Unveiling critical genes and molecular subtypes in ovarian cancer: insights into tumor immunity and carbohydrate-lipid metabolism

Introduction

Ovarian cancer (OC) seriously threatens the health of women globally. Because of the lack of reliable early diagnostic methods, most patients are first diagnosed at advanced stages. It is difficult to diagnose OC in its early stages, and it is common to ignore the progression. The overall survival (OS) of OC has not exceeded 30%, even though the technology of clinical diagnosis and treatment has improved (1,2). Therefore, there is an urgent need to explore its pathogenesis and seek more efficient early detection methods and accurate molecular markers.

Alterations in the genome of various cells are linked to the dysregulation of tumor immune processes. These abnormal changes encourage cancer cells to proliferate and eventually metastasize to other organs. The development of dysregulation of the anti-tumor immune process usually leads to cancer progression, metastasis, poor prognosis, and immunotherapy failure (3). Immune checkpoint inhibitors (ICIs) are widely used to treat various cancers, but their effectiveness in treating OC has been limited (4). Therefore, it is important to deeply explore the immunomodulatory characteristics of OC and find new targets conducive to immunotherapy.

Recently, a number of investigations employing transcriptomics have explored the expression of critical genes within the biological behaviors of OC. These studies, including large-scale datasets such as The Cancer Genome Atlas (TCGA) and Gene Expression Omnibus (GEO), have revealed distinct molecular subtypes and identified key drivers related to proliferation, metastasis, and chemoresistance (5-7). For instance, certain classifications based on immune-related gene expression or mesenchymal signatures have been associated with clinical outcomes and therapeutic responses (8). However, most existing work has primarily focused on isolated pathways or generalized gene signatures, while the integration of immunological features with metabolic processes-especially carbohydrate-lipid metabolism-remains insufficiently addressed. Given the crucial role of metabolic reprogramming in modulating the tumor microenvironment and immune escape, our study seeks to fill this gap by constructing novel molecular subtypes and a prognostic risk model based on metabolism-related genes, thus offering new insights into the interplay between immunometabolism and OC progression.

At present, many studies have shown that all kinds of OC do not have a single molecular signature, but can be divided into multiple molecular subtypes. Patients with different molecular subtypes exhibited significant differences in survival rates (9-11). However, the robustness of these studies remains poor because different algorithms were used between the studies and cross-validation was lacking. Furthermore, stratifying patients by these molecular subtypes in clinical trials has not shown a difference in response rates to ICI treatment (12). These findings highlight the important impact of tumor immunity on OC and highlight the need to explore its heterogeneity more accurately.

Carbohydrate-lipid metabolism plays critical roles in tumorigenesis and tumor progression. These metabolic alterations not only facilitate the growth and survival of cancer cells but also contribute to the development of a tumor microenvironment that supports immune evasion and metastasis (13). Targeting these metabolic pathways offers potential therapeutic avenues for disrupting the metabolic flexibility and adaptability of cancer cells.

In this study, we obtained microarray datasets from the GEO (https://www.ncbi.nlm.nih.gov/geo/), comprising 156 samples in total, including 122 OC samples and 34 ovarian epithelial samples. After that, we obtained 6 core signatures associated with tumor-immune and fatty acid synthesis, which underwent comprehensive analysis. Finally, we established 3 molecular subtypes based on them and found different clinical characteristics among them. To sum up, our study offers robust molecular signatures and subtypes for detection and prognosis, along with potential immunotherapy targets for the treatment of OC. We present this article in accordance with the TRIPOD reporting checklist (available at https://tcr.amegroups.com/article/view/10.21037/tcr-2025-745/rc).

Methods

Data processing

In this study, we selected three GEO datasets-GSE18520 (14), GSE14407 (14), and GSE66957 (15)-based on the following criteria: (I) relevance to OC transcriptomic profiling, (II) adequate sample size to ensure statistical power, and (III) compatibility in terms of platform and data processing to allow for effective integration and comparison (Table 1). These datasets provide high-quality gene expression data derived from OC tissues and normal controls, enabling us to perform differential expression analysis and robust identification of candidate genes. Moreover, the inclusion of multiple independent cohorts enhances the generalizability and reliability of our findings. Subsequently, the OC transcriptomic profiling in TCGA was downloaded from UCSCXena (https://xena.ucsc.edu/) and used as an independent validation cohort to assess the robustness and prognostic value of the derived gene signatures. This design allowed for the integration of multiple data sources and enhanced the reliability of our findings. We obtained OS data for OC patients from Kaplan-Meier (KM) plotter (https://kmplot.com/analysis/), where OC samples were categorized into two groups according to gene expression levels. The study was conducted in accordance with the Declaration of Helsinki and its subsequent amendments.

