A risk model constructed from five copper death-related genes for predicting prognosis and immune microenvironment infiltration in ovarian cancer patients

doi:10.21203/rs.3.rs-2251360/v1

Download PDF

Research Article

A risk model constructed from five copper death-related genes for predicting prognosis and immune microenvironment infiltration in ovarian cancer patients

https://doi.org/10.21203/rs.3.rs-2251360/v1

This work is licensed under a CC BY 4.0 License

Version 1

posted

You are reading this latest preprint version

Purpose

Copper-dependent death is a novel mode of cell death. The prognostic value of copper death-associated genes in ovarian cancer needs further elucidation. In this study, a risk model based on copper death-related genes was identified for predicting prognosis and immunotherapy sensitivity in ovarian cancer patients.

Methods

Copper death-related genes were obtained according to literature reports. Transcriptome and prognostic information of ovarian cancer patients were obtained from TCGA database. Based on the expression of 10 copper death-associated genes, ovarian cancer patients from TCGA were typed by consensus clustering algorithm. Multi-gene risk profiles were developed from the TCGA cohort using least absolute shrinkage and selection operator (LASSO) regression analysis, followed by external validation set validation using the GEO database. Immunological correlation of the models was calculated by the CIBERSORTx algorithm and ESTIMATE algorithm, and drug sensitivity analysis was performed for patients with different risk models.

Results

Based on the expression of 10 copper death-related genes, 367 TCGA-OV cases could be better classified into two subtypes. 23 inter-subgroup differential genes screened in TCGA-OV were corrected with GEO data and 15 intersecting genes were extracted. Univariate COX analysis showed a total of 5 inter-subgroup difference genes associated with ovarian cancer survival (P < 0.05), and lasso regression finally screened 5 genes for the construction of risk models. GO and KEGG functional analysis of differential genes between high and low risk groups showed that multiple immune components and immunomodulatory pathways were enriched to. Further analysis of the immune status between the high and low risk groups suggested that the low risk group had more active immune cell infiltration and multiple immune checkpoint enrichment. Drug sensitivity analysis suggested that patients in the risk group had higher sensitivity to multiple chemotherapeutic agents.

Conclusion

Copper death-related genes may be used as a new biomarker for predicting ovarian cancer prognosis, and treatment targeting copper death genes may be a new therapeutic modality to achieve efficient remission in OC.

Ovarian cancer

risk model

copper death

immune infiltration

Ovarian cancer (OC) is the eighth most common cause of cancer death in women worldwide. The pathological subtype of ovarian cancer is 90% epithelial. There are five main tissue types of epithelial ovarian cancer, of which high-grade plasmacytotic ovarian cancer (HGSOC) is the most common. Because ovarian cancer is usually diagnosed at an advanced clinical stage with a poor prognosis, the 5-year relative survival rate is only 29%(1-3).The standard treatment for OC is initial tumor reduction surgery with platinum-based chemotherapy or neoadjuvant chemotherapy; however, this treatment is only effective in a minority of patients and the prognosis for OC remains poor. Early diagnosis is an important factor in improving the low survival rate of OC (4), therefore, finding a biomarker for predicting the persistence of ovarian cancer would be an important way to reduce mortality and increase the clinical benefit of ovarian cancer.

Copper-dependent death is a novel regulatory cell death modality (5) that differs from oxidative stress-related cell death modalities (apoptosis, iron death, and necrotizing apoptosis)(6). Copper-dependent death occurs through direct binding of copper to the lipid acylated components of the tricarboxylic acid (TCA) cycle, leading to lipid acylated protein aggregation and loss of iron-sulfur cluster proteins, which results in proteotoxic stress and ultimately cell death (7, 8).Thus, copper-dependent death is expected to be a new therapeutic approach (9). The relationship between copper-dependent death genes and the prognosis of OC patients is unclear, and determining the molecular characteristics of copper-dependent death genes may help to elucidate the causes of heterogeneity in the origin of OC.

In this study, 379 OC patients were classified into 2 subtypes based on 10 copper-dependent death-associated genes, and survival and immune infiltration differences among the subtypes were investigated. A risk model of copper death-related genes was also developed to further evaluate the effect of copper death-related genes on the prognosis of ovarian cancer patients. The results showed that copper death-related gene scores were important in determining the prognosis, TME cell infiltration characteristics, and immunotherapy effects in OC patients.

