Multi-Omic Analysis of Glioma Identifies RRM2 as a Prognostic Biomarker Associated With Tumor Stemness and Cell Cycle

Abstract

Introduction

Glioma is the most common primary malignant tumor of the central nervous system and remains associated with poor prognosis and limited effective biomarkers. Ribonucleotide reductase M2 (RRM2), a key enzyme involved in DNA synthesis, is frequently upregulated in multiple cancers; however, its pan-cancer characteristics and clinical significance in glioma remain incompletely defined.

Methods

This retrospective study integrated data from The Cancer Genome Atlas (TCGA), Therapeutically Applicable Research to Generate Effective Treatments (TARGET), and Genotype-Tissue Expression (GTEx) to characterize RRM2 expression patterns and prognostic relevance across 34 solid tumor types. Glioma (GBMLGG) was further analyzed to assess associations between RRM2 expression and Mutant-Allele Tumor Heterogeneity (MATH), stemness indices, clinicopathological features, and survival outcomes. An RRM2-based prognostic nomogram was constructed and externally validated. Differential expression analysis combined with Kyoto Encyclopedia of Genes and Genomes (KEGG) and Gene Set Enrichment Analysis (GSEA) was performed to explore RRM2-associated biological pathways.

Results

RRM2 was significantly upregulated in 33 of 34 cancer types, with the strongest prognostic effect observed in GBMLGG. In glioma, RRM2 expression was markedly higher than in normal brain tissue (P < 0.001) and showed high diagnostic accuracy (AUC = 0.967). Elevated RRM2 expression was associated with advanced age, higher WHO grade, and Isocitrate Dehydrogenase (IDH) wild-type status, and predicted significantly poorer overall survival, disease-specific survival, and progression-free interval (all P < 0.001). A nomogram incorporating RRM2 expression, age, WHO grade, and IDH status demonstrated robust predictive performance for 1-, 3-, and 5-year survival. RRM2 expression was positively correlated with tumor stemness and negatively correlated with intratumoral heterogeneity, and was associated with enrichment of cell cycle and immune-related pathways.

Conclusion

RRM2 is markedly overexpressed in glioma and is strongly associated with tumor aggressiveness and poor prognosis. These findings support the potential value of RRM2 as a prognostic biomarker in glioma and suggest that it is associated with stemness-related and cell cycle-related biological features. Further studies are needed to clarify its mechanistic and therapeutic relevance.

Keywords

glioma RRM2 pan-cancer analysis prognostic biomarker nomogram

1. Introduction

Gliomas comprise approximately 40–50% of primary malignant brain tumors in adults, with glioblastoma (GBM) accounting for nearly 49% of all gliomas.¹ The global incidence of glioma continues to rise. Despite advances in neurosurgical techniques and the integration of radiotherapy and chemotherapy, the prognosis remains poor; even with standard-of-care treatment, the median survival time is approximately 15 months, and the 5-year survival rate for GBM patients is below 10%.² These outcomes highlight the urgent need for more effective biomarkers and therapeutic strategies.

Ribonucleotide reductase M2 (RRM2) is one of the catalytic subunits of ribonucleotide reductase (RNR), an essential enzyme responsible for converting ribonucleotides into deoxyribonucleotides required for DNA synthesis and repair.³ In normal tissues, RRM2 expression is tightly regulated; however, it is frequently upregulated in various cancers.⁴ Previous studies have shown that RRM2 is highly expressed in multiple malignancies, including breast cancer, bladder cancer, and colorectal cancer,^5-7 and that its overexpression is associated with enhanced DNA replication and repair in rapidly proliferating tumor cells. Consistent with these findings, RRM2 overexpression has been reported in a variety of malignancies and has been associated with tumor progression and unfavorable clinical outcomes.^5-12 Accumulating evidence further indicates that RRM2 may exert oncogenic functions in multiple cancer types.^8-12 Moreover, elevated RRM2 expression has been reported to correlate with unfavorable survival outcomes and increased recurrence rates in several cancers, including lung cancer,⁸ hepatocellular carcinoma,⁹ gastric cancer,¹⁰ breast cancer and colorectal cancer,¹¹ as well as pancreatic cancer.¹²

I n this study, we conducted a comprehensive pan-cancer analysis of RRM2, integrating DNA methylation and transcriptomic data to characterize its expression patterns and associations with prognosis, intratumoral heterogeneity, and stemness. We further focused on glioma to evaluate the clinical significance of RRM2 as a potential prognostic biomarker. Specifically, we validated RRM2 expression in tumor tissues, examined its associations with clinicopathological features and survival outcomes, and developed a prognostic nomogram to improve individualized survival prediction. These analyses provide additional insight into the potential clinical relevance of RRM2 in glioma.