Screening and ranking for differentially expressed genes (DEGs) based on expression levels

Differential expression analysis was conducted to filter out significantly dysregulated genes in OC compared to normal controls. Utilizing robust statistical methods, such as “Linear Models for Microarray Data (LIMMA) (https://www.bioconductor.org/packages/release/bioc/html/limma.html)”, “data. table (https://CRAN.R-project.org/package=data.table)” and “dplyr (https://dplyr.tidyverse.org/reference/dplyr-package.html)” packages, gene expression profiles from GEO were compared between OC samples and corresponding normal tissues. DEGs were characterized by log₂|fold change (FC)| >2 and adjusted P<0.01. After that, DEGs were ranked via robust rank aggregation (RRA) (16).

Enrichment analysis of DEGs

DEGs were classified as core molecules involved in immune or fatty acid metabolic procession via Gene Ontology (GO) Enrichment Analysis, Gene Set Variation Analysis (GSVA), or Gene Set Enrichment Analysis (GSEA). The Database for Annotation, Visualization, and Integrated Discovery (DAVID) was used to annotate the DEGs and to identify significantly enriched GO terms (17). The GO categories including biological process (BP), molecular function (MF), and cellular component (CC) were used in annotating DEGs.

GSVA is used to assess the enrichment of gene sets (18). GSEA is used to ascertain whether a pre-defined set of genes exhibits statistically significant, concordant differences between two biological states (19). In this study, gene sets associated with carbohydrate-lipid metabolism were obtained from The Molecular Signatures Database (MSigDB) v7.0. Terms with P<0.05 were deemed statistically significant.

Construction of protein-protein interaction (PPI) networks

PPI networks were established to further explore the molecular interactions among the DEGs via the STRING database (https://cn.string-db.org/) (20).

Identification of molecular subtypes in OC

We conducted unsupervised hierarchical clustering using the R package “ConsensusClusterPlus” (21) to delineate expression patterns of signatures in OC. Molecular subtypes were identified using the following parameters: maxK =4, reps =100, pItem =0.8, pFeature =1, and distance = Pearson. Utilizing unsupervised hierarchical clustering with these parameters on all genes resulting from the RRA, distinct molecular subtypes were delineated.

Quantitative analysis of tumor immune environment

The Cell-type Identification By Estimating Relative Subsets Of RNA Transcripts (CIBERSORT) algorithm was applied to assess the immune landscape and the condition of the tumor immune environment in OC (22). Additionally, the Estimation of Stromal and Immune cells in Malignant Tumor tissues using Expression data (ESTIMATE) method is applied to evaluate the infiltration of immune and tumor cells, which employed in our previous work (23).

Development of risk models tailored to distinct subtypes

To develop prognostic models tailored to the distinct molecular subtypes of OC, the least absolute shrinkage and selection operator (LASSO), implemented via the “glmnet” package in R, was then applied to construct risk models for each subtype (24). The performance of the risk models was evaluated using Kaplan-Meier survival analysis to assess their predictive power and discriminative ability.

Statistical analysis

All data preprocessing and statistical analyses were conducted using R (version 4.2.2) (25) within the RStudio environment (26). GraphPad Prism v8.0 (GraphPad Software, San Diego, CA, USA) was used for drawing figures. The differences among the different groups were assessed using both the two-tailed unpaired t-test and Fisher’s exact test. Means and SEM were used to display the variables. Statistical significance was defined as P<0.05 for every comparison.

Results

Screening, ranking and GO enrichment analysis of DEGs in OC

The normalized results of GEO datasets are displayed in Figure 1A. Screening the GSE18520 dataset resulted in 1,401 DEGs, comprising 588 upregulated and 813 downregulated genes. A total of 4,965 DEGs were identified from the GSE66957 dataset, consisting of 1,421 upregulated genes and 3,544 downregulated genes. Besides, 2,740 DEGs were identified from the GSE14407 dataset, with 1,743 upregulated genes and 997 downregulated genes (Figure 1B and available online: https://cdn.amegroups.cn/static/public/tcr-2025-745-1.xlsx).

