Skip to main content
NCBI home page
Journal of Cancer Research and Clinical Oncology logoLink to Journal of Cancer Research and Clinical Oncology
. 2023 Aug 16;149(16):14673–14689. doi: 10.1007/s00432-023-05267-z

Identification of ferroptosis-related molecular subtypes and a methylation-related ferroptosis gene prognostic signature in cervical squamous cell carcinoma

Lijun Yu 1,2,#, Zhenwei Gao 1,#, Zeyu Li 3, Ping Liu 1,✉, Ya Gao 2, Gang Liang 4
PMCID: PMC11796721  PMID: 37584713

Abstract

Purpose

We aimed to investigate the molecular characteristics of cervical squamous cell carcinoma (CESC) by analyzing ferroptosis-related gene (FRG) expression data to predict prognosis.

Methods

Gene expression and clinicopathological data of patients with CESC were collected from the Cancer Genome Atlas and the Genotype-Tissue Expression databases. Using Cox regression analysis, we identified 21 FRGs associated with prognosis. Cluster analysis categorized patients into subgroups based on these genes and compared their clinicopathological, biological, and immune infiltration features. FRG methylation levels were examined, and a risk model based on such FRG methylation levels was constructed using LASSO and Cox regression analyses. The model’s predictive capacity was validated, and the relationships between the risk score and immune infiltration, tumor microenvironment, and drug sensitivity were explored. FRG methylation in CESC tissues was validated by immunohistochemistry.

Results

We identified 21 FRGs associated with CESC prognosis. Patients were stratified into two subtypes based on these genes, they showed differences in prognosis, immune cell types, and immune checkpoint expression. A three-gene risk score (including AQP3, MGST1, and TFRC) was generated, and the low-risk group showed better overall survival. The high-risk and low-risk groups differed in terms of immune infiltration, gene mutations, and drug sensitivity. Experimental validation confirmed the upregulation of AQP3 and TFRC, whereas MGST1 expression was not significantly altered in CESC tissues compared with that in normal cervical tissues.

Conclusion

This study highlights the potential role of FRG methylation in predicting CESC prognosis and provides a personalized assessment of immune responses in patients with CESC.

Keywords: Cervical squamous cell carcinoma, Ferroptosis, Prognosis, Tumor immune microenvironment, Tumor mutation burden

Introduction

Cervical cancer, with approximately 604,000 new cases and over 342,000 related deaths worldwide in 2020, is the fourth most common malignant tumor and the fourth leading cause of cancer-related deaths in women (Sung et al. 2021). Despite the tremendous progress that has been made in therapies, including surgery, radiation, and chemotherapy, the prognosis of some patients remains poor even after standard treatment and approximately 75% of patients experience disease progression and/or recurrence (Kim et al. 2017; Cohen et al. 2019). Considering that medical technology, therapeutics, and related areas are continuously improving, it is essential to explore the tumor markers for predicting prognosis and identifying new therapeutic targets.

Over the past few years, research on tumor ferroptosis has rapidly increased. In 2012, Dixon et al. (2012) first defined ferroptosis, an iron-dependent and caspase-independent form of nonapoptotic cell death. Ferroptosis, distinct from classical programmed cell death, is characterized by mitochondrial shrinkage and increased mitochondrial membrane density (morphological features), accumulation of iron and lipid reactive oxygen species (L-ROS) (biochemical changes), and the involvement of a unique set of genes (genetic element) (Hassannia et al. 2019; Yee et al. 2020). Notably, the dysregulation of iron metabolism can increase the risk of cancer and promote tumor proliferation (Bogdan et al. 2016). Cancer cells require high iron concentrations for their rapid proliferation, suggesting that they are more vulnerable to ferroptosis than normal cells (Manz et al. 2016; Li and Li 2020). Multiple studies have demonstrated the pivotal role of ferroptosis in killing tumor cells and impeding tumor growth in many malignant tumors (Yamaguchi et al. 2013; Ooko et al. 2015), particularly in tumors with a high tendency for metastasis and resistance to conventional therapies (Hangauer et al. 2017; Xu et al. 2019). Significant efforts have been made to target the regulatory mechanisms of iron-dependent death in cancer cells in order to develop novel therapies (Sun et al. 2016). However, the function and regulatory mechanisms of ferroptosis in tumor biology, as well as its association with the tumor immune microenvironment, remain unclear.

With the tremendous increase in transcriptome sequencing data of tumor tissues, numerous prognostic models based on ferroptosis-related genes (FRGs) have been successfully established, including those for hepatocellular carcinoma (Wan et al. 2022), lung adenocarcinoma (Cheng et al. 2022), head and neck squamous cell carcinoma (Chen et al. 2022), and cervical cancer (Qin et al. 2022). These models have the potential to predict prognosis and screen for therapeutic molecular targets. Cervical cancer consists of two main subtypes: squamous cell carcinoma (CESC) and adenocarcinoma. CESC is the most common histological subtype, accounting for approximately 75% of all cervical cancer cases (Small et al. 2017). To date, most research has focused on investigating the association between ferroptosis and cervical cancer (Qi et al. 2021; Yang and Al-Hendy 2022), whereas studies specifically exploring the relationship between ferroptosis and CESC are limited. Moreover, studies on the impact of ferroptosis on the survival of patients with CESC have been relatively scarce.