2. Materials and Methods

The reporting of this study conforms to the REMARK (Reporting Recommendations for Tumor Marker Prognostic Studies) guidelines.¹³

2.1. Differential Expression Analysis of RRM2

Pan-cancer expression analysis: We downloaded the uniformly processed pan-cancer dataset (TCGA–TARGET–GTEx) from the UCSC Xena database. Data from different sources in UCSC Xena had already been processed through a unified pipeline and standardized by the platform. We directly used the normalized expression matrix provided by UCSC Xena for subsequent analyses. All expression data were transformed using log₂(x + 1). The mRNA expression levels of RRM2 were compared between tumor tissues and their corresponding normal tissues across 34 cancer types. Cancer types with fewer than three samples were excluded from the analysis. The numbers of tumor and normal samples for each cancer type are summarized in Supplementary Table 1. Differences in expression between tumor and normal tissues were calculated using R software (version 3.6.4). Statistical significance was evaluated using the unpaired Wilcoxon rank-sum test or the paired Wilcoxon signed-rank test, as appropriate.

TCGA glioma cohort analysis: For the glioma cohort analysis, RNA-seq data processed by the STAR pipeline and corresponding clinical data for the TCGA-GBM and TCGA-LGG projects were downloaded from the TCGA database (https://portal.gdc.cancer.gov), and FPKM-format expression data were extracted for subsequent analyses. To improve comparability, expression data from the TCGA glioma cohorts were also transformed using log₂(x + 1) before downstream analyses. The TCGA glioma cohorts were analyzed separately from the pan-cancer dataset rather than being directly merged into a single expression matrix.

2.2. Prognostic Analysis of RRM2 Expression

Uniformly normalized pan-cancer expression data were sourced from the TCGA–TARGET–GTEx database, from which RRM2 expression values were extracted for each sample. High-quality survival data for TCGA cohorts were obtained from a previously published pan-cancer prognostic study.¹⁰ Samples with follow-up durations of less than 30 days were excluded from the analysis. All expression data underwent log₂ (x + 1) transformation, and cancer types with fewer than 10 samples were also excluded. Ultimately, expression data, along with corresponding overall survival (OS) and disease-specific survival (DSS) information, were included for 44 cancer types in the analysis.

2.3. Association of RRM2 Expression With Tumor Heterogeneity and Stemness

Uniformly normalized pan-cancer expression data were retrieved from the TCGA–TARGET–GTEx database. Somatic mutation data (level 4 Simple Nucleotide Variation) for all TCGA samples, processed using MuTect2 (DOI: 10.1038/nature08822), were obtained from the Genomic Data Commons (GDC). Tumor mutation burden (TMB) was calculated using the tmb function in the R package maftools (version 2.8.05) and integrated with gene expression data. DNA methylation-based stemness score (DNAss), derived from DNA methylation features, was obtained from a prior study¹¹ and combined with the expression data. All datasets underwent log₂ (x + 1) transformation, and cancer types with fewer than three samples were excluded. Ultimately, data from 37 cancer types were included. Pearson correlation analysis was conducted to assess the associations between RRM2 expression and Mutant-Allele Tumor Heterogeneity (MATH) as well as stemness indices: DNAss, Epigenetically Regulated RNA Expression-based Methylation Stemness Score (EREG-METHss), Differentially Methylated Probes-based Stemness Score (DMPss), and Enhancer-based Stemness Score (ENHss).

2.4. Prognostic Validation and Nomogram Construction

RNA-seq expression data and corresponding clinical information for glioma (GBMLGG) were obtained from the TCGA database (https://portal.gdc.cancer.gov). RRM2 expression levels were analyzed across various clinicopathological subgroups, including sex, age, World Health Organization (WHO) grade, 1p/19q codeletion status, Isocitrate Dehydrogenase (IDH) mutation status, histological subtype, treatment outcome, and survival outcomes (OS, DSS, and Progression-free Interval [PFI]). The relationship between RRM2 expression and survival outcomes was assessed. Time-dependent Receiver Operating Characteristic (ROC) curves were generated to evaluate the predictive performance of RRM2 for 1-, 3-, and 5-year OS, DSS, and PFI, with the area under the curve (AUC) calculated. The CGGA cohort (Chinese Glioma Genome Atlas, https://www.cgga.org.cn/) was used as an independent external dataset to validate the prognostic value of RRM2. Independent prognostic factors identified through univariate and multivariate Cox regression analyses were employed to construct a nomogram model, which was subsequently validated using the CGGA and institutional cohorts.

2.5. Western Blot Analysis

Human glioblastoma tumor tissues and corresponding adjacent non-tumorous tissues were obtained from patients who underwent surgical resection at our center. Written informed consent was obtained from all participants or their legal guardians, allowing the use of their clinical data and tissue specimens for research purposes. All patient data were fully de-identified prior to analysis to ensure that no individual could be identified from the reported information. Total proteins were extracted from these tissues using RIPA lysis buffer (Beyotime, Shanghai, China) supplemented with protease inhibitors (Solarbio, China), and protein concentrations were determined with a BCA assay kit (Solarbio). Equal amounts of protein (30 μg per lane) were separated by 10% SDS–PAGE and subsequently transferred onto PVDF membranes (Millipore, MA, USA). After blocking, the membranes were incubated overnight at 4°C with primary antibodies against RRM2 (1:2000, Bioss) and GAPDH (1:10000, Proteintech). Following a 1-hour incubation with appropriate secondary antibodies, protein bands were visualized using an enhanced chemiluminescence (ECL) detection kit (Biosharp, China).