Figure 1 Landscape of DEGs in OC derived from GEO and TCGA. (A) Normalization of datasets in GEO. Pre-normalized data is displayed in blue bars, while red bars show the processed data. (B) Volcano plots show differential expression analysis among datasets in GEO. Red points denote upregulated genes with |fold change| >2 and corrected P<0.01, green points indicate downregulated genes with similar criteria, and black points represent genes with no significant difference. (C) Heatmap displaying the top 20 up- and downregulated genes from different microarrays based on RRA analysis. Red indicates logFC >0, blue indicates logFC <0, and the numbers within the boxes represent logFC values. (D) GO enrichment analysis of DEGs in OC. (E) PPI networks of DEGs constructed using the STRING database. DEGs, differentially expressed genes; FC, fold change; GEO, Gene Expression Omnibus; GO, Gene Ontology; OC, ovarian cancer; PPI, protein-protein interaction; RRA, robust rank aggregation; TCGA, The Cancer Genome Atlas.

After that, the DEGs were ranked using the “RobustRankAggreg” package (27). A total of 199 DEGs were identified, including 101 upregulated genes and 98 downregulated genes, which are listed in Table S1. Furthermore, the “pheatmap” package (28) was utilized to generate a heatmap illustrating the expression patterns of the top 20 up-and-down-regulated genes (Figure 1C). The top 20 upregulated and downregulated DEGs in OC were analyzed using the STRING (29) to construct PPI networks, as depicted in Figure 1D.

Next, we performed GO enrichment analysis in DEGs of OC to obtain tumor immune-related gene sets (Figure 1E and Table S2). We found that some DEGs were concentrated in the immune process (GO: 0002376, GO: 0071357, and GO: 0032729). These results indicate that these DEGs may be significantly associated with immune regulation. We continued to explore the clinical significance of core signatures (RYBP, RNF2, RGL2, RCOR3, SMURF2, and SESN3) in OC. Interestingly, high expression of RGL2 was linked to prolonged OS (as depicted in Figure 2). These findings suggest that the DEGs identified may play a role in the tumor immune processes in OC, and could potentially impact the outcomes of OC patients.

Figure 2 Critical signatures associated with OS among OC patients. OS data downloaded from KM plotter and processed. HR, hazard ratio; KM, Kaplan-Meier; OC, ovarian cancer; OS, overall survival.

Identifying specific subtypes with features of different clinical characteristics based on critical signatures

Next, we conducted unsupervised consensus clustering using the expression profiles of 6 key signatures, utilizing the R package “ConsensusClusterPlus” (Figure 3A and available online: https://cdn.amegroups.cn/static/public/tcr-2025-745-2.xlsx). We derived 3 subtypes and explored their clinical characteristics. Subtype-B has the poorest OS time compared with the other 2 subtypes (median survival =39.97 months, Figure 3B,3C). By contrast, Subtype-C has the longest OS time (median survival =48.07 months, Figure 3B,3C). There was a notable disparity in survival duration among these subtypes (Log-rank test P=0.02). Next, we continued to explore the distinct clinical characteristics among these 3 subtypes. Interestingly, we found that the late International Federation of Gynecology and Obstetrics (FIGO) stage, high World Health Organization (WHO) grade, and TP53 mutations were positively associated with Subtype B, which predicted poor outcomes. However, the OC patients belonging to Subtype-C may live longer than Subtype-B because of a significant positive correlation with the early FIGO stage, low WHO grade, and stable TP53 status (Figure 3D-3F). Besides, the Subtype-A, which has no significance in OC patients, has a higher proportion of elder patients than the other two subtypes (Figure 3G).

Figure 3 Validation of OC subtypes with distinct prognosis and clinical characteristics. (A) Result of unsupervised consensus clustering. (B,C) OS among distinct molecular subtypes in OC. (D-G) Subtype-specific clinical features of subtype-B and -C shown in different colors, including the following: (D) FIGO stage, (E) WHO grade, (F) TP53 status, (G) age. FIGO, International Federation of Gynecology and Obstetrics; OC, ovarian cancer; OS, overall survival; TCGA-OV, The Cancer Genome Atlas Ovarian Cancer; WHO, World Health Organization.

