Vasoactive intestinal peptide associated 8-gene signature with prognostic and immune associations in melanoma.

Vasoactive intestinal peptide associated 8-gene signature with prognostic and immune associations in melanoma - PMC Skip to main content An official website of the United States government Here's how you know Here's how you know Search Log in Primary site navigation Search Log in Search PMC Full-Text Archive Search in PMC Journal List User Guide As a library, NLM provides access to scientific literature. Inclusion in an NLM database does not imply endorsement of, or agreement with, the contents by NLM or the National Institutes of Health. Learn more: PMC Disclaimer | PMC Copyright Notice Medicine (Baltimore) . 2026 May 29;105(22):e48961. doi: 10.1097/MD.0000000000048961 Vasoactive intestinal peptide associated 8-gene signature with prognostic and immune associations in melanoma Xiaogang Hong a, * , Xiuzhen Zheng a Author information Article notes Copyright and License information a Department of Ophthalmology, People’s Hospital of Kaihua, Quzhou, Zhejiang, China. * Correspondence: Xiaogang Hong, Department of Ophthalmology, People’s Hospital of Kaihua, No. 59, Fenghuang Middle Road, Qinyang Office, Huabu, Kaihua, Quzhou, Zhejiang 324300, China (e-mail: Hxg-1027@163.com ). Received 2025 Aug 31; Revised 2026 Apr 19; Accepted 2026 May 5; Collection date 2026 May 29. Copyright © 2026 the Author(s). Published by Wolters Kluwer Health, Inc. This is an open access article distributed under the Creative Commons Attribution License 4.0 (CCBY) , which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited. PMC Copyright notice PMCID: PMC13225499  PMID: 42216340 Abstract Vasoactive intestinal peptide (VIP) plays a multifaceted role in cancer biology, yet its prognostic and immunological implications in melanoma remain underexplored. This study aimed to construct a VIP-related gene signature for risk stratification and to explore its potential associations with the tumor immune microenvironment and drug sensitivity in melanoma. RNA-sequencing data and clinical information were obtained from TCGA-SKCM (n = 477, training set) and GSE65904 (n = 209, validation set). A VIP-related prognostic signature (VIPsig) was developed using stepwise Cox regression and Gradient Boosting Machine algorithms. Patients were stratified into high- and low-risk groups based on the median risk score. Somatic mutations, immune infiltration, drug sensitivity, and functional enrichment were further analyzed. A nomogram integrating clinical factors was developed and validated. Single-cell RNA-seq data ( GSE115978 ) were reanalyzed to characterize cellular expression patterns. A preliminary pan-cancer analysis indicated potential dysregulation of VIP pathway genes. In melanoma, we identified an 8-gene signature ( CD28 , CD80 , CD86 , CTLA4 , FAS , IFNG , IL10 , and IL12A ) associated with survival. In the training set, high-risk patients exhibited significantly worse overall survival (OS) compared to low-risk patients (HR = 2.96, 95% CI: 2.24–3.92, P < .001). This was validated in the GSE65904 cohort (HR = 1.87, 95% CI: 1.26–2.77, P = .002). The VIPsig demonstrated discriminative ability with 1-, 3-, and 5-year AUCs of 0.760 (95% CI: 0.675–0.845, P < .001), 0.749 (95% CI: 0.692–0.807, P < .001), and 0.770 (95% CI: 0.717–0.823, P < .001) in the training set, respectively. Functionally, the signature was significantly associated with immune cell infiltration and immune checkpoint expression. In silico drug sensitivity analysis suggested potential associations with estimated IC50 values for several agents, though these findings require experimental validation. A nomogram integrating VIPsig with clinicopathological factors showed improved net benefit in decision curve analysis. The VIP-based signature demonstrates robust prognostic value for melanoma survival and reflects the tumor immune microenvironment status. Exploratory analyses suggest potential associations with in silico drug sensitivity estimates; however, its utility for treatment response prediction remains unvalidated and warrants further investigation in prospective melanoma cohorts treated with immune checkpoint inhibitors. Keywords: immune cell infiltration, machine learning, melanoma, nomogram, vasoactive intestinal peptide 1. Introduction Cutaneous melanoma is an aggressive malignancy originating from melanocytes, with a steadily rising global incidence over the past decades. [ 1 , 2 ] It accounts for the majority of skin cancer-related deaths due to its high metastatic potential and resistance to conventional therapies. [ 3 , 4 ] Current estimates indicate over 3,00,000 new cases are diagnosed worldwide annually, representing a significant public health burden. [ 1 ] Although immune checkpoint inhibitors (ICIs) have revolutionized melanoma treatment, [ 5 , 6 ] a significant subset of patients exhibits intrinsic or acquired resistance. [ 7 , 8 ] Several gene signatures have been developed to predict prognosis and treatment outcomes, focusing on tumor mutation burden, T-cell exclusion, or inflammatory cytokine profiles. [ 9 - 11 ] While these models have significantly advanced the field, they primarily capture general characteristics of the tumor immune microenvironment (TIME) and often overlook the neuroimmunological axis. [ 12 ] The crosstalk between the nervous and immune systems, particularly through neuropeptides, has emerged as a critical but underexplored regulator of tumor progression and immune tolerance. Therefore, existing signatures may lack the specificity to capture immune evasion mechanisms driven by neuroendocrine signaling, representing a critical knowledge gap. The vasoactive intestinal peptide (VIP) is a multifunctional neuropeptide that exerts potent immunomodulatory effects through binding to its receptors, VPAC1 and VPAC2 . Beyond its physiological roles, VIP signaling has been implicated in the pathogenesis of various cancers. It can directly influence tumor cell proliferation, migration, and survival in contexts such as breast, prostate, and colorectal cancer. [ 13 - 16 ] More notably, VIP is a pivotal regulator of immune homeostasis within the tumor microenvironment (TME). [ 17 ] It promotes a shift towards an anti-inflammatory phenotype by stimulating regulatory T-cell (Treg) differentiation, inhibiting pro-inflammatory cytokine production (e.g., TNF-α, IL-6), and alternatively activating macrophages. [ 18 - 20 ] This immunosuppressive activity suggests a role for VIP in facilitating tumor immune evasion. However, its specific function and clinical relevance in melanoma – a malignancy profoundly shaped by the TIME – are complex and not fully elucidated. [ 21 ] Emerging evidence suggests that VIP and its related genes may influence melanoma progression and response to therapy, [ 22 ] positioning them as potential biomarkers and therapeutic targets worthy of systematic investigation. Based on this biological premise, we hypothesized that the dysregulation of VIP-related genes constitutes a distinct “neuro-immune” signature that correlates with aggressive melanoma phenotypes and immune suppression, offering prognostic information nonredundant to existing markers. To test this hypothesis, this study addressed 3 prespecified objectives: to construct a robust VIP-based gene signature (VIPsig) using machine learning algorithms in bulk transcriptomic cohorts; to validate the prognostic accuracy of VIPsig and its association with the immune landscape; and to perform an exploratory analysis of its associations with drug sensitivity. It is important to note that this study utilizes retrospective multi-cohort data to establish prognostic associations, and the findings regarding therapeutic sensitivity are hypothesis-generating, warranting further validation in prospective clinical trials. 