2.6. Immunohistochemistry

For immunohistochemical analysis, paraffin-embedded glioma tissue specimens were obtained from patients who underwent surgical resection at our center. Written informed consent was obtained from all participants or their legal guardians, and all patient information was anonymized and de-identified prior to analysis. These samples were independent from the fresh postoperative glioma tissues used for Western blot analysis. The specimens were sectioned into 4-μm-thick slices, mounted on slides, and baked. The sections were deparaffinized in xylene and rehydrated through a series of graded ethanol solutions. Antigen retrieval was performed using sodium citrate buffer according to the manufacturer’s instructions. The sections were then incubated overnight at 4 °C with a primary antibody against RRM2 (1:200, Bioss). After incubation with the appropriate secondary antibodies, DAB was used for visualization, and hematoxylin was applied for counterstaining. Images were captured using a light microscope, and quantitative analysis was performed using ImageJ software (National Institutes of Health, Bethesda, MD, USA).

2.7. Functional Enrichment and Mechanistic Analysis

Differentially expressed genes (DEGs) between high and low RRM2 expression groups in TCGA glioma samples were identified based on the criteria |log₂FC| > 1 and P < 0.05. Subsequently, Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway analysis and Gene Set Enrichment Analysis (GSEA) were conducted. A Normalized Enrichment Score (NES) greater than 1 and P < 0.05 were deemed statistically significant. Correlations between RRM2 and intersecting genes were analyzed using the Xiantao Academic online platform (https://www.xiantaozi.com/).

2.8. Statistical Analysis

All statistical analyses were performed using R software (version 3.6.4). Comparisons between two groups were performed using the Wilcoxon rank-sum test or Wilcoxon signed-rank test, as appropriate, whereas comparisons among multiple groups employed the Kruskal–Wallis test. Survival analyses were conducted using the Kaplan–Meier method with log-rank tests, and univariate and multivariate analyses were performed using Cox proportional hazards regression models. A two-sided P < 0.05 was considered statistically significant (Figure 1).

Figure 1.

Study of diagram

3. Results

3.1. Pan-Cancer Analysis of RRM2, Tumor Heterogeneity, and Stemness

To systematically evaluate the expression profile of RRM2 across various cancers, we utilized data from the UCSC Xena platform (TCGA–TARGET–GTEx) to compare RRM2 expression levels between tumor tissues and their corresponding normal tissues across 34 cancer types. RRM2 was significantly upregulated in 33 of the 34 tumor types, including GBM, GBMLGG, low-grade glioma (LGG), UCEC, CESC, and LUAD, with KICH being the only exception (Figure 2A). Pan-cancer survival analyses further indicated that high RRM2 expression was significantly associated with unfavorable DSS in 17 cancer types, such as GBMLGG, KIPAN, LGG, KIRP, ACC, and LUAD (Figure 2B), and with poorer OS in 13 cancer types, including GBMLGG, LGG, KIPAN, ACC, LUAD, and LIHC (Figure 2C). Among these, the prognostic impact of RRM2 was most pronounced and robust in GBMLGG.

Figure 2.

Pan-cancer analyses of RRM2 expression, survival, heterogeneity, and stemness. (A) RRM2 expression in tumor and normal tissues across 34 cancer types (UCSC Xena, TCGA–TARGET–GTEx). (B–C) Survival analyses of RRM2 across cancers using DSS and OS. (D) Relationship between RRM2 expression and the MATH heterogeneity index. (E–H) Relationships between RRM2 expression and stemness indices (DNAss, EREG-METHss, DMPss, ENHss) in GBM and LGG

Given this strong association, we further investigated the relationship between RRM2 expression and tumor biological features. RRM2 expression exhibited a significant negative correlation with the tumor heterogeneity index MATH (Figure 2D) and showed significant positive correlations with multiple tumor stemness indices, including DNAss, EREG-METHss, DMPss, and ENHss (Figure 2E–G). These trends were consistently observed in both GBM and LGG. Collectively, these findings indicate that RRM2 expression is associated with tumor stemness characteristics and intratumoral heterogeneity in glioma.

3.2. Expression, Diagnostic Value, and Prognostic Significance of RRM2 in Glioma

We systematically analyzed the expression characteristics of RRM2 in glioma. Utilizing the UCSC Xena dataset (TCGA–TARGET–GTEx), we found that RRM2 expression was significantly elevated in glioma tissues (n = 697) compared to normal brain tissues (n = 10) (Figure 3A, P < 0.001). ROC analysis indicated a high diagnostic accuracy for RRM2 in differentiating tumor from normal tissue (AUC = 0.967, 95% CI: 0.958–0.976; Figure 3B). Furthermore, associations between RRM2 expression and clinicopathological characteristics revealed significant differences across age groups, WHO grades, 1p/19q codeletion status, and IDH mutation status (Figure 3C, E–G; all P < 0.001). However, no significant difference was noted between sexes (Figure 3D, P = 0.2588). Additionally, RRM2 expression differed significantly among histological subtypes, initial treatment outcomes, and DSS and OS event groups. Specifically, higher RRM2 expression was observed in more aggressive histological subtypes, in patients with unfavorable initial treatment outcomes, and in those who experienced DSS or OS events. (Figure 3H–K; all P < 0.001).