Distinct metabolic processes and tumor immune environment among OC subtypes

Furthermore, we performed GSVA on Subtype-B and Subtype-C. Significantly, there is a positive correlation between Subtype-B and glucose-lipid metabolic pathways, including KEGG_CITRATE_CYCLE_TCA_CYCLE, KEGG_CYSTEINE_AND_METHIONINE_METABOLISM, KEGG_DNA_REPLICATION, KEGG_FRUCTOSE_AND_MANNOSE_METABOLISM, KEGG_GALACTOSE_METABOLISM, KEGG_GLUTATHIONE_METABOLISM and KEGG_GLYCOLYSIS_GLUCONEOGENESIS and so on. However, we found that the GSVA score of KEGG_PEROXISOME in Subtype-B was lower than Subtype-C, which is considered the indicator of ferroptosis mediated by lipid accumulation (Figure 4A and available online: https://cdn.amegroups.cn/static/public/tcr-2025-745-3.xlsx). Besides, Subtype-C has a higher GSVA score than Subtype-B, which includes GOBP_IMMUNE_EFFECTOR_PROCESS, GOBP_IMMUNE_RESPONSE, GOBP_IMMUNE_SYSTEM_DEVELOPMENT and GOBP_INNATE_IMMUNE_RESPONSE (Figure 4A and available online: https://cdn.amegroups.cn/static/public/tcr-2025-745-4.xlsx). Recognizing the pivotal role of the tumor immune microenvironment in tumorigenesis, we investigated the relationship between immune cell infiltration and molecular subtypes in OC. Among all subtypes, Subtype-C exhibited the highest and most significant correlation with infiltration of T cells, Monocyte, NK cells, and Macrophages, which are involved in the anti-tumor process (Figure 4B). Next, we compared the immune and stromal scores of Subtype-B with Subtype-C via ESTIMATE. As shown in Figure 4A, the immune score in Subtype-C was higher than Subtype-B, which has a higher stromal score (Figure 4C).

Figure 4 Distinct OC subtypes with different metabolic processes and immune conditions (A) Heatmaps show the result of the KEGG pathway and GO ontology enrichment via GSVA in different OC subtypes. (B) Evaluating 22 infiltrating immune cells in different OC subtypes. (C) Calculation of immune and stromal scores in different OC subtypes. Student’s t-test, KEGG, Kyoto Encyclopedia of Genes and Genomes; GEO, Gene Expression Omnibus; GO, Gene Ontology; GSVA, Gene Set Variation Analysis; OC, ovarian cancer; TCGA, The Cancer Genome Atlas.

Construction of risk models and groups based on critical signatures

To summarize the further investigation into the prognostic value of six critical signatures, a risk model was constructed based on their expression levels using the LASSO algorithm, achieving optimal performance at lambda =0.01254. Subsequently, the risk model was refined by selecting 4 signatures (Figure 5A). The risk signature formula was defined as follows: risk score = RYBP×0.105037 + RNF2×(−0.02080592) + SMURF2×(−0.06235004) + SESN3×0.01864498. Based on the calculated risk scores, OC patients from TCGA were stratified into high- and low-risk groups, and OS was evaluated accordingly. The high-risk group exhibited significantly shorter OS compared to the low-risk group (Log-rank test P<0.001) (Figure 5B and available online: https://cdn.amegroups.cn/static/public/tcr-2025-745-5.xlsx). Furthermore, signatures associated with fatty acid synthesis, which includes acetyl-CoA carboxylase alpha (ACACA), acetyl-CoA carboxylase beta (ACACB), acyl-CoA synthetase family member 3 (ACSF3), 3-oxoacyl-ACP Synthase of mitochondrial (OXSM), hydroxysteroid 17-beta dehydrogenase 8 (HSD17B8), and hydroxyacyl-thioester dehydratase type 2 (HTD2), were upregulated in the high-risk group (Figure 5C). GSEA revealed significant enrichment of critical molecules in the Kirsten rat sarcoma viral oncogene homolog (KRAS) signaling pathway, Hedgehog signaling pathway, and Wingless/Integrated (Wnt)/β-catenin pathway in the high-risk group (Figure 5D and available online: https://cdn.amegroups.cn/static/public/tcr-2025-745-6.xlsx). Additionally, OC patients in the high-risk group were characterized by advanced FIGO stage, high WHO grade, and tumor protein P53 (TP53)/breast cancer type 1 susceptibility protein (BRCA1) mutations (Figure 5E).

Figure 5 The risk model based on six critical signatures and distinct characteristics in different risk score groups. (A) LASSO dimensionality reduction and locus plot of six genes in prognosis model. (B) Overall survival plot for high- and low-risk score groups. Red indicates the high-risk score group and blue indicates the low-risk score group. (C) Expression of ten genes associated with fatty acid synthesis. Red indicates the high-risk score group and blue indicates the low-risk score group. Student’s t-test. (D) GSEA of high- and low-risk score groups. (E) Distinct clinical characteristics in high- and low-risk score groups. FIGO, International Federation of Gynecology and Obstetrics; GSEA, Gene Set Enrichment Analysis; LASSO, least absolute shrinkage and selection operator; TCGA-OV, The Cancer Genome Atlas Ovarian Cancer; WHO, World Health Organization.