2. Materials and methods 2.1. Data acquisition and processing RNA-sequencing data (FPKM values) for skin cutaneous melanoma (SKCM, n = 477) were obtained from The Cancer Genome Atlas (TCGA, https://portal.gdc.cancer.gov/ ) repository and used as the training cohort. The validation cohort GSE65904 (n = 209) was obtained from the Gene Expression Omnibus ( https://www.ncbi.nlm.nih.gov/geo/ ). Each dataset underwent rigorous filtering to exclude samples with incomplete clinical or survival information. Batch effect correction was performed by merging the 2 datasets and applying the ComBat algorithm (sva package v3.56.0; Boston University) to the combined expression matrix, as detailed in the uploaded analysis code. The corrected matrix was then split back into TCGA-SKCM and GSE65904 portions. Importantly, all parameters for VIPsig construction (gene filtering, StepCox[forward] selection, and Gradient Boosting Machine [GBM] coefficient estimation) were derived exclusively from the TCGA-SKCM training set. The GSE65904 cohort was used to evaluate the performance of the generated candidate models, thereby informing the comparison of machine learning algorithms and the final selection of the optimal prognostic signature. Thirteen VIP-related genes from the WP_CONTROL_OF_IMMUNE_TOLERANCE_BY_VASOACTIVE_INTESTINAL_PEPTIDE gene set were retrieved from the MSigDB ( https://www.gsea-msigdb.org/gsea/msigdb ) database ( Table S1 , Supplemental Digital Content). [ 23 ] The CONSORT diagram was illustrated in Figure S1 , Supplemental Digital Content. 2.2. Pan-cancer analysis The TCGAplot package (version 8.0.0; Huazhong University of Science and Technology) was employed for pan-cancer analysis. [ 24 ] Gene set variation analysis was used to calculate enrichment scores for the VIP gene set. Wilcoxon tests were applied to compare scores between tumor and normal tissues. Spearman’s correlation analysis was conducted to evaluate associations between the VIP gene set score and tumor mutational burden (TMB) or microsatellite instability (MSI). Cox proportional hazards regression was used to assess the relationship between the VIP score and overall survival across cancer types, visualized via forest plots. 2.3. Machine learning model construction The Mime1 package (version 0.0.0.9000; Central South University) [ 25 ] was used to develop and evaluate machine learning-based integrative models from transcriptomic data. Model construction and performance evaluation were conducted using the ML.Dev.Prog.Sig() function within a 10-fold cross-validation framework. Regarding parameter configuration, nodesize – a critical hyperparameter for the random survival forest model was set to 5, and unicox.filter.for.candi was set to TRUE, while all other parameters retained their default values. The procedure initially identified VIP-related genes significantly associated with overall survival (OS) in the TCGA-SKCM cohort. Subsequently, based on the expression profiles of these candidate genes, 117 combinatorial models were established by integrating 10 machine learning algorithms: random survival forest, Enet, StepCox, CoxBoost, plsRcox, superpc, GBM, survivalsvm, Ridge, and Lasso. Finally, the prognostic risk score, designated as VIPsig, was calculated for each model. Following the standard model selection procedure recommended by the Mime framework, [ 26 ] the optimal model was selected based on the highest mean concordance index (C-index) across the validation cohort(s). This approach prioritizes generalizability and reduces the risk of overfitting by evaluating model performance on data not used for parameter estimation, consistent with established practices in machine learning-based prognostic model development. [ 26 - 28 ] Importantly, while all model parameters (gene coefficients and weights) were estimated exclusively using the TCGA-SKCM cohort, the GSE65904 cohort was used as an integral part of the selection process to identify the algorithm with the highest generalizability. Samples in both TCGA-SKCM and GSE65904 cohorts were stratified into high- and low-risk groups based on the median value of the VIPsig. Kaplan–Meier survival analysis was used to compare overall survival between risk groups. Time-dependent receiver operating characteristic (ROC) curves were constructed to evaluate discriminative performance at 1, 3, and 5 years. The area under the curve (AUC), its 95% confidence interval, and the associated P -value were computed using the DeLong nonparametric method as implemented in the pROC R package (version 1.18.5; University of Geneva). [ 29 ] The P -value tests the null hypothesis H 0 : AUC = 0.5 (no discriminative ability) against the alternative H 1 : AUC ≠ 0.5. Spearman’s rank correlation was used for all correlation analyses throughout this study, as the variables assessed (risk scores, immune infiltration estimates, IC50 values, and genomic instability metrics) do not assume normal distribution. All correlation coefficients reported in figures and text refer to Spearman’s ρ unless explicitly stated otherwise. The IMvigor210 cohort (metastatic urothelial carcinoma treated with anti-PD-L1 agent atezolizumab) was analyzed using the IMvigor210CoreBiologies package (version 1.0.1). The association between VIPsig and objective response to immunotherapy (complete/partial response vs stable/progressive disease) was assessed using the Wilcoxon rank-sum test and ROC analysis. 2.4. Somatic mutation analysis Somatic mutation data were analyzed using the R package maftools (v2.24.0; National University of Singapore). [ 30 ] Mutational signatures were extracted via Non-Negative Matrix Factorization and matched against the COSMIC legacy database to quantify UV exposure (Signature 7). Tumor Mutational Burden (TMB) was calculated as non-synonymous mutations per megabase. To assess the association between the Risk Score and TMB, partial Spearman’s correlation analysis was performed using the ppcor (version 1.1; Wayne State University) package, adjusting for tumor stage, sample type, and UV signature contribution. Additionally, multivariable logistic regression was employed to evaluate the independent association between risk groups and the top 20 mutated genes, correcting for the aforementioned covariates. Results were reported as odds ratios with Benjamini–Hochberg adjusted P -values. 2.5. Immune landscape and drug sensitivity analysis The IOBR package (version 0.99.0; Southern Medical University) was used to evaluate the immune landscape. [ 31 ] CIBERSORT estimated infiltration levels of 22 immune cell types. Immunophenoscore (IPS) was calculated using the IPS algorithm, and stromal/immune/ESTIMATE scores along with tumor purity were computed via the ESTIMATE algorithm. Group differences were assessed using Wilcoxon tests, and Spearman’s correlation analyzed associations between VIPsig (and its component genes) and IPS, ESTIMATE scores, and immune checkpoint gene expression. Following the assessment of the immune landscape, we extended our analysis to explore potential therapeutic vulnerabilities associated with the VIPsig using the pRRophetic R package (version 0.5; University of Chicago). [ 32 ] This algorithm employs ridge regression models trained on the Genomics of Drug Sensitivity in Cancer database, which encompasses extensive pharmacogenomic data from a wide panel of cancer cell lines. We estimated the half-maximal inhibitory concentration (IC50) for each sample – where a lower IC50 indicates higher predicted sensitivity Group differences in drug sensitivity were compared via Wilcoxon test. 2.6. Differential expression and enrichment analysis The limma package (version 3.64.3; WEHI) was used to identify differentially expressed genes (DEGs) between high- and low-risk groups, with thresholds of |log 2 (fold change)| > 1 and adjusted P -value < .05. [ 33 ] Gene Ontology (GO) and Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway enrichment analyses were performed on DEGs using clusterProfiler package (version 4.16.0; Southern Medical University). [ 34 ] Gene Set Enrichment Analysis was conducted using the Hallmark gene sets (h.all.v2025.1.Hs.symbols.gmt) from MSigDB to identify coordinated phenotypic changes. 