Data acquisition

Transcriptomic and clinical data of 379 OC samples and 88 GTEX normal ovarian tissues from the TCGA database were downloaded through the UCSC online website (http://www.genome.ucsc.edu/), and immune subtype scoring files as well as RNA stem cell scoring files in pan-cancer were downloaded. In addition, external validation data (including transcriptomic and clinical data) for 554 OC patients were obtained from the Gene Expression Omnibus (GEO) database (https://www.ncbi.nlm.nih.gov/geo/; GSE140082, GSE53963). A collection of 10 copper death-associated genes was reported in the literature (8), genes are presented in Supplementary 1.

Consensual clustering analysis

Based on the expression of 10 copper death-related genes, 379 TCGA-OV cases were clustered using the ConsenSuClusterPlus R package. Filtered by |logFC|≥1 and adj.P. Value ≤ 0.05, the "limma" package was used to screen for inter-typical differences in genes.

GEO and TCGA data correction

The R package "sva" "limma" was used to correct between GEO and TCGA data and extract intersecting genes.

Development and validation of a prognostic model for copper death-related genes

Inter-subtype differential genes were corrected with GEO database data and extracted intersecting genes. One-way COX analysis was used to screen prognosis-related inter-subtype differential genes, LASSO regression analysis, which minimizes the risk of overfitting, was used to construct the model. The gene corresponding to the minimum point of cross-validation error was selected as the characteristic gene, and based on the expression of the selected gene and the LASSO Cox regression analysis. The individualized risk scores were obtained based on the expression of the selected genes and the regression coefficients estimated in the LASSO Cox regression analysis. The risk score formula = sum (expression of the gene × corresponding coefficient), and the patient clusters were divided into high and low risk groups according to the median risk score. The prognostic ability of the model was predicted using ROC curves and column line plots.

Immune Environment Analysis

The CIBERSORTx algorithm was used to quantify the proportion of immune cells in ovarian cancer. The ESTIMATE algorithm was used to assess the immune and interstitial scores of each patient. Single sample gene set enrichment analysis (SSGSEA) algorithm was used to determine the level of immune cell infiltration in the immune microenvironment of ovarian cancer. Immune subtypes and immune checkpoints in tumor tissue were analyzed using the R package "reshape2" "ggpubr" (10).

Stem cell correlation and drug sensitivity analysis

Risk score and mRNA stem cell correlations were analyzed using the R package "ggExtra". Sensitivity of multiple chemotherapeutic agents between high and low risk groups was analyzed using the R package "pRROphetic".

Risk gene expression analysis

Using the R package "ggplot2", the differential expression of risk genes was analyzed and the expression of risk genes at the protein level was obtained from the Human Protein Atlas database (https://www.proteinatlas.org/).

Statistical analysis

All statistical analyses were performed using R version 4.1.0. Statistically processed, P < 0.05.

Ovarian cancer staging and interstage immune cell infiltration based on copper death-associated genes

The flow chart of this study is shown in Fig. 1.

To investigate the relationship and function of copper death-related genes in ovarian cancer, the PPI network analyzed the interrelationship of 10 copper death-related genes, and the results suggested a close relationship between 8 copper death genes (Fig. 2A). The survival curve results suggested that five copper death-related genes (LIAS, LIPT1, MTF1, DLD, and PDHB) were significantly associated with patient prognosis (Fig. 2B-F). TCGA-OV patient samples were clustered and analyzed by the expression levels of 10 copper death-related genes to identify two different subtypes (Fig. 2G). Principal component analysis showed significant differences in transcriptome data between these two subtypes for subtype A (n = 265) and subtype B (n = 114) (Fig. 2H). Tumor immune microenvironment scores obtained by ESTIMATE showed that the immune scores and estimated scores were higher than those in group B, suggesting that group A had a higher immune cell content (Fig. 2I). Further analysis of 23 human immune cells in ovarian cancer by the CiberSort algorithm for inter-subpopulation infiltration revealed a large number of activated immune cell infiltrates in the group A, such as activated B cells, activated CD4 and CD8 T cells and neutrophils (Fig. 2J). Finally inter-subpopulation GSVA analysis suggested that the subpopulation A was significantly enriched in immune activation pathways, including autophagy regulatory pathways, primary immunodeficiency, and natural killer cell-mediated cytotoxicity (Fig. 2K).

Construction and validation of a risk model for copper death-related genes

To construct a more accurate and highly available risk model, 23 inter-subtype differential genes were screened by applying the screening conditions of |logFC|≥1 and adj. P.Value < 0.05 (Fig. 3A). We used the R package "Sva" to correct the 23 inter-subtype difference genes from the patient data in the GEO database and extracted the intersection genes (n = 15).Removed the data with survival time less than 30 days, and used the TCGA data as the training set (n = 378) ,the GEO data as the validation set (n = 554). Single-factor Cox analysis of 15 inter-subtype differential genes finally obtained 5 prognostic differential genes associated with copper death (Fig. 3B), and Lasso regression analysis and multifactor Cox analysis of 5 prognostic differential genes finally obtained 5 (RARRES1, IFI27, IFI6, CXCL10, PI3) risk genes, including 2 high-risk genes (RARRES1, PI3) and 3 low-risk genes (IFI27, IFI6, CXCL10).