Discussion

OC remains one of the deadliest gynecological malignancies, boasting the highest mortality rate among its counterparts (30). Therefore, the identification of novel therapeutic targets and molecular subtypes is imperative in the fight against OC.

In this study, we validated a series of DEGs associated with tumor microenvironment. It is known that OC has no significantly effective responses for anti-tumor immunotherapy (31). Then regulation of these immune cells may be influenced by the six biomarkers identified in our study.

RING1 and YY1 binding protein (RYBP) has been reported to sustain the transcriptionally suppressed status of numerous genes. The role of anti-tumor of RYBP has been reported in several kinds of cancers (32,33). Maybee et al. found that up-regulation of RYBP hinders osteosarcoma migration (34). Related to opposite findings in our research, we speculate that perhaps the up-regulation of RYBP inhibits the transcription of parts of oncogenes while also hindering the expression of a series of oncogenes, leading to disruption of the anti-tumor process and the occurrence of cancer. Ring finger protein 2 (RNF2), which is the E3 ubiquitin-protein ligase, has been regarded as a negative regulator of anti-tumor immunity in breast cancer (35). Ral guanine nucleotide dissociation stimulator-like 2 (RGL2) is the putative effector of Ras and/or Rap protein. It has been reported that RGL2 drives metastasis and growth in many kinds of cancers (36,37). Sun et al. revealed that inhibition of RGL2 suppressed the Ral activity of pancreatic cancer cells, which displayed slower growth and invasion (38). REST corepressor 3 (RCOR3) is recognized as a component of a corepressor complex responsible for transcriptional repression. Xue et al. found that the level of serum RCOR3 reflects the emergence of hepatitis, potentially serving as a biomarker for patients (39). Our research suggests that elevated RCOR3 expression is correlated with reduced OS time in OC patients, which might promote the cancer development via activating inflammatory signaling pathways. SMAD ubiquitination regulatory factor 2 (SMURF2) receives ubiquitin from an E2 ubiquitin-conjugating enzyme in a thioester form, subsequently transferring the ubiquitin directly to its targeted substrates. It is reported that SMURF2 enhanced cellular transformation and tumorigenesis of HeLa cells with TGF-β, which interacts with AIMP2/p38 (40). Ramkumar et al. found that down-regulation of SMURF2 promotes YY1-mediated trans-activation of c-Myc and B-cell proliferation (41). Sestrin 3 (SESN3) is regarded as an intracellular leucine sensor that negatively regulates the TORC1 signaling pathway. Fan, Liyuan et al. found that miR-194-3p/SESN3 axis is the potential modulatory mechanism by influences the regulation of autophagy, macroautophagy, and chaperone-mediated autophagy (42).

To investigate the heterogeneity of OC comprehensively, we divided the samples into 3 subtypes based on the level of expression of the 6 critical signatures. Interestingly, we found that Subtype-B and Subtype-C have significant distinctions in clinical characteristics and molecular regulation. According to the result of GSVA, we surmise that the stimulated glucose-lipid metabolism and cell division result in the poor prognosis of OC patients in Subtype-B. Meanwhile, OC samples are more activated in immune response and infiltration of immune cells in Subtype-C, which suggested that glucose-lipid metabolism might inhibit the anti-tumor process via suppressing the immune response and infiltration of immune cells.

We developed a risk model for OC samples, leveraging the expression profiles of 6 significant genes alongside OS data from OC patients. Apart from the adverse outcomes and clinical characteristics observed in the high-risk cohort, our findings also unveiled potential enhancements in fatty acid synthesis and associated pathways within this subgroup, which has been implicated in conferring resistance to various cancer therapies, including chemotherapy, radiotherapy, and immunotherapy (43-45). Previous studies have underscored the interplay between IFNγ signaling and specific fatty acids as a natural mechanism promoting tumor ferroptosis (46). Thus, the construction of a risk model centered on critical immune signatures facilitated an in-depth exploration of the intricate relationship between tumor immunity and glucose or fatty acid metabolism.