Figure 3.

RRM2 expression landscape and survival analyses in glioma. (A) RRM2 expression in glioma (n = 697) and normal brain tissues (n = 10) from UCSC Xena (TCGA–TARGET–GTEx). (B) ROC curve evaluating the ability of RRM2 to distinguish tumor from normal tissue. (C–G) Distribution of RRM2 expression across clinical and molecular subgroups (age, WHO grade, 1p/19q codeletion, IDH status, and sex).(H–K) RRM2 expression across histological subtypes, initial therapy outcomes, and survival status groups.(L–Q) Kaplan–Meier curves and time-dependent ROC analyses for OS, DSS, and PFI in TCGA, with validation in CGGA cohorts.(U–V) RRM2 expression in paired glioma and adjacent tissues from the institutional cohort.(W–X) Representative immunohistochemistry images and scoring of RRM2 staining in glioma samples

Survival analyses indicated that patients exhibiting high RRM2 expression experienced significantly shorter OS, DSS, and PFI compared to those with low expression (Figure 3L, N, P; all P < 0.001). Time-dependent ROC analyses further illustrated the favorable predictive performance of RRM2 for 1-, 3-, and 5-year OS, DSS, and PFI (Figure 3M, O, Q). These findings were consistently validated in external cohorts CGGA301, CGGA325, and CGGA693, where high RRM2 expression was similarly linked to significantly poorer survival outcomes (Figure 3I–T; all P < 0.001).

Additionally, analysis of 16 paired glioblastoma and adjacent non-tumorous tissues from our institution confirmed a significant elevation of RRM2 expression in tumor tissues (P = 0.0281; Figure 3U–V). Immunohistochemical analysis of 57 glioma samples further revealed marked differences in RRM2 staining intensity among samples (Figure 3W–X), consistent with findings from public datasets. The clinicopathological characteristics of the 57 glioma patients used for immunohistochemistry are summarized in Table 1. Collectively, these results indicate that RRM2 is significantly overexpressed in glioma and is associated with adverse clinicopathological features and poor prognosis.

Table 1.

Clinicopathological Characteristics of Patients Included in the IHC Cohort (n = 57)

Characteristics		Number of cases (n=57)
Age	<60	43(75.44%)
Age	≥60	14(24.56%)
Gender	Male	36(63.16%)
Gender	Female	21(36.84%)
WHO Grade	2 Grade	17(29.82%)
	3 Grade	16(28.07%)
	4 Grade	24(42.11%)
IDH1	Mutant	20(35.09%)
	Wild-type	30(52.63%)
	NA	7(12.28%)
BREAF V600E	Mutant	0(0.00%)
	Wild-type	43(75.44%)
	NA	14(24.56%)
MGMT	Methylated	24(42.10%)
	Unmethylated	11(19.30%)
	NA	22(38.60%)
TERT	Mutant	19(33.33%)
	Wild-type	23(40.35%)
	NA	15(26.32%)

3.3. Construction and Validation of the Prognostic Nomogram

Univariate Cox regression analysis utilizing TCGA data identified age, sex, WHO grade, IDH status, and RRM2 expression as factors significantly associated with overall survival (Table 2). Subsequent multivariate Cox regression analysis revealed that age > 60 years (P = 0.009), high RRM2 expression (P = 0.003), WHO grade III/IV (P < 0.001), and IDH wild-type status (P < 0.001) were independent adverse prognostic factors; however, sex did not achieve statistical significance (P = 0.054).

Table 2.

Univariate and Multivariate Analyses of Age, Gender, WHO Grade, IDH Status, and RRM2 Expression

Characteristics	Total(N)	Univariate analysis		Multivariate analysis
Characteristics	Total(N)	HR(95% CI)	P value	HR(95% CI)	P value
Gender	698
Female	297	Reference		Reference
Male	401	1.250 (0.979 - 1.595)	0.073	1.302 (0.996 - 1.702)	0.054
Age	698
<= 60	555	Reference		Reference
> 60	143	4.696 (3.620 - 6.093)	< 0.001	1.493 (1.104 - 2.019)	0.009
WHO grade	636
G2&G3	468	Reference		Reference
G4	168	9.538 (7.243 - 12.560)	< 0.001	2.456 (1.702 - 3.545)	< 0.001
IDH status	688
WT	246	Reference		Reference
Mut	442	0.116 (0.089 - 0.151)	< 0.001	0.254 (0.174 - 0.369)	< 0.001
RRM2	698
Low	349	Reference		Reference
High	349	4.421 (3.336 - 5.858)	< 0.001	1.753 (1.214 - 2.532)	0.003

Utilizing these independent prognostic variables, a nomogram was developed to predict 1-, 3-, and 5-year survival probabilities in glioma patients (Figure 4A). Calibration curves demonstrated excellent concordance between predicted and observed survival outcomes (Figure 4B). Risk stratification based on the total nomogram score indicated significantly poorer survival in the high-risk group compared to the low-risk group (HR = 7.94, 95% CI: 5.78–10.91, P < 0.001; Figure 4C), thereby highlighting strong prognostic discrimination.