In this investigation, our objective was to examine the prognostic significance of FRGs and establish an innovative three-FRG-based risk model, while exploring the clinical implications of these genes in CESC. Initially, we scrutinized the RNA profiles and clinical data of patients with CESC sourced from the Cancer Genome Atlas (TCGA). Subsequently, differentially expressed FRGs that correlated with overall survival (OS) were used for cluster analysis and the development of a novel prognostic model. We observed that patient prognosis, pathway enrichment and immune infiltration exhibited marked differences across subtype and risk classifications. The ferroptosis-related (FR) risk model was identified as an independent prognostic determinant using univariate and multivariate Cox regression analyses. Furthermore, the IC50 values of the targeted therapeutic agents for CESC displayed a significant disparity between the high- and low-risk groups, implying that patients in distinct risk strata possess varying drug sensitivities. To further validate these findings, we conducted immunohistochemistry (IHC) assays on clinical specimens to compare the expression levels in tumor and normal tissues of the genes included in the risk model. The results substantiated the utility of the risk signature for prognosis prediction and selection of patients for immunotherapies, highlighting its potential clinical significance. Our study provides novel insights into the prediction and treatment of CESC.

Materials and methods

Data collection

RNA sequencing (RNA-seq) data from 255 tumor and 2 normal samples and the corresponding clinical information of patients with CESC, was obtained from TCGA database (https://portal.gdc.cancer.gov/repository/). In addition, RNA-seq data from 6 normal samples were downloaded from the Genotype-Tissue Expression (GTEx) database (https://gtexportal.org/home/). In total, 483 validated FRGs were identified using FerrDb (http://www.zhounan.org/ferrdb/).

Identification of differentially expressed FRGs and functional enrichment analysis

Raw expression data were quantile-normalized using the R software. The “limma” package in RStudio was used to identify significantly differentially expressed genes (DEGs) between tumor and adjacent normal tissues using the Wilcoxon test, with parameters set at ∣ logFC ∣ > 1.5 and a p value < 0.05 cut-off. The intersection of DEGs and FRGs was visualized using a Venn diagram. We used the topGO R package for GO analyses and the clusterProfiler R package for KEGG analyses of differentially expressed FRGs, and the data were visualized in R using the “org.db” and “ggplot2” packages. All GO and KEGG terms with p and q values < 0.05 were considered significantly enriched.

Identification and cluster analysis of FR DEGs associated with OS

Univariate Cox analysis of OS was performed to identify the survival-related FRGs associated with significant prognostic values using the “survival” package in R: p values ≤ 0.05 were considered statistically significant. Cluster analysis was performed using the R package “ConsensusClusterPlus” to identify FR molecular subtypes, and Kaplan–Meier analysis was conducted to compare the prognosis between the two resulting clusters. The relationship between clusters and clinical parameters was visualized through heat maps using a color range from blue to red.

Immune cell infiltration and immune microenvironment evaluation

CIBERSORT uses a machine learning approach to estimate the cellular composition of complex tissues in each sample based on the expression profiles of a set of reference cell types (Newman et al. 2015). The expression feature leukocyte signature matrix 22 (LM22) was used as a reference for the analysis of human data. The total immune infiltrate in each sample and immune cell subset were estimated using CIBERSORT with the LM22 gene set from the CIBERSORT database. Single-sample gene set enrichment analysis (ssGSEA) scores were used to calculate the degree of enrichment of each gene set in each sample, which was determined based on the expression levels of 28 immune cell-specific marker genes (Finotello and Trajanoski 2018). After defining the immune cell-associated gene sets, the enrichment score of these gene sets represents the density of tumor-infiltrating immune cells. Based on the CIBERSORT and ssGSEA algorithms, we analyzed the differences in immune cell infiltration between CESCs with different ferroptosis molecular subtypes. The relationship between different ferroptosis molecular subtypes and tumor immune infiltration, as well as the association between risk scores and tumor-infiltrating immune cells, was investigated.

We measured the immune, stromal and estimated scores between samples using ESTIMATE (Yoshihara et al. 2013), a gene expression signature-based method that assesses immune cell infiltration and tumor microenvironment from gene expression data.

Gene set variation analysis (GSVA) and immune checkpoint analysis

GSVA is a nonparametric and unsupervised method for evaluating the enrichment of transcriptome gene sets. It transforms the gene level into the pathway level by performing a comprehensive scoring of the gene set of interest, thus assessing the biological function of the samples. In this study, the R package “GSVA” was used to assess the enrichment of pathways for the two identified ferroptosis subtypes. All marker gene sets were downloaded from the Molecular Signature Database (MSigDB). Statistical significance was set to adjusted p values ≤ 0.05.