Intratumoral heterogeneity (ITH) in OC has been widely recognized and poses a significant challenge to the interpretation of bulk transcriptomic data. While our study provides a comprehensive overview of molecular subtypes based on bulk-level signatures, it does not capture the potential influence of rare but functionally significant tumor cell populations, as demonstrated in recent single-cell studies (47,48). Genes such as interleukin-4 (IL-4) may be expressed only in a minor fraction of cancer cells, yet exert substantial effects on the tumor microenvironment and responses to immunotherapy. Future investigations incorporating single-cell or spatial transcriptomics will be essential to validate our findings and refine molecular subtype definitions, particularly with respect to cellular heterogeneity and localized immune modulation.

This study provides novel insights into the molecular landscape of OC by integrating immunological characteristics with carbohydrate-lipid metabolic signatures-an area that remains underexplored in the current literature. Unlike previous studies that have focused primarily on immune phenotypes or metabolic pathways in isolation, our work identifies molecular subtypes based on the intersection of these two dimensions. Furthermore, we developed and validated a risk prediction model using four metabolism-related genes that are not only prognostically relevant but also functionally linked to tumor immunity. To our knowledge, this is one of the first studies to propose a subtype classification and risk stratification framework grounded in immunometabolic gene expression patterns in OC. These findings expand our understanding of tumor heterogeneity and may offer a foundation for personalized therapeutic strategies targeting metabolic vulnerabilities and immune evasion.

Conclusions

In conclusion, this study identified key gene signatures related to carbohydrate-lipid metabolism that define distinct molecular subtypes of OC. A risk prediction model based on four critical genes was developed and validated, demonstrating strong prognostic value. These findings provide new insights into the immunometabolic landscape of OC and may inform future personalized therapeutic strategies.

Acknowledgments

None.

Footnote

Reporting Checklist: The authors have completed the TRIPOD reporting checklist. Available at https://tcr.amegroups.com/article/view/10.21037/tcr-2025-745/rc

Peer Review File: Available at https://tcr.amegroups.com/article/view/10.21037/tcr-2025-745/prf

Funding: None.

Conflicts of Interest: All authors have completed the ICMJE uniform disclosure form (available at https://tcr.amegroups.com/article/view/10.21037/tcr-2025-745/coif). The authors have no conflicts of interest to declare.

Ethical Statement: The authors are accountable for all aspects of the work in ensuring that questions related to the accuracy or integrity of any part of the work are appropriately investigated and resolved. The study was conducted in accordance with the Declaration of Helsinki and its subsequent amendments.

Open Access Statement: This is an Open Access article distributed in accordance with the Creative Commons Attribution-NonCommercial-NoDerivs 4.0 International License (CC BY-NC-ND 4.0), which permits the non-commercial replication and distribution of the article with the strict proviso that no changes or edits are made and the original work is properly cited (including links to both the formal publication through the relevant DOI and the license). See: https://creativecommons.org/licenses/by-nc-nd/4.0/.

