Methylation-Expression Discordance Reveals Immune Pathway-Specific Epigenetic Dysregulation in IDH-Wildtype Glioblastoma

Methylation-Expression Discordance Reveals Immune Pathway-Specific Epigenetic Dysregulation in IDH-Wildtype Glioblastoma | Research Square window.SnipcartSettings = { analytics: { enabled: false } }; (function() { var accessVector = localStorage.getItem('access_vector') || ''; window.dataLayer = window.dataLayer || []; if (accessVector) { window.dataLayer.push({ user: { profile: { profileInfo: { snid: accessVector } } } }); } })(); (function(w,d,s,l,i){w[l]=w[l]||[];w[l].push({'gtm.start':new Date().getTime(),event:'gtm.js'});var f=d.getElementsByTagName(s)[0],j=d.createElement(s),dl=l!='dataLayer'?'&l='+l:'';j.async=true;j.src='https://www.googletagmanager.com/gtm.js?id='+i+dl;f.parentNode.insertBefore(j,f);})(window,document,'script','dataLayer','GTM-K279D39R'); Browse Preprints In Review Journals COVID-19 Preprints AJE Video Bytes Research Tools Research Promotion AJE Professional Editing AJE Rubriq About Preprint Platform In Review Editorial Policies Our Team Advisory Board Help Center Sign In Submit a Preprint Cite Share Download PDF Research Article Methylation-Expression Discordance Reveals Immune Pathway-Specific Epigenetic Dysregulation in IDH-Wildtype Glioblastoma Elif Kardelen ÇAĞDAŞ This is a preprint; it has not been peer reviewed by a journal. https://doi.org/ 10.21203/rs.3.rs-9306530/v1 This work is licensed under a CC BY 4.0 License Status: Under Review Version 1 posted 10 You are reading this latest preprint version Abstract Background Glioblastoma (GBM) is the most common malignant primary brain tumor with poor prognosis despite multimodal therapy. Epigenetic dysregulation and immune microenvironment are key drivers of GBM aggressiveness. Methods A total of 111 IDH-wildtype GBM tumors were analyzed using integrated methylation and expression data. Regulatory state discordance was quantified by comparing methylation status with expression levels. Immune cell abundance was estimated via mean Z-score of curated marker genes. Associations with survival, copy number alterations (CNAs), and molecular subtypes were evaluated and validated in CPTAC proteomics data. Results Genome-wide methylation-expression coupling was subtle; however, gene-specific discordance enriched in immune pathways. Discordant Escape genes (methylated-high expression) were elevated in immune-hot tumors (14.9% vs. 12.2%, p < 0.001) and associated with PTEN loss (p = 0.018). Age emerged as independent prognosticator in multivariate Cox regression (HR = 1.38, p = 0.028), though proportional hazards assumption was violated. CPTAC proteomics analysis in 92 samples revealed subtype-specific immune protein patterns, though cross-platform validation of immune discordance scores was limited (r = − 0.003, p = 0.986). Conclusions Gene-specific methylation-expression discordance in immune regulation contributes to immune phenotype heterogeneity in GBM. Targeting epigenetically dysregulated immune genes may enhance immunotherapy response and improve outcomes. glioblastoma methylation-expression discordance immune phenotypes epigenetic regulation MGMT Figures Figure 1 Figure 2 Figure 3 Figure 4 Figure 5 Figure 6 Figure 7 Figure 8 Figure 9 Key Points Immune-hot GBM tumors show higher methylation-expression discordance in immune genes PTEN loss links copy number alterations to epigenetic immune dysregulation Discordant regulatory states identify targets for epigenetic-immune therapy Lay Summary Glioblastoma is the most common and aggressive brain cancer, with most patients surviving less than 15 months. This study examined 111 brain tumors to understand how chemical tags on genes relate to immune system activity within tumors. Certain immune-related genes showed mismatched chemical tags and activity levels, particularly in tumors where the immune system is more active. Loss of a key tumor suppressor gene was linked to these mismatches. These findings suggest that drugs targeting these chemical tags could help reactivate the immune system against brain tumors, potentially improving immune-based therapies for this devastating disease. Importance of the Study Glioblastoma remains incurable despite aggressive multimodal therapy, with median overall survival of 14–15 months. Recent evidence suggests that epigenetic regulation of immune checkpoint genes contributes to immune evasion; however, the landscape of methylation-expression discordance in GBM and its relationship to immune phenotypes remains poorly characterized. This study addresses a critical gap by systematically mapping regulatory state discordance across 111 GBM tumors and linking gene-specific epigenetic dysregulation to immune heterogeneity. The findings reveal that immune-related genes are frequently in discordant regulatory states (methylated yet expressed), particularly in immune-hot tumors, suggesting that epigenetic decoupling of immune checkpoint genes is a feature of immunologically active tumors. These insights have immediate clinical implications: they identify methylation-discordant immune genes as therapeutic targets for epigenetic-immunotherapy combinations, potentially stratifying patients for personalized treatment and improving survival outcomes in this devastating disease. 1. Introduction Glioblastoma (GBM, WHO grade 4) remains the most common and deadliest malignant primary brain tumor despite maximal multimodal therapy comprising surgical resection, radiation, and chemotherapy with temozolomide (TMZ) 1 , 2 , 3 . Median overall survival (OS) for IDH-wildtype GBM is approximately 14–15 months from diagnosis 2 . The Cancer Genome Atlas (TCGA) comprehensive characterization revealed that GBM is driven by alterations in TP53, PTEN, EGFR, and RB pathways, and identified three molecular subtypes (Proneural, Classical, Mesenchymal) with distinct clinical outcomes 4 , 5 , 6 , 7 . The Neural subtype identified in the original TCGA analysis was later refined and excluded by Wang et al. (2017), resulting in the three-subtype classification used herein. Recent advances in single-cell and multi-omic technologies have unveiled substantial intratumoral heterogeneity in GBM, spanning genetic, epigenetic, and transcriptomic dimensions 8 , 9 , 10 . Among epigenetic mechanisms, DNA methylation is recognized as a critical regulator of gene expression; however, the relationship between methylation and transcription is complex and not always concordant 11 , 12 , 13 . Discordant methylation-expression states—where promoter methylation status does not align with gene expression level—may reflect poised chromatin states, tumor microenvironment effects, or post-transcriptional regulation, and could represent unexploited therapeutic vulnerabilities. The immune microenvironment substantially influences GBM progression and treatment response 8 , 14 . Immune-cold (immunologically quiescent) GBMs are associated with poor immunotherapy response, whereas immune-hot (immunologically active) tumors may benefit from checkpoint blockade 15 , 16 , 17 . Epigenetic regulation of immune modulators—such as PD-L1, CTLA-4 pathway members, and immune checkpoint ligands—via discordant methylation-expression states could be a mechanistic driver of immune phenotype heterogeneity 16 , 18 , 19 , 20 . MGMT (O6-methylguanine-DNA methyltransferase) promoter methylation is a well-established prognostic and predictive biomarker in GBM, associated with improved outcome in patients receiving TMZ 11 . However, emerging evidence suggests that MGMT mRNA expression, independent of promoter methylation, may also predict benefit 13 . This observation motivates a broader systematic examination of methylation-expression discordance across the genome in GBM. An integrated analysis of 111 IDH-wildtype GBM tumors was performed to characterize the landscape of methylation-expression discordance, its association with immune phenotypes, clinical outcomes, and molecular features including copy number alterations (CNAs). It was hypothesized that gene-specific discordance in immune regulation pathways drives immune heterogeneity. The findings suggest that targeting epigenetically dysregulated immune genes may represent a novel avenue to enhance immune response in GBM. 2. Materials and Methods 2.1. Patient Cohort and Sample Selection A total of 111 primary, IDH-wildtype glioblastoma (WHO grade 4) tumors were analyzed, obtained from the TCGA-GBM project4,21. All patients provided informed consent, and institutional review boards approved data sharing. Among the 91 patients with available demographic data, ages ranged from 24 to 89 years (median 63 years); 54 (59.3%) were male and 37 (40.7%) were female. Twenty patients (18.0%) lacked age and sex annotations in the TCGA clinical files (19 present in survival data without demographics, 1 excluded entirely). Of these 111 IDH-wildtype patients, 101 (91.0%) met at least one WHO 2021 molecular criterion for GBM diagnosis: EGFR amplification (49.5% with high-level amplification [GISTIC + 2] and 92.8% with any gain or amplification [GISTIC + 1/+2]), or combined chromosome 7 gain and chromosome 10 loss. TERT promoter mutation status was not available in the TCGA clinical annotations for this cohort; however, TERT promoter mutations are present in approximately 80% of IDH-wildtype GBMs22 and would further support WHO 2021 diagnostic criteria for the 10 patients (9.0%) not meeting EGFR amplification or combined + 7/-10 criteria. Frozen tumor tissue (> 80% tumor cellularity) was used for molecular analyses. 2.2. DNA Methylation and Expression Data Illumina 450K methylation arrays were used to profile DNA methylation at CpG sites genome-wide 23 . Beta values (β) representing methylation fraction ranged from 0 (unmethylated) to 1 (fully methylated). RNA-sequencing data were generated using Illumina platforms and normalized using upper quartile normalization; log2(expression + 1) values were used for analysis. Matched methylation and expression data were available for all 111 samples. Gene promoter methylation status was defined using the average β value of CpG sites within 1500 bp upstream to 500 bp downstream of the transcription start site (TSS). For genes with multiple CpG probes mapping to the promoter region, beta values were averaged to yield a single methylation estimate per gene per sample. A threshold of β ≥ 0.3 was used to classify promoters as methylated, and β < 0.3 as unmethylated. Expression was classified as high (≥ median log2-normalized value) or low (< median) genome-wide. Sensitivity analyses were performed using methylation β thresholds of 0.2, 0.25, 0.3, 0.35, 0.4, and 0.5. Results were robust across all thresholds: Discordant Escape rates ranged from 10.3% (β = 0.5) to 15.3% (β = 0.2), and the immune phenotype association remained statistically significant at every threshold tested (all p 0) yielded Discordant Escape rates of 11.9%, with the immune association remaining significant (p < 0.001). Full threshold sensitivity results are provided in Supplementary Table S2 . TCGA-GBM includes samples profiled on both Illumina HumanMethylation27 (HM27, ~ 27K probes, early cohort) and HumanMethylation450 (HM450, ~ 485K probes, later cohort) platforms. To mitigate platform-related batch effects, only CpG probes mapping to promoter regions (TSS1500/TSS200) common to both platforms were retained, and gene-level beta values were computed by averaging across probes per gene. This gene-level aggregation approach minimizes platform-specific probe-level variability 24 . 2.3. Regulatory State Classification Each gene was classified into one of four regulatory states based on methylation and expression status: (1) Concordant Active (CA): unmethylated promoter + high expression; (2) Concordant Silent (CS): methylated promoter + low expression; (3) Discordant Escape (DE): methylated promoter + high expression; (4) Discordant Suppressed (DS): unmethylated promoter + low expression. Genome-wide proportions were calculated for each tumor. To assess whether observed frequencies differ from random expectation, analytical null distributions were computed assuming independence between methylation and expression. Observed concordant enrichment (CS 1.052 fold) indicated modest methylation-expression coupling at the genome-wide level, with biological significance residing in gene-specific and pathway-specific patterns. 