Immune checkpoint-related gene expression levels may be associated with therapeutic responsiveness to immune checkpoint inhibitors. By analyzing gene expression between the two different subtypes, the correlation between molecular subtypes and immune checkpoints was investigated.

UALCAN database analysis

UALCAN, based on TCGA can help scholars analyze epigenetic data of 31 common cancers online, and identify potential regulators of gene expression by these mechanisms. Additionally, the website can be used to analyze methylation levels online (Chandrashekar et al. 2017). Therefore, we used UALCAN to evaluate the epigenetic regulation of 21 FRGs via promoter methylation. We also analyzed the methylation levels of five hub genes among these 21 FRGs.

Construction and validation of a prognostic FRG signature in CESC

FR DEGs associated with survival, determined by univariate Cox analyses, were selected to construct a prognostic risk model using LASSO regression and multivariate Cox analysis. After incorporating the expression value of each specific gene, a risk score formula was built for each patient, weighted by the estimated regression coefficients after the LASSO regression analysis. According to the risk score formula, patients were classified into high- and low-risk groups using the median risk score as the cut-off value. We assessed the differences in survival using the Kaplan–Meier method. Receiver operating characteristic (ROC) curves were constructed and the area under the curve (AUC) was calculated using the R package “survival ROC.” We then visualized the expression heat map, risk score distribution, and survival time associated with the risk scores. To assess whether the FR risk model was an independent prognostic factor, univariate and multivariate Cox regression analyses were performed for risk scores and clinical variables (age: stage: T, N, and M: and grade). Statistical significance was set at p < 0.05.

Nomogram model establishment

Combining the prognostic model and the clinical characteristics of patients with CESC, including age, grade, and T, N, and M stages, a nomogram was constructed to better predict prognosis. Based on the nomogram, the total score for each patient was calculated. ROC curves were constructed and AUC values were used to verify the accuracy of the model using the R package “survival ROC.”

Mutation analysis

SNP-related data of processed CESC were downloaded, and mutated genes were obtained from the downloaded SNP CESC sample VarScan data. The top 20 mutation frequency genes were then selected for comparing and determining the differences between the mutated genes of the low- and high-risk groups of patients: the mutation landscape was visualized using “ComplexHeatmap” in R. The relationship between the risk and the tumor mutational burden (TMB) scores was analyzed, and a risk score survival analysis was obtained by redistributing the TMB score.

Prediction of drug sensitivity

We estimated the therapeutic response of cancer cells to known chemotherapeutic agents using the “pRRophetic” package. We simulated the content of each chemotherapeutic agent in CESC specimens by constructing a ridge regression model according to the Genomics of Drug Sensitivity in Cancer (GDSC) database (www.cancerRxgene.org) and transcriptome data, and calculated the half-maximal inhibitory concentration (IC50).

Experimental validation

To validate the identified methylated FRGs, we conducted IHC analysis for three prognostic genes in 20 tumor tissues and adjacent non-cancerous tissues. Specimens were obtained from patients with CESC who underwent primary surgical resection between 2019 and 2022 at the First Hospital of Shanxi Medical University. This study was approved by the Clinical Research Ethics Committee of First Hospital of Shanxi Medical University. IHC was performed using antibodies against AQP3 (1:500; Ab215853; Abcam, Cambridge, UK), MGST1 (1:1000; Ab131059; Abcam), and TFRC (1:500; Ab214039; Abcam) in 4-μm thick paraffin-embedded sections according to the manufacturer’s instructions. Two independent pathologists blinded to the clinical information evaluated and randomly selected five regions and scored the immunostaining.

Statistical analysis

All statistical analyses were performed using R version 4.1.2. Univariate and multivariate Cox regression analyses were performed to identify independent prognostic factors for OS. Survival analysis was conducted using the Cox univariate regression analysis. The predictive performance of the OS prognostic model was assessed by ROC analysis and calculation of the AUC. A correlation analysis of the FRGs was performed using the Pearson correlation method. The Wilcoxon test was used to compare tumor-infiltrating cells. Data were visualized using the “ggplot2,” “pheatmap,” and “forestplot” packages. Statistical significance was set p < 0.05.

Results

Identification of FR DEGs in CESC

The flowchart of the study is shown in Fig. 1. We identified 2649 DEGs in the TCGA dataset, including 1162 upregulated and 1487 downregulated genes (Fig. 2a). Besides, we identified 483 FRGs in the FerrDb database. After intersecting the DEGs and FRGs, we obtained 77 FR DEGs (Fig. 2b), and generated a heat map based on their expression (Fig. 2c).

Fig. 1.

Fig. 1

Open in a new tab

The flowchart of the study

Fig. 2.