Figure 4.

Nomogram establishment and validation for survival prediction in glioma. (A) Nomogram incorporating RRM2 expression and clinicopathological variables to estimate 1-, 3-, and 5-year overall survival in TCGA. (B) Calibration plots for nomogram-predicted survival. (C) Kaplan–Meier curves for nomogram-based risk stratification. (D) Time-dependent ROC curves for the nomogram in TCGA. (E–G) External validation in CGGA325, including calibration, risk-group survival curves, and time-dependent ROC analyses

The predictive performance of the nomogram was further assessed through ROC analysis, which yielded AUCs of 0.880, 0.925, and 0.863 for 1-, 3-, and 5-year survival, respectively (Figure 4D). External validation in the CGGA325 cohort demonstrated strong calibration (Figure 4E) and similarly poor survival outcomes in the high-risk group (HR = 7.36, 95% CI: 4.99–10.86, P < 0.001; Figure 4F), with corresponding AUCs of 0.832, 0.890, and 0.886 (Figure 4G). These findings suggest that the nomogram integrating RRM2 with essential clinicopathological factors has strong predictive performance for glioma prognosis.

3.4. RRM2-Associated Pathways and Co-Expressed Genes in GBMLGG

To further explore RRM2-associated biological pathways in GBMLGG, we analyzed transcriptomic differences between the high- and low-RRM2 expression groups. A total of 274 significantly differentially expressed genes (DEGs) were identified (FDR < 0.05, |log₂FC| > 1), including 142 upregulated and 132 downregulated genes (Figure 5A). KEGG pathway analysis of the upregulated genes, together with overlapping results from GSEA, identified 10 significantly enriched pathways (Figure 5B), mainly related to cell cycle regulation, complement and coagulation cascades, focal adhesion, chemokine signaling, and homologous recombination.

Figure 5.

Differentially expressed genes, functional enrichment, and candidate genes related to RRM2 in brain tumors. (A) Volcano plot of differentially expressed genes between brain tumors and normal brain tissues (FDR < 0.05, |log₂FC| > 1). (B) KEGG pathway enrichment of upregulated genes and intersection with GSEA results. (C–G) Expression levels of the identified hub genes (PLAT, CXCL10, CXCL9, IL7, and CXCL11) in tumor and adjacent tissues.(H–L) Kaplan–Meier overall survival analysis of the identified hub genes (PLAT, CXCL10, CXCL9, IL7, and CXCL11).(M–Q) Correlation analysis between RRM2 and the identified hub genes (PLAT, CXCL10, CXCL9, IL7, and CXCL11)

Correlation analyses identified five genes most strongly associated with RRM2 expression: Plasminogen Activator, Tissue Type (PLAT) (R = 0.605), C-X-C Motif Chemokine Ligand 10 (CXCL10) (R = 0.546), CXCL9 (R = 0.532), Interleukin-7 (IL7) (R = 0.546), and CXCL11 (R = 0.509) (all P < 0.001). Kaplan–Meier survival analyses indicated that high expression of these genes was significantly associated with shorter overall survival in glioma patients (Figure 5H–L, P < 0.001), further supporting their association with unfavorable prognosis. Expression validation showed no significant difference in PLAT expression between tumor and adjacent tissues (Figure 5M, P = 0.3953), whereas CXCL10, CXCL9, IL7, and CXCL11 were significantly upregulated in tumor tissues (Figure 5N–Q; all P < 0.001). Collectively, these findings suggest that RRM2 expression is associated with molecular features related to chemokine signaling and cell cycle regulation in GBMLGG. However, these observations are based on correlation and enrichment analyses and do not establish a direct regulatory relationship.

4. Discussion

Glioma is one of the most common primary malignant tumors of the central nervous system, accounting for approximately 40–50% of intracranial tumors.¹⁴ Among these, glioblastoma (GBM) represents the most aggressive subtype, with a median survival of only 12–15 months and a 5-year survival rate below 5%.¹⁵ Despite advances in surgical resection combined with radiotherapy and temozolomide chemotherapy, tumor recurrence remains almost inevitable because of intratumoral heterogeneity, highly invasive growth, and therapeutic resistance. Currently, there is still no consensus regarding the optimal treatment strategy for recurrent glioma.¹⁶ Moreover, emerging molecular targeted therapies and immunotherapies have shown limited efficacy, partly because of the blood-brain barrier, compensatory signaling pathways, and the immunosuppressive tumor microenvironment.^17,18 Consequently, given the complex and heterogeneous molecular landscape of glioma, multiple genetic and epigenetic alterations contribute to tumor progression. Therefore, identifying these molecular drivers and exploring their translational potential remain important research priorities.