Risk score = (0.1415*RARRES1 expression) + (-0.0284*IFI27 expression) + (-0.0553*IFI6 expression) + (-0.1678*CXCL10 expression) + (0.1328*PI3 expression).

Kaplan-Meier survival curves between the high and low-risk groups for the TCGA training set showed that overall survival was significantly higher for patients in the low-risk group (Fig. 3C). Risk scores in the TCGA group were indicated to be significantly associated with survival in both univariate and multifactorial COX regression analyses (HR = 2.907, 95% CI = 1.929–4.380, p < 0.001; HR = 2.816, 95% CI = 1.856–4.271, P < 0.001) (Fig. 3D, E). The Kaplan-Meier survival curves for the high and low -risk groups of the GEO validation set showed that the overall survival of patients in the low-risk group was significantly higher (Fig. 3F). In both univariate and multifactorial COX regression analyses, the risk scores in the GEO group were significantly associated with survival (HR = 1.705, 95% CI = 1.100-2.641, P = 0.017; HR = 1.650, 95% CI = 1.048–1.473, P = 0.013) (Fig. 3G, H).

Prognostic value based on risk model

To analyze the prognostic value of the model for ovarian cancer, we constructed a nomogram containing three clinical factors for predicting 1, 2, and 3-year survival based on TCGA clinical data, and the calibration curves verified the accuracy of the nomogram (Fig. 4A, B). A nomogram was constructed based on GEO clinical data with calibration curves (Fig. 4C, D). The area under the ROC curve of the risk model showed that the risk model had some reliability (AUC = 0.626 (Fig. 4E). The DCA curve plot showed that the risk score had more clinical benefit than other clinical factors (Fig. 4F), The C index indicated that the nomogram had a good predictive value (Fig. 4G). Further analysis of the correlation between risk score and clinicopathology, the survival curves showed that patients in the high-risk group had poorer survival under different pathological conditions (Fig. 4H-K).

Correlation of the model with the immune microenvironment

To elucidate the biological functions and pathways associated with risk scores, GO, KEGG enrichment analysis was performed using differential genes between the high-risk and low-risk groups (Fig. 5A, B), and the results showed that differential genes are involved in multiple immune components such as cytokine activation, chemokine activation and chemokine receptor binding, and in multiple immune regulatory pathways. Next, we analyzed the correlation between risk genes and 10 copper death genes, and the results showed that only 9 genes may be correlated with risk genes. Copper death-related genes (CDKN2A, FDX1) were significantly positively correlated with most risk genes, while LIAS was significantly negatively correlated. (Fig. 5C). When the five risk genes of the model were evaluated in relation to immune cells for abundance, a variety of macrophages and T cells were observed to be significantly correlated with these five genes (Fig. 5D). When immune function analysis was performed between high and low -risk groups, the results showed that multiple immunomodulatory responses showed a significant expression of activation in the high-risk group (Fig. 5E). In the immune subtype analysis, four risk genes (IFI27, IFI6, CXCL10, PI3) exhibited significant differences (Fig. 5F). In addition, we investigated the association between 21 immune checkpoints and risk models, and the results showed that multiple immune checkpoint genes were mainly enriched in the low-risk group (Fig. 5G). Finally, the TME scores between the high and low risk groups showed that the low-risk group had more immune cells (Fig. 5H). Using the CIBERSORTx algorithm, we observed that fibroblasts and neutrophil (white) cells were mainly enriched in the high-risk group, and cytotoxic lymphocytes, NK cells, T cells, B-series cells and CD8 + T cells were mainly enriched in the low-risk group (Fig. 5I-O).

Comprehensive analysis of risk score models

The sensitivity of the risk model as a biomarker for predicting treatment response in ovarian cancer patients was assessed using the R package "pRROphetic". The results showed that patients with low-risk values may have higher sensitivity to Doxorubicin, Etoposide, Cyclopamine, etc.; patients with high-risk values may have higher sensitivity to Imatinib, PHA.665752, X17.AAG, etc. (Fig. 6A-F). The correlation between the risk model and mRNA stem cells was assessed using the R package "ggExtra", and the results showed that the risk score was negatively correlated with stem cell scoring (R=-0.33, P = 3e-08) (Fig. 6G). Next, we analyzed the differential expression of risk genes in the transcriptome (Fig. 6H), and in the protein level (Fig. 6I).

