elevated in 11 cancer types (CHOL, COAD, ESCA, HNSC, KIRC, KIRP, LIHC, LUAD, LUSC, STAD, THCA) and downregulated in 4 (BLCA, BRCA, CESC, UCEC). Aberrant GRB10 expression was strongly associated with adverse prognosis in LGG, CESC, COAD, and LUAD for OS, DSS, and PFI, a finding validated externally in GEO colon cancer cohorts. Conversely, it served as a protective factor in KIRC. Functionally, GRB10 was implicated in epithelial-mesenchymal transition (EMT), evidenced by positive correlations with EMT pathway scores and key regulators (ZEB1, ZEB2, SNAI1, SNAI2). GRB10 expression robustly correlated with an immunosuppressive tumor microenvironment (TME), evidenced by negative enrichment of immune response pathways (e.g., IFN-α/γ) and complex associations with immune-related genes. Immune infiltration analysis revealed consistent positive correlations with CD4⁺ memory-resting T cells and negative correlations with CD4⁺ effector memory T cells across most cancers. Critically, high GRB10 expression predicted significantly shorter survival and poorer response rates in multiple immunotherapy-treated cohorts (urothelial carcinoma, melanoma, gastric cancer). GRB10 also showed significant associations with TMB in several cancers, and its protein interaction network was enriched in PI3K-Akt, FoxO, and Rap1 signaling pathways.ConclusionOur pan-cancer analysis establishes GRB10 as a key facilitator of tumor progression, linked to EMT and an immunosuppressive microenvironment, and nominates it as a novel biomarker for adverse prognosis and resistance to immune checkpoint blockade therapy.Supplementary InformationThe online version contains supplementary material available at 10.1007/s12672-025-04333-x. Introduction Cancer remains a paramount global health challenge of the 21st century, accounting for nearly one-sixth of all deaths worldwide [1]. Although advances in targeted therapies, including immune checkpoint inhibitors (ICIs), have improved outcomes for some patients [2, 3], therapeutic resistance driven by tumor heterogeneity and the immunosuppressive tumor microenvironment (TME) continues to pose a major obstacle [4]. Identifying novel biomarkers and regulatory genes capable of predicting response to immunotherapy is therefore essential for improving treatment efficacy. Growth Factor Receptor-Binding Protein 10 (GRB10) encodes a multifunctional adaptor protein containing Pleckstrin Homology (PH) and Src Homology 2 (SH2) domains, which are critical for signal transduction downstream of receptor tyrosine kinases. It influences cell proliferation, differentiation, and metabolism by interacting with various tyrosine kinase receptors and other phosphoprotein partners [5–7]. Emerging evidence suggests context-dependent roles for GRB10 in oncogenesis. For example, it acts as an oncogene in gastric cancer, where high expression correlates with poor survival and its knockdown inhibits proliferation and migration [8]. In osteosarcoma, Rab Interacting Lysosomal Protein (RILP) suppresses tumor progression by inhibiting the GRB10-mediated PI3K/AKT/mTOR pathway [9]. Conversely, in prostate cancer, GRB10 promotes tumor growth by sustaining androgen receptor (AR) activity through interaction with the PP2A complex [10]. GRB10 also demonstrates potential oncogenic or tumor-suppressive functions in other cancers, such as in small cell lung cancer [11], acute myeloid leukemia [12], skin cancer [13], and glioma [14], where its expression has been linked to hypoxia and epithelial-mesenchymal transition (EMT) signatures. These collective findings suggest that GRB10 may serve as a prognostic biomarker and therapeutic target, yet most current studies are confined to single cancer types. Conventional cancer research often focuses on individual tumor types, which can limit the discovery of fundamental oncogenic mechanisms. The advent of large-scale consortia such as The Cancer Genome Atlas (TCGA) has enabled pan-cancer analysis, a systematic approach to comparing molecular data across diverse malignancies [15]. By integrating multi-omics profiles, this strategy transcends tissue-specific contexts to reveal conserved drivers of tumorigenesis, including shared genetic alterations, pathway dysregulation, and microenvironmental remodeling. Pan-cancer analysis thus provides a powerful framework for discriminating functional drivers from passenger events and for identifying biomarkers and therapeutic targets with broad clinical relevance [16, 17]. Recently, a bioinformatics study by Yang et al. identified GRB10 as dysregulated across both depression and pan-cancer, highlighting its potential