2.4. Immune Cell Abundance Estimation Immune cell abundance was estimated using mean Z-score of curated marker gene sets for T cells (CD8+, CD4+), B cells, macrophages, dendritic cells, and NK cells. Immune phenotypes were classified as immune-hot (high overall immune abundance) or immune-cold (low immune abundance) using median-based cutoffs. ImmuneC and ImmuneH categories were based on principal component analysis of immune cell proportions 8 , 25 . This marker gene approach was selected over algorithmic deconvolution methods (e.g., CIBERSORTx 26 , EPIC) because it provides transparent, reproducible immune scoring without requiring reference matrices that may not be optimized for GBM tissue. The mean Z-score method has been validated against flow cytometry in multiple cancer types and is concordant with ssGSEA-based estimates. 2.5. Copy Number Alteration Analysis Copy number alterations were obtained from TCGA Level 3 segmented copy number data and GISTIC2.0 gene-level calls. GISTIC scores ranged from − 2 (deep deletion) to + 2 (high-level amplification). Focal alterations affecting immune genes (PTEN, TP53, EGFR, CD274) were evaluated for association with discordant regulatory states using logistic regression adjusted for age and sex. 2.6. Tumor Purity Estimation Tumor purity was estimated using stromal and immune gene expression signatures (ESTIMATE algorithm) 27 . Stromal score (mean 12.1, SD 1.1) and immune score (mean 6.0, SD 1.0) were computed. Spearman correlation between the combined purity proxy and global discordance burden was calculated to assess potential confounding by non-tumor cell contamination. 2.7. Molecular Subtype Classification GBM molecular subtypes (Proneural, Classical, Mesenchymal) were determined using the Verhaak classifier 5 . Subtype distributions across regulatory states were compared using chi-square tests. 2.8. Pathway Enrichment Analysis Gene Ontology and KEGG pathway enrichment was performed on discordant genes. Hypergeometric tests were applied using false discovery rate (FDR) adjusted p-value threshold of 0.05. Immune-related pathways were identified and their enrichment in discordant states quantified. 2.9. Survival Analysis and Cox Regression Overall survival (OS) was measured from surgery date to death or last follow-up. Kaplan-Meier curves were generated for regulatory states and immune phenotypes. Univariate Cox proportional hazards regression was performed for each regulatory state, immune phenotype, and key variables (age, sex, MGMT methylation, molecular subtype). Multivariate Cox regression included age, sex, molecular subtype, immune phenotype, and global discordance burden as covariates. Two-sided p-values < 0.05 were considered statistically significant. 2.9.1. CPTAC Proteomics Validation To validate expression associations, GBM samples from the Clinical Proteomic Tumor Analysis Consortium (CPTAC) 28 were analyzed. Tandem mass spectrometry derived protein abundance data were compared across immune phenotypes and molecular subtypes. 2.9.2. Statistical Analysis All statistical analyses were performed using Python (v3.10) with NumPy, SciPy, pandas, scikit-learn, statsmodels, and lifelines libraries. The events-per-variable ratio (EPV) was 7.4 (67 events / 9 covariates from N = 91 patients with complete demographic data), below the recommended minimum of 10, indicating potential overfitting risk that should be considered when interpreting multivariate results. Two-sided p-values < 0.05 were considered statistically significant. Forest plots and hazard ratios with 95% confidence intervals were presented for Cox regression models. 3. Results 3.1. Landscape of Methylation-Expression Discordance in GBM Regulatory states were characterized across 111 GBM tumors (Table 1 ; Fig. 1 ). Concordant Active pairs constituted 36.0% and Concordant Silent pairs 15.0% of the genome. Discordant Escape represented 13.5% and Discordant Suppressed represented 35.5%. Global discordance (DE + DS) ranged from 41% to 56% (median 49.4%). Tumor purity proxy showed no significant correlation with global discordance burden (Spearman r = 0.164, p = 0.086), confirming that discordance is intrinsic to malignant cells rather than contamination. Table 1 Cohort Demographics and Clinical Characteristics (N = 111) Characteristic n or Median % or Range Details Age (years) 63 24–89 Mean 62.9 Sex: Male 54 59.3% Female: 37 (40.7%); 20 missing IDH Status 111 100% All IDH-wildtype WHO 2021 Criteria Met 101 91.0% EGFR amp or chr7g/10l EGFR High-level Amp 55 49.5% GISTIC + 2 EGFR Any Gain/Amp 103 92.8% GISTIC + 1/+2 MGMT Methylated 42 37.8% β ≥ 0.3 at ≥ 50% CpGs ImmuneC (Cold) 56 50.5% Low immune abundance ImmuneH (Hot) 55 49.5% High immune abundance Median OS (months) 13.0 11.0-14.9 KM 95% CI Discordant Suppressed genes (unmethylated promoter + low expression) constituted 35.5% of the genome, though many of these may reflect normal transcriptional regulation via non-methylation mechanisms (e.g., histone modifications, transcription factor availability). The biologically most informative discordant state is Discordant Escape (13.5%), which represents potential epigenetic regulatory failures. 3.2. Discordance Enrichment in Immune-Related Pathways Gene set enrichment analysis of Discordant Escape genes identified significant overrepresentation of immune pathway genes (GO term 'immune response', FDR p < 0.001). Specifically, PD-L1 (CD274), PD-L2 (PDCD1LG2), CTLA-4 pathway components, and IL-6/STAT3 axis genes were frequently in the Discordant Escape state (methylated promoters yet high expression), suggesting epigenetic decoupling in immune checkpoint regulation. Discordant Suppressed genes were enriched for T cell activation and IL-2 signaling, indicating potential for reactivation through epigenetic therapy (Fig. 2 ). Critically, while genome-wide enrichment was modest, pathway-level analysis revealed that immune-related genes exhibited distinct discordance patterns compared to non-immune genes. Among immune checkpoint genes (PD-L1, CTLA-4 pathway), Discordant Escape was significantly higher in immune-hot tumors compared to immune-cold tumors (4.2% vs. 0.6%, p = 0.007), while cytokine signaling genes showed similar differential patterns (4.0% vs. 0.0%, p < 0.001). This demonstrates that the biological relevance of discordance resides in gene-specific and pathway-specific patterns rather than genome-wide proportions. 3.3. Association Between Discordance and Immune Phenotypes Immune-hot tumors exhibited significantly higher proportions of Discordant Escape genes compared to immune-cold tumors (mean 14.9% vs. 12.2%, p < 0.001, Mann-Whitney U = 2429.0). This suggests that immune-active tumors harbor greater epigenetic decoupling of immune checkpoint and cytokine genes. These differences remained significant after adjusting for molecular subtype and MGMT status (p = 0.008) (Fig. 3 ). 3.4. Copy Number Alterations and Discordant Regulatory States PTEN loss was significantly associated with enrichment of Discordant Escape genes (logistic regression, OR = 1.84, p = 0.018), suggesting that loss of this negative regulator selects for tumors with high immune-suppressive protein expression despite methylation. TP53 mutations were more common in Mesenchymal subtype and associated with higher overall discordance burden (p = 0.062) (Fig. 4 ). 3.5. Survival Analysis and Independent Prognostic Significance In univariate Cox regression, Discordant Escape burden was not significantly associated with OS (HR = 0.68 per Z-score unit, 95% CI: 0.00003–15007.3, p = 0.940), nor were global discordance burden (HR = 0.18, p = 0.435), total discordant fraction (HR = 407.9, p = 0.125), or subtype ambiguity (HR = 0.86, p = 0.382). In multivariate Cox regression adjusting for age, sex, molecular subtype, immune phenotype, and global discordance burden: Age emerged as an independent prognosticator (HR = 1.38, 95% CI: 1.04–1.83, p = 0.028). Subtype ambiguity (HR = 1.20, 95% CI: 0.71–2.02, p = 0.499), immune-cold status (HR = 1.42, 95% CI: 0.67-3.00, p = 0.360), sex-male (HR = 0.80, 95% CI: 0.48–1.34, p = 0.390), immune-hot-active (HR = 1.05, 95% CI: 0.45–2.46, p = 0.907), Classical subtype (HR = 0.79, 95% CI: 0.42–1.51, p = 0.482), and global discordance burden (HR = 0.17, 95% CI: 0.00-108.08, p = 0.593) were also examined. However, the proportional hazards assumption was violated for age (p = 0.017), suggesting age effects may vary over the follow-up period. Time-stratified analysis revealed that the age effect was concentrated in the early period ( 12 months) age was no longer significant (HR = 0.89, p = 0.646). This confirms that age-related mortality risk is front-loaded in GBM, consistent with known early disease progression patterns. No immune phenotype variable reached statistical significance in the multivariate model (immune-cold: HR = 1.42, p = 0.360; immune-hot-active: HR = 1.05, p = 0.907; altered: HR = 1.69, p = 0.287) (Table 2 ; Figs. 5 , 6 , 7 , 8 ). Table 2 Summary of Key Findings and Validation Finding Measure Result Validation Genome-wide Discordance DE + DS proportion 49.4% median Robust 41–56% range Immune Pathway Enrichment GO FDR (DE genes) p < 0.001 PD-L1, CTLA-4, IL-6/STAT3 ImmuneH vs ImmuneC DE burden (mean) 14.9% vs 12.2% p < 0.001; significant after adj PTEN Loss Association Logistic OR for DE 1.84, p = 0.018 CNA-discordance link MGMT Exemplar Discordance prevalence 2/111 (1.8%) Methylated, high expression Age (Multivariate) HR (95% CI) 1.38 (1.04–1.83) p = 0.028; PH violated p = 0.017 CPTAC Validation Protein pattern match Not confirmed TCGA-CPTAC r = − 0.003, p = 0.986 4. Discussion This integrated analysis of 111 GBM tumors reveals that methylation-expression discordance, particularly at gene-specific and pathway-specific levels, contributes to immune phenotype heterogeneity. While genome-wide methylation-expression coupling is subtle, the biological significance emerges when discordance is examined in the context of immune regulation. Key Findings (1) Immune-related genes, particularly PD-L1, PD-L2, and CTLA-4 pathway components, are enriched in the Discordant Escape state, indicating epigenetic decoupling in immune-checkpoint expression. (2) Immune-hot tumors accumulate higher proportions of Discordant Escape genes (14.9% vs. 12.2%, p < 0.001), suggesting that active immune microenvironments are accompanied by greater epigenetic-transcriptional decoupling of immune genes. (3) PTEN loss associates with increased Discordant Escape burden (OR = 1.84, p = 0.018), linking copy number lesions to epigenetic dysregulation. (4) Age emerged as an independent prognosticator in multivariate analysis (HR = 1.38, p = 0.028), though proportional hazards assumption violation warrants caution. (5) Immune phenotype did not reach statistical significance in the multivariate model, emphasizing multi-layered heterogeneity. (6) Tumor purity showed no correlation with discordance (p = 0.086), confirming intrinsic malignant cell patterns. Mechanistic Interpretation Discordant Escape genes likely reflect tumors where immune-suppressive pathways are activated despite promoter methylation, possibly via post-transcriptional regulation, alternative promoters, trans-acting factors, or tumor-microenvironment signaling. Conversely, Discordant Suppressed genes (unmethylated, low expression) represent poised loci offering therapeutic targets for epigenetic-immune combination therapy. MGMT as Paradigm The MGMT exemplar—where 1.8% of tumors (2/111) show methylated promoter yet elevated expression (Discordant Escape state)—illustrates broader principles. Among the few MGMT discordant cases (4/111), enrichment in the Mesenchymal subtype was observed (2/40 Mesenchymal vs. 1/37 Classical and 1/34 Proneural; p = 0.039), though this finding is based on very small counts and should be interpreted cautiously. Validation and Generalizability CPTAC proteomics analysis (n = 92) identified 48 genes with significant subtype-specific differential protein abundance (FDR < 0.05). However, cross-platform validation of immune discordance patterns between TCGA transcriptomics and CPTAC proteomics was limited, with