In the study, we systematically investigated the expression of 10 copper death-related genes in OC and their relationship with survival. A new prognostic model containing 5 risk genes was constructed for the first time and validated in an external cohort. Functional analysis between high and low risk showed an abundance of immune-related pathways, and the low-risk group was closely associated with multiple immune cells and immune checkpoints.

Currently, no copper death-related genes have been found to explore their anti-ovarian cancer activity and immune activity based on copper death-related genes. To explore the research value of copper death-related genes in OC, we observed that more than half of the 10 copper death-related genes were associated with OS in OC by univariate COX regression analysis. By consensus clustering analysis, it was observed that the 10 copper death-related genes could significantly classify OC samples into two regulatory patterns, the classification results showed that the group A was closely associated with immune cells and immune pathways. These results significantly indicate the potential role of copper death-related genes in OC and the possibility of using these copper death-related genes to build prognostic models. In this study, we finally obtained prognostic models constructed from five (RARRES1, IFI27, IFI6, CXCL10, PI3) risk genes. We performed GO and KEGG enrichment analysis of differential genes between different risk groups, and found that many biological processes and pathways related to immunity were enriched. Correlation analysis of five risk genes with 10 copper death-related genes and multiple human immune cells showed that the risk model constructed by copper death was closely related to tumor immune function. Currently, although immunotherapy has brought benefits to patients with malignant tumors, the beneficiaries are only a small percentage of patients, and a comprehensive understanding of the tumor immune environment and the search for new ones are important initiatives to exert immunotherapy (11–13). Risk model-based analysis of the TME immune microenvironment, immune cell infiltration analysis and immune checkpoint analysis consistently showed that the low-risk group had more lively immune activity. The higher survival in the low-risk group is, we speculate, associated with an increase in anti-inflammatory-associated macrophages, T cells, and dendritic cells (14–16), and the higher risk score is associated with impaired antitumor immunity, such as activity of type I interferon response, cytotoxicity, pro-inflammatory activity, and immune checkpoint activity (17–19). An important current research hotspot suggests that immune checkpoint inhibitors in combination with chemotherapeutic modalities may enhance cancer treatment effects by initiating immunostimulatory pathways (20, 21). Based on the risk model, we propose appropriate chemotherapeutic agents for patients in different risk groups. Cancer stem cells (CSCS) are a population of cells within a tumor with the ability to self-renew, differentiate, and generate tumor tissue to survive in an evasive manner from anticancer therapy, and targeting CSC therapy may prevent cancer recurrence (22, 23). CSC-based markers have been widely used for diagnosis, treatment, and prediction of poor tumor prognosis (23, 24). The risk model we constructed showed a significant negative correlation with tumor stem cells, suggesting that the risk model could interact with tumor stemness biomarkers to jointly participate in the immunotherapy of OC and predict the prognostic factors of OC.

In summary, risk scores represent patients with different clinical characteristics, with significant differences in immune status between patients in the high and low risk groups, and risk scores are associated with patients' response to immunotherapy. The copper death gene risk model can be used as a predictor to assess the clinical prognosis and immunotherapeutic response of OC. The present study has several limitations: first, we constructed a predictive model and validated it with public data. More prospective real-world data are needed to validate its clinical utility. Second, the association between risk scores and immune activity has not been validated in experiments.

Ethical approval and consent to participate (not applicable)

Consent to publication

Written informed consent for publication was obtained from all participants.

Availability of data and materials

The original data supporting the conclusions of this manuscript will be available by the author without undue reservation.

Competing interests

All authors do not have any possible conflicts of interest.

Funding

This study was supported by a grant from the Key Project of Anhui Provincial Education Department (kj2019A0363).

Authors' contributions

Prof. HYJ and Director LJ guided and reviewed the article; CSR and LY analyzed and interpreted all the data, and WYL helped to obtain the data. All read and approved the final manuscript.

Acknowledgements

Sincerely thanks to everyone who participated in this study.