2.7. Nomogram construction The rms (version 8.1-0; Vanderbilt University) and survival (version 3.8-3; Mayo Clinic) packages were used to develop and validate a clinical prediction model. The model integrated VIPsig and clinicopathological features (age, sex, pT stage, pN stage, pM stage, pathologic stage, Clark level, and radiotherapy). Univariate Cox regression ( P < .05) identified variables associated with OS, which were included in a multivariate Cox model to determine independent prognostic factors. For Cox regression analyses and nomogram construction, staging and grading variables were coded as ordered numeric integers as follows: pT stage (T1 = 1, T2 = 2, T3 = 3, T4 = 4), pN stage (N0 = 0, N1 = 1, N2 = 2, N3 = 3), pM stage (M0 = 0, M1 = 1), pathologic stage (Stage I = 1, Stage II = 2, Stage III = 3, Stage IV = 4), and Clark level (Level I = 1, Level II = 2, Level III = 3, Level IV = 4, Level V = 5). This ordered numeric coding approach assumes a monotonic relationship between increasing stage/grade and survival risk. Age and VIPsig risk score were included as continuous variables. A nomogram was constructed using the cph() and nomogram() functions to predict 1‑, 3‑, and 5‑year OS. Internal validation was performed via bootstrap resampling (1000 repetitions). Calibration curves and decision curve analysis assessed predictive accuracy and clinical utility, respectively. Time-dependent ROC curves evaluated discrimination ability at 1, 3, and 5 years. 2.8. Single-cell RNA sequencing analysis The scRNA-seq dataset ( GSE115978 ) was processed using the Seurat R package (version 5.4.0; New York Genome Center). [ 35 ] Raw count matrices were filtered to exclude genes expressed in fewer than 3 cells and cells with fewer than 200 detected features. Gene expression was normalized using the global-scaling LogNormalize method (scale factor = 10,000), followed by the identification of the top 2000 variable features via variance-stabilizing transformation. Principal component analysis was performed on scaled data, and the first 15 principal components were selected for nonlinear dimensionality reduction using t-distributed Stochastic Neighbor Embedding (t-SNE). Cell annotations were derived from the original study’s metadata and rigorously validated by visualizing the expression of canonical lineage markers: PMEL / MLANA (Melanoma), CD8A / CD3D (CD8+ T cells), CD4 / IL7R (CD4+ T cells), GNLY / NKG7 (NK cells), MS4A1 (B cells), CD14 (Myeloid), PECAM1 (Endothelial), and COL1A1 (Cancer-associated fibroblasts). To quantify the risk signature at single-cell resolution, a composite Risk Score was calculated for each cell using the AddModuleScore function based on the model-derived gene signature. Statistical differences in risk scores between treatment groups (Naive vs Treated) within malignant subpopulations were assessed using the Wilcoxon rank-sum test. 3. Results 3.1. Pan-cancer analysis of VIP-related gene sets The enrichment score of the VIP gene set was calculated using the Gene set variation analysis algorithm. The results revealed that this score was dysregulated across multiple cancer types, showing significant upregulation or downregulation. Compared to normal tissues, the score was significantly decreased in COAD, LIHC, LUAD, and LUSC, while it was markedly increased in KICH, KIRC, KIRP, and THCA (Fig. 1 A). Furthermore, correlation analysis demonstrated that the VIP score was significantly associated with TMB in BRCA, COAD, LIHC, LUAD, and OV (Fig. 1 B), and with MSI in BRCA, CHOL, COAD, KIRP, LIHC, LUSC, OV, TGCT, and THCA (Fig. 1 C). Cox analysis indicated that the VIP score was significantly correlated with overall survival in KIRC (HR = 2.271, 95% CI: 1.538–3.353, P < .001) and SKCM (HR = 0.509, 95% CI: 0.347–0.747; Fig. 1 D). These findings suggest that the VIP gene set is involved in tumorigenesis and progression across multiple cancer types, exhibiting pleiotropic effects. Figure 1. Open in a new tab Pan-cancer analysis of GSVA scores for the VIP gene set. (A) Wilcoxon test comparing VIP scores between tumor and normal tissues. Spearman’s rank correlation analysis between VIP score and (B) TMB, and (C) MSI. (D) Forest plot of Cox analysis evaluating the association between VIP score and overall survival across cancers. * P < .05, ** P < .01, **** P < .0001. GSVA = Gene Set Variation Analysis, MSI = microsatellite instability, Ns = not significant, TMB = tumor mutational burden, VIP = vasoactive intestinal peptide. 3.2. Construction and evaluation of the VIPsig risk signature Cox analysis identified 8 genes significantly associated with overall survival in melanoma: CD28 , CD80 , CD86 , CTLA4 , FAS , IFNG , IL10 , and IL12A . Subsequently, 117 machine learning models were constructed based on these genes (Fig. 2 A). Among 117 machine learning models, the StepCox[forward] + GBM algorithm achieved the highest mean C-index across the validation cohort and was therefore selected as the optimal model, following the Mime framework’s recommended selection criterion. [ 26 ] The performance of the top 10 candidate models in both training and validation cohorts is detailed in Table S2 , Suppplemental Digital Content, confirming consistent performance of the selected model across datasets. This optimal model was designated as VIPsig. Kaplan–Meier survival analysis demonstrated that patients in the high-risk group (n = 227) had significantly worse OS compared to the low-risk group (n = 226) in the training cohort (HR = 2.96, 95% CI: 2.24–3.92, P < .001; Fig. 2 B). This finding was validated in the external GSE65904 cohort (HR = 1.87, 95% CI: 1.26–2.77, P = .002; Fig. 2 C). ROC curve analysis indicated robust predictive performance (Fig. 2 D). In the TCGA-SKCM cohort, the AUCs for 1-, 3-, and 5-year OS were 0.760 (95% CI: 0.675–0.845, P < .001 vs null AUC = 0.5, DeLong test), 0.749 (95% CI: 0.692–0.807, P < .001, DeLong test), and 0.770 (95% CI: 0.717–0.823, P < .001, DeLong test), respectively. In the GSE65904 cohort, the AUCs for 1-, 3-, and 5-year OS were 0.629 (95% CI: 0.533–0.725, P = .009, DeLong test), 0.686 (95% CI: 0.600–0.771, P < .001, DeLong test), and 0.604 (95% CI: 0.489–0.719, P = .077, DeLong test), respectively. Figure 2. Open in a new tab Construction of the VIPsig risk signature based on VIP-related genes in melanoma. (A) Distribution of C-index values across 117 machine learning-based signature models. Kaplan–Meier survival analysis comparing overall survival between high- and low-risk groups in the (B) TCGA-SKCM and (C) GSE65904 cohorts. (D) ROC curves showing the area under the curve (AUC) of VIPsig for predicting 1-, 3-, and 5-year overall survival. ROC = receiver operating characteristic, VIP = vasoactive intestinal peptide. To rigorously validate the robustness and broad applicability of VIPsig, we performed a series of prespecified sensitivity and subgroup analyses. As illustrated in the forest plot ( Fig. S2A , Supplemental Digital Content), the prognostic power of VIPsig remained highly consistent and statistically significant across strictly defined sensitivity cohorts. Furthermore, to verify the diagnostic stability across diverse clinical scenarios, we conducted stratified time-dependent ROC analyses ( Fig. S2B , Supplemental Digital Content). The VIPsig demonstrated stable predictive accuracy (1-, 3-, and 5-year AUCs) across distinct patient strata, including gender, Clark level, sample type, and pathological stages (Stage I + II vs III + IV), confirming its reliability independent of confounding clinical variables. 3.3. Association between VIPsig and clinicopathological features in melanoma patients We next investigated the relationship between VIPsig and clinicopathological characteristics of melanoma patients. A significant positive correlation was observed between VIPsig and patient age (ρ = 0.14, P = .002; Fig. 3 A). Subgroup analysis based on clinicopathological features revealed no significant difference in VIPsig between gender subgroups (Fig. 3 B), whereas significant differences were detected across pathological stages ( P = 2.7e−06; Fig. 3 C), T stages ( P = 1.4e−08; Fig. 3 D), and Clark levels ( P = 8.2e−05; Fig. 3 F). No significant difference was observed across N stages (Fig. 3 E). Additionally, metastatic melanoma patients exhibited significantly lower VIPsig scores compared to those with primary tumors ( P = 2.3e−09; Fig. 3 G). These results indicate that VIPsig is associated with age and pathological features in melanoma patients. Figure 3. Open in a new tab Spearman’s rank correlation between VIPsig and clinicopathological features in melanoma. (A) Scatter plot of Spearman correlation analysis between patient age and VIPsig. Kruskal–Wallis test comparing VIPsig across subgroups of (B) gender, (C) pathological stage, (D) T stage, (E) N stage, (F) Clark level, and (G) sample type. VIP = vasoactive intestinal peptide. 