broader disease relevance [18]. While this work offers valuable initial descriptive evidence, a deeper mechanistic understanding of GRB10’s role in oncogenic processes, its interplay with the tumor immune microenvironment, and its impact on immunotherapy efficacy remains largely unexplored. Leveraging TCGA and Gene Expression Omnibus (GEO) datasets, we systematically delineate the pan-cancer landscape of GRB10 dysregulation and evaluate its association with prognosis, functional pathways, immune infiltration, tumor mutational burden, and immunotherapy response. Our integrated analysis aims to establish GRB10 as a potential pan-cancer regulator and biomarker with implications for targeted and immune-based treatment strategies. Materials and methods Data acquisition and processing Transcriptome data, somatic mutation data, and corresponding clinical information for 33 cancer types from The Cancer Genome Atlas (TCGA) were acquired using the R package TCGAbiolinks (v2.34.0) [19]. Detailed prognostic endpoints, including Disease-Specific Survival (DSS), Disease-Free Interval (DFI), and Progression-Free Interval (PFI), were obtained from the UCSC XENA platform. To ensure data quality, we retained only primary tumor samples and excluded Formalin-Fixed Paraffin-Embedded (FFPE) specimens due to their potentially compromised sequencing quality [20]. Independent colon cancer gene expression datasets (GSE18105, GSE44076, GSE44861, GSE17536, GSE41258) were downloaded from the Gene Expression Omnibus (GEO) using GEOquery (v2.62.2) [21] and normalized with the limma package (v3.40.0) [22]. Data from immunotherapy-treated cohorts were sourced from the BEST database [23]. The GRB10 protein-protein interaction network was constructed using GeneMANIA [24]. Gene set enrichment and variation analysis Gene Set Enrichment Analysis (GSEA) was conducted with the clusterProfiler package (v4.10.1) [25]. Kyoto Encyclopedia of Genes and Genomes (KEGG) and Gene Ontology (GO) over-representation analyses (ORA) were performed using the enrichKEGG and enrichGO functions, respectively. For both GSEA and ORA, the False Discovery Rate (FDR) was controlled using the Benjamini-Hochberg (BH) procedure, and an adjusted p-value of less than 0.05 was defined as statistically significant. Hallmark and EMT-related gene sets were retrieved from the Molecular Signatures Database (MSigDB, v2024.1) [26]. Specifically, GOBP_EPITHELIAL_TO_MESENCHYMAL_TRANSITION and GOBP_MESENCHYMAL_TO_EPITHELIAL_TRANSITION were sourced from the C5:GO collection. Gene Set Variation Analysis (GSVA) was carried out using the GSVA package (v1.50.5) [27] to compute sample-level enrichment scores for the specified gene sets. Immune microenvironment characterization The ESTIMATE algorithm (v1.0.13) was employed to calculate immune, stromal, and estimate scores via its estimateScore function [28]. The abundances of specific immune cell populations were estimated using the IOBR package (v0.99.0) [29], which integrates multiple deconvolution methodologies, including CIBERSORT, EPIC, TIMER, and xCell, through its deconvo_tme function. Somatic mutation analysis Tumor Mutational Burden (TMB) was calculated as the total number of mutations per megabase from TCGA somatic mutation MAF files using the maftools package (v2.18.0) [30]. The mutation frequency and types of GRB10 across different cancers were summarized and visualized via the cBioPortal platform [31]. Statistical analysis All statistical analyses were performed in the R software(v4.3.0). Comparisons between two groups for continuous variables were assessed using the independent Student’s t-test for normally distributed data and the Wilcoxon rank-sum test for non-normally distributed data. For comparisons across three or more groups, one-way ANOVA or the Kruskal-Wallis test was applied as appropriate. Univariate and multivariate Cox proportional hazards regression models, along with Kaplan-Meier survival analysis, were implemented using the survival package (v3.5.5). Time-dependent receiver operating characteristic (ROC) analysis was conducted using the timeROC package (v0.4). Correlations between variables were evaluated using Spearman’s rank correlation coefficient. To account for multiple hypothesis testing in large-scale pan-cancer analyses, the FDR was controlled using the BH method within the context of each specific analysis. A result was considered significant if the FDR-corrected q-value was below 0.05. Significance levels are denoted as follows: *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001. Results Pan-cancer dysregulation and prognostic impact of GRB10 