Fig. 2

Open in a new tab

Differentially expressed gene (DEG) analysis in patients with CESC. a A volcano plot showed differentially expressed genes (DEGs) in TCGA. b Venn diagram showing ferroptosis-related DEGs. c Heat map of ferroptosis-related DEGs in normal and tumor samples

Enrichment analysis of FR DEGs

To gain insight into the potential functional role of FR DEGs, we performed enrichment analyses of known biological functions and pathways. Kyoto Encyclopedia of Genes and Genomes (KEGG) analysis showed significant enrichment of many related pathways, including fluid shear stress, atherosclerosis, and ferroptosis (Fig. 3a). Regarding biological processes (BP), FR DEGs were significantly enriched in the response to metal ions, regulation of epithelial cell proliferation, and maintenance of location. In terms of cellular components (CC), the analysis indicated that DEGs were predominantly localized in the basal part of the cell, basal plasma membrane, and basolateral plasma membrane. Regarding molecular functions, the majority of DEGs were enriched in iron ion binding and ferrous iron binding (Fig. 3b).

Fig.3.

Fig.3

Open in a new tab

KEGG and GO analysis for ferroptosis-related differentially expressed genes a KEGG and b GO

Identification of ferroptosis molecular subtypes in CESC

Using univariate Cox analysis, we identified 21 FRGs associated with prognosis, 12 of which could be considered protective factors and 9 that could be considered risk factors (Fig. 4a). Moreover, correlation analyses showed that the expression of most genes was correlated with FRG expression in CESC tissues (Fig. 4b). Furthermore, we performed a cluster analysis based on the 21 FR prognostic genes, and found that patients with CESC clustered into two subgroups with good internal consistency and excellent stability (Fig. 4c–f). A survival analysis showed that patients from cluster1 had better outcomes than those from cluster2 (Fig. 4g), verifying the results from our cluster analysis.

Fig. 4.

Fig. 4

Open in a new tab

Patients were divided into two subtypes based on the ferroptosis-related DEGs associated with OS (a). The prognostic analysis for 21 ferroptosis-related genes in the TCGA cohort of cervical cancer using a univariate Cox regression model. Hazard ratio > 1 represented risk factors for survival and hazard ratio < 1 represented protective factors for survival (b). Correlation analysis of the 21 ferroptosis-related genes based on their expression in CESC tissues. Red indicates a positive correlation; blue indicates a negative correlation (c). Patients with CESC were stratified into two subtypes according to the consensus clustering matrix (k = 2). d Consensus CDF for k = 2–9. e Increment in area under CDF curve for k = 2–9. f The tracking plot for k = 2–9. g Kaplan–Meier overall survival (OS) curves of the two clusters. h Heatmap visualization of differentially expressed genes in two subtypes (clusters1-2), along with age, tumor grade, and TNM stage

We then conducted a correlational predictive analysis of cancer treatment in these two clusters and visualized the clinicopathological and genetic features of the two clusters using a heat map. As shown in Fig. 4h, cluster1 exhibited a notably lower cancer grade compared to that of cluster2 (p < 0.05). Moreover, our findings indicated an elevated expression of MIR9-3HG, TP63, AQP3 and other genes in patients from clusters1, thus affirming the credibility of our cluster analysis and further implying the association between the expression of these genes and a favorable prognosis in patients with CESC.

Differences in pathway enrichment and immune infiltration between patients with different ferroptosis molecular subtypes

To explore the differences in biological behaviors between the two ferroptosis subtypes patient groups, we conducted a GSVA enrichment analysis (Fig. 5a). The cluster1 subtype exhibited significant enrichment in pyrimidine metabolism, dilated cardiomyopathy, and focal adhesion, whereas the cluster2 subtype showed prominent enrichment in proximal tubule bicarbonate reclamation, the GnRH signaling pathway, linoleic acid metabolism, and aldosterone-regulated sodium reabsorption. By analyzing the relationship between different ferroptosis subtypes and tumor immune infiltration, we further explored the potential molecular mechanisms by which ferroptosis molecular subtypes influence the progression of CESC. Using the CIBERSORT algorithm, we identified 22 immune cell types and composed a heatmap to visualize the relative abundance of immune-infiltrating cell subtypes (Fig. 5b). We observed differences in immune infiltration between macrophage subsets of the ferroptosis subtypes. CD4 memory resting T cells, regulatory T cells (Tregs), resting dendritic cells and resting mast cells were significantly enriched in cluster1, while CD4 memory activated T cells, resting NK cells, and M0 macrophages were significantly enriched in cluster2(Fig. 5c–j). These results were further validated using ssGSEA. Immature dendritic cells, macrophages, parainflammation, plasmocytoid dendritic cells, type I IFN response and type II IFN response were significantly enriched in cluster1(Fig. 5k, m–r). Hence, we postulated that the adverse prognosis of the cluster2 subtype could be attributed to activation of the extracellular matrix in the tumor microenvironment. Given that the two clusters differed significantly in terms of immune infiltration, we assessed the expression of immune checkpoints in the different subgroups. Immune checkpoints TNFRSF25, TNFRSF18, BTLA, CD200R1, CD40LG, TNFSF9, TNFSF18, and CD44 showed elevated activity in cluster1, whereas CD276 and PDCD1LG2 exhibited increased activity in cluster2, highlighting significant variations in immune checkpoints between subtypes. (Fig. 5l).