References

1. Webb PM, Jordan SJ. Global epidemiology of epithelial ovarian cancer. Nat Rev Clin Oncol 2024;21:389-400. [Crossref] [PubMed]
2. Han GYQ, Alexander M, Gattozzi J, et al. Ecological and evolutionary dynamics to design and improve ovarian cancer treatment. Clin Transl Med 2024;14:e70012. [Crossref] [PubMed]
3. Rosario SR, Long MD, Chilakapati S, et al. Integrative multi-omics analysis uncovers tumor-immune-gut axis influencing immunotherapy outcomes in ovarian cancer. Nat Commun 2024;15:10609. [Crossref] [PubMed]
4. Ghisoni E, Morotti M, Sarivalasis A, et al. Immunotherapy for ovarian cancer: towards a tailored immunophenotype-based approach. Nat Rev Clin Oncol 2024;21:801-17. [Crossref] [PubMed]
5. Sun C, Li X, Teng Q, et al. Targeting platinum-resistant ovarian cancer by disrupting histone and RAD51 lactylation. Theranostics 2025;15:3055-75. [Crossref] [PubMed]
6. Wang Y, Zhao G, Condello S, et al. Frizzled-7 Identifies Platinum-Tolerant Ovarian Cancer Cells Susceptible to Ferroptosis. Cancer Res 2021;81:384-99. [Crossref] [PubMed]
7. Ma S, Li R, Li G, et al. Identification of a G-protein coupled receptor-related gene signature through bioinformatics analysis to construct a risk model for ovarian cancer prognosis. Comput Biol Med 2024;178:108747. [Crossref] [PubMed]
8. Elorbany S, Berlato C, Carnevalli LS, et al. Immunotherapy that improves response to chemotherapy in high-grade serous ovarian cancer. Nat Commun 2024;15:10144. [Crossref] [PubMed]
9. Davidson NR, Barnard ME, Hippen AA, et al. Molecular Subtypes of High-Grade Serous Ovarian Cancer across Racial Groups and Gene Expression Platforms. Cancer Epidemiol Biomarkers Prev 2024;33:1114-25. [Crossref] [PubMed]
10. Da-Anoy A, Kang EY, Lee CH, et al. Molecular Surrogate Subtypes of Ovarian and Peritoneal Low-grade Serous Carcinoma. Int J Gynecol Pathol 2024;43:617-25. [Crossref] [PubMed]
11. Zhou M, Pi J, Zhao Y. Integrative Multi-Omics Analysis Reveals Molecular Subtypes of Ovarian Cancer and Constructs Prognostic Models. J Immunother 2025;48:197-208. [Crossref] [PubMed]
12. Gupta R, Kumar R, Penn CA, et al. Immune evasion in ovarian cancer: implications for immunotherapy and emerging treatments. Trends Immunol 2025;46:166-81. [Crossref] [PubMed]
13. Ma Y, Kroemer G. The cancer-immune dialogue in the context of stress. Nat Rev Immunol 2024;24:264-81. [Crossref] [PubMed]
14. Donninger H, Bonome T, Radonovich M, et al. Whole genome expression profiling of advance stage papillary serous ovarian cancer reveals activated pathways. Oncogene 2004;23:8065-77. [Crossref] [PubMed]
15. Marchion DC. Pathways of ovarian cancer development. GEO DataSets 2015. Available online: https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE66957
16. A J, Zhang B, Zhang Z, et al. Novel Gene Signatures Predictive of Patient Recurrence-Free Survival and Castration Resistance in Prostate Cancer. Cancers (Basel) 2021;13:917.
17. Sherman BT. DAVID Knowledgebase: a gene-centered database integrating heterogeneous gene annotation resources to facilitate high-throughput gene functional analysis. BMC Bioinformatics 2007;8:426. [Crossref] [PubMed]
18. Hänzelmann S, Castelo R, Guinney J. GSVA: gene set variation analysis for microarray and RNA-seq data. BMC Bioinformatics 2013;14:7. [Crossref] [PubMed]
19. Subramanian A, Tamayo P, Mootha VK, et al. Gene set enrichment analysis: a knowledge-based approach for interpreting genome-wide expression profiles. Proc Natl Acad Sci U S A 2005;102:15545-50. [Crossref] [PubMed]
20. von Mering C, Huynen M, Jaeggi D, et al. STRING: a database of predicted functional associations between proteins. Nucleic Acids Res 2003;31:258-61. [Crossref] [PubMed]
21. Wilkerson MD, Hayes DN. ConsensusClusterPlus: a class discovery tool with confidence assessments and item tracking. Bioinformatics 2010;26:1572-3. [Crossref] [PubMed]
22. Newman AM, Liu CL, Green MR, et al. Robust enumeration of cell subsets from tissue expression profiles. Nat Methods 2015;12:453-7. [Crossref] [PubMed]
23. Wang X, Xie W, Zhao D, et al. Molecular Subtypes of Ovarian Cancer Based on Lipid Metabolism and Glycolysis Reveals Potential Therapeutic Targets. Front Biosci (Landmark Ed) 2023;28:253. [Crossref] [PubMed]
24. Engebretsen S, Bohlin J. Statistical predictions with glmnet. Clin Epigenetics 2019;11:123. [Crossref] [PubMed]