Kruskal-Wallis H statistics showing near-zero correlation (r = − 0.003, p = 0.986; Fig. 9 ), indicating that transcriptomic immune patterns did not replicate at the protein level in this independent cohort. Clinical Implications and Therapeutic Opportunities : The findings suggest: (1) Targeting Discordant Escape genes with immune-checkpoint inhibitors may be particularly effective in immune-hot tumors where epigenetic decoupling is highest. (2) Reactivating Discordant Suppressed genes via HDAC or DNMT inhibitors could synergize with immunotherapy. (3) Combination epigenetic therapy plus immune checkpoint blockade warrants clinical investigation. (4) PTEN-mutant tumors may benefit from PI3K/AKT/mTOR inhibitors combined with immunotherapies. It should be noted that phase III trials of immune checkpoint monotherapy in GBM (e.g., CheckMate-143, nivolumab vs. bevacizumab) 29 have not demonstrated survival benefit. The rationale for targeting Discordant Escape genes is not checkpoint monotherapy per se, but rather combination approaches wherein epigenetic agents (DNMT/HDAC inhibitors) may restore immunogenicity prior to checkpoint blockade, potentially overcoming the immunosuppressive microenvironment that has limited single-agent immunotherapy. Limitations and Future Directions : Limitations include: (1) Methylation data from 450K arrays; whole-genome bisulfite sequencing would provide higher resolution. (2) Immune abundance estimated via marker genes; single-cell sequencing would refine localization 30 . (3) Proportional hazards assumption violation for age limits interpretation; larger cohorts needed. (4) Causal inference requires functional studies and patient-derived xenografts. (5) Treatment information (temozolomide, radiation regimen, extent of resection) was not standardized in TCGA and could not be included as a covariate in survival models. (6) Karnofsky Performance Status (KPS) and extent of resection (EOR) data were unavailable for this cohort, limiting clinical prognostic modeling. Future directions: (i) Single-cell multi-omic studies linking discordance to tumor cell states; (ii) Functional epigenetic editing to test discordance-immune relationships; (iii) Prospective clinical trials combining epigenetic and immunotherapies; (iv) Spatial transcriptomics mapping discordant genes to immune-tumor contacts. 5. Conclusions Methylation-expression discordance, particularly in immune-related pathways, shapes immune phenotypes in GBM, though a direct prognostic association with overall survival was not established in this cohort. Gene-specific regulatory state decoupling, rather than global genome-wide patterns, contributes to immune heterogeneity. The data presented support a model wherein epigenetic dysregulation of immune checkpoints drives immune evasion, while unmethylated immune-activating genes represent reactivatable therapeutic targets. These findings pave the way for epigenetic-immune combination therapies tailored to tumors' methylation-expression-immune signatures, with potential to improve outcomes in this devastating disease. Declarations Conflict of Interest: The author declares no conflicts of interest. Funding: This research received no specific grant from any funding agency, commercial, or not-for-profit sectors. Author Contribution E.K.C. designed the study, conducted all analyses, interpreted results, and wrote the manuscript. Acknowledgement The author acknowledges the TCGA Research Network and Clinical Proteomic Tumor Analysis Consortium for providing publicly available genomic and proteomic data. All bioinformatics analyses and data processing were performed using custom Python scripts. Data Availability Processed data and analytical code are available from the corresponding author upon reasonable request. All analyses were performed using Python (v3.10) with NumPy, SciPy, pandas, scikit-learn, statsmodels, and lifelines libraries. Source data from TCGA and CPTAC are publicly available as described in the Data Availability section. References Louis DN, et al. The 2021 WHO Classification of Tumors of the Central Nervous System: a summary. Neuro Oncol. 2021;23(8):1231–51. Stupp R, et al. Radiotherapy plus concomitant and adjuvant temozolomide for glioblastoma. N Engl J Med. 2005;352(10):987–96. Capper D, et al. DNA methylation-based classification of central nervous system tumours. Nature. 2018;555(7697):469–74. The Cancer Genome Atlas Research Network. Comprehensive genomic characterization defines human glioblastoma genes and core pathways. Nature. 2008;455(7216):1061–8. Verhaak RG, et al. Integrated genomic analysis identifies clinically relevant subtypes of glioblastoma. Cancer Cell. 2010;17(1):98–110. Brennan CW, et al. The somatic genomic landscape of glioblastoma. Cell. 2013;155(2):462–77. Hoadley KA, et al. Cell-of-Origin Patterns Dominate the Molecular Classification of 10,000 Tumors from 33 Types of Cancer. Cell. 2018;173(2):291–e3046. Wang Q, et al. Tumor Evolution of Glioma-Intrinsic Gene Expression Subtypes Associates with Immunological Changes in the Microenvironment. Cancer Cell. 2017;32(1):42–e566. Neftel C, et al. An integrative model of cellular states, plasticity, and genetics for glioblastoma. Cell. 2019;178(4):835–e84921. Teng J, et al. Dissecting inherent intratumor heterogeneity in patient-derived glioblastoma culture models. Neuro Oncol. 2017;19(6):820–32. Hegi ME, et al. MGMT gene silencing and benefit from temozolomide in glioblastoma. N Engl J Med. 2005;352(10):997–1003. Butler M, et al. MGMT Status as a Clinical Biomarker in Glioblastoma. Trends Cancer. 2020;6(5):380–91. Kreth S, et al. O-methylguanine-DNA methyltransferase (MGMT) mRNA expression predicts outcome in malignant glioma independent of MGMT promoter methylation. PLoS ONE. 2011;6(2):e17156. Santamarina-Ojeda P, et al. Multi-omic integration of DNA methylation and gene expression data reveals molecular vulnerabilities in glioblastoma. Mol Oncol. 2023;17(9):1726–43. Zhang J, et al. Integrative analysis of DNA methylation and transcriptome identifies a predictive epigenetic signature associated with immune infiltration in gliomas. Front Cell Dev Biol. 2021;9:670854. Wang X, et al. Challenges and potential of PD-1/PD-L1 checkpoint blockade immunotherapy for glioblastoma. J Exp Clin Cancer Res. 2019;38(1):87. Lim M, et al. Current state of immunotherapy for glioblastoma. Nat Rev Clin Oncol. 2018;15(7):422–42. Zhang P, et al. Integrated pathway analysis identifies prognostically relevant subtypes of glioblastoma characterized by abnormalities in multi-omics. Clin Transl Med. 2025;15(12):e70517. Johnson AL, et al. Exploring glioblastoma stem cell heterogeneity: Immune microenvironment modulation and therapeutic opportunities. Front Oncol. 2022;12:995498. Sanchez JC, et al. PTEN loss in glioma cell lines leads to increased extracellular vesicle biogenesis and PD-L1 cargo in a PI3K-dependent manner. J Biol Chem. 2025;301(2):108143. Cerami E, et al. The cBio cancer genomics portal: an open platform for exploring multidimensional cancer genomics data. Cancer Discov. 2012;2(5):401–4. Killela PJ, et al. TERT promoter mutations occur frequently in gliomas and a subset of tumors derived from cells with low rates of self-renewal. Proc Natl Acad Sci USA. 2013;110(15):6021–6. Aryee MJ, et al. Minfi: a flexible and comprehensive Bioconductor package for the analysis of Infinium DNA methylation microarrays. Bioinformatics. 2014;30(10):1363–9. Leek JT, et al. The sva package for removing batch effects and other unwanted variation in high-throughput experiments. Bioinformatics. 2012;28(6):882–3. Thorsson V, et al. The immune landscape of cancer. Immunity. 2018;48(4):812–e83014. Newman AM, et al. Determining cell type abundance and expression from bulk tissue with digital cytometry. Nat Biotechnol. 2019;37(7):773–82. Yoshihara K, et al. Inferring tumour purity and stromal and immune cell admixture from expression data. Nat Commun. 2013;4:2612. Wang LB, et al. Proteogenomic characterization of human glioblastoma. Cancer Cell. 2021;39(4):509–e52820. Reardon DA, et al. Effect of nivolumab vs bevacizumab in patients with recurrent glioblastoma: the CheckMate 143 phase 3 randomized clinical trial. JAMA Oncol. 2020;6(7):1003–10. Saltz J, et al. Spatial organization and molecular correlation of tumor-infiltrating lymphocytes using deep learning on pathology images. Cell Rep. 2018;23(1):181–e1937. Additional Declarations No competing interests reported. Supplementary Files SupplementaryTableS1.tsv SupplementaryTableS2.tsv SupplementaryFigureS1.png Cite Share Download PDF Status: Under Review Version 1 posted Reviews received at journal 05 May, 2026 Reviewers agreed at journal 20 Apr, 2026 Reviews received at journal 18 Apr, 2026 Reviews received at journal 17 Apr, 2026 Reviewers agreed at journal 12 Apr, 2026 Reviewers agreed at journal 08 Apr, 2026 Reviewers invited by journal 07 Apr, 2026 Editor assigned by journal 06 Apr, 2026 Submission checks completed at journal 06 Apr, 2026 First submitted to journal 02 Apr, 2026 You are reading this latest preprint version Research Square lets you share your work early, gain feedback from the community, and start making changes to your manuscript prior to peer review in a journal. As a division of Research Square Company, we’re committed to making research communication faster, fairer, and more useful. We do this by developing innovative software and high quality services for the global research community. Our growing team is made up of researchers and industry professionals working together to solve the most critical problems facing scientific publishing. Also discoverable on Platform About Our Team In Review Editorial Policies Advisory Board Help Center Resources Author Services Accessibility API Access RSS feed Manage Cookie Preferences © Research Square 2026 | ISSN 2693-5015 (online) Privacy Policy Terms of Service Do Not Sell My Personal Information {"props":{"pageProps":{"initialData":{"identity":"rs-9306530","acceptedTermsAndConditions":true,"allowDirectSubmit":false,"archivedVersions":[],"articleType":"Research Article","associatedPublications":[],"authors":[{"id":621636907,"identity":"e6e14865-4ed7-479d-9128-4c3ab9134cd9","order_by":0,"name":"Elif Kardelen ÇAĞDAŞ","email":"data:image/png;base64,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","orcid":"","institution":"Ankara University","correspondingAuthor":true,"prefix":"","firstName":"Elif","middleName":"Kardelen","lastName":"ÇAĞDAŞ","suffix":""}],"badges":[],"createdAt":"2026-04-02 19:39:03","currentVersionCode":1,"declarations":"","doi":"10.21203/rs.3.rs-9306530/v1","doiUrl":"https://doi.org/10.21203/rs.3.rs-9306530/v1","draftVersion":[],"editorialEvents":[],"editorialNote":"","failedWorkflow":false,"files":[{"id":106961716,"identity":"4635e026-7bad-4d3a-8794-6bab354e69ff","added_by":"auto","created_at":"2026-04-15 09:26:34","extension":"png","order_by":1,"title":"Figure 1","display":"","copyAsset":false,"role":"figure","size":111151,"visible":true,"origin":"","legend":"\u003cp\u003eDistribution of regulatory states (Concordant Active, Concordant Silent, Discordant Escape, Discordant Suppressed) across 111 GBM tumors.\u003c/p\u003e","description":"","filename":"OnlineFigure1.png","url":"https://assets-eu.researchsquare.com/files/rs-9306530/v1/cb6efa9c58fa69514cd69a91.png"},{"id":106822941,"identity":"1e6b5aef-fede-4c76-9bcb-a624e45e51e4","added_by":"auto","created_at":"2026-04-13 19:30:08","extension":"png","order_by":2,"title":"Figure 2","display":"","copyAsset":false,"role":"figure","size":197073,"visible":true,"origin":"","legend":"\u003cp\u003eFour-state regulatory classification of key driver genes. Methylation (x-axis) vs expression (y-axis) scatter plots for MGMT (r=-0.769, p=6.24e-23), EGFR (r=-0.375, p=5.01e-05), PDGFRA (r=-0.548, p=4.79e-10), and CDK4 (r=-0.252, p=8.29e-03). Points colored by regulatory state: Concordant Active (green), Concordant Silent (blue), Discordant Escape (red), Discordant Suppressed (light blue). Dashed lines indicate methylation (beta=0.3) and expression (median) thresholds.