3.4. VIPsig and the immune landscape of melanoma To robustly characterize the immune contexture associated with VIPsig, we performed a cross-platform deconvolution analysis using CIBERSORT, MCP-counter, Quantiseq, and TIMER. Despite methodological differences, consistent patterns emerged: the VIPsig risk score was significantly negatively correlated with the infiltration of antitumor effector cells, specifically CD8+ T cells and activated NK cells, across multiple algorithms (Fig. 4 A). Conversely, a positive correlation was observed with pro-tumorigenic cell types, such as M0 and M2 macrophages. Furthermore, the VIPsig risk score exhibited a strong negative correlation with the expression of all analyzed immune checkpoint genes (e.g., CD274 , CTLA4 , PDCD1 ), whereas its constituent genes showed the opposite trend (Fig. 4 B). Consistent with an “immune-cold” phenotype, VIPsig was significantly positively correlated with Tumor Purity (ρ = 0.73) and negatively correlated with Immune Score (ρ = −0.77), Stromal Score (ρ = −0.54), and ESTIMATE Score (ρ = −0.73; Fig. 4 C). Interestingly, a weak positive correlation was observed with IPS (ρ = 0.19). Figure 4. Open in a new tab The immune landscape of the VIPsig risk signature. (A) Cross-validation of immune cell infiltration correlations using 4 independent algorithms (CIBERSORT, MCPcounter, Quantiseq, and TIMER). (B) Dot plot illustrating the correlation between the VIPsig Risk Score (and its constituent genes) and key immune checkpoint genes (ICGs). Asterisks denote FDR-adjusted significance levels (*FDR < 0.05, **FDR < 0.01, ***FDR < 0.001). (C) Lollipop chart showing the correlation of the Risk Score with Tumor Purity, Stromal Score, Immune Score, ESTIMATE Score, and immunophenoscore (IPS). ICG = immune checkpoint gene, FDR = false discovery rate, VIP = vasoactive intestinal peptide. 3.5. VIPsig and drug sensitivity in melanoma We further performed an exploratory in silico analysis to examine the association between VIPsig and predicted drug sensitivity profiles. To ensure statistical reliability, all associations were corrected for multiple testing (false discovery rate [FDR]). As shown in Figure 5 A, VIPsig was positively correlated with the estimated IC50 values of 21 drugs (ρ > 0, red), indicating potential resistance in high-risk patients to agents such as Methotrexate, Camptothecin, and Gemcitabine. Conversely, VIPsig was negatively correlated with the IC50 values of 13 drugs (ρ < 0, blue), suggesting that high-risk patients might be more sensitive to targeted therapies including Bicalutamide, Imatinib, and Elesclomol. These computational predictions were further corroborated by comparing IC50 distributions, where the high-risk group demonstrated significantly lower IC50 values for Bicalutamide (ρ = −0.55, P < 2.2e−16) and Imatinib (ρ = −0.43, P < 2.2e−16) compared to the low-risk group (Fig. 5 B). Figure 5. Open in a new tab Drug sensitivity and resistance prediction. (A) Lollipop chart displaying the Spearman correlation between the VIPsig Risk Score and the predicted IC50 values of therapeutic agents. (B) Boxplots validating the IC50 differences for the top correlated drugs between low- and high-risk groups. VIP = vasoactive intestinal peptide. 3.6. VIPsig and gene expression in melanoma To elucidate the biological processes and pathways associated with VIPsig, we performed differential gene expression analysis between high- and low-risk groups, identifying 1338 DEGs (Fig. 6 A). Enrichment analysis revealed that these genes were involved in KEGG pathways such as cytokine-cytokine receptor interaction, cell adhesion molecules, and intestinal immune network for IgA production (Fig. 6 B). GO enrichment indicated associations with biological processes including immune effector process, leukocyte cell–cell adhesion, and regulation of T-cell activation (Fig. 6 C). Hallmark Gene Set Enrichment Analysis further demonstrated significant activation of apoptosis, IL2-STAT5 signaling, IL6-JAK-STAT3 signaling, inflammatory response, and interferon-related pathways in the high-risk group (Fig. 6 D). Figure 6. Open in a new tab Cellular pathways and biological processes associated with VIPsig. (A) Volcano plot of differentially expressed genes (|log 2 (fold change)| > 1 and adjusted P < .05) between high- and low-risk groups. (B) KEGG pathway and (C) GO term enrichment analyses of VIPsig-related differentially expressed genes. (D) Gene Set Enrichment Analysis (GSEA) of Hallmark pathways. GSEA = Gene Set Enrichment Analysis, GO = Gene Ontology, KEGG = Kyoto Encyclopedia of Genes and Genomes, VIP = vasoactive intestinal peptide. 3.7. VIPsig and tumor mutational burden in melanoma We visualized the mutation profiles of the top frequently mutated genes – such as TTN , MUC16 , DNAH5 , BRAF , and PCLO – in the high- and low-risk groups (Fig. 7 A). To rigorously assess whether specific mutations were enriched in either risk group, we calculated odds ratios with FDR adjustment. As shown in Figure 7 B, no individual genes showed statistically significant differences in mutation frequency between the risk groups (FDR > 0.05), suggesting that the prognostic value of VIPsig is not driven by specific driver mutations. Furthermore, we investigated the relationship between VIPsig and genomic instability metrics while accounting for confounding factors. The risk score showed a marginal, nonsignificant negative trend with the UV signature contribution (ρ = −0.095, P = .057; Figure 7 C), whereas TMB was strongly correlated with UV exposure as expected (Fig. 7 D). Crucially, while TMB appeared slightly lower in the high-risk group, partial correlation analysis adjusting for tumor stage, sample type, and UV signature revealed no significant independent association between VIPsig and TMB (Partial ρ = −0.08, P = .117; Fig. 7 E). These findings indicate that the prognostic capacity of VIPsig is independent of tumor mutational burden and UV exposure history. Figure 7. Open in a new tab Genomic landscape and multivariable analysis of the VIPsig risk signature. (A) Co-oncoplot comparing the somatic mutation landscape of the top 20 most frequently mutated genes between the high-risk (n = 213) and low-risk (n = 193) groups. Mutation types are color-coded. (B) Forest plot displaying the odds ratios (OR) for the mutation frequencies of the top genes comparing high- versus low-risk groups. Statistical significance was assessed using Fisher’s exact test with FDR correction; “ns” indicates no significant difference ( P > .05). (C) Scatter plot showing the correlation between the VIPsig risk score and the UV signature contribution (COSMIC Signature 7). The correlation was assessed using Spearman’s rank correlation test. (D) Scatter plot illustrating the positive correlation between tumor mutational burden (TMB) and UV signature contribution. (E) Boxplot comparing TMB levels between high- and low-risk groups. FDR = false discovery rate, OR = odds ratio, TMB = tumor mutational burden, VIP = vasoactive intestinal peptide. 