We analyzed GRB10 expression in 21 TCGA cancer types with at least five matched normal samples. GRB10 was significantly upregulated in 11 cancer types, including cholangiocarcinoma (CHOL), colon adenocarcinoma (COAD), esophageal carcinoma (ESCA), head and neck squamous cell carcinoma (HNSC), kidney renal clear cell carcinoma (KIRC), kidney renal papillary cell carcinoma (KIRP), liver hepatocellular carcinoma (LIHC), lung adenocarcinoma (LUAD), lung squamous cell carcinoma (LUSC), stomach adenocarcinoma (STAD), and thyroid cancer (THCA). Conversely, its expression was downregulated in bladder urothelial carcinoma (BLCA), breast invasive carcinoma (BRCA), cervical squamous cell carcinoma and endocervical adenocarcinoma (CESC), and uterine corpus endometrial carcinoma (UCEC) (Fig. 1A). Furthermore, GRB10 expression exhibited stage-dependent variations, decreasing with lower pathological stage in BLCA, KIRC, and THCA, while increasing with higher stage in COAD (Fig. 1C ). Fig. 1GRB10 expression and prognostic impact in pan-cancer. A Differential expression of GRB10 between tumors and adjacent normal tissues in pan-cancer. B Summary of the correlation between expression of GRB10 with OS, DSS, DFI and PFI, (left), and Forest plot summarizing univariate Cox regression analysis of GRB10 association with OS, (right). Cancer types highlighted in red indicate that GRB10 serves as a significant risk factor, (Hazard Ratio > 1; p < 0.05), while those in blue indicate a significant protective role, (Hazard Ratio < 1; p < 0.05). The exact hazard ratio, (HR) and 95% confidence interval for each cancer type are provided. C Differential expression of GRB10 in different tumor pathological stage. D Validation of GRB10 overexpression in three colon cancer datasets. E Kaplan-Meier survival curve of GRB10 in two independent colon cancer datasets, (top 30% vs. bottom 30% by GRB10 expression). Risk tables beneath each curve indicate the number of patients at risk over time in the respective GRB10 expression groups GRB10 expression and prognostic impact in pan-cancer. A Differential expression of GRB10 between tumors and adjacent normal tissues in pan-cancer. B Summary of the correlation between expression of GRB10 with OS, DSS, DFI and PFI, (left), and Forest plot summarizing univariate Cox regression analysis of GRB10 association with OS, (right). Cancer types highlighted in red indicate that GRB10 serves as a significant risk factor, (Hazard Ratio > 1; p < 0.05), while those in blue indicate a significant protective role, (Hazard Ratio < 1; p < 0.05). The exact hazard ratio, (HR) and 95% confidence interval for each cancer type are provided. C Differential expression of GRB10 in different tumor pathological stage. D Validation of GRB10 overexpression in three colon cancer datasets. E Kaplan-Meier survival curve of GRB10 in two independent colon cancer datasets, (top 30% vs. bottom 30% by GRB10 expression). Risk tables beneath each curve indicate the number of patients at risk over time in the respective GRB10 expression groups Univariate Cox regression analysis across all 33 TCGA cancers identified GRB10 as a significant adverse prognostic factor for overall survival (OS) in uveal melanoma (UVM), low-grade glioma (LGG), CESC, COAD, STAD, LUAD, UCEC, and LUSC. Notably, GRB10 served as an independent risk factor for OS, disease-specific survival (DSS), and progression-free interval (PFI) in LGG, CESC, COAD, and LUAD (Fig. 1B). In contrast, GRB10 expression was associated with a protective effect in KIRC. Validation in independent GEO colon cancer cohorts, alongside immunohistochemical data, confirmed GRB10 overexpression in tumor tissues (Fig. 1D and Supplementary Fig. 1) and its association with significantly shorter patient survival (Fig. 1E). To determine whether the prognostic value of GRB10 was independent of classic clinical and immunological variables, we performed multivariable Cox regression analyses in two independent colon cancer cohorts (GSE41258 and GSE161158; Supplementary Fig. 2). After adjusting for tumor TNM stage and the expression of the immune checkpoint genes CTLA4 and CD274 (PD-L1), high GRB10 expression remained a statistically significant independent predictor of poor survival in the GSE41258 cohort (HR = 1.66, p = 0.02). A concordant, strong trend was observed in the GSE161158 cohort (HR = 1.64, p = 0.078). The consistent direction and magnitude of the GRB10 effect across both cohorts following multivariable adjustment reinforce its potential as an independent prognostic biomarker. GRB10 associates with EMT and immune pathway alterations Gene Set Enrichment