Fig. 5.

Fig. 5

Open in a new tab

Differences of biological features and immune profiles of the two subgroups. a Differences in biological behavior between two subtypes ranked by t value of GSVA score. b The percent of 22 types of tumor-infiltrating immune cell in cluster1 and cluster 2. c–j Comparison of the proportions of tumor-infiltrating immune cells between cluster1 and cluster2. k, m–r Comparison of immune cell infiltration among in cluster 1 and cluster 2 subtype using ssGSEA analysis. l Differential expression of immune checkpoints between the two clusters (GSVA, gene set variation analysis)

Construction of a risk model and identification of a FR prognostic signature

DNA methylation is an important epigenetic process that affects pretranscriptional genetic imprinting, genomic stability, and cell fate. To demonstrate the potential mechanism of aberrant regulation of the 21 FRGs in CESC tissues, a methylation expression level analysis was conducted using UALCAN. The results revealed that the mean methylation levels of AQP3, GCH1, MGST1, TFRC, and PTPN6 were significantly lower in CESC tissues than in normal peritumoral tissues (Fig. 6a–e).

Fig. 6.

Fig. 6

Open in a new tab

Constructing a prognostic model and analyzing the independent prognostic potential. a–e The methylation levels of AQP3, GCH1, MGST1, TFRC, and PTPN6 in CESC and peri-tumor tissues were determined using the UALCAN database. f Optimal parameter (λ) selected in the LASSO Cox regression model based on the minimum criteria. g The LASSO coefficient profiles of the five methylation-related ferroptosis genes signature. h Kaplan–Meier curves of survival status and survival time. i, j Distribution of overall survival status and risk scores. k Heat map of mRNA expression of three selected methylation-related ferroptosis genes in high- and low-score samples. l–m Univariate and multivariate Cox regression confirmed that the risk score was independent prognostic factor. n ROC curves showed the potential of prognostic methylation-related ferroptosis genes signature in predicting 1, 3, and 5 years overall survival (OS)

We used LASSO regression analysis to select FRGs optimally associated with survival; then multivariate Cox regression analysis was utilized to establish a prognostic model of CESC using the selected genes: AQP3, MGST1 and TFRC (Fig. 6f, g). The risk score for each sample was calculated as follows: (-0.1365 × expression level of AQP3) + (0.1667 × expression level of MGST1) + (0.2842 × expression level of TFRC). Patients were divided into high- and low-score groups according to the median risk score. Kaplan–Meier survival curves were plotted with the log-rank test to explore the relationship between the risk score and survival. The results showed that the high-risk patients had significantly shorter survival rates than the low-risk patients (Fig. 6h). Together, the risk score and OS status showed that the number of patients who had died directly correlated with risk scores. (Fig. 6i, j). Additionally, differences in AQP3, MGST1 and TFRC expression in the high- and low-risk groups were visualized through heat maps (Fig. 6k). Collectively, these results suggest that AQP3, MGST1 and TFRC FRGs could be used as prognostic signatures for patients with CESC.

We further evaluated the performance of the three-gene FR model in predicting CESC prognosis. To better predict the OS of patients with CESC, the risk score and clinical characteristics, such as age and grade, were included in the univariate and multivariate Cox regression analyses. Cox regression analysis showed that the risk score was an independent prognostic factor for patients with CESC (p < 0.05) (Fig. 6l, m). ROC curves were used to assess the prognostic power of FR signatures. The AUCs were 0.729, 0.785, and 0.631 for 1-year, 2-year, and 3-year survival, respectively (Fig. 6n). A correlation analysis between the risk score and the clinical characteristics of patients with CESC was then conducted (Fig. 7a–e). The results showed that the risk score expression was higher in the T3-4 stage compared to that in the T1-2 stage. Besides, the risk score was not significantly correlated with age, grade, M stage, or N stage. A nomogram incorporating the clinicopathological characteristics and risk scores was developed to predict the probability of survival in patients with CESC (Fig. 7f). Similarly, ROC curve analyses showed that the AUCs of the nomogram at 1, 3, and 5 years were 0.721, 0.791, and 0.763, respectively (Fig. 7g). These results indicated that the hybrid nomogram had a consistent and robust ability to predict the prognosis of patients with CESC.

Fig. 7.

Fig. 7

Open in a new tab

Relationship between the risk groups and clinical features, and establishment and validation of the nomogram. a Correlation between risk score and age. b Correlation between risk score and grade. c–e Correlation between risk score and TNM stage. f Nomogram predicting the 1, 3, and 5 years overall survival in CESC patients. g The ROC curves of the nomogram estimate the prognostic value of the nomogram

FR risk model scores correlate with immune cell infiltration in CESC

To elucidate the differences in immune cells infiltration between the high- and low-risk groups, we compared their stromal, immune, tumor purity, and ESTIMATE scores, and found significant differences among them (p < 0.5) (Fig. 8a–d). We applied the CIBERSORT algorithm to conduct immunological tests of the patients to provide a better understanding of their relationship with the FR risk model. The results showed that the immune cells that differed the most in abundancy between the high- and low-risk groups were M1 macrophages (Fig. 8h). Additionally, the risk score was significantly positively associated with CD4 naïve T cell, monocyte, and naïve B cell levels, and significantly negatively associated with CD8 T cell levels (Fig. 8e–g, i). A heat map was used to visualize the correlation between the three genes of the FR risk model and the immune microenvironment scores (Fig. 8j).