3.8. Nomogram for melanoma patients To construct a nomogram, we first performed Cox analysis. Univariate Cox analysis identified VIPsig, age, pT, pN, pStage, Clark level, and radiotherapy as significantly associated with overall survival. Multivariate Cox analysis confirmed VIPsig, age, pT, pN, Clark level, pStage, and radiotherapy as independent prognostic factors (Table 1 ). These factors were incorporated into the nomogram (Fig. 8 A). Calibration curves demonstrated close agreement between predicted and observed 1-, 3-, and 5-year survival probabilities in both training and validation sets, indicating good model calibration (Fig. 8 B). Decision curve analysis showed that the nomogram provided higher net benefit across threshold probabilities of 0.2 to 0.8 compared to other prognostic features, supporting its clinical utility (Fig. 8 C). ROC analysis revealed that the nomogram predicted 1-, 3-, and 5-year overall survival with AUC values of 0.834, 0.771, and 0.774, respectively, indicating strong discriminative ability (Fig. 8 D). Table 1. Univariate and multivariate cox regression analysis evaluated the prognostic relevance. Characteristics Univariate Cox Multivariate Cox HR [95% CI] P -value HR[95% CI] P VIPsig 6.503 [3.906–10.827] 6.09E−13 6.045 [3.4–10.746] 8.82E−10 Gender 0.916 [0.64–1.31] .629 0.891 [0.607–1.309] .558 Age 1.021 [1.01–1.033] .000222 1.015 [1.003–1.027] .0158 pT 1.702 [1.408–2.056] 3.70E−08 1.893 [1.439–2.49] 4.99E−06 pN 1.672 [1.419–1.969] 7.71E−10 2.301 [1.72–3.079] 1.96E−08 pstage 1.537 [1.261–1.874] 2.10E−05 4.061 [1.55–10.64] .00434 clark_level 1.889 [1.465–2.435] 9.44E−07 0.431 [0.274–0.676] .000254 pM 2.168 [0.953–4.931] .0649 1.077 [0.79–1.47] .638 Radiotherapy 0.271 [0.119–0.616] .00183 0.387 [0.166–0.901] .0277 Open in a new tab Staging and grading variables were coded as ordered numeric integers in the Cox regression model: pT stage (T1 = 1, T2 = 2, T3 = 3, T4 = 4), pN stage (N0 = 0, N1 = 1, N2 = 2, N3 = 3), pM stage (M0 = 0, M1 = 1), pathologic stage (Stage I = 1, Stage II = 2, Stage III = 3, Stage IV = 4), and Clark level (Level I = 1, Level II = 2, Level III = 3, Level IV = 4, Level V = 5). Age and VIPsig risk score were modeled as continuous variables. CI = confidence interval, HR = hazard ratio, VIPsig = vasoactive intestinal peptide-related gene signature. Figure 8. Open in a new tab Nomogram for melanoma patients based on VIPsig and other clinicopathological features. (A) Nomogram incorporating independent prognostic factors identified by multivariate Cox analysis: VIPsig, age, pT, pN, Clark level, pStage, and radiotherapy. (B) Calibration curves for 1-, 3-, and 5-year overall survival predictions. (C) Decision curve analysis comparing the nomogram with other prognostic factors for 1-year overall survival prediction. (D) ROC curves of the nomogram for predicting 1-, 3-, and 5-year overall survival. ROC = receiver operating characteristic, VIP = vasoactive intestinal peptide. 3.9. Single-cell expression of VIPsig-related genes To validate the cell-type specificity of VIPsig and address potential confounding by tumor purity, we reanalyzed the single-cell RNA-seq dataset ( GSE115978 ) using expert-curated annotations. Clustering analysis identified 8 biologically distinct cell populations: malignant melanoma cells, CD8+ T cells, CD4+ T cells, NK cells, B cells, macrophages, endothelial cells, and CAFs (Fig. 9 A). The accuracy of these annotations was confirmed by the expression of canonical markers, such as PMEL for malignant cells and CD3D for T cells ( Fig. S3 , Supplemental Digital Content), correcting previous automated labeling artifacts. Module scoring revealed that VIPsig-related genes were predominantly expressed in the immune compartment, particularly in CD8+ T cells, CD4+ T cells, and macrophages, with relatively lower expression in malignant and stromal cells (Fig. 9 B, C). This observation aligns with our bulk RNA-seq findings, confirming that VIPsig largely reflects the immune infiltration status of the tumor. Crucially, to determine if differences persist after accounting for sampling heterogeneity, we analyzed the score exclusively within the malignant cell subpopulation. As shown in Figure 9 D, a statistically significant difference in VIPsig scores was observed between treatment-naive and posttreatment malignant cells ( P < .01). This suggests that while the signature is driven by immune infiltration, it also captures subtle intrinsic alterations within the tumor cells that are modulated by therapeutic pressure. Figure 9. Open in a new tab Single-Cell Transcriptomic Analysis of VIPsig in Melanoma ( GSE115978 ). (A) t-SNE visualization of single cells from melanoma patients. Cells are colored by rigorously annotated cell types, including CD8+ T cells (T.CD8), CD4+ T cells (T.CD4), NK cells, B cells, Macrophages, Endothelial cells (Endo.), Cancer-associated fibroblasts (CAF), and Malignant melanoma cells (Mal). (B) Feature plot displaying the distribution of the VIPsig gene signature expression across the single-cell landscape. (C) Violin plot comparing the VIPsig module score across different cell types. (D) Violin plot analyzing the VIPsig score exclusively within the Malignant cell population, comparing treatment-naive versus posttreatment samples. Statistical significance was assessed using the Wilcoxon rank-sum test (** P < .01). CAF = Cancer-associated fibroblasts, VIP = vasoactive intestinal peptide. To further evaluate the clinical relevance of VIPsig in the context of immunotherapy, we analyzed the IMvigor210 cohort. Although the high-risk group showed a trend towards shorter overall survival ( P = .093, Fig. S4A , Supplemental Digital Content) and the direct prediction of binary drug response was limited in this specific cohort (AUC = 0.543, Fig. S4B, C , Supplemental Digital Content), the biological nature of the signature was strongly validated by its association with tumor immune phenotypes. Consistent with our single-cell findings where VIPsig was associated with malignant features rather than immune effector function, the Risk Score was significantly elevated in the “immune-desert” phenotype and lowest in the “immune-inflamed” phenotype ( P < 2.2e−16, Fig. S4D , Supplemental Digital Content). This confirms that a high VIPsig robustly characterizes an immunologically “cold” or immune-excluded tumor microenvironment, which provides a mechanistic basis for the potential resistance to immune checkpoint blockade. 4. Discussion In this study, we addressed the critical challenge of stratifying melanoma prognosis by developing a novel VIP-related gene signature (VIPsig). While the immunomodulatory role of VIP is recognized in autoimmune contexts, its translational potential in oncology has remained underexplored. By integrating machine learning algorithms with transcriptomic data from TCGA-SKCM and GSE65904 cohorts, we identified a robust 8-gene signature that effectively predicts overall survival. Beyond numerical risk stratification, our multi-omics analysis – spanning bulk RNA-seq, single-cell resolution, and immunotherapy cohorts – unveiled that VIPsig serves as a surrogate marker for the tumor immune microenvironment, specifically distinguishing between immune-inflamed and immune-desert phenotypes. Compared to established signatures, [ 9 - 11 ] VIPsig complements these by delivering stage-agnostic prognostic power (AUCs 0.76–0.77) in metastatic/mixed cohorts and by highlighting VIP-linked costimulatory inhibition and immune exclusion detectable in bulk RNA assays, which may inform future biomarker research in combination with other markers. The 8 genes in VIPsig are predominantly immune-related, encoding molecules critical for T-cell activation, costimulation, apoptosis, and cytokine signaling. CD28 , CD80 , and CD86 form the core of the costimulatory axis in T-cell priming: CD80 and CD86 (B7 ligands) bind CD28 to promote effector T-cell responses, while their interaction with CTLA4 delivers inhibitory signals to maintain immune homeostasis. [ 36 ] In melanoma, transfection of melanoma cells with CD80 or CD86 enhances antitumor immune responses by promoting T-cell proliferation, underscoring their role in bridging innate and adaptive immunity. [ 37 ] Clinicopathologic studies further indicate that elevated CD80 and CD86 expression correlates with better prognosis in melanoma, potentially reflecting enhanced antigen presentation and immune activation – patterns mirrored in our VIPsig’s negative correlation with pro-tumor macrophages and positive association with immune checkpoints. [ 38 ] CTLA4 , a key inhibitory receptor, is upregulated by IFNG signaling in melanocytes and melanoma cells, creating a feedback loop that dampens T-cell effector function and fosters immune exclusion. [ 39 , 40 ] This is evidenced by the efficacy of CTLA4 blockade (e.g., ipilimumab) in melanoma, where disrupting CTLA4 - CD80 / CD86 interactions reinvigorates exhausted T cells, though IL-21 and IL-17 signaling modulates this response to enhance checkpoint therapy outcomes. [ 41 ] FAS (CD95 ) mediates apoptosis in activated T cells and tumor cells, linking to stromal-immune crosstalk by regulating FasL-FAS interactions in the TME, which can promote tumor cell survival if dysregulated. [ 36 ] Cytokines like IFNG (interferon-gamma) drive pro-inflammatory Th1 responses essential