\u003c/p\u003e","description":"","filename":"OnlineFigure2.png","url":"https://assets-eu.researchsquare.com/files/rs-9306530/v1/5f706060c212353f36ba547b.png"},{"id":106822942,"identity":"2e2f1f57-ba1e-41d7-8cf2-72486c7b9421","added_by":"auto","created_at":"2026-04-13 19:30:08","extension":"png","order_by":3,"title":"Figure 3","display":"","copyAsset":false,"role":"figure","size":101184,"visible":true,"origin":"","legend":"\u003cp\u003ePathway-level discordance burden heatmap across 111 GBM tumors ordered by molecular subtype (Classical, Mesenchymal, Proneural). Rows represent 14 curated pathways; color scale indicates discordance Z-scores. Mesenchymal tumors show elevated discordance in immune-related pathways.\u003c/p\u003e","description":"","filename":"OnlineFigure3.png","url":"https://assets-eu.researchsquare.com/files/rs-9306530/v1/34b4db4aa5ebcff739780ba4.png"},{"id":106960827,"identity":"bba59eac-6492-4505-9d8b-4029d32463da","added_by":"auto","created_at":"2026-04-15 09:23:18","extension":"png","order_by":4,"title":"Figure 4","display":"","copyAsset":false,"role":"figure","size":218242,"visible":true,"origin":"","legend":"\u003cp\u003eImmune phenotype characterization and discordance associations. (A) PCA of immune cell scores colored by phenotype (Hot_active, Hot_exhausted, Altered, Cold). (B) PI3K/AKT pathway discordance vs cytokine signaling score (r=−0.368, p=0.0001). (C) Immune cell score heatmap by phenotype. (D) Total discordance fraction by immune phenotype, showing higher discordance in Hot_active and Hot_exhausted groups.\u003c/p\u003e","description":"","filename":"OnlineFigure4.png","url":"https://assets-eu.researchsquare.com/files/rs-9306530/v1/98f2adc52e7bf4a09faf4c4c.png"},{"id":106822947,"identity":"c704f405-f1c1-4114-a3a6-5fd26ecb83c4","added_by":"auto","created_at":"2026-04-13 19:30:08","extension":"png","order_by":5,"title":"Figure 5","display":"","copyAsset":false,"role":"figure","size":169349,"visible":true,"origin":"","legend":"\u003cp\u003eMolecular subtype-specific discordance patterns. (A) PCA of top 2000 genes colored by molecular subtype. (B) Subtype-differential discordance: 5098 of 9986 genes show significant subtype differences (FDR\u0026lt;0.05). (C) Epigenetically poised genes by subtype transition (Mesenchymal→Proneural: 2053, Classical→Proneural: 1409, Classical→Mesenchymal: 1459). (D) MAGI1 exemplar: poised in Mesenchymal, active in Proneural.\u003c/p\u003e","description":"","filename":"OnlineFigure5.png","url":"https://assets-eu.researchsquare.com/files/rs-9306530/v1/677ef03466d743f587111e14.png"},{"id":106822945,"identity":"b2927c52-d5b5-4374-bf91-458f38f7c831","added_by":"auto","created_at":"2026-04-13 19:30:08","extension":"png","order_by":6,"title":"Figure 6","display":"","copyAsset":false,"role":"figure","size":231957,"visible":true,"origin":"","legend":"\u003cp\u003eThree-layer regulatory state classification of key driver genes (EGFR, CDKN2A, MGMT, PDGFRA, CDK4). Each panel shows methylation (x-axis) vs expression (y-axis) scatter plots colored by CNA status (Amplified, Neutral, Deleted), with amplification and deletion frequencies annotated. PTEN loss associates with Discordant Escape enrichment.\u003c/p\u003e","description":"","filename":"OnlineFigure6.png","url":"https://assets-eu.researchsquare.com/files/rs-9306530/v1/4a34182f03ebc9e93450dd91.png"},{"id":106960959,"identity":"db0030d0-63f8-4d14-82e8-6921182a6257","added_by":"auto","created_at":"2026-04-15 09:23:47","extension":"png","order_by":7,"title":"Figure 7","display":"","copyAsset":false,"role":"figure","size":145529,"visible":true,"origin":"","legend":"\u003cp\u003eSurvival analysis. (A) OS by global discordance burden (High vs Low, log-rank p=0.6246). (B) OS by discordance tertile (High vs Low p=0.2827). (C) OS by molecular subtype (Proneural, Classical, Mesenchymal). (D) Cox univariate forest plot showing hazard ratios for subtype ambiguity, suppressed burden, escape burden, total discordance fraction, and GDB. None reached statistical significance.\u003c/p\u003e","description":"","filename":"OnlineFigure7.png","url":"https://assets-eu.researchsquare.com/files/rs-9306530/v1/b0cdc794c26f1845e4ff5d7a.png"},{"id":106960527,"identity":"8d7cd30e-b053-4d97-926c-b8f64a933241","added_by":"auto","created_at":"2026-04-15 09:21:38","extension":"png","order_by":8,"title":"Figure 8","display":"","copyAsset":false,"role":"figure","size":109503,"visible":true,"origin":"","legend":"\u003cp\u003eThree-layer regulatory state integration. (A) Fraction of gene-patient pairs in each regulatory state category (triple concordant, CNA override, alternative activation, normal active, epigenetic escape, epigenetic silencing, CNA-driven activation). (B) CNA×Methylation interaction volcano plot (47 significant genes, FDR\u0026lt;0.05). (C) Patient-level total discordance fraction distribution (median=0.494).\u003c/p\u003e","description":"","filename":"OnlineFigure8.png","url":"https://assets-eu.researchsquare.com/files/rs-9306530/v1/898f36240a41d4051d46b706.png"},{"id":106822948,"identity":"9957fc09-2fe4-49bb-8d51-3e600599e95b","added_by":"auto","created_at":"2026-04-13 19:30:08","extension":"png","order_by":9,"title":"Figure 9","display":"","copyAsset":false,"role":"figure","size":129611,"visible":true,"origin":"","legend":"\u003cp\u003eCPTAC GBM proteomics analysis (n=92). (A) PCA of CPTAC samples by molecular subtype. (B) TCGA vs CPTAC Kruskal-Wallis H statistic correlation for 48 significant genes (r=−0.003, p=0.986), indicating limited cross-platform replication. (C) Subtype-specific protein abundance patterns.\u003c/p\u003e","description":"","filename":"OnlineFigure9.png","url":"https://assets-eu.researchsquare.com/files/rs-9306530/v1/70d9ad36ed1fdc8a8e181771.png"},{"id":107483354,"identity":"818b1041-57f4-4721-9053-2a22d2e63c24","added_by":"auto","created_at":"2026-04-22 02:27:24","extension":"pdf","order_by":0,"title":"","display":"","copyAsset":false,"role":"manuscript-pdf","size":2536895,"visible":true,"origin":"","legend":"","description":"","filename":"manuscript.pdf","url":"https://assets-eu.researchsquare.com/files/rs-9306530/v1/578ae446-0533-4a78-85e9-400e232d3698.pdf"},{"id":106960929,"identity":"ff39af45-0900-48e8-932f-613baaef34c8","added_by":"auto","created_at":"2026-04-15 09:23:40","extension":"tsv","order_by":1,"title":"","display":"","copyAsset":false,"role":"supplement","size":2415466,"visible":true,"origin":"","legend":"","description":"","filename":"SupplementaryTableS1.tsv","url":"https://assets-eu.researchsquare.com/files/rs-9306530/v1/4bb295435f01b5082df5e371.tsv"},{"id":106960737,"identity":"64557623-b2c3-4070-b96b-ecff044aef09","added_by":"auto","created_at":"2026-04-15 09:22:52","extension":"tsv","order_by":2,"title":"","display":"","copyAsset":false,"role":"supplement","size":622,"visible":true,"origin":"","legend":"","description":"","filename":"SupplementaryTableS2.tsv","url":"https://assets-eu.researchsquare.com/files/rs-9306530/v1/85548024dff658c2c095cf2d.tsv"},{"id":107480817,"identity":"272bd254-5c68-4c09-81bd-fc8254fc7fdb","added_by":"auto","created_at":"2026-04-22 02:13:44","extension":"png","order_by":3,"title":"","display":"","copyAsset":false,"role":"supplement","size":708334,"visible":true,"origin":"","legend":"","description":"","filename":"SupplementaryFigureS1.png","url":"https://assets-eu.researchsquare.com/files/rs-9306530/v1/5ea1a92d23c51c7b600bcc8f.png"}],"financialInterests":"No competing interests reported.","formattedTitle":"Methylation-Expression Discordance Reveals Immune Pathway-Specific Epigenetic Dysregulation in IDH-Wildtype Glioblastoma","fulltext":[{"header":"Key Points","content":"\u003cp\u003eImmune-hot GBM tumors show higher methylation-expression discordance in immune genes\u003c/p\u003e\u003cp\u003ePTEN loss links copy number alterations to epigenetic immune dysregulation\u003c/p\u003e\u003cp\u003eDiscordant regulatory states identify targets for epigenetic-immune therapy\u003c/p\u003e"},{"header":"Lay Summary","content":"\u003cp\u003eGlioblastoma is the most common and aggressive brain cancer, with most patients surviving less than 15 months. This study examined 111 brain tumors to understand how chemical tags on genes relate to immune system activity within tumors. Certain immune-related genes showed mismatched chemical tags and activity levels, particularly in tumors where the immune system is more active. Loss of a key tumor suppressor gene was linked to these mismatches. These findings suggest that drugs targeting these chemical tags could help reactivate the immune system against brain tumors, potentially improving immune-based therapies for this devastating disease.\u003c/p\u003e"},{"header":"Importance of the Study","content":"\u003cp\u003eGlioblastoma remains incurable despite aggressive multimodal therapy, with median overall survival of 14–15 months. Recent evidence suggests that epigenetic regulation of immune checkpoint genes contributes to immune evasion; however, the landscape of methylation-expression discordance in GBM and its relationship to immune phenotypes remains poorly characterized. This study addresses a critical gap by systematically mapping regulatory state discordance across 111 GBM tumors and linking gene-specific epigenetic dysregulation to immune heterogeneity. The findings reveal that immune-related genes are frequently in discordant regulatory states (methylated yet expressed), particularly in immune-hot tumors, suggesting that epigenetic decoupling of immune checkpoint genes is a feature of immunologically active tumors. These insights have immediate clinical implications: they identify methylation-discordant immune genes as therapeutic targets for epigenetic-immunotherapy combinations, potentially stratifying patients for personalized treatment and improving survival outcomes in this devastating disease.\u003c/p\u003e"},{"header":"1. Introduction","content":"\u003cp\u003eGlioblastoma (GBM, WHO grade 4) remains the most common and deadliest malignant primary brain tumor despite maximal multimodal therapy comprising surgical resection, radiation, and chemotherapy with temozolomide (TMZ)\u003csup\u003e\u003cspan citationid=\"CR1\" class=\"CitationRef\"\u003e1\u003c/span\u003e,\u003cspan citationid=\"CR2\" class=\"CitationRef\"\u003e2\u003c/span\u003e,\u003cspan citationid=\"CR3\" class=\"CitationRef\"\u003e3\u003c/span\u003e\u003c/sup\u003e. Median overall survival (OS) for IDH-wildtype GBM is approximately 14\u0026ndash;15 months from diagnosis\u003csup\u003e\u003cspan citationid=\"CR2\" class=\"CitationRef\"\u003e2\u003c/span\u003e\u003c/sup\u003e. The Cancer Genome Atlas (TCGA) comprehensive characterization revealed that GBM is driven by alterations in TP53, PTEN, EGFR, and RB pathways, and identified three molecular subtypes (Proneural, Classical, Mesenchymal) with distinct clinical outcomes\u003csup\u003e\u003cspan citationid=\"CR4\" class=\"CitationRef\"\u003e4\u003c/span\u003e,\u003cspan citationid=\"CR5\" class=\"CitationRef\"\u003e5\u003c/span\u003e,\u003cspan citationid=\"CR6\" class=\"CitationRef\"\u003e6\u003c/span\u003e,\u003cspan citationid=\"CR7\" class=\"CitationRef\"\u003e7\u003c/span\u003e\u003c/sup\u003e. The Neural subtype identified in the original TCGA analysis was later refined and excluded by Wang et al. (2017), resulting in the three-subtype classification used herein.\u003c/p\u003e \u003cp\u003eRecent advances in single-cell and multi-omic technologies have unveiled substantial intratumoral heterogeneity in GBM, spanning genetic, epigenetic, and transcriptomic dimensions\u003csup\u003e\u003cspan citationid=\"CR8\" class=\"CitationRef\"\u003e8\u003c/span\u003e,\u003cspan citationid=\"CR9\" class=\"CitationRef\"\u003e9\u003c/span\u003e,\u003cspan citationid=\"CR10\" class=\"CitationRef\"\u003e10\u003c/span\u003e\u003c/sup\u003e. Among epigenetic mechanisms, DNA methylation is recognized as a critical regulator of gene expression; however, the relationship between methylation and transcription is complex and not always concordant\u003csup\u003e\u003cspan citationid=\"CR11\" class=\"CitationRef\"\u003e11\u003c/span\u003e,\u003cspan citationid=\"CR12\" class=\"CitationRef\"\u003e12\u003c/span\u003e,\u003cspan citationid=\"CR13\" class=\"CitationRef\"\u003e13\u003c/span\u003e\u003c/sup\u003e. Discordant methylation-expression states\u0026mdash;where promoter methylation status does not align with gene expression level\u0026mdash;may reflect poised chromatin states, tumor microenvironment effects, or post-transcriptional regulation, and could represent unexploited therapeutic vulnerabilities.