Fig. 8.

Fig. 8

Open in a new tab

The immune status of patients with CESC between high and low risk score groups. a–d Differences in ESTIMATEScore, StromalScore, ImmuneScore, and TumorPurity between high- and low-risk patients. e–g Correlation between risk score and immune cells. h Differential expression of immune cells in low- and high-risk groups. i Correlation between risk score and 22 kinds of immune infiltration cells. j Correlation between three genes expression of risk model and immune cell infiltration. Red squares indicate positive correlation, and blue squares indicate negative correlation; the deeper color squares indicate stronger correlations

Gene mutation analyses

The relationship between genetic variation in the risk score subtypes was determined by exploring the correlation between TMB and risk scores. We visualized the top 20 gene mutations in the low- and high-risk groups after sorting gene mutation frequencies. As shown in Fig. 9a, b, TTN, PIK3CA, and KMT2C exhibited the highest mutation frequencies. By analyzing the differences between the risk and TMB scores, we found that they had a low correlation (Fig. 9c). Furthermore, no significant differences were found between the TMB of the two groups and no survival time differences were found between the groups, either (Fig. 9d, e). When the scores were jointly analyzed, it was observed that the prognosis was significantly worse in the high-risk/low-TMB group than in the low-risk/high-TMB group (Fig. 9f). These data suggest that risk signatures may be related to somatic mutations that affect tumor progression.

Fig. 9.

Fig. 9

Open in a new tab

TMB and somatic mutation between the low- and high-risk groups. a High-risk score oncoPrint map. b Low-risk score oncoPrint map. c The correlations between the risk score and TMB. d Comparison of TMB between the high- and low-risk score groups. e The Kaplan–Meier curves for high and low TMB groups. f The Kaplan–Meier curves for patients stratified by TMB and risk score

Predicting the response to small drug molecules

To analyze the differences in resistance potential between the two risk groups, we used the “pRRophetic” package to compare the estimated IC50 levels of chemotherapeutics or inhibitors of the two groups. The results showed that the IC50s of BMS345541, erlotinib, KIN001-102, lapatinib, phenformin, WZ-1-84, XL-184, Z-LLNIe-CHO, and ZSTK474 were significantly higher in the high-risk group than in the low-risk group, whereas that of pyrimethamine was significantly lower in the high-risk group than in the low-risk group. These data suggested that the patients in the high-risk group may benefit from pyrimethamine treatment. Conversely, BMS345541, erlotinib, KIN001-102, lapatinib, phenformin, WZ-1-84, XL-184, Z-LLNIe-CHO, and ZSTK474 may be candidates for treating patients in the low-risk group (Fig. 10a–j).

Fig. 10.

Fig. 10

Open in a new tab

Prediction of the drug sensitivity in patients with CESC. a–j The boxplot shows the mean difference in the estimated IC50 values for 10 representative drugs (BMS345541, Erlotinib, KIN001-102, Lapatinib, Phenformin, Pyrimethamine, WZ-1-84, XL-184, Z-LLNIe-CHO, ZSTK474) between the two risk groups

Histopathologic validation of the three methylation-related FRGs

IHC results confirmed that AQP3 and TFRC proteins were highly expressed in CESC tissues (77.5% and 67.5%, respectively), while the expression of MGST1 in tumor and tumor-adjacent normal tissues was not significantly different (Fig. 11). Interestingly, this was consistent with the results of the bioinformatics analysis, except for those of the MGST1 protein, which was reported as highly expressed in tumor-adjacent tissues according with data from the TCGA database. This may be linked to variations in the study populations and our limited sample size, which needs to be expanded in future.

Fig. 11.

Fig. 11

Open in a new tab

Immunohistochemistry showing the expression of a AQP3, b MGST1, and c TFRC in CESC and normal tissues

Discussion

Ferroptosis has emerged as a potential tumor-suppression mechanism in various cancers, including CESC. In this study, we systematically investigated the expression of FRGs in CESC using TCGA and GTEx cohorts. We found 77 FRGs that were differentially expressed in tumor and peritumor tissues. Pathway analysis revealed an enrichment of FRGs in CESC, suggesting their potential role in the disease. Moreover, we identified that 21 of these 77 FRGs were significantly associated with OS in patients with CESC. Further cluster analysis divided the patients into two clusters, each with distinct characteristics and potentially different treatment responses and prognoses.