for anti-melanoma immunity, while IL12A promotes NK and T-cell activation; conversely, IL10 acts as an immunosuppressive factor that inhibits dendritic cell maturation and fosters regulatory T cells (Tregs), contributing to immune tolerance in advanced melanoma. [ 42 ] These genes collectively participate in stromal-immune crosstalk: for instance, melanoma-associated fibroblasts (CAFs) can upregulate CTLA4 and IL10 to exclude effector cells from the tumor core, aligning with our single-cell findings of VIPsig expression in immune compartments and its correlation with tumor purity. [ 43 ] VIP may indirectly regulate this network via VPAC receptors on immune cells, amplifying CTLA4 -mediated inhibition and IL10 -driven suppression to create an immune-desert phenotype. Co-expression correlations in our data and the observed negative association with CD8+ T and NK cell infiltration support this, but direct VIP knockout studies in melanoma models are needed to confirm causality, distinguishing our correlative findings from mechanistic proof. The immune landscape analysis robustly characterized VIPsig as a marker of an immunologically “cold” or immune-excluded TME. Cross-platform deconvolution consistently revealed that a high VIPsig score was associated with reduced infiltration of antitumor effector cells (e.g., CD8+ T cells, activated NK cells) and enriched pro-tumorigenic populations (e.g., M0/M2 macrophages). This aligns with the negative correlation between VIPsig and stromal/immune scores, and its positive correlation with tumor purity, collectively depicting a TME less conducive to immune surveillance. Notably, this suppressive net effect captured by the integrated VIPsig contrasts with the pro-inflammatory functions of its individual components (e.g., CD28 , CD80 / CD86 , IFNG , IL12A ), which typically promote immune activation. This paradox may be resolved by considering VIP-mediated immunoregulatory feedback. [ 17 ] VIP signaling is known to induce tolerance, potentially creating a context where the co-expression of these activation-associated genes reflects a dysregulated or exhausted immune state rather than productive immunity. Our single-cell analysis corroborates this by showing that VIPsig-related gene expression is predominantly localized to the immune compartment, confirming the signature’s reflection of immune contexture. Importantly, a significant difference in the VIPsig score was also detected within the malignant cell subpopulation between treatment-naive and posttreatment conditions, suggesting it may also capture therapy-induced intrinsic alterations. The clinical relevance of this “cold” phenotype was further supported by analysis of the IMvigor210 cohort, where a high VIPsig score was significantly associated with the “immune-desert” phenotype. Thus, VIPsig appears to identify melanoma patients with a TME characterized by immune exclusion and potential diminished responsiveness to ICIs, [ 44 , 45 ] warranting further clinical investigation. The exploratory in silico drug sensitivity findings, while hypothesis-generating, may derive from the interplay between VIPsig genes and drug mechanisms. It is important to emphasize that these computational predictions are based on cell line-derived pharmacogenomic data and have not been validated in clinical melanoma samples or prospective trials. For instance, IFNG and IL12A, which promote inflammatory signaling, could enhance responses to immunomodulatory drugs like lenalidomide, which stimulates cytokine production and NK cell activity. [ 46 , 47 ] Conversely, the negative correlation with imatinib might relate to CTLA4 and CD28’s roles in T-cell signaling, potentially modulating tyrosine kinase pathways in the TME. [ 48 - 50 ] FAS-mediated apoptosis could influence sensitivity to DNA-damaging agents like gemcitabine, [ 51 ] where high VIPsig (potentially reduced FAS activity) might confer resistance. The immunosuppressive TME in high-VIPsig tumors could also indirectly affect drug efficacy by altering drug delivery or immune-mediated clearance. It is important to note that the initial inverse trend observed between VIPsig and TMB was not statistically significant after adjusting for key covariates including tumor stage and UV exposure (Partial ρ = −0.08, P = .117). Thus, the drug sensitivity patterns predicted for high-VIPsig tumors are likely mediated by mechanisms other than mutational load, such as the specific immune-excluded phenotype this signature captures. Differential expression analysis between high- and low-VIPsig groups identified 1338 genes, with enrichment analyses illuminating underlying biological processes. KEGG pathway enrichment pointed to cytokine-cytokine receptor interaction, cell adhesion molecules, and intestinal immune network for IgA production, emphasizing immune regulation. GO terms highlighted immune effector processes, leukocyte cell-cell adhesion, and regulation of T-cell activation, reinforcing VIPsig’s immunomodulatory role. Hallmark gene set enrichment revealed activation of apoptosis, IL2-STAT5 signaling, IL6-JAK-STAT3 signaling, inflammatory response, and interferon-related pathways in the high-VIPsig group. These enrichments tie directly to VIPsig genes: IL10 and IL12A are central to cytokine networks, modulating inflammatory responses [ 52 , 53 ] ; IFNG drives interferon pathways [ 54 ] ; and CTLA4 / CD28 regulate T-cell activation via STAT signaling. [ 49 , 55 ] The activation of apoptosis and inflammatory pathways in high-VIPsig tumors might represent a compensatory response to immune suppression, or alternatively, a tumor-promoting mechanism via chronic inflammation. This duality mirrors VIP’s pleiotropic effects, where it can suppress acute inflammation but sustain low-grade chronic states favorable for tumor growth. In acknowledging the potential limitations of the study, we recognize several key areas for improvement. First, the retrospective design limits causal inference; while VIPsig’s associations with TME and survival are consistent across cohorts, prospective clinical trials are essential to assess its utility in real-time decision-making. Second, although we adjusted for confounders like stage and purity, unmeasured variables (e.g., microbiome influences on IL10 signaling) could bias results. Third, the focus on bulk and single-cell RNA-seq overlooks posttranscriptional regulation (e.g., miRNA modulation of CTLA4), and the IMvigor210 validation showed only marginal OS differences ( P = .093), highlighting cohort-specific variability. Furthermore, it should be noted that the joint application of ComBat for batch-effect correction across the combined cohorts may have allowed some distributional information from the validation set into the preprocessing stage, which could potentially result in a slight overestimation of the signature’s performance in external validation. Fourth, in silico analyses, while computationally efficient, cannot fully replicate clinical responses; experimental validation of VIP-VPAC targeting in VIPsig-stratified models is crucial. Finally, generalizability to non-Caucasian populations or uveal melanoma remains untested, as TCGA is predominantly skin melanoma. Critically, while VIPsig demonstrated strong prognostic associations, its predictive utility for treatment response – particularly to ICIs – remains unproven. The IMvigor210 cohort analysis showed only marginal survival differences ( P = .093) and limited binary response discrimination (AUC = 0.543), highlighting that translation to therapeutic decision-making requires validation in melanoma-specific immunotherapy cohorts. Future studies should integrate proteomics and spatial transcriptomics to dissect VIP-driven stromal-immune interactions more granularly. 