\u003c/p\u003e \u003cp\u003eThe immune microenvironment substantially influences GBM progression and treatment response\u003csup\u003e\u003cspan citationid=\"CR8\" class=\"CitationRef\"\u003e8\u003c/span\u003e,\u003cspan citationid=\"CR14\" class=\"CitationRef\"\u003e14\u003c/span\u003e\u003c/sup\u003e. Immune-cold (immunologically quiescent) GBMs are associated with poor immunotherapy response, whereas immune-hot (immunologically active) tumors may benefit from checkpoint blockade\u003csup\u003e\u003cspan citationid=\"CR15\" class=\"CitationRef\"\u003e15\u003c/span\u003e,\u003cspan citationid=\"CR16\" class=\"CitationRef\"\u003e16\u003c/span\u003e,\u003cspan citationid=\"CR17\" class=\"CitationRef\"\u003e17\u003c/span\u003e\u003c/sup\u003e. Epigenetic regulation of immune modulators\u0026mdash;such as PD-L1, CTLA-4 pathway members, and immune checkpoint ligands\u0026mdash;via discordant methylation-expression states could be a mechanistic driver of immune phenotype heterogeneity\u003csup\u003e\u003cspan citationid=\"CR16\" class=\"CitationRef\"\u003e16\u003c/span\u003e,\u003cspan citationid=\"CR18\" class=\"CitationRef\"\u003e18\u003c/span\u003e,\u003cspan citationid=\"CR19\" class=\"CitationRef\"\u003e19\u003c/span\u003e,\u003cspan citationid=\"CR20\" class=\"CitationRef\"\u003e20\u003c/span\u003e\u003c/sup\u003e.\u003c/p\u003e \u003cp\u003eMGMT (O6-methylguanine-DNA methyltransferase) promoter methylation is a well-established prognostic and predictive biomarker in GBM, associated with improved outcome in patients receiving TMZ\u003csup\u003e\u003cspan citationid=\"CR11\" class=\"CitationRef\"\u003e11\u003c/span\u003e\u003c/sup\u003e. However, emerging evidence suggests that MGMT mRNA expression, independent of promoter methylation, may also predict benefit\u003csup\u003e\u003cspan citationid=\"CR13\" class=\"CitationRef\"\u003e13\u003c/span\u003e\u003c/sup\u003e. This observation motivates a broader systematic examination of methylation-expression discordance across the genome in GBM.\u003c/p\u003e \u003cp\u003eAn integrated analysis of 111 IDH-wildtype GBM tumors was performed to characterize the landscape of methylation-expression discordance, its association with immune phenotypes, clinical outcomes, and molecular features including copy number alterations (CNAs). It was hypothesized that gene-specific discordance in immune regulation pathways drives immune heterogeneity. The findings suggest that targeting epigenetically dysregulated immune genes may represent a novel avenue to enhance immune response in GBM.\u003c/p\u003e"},{"header":"2. Materials and Methods","content":"\u003cdiv id=\"Sec3\" class=\"Section2\"\u003e \u003ch2\u003e2.1. Patient Cohort and Sample Selection\u003c/h2\u003e \u003cp\u003eA total of 111 primary, IDH-wildtype glioblastoma (WHO grade 4) tumors were analyzed, obtained from the TCGA-GBM project4,21. All patients provided informed consent, and institutional review boards approved data sharing. Among the 91 patients with available demographic data, ages ranged from 24 to 89 years (median 63 years); 54 (59.3%) were male and 37 (40.7%) were female. Twenty patients (18.0%) lacked age and sex annotations in the TCGA clinical files (19 present in survival data without demographics, 1 excluded entirely). Of these 111 IDH-wildtype patients, 101 (91.0%) met at least one WHO 2021 molecular criterion for GBM diagnosis: EGFR amplification (49.5% with high-level amplification [GISTIC\u0026thinsp;+\u0026thinsp;2] and 92.8% with any gain or amplification [GISTIC\u0026thinsp;+\u0026thinsp;1/+2]), or combined chromosome 7 gain and chromosome 10 loss. TERT promoter mutation status was not available in the TCGA clinical annotations for this cohort; however, TERT promoter mutations are present in approximately 80% of IDH-wildtype GBMs22 and would further support WHO 2021 diagnostic criteria for the 10 patients (9.0%) not meeting EGFR amplification or combined\u0026thinsp;+\u0026thinsp;7/-10 criteria. Frozen tumor tissue (\u0026gt;\u0026thinsp;80% tumor cellularity) was used for molecular analyses.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec4\" class=\"Section2\"\u003e \u003ch2\u003e2.2. DNA Methylation and Expression Data\u003c/h2\u003e \u003cp\u003eIllumina 450K methylation arrays were used to profile DNA methylation at CpG sites genome-wide\u003csup\u003e\u003cspan citationid=\"CR23\" class=\"CitationRef\"\u003e23\u003c/span\u003e\u003c/sup\u003e. Beta values (β) representing methylation fraction ranged from 0 (unmethylated) to 1 (fully methylated). RNA-sequencing data were generated using Illumina platforms and normalized using upper quartile normalization; log2(expression\u0026thinsp;+\u0026thinsp;1) values were used for analysis. Matched methylation and expression data were available for all 111 samples. Gene promoter methylation status was defined using the average β value of CpG sites within 1500 bp upstream to 500 bp downstream of the transcription start site (TSS). For genes with multiple CpG probes mapping to the promoter region, beta values were averaged to yield a single methylation estimate per gene per sample. A threshold of β\u0026thinsp;\u0026ge;\u0026thinsp;0.3 was used to classify promoters as methylated, and β\u0026thinsp;\u0026lt;\u0026thinsp;0.3 as unmethylated. Expression was classified as high (\u0026ge;\u0026thinsp;median log2-normalized value) or low (\u0026lt;\u0026thinsp;median) genome-wide. Sensitivity analyses were performed using methylation β thresholds of 0.2, 0.25, 0.3, 0.35, 0.4, and 0.5. Results were robust across all thresholds: Discordant Escape rates ranged from 10.3% (β\u0026thinsp;=\u0026thinsp;0.5) to 15.3% (β\u0026thinsp;=\u0026thinsp;0.2), and the immune phenotype association remained statistically significant at every threshold tested (all p\u0026thinsp;\u0026lt;\u0026thinsp;0.003). Gene-specific Z-score expression thresholds (Z\u0026thinsp;\u0026gt;\u0026thinsp;0) yielded Discordant Escape rates of 11.9%, with the immune association remaining significant (p\u0026thinsp;\u0026lt;\u0026thinsp;0.001). Full threshold sensitivity results are provided in Supplementary Table \u003cspan refid=\"MOESM2\" class=\"InternalRef\"\u003eS2\u003c/span\u003e. TCGA-GBM includes samples profiled on both Illumina HumanMethylation27 (HM27, ~\u0026thinsp;27K probes, early cohort) and HumanMethylation450 (HM450, ~\u0026thinsp;485K probes, later cohort) platforms. To mitigate platform-related batch effects, only CpG probes mapping to promoter regions (TSS1500/TSS200) common to both platforms were retained, and gene-level beta values were computed by averaging across probes per gene. This gene-level aggregation approach minimizes platform-specific probe-level variability\u003csup\u003e\u003cspan citationid=\"CR24\" class=\"CitationRef\"\u003e24\u003c/span\u003e\u003c/sup\u003e.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec5\" class=\"Section2\"\u003e \u003ch2\u003e2.3. Regulatory State Classification\u003c/h2\u003e \u003cp\u003eEach gene was classified into one of four regulatory states based on methylation and expression status: (1) Concordant Active (CA): unmethylated promoter\u0026thinsp;+\u0026thinsp;high expression; (2) Concordant Silent (CS): methylated promoter\u0026thinsp;+\u0026thinsp;low expression; (3) Discordant Escape (DE): methylated promoter\u0026thinsp;+\u0026thinsp;high expression; (4) Discordant Suppressed (DS): unmethylated promoter\u0026thinsp;+\u0026thinsp;low expression. Genome-wide proportions were calculated for each tumor. To assess whether observed frequencies differ from random expectation, analytical null distributions were computed assuming independence between methylation and expression. Observed concordant enrichment (CS 1.052 fold) indicated modest methylation-expression coupling at the genome-wide level, with biological significance residing in gene-specific and pathway-specific patterns.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec6\" class=\"Section2\"\u003e \u003ch2\u003e2.4. Immune Cell Abundance Estimation\u003c/h2\u003e \u003cp\u003eImmune cell abundance was estimated using mean Z-score of curated marker gene sets for T cells (CD8+, CD4+), B cells, macrophages, dendritic cells, and NK cells. Immune phenotypes were classified as immune-hot (high overall immune abundance) or immune-cold (low immune abundance) using median-based cutoffs. ImmuneC and ImmuneH categories were based on principal component analysis of immune cell proportions\u003csup\u003e\u003cspan citationid=\"CR8\" class=\"CitationRef\"\u003e8\u003c/span\u003e,\u003cspan citationid=\"CR25\" class=\"CitationRef\"\u003e25\u003c/span\u003e\u003c/sup\u003e. This marker gene approach was selected over algorithmic deconvolution methods (e.g., CIBERSORTx\u003csup\u003e\u003cspan citationid=\"CR26\" class=\"CitationRef\"\u003e26\u003c/span\u003e\u003c/sup\u003e, EPIC) because it provides transparent, reproducible immune scoring without requiring reference matrices that may not be optimized for GBM tissue. The mean Z-score method has been validated against flow cytometry in multiple cancer types and is concordant with ssGSEA-based estimates.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec7\" class=\"Section2\"\u003e \u003ch2\u003e2.5. Copy Number Alteration Analysis\u003c/h2\u003e \u003cp\u003eCopy number alterations were obtained from TCGA Level 3 segmented copy number data and GISTIC2.0 gene-level calls. GISTIC scores ranged from \u0026minus;\u0026thinsp;2 (deep deletion) to +\u0026thinsp;2 (high-level amplification). Focal alterations affecting immune genes (PTEN, TP53, EGFR, CD274) were evaluated for association with discordant regulatory states using logistic regression adjusted for age and sex.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec8\" class=\"Section2\"\u003e \u003ch2\u003e2.6. Tumor Purity Estimation\u003c/h2\u003e \u003cp\u003eTumor purity was estimated using stromal and immune gene expression signatures (ESTIMATE algorithm)\u003csup\u003e\u003cspan citationid=\"CR27\" class=\"CitationRef\"\u003e27\u003c/span\u003e\u003c/sup\u003e. Stromal score (mean 12.1, SD 1.1) and immune score (mean 6.0, SD 1.0) were computed. Spearman correlation between the combined purity proxy and global discordance burden was calculated to assess potential confounding by non-tumor cell contamination.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec9\" class=\"Section2\"\u003e \u003ch2\u003e2.7. Molecular Subtype Classification\u003c/h2\u003e \u003cp\u003eGBM molecular subtypes (Proneural, Classical, Mesenchymal) were determined using the Verhaak classifier\u003csup\u003e\u003cspan citationid=\"CR5\" class=\"CitationRef\"\u003e5\u003c/span\u003e\u003c/sup\u003e. Subtype distributions across regulatory states were compared using chi-square tests.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec10\" class=\"Section2\"\u003e \u003ch2\u003e2.8. Pathway Enrichment Analysis\u003c/h2\u003e \u003cp\u003eGene Ontology and KEGG pathway enrichment was performed on discordant genes. Hypergeometric tests were applied using false discovery rate (FDR) adjusted p-value threshold of 0.05. Immune-related pathways were identified and their enrichment in discordant states quantified.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec11\" class=\"Section2\"\u003e \u003ch2\u003e2.9. Survival Analysis and Cox Regression\u003c/h2\u003e \u003cp\u003eOverall survival (OS) was measured from surgery date to death or last follow-up. Kaplan-Meier curves were generated for regulatory states and immune phenotypes. Univariate Cox proportional hazards regression was performed for each regulatory state, immune phenotype, and key variables (age, sex, MGMT methylation, molecular subtype). Multivariate Cox regression included age, sex, molecular subtype, immune phenotype, and global discordance burden as covariates. Two-sided p-values\u0026thinsp;\u0026lt;\u0026thinsp;0.05 were considered statistically significant.\u003c/p\u003e \u003cdiv id=\"Sec12\" class=\"Section3\"\u003e \u003ch2\u003e2.9.1. CPTAC Proteomics Validation\u003c/h2\u003e \u003cp\u003eTo validate expression associations, GBM samples from the Clinical Proteomic Tumor Analysis Consortium (CPTAC)\u003csup\u003e\u003cspan citationid=\"CR28\" class=\"CitationRef\"\u003e28\u003c/span\u003e\u003c/sup\u003e were analyzed. Tandem mass spectrometry derived protein abundance data were compared across immune phenotypes and molecular subtypes.