Analysis (GSEA) comparing tumors with high versus low GRB10 expression revealed significant downregulation of oxidative phosphorylation in several cancers, including BRCA, COAD, DLBC, KIRP, MESO, PCPG, PRAD, SKCM, and THYM. A prominent finding was the consistent and significant enrichment of the epithelial-mesenchymal transition (EMT) pathway in high-GRB10 subgroups across the vast majority of cancer types analyzed (Fig. 2A). This association was corroborated by Gene Set Variation Analysis (GSVA), which demonstrated significant positive correlations between GRB10 expression and enrichment scores for the EMT pathway, as well as with the expression of key EMT transcription factors (ZEB1, ZEB2, SNAI1, SNAI2). In parallel, a significant negative correlation was observed with the mesenchymal-epithelial transition (MET) score (Fig. 2B). Concurrently, immune-activating pathways, such as interferon-alpha/gamma response and inflammatory response, were negatively enriched in high-GRB10 subgroups, particularly in CESC, CHOL, HNSC, LAML, LGG, LIHC, LUAD, LUSC, SKCM, STAD, TGCT, and THCA. Fig. 2Functional analysis of GRB10. A GSEA results for hallmark pathways comparing high vs. low GRB10 expression tumors, (top 30% vs. bottom 30%). The circle color represents the normalized enrichment score, (NES), and the circle size indicates the statistical significance, (-log₁₀[False Discovery Rate, (FDR)]). B Heatmap depicting correlations between GRB10 expression, EMT/MET scores, and key EMT transcription factors, (ZEB1, ZEB2, SNAI1, SNAI2) across TCGA cancers. The color bar represents the Spearman’s correlation coefficient, and statistical significance is denoted by asterisks:: *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001 Functional analysis of GRB10. A GSEA results for hallmark pathways comparing high vs. low GRB10 expression tumors, (top 30% vs. bottom 30%). The circle color represents the normalized enrichment score, (NES), and the circle size indicates the statistical significance, (-log₁₀[False Discovery Rate, (FDR)]). B Heatmap depicting correlations between GRB10 expression, EMT/MET scores, and key EMT transcription factors, (ZEB1, ZEB2, SNAI1, SNAI2) across TCGA cancers. The color bar represents the Spearman’s correlation coefficient, and statistical significance is denoted by asterisks:: *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001 GRB10 correlates with an immunosuppressive microenvironment To elucidate the role of GRB10 in shaping the tumor immune microenvironment (TME), we systematically evaluated its correlation with a comprehensive panel of immune-related genes. GRB10 expression showed significant positive correlations with numerous immune-activating genes in cancers such as COAD, DLBC, HNSC, KIRC, KIRP, LIHC, LUAD, PAAD, PCPG, PRAD, STAD, THYM, UCEC, and UVM. The expression of CD276, ENTPD1, PVR, and ULBP1 was positively correlated with GRB10 in most cancer types. GRB10 also correlated positively with several immunosuppressive genes (e.g., KDR, TGFBR1) in HNSC, KIRC, PAAD, PCPG, and PRAD, while showing negative correlations in LGG and THCA (Fig. 3). Strong positive correlations with chemokines, chemokine receptors, and MHC-related genes were evident in KIRC, KIRP, PAAD, PCPG, PRAD, and UVM, which contrasted with the negative correlations observed in THCA. A similar cancer-type-specific pattern was observed for immune checkpoint-related genes. Fig. 3GRB10 Correlates with Immune-Related Gene Expression. Scatter plots illustrating correlations between GRB10 expression and representative genes across functional categories: Immune-activating, Immune-suppressive, Immune checkpoint, Chemokines, Chemokine receptors, and MHC-related in selected TCGA cancers. The color represents the Spearman’s correlation coefficient, and the circle size is proportional to the statistical significance (-log₁₀[p-value]) GRB10 Correlates with Immune-Related Gene Expression. Scatter plots illustrating correlations between GRB10 expression and representative genes across functional categories: Immune-activating, Immune-suppressive, Immune checkpoint, Chemokines, Chemokine receptors, and MHC-related in selected TCGA cancers. The color represents the Spearman’s correlation coefficient, and the circle size is proportional to the statistical significance (-log₁₀[p-value]) Immune infiltration analysis revealed significant positive correlations between GRB10 and overall immune scores in BRCA, KIRC, KIRP, LIHC, LUAD, LUSC, OV, PAAD, and PRAD (Fig. 4A). Crucially, CIBERSORT analysis indicated a near-universal significant positive correlation between GRB10 and CD4⁺ memory-resting T cell infiltration (Fig. 4B). Validation using EPIC, TIMER, and xCell methods confirmed that while GRB10 correlated positively with total CD4⁺ T cells in ACC, KIRC, PAAD, PCPG, and PRAD, it exhibited a consistent negative correlation with CD4⁺ effector memory T cell infiltration levels across 18 cancer types (Fig. 4C). Collectively, these findings strongly suggest that GRB10 expression is associated with an immunosuppressive TME. Fig. 4GRB10 expression alters immune cell infiltration. A Correlation between GRB10 expression and immune score across TCGA cancers. B Correlation between GRB10 expression and immune cell infiltration levels (CIBERSORT) across TCGA cancers. The color represents the Spearman’s correlation coefficient, and the circle size is proportional to the statistical significance (-log₁₀[p-value]). C Heatmap showing correlations between GRB10 expression and CD4⁺ T cell subtype abundances estimated by multiple algorithms (EPIC, TIMER, xCell). The color bar represents the Spearman’s correlation coefficient, and statistical significance is denoted by asterisks: *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001 GRB10 expression alters immune cell infiltration. A Correlation between GRB10 expression and immune score across TCGA cancers. B Correlation between GRB10 expression and immune cell infiltration levels (CIBERSORT) across TCGA cancers. The color represents the Spearman’s correlation coefficient, and the circle size is proportional to the statistical significance (-log₁₀[p-value]). C Heatmap showing correlations between GRB10 expression and CD4⁺ T cell subtype abundances estimated by multiple algorithms (EPIC, TIMER, xCell). The color bar represents the Spearman’s correlation coefficient, and statistical significance is denoted by asterisks: *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001 GRB10 predicts immunotherapy resistance Given the critical role of the TME in determining the efficacy of immune checkpoint inhibitors (ICIs) [32], we evaluated the predictive value of GRB10. In cohorts of ICI-treated patients with urothelial cancer (IMvigor210), melanoma (VanAllen), and metastatic gastric cancer (Kim), patients with low GRB10 expression exhibited significantly higher survival rates and longer survival times compared to those with high GRB10 expression (Fig. 5A). Furthermore, non-responders to immunotherapy in the IMvigor210, Riaz (melanoma), and Amato (melanoma) cohorts displayed significantly higher GRB10 expression levels than responders (Fig. 5B). These results establish GRB10 as a potential biomarker for predicting resistance to ICI therapy. To quantify its additive predictive power, we compared time-dependent receiver operating characteristic (ROC) curves for models with and without GRB10 (Supplementary Fig. 3). The inclusion of GRB10 resulted in a consistent, albeit modest, increase in the area under the curve. Fig. 5GRB10 Predicts Immunotherapy Resistance. A Kaplan-Meier survival curves comparing ICI-treated patients with high vs. low GRB10 expression in urothelial carcinoma (IMvigor210), melanoma (VanAllen), and metastatic gastric cancer (Kim) cohorts. B GRB10 expression levels in responders vs. non-responders to ICI therapy in urothelial carcinoma (IMvigor210) and melanoma (Riaz, Amato) cohorts GRB10 Predicts Immunotherapy Resistance. A Kaplan-Meier survival curves comparing ICI-treated patients with high vs. low GRB10 expression in urothelial carcinoma (IMvigor210), melanoma (VanAllen), and metastatic gastric cancer (Kim) cohorts. B GRB10 expression levels in responders vs. non-responders to ICI therapy in urothelial carcinoma (IMvigor210) and melanoma (Riaz, Amato) cohorts Genomic alterations and interaction network of GRB10 GRB10 expression demonstrated significant positive correlations with tumor mutational burden (TMB) in LGG, SARC, and THYM, and significant negative correlations in BLCA, PAAD, STAD, THCA, and UVM (Fig. 6A). The overall mutation frequency of the GRB10 gene in the TCGA cohort was low. Uterine corpus endometrial carcinoma (UCEC) exhibited the highest mutation frequency, which was predominantly characterized by nonsense mutations (Fig. 6B). Protein-protein interaction network analysis using GeneMANIA revealed that GRB10-interacting partners are significantly enriched in key biological processes and pathways. These include transmembrane receptor protein tyrosine kinase signaling, insulin receptor signaling, PI3K-Akt signaling, cellular response to hormone stimulus, regulation of glycogen biosynthetic process, regulation of longevity, FoxO signaling, and Rap1 signaling (Fig. 6C-E). Fig. 