Epigenetic alterations, particularly DNA methylation, play crucial roles in tumorigenesis (Nagase and Ghosh 2008; Woloszynska-Read et al. 2008; Reiter et al. 2019). Differentially methylated genes are potential cancer driver genes and may serve as diagnostic and prognostic biomarkers (Manolakos et al. 2014). Hence, we used the UALCAN database to investigate the DNA methylation patterns of the 21 prognosis-related FRGs and identified 5 methylation-related ferroptosis genes (AQP3, GCH1, MGST1, TFRC, and PTPN6). Among them, AQP3, MGST1 and TFRC were used to construct a novel methylation-related FRGs risk model using LASSO and multivariate Cox regression analyses. Although these three methylation-related ferroptosis genes have been reported to be dysregulated in cancers and other disorders, their exact biological roles in CESC have not been completely elucidated.

AQP3 (aquaporin3) is an aquaglyceroporin that transports water and small molecules such as glycerol. Previous studies have confirmed that AQP3 plays an important role in the progression and metastasis of a variety of cancers (Wang et al. 2015), including lung cancer (Machida et al. 2011), colon cancer (Li et al. 2013), pancreatic ductal adenocarcinoma (Direito et al. 2017), and hepatocellular carcinoma (Guo et al. 2013). In our study, AQP3 was demonstrated to be a protective factor with high expression in tumor tissues. MGST1 (microsomal glutathione S-transferase 1) is a membrane-bound transferase involved in oxidative stress regulation (Morgenstern et al. 2011). A previous study found that MGST1 expression activation helps pancreatic ductal adenocarcinoma cells resist ferroptosis, both in vitro and in vivo (Kuang et al. 2021). This finding is consistent with our results. We found that high MGST1 expression was associated with poor prognosis in patients with CESC. TFRC (transferrin receptor) encodes a classical transferrin receptor that is necessary for cellular iron uptake (Yuan et al. 2020). It is upregulated in pancreatic cancer and breast cancers, and is associated with poor survival rates and tumor immunology, suggesting that targeting TFRC may be a potential anticancer treatment (Chen et al. 2021; Yang et al. 2022a, b). In our study, high TFRC expression in tumor tissues was associated with worse prognosis of patients with CESC. In our study, AQP3, MSGT1 and TFRC were confirmed to be differentially expressed in tumor and normal tissues, highlighting their importance in CESC and ferroptosis regulation.

The predictive model based on the methylation-related FRGs effectively predicted the outcomes of patients with CESC. Patients were categorized into low- and high-risk groups based on their risk score values (high risk indicated a worse prognosis). Cox univariate and multivariate analyses showed that the risk score was as an independent prognostic factor for CESC. The nomogram, which incorporated clinical information, facilitated the calculation of the survival rate. Notably, AUC values of 0.721, 0.791 and 0.763 for predicting 1-, 3-, and 5-year survival, respectively, indicated that the model had a robust discriminatory ability for predicting patient survival.

Tumor-infiltrating immune cells(TILs) exert anti-tumor effects by releasing cytokines that facilitate ferroptosis in tumor cells (Gao et al. 2022). Previous studies indicated a close correlation between the efficacy of immunotherapies and ferroptosis in various cancers (Liao et al. 2022). TILs can be categorized into two groups: immunosuppressive and anti-tumor. Tregs, monocytic myeloid-derived suppressor cells and polymorphonuclear MDSCs are the primary immunosuppressive cell types associated with an unfavorable prognosis in most solid cancers. In contrast, the major anti-tumor immune cells include cytotoxic CD8 + T cells, NK cells, and Th1 CD4 + T cells: they are typically associated with a favorable prognosis (Gorvel and Olive 2023). Using CIBERSORT technology, we identified two subgroups: cluster1 (with a better prognosis) and cluster2. Cluster1 cells showed higher levels of CD4 memory resting T cells, Tregs, resting dendritic cells, and mast cells; whereas cluster2 cells displayed elevated levels of CD4 memory activated T cells, resting NK cells, and M0 macrophage. CD8 + T cells, which are essential for anti-tumor immunity (Imbert et al. 2020), exhibited a negative correlation with the risk score in the high-risk population, potentially explaining their reduced effectiveness.

Research has demonstrated that immune checkpoint inhibitors enable the host immune system to identify and eliminate tumor cells (Cristescu et al. 2018). Immune checkpoint inhibitors have demonstrated remarkable efficacy in cervical cancer (Minion and Tewari 2018); however, response rates vary among patients. Specific immune checkpoint genes, such as CD276 and PDCD1LG2, were significantly upregulated in cluster2, suggesting opportunities to enhance immune responsiveness and improve outcomes. Notably, different tumor subtypes may respond differently to immunotherapy (Kodach and Peppelenbosch 2021).

Recent studies suggest that genetic variation correlates with immunotherapy responses (Burr et al. 2017). Significant disparities in risk scores and mutation data were observed between high- and low-risk patients at the transcriptional level. In our study, PIK3CA gene variation was notably elevated in the low-risk group, whereas that of TTN was significantly elevated in the high-risk group. Moreover, TMB has been linked to improved immunotherapy response (Eroglu et al. 2018; Hellmann et al. 2019), showing lower subsequent survival rates with increased TMB incidence. Additionally, the cBioPortal database revealed a high frequency of mutations in the TFRC model gene in CESCs, with amplification being the most common. In summary, there is a complex interaction between immune ferroptosis and tumor mutations.