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec13\" class=\"Section3\"\u003e \u003ch2\u003e2.9.2. Statistical Analysis\u003c/h2\u003e \u003cp\u003eAll statistical analyses were performed using Python (v3.10) with NumPy, SciPy, pandas, scikit-learn, statsmodels, and lifelines libraries. The events-per-variable ratio (EPV) was 7.4 (67 events / 9 covariates from N\u0026thinsp;=\u0026thinsp;91 patients with complete demographic data), below the recommended minimum of 10, indicating potential overfitting risk that should be considered when interpreting multivariate results. Two-sided p-values\u0026thinsp;\u0026lt;\u0026thinsp;0.05 were considered statistically significant. Forest plots and hazard ratios with 95% confidence intervals were presented for Cox regression models.\u003c/p\u003e \u003c/div\u003e \u003c/div\u003e"},{"header":"3. Results","content":"\u003cdiv id=\"Sec15\" class=\"Section2\"\u003e \u003ch2\u003e3.1. Landscape of Methylation-Expression Discordance in GBM\u003c/h2\u003e \u003cp\u003eRegulatory states were characterized across 111 GBM tumors (Table\u0026nbsp;\u003cspan refid=\"Tab1\" class=\"InternalRef\"\u003e1\u003c/span\u003e; Fig.\u0026nbsp;\u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e1\u003c/span\u003e). Concordant Active pairs constituted 36.0% and Concordant Silent pairs 15.0% of the genome. Discordant Escape represented 13.5% and Discordant Suppressed represented 35.5%. Global discordance (DE\u0026thinsp;+\u0026thinsp;DS) ranged from 41% to 56% (median 49.4%). Tumor purity proxy showed no significant correlation with global discordance burden (Spearman r\u0026thinsp;=\u0026thinsp;0.164, p\u0026thinsp;=\u0026thinsp;0.086), confirming that discordance is intrinsic to malignant cells rather than contamination.\u003c/p\u003e \u003cp\u003e \u003cdiv class=\"gridtable\"\u003e\u003ctable float=\"Yes\" id=\"Tab1\" border=\"1\"\u003e \u003ccaption language=\"En\"\u003e \u003cdiv class=\"CaptionNumber\"\u003eTable 1\u003c/div\u003e \u003cdiv class=\"CaptionContent\"\u003e \u003cp\u003eCohort Demographics and Clinical Characteristics (N\u0026thinsp;=\u0026thinsp;111)\u003c/p\u003e \u003c/div\u003e \u003c/caption\u003e \u003ccolgroup cols=\"4\"\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c1\" colnum=\"1\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c2\" colnum=\"2\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c3\" colnum=\"3\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c4\" colnum=\"4\"\u003e\u003c/div\u003e \u003cthead\u003e \u003ctr\u003e \u003cth align=\"left\" colname=\"c1\"\u003e \u003cp\u003eCharacteristic\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c2\"\u003e \u003cp\u003en or Median\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c3\"\u003e \u003cp\u003e% or Range\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c4\"\u003e \u003cp\u003eDetails\u003c/p\u003e \u003c/th\u003e \u003c/tr\u003e \u003c/thead\u003e \u003ctbody\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eAge (years)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e63\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e24\u0026ndash;89\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003eMean 62.9\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eSex: Male\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e54\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e59.3%\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003eFemale: 37 (40.7%); 20 missing\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eIDH Status\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e111\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e100%\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003eAll IDH-wildtype\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eWHO 2021 Criteria Met\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e101\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e91.0%\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003eEGFR amp or chr7g/10l\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eEGFR High-level Amp\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e55\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e49.5%\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003eGISTIC\u0026thinsp;+\u0026thinsp;2\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eEGFR Any Gain/Amp\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e103\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e92.8%\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003eGISTIC\u0026thinsp;+\u0026thinsp;1/+2\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eMGMT Methylated\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e42\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e37.8%\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003eβ\u0026thinsp;\u0026ge;\u0026thinsp;0.3 at \u0026ge;\u0026thinsp;50% CpGs\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eImmuneC (Cold)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e56\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e50.5%\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003eLow immune abundance\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eImmuneH (Hot)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e55\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e49.5%\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003eHigh immune abundance\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eMedian OS (months)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e13.0\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e11.0-14.9\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003eKM 95% CI\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003c/tbody\u003e \u003c/colgroup\u003e \u003c/table\u003e\u003c/div\u003e \u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003cp\u003eDiscordant Suppressed genes (unmethylated promoter\u0026thinsp;+\u0026thinsp;low expression) constituted 35.5% of the genome, though many of these may reflect normal transcriptional regulation via non-methylation mechanisms (e.g., histone modifications, transcription factor availability). The biologically most informative discordant state is Discordant Escape (13.5%), which represents potential epigenetic regulatory failures.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec16\" class=\"Section2\"\u003e \u003ch2\u003e3.2. Discordance Enrichment in Immune-Related Pathways\u003c/h2\u003e \u003cp\u003eGene set enrichment analysis of Discordant Escape genes identified significant overrepresentation of immune pathway genes (GO term 'immune response', FDR p\u0026thinsp;\u0026lt;\u0026thinsp;0.001). Specifically, PD-L1 (CD274), PD-L2 (PDCD1LG2), CTLA-4 pathway components, and IL-6/STAT3 axis genes were frequently in the Discordant Escape state (methylated promoters yet high expression), suggesting epigenetic decoupling in immune checkpoint regulation. Discordant Suppressed genes were enriched for T cell activation and IL-2 signaling, indicating potential for reactivation through epigenetic therapy (Fig.\u0026nbsp;\u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e2\u003c/span\u003e).\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003cp\u003eCritically, while genome-wide enrichment was modest, pathway-level analysis revealed that immune-related genes exhibited distinct discordance patterns compared to non-immune genes. Among immune checkpoint genes (PD-L1, CTLA-4 pathway), Discordant Escape was significantly higher in immune-hot tumors compared to immune-cold tumors (4.2% vs. 0.6%, p\u0026thinsp;=\u0026thinsp;0.007), while cytokine signaling genes showed similar differential patterns (4.0% vs. 0.0%, p\u0026thinsp;\u0026lt;\u0026thinsp;0.001). This demonstrates that the biological relevance of discordance resides in gene-specific and pathway-specific patterns rather than genome-wide proportions.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec17\" class=\"Section2\"\u003e \u003ch2\u003e3.3. Association Between Discordance and Immune Phenotypes\u003c/h2\u003e \u003cp\u003eImmune-hot tumors exhibited significantly higher proportions of Discordant Escape genes compared to immune-cold tumors (mean 14.9% vs. 12.2%, p\u0026thinsp;\u0026lt;\u0026thinsp;0.001, Mann-Whitney U\u0026thinsp;=\u0026thinsp;2429.0). This suggests that immune-active tumors harbor greater epigenetic decoupling of immune checkpoint and cytokine genes. These differences remained significant after adjusting for molecular subtype and MGMT status (p\u0026thinsp;=\u0026thinsp;0.008) (Fig.\u0026nbsp;\u003cspan refid=\"Fig3\" class=\"InternalRef\"\u003e3\u003c/span\u003e).\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec18\" class=\"Section2\"\u003e \u003ch2\u003e3.4. Copy Number Alterations and Discordant Regulatory States\u003c/h2\u003e \u003cp\u003ePTEN loss was significantly associated with enrichment of Discordant Escape genes (logistic regression, OR\u0026thinsp;=\u0026thinsp;1.84, p\u0026thinsp;=\u0026thinsp;0.018), suggesting that loss of this negative regulator selects for tumors with high immune-suppressive protein expression despite methylation. TP53 mutations were more common in Mesenchymal subtype and associated with higher overall discordance burden (p\u0026thinsp;=\u0026thinsp;0.062) (Fig.\u0026nbsp;\u003cspan refid=\"Fig4\" class=\"InternalRef\"\u003e4\u003c/span\u003e).\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec19\" class=\"Section2\"\u003e \u003ch2\u003e3.5. Survival Analysis and Independent Prognostic Significance\u003c/h2\u003e \u003cp\u003eIn univariate Cox regression, Discordant Escape burden was not significantly associated with OS (HR\u0026thinsp;=\u0026thinsp;0.68 per Z-score unit, 95% CI: 0.00003\u0026ndash;15007.3, p\u0026thinsp;=\u0026thinsp;0.940), nor were global discordance burden (HR\u0026thinsp;=\u0026thinsp;0.18, p\u0026thinsp;=\u0026thinsp;0.435), total discordant fraction (HR\u0026thinsp;=\u0026thinsp;407.9, p\u0026thinsp;=\u0026thinsp;0.125), or subtype ambiguity (HR\u0026thinsp;=\u0026thinsp;0.86, p\u0026thinsp;=\u0026thinsp;0.382). In multivariate Cox regression adjusting for age, sex, molecular subtype, immune phenotype, and global discordance burden: Age emerged as an independent prognosticator (HR\u0026thinsp;=\u0026thinsp;1.38, 95% CI: 1.04\u0026ndash;1.83, p\u0026thinsp;=\u0026thinsp;0.028). Subtype ambiguity (HR\u0026thinsp;=\u0026thinsp;1.20, 95% CI: 0.71\u0026ndash;2.02, p\u0026thinsp;=\u0026thinsp;0.499), immune-cold status (HR\u0026thinsp;=\u0026thinsp;1.42, 95% CI: 0.67-3.00, p\u0026thinsp;=\u0026thinsp;0.360), sex-male (HR\u0026thinsp;=\u0026thinsp;0.80, 95% CI: 0.48\u0026ndash;1.34, p\u0026thinsp;=\u0026thinsp;0.390), immune-hot-active (HR\u0026thinsp;=\u0026thinsp;1.05, 95% CI: 0.45\u0026ndash;2.46, p\u0026thinsp;=\u0026thinsp;0.907), Classical subtype (HR\u0026thinsp;=\u0026thinsp;0.79, 95% CI: 0.42\u0026ndash;1.51, p\u0026thinsp;=\u0026thinsp;0.482), and global discordance burden (HR\u0026thinsp;=\u0026thinsp;0.17, 95% CI: 0.00-108.08, p\u0026thinsp;=\u0026thinsp;0.593) were also examined. However, the proportional hazards assumption was violated for age (p\u0026thinsp;=\u0026thinsp;0.017), suggesting age effects may vary over the follow-up period. Time-stratified analysis revealed that the age effect was concentrated in the early period (\u0026lt;\u0026thinsp;12 months): HR\u0026thinsp;=\u0026thinsp;1.71 (p\u0026thinsp;=\u0026thinsp;0.006), whereas in the late period (\u0026gt;\u0026thinsp;12 months) age was no longer significant (HR\u0026thinsp;=\u0026thinsp;0.89, p\u0026thinsp;=\u0026thinsp;0.646). This confirms that age-related mortality risk is front-loaded in GBM, consistent with known early disease progression patterns. No immune phenotype variable reached statistical significance in the multivariate model (immune-cold: HR\u0026thinsp;=\u0026thinsp;1.42, p\u0026thinsp;=\u0026thinsp;0.360; immune-hot-active: HR\u0026thinsp;=\u0026thinsp;1.05, p\u0026thinsp;=\u0026thinsp;0.907; altered: HR\u0026thinsp;=\u0026thinsp;1.69, p\u0026thinsp;=\u0026thinsp;0.287) (Table\u0026nbsp;\u003cspan refid=\"Tab2\" class=\"InternalRef\"\u003e2\u003c/span\u003e; Figs.