Pan-cancer analyses based on TCGA and other public datasets revealed that RRM2 is significantly upregulated across multiple solid tumors, with particularly strong associations with poor prognosis in glioma and kidney renal papillary cell carcinoma. This cross-cancer consistency suggests that RRM2 may be associated with biological programs related to cell proliferation and replication stress. Notably, the prognostic significance of RRM2 was especially prominent in glioma, where elevated expression was associated with a substantially increased mortality risk (HR > 2, P < 0.001), supporting its potential relevance as a prognostic biomarker in glioma.

At the biological level, we observed a positive correlation between RRM2 expression and tumor stemness indices, together with a negative correlation with the heterogeneity index MATH. However, these findings are based on correlation analyses and do not establish a causal or mechanistic relationship between RRM2 expression and intratumoral heterogeneity. Moreover, MATH reflects only one aspect of intratumoral heterogeneity based on mutant allele frequency dispersion and does not capture the full complexity of tumor evolution.^19-23 Therefore, the association between high RRM2 expression, lower MATH scores, and poor prognosis should be interpreted cautiously as a correlation-based observation, rather than direct evidence of a specific biological or evolutionary state, and warrants further mechanistic investigation.

Integrating TCGA data with clinicopathological features further highlights the clinical relevance of RRM2 in glioma. RRM2 expression was significantly higher in glioma tissues than in normal brain tissues (P < 0.001) and demonstrated strong diagnostic performance.²⁴ Moreover, elevated RRM2 expression was associated with established adverse prognostic factors, including advanced age, higher WHO grade, and IDH wild-type status.²⁵ Survival analyses showed that patients with high RRM2 expression had significantly shorter overall survival and progression-free interval (P < 0.001). Multivariate Cox analysis confirmed RRM2 as an independent prognostic factor after adjustment for major clinical variables. Importantly, Western blot and immunohistochemical analyses of clinical glioma specimens validated the overexpression of RRM2 at the protein level, supporting the reliability of the transcriptomic findings.

From a translational perspective, the prognostic nomogram integrating RRM2 expression with key clinical variables demonstrated strong predictive performance across multiple cohorts. Previous studies have shown that incorporating molecular biomarkers into prognostic models can improve risk stratification and clinical decision-making. Together, these findings suggest that RRM2 is a potential prognostic biomarker in glioma and may also have functional relevance, while its therapeutic relevance requires further investigation.²⁶

Pathway enrichment analyses indicated that RRM2-associated genes were significantly enriched in cell cycle regulation, homologous recombination repair, and chemokine signaling pathways. The enrichment of cell cycle pathways is consistent with the established role of RRM2 in dNTP synthesis and DNA replication. Previous studies have shown that RRM2 upregulation is associated with cell-cycle progression in several cancers, including lung adenocarcinoma.²⁷ Furthermore, enrichment of homologous recombination pathways suggests that RRM2 expression may be associated with DNA damage repair-related programs that are relevant to treatment response following radiotherapy or chemotherapy.^4,28-30

Correlation analyses further identified several genes strongly associated with RRM2 expression, including PLAT, CXCL9, CXCL10, CXCL11, and IL7, all of which were significantly linked to poorer survival in glioma patients.^31-33 Notably, CXCL9, CXCL10, and CXCL11 belong to the CXCR3 chemokine axis and are involved in immune cell recruitment and inflammatory signaling within the tumor microenvironment, whereas IL7 plays an important role in lymphocyte survival and immune regulation. PLAT has also been implicated in extracellular matrix remodeling and tumor invasion. These findings suggest that RRM2 expression is associated with a broader molecular context involving both tumor proliferation-related and immune microenvironment-related features. Nevertheless, these associations were derived from correlation-based analyses, and further in vitro and in vivo studies are needed to determine whether RRM2 directly regulates these chemokine-related pathways. In the present study, these co-expressed genes were analyzed primarily to characterize the molecular context associated with RRM2 expression, rather than to construct an additional multi-gene prognostic model.

Despite these findings, several limitations should be acknowledged. First, functional experiments directly evaluating the effects of RRM2 on tumor stemness, invasiveness, and treatment response were not performed. Second, the regulatory relationships between RRM2 and the identified co-expressed genes remain to be experimentally validated. Third, potential differences in RRM2 function across molecular subtypes and spatial heterogeneity within gliomas were not fully explored. In addition, several biological interpretations in this study were based on correlation and enrichment analyses and therefore should be considered exploratory rather than evidence of direct causal mechanisms. Furthermore, the CGGA cohort represents a geographically distinct population from the TCGA cohort, and potential heterogeneity related to genetic background, lifestyle, dietary habits, and clinical management may therefore exist. Thus, the CGGA validation should be interpreted as supportive evidence for the reproducibility of our findings in an independent external cohort, rather than proof of identical effects across all populations. Finally, the prognostic model would benefit from further validation in larger multicenter and multi-ethnic cohorts.