5. Conclusion In conclusion, VIPsig represents a novel prognostic biomarker in melanoma that is closely associated with immune suppression and inflammatory pathway activity. The exploratory in silico analyses suggest potential associations with drug sensitivity estimates, but these findings are hypothesis-generating and do not constitute validated treatment response predictions. Notably, although VIPsig was associated with immune phenotypes in the IMvigor210 cohort, its direct predictive value for immunotherapy response was limited (AUC = 0.543), underscoring the need for validation in melanoma-specific cohorts treated with ICIs. By bridging bulk and single-cell insights, this signature offers a framework for future mechanistic studies to unravel VIP’s roles in tumor immunity and to prospectively evaluate its clinical utility. Author contributions Conceptualization: Xiaogang Hong, Xiuzhen Zheng. Data curation: Xiaogang Hong, Xiuzhen Zheng. Formal analysis: Xiaogang Hong, Xiuzhen Zheng. Funding acquisition: Xiaogang Hong, Xiuzhen Zheng. Investigation: Xiaogang Hong, Xiuzhen Zheng. Methodology: Xiaogang Hong, Xiuzhen Zheng. Project administration: Xiaogang Hong, Xiuzhen Zheng. Resources: Xiaogang Hong, Xiuzhen Zheng. Software: Xiaogang Hong, Xiuzhen Zheng. Supervision: Xiaogang Hong, Xiuzhen Zheng. Validation: Xiaogang Hong, Xiuzhen Zheng. Visualization: Xiaogang Hong, Xiuzhen Zheng. Writing – original draft: Xiaogang Hong, Xiuzhen Zheng. Writing – review & editing: Xiaogang Hong, Xiuzhen Zheng. medi-105-e48961-s001.xlsx (9.4KB, xlsx) Open in a new tab medi-105-e48961-s003.xlsx (15.4KB, xlsx) Open in a new tab Open in a new tab Open in a new tab Abbreviations: DEG differentially expressed genes FDR false discovery rate GO Gene Ontology IPS immunophenoscore KEGG Kyoto Encyclopedia of Genes and Genomes MSI microsatellite instability OS overall survival ROC receiver operating characteristic TCGA The Cancer Genome Atlas TIME tumor immune microenvironment TMB tumor mutational burden Treg regulatory T-cell VIP vasoactive intestinal peptide The authors have no funding and conflicts of interest to disclose. All data generated or analyzed during this study are included in this published article (and its supplementary information files). Supplemental Digital Content is available in the online version of this article ( http://dx.doi.org/10.1097/MD.0000000000048961 ). How to cite this article: Hong X, Zheng X. Vasoactive intestinal peptide associated 8-gene signature with prognostic and immune associations in melanoma. Medicine 2026;105:22(e48961). References [1]. Karlsson MJ, Costa SF, Tebani A, et al. Inflammation and apolipoproteins are potential biomarkers for stratification of cutaneous melanoma patients for immunotherapy and targeted therapy. Cancer Res. 2021;81:2545–55. [ DOI ] [ PubMed ] [ Google Scholar ] [2]. Mao R, Yang F, Zhang T, Li J. Landscape and clinical significance of immune checkpoint in cutaneous melanoma. Front Immunol. 2021;12:756282. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] [3]. Bida M, Miya TV, Hull R, Dlamini Z. Tumor-infiltrating lymphocytes in melanoma: from prognostic assessment to therapeutic applications. Front Immunol. 2024;15:1497522. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] [4]. Hodi FS, Wolchok JD, Schadendorf D, et al. TMB and inflammatory gene expression associated with clinical outcomes following immunotherapy in advanced melanoma. Cancer Immunol Res. 2021;9:1202–13. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] [5]. Ascierto PA, Tang H, Dolfi S, et al. Effect of prior and first-line immunotherapy on baseline immune biomarkers and modulation of the tumor microenvironment in response to nivolumab and relatlimab combination therapy in patients with melanoma from RELATIVITY-020. J Immunother Cancer. 2025;13:e009773. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] [6]. Liu D, Wei B, Liang L, et al. The circadian clock component RORA Increases immunosurveillance in melanoma by inhibiting PD-L1 expression. Cancer Res. 2024;84:2265–81. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] [7]. Pavlick AC, Ariyan CE, Buchbinder EI, et al. Society for Immunotherapy of Cancer (SITC) clinical practice guideline on immunotherapy for the treatment of melanoma, version 3.0. J Immuno Ther Cancer. 2023;11:e006947. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] [8]. Vasudevan HN, Delley C, Chen WC, et al. Molecular features of resected melanoma brain metastases, clinical outcomes, and responses to immunotherapy. JAMA Netw Open. 2023;6:e2329186. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] [9]. Aung TN, Warrell J, Martinez-Morilla S, et al. Spatially informed gene signatures for response to immunotherapy in melanoma. Clin Cancer Res. 2024;30:3520–32. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] [10]. Garg M, Couturier DL, Nsengimana J, et al. Tumour gene expression signature in primary melanoma predicts long-term outcomes. Nat Commun. 1137;12:2021. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] [11]. Ayers M, Lunceford J, Nebozhyn M, et al. IFN-γ-related mRNA profile predicts clinical response to PD-1 blockade. J Clin Invest. 2017;127:2930–40. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] [12]. Yang M, Zheng C, Miao Y, et al. BTLA promoter hypomethylation correlates with enhanced immune cell infiltration, favorable prognosis, and immunotherapy response in melanoma. J ImmunoTher Cancer. 2025;13:e009841. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] [13]. Asano S, Ono A, Sakamoto K, et al. Vasoactive intestinal peptide receptor 2 signaling promotes breast cancer cell proliferation by enhancing the ERK pathway. Peptides. 2023;161:170940. [ DOI ] [ PubMed ] [ Google Scholar ] [14]. Asano S, Yamasaka M, Ozasa K, et al. Vasoactive intestinal peptide-VIPR2 signaling regulates tumor cell migration. Front Oncol. 2022;12:852358. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] [15]. Gkonos PJ, Ashby MH, Andrade AA. Vasoactive intestinal peptide stimulates prostate-specific antigen secretion by LNCaP prostate cancer cells. Regul Pept. 1996;65:153–7. [ DOI ] [ PubMed ] [ Google Scholar ] [16]. Kittikulsuth W, Nakano D, Kitada K, et al. Vasoactive intestinal peptide blockade suppresses tumor growth by regulating macrophage polarization and function in CT26 tumor-bearing mice. Sci Rep. 2023;13:927. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] [17]. Pozo D, Gonzalez-Rey E, Chorny A, Anderson P, Varela N, Delgado M. Tuning immune tolerance with vasoactive intestinal peptide: a new therapeutic approach for immune disorders. Peptides. 2007;28:1833–46. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] [18]. Chen L, Yuan W, Chen Z, et al. Vasoactive intestinal peptide represses activation of tumor-associated macrophages in gastric cancer via regulation of TNFα, IL-6, IL-12 and iNOS. Int J Oncol. 2015;47:1361–70. [ DOI ] [ PubMed ] [ Google Scholar ] [19]. Li G, Wu K, Tao K, et al. Vasoactive intestinal peptide induces CD14+HLA-DR‑/low myeloid-derived suppressor cells in gastric cancer. Mol Med Rep. 2015;12:760–8. [ DOI ] [ PubMed ] [ Google Scholar ] [20]. Li JM, Darlak KA, Southerland L, et al. VIPhyb, an antagonist of vasoactive intestinal peptide receptor, enhances cellular antiviral immunity in murine cytomegalovirus infected mice. PLoS One. 2013;8:e63381. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] [21]. Wang W, Yang H, Passang T, et al. Targeting the VIP-VPAC pathway in melanoma models inhibits tumor growth and liver metastasis. Cancer Lett. 2025;628:217855. [ DOI ] [ PubMed ] [ Google Scholar ] [22]. Yuan XH, Yao C, Oh JH, et al. Vasoactive intestinal peptide stimulates melanogenesis in B16F10 mouse melanoma cells via CREB/MITF/tyrosinase signaling. Biochem Biophys Res Commun. 2016;477:336–42. [ DOI ] [ PubMed ] [ Google Scholar ] [23]. Liberzon A, Birger C, Thorvaldsdóttir H, Ghandi M, Mesirov JP, Tamayo P. The Molecular Signatures Database (MSigDB) hallmark gene set collection. Cell Syst. 2015;1:417–25. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] [24]. Liao C, Wang X. TCGAplot: an R package for integrative pan-cancer analysis and visualization of TCGA multi-omics data. BMC Bioinf. 2023;24:483. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] [25]. Liu H, Zhang W, Zhang Y, Li X, Wanggou S. Constructing and visualizing models using mime-based machine-learning framework. J Vis Exp. 2025;22. [ DOI ] [ PubMed ] [ Google Scholar ] [26]. Liu H, Zhang W, Zhang Y, et al. A flexible machine-learning framework to construct and visualize models for clinical characteristics prediction and feature selection. Comput Struct Biotechnol J. 2024;23:2798–810. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] [27]. Wu X, Li C. Development of a prognostic model for breast cancer patients based on intratumoral tumor-infiltrating lymphocytes using machine learning algorithms. Discov Oncol. 2025;16:762. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] [28]. Zhao Y, Xun D, Chen J, Qi X. A novel machine learning-based immune prognostic signature for improving clinical outcomes and guiding therapy in colorectal cancer: an integrated bioinformatics and experimental study. BMC Cancer. 2025;25:65. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] [29]. DeLong ER, DeLong DM, Clarke-Pearson DL. Comparing the areas under two or more correlated receiver operating characteristic curves: a nonparametric approach. Biometrics. 1988;44:837–45. [ PubMed ] [ Google Scholar ] [30]. Mayakonda A, Lin DC, Assenov Y, Plass C, Koeffler HP. Maftools: efficient and comprehensive analysis of somatic variants in cancer. Genome Res. 2018;28:1747–56. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] [31]. Zeng D, Ye Z, Shen R, et al. IOBR: multi-omics immuno-oncology biological research to decode tumor microenvironment and signatures. Front Immunol. 2021;12:687975. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] [32]. Geeleher P, Cox N, Huang RS. pRRophetic: an R package for prediction of clinical chemotherapeutic response from tumor gene expression levels. PLoS One. 2014;9:e107468. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] [33]. Ritchie ME, Phipson B, Wu D, et al. limma powers differential expression analyses for RNA-sequencing and microarray studies. Nucleic Acids Res. 2015;43:e47. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] [34]. Yu G, Wang LG, Han Y, He QY. clusterProfiler: an R package for comparing biological themes among gene clusters. OMICS. 2012;16:284–7. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] [35]. Hao Y, Stuart T, Kowalski MH, et al. Dictionary learning for integrative, multimodal and scalable single-cell analysis. Nat Biotechnol. 2024;42:293–304. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] [36]. Hodi FS. Cytotoxic T-lymphocyte-associated antigen-4. Clin Cancer Res. 2007;13(18 Pt 1):5238–42. [ DOI ] [ PubMed ] [ Google Scholar ] [37]. Rudy W, Gückel B, Siebels M, Lindauer M, Meuer SC, Moebius U. Differential function of CD80- and CD86-transfected human melanoma cells in the presence of IL-12 and IFN-gamma. Int Immunol. 1997;9:853–60. [ DOI ] [ PubMed ] [ Google Scholar ] [38]. Aytekin EC, Unal B, Bassorgun CI, Ozkan O. Clinicopathologic evaluation of CD80, CD86, and PD-L1 expressions with immunohistochemical methods in malignant melanoma patients. Turk Patoloji Derg. 2024;40:16–26. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] [39]. Mo X, Zhang H, Preston S, et al. Interferon-γ signaling in melanocytes and melanoma cells regulates expression of CTLA-4. Cancer Res. 2018;78:436–50. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] [40]. Váraljai R, Zimmer L, Al-Matary Y, et al. Interleukin 17 signaling supports clinical benefit of dual CTLA-4 and PD-1 checkpoint inhibition in melanoma. Nat Cancer. 2023;4:1292–308. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] [41]. Zhang Z, Langenbach M, Sagar S, et al. Efficacy of CTLA-4 checkpoint therapy is dependent on IL-21 signaling to mediate cytotoxic reprogramming of PD-1(+)CD8(+) T cells. Nat Immunol. 2025;26:92–104. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] [42]. Tsai AK, Khan AY, Worgo CE, Wang LL, Liang Y, Davila E. A multikinase and DNA-PK inhibitor combination immunomodulates melanomas, suppresses tumor progression, and enhances immunotherapies. Cancer Immunol Res. 2017;5:790–803. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] [43]. Loser K, Scherer A, Krummen MB, et al. An important role of CD80/CD86-CTLA-4 signaling during photocarcinogenesis in mice. J Immunol. 2005;174:5298–305. [ DOI ] [ PubMed ] [ Google Scholar ] [44]. Ravindranathan S, Passang T, Li JM, et al. Targeting vasoactive intestinal peptide-mediated signaling enhances response to immune checkpoint therapy in pancreatic ductal adenocarcinoma. Nat Commun. 6418;13:2022. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] [45]. Waller EK, Chaudagar K. Pancreatic cancer cell-intrinsic transglutaminase-2 promotes T cell suppression through microtubule-dependent secretion of immunosuppressive cytokines. J ImmunoTher Cancer. 2025;13:e012188. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] [46]. Ishiguro K, Kitajima H, Niinuma T, et al. DOT1L inhibition reprograms innate immunity to potentiate immunomodulatory drug responses in multiple myeloma. Cancer Lett. 2025;631:217941. [ DOI ] [ PubMed ] [ Google Scholar ] [47]. Zebley CC, Abdelsamed HA, Ghoneim HE, et al. Proinflammatory cytokines promote TET2-mediated DNA demethylation during CD8 T cell effector differentiation. Cell Rep. 2021;37:109796. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] [48]. Daly RJ, Scott AM, Klein O, Ernst M. Enhancing therapeutic anti-cancer responses by combining immune checkpoint and tyrosine kinase inhibition. Mol Cancer. 2022;21:189. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] [49]. Lin S, Cheng L, Ye W, et al. Chimeric CTLA4-CD28-CD3z T Cells potentiate antitumor activity against CD80/CD86-Positive B cell malignancies. Front Immunol. 2021;12:642528. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] [50]. Dugage DM, Jones RL, Trent J, Champiat S, Dumont S. Beyond the driver mutation: immunotherapies in gastrointestinal stromal tumors. Front Immunol. 2021;12:715727. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] [51]. Sugihara E, Hashimoto N, Osuka S, et al. The Inhibitor of apoptosis protein livin confers resistance to fas-mediated immune cytotoxicity in refractory lymphoma. Cancer Res. 2020;80:4439–50. [ DOI ] [ PubMed ] [ Google Scholar ] [52]. Carlini V, Noonan DM, Abdalalem E, et al. The multifaceted nature of IL-10: regulation, role in immunological homeostasis and its relevance to cancer, COVID-19 and post-COVID conditions. Front Immunol. 2023;14:1161067. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] [53]. Kureshi CT, Dougan SK. Cytokines in cancer. Cancer Cell. 2025;43:15–35. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] [54]. Kupke P, Adenugba A, Schemmerer M, et al. Immunomodulation of natural killer cell function by ribavirin involves TYK-2 activation and subsequent increased IFN-γ secretion in the context of in vitro hepatitis E virus infection. Cells. 2023;12:453. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] [55]. Iglesias M, Khalifian S, Oh BC, et al. A short course of tofacitinib sustains the immunoregulatory effect of CTLA4-Ig in the presence of inflammatory cytokines and promotes long-term survival of murine cardiac allografts. Am J Transplant. 2021;21:2675–87. [ DOI ] [ PubMed ] [ Google Scholar ] Associated Data This section collects any data citations, data availability statements, or supplementary materials included in this article. Supplementary Materials medi-105-e48961-s001.xlsx (9.4KB, xlsx) medi-105-e48961-s003.xlsx (15.4KB, xlsx) Articles from Medicine are provided here courtesy of Wolters Kluwer Health ACTIONS View on publisher site PDF (5.4 MB) Cite Collections Permalink RESOURCES Similar articles Cited by other articles Links to NCBI Databases Follow NCBI NCBI on X (formerly known as Twitter) NCBI on Facebook NCBI on LinkedIn NCBI on GitHub NCBI RSS feed Connect with NLM NLM on X (formerly known as Twitter) NLM on Facebook NLM on YouTube National Library of Medicine 8600 Rockville Pike Bethesda, MD 20894 Web Policies FOIA HHS Vulnerability Disclosure Help Accessibility Careers NLM NIH HHS USA.gov Back to Top

Related documents

• Variations of Immune Infiltrate and Cell Plasticity Markers in Treated Metastatic Melanoma Patients#405147
• A single-cell and machine learning framework identifies CAFs-associated signatures linking stromal heterogeneity to immune regulation in pancreatic cancer.#360644
• Adjuvant Therapy of Skin Melanoma With Alpha Interferon and Naderin#651236
• Identifying the Predictive Factors of Response to PD-1 or PD-L1 Antagonists#842712
• From Germline Variants to Tumor Outcome: GWAS-Based Functional Genomics Prioritizes Colorectal Cancer Susceptibility Genes and Links SMAD9 to Prognosis.#476905