25. Null RCTR, Team R, Null RCT, et al. R: A language and environment for statistical computing. Computing 2011;1:12-21.
26. Team R. RStudio: Integrated development environment for R. RStudio, Inc., Boston; 2014.
27. Kolde R, Laur S, Adler P, et al. Robust rank aggregation for gene list integration and meta-analysis. Bioinformatics 2012;28:573-80. [Crossref] [PubMed]
28. Vacanti NM. The Fundamentals of Constructing and Interpreting Heat Maps. Methods Mol Biol 2019;1862:279-91. [Crossref] [PubMed]
29. Szklarczyk D, Gable AL, Lyon D, et al. STRING v11: protein-protein association networks with increased coverage, supporting functional discovery in genome-wide experimental datasets. Nucleic Acids Res 2019;47:D607-13. [Crossref] [PubMed]
30. Siegel RL, Kratzer TB, Giaquinto AN, et al. Cancer statistics, 2025. CA Cancer J Clin 2025;75:10-45. [Crossref] [PubMed]
31. Kurnit KC, Odunsi K. Harnessing Antitumor Immunity in Ovarian Cancer. Cold Spring Harb Perspect Med 2024;14:a041336. [Crossref] [PubMed]
32. Li Q, Chen J, Liang F, et al. RYBP modulates embryonic neurogenesis involving the Notch signaling pathway in a PRC1-independent pattern. Stem Cell Reports 2021;16:2988-3004. [Crossref] [PubMed]
33. Jin FE, Xie B, Xian HZ, et al. Knockdown of miR-125b-5p inhibits the proliferation and invasion of gastric carcinoma cells by targeting RYBP. Kaohsiung J Med Sci 2021;37:863-71. [Crossref] [PubMed]
34. Maybee DV, Psaras AM, Brooks TA, et al. RYBP Sensitizes Cancer Cells to PARP Inhibitors by Regulating ATM Activity. Int J Mol Sci 2022;23:11764. [Crossref] [PubMed]
35. Zhang Z, Luo L, Xing C, et al. RNF2 ablation reprograms the tumor-immune microenvironment and stimulates durable NK and CD4(+) T-cell-dependent antitumor immunity. Nat Cancer 2021;2:1018-38. [Crossref] [PubMed]
36. Cheng Q, Wu Y, Xia H, et al. RGL2 as an age-dependent factor regulates colon cancer progression. Comput Struct Biotechnol J 2021;19:2190-201. [Crossref] [PubMed]
37. O Santos A. Parrini MC, Camonis J. RalGPS2 Is Essential for Survival and Cell Cycle Progression of Lung Cancer Cells Independently of Its Established Substrates Ral GTPases. PLoS One 2016;11:e0154840. [Crossref] [PubMed]
38. Sun MS, Yuan LT, Kuei CH, et al. RGL2 Drives the Metastatic Progression of Colorectal Cancer via Preventing the Protein Degradation of β-Catenin and KRAS. Cancers (Basel) 2021;13:1763. [Crossref] [PubMed]
39. Xue JH, Zheng M, Xu XW, et al. Involvement of REST corepressor 3 in prognosis of human hepatitis B. Acta Pharmacol Sin 2011;32:1019-24. [Crossref] [PubMed]
40. Malonis RJ, Fu W, Jelcic MJ, et al. RNF11 sequestration of the E3 ligase SMURF2 on membranes antagonizes SMAD7 down-regulation of transforming growth factor β signaling. J Biol Chem 2017;292:7435-51. [Crossref] [PubMed]
41. Ramkumar C, Cui H, Kong Y, et al. Smurf2 suppresses B-cell proliferation and lymphomagenesis by mediating ubiquitination and degradation of YY1. Nat Commun 2013;4:2598. [Crossref] [PubMed]
42. Fan L, Li B, Li Z, et al. Identification of Autophagy Related circRNA-miRNA-mRNA-Subtypes Network With Radiotherapy Responses and Tumor Immune Microenvironment in Non-small Cell Lung Cancer. Front Genet 2021;12:730003. [Crossref] [PubMed]
43. Yu Y, Yu J, Ge S, et al. Novel insight into metabolic reprogrammming in cancer radioresistance: A promising therapeutic target in radiotherapy. Int J Biol Sci 2023;19:811-28. [Crossref] [PubMed]
44. de Freitas Rodrigues J, Philippsen HK, Dolabela MF, et al. The Potential of DHA as Cancer Therapy Strategies: A Narrative Review of In Vitro Cytotoxicity Trials. Nutrients 2023;15:2006. [Crossref] [PubMed]
45. Wang Y, Guo Z, Isah AD, et al. Lipid metabolism and tumor immunotherapy. Front Cell Dev Biol 2023;11:1187989. [Crossref] [PubMed]
46. Liao P, Wang W, Wang W, et al. CD8(+) T cells and fatty acids orchestrate tumor ferroptosis and immunity via ACSL4. Cancer Cell 2022;40:365-378.e6. [Crossref] [PubMed]
47. Mollaoglu G, Tepper A, Falcomatà C, et al. Ovarian cancer-derived IL-4 promotes immunotherapy resistance. Cell 2024;187:7492-7510.e22. [Crossref] [PubMed]
48. Vázquez-García I, Uhlitz F, Ceglia N, et al. Ovarian cancer mutational processes drive site-specific immune evasion. Nature 2022;612:778-86. [Crossref] [PubMed]