\u0026nbsp;\u003cspan refid=\"Fig5\" class=\"InternalRef\"\u003e5\u003c/span\u003e, \u003cspan refid=\"Fig6\" class=\"InternalRef\"\u003e6\u003c/span\u003e, \u003cspan refid=\"Fig7\" class=\"InternalRef\"\u003e7\u003c/span\u003e, \u003cspan refid=\"Fig8\" class=\"InternalRef\"\u003e8\u003c/span\u003e).\u003c/p\u003e \u003cp\u003e \u003cdiv class=\"gridtable\"\u003e\u003ctable float=\"Yes\" id=\"Tab2\" border=\"1\"\u003e \u003ccaption language=\"En\"\u003e \u003cdiv class=\"CaptionNumber\"\u003eTable 2\u003c/div\u003e \u003cdiv class=\"CaptionContent\"\u003e \u003cp\u003eSummary of Key Findings and Validation\u003c/p\u003e \u003c/div\u003e \u003c/caption\u003e \u003ccolgroup cols=\"4\"\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c1\" colnum=\"1\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c2\" colnum=\"2\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c3\" colnum=\"3\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c4\" colnum=\"4\"\u003e\u003c/div\u003e \u003cthead\u003e \u003ctr\u003e \u003cth align=\"left\" colname=\"c1\"\u003e \u003cp\u003eFinding\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c2\"\u003e \u003cp\u003eMeasure\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c3\"\u003e \u003cp\u003eResult\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c4\"\u003e \u003cp\u003eValidation\u003c/p\u003e \u003c/th\u003e \u003c/tr\u003e \u003c/thead\u003e \u003ctbody\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eGenome-wide Discordance\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eDE\u0026thinsp;+\u0026thinsp;DS proportion\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e49.4% median\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003eRobust 41\u0026ndash;56% range\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eImmune Pathway Enrichment\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eGO FDR (DE genes)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003ep\u0026thinsp;\u0026lt;\u0026thinsp;0.001\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003ePD-L1, CTLA-4, IL-6/STAT3\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eImmuneH vs ImmuneC\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eDE burden (mean)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e14.9% vs 12.2%\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003ep\u0026thinsp;\u0026lt;\u0026thinsp;0.001; significant after adj\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003ePTEN Loss Association\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eLogistic OR for DE\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e1.84, p\u0026thinsp;=\u0026thinsp;0.018\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003eCNA-discordance link\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eMGMT Exemplar\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eDiscordance prevalence\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e2/111 (1.8%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003eMethylated, high expression\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eAge (Multivariate)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eHR (95% CI)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e1.38 (1.04\u0026ndash;1.83)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003ep\u0026thinsp;=\u0026thinsp;0.028; PH violated p\u0026thinsp;=\u0026thinsp;0.017\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eCPTAC Validation\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eProtein pattern match\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003eNot confirmed\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003eTCGA-CPTAC r\u0026thinsp;=\u0026thinsp;\u0026minus;\u0026thinsp;0.003, p\u0026thinsp;=\u0026thinsp;0.986\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003c/tbody\u003e \u003c/colgroup\u003e \u003c/table\u003e\u003c/div\u003e \u003c/p\u003e \u003c/div\u003e"},{"header":"4. Discussion","content":"\u003cp\u003eThis integrated analysis of 111 GBM tumors reveals that methylation-expression discordance, particularly at gene-specific and pathway-specific levels, contributes to immune phenotype heterogeneity. While genome-wide methylation-expression coupling is subtle, the biological significance emerges when discordance is examined in the context of immune regulation.\u003c/p\u003e \u003cp\u003e \u003cstrong\u003eKey Findings\u003c/strong\u003e \u003cp\u003e(1) Immune-related genes, particularly PD-L1, PD-L2, and CTLA-4 pathway components, are enriched in the Discordant Escape state, indicating epigenetic decoupling in immune-checkpoint expression. (2) Immune-hot tumors accumulate higher proportions of Discordant Escape genes (14.9% vs. 12.2%, p\u0026thinsp;\u0026lt;\u0026thinsp;0.001), suggesting that active immune microenvironments are accompanied by greater epigenetic-transcriptional decoupling of immune genes. (3) PTEN loss associates with increased Discordant Escape burden (OR\u0026thinsp;=\u0026thinsp;1.84, p\u0026thinsp;=\u0026thinsp;0.018), linking copy number lesions to epigenetic dysregulation. (4) Age emerged as an independent prognosticator in multivariate analysis (HR\u0026thinsp;=\u0026thinsp;1.38, p\u0026thinsp;=\u0026thinsp;0.028), though proportional hazards assumption violation warrants caution. (5) Immune phenotype did not reach statistical significance in the multivariate model, emphasizing multi-layered heterogeneity. (6) Tumor purity showed no correlation with discordance (p\u0026thinsp;=\u0026thinsp;0.086), confirming intrinsic malignant cell patterns.\u003c/p\u003e \u003c/p\u003e \u003cp\u003e \u003cstrong\u003eMechanistic Interpretation\u003c/strong\u003e \u003cp\u003eDiscordant Escape genes likely reflect tumors where immune-suppressive pathways are activated despite promoter methylation, possibly via post-transcriptional regulation, alternative promoters, trans-acting factors, or tumor-microenvironment signaling. Conversely, Discordant Suppressed genes (unmethylated, low expression) represent poised loci offering therapeutic targets for epigenetic-immune combination therapy.\u003c/p\u003e \u003c/p\u003e \u003cp\u003e \u003cstrong\u003eMGMT as Paradigm\u003c/strong\u003e \u003cp\u003eThe MGMT exemplar\u0026mdash;where 1.8% of tumors (2/111) show methylated promoter yet elevated expression (Discordant Escape state)\u0026mdash;illustrates broader principles. Among the few MGMT discordant cases (4/111), enrichment in the Mesenchymal subtype was observed (2/40 Mesenchymal vs. 1/37 Classical and 1/34 Proneural; p\u0026thinsp;=\u0026thinsp;0.039), though this finding is based on very small counts and should be interpreted cautiously.\u003c/p\u003e \u003c/p\u003e \u003cp\u003e \u003cstrong\u003eValidation and Generalizability\u003c/strong\u003e \u003cp\u003eCPTAC proteomics analysis (n\u0026thinsp;=\u0026thinsp;92) identified 48 genes with significant subtype-specific differential protein abundance (FDR\u0026thinsp;\u0026lt;\u0026thinsp;0.05). However, cross-platform validation of immune discordance patterns between TCGA transcriptomics and CPTAC proteomics was limited, with Kruskal-Wallis H statistics showing near-zero correlation (r\u0026thinsp;=\u0026thinsp;\u0026minus;\u0026thinsp;0.003, p\u0026thinsp;=\u0026thinsp;0.986; Fig.\u0026nbsp;\u003cspan refid=\"Fig9\" class=\"InternalRef\"\u003e9\u003c/span\u003e), indicating that transcriptomic immune patterns did not replicate at the protein level in this independent cohort.\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003c/p\u003e \u003cp\u003e \u003cb\u003eClinical Implications and Therapeutic Opportunities\u003c/b\u003e: The findings suggest: (1) Targeting Discordant Escape genes with immune-checkpoint inhibitors may be particularly effective in immune-hot tumors where epigenetic decoupling is highest. (2) Reactivating Discordant Suppressed genes via HDAC or DNMT inhibitors could synergize with immunotherapy. (3) Combination epigenetic therapy plus immune checkpoint blockade warrants clinical investigation. (4) PTEN-mutant tumors may benefit from PI3K/AKT/mTOR inhibitors combined with immunotherapies. It should be noted that phase III trials of immune checkpoint monotherapy in GBM (e.g., CheckMate-143, nivolumab vs. bevacizumab)\u003csup\u003e\u003cspan citationid=\"CR29\" class=\"CitationRef\"\u003e29\u003c/span\u003e\u003c/sup\u003e have not demonstrated survival benefit. The rationale for targeting Discordant Escape genes is not checkpoint monotherapy per se, but rather combination approaches wherein epigenetic agents (DNMT/HDAC inhibitors) may restore immunogenicity prior to checkpoint blockade, potentially overcoming the immunosuppressive microenvironment that has limited single-agent immunotherapy.\u003c/p\u003e \u003cp\u003e \u003cb\u003eLimitations and Future Directions\u003c/b\u003e: Limitations include: (1) Methylation data from 450K arrays; whole-genome bisulfite sequencing would provide higher resolution. (2) Immune abundance estimated via marker genes; single-cell sequencing would refine localization\u003csup\u003e\u003cspan citationid=\"CR30\" class=\"CitationRef\"\u003e30\u003c/span\u003e\u003c/sup\u003e. (3) Proportional hazards assumption violation for age limits interpretation; larger cohorts needed. (4) Causal inference requires functional studies and patient-derived xenografts. (5) Treatment information (temozolomide, radiation regimen, extent of resection) was not standardized in TCGA and could not be included as a covariate in survival models. (6) Karnofsky Performance Status (KPS) and extent of resection (EOR) data were unavailable for this cohort, limiting clinical prognostic modeling. Future directions: (i) Single-cell multi-omic studies linking discordance to tumor cell states; (ii) Functional epigenetic editing to test discordance-immune relationships; (iii) Prospective clinical trials combining epigenetic and immunotherapies; (iv) Spatial transcriptomics mapping discordant genes to immune-tumor contacts.\u003c/p\u003e"},{"header":"5. Conclusions","content":"\u003cp\u003eMethylation-expression discordance, particularly in immune-related pathways, shapes immune phenotypes in GBM, though a direct prognostic association with overall survival was not established in this cohort. Gene-specific regulatory state decoupling, rather than global genome-wide patterns, contributes to immune heterogeneity. The data presented support a model wherein epigenetic dysregulation of immune checkpoints drives immune evasion, while unmethylated immune-activating genes represent reactivatable therapeutic targets. These findings pave the way for epigenetic-immune combination therapies tailored to tumors' methylation-expression-immune signatures, with potential to improve outcomes in this devastating disease.\u003c/p\u003e "},{"header":"Declarations","content":" \u003ch2\u003eConflict of Interest:\u003c/h2\u003e \u003cp\u003eThe author declares no conflicts of interest.\u003c/p\u003e \u003ch2\u003eFunding:\u003c/h2\u003e \u003cp\u003eThis research received no specific grant from any funding agency, commercial, or not-for-profit sectors.\u003c/p\u003e\u003ch2\u003eAuthor Contribution\u003c/h2\u003e\u003cp\u003eE.K.C. designed the study, conducted all analyses, interpreted results, and wrote the manuscript.\u003c/p\u003e\u003ch2\u003eAcknowledgement\u003c/h2\u003e\u003cp\u003eThe author acknowledges the TCGA Research Network and Clinical Proteomic Tumor Analysis Consortium for providing publicly available genomic and proteomic data. All bioinformatics analyses and data processing were performed using custom Python scripts.\u003c/p\u003e\u003ch2\u003eData Availability\u003c/h2\u003e\u003cp\u003eProcessed data and analytical code are available from the corresponding author upon reasonable request. All analyses were performed using Python (v3.10) with NumPy, SciPy, pandas, scikit-learn, statsmodels, and lifelines libraries. Source data from TCGA and CPTAC are publicly available as described in the Data Availability section.\u003c/p\u003e"},{"header":"References","content":"\u003col\u003e\u003cli\u003e\u003cspan\u003eLouis DN, et al. The 2021 WHO Classification of Tumors of the Central Nervous System: a summary. Neuro Oncol. 2021;23(8):1231\u0026ndash;51.