6GRB10 Genetic Alterations and Protein Interaction Network. A Radar plot showing correlation between GRB10 expression and Tumor Mutation Burden across TCGA cancers. B Summary of GRB10 mutation frequency and types across the TCGA pan-cancer cohort. C Protein-Protein Interaction network of GRB10. D Physical and predicted interaction partners of GRB10. E Functional enrichment analysis (GO Biological Process, KEGG pathways) of GRB10-interacting proteins GRB10 Genetic Alterations and Protein Interaction Network. A Radar plot showing correlation between GRB10 expression and Tumor Mutation Burden across TCGA cancers. B Summary of GRB10 mutation frequency and types across the TCGA pan-cancer cohort. C Protein-Protein Interaction network of GRB10. D Physical and predicted interaction partners of GRB10. E Functional enrichment analysis (GO Biological Process, KEGG pathways) of GRB10-interacting proteins Discussion This comprehensive pan-cancer analysis establishes GRB10 as a pivotal regulator at the intersection of oncogenic signaling and immunosuppression. While Yang et al. previously reported GRB10 dysregulation across cancers [18], our study provides deeper mechanistic and clinical insights by systematically linking its expression to EMT, immune evasion, and ultimately, resistance to immune checkpoint blockade. A critical finding is the strong association between GRB10 and EMT, a well-established driver of metastasis, therapeutic resistance, and stem-like properties [33]. The consistent enrichment of EMT pathways in high-GRB10 subgroups, along with positive correlations across diverse cancers with key transcription factors (ZEB1, ZEB2, SNAI1, SNAI2) and negative associations with mesenchymal-epithelial transition (MET) scores, collectively implicate GRB10 in promoting and maintaining aggressive mesenchymal states. This finding aligns with prior evidence linking GRB10 to hypoxia-induced EMT in glioma [14], suggesting a potentially conserved oncogenic role across malignancies. Given GRB10’s established function in insulin/IGF-1 and PI3K-AKT signaling [34], we propose that it may act as an adaptor node transducing growth factor signals into EMT activation, thereby facilitating the aggressive mesenchymal phenotype observed in multiple cancers. Beyond its role in cellular plasticity, GRB10 expression correlated strongly with immunosuppressive features in the tumor microenvironment. The significant suppression of interferon response pathways, coupled with a reproducible shift in CD4⁺ T cell polarization—specifically, an increase in resting-memory populations and a decrease in effector memory subsets—suggests that GRB10 contributes to an immune-evasive niche. In light of emerging evidence that growth factor pathways (e.g., insulin/IGF, PI3K-AKT) can shape immune cell function and exhaustion [35–37], GRB10 may serve as a critical bridge connecting oncogenic signaling to immune evasion, thereby providing a compelling mechanistic explanation for its role in therapy resistance. The translational impact of these findings is underscored by GRB10’s consistent performance as a predictive biomarker across multiple immunotherapy cohorts. Patients with high GRB10 expression showed significantly poorer survival and response rates in urothelial carcinoma, melanoma, and gastric cancer cohorts. While multivariable analyses confirmed GRB10 provides independent prognostic information beyond standard clinical variables, its incremental value for improving prediction accuracy was modest (ΔAUC < 0.02). This suggests GRB10’s clinical utility may lie not in dramatically enhancing outcome prediction, but in identifying a distinct biological subset of patients with EMT-driven, immune-evasive tumors that might benefit from tailored combination therapies. We acknowledge several limitations. The bioinformatic nature of our study, while strengthened by validation across multiple independent cohorts, necessitates experimental confirmation of GRB10’s mechanistic contributions to EMT programming, immune suppression, and therapeutic resistance. Furthermore, prospective validation in multi-cancer clinical trials will be essential to establish GRB10’s utility as a stratification biomarker. In conclusion, our work positions GRB10 as a coordinator of tumor progression that simultaneously drives EMT and fosters immunosuppression. These findings nominate GRB10 not only as a biomarker for identifying patients at high risk of immunotherapy resistance, but also as a compelling therapeutic target worthy of future investigation. Supplementary Information Supplementary material 1. Supplementary material 1.