Author information

1 Department of Gynecologic Oncology, The First Affiliated Hospital of Bengbu Medical College, Bengbu, Anhui, 233099, China

2 Department of Gastrointestinal Surgery, The First Affiliated Hospital of Bengbu Medical College, Bengbu, Anhui, 233099, China;

3 College of Life Sciences, Bengbu Medical College, Bengbu, Anhui, 233030, China;

Reid BM, et al. Epidemiology of ovarian cancer: a review. Cancer biology & medicine. 2017;14(1):9-32.
Jones MR, et al. Genetic epidemiology of ovarian cancer and prospects for polygenic risk prediction. Gynecologic oncology. 2017;147(3):705-13.
Xiao Y, et al. Multi-omics approaches for biomarker discovery in early ovarian cancer diagnosis. EBioMedicine. 2022;79:104001.
El Bairi K, et al. Emerging diagnostic, prognostic and therapeutic biomarkers for ovarian cancer. Cellular oncology (Dordrecht). 2017;40(2):105-18.
Wang Y, et al. Cuproptosis: a new form of programmed cell death. Cellular & molecular immunology. 2022.
Tang D, et al. Cuproptosis: a copper-triggered modality of mitochondrial cell death. Cell research. 2022;32(5):417-8.
Li SR, et al. Cuproptosis: lipoylated TCA cycle proteins-mediated novel cell death pathway. Signal transduction and targeted therapy. 2022;7(1):158.
Tsvetkov P, et al. Copper induces cell death by targeting lipoylated TCA cycle proteins. Science (New York, NY). 2022;375(6586):1254-61.
Cobine PA, Brady DC. Cuproptosis: Cellular and molecular mechanisms underlying copper-induced cell death. Molecular cell. 2022;82(10):1786-7.
Xu D, et al. Identification of immune subtypes and prognosis of hepatocellular carcinoma based on immune checkpoint gene expression profile. Biomedicine & pharmacotherapy = Biomedecine & pharmacotherapie. 2020;126:109903.
Gnjatic S, et al. Identifying baseline immune-related biomarkers to predict clinical outcome of immunotherapy. Journal for immunotherapy of cancer. 2017;5:44.
Nixon AB, et al. Peripheral immune-based biomarkers in cancer immunotherapy: can we realize their predictive potential? Journal for immunotherapy of cancer. 2019;7(1):325.
Li B, et al. Immune Checkpoint Inhibitors: Basics and Challenges. Current medicinal chemistry. 2019;26(17):3009-25.
Leavenworth JW, et al. Editorial: Immune Cell Lineage Reprogramming in Cancer. Frontiers in immunology. 2021;12:838464.
Parker ME, Ciofani M. Regulation of γδ T Cell Effector Diversification in the Thymus. Frontiers in immunology. 2020;11:42.
Cui K, et al. Increased interleukin-17A-producing γδT cells predict favorable survival in elderly patients with LUAD and LUSC. Journal of cancer research and clinical oncology. 2021;147(11):3289-98.
Oshi M, et al. Immune cytolytic activity is associated with reduced intra-tumoral genetic heterogeneity and with better clinical outcomes in triple negative breast cancer. American journal of cancer research. 2021;11(7):3628-44.
Zhang F, et al. Type I Interferon Response in Radiation-Induced Anti-Tumor Immunity. Seminars in radiation oncology. 2020;30(2):129-38.
Mazewski C, et al. Type I Interferon (IFN)-Regulated Activation of Canonical and Non-Canonical Signaling Pathways. Frontiers in immunology. 2020;11:606456.
Galluzzi L, et al. Immunostimulation with chemotherapy in the era of immune checkpoint inhibitors. Nature reviews Clinical oncology. 2020;17(12):725-41.
Zouein J, et al. The combination of immune checkpoint inhibitors and chemotherapy in advanced non-small-cell lung cancer: the rational choice. Immunotherapy. 2022;14(2):155-67.
Barbato L, et al. Cancer Stem Cells and Targeting Strategies. Cells. 2019;8(8).
Dzobo K, et al. Cancer Stem Cell Markers in Relation to Patient Survival Outcomes: Lessons for Integrative Diagnostics and Next-Generation Anticancer Drug Development. Omics : a journal of integrative biology. 2021;25(2):81-92.
Walcher L, et al. Cancer Stem Cells-Origins and Biomarkers: Perspectives for Targeted Personalized Therapies. Frontiers in immunology. 2020;11:1280.

No competing interests reported.

Supplementary1.docx

Download PDF

Version 1

posted

You are reading this latest preprint version

A risk model constructed from five copper death-related genes for predicting prognosis and immune microenvironment infiltration in ovarian cancer patients

Status:

Version 1

Abstract

Purpose

Methods

Results

Conclusion

Figures

Introduction

Materials And Methods

Data acquisition

Consensual clustering analysis

GEO and TCGA data correction

Immune Environment Analysis

Stem cell correlation and drug sensitivity analysis

Risk gene expression analysis

Statistical analysis

Results

Discussion

Declarations

References

Additional Declarations

Supplementary Files

Status:

Version 1