Future studies integrating glioma stem cell models, orthotopic animal models, and single-cell sequencing approaches may provide deeper insights into the molecular networks associated with RRM2 and clarify its potential role in replication stress-related programs and tumor evolution. Such studies may also help determine whether RRM2 inhibition, alone or in combination with conventional therapies, could serve as an effective therapeutic strategy for specific glioma subtypes. Ultimately, these efforts may facilitate the development of more precise diagnostic and therapeutic strategies for glioma.

5. Conclusion

RRM2 is markedly overexpressed in glioma and is strongly associated with tumor aggressiveness and poor prognosis. These findings support the potential value of RRM2 as a prognostic biomarker and suggest that it is associated with stemness-related and cell-cycle-related biological features in glioma. Further studies are needed to clarify its mechanistic and therapeutic relevance.

Supplemental Material

Supplemental Material - Multi-Omic Analysis of Glioma Identifies RRM2 as a Prognostic Biomarker Associated With Tumor Stemness and Cell Cycle–Related Pathways

Supplemental Material for Multi-Omic Analysis of Glioma Identifies RRM2 as a Prognostic Biomarker Associated With Tumor Stemness and Cell Cycle–Related Pathways by Jiayuan Li, Liubing Hou, Xuetao Han, Yu Wang, Xiang Zhan, Zizhou Zhang, Ge Zhang, Mengting Zhang, Weijing Cai, Huandi Zhou, Xiaoying Xue in Technology in Cancer Research & Treatment

Footnotes

ORCID iDs

Huandi Zhou

Xiaoying Xue

Ethical Considerations

This study was conducted in accordance with the ethical principles of the Declaration of Helsinki. The study protocol was approved by the Research Ethics Committee (Approval No. 2025-R914; approval date: 25 November 2025). Human glioma tissues and corresponding adjacent non-tumorous tissues used for Western blot and immunohistochemistry analyses were obtained from patients who underwent surgical resection at our center. Written informed consent was obtained from all participants or their legal guardians prior to tissue collection.

Author contributions

Jiayuan Li: Methodology, Writing-review & editing;

Liubing Hou: Validation, Writing-original draft;

Xuetao Han: Methodology, Validation, Funding acquisition;

Yu Wang and Xiang Zhan: Methodology, Data curation;

Zizhou Zhang: Data curation; Ge Zhang, Mengting Zhang and Weijing Cai: Investigation;

Huandi Zhou: Methodology, Supervision, Funding acquisition;

Xiaoying Xue: Methodology, Supervision, Conceptualization, Funding acquisition, Resources.

Funding

The authors disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This research was supported by Science Research Project of Hebei Education Department (CXZX2025006), Hebei Natural Science Foundation (Grant Number H2025206307), and the S&T Program of Hebei.

Declaration of Conflicting Interests

The authors declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Data Availability Statement

The authors confirm that the data supporting the findings of this study are available within the article and its supplementary materials and further inquiries can be directed to the corresponding authors.*

Supplemental Material

Supplemental material for this article is available online.

Appendix

References

Yang

Wang

Zhou

. Unveiling the role of RNA methylation in glioma: Mechanisms, prognostic biomarkers, and therapeutic targets. Cell Signal. 2024;124:111380.

Song

Lim

Monje

. Complex neural-immune interactions shape glioma immunotherapy. Immunity. 2025;58(5):1140-1160.

Yang

Zhang

Wang

, et al. Study on the effects and mechanism of RRM2 on three gynecological malignancies. Cell Signal. 2025;129:111674.

Zuo

Zhou

Chang

, et al. Ribonucleotide reductase M2 (RRM2): Regulation, function and targeting strategy in human cancer. Genes Dis. 2024;11(1):218-233.

Abdel-Rahman

Mahfouz

Habashy

. RRM2 expression in different molecular subtypes of breast cancer and its prognostic significance. Diagn Pathol. 2022;17(1):1.

Guo

Zhao

Bai

, et al. RRM2 Is a Putative Biomarker and Promotes Bladder Cancer Progression via PI3K/AKT/mTOR Pathway. J Cell Physiol. 2025;240(1):e31501.

Yazdanian

Mobarra

Fazel

, et al. Ribonucleotide-diphosphate reductase subunit M2 (RRM2) expression and colorectal cancer invasiveness: a potential prognostic biomarker. Mol Biol Rep. 2025;52(1):447.

Jin

Nuerlan

Wang

Yang

Guo

. High expression of RRM2 as an independent predictive factor of poor prognosis in patients with lung adenocarcinoma. Aging (Albany NY). 2020;13(3):3518-3535.

Zhan

Liu

Duan

, et al. Targeting HCG18 counteracts ferroptosis resistance via blocking the miR-30a-5p/RRM2/GSS pathway in hepatocellular carcinoma. Int J Biol Sci. 2025;21(6):2550-2567.

10.

Ping

Jia

Tian

. Integration of scRNA-seq and ST-seq identifies hyperproliferative RRM2+ cells features and therapeutic targets in gastric cancer. J Transl Med. 2025;23(1):795.