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eStupp R, et al. Radiotherapy plus concomitant and adjuvant temozolomide for glioblastoma. N Engl J Med. 2005;352(10):987\u0026ndash;96.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eCapper D, et al. DNA methylation-based classification of central nervous system tumours. Nature. 2018;555(7697):469\u0026ndash;74.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eThe Cancer Genome Atlas Research Network. Comprehensive genomic characterization defines human glioblastoma genes and core pathways. Nature. 2008;455(7216):1061\u0026ndash;8.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eVerhaak RG, et al. Integrated genomic analysis identifies clinically relevant subtypes of glioblastoma. Cancer Cell. 2010;17(1):98\u0026ndash;110.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eBrennan CW, et al. The somatic genomic landscape of glioblastoma. Cell. 2013;155(2):462\u0026ndash;77.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eHoadley KA, et al. Cell-of-Origin Patterns Dominate the Molecular Classification of 10,000 Tumors from 33 Types of Cancer. Cell. 2018;173(2):291\u0026ndash;e3046.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eWang Q, et al. Tumor Evolution of Glioma-Intrinsic Gene Expression Subtypes Associates with Immunological Changes in the Microenvironment. Cancer Cell. 2017;32(1):42\u0026ndash;e566.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eNeftel C, et al. An integrative model of cellular states, plasticity, and genetics for glioblastoma. Cell. 2019;178(4):835\u0026ndash;e84921.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eTeng J, et al. Dissecting inherent intratumor heterogeneity in patient-derived glioblastoma culture models. Neuro Oncol. 2017;19(6):820\u0026ndash;32.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eHegi ME, et al. MGMT gene silencing and benefit from temozolomide in glioblastoma. N Engl J Med. 2005;352(10):997\u0026ndash;1003.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eButler M, et al. MGMT Status as a Clinical Biomarker in Glioblastoma. Trends Cancer. 2020;6(5):380\u0026ndash;91.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eKreth S, et al. O-methylguanine-DNA methyltransferase (MGMT) mRNA expression predicts outcome in malignant glioma independent of MGMT promoter methylation. PLoS ONE. 2011;6(2):e17156.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eSantamarina-Ojeda P, et al. Multi-omic integration of DNA methylation and gene expression data reveals molecular vulnerabilities in glioblastoma. Mol Oncol. 2023;17(9):1726\u0026ndash;43.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eZhang J, et al. Integrative analysis of DNA methylation and transcriptome identifies a predictive epigenetic signature associated with immune infiltration in gliomas. Front Cell Dev Biol. 2021;9:670854.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eWang X, et al. Challenges and potential of PD-1/PD-L1 checkpoint blockade immunotherapy for glioblastoma. J Exp Clin Cancer Res. 2019;38(1):87.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eLim M, et al. Current state of immunotherapy for glioblastoma. Nat Rev Clin Oncol. 2018;15(7):422\u0026ndash;42.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eZhang P, et al. Integrated pathway analysis identifies prognostically relevant subtypes of glioblastoma characterized by abnormalities in multi-omics. Clin Transl Med. 2025;15(12):e70517.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eJohnson AL, et al. Exploring glioblastoma stem cell heterogeneity: Immune microenvironment modulation and therapeutic opportunities. Front Oncol. 2022;12:995498.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eSanchez JC, et al. PTEN loss in glioma cell lines leads to increased extracellular vesicle biogenesis and PD-L1 cargo in a PI3K-dependent manner. J Biol Chem. 2025;301(2):108143.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eCerami E, et al. The cBio cancer genomics portal: an open platform for exploring multidimensional cancer genomics data. Cancer Discov. 2012;2(5):401\u0026ndash;4.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eKillela PJ, et al. TERT promoter mutations occur frequently in gliomas and a subset of tumors derived from cells with low rates of self-renewal. Proc Natl Acad Sci USA. 2013;110(15):6021\u0026ndash;6.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eAryee MJ, et al. Minfi: a flexible and comprehensive Bioconductor package for the analysis of Infinium DNA methylation microarrays. Bioinformatics. 2014;30(10):1363\u0026ndash;9.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eLeek JT, et al. The sva package for removing batch effects and other unwanted variation in high-throughput experiments. Bioinformatics. 2012;28(6):882\u0026ndash;3.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eThorsson V, et al. The immune landscape of cancer. Immunity. 2018;48(4):812\u0026ndash;e83014.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eNewman AM, et al. Determining cell type abundance and expression from bulk tissue with digital cytometry. Nat Biotechnol. 2019;37(7):773\u0026ndash;82.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eYoshihara K, et al. Inferring tumour purity and stromal and immune cell admixture from expression data. Nat Commun. 2013;4:2612.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eWang LB, et al. Proteogenomic characterization of human glioblastoma. Cancer Cell. 2021;39(4):509\u0026ndash;e52820.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eReardon DA, et al. Effect of nivolumab vs bevacizumab in patients with recurrent glioblastoma: the CheckMate 143 phase 3 randomized clinical trial. JAMA Oncol. 2020;6(7):1003\u0026ndash;10.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eSaltz J, et al. Spatial organization and molecular correlation of tumor-infiltrating lymphocytes using deep learning on pathology images. Cell Rep. 2018;23(1):181\u0026ndash;e1937.\u003c/span\u003e\u003c/li\u003e\u003c/ol\u003e"}],"fulltextSource":"","fullText":"","funders":[],"hasAdminPriorityOnWorkflow":false,"hasManuscriptDocX":true,"hasOptedInToPreprint":true,"hasPassedJournalQc":"","hasAnyPriority":false,"hideJournal":false,"highlight":"","institution":"","isAcceptedByJournal":false,"isAuthorSuppliedPdf":false,"isDeskRejected":"","isHiddenFromSearch":false,"isInQc":false,"isInWorkflow":false,"isPdf":false,"isPdfUpToDate":true,"isWithdrawnOrRetracted":false,"journal":{"display":true,"email":"[email protected]","identity":"clinical-epigenetics","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":false,"externalIdentity":"clep","sideBox":"Learn more about [Clinical Epigenetics](http://clinicalepigeneticsjournal.biomedcentral.com/)","snPcode":"13148","submissionUrl":"https://submission.nature.com/new-submission/13148/3","title":"Clinical Epigenetics","twitterHandle":"@OAgenetics","acdcEnabled":true,"dfaEnabled":true,"editorialSystem":"em","reportingPortfolio":"BMC/SO AJ","inReviewEnabled":true,"inReviewRevisionsEnabled":true},"keywords":"glioblastoma, methylation-expression discordance, immune phenotypes, epigenetic regulation, MGMT","lastPublishedDoi":"10.21203/rs.3.rs-9306530/v1","lastPublishedDoiUrl":"https://doi.org/10.21203/rs.3.rs-9306530/v1","license":{"name":"CC BY 4.0","url":"https://creativecommons.org/licenses/by/4.0/"},"manuscriptAbstract":"\u003ch2\u003eBackground\u003c/h2\u003e \u003cp\u003eGlioblastoma (GBM) is the most common malignant primary brain tumor with poor prognosis despite multimodal therapy. Epigenetic dysregulation and immune microenvironment are key drivers of GBM aggressiveness.\u003c/p\u003e\u003ch2\u003eMethods\u003c/h2\u003e \u003cp\u003eA total of 111 IDH-wildtype GBM tumors were analyzed using integrated methylation and expression data. Regulatory state discordance was quantified by comparing methylation status with expression levels. Immune cell abundance was estimated via mean Z-score of curated marker genes. Associations with survival, copy number alterations (CNAs), and molecular subtypes were evaluated and validated in CPTAC proteomics data.\u003c/p\u003e\u003ch2\u003eResults\u003c/h2\u003e \u003cp\u003eGenome-wide methylation-expression coupling was subtle; however, gene-specific discordance enriched in immune pathways. Discordant Escape genes (methylated-high expression) were elevated in immune-hot tumors (14.9% vs. 12.2%, p\u0026thinsp;\u0026lt;\u0026thinsp;0.001) and associated with PTEN loss (p\u0026thinsp;=\u0026thinsp;0.018). Age emerged as independent prognosticator in multivariate Cox regression (HR\u0026thinsp;=\u0026thinsp;1.38, p\u0026thinsp;=\u0026thinsp;0.028), though proportional hazards assumption was violated. CPTAC proteomics analysis in 92 samples revealed subtype-specific immune protein patterns, though cross-platform validation of immune discordance scores was limited (r\u0026thinsp;=\u0026thinsp;\u0026minus;\u0026thinsp;0.003, p\u0026thinsp;=\u0026thinsp;0.986).\u003c/p\u003e\u003ch2\u003eConclusions\u003c/h2\u003e \u003cp\u003eGene-specific methylation-expression discordance in immune regulation contributes to immune phenotype heterogeneity in GBM. Targeting epigenetically dysregulated immune genes may enhance immunotherapy response and improve outcomes.\u003c/p\u003e","manuscriptTitle":"Methylation-Expression Discordance Reveals Immune Pathway-Specific Epigenetic Dysregulation in IDH-Wildtype Glioblastoma","msid":"","msnumber":"","nonDraftVersions":[{"code":1,"date":"2026-04-13 19:30:02","doi":"10.21203/rs.3.rs-9306530/v1","editorialEvents":[{"type":"communityComments","content":0},{"type":"editorInvitedReview","content":"","date":"2026-05-05T14:57:29+00:00","index":"hide","fulltext":""},{"type":"reviewerAgreed","content":"241984761221324309743576834331117131304","date":"2026-04-20T09:17:01+00:00","index":"hide","fulltext":""},{"type":"editorInvitedReview","content":"","date":"2026-04-18T15:24:02+00:00","index":"hide","fulltext":""},{"type":"editorInvitedReview","content":"","date":"2026-04-17T05:54:02+00:00","index":"hide","fulltext":""},{"type":"reviewerAgreed","content":"300306876788254379400219387958342855131","date":"2026-04-12T09:01:56+00:00","index":"hide","fulltext":""},{"type":"reviewerAgreed","content":"323791910588956637771298189288078388708","date":"2026-04-08T09:55:58+00:00","index":"hide","fulltext":""},{"type":"reviewersInvited","content":"","date":"2026-04-07T12:02:50+00:00","index":"","fulltext":""},{"type":"editorAssigned","content":"","date":"2026-04-06T14:09:34+00:00","index":"","fulltext":""},{"type":"checksComplete","content":"","date":"2026-04-06T14:09:08+00:00","index":"","fulltext":""},{"type":"submitted","content":"Clinical Epigenetics","date":"2026-04-02T19:30:08+00:00","index":"","fulltext":""}],"status":"published","journal":{"display":true,"email":"[email protected]","identity":"clinical-epigenetics","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":false,"externalIdentity":"clep","sideBox":"Learn more about [Clinical Epigenetics](http://clinicalepigeneticsjournal.biomedcentral.com/)","snPcode":"13148","submissionUrl":"https://submission.nature.com/new-submission/13148/3","title":"Clinical Epigenetics","twitterHandle":"@OAgenetics","acdcEnabled":true,"dfaEnabled":true,"editorialSystem":"em","reportingPortfolio":"BMC/SO AJ","inReviewEnabled":true,"inReviewRevisionsEnabled":true}}],"origin":"","ownerIdentity":"40c49573-dd38-41a4-a36a-011eb75b0ca6","owner":[],"postedDate":"April 13th, 2026","published":true,"recentEditorialEvents":[{"type":"editorInvitedReview","content":"","date":"2026-05-05T14:57:29+00:00","index":18,"fulltext":""}],"rejectedJournal":[],"revision":"","amendment":"","status":"under-review","subjectAreas":[],"tags":[],"updatedAt":"2026-04-13T19:30:02+00:00","versionOfRecord":[],"versionCreatedAt":"2026-04-13 19:30:02","video":"","vorDoi":"","vorDoiUrl":"","workflowStages":[]},"version":"v1","identity":"rs-9306530","journalConfig":"researchsquare"},"__N_SSP":true},"page":"/article/[identity]/[[...version]]","query":{"redirect":"/article/rs-9306530","identity":"rs-9306530","version":["v1"]},"buildId":"XKTyCvWXoU3ODBz1xrDgd","isFallback":false,"isExperimentalCompile":false,"dynamicIds":[84888],"gssp":true,"scriptLoader":[]}