11.

Moon

Lee

Shah

Yoon

Woo

. RRM2 Is a CTNNB1 Transport Regulator Promoting Colon Cancer Progression. Anticancer Res. 2024;44(6):2471-2485.

12.

Shan

Wang

, et al. Ribonucleotide reductase M2 subunit silencing suppresses tumorigenesis in pancreatic cancer via inactivation of PI3K/AKT/mTOR pathway. Pancreatology. 2022;22(3):401-413.

13.

McShane

Altman

Sauerbrei

Taube

Gion

Clark

Statistics Subcommittee of the NCI-EORTC Working Group on Cancer Diagnostics . REporting recommendations for tumour MARKer prognostic studies (REMARK). Eur J Cancer. 2005;41(12):1690-1696.

14.

Cella

Bosio

Persico

, et al. PARP inhibitors in gliomas: Mechanisms of action, current trends and future perspectives. Cancer Treat Rev. 2024;131:102850.

15.

Sahoo

Paikray

Tripathy

Fernandes

Dilnawaz

. Advancements in nanotheranostics for glioma therapy. Naunyn Schmiedebergs Arch Pharmacol. 2025;398(3):2587-2608.

16.

Liu

Zhou

Ali

Lathia

Chen

. Immunotherapy for glioblastoma: current state, challenges, and future perspectives. Cell Mol Immunol. 2024;21(12):1354-1375.

17.

Krupa

Pienkowski

Tankiewicz-Kwedlo

Lyson

. Targeting the kynurenine pathway in gliomas: Insights into pathogenesis, therapeutic targets, and clinical advances. Biochim Biophys Acta Rev Cancer. 2025;1880(3):189343.

18.

Brown

Hibbard

Alizadeh

, et al. Locoregional delivery of IL-13Rα2-targeting CAR-T cells in recurrent high-grade glioma: a phase 1 trial. Nat Med. 2024;30(4):1001-1012.

19.

Parsons

. Clonal expansion of cancer driver gene mutants investigated using advanced sequencing technologies. Mutat Res Rev Mutat Res. 2024;794:108514.

20.

Manini

Laruelle

Rocha

López

. Convergent insights into intratumor heterogeneity. Trends Cancer. 2024;10(1):12-14.

21.

McGranahan

Swanton

. Clonal Heterogeneity and Tumor Evolution: Past, Present, and the Future. Cell. 2017;168(4):613-628.

22.

Palm

Elemans

Beltman

. Heritable tumor cell division rate heterogeneity induces clonal dominance. PLoS Comput Biol. 2018;14(2):e1005954.

23.

Noorbakhsh

Kim

Namburi

Chuang

. Distribution-based measures of tumor heterogeneity are sensitive to mutation calling and lack strong clinical predictive power. Sci Rep. 2018;8(1):11445.

24.

Corrales-Guerrero

Cui

Castro-Aceituno

, et al. Inhibition of RRM2 radiosensitizes glioblastoma and uncovers synthetic lethality in combination with targeting CHK1. Cancer Lett. 2023;570:216308.

25.

Louis

Perry

Wesseling

, et al. The 2021 WHO Classification of Tumors of the Central Nervous System: a summary. Neuro Oncol. 2021;23(8):1231-1251.

26.

Sun

Yang

Zhang

, et al. RRM2 is a potential prognostic biomarker with functional significance in glioma. Int J Biol Sci. 2019;15(3):533-543.

27.

Yang

Liu

, et al. SSBP1 positively regulates RRM2, affecting epithelial mesenchymal transition and cell cycle arrest in human lung adenocarcinoma cells. Cell Signal. 2025;127:111552.

28.

Giang

Lee

, et al. Targeting of RRM2 suppresses DNA damage response and activates apoptosis in atypical teratoid rhabdoid tumor. J Exp Clin Cancer Res. 2023;42(1):346.

29.

Shi

Zhang

Wang

. High RRM2 expression has poor prognosis in specific types of breast cancer. PLoS One. 2022;17(3):e0265195.

30.

Qiu

, et al. Targeting E2F8 sensitizes gemcitabine-resistant gallbladder cancer to PARP inhibitors by disrupting RRM2-driven DNA repair. J Exp Clin Cancer Res. 2025;45(1):19.

31.

Atanaki

Mirsadeghi

Manesh

Kavousi

. Integrative analysis of single-cell transcriptomic and multilayer signaling networks in glioma reveal tumor progression stage. Front Genet. 2024;15:1446903.

32.

Zheng

Chen

, et al. Calycosin (CA) inhibits proliferation, migration and invasion by suppression of CXCL10 signaling pathway in glioma. Aging (Albany NY). 2024;16(5):4191-4203.

33.

Chan

TYH

Wong

JSY

Kiang

Sun

CWY

Leung

. The duality of CXCR3 in glioblastoma: unveiling autocrine and paracrine mechanisms for novel therapeutic approaches. Cell Death Dis. 2023;14(12):835.

Supplementary Material

Please find the following supplemental material available below.

For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.

For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.

0.80 MB

0.00 MB