The survival of patients with CESC varies between the high-risk and low-risk groups, as a result of their response to chemotherapy. Patients with advanced or recurrent CESC tend to have longer survival after chemotherapy. Insensitive patients may require a change in the treatment strategy to enhance its efficacy; however, relevant clinical information is lacking. We analyzed the drug sensitivity and identified pyrimethamine as a potential therapeutic candidate for high-risk populations. In addition, erlotinib, KIN001-102, lapatinib, phenformin, WZ-1-84, XL-184, and Z-LLNIe-CHO are promising therapeutic agents for low-risk populations.

In conclusion, our study contributes to the growing understanding of ferroptosis in CESC and highlights the significance of methylation-related ferroptosis genes in prognostication. The identification of a three-gene methylation-related FRGs risk model provides a promising avenue for further investigation and potential therapeutic targeting.

This study had certain limitations. First, the research data were sourced from public databases, namely TCGA and GTEx, leading to some information being either unavailable or incomplete. Consequently, additional prospective studies are imperative to elucidate the clinical significance of methylation-related ferroptosis gene signatures. Second, further validation of our signature using an expanded sample size is required. Finally, a more comprehensive understanding of the specific molecular mechanisms underlying the methylation-related ferroptosis gene signature in the pathogenesis of CESC requires in-depth molecular biology experiments. In general, our retrospective study needs to be confirmed using further experimental and clinical data.

Conclusion

In this study, we divided 21 prognosis-related genes into two distinct subgroups based on their characteristics and successfully established a robust prognosis prediction model centered on three methylation-related ferroptosis genes, demonstrating their independent predictive capabilities. Our primary objectives were to predict the survival status, understand the immunological environment, and gauge the immunotherapeutic response in patients with CESC, thereby offering valuable insights for the development of novel treatment strategies. Additionally, we explored tumor mutation status within the risk model. Gene and functional enrichment analyses were performed to investigate the intricate relationships among the model genes. The lack of validation in an external cohort is a limitation of this study, and further studies are required to elucidate its clinical value. Future studies should focus on validating the prognostic model in external cohorts to ascertain its clinical efficacy. This study may pave the way for the exploration of novel avenues to understand the molecular mechanisms and develop targeted therapies for CESC.

Acknowledgements

We thank all quoted authors for their contributions to this work.

Author contributions

LY contributed to the study conception and design. Material preparation, data collection and analysis were performed by LY and ZG. YG and GL performed the experiment. LY and ZL wrote the first draft of the manuscript altogether and all authors commented on previous versions of the manuscript. All authors read and approved the final manuscript.

Funding

The research leading to these results has received no specific grant.

Data availability

The data that support the findings of this study are available on request from the corresponding author Ping Liu, upon reasonable request.

Declarations

Conflict of interest

The authors have no relevant financial or non-financial interests to disclose.

Ethical approval

The study was approved by the Clinical Research Ethics Committee of the first Hospital of Shanxi Medical University.

Consent for publication

Freely and informed consent was obtained from all authors to participate in the study.

Footnotes

Publisher's Note

Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Lijun Yu and Zhenwei Gao contributed equally to this work.

References

  1. Bogdan AR, Miyazawa M, Hashimoto K, Tsuji Y (2016) Regulators of iron homeostasis: new players in metabolism, cell death, and disease. Trends Biochem Sci 41(3):274–286 [DOI] [PMC free article] [PubMed] [Google Scholar]
  2. Burr ML, Sparbier CE, Chan YC, Williamson JC, Woods K, Beavis PA, Lam EYN, Henderson MA, Bell CC, Stolzenburg S, Gilan O, Bloor S, Noori T, Morgens DW, Bassik MC, Neeson PJ, Behren A, Darcy PK, Dawson SJ, Voskoboinik I, Trapani JA, Cebon J, Lehner PJ, Dawson MA (2017) CMTM6 maintains the expression of PD-L1 and regulates anti-tumour immunity. Nature 549(7670):101–105 [DOI] [PMC free article] [PubMed] [Google Scholar]
  3. Chandrashekar DS, Bashel B, Balasubramanya SAH, Creighton CJ, Ponce-Rodriguez I, Chakravarthi B, Varambally S (2017) UALCAN: a portal for facilitating tumor subgroup gene expression and survival analyses. Neoplasia 19(8):649–658 [DOI] [PMC free article] [PubMed] [Google Scholar]
  4. Chen F, Fan Y, Hou J, Liu B, Zhang B, Shang Y, Chang Y, Cao P, Tan K (2021) Integrated analysis identifies TfR1 as a prognostic biomarker which correlates with immune infiltration in breast cancer. Aging (albany NY) 13(17):21671–21699 [DOI] [PMC free article] [PubMed] [Google Scholar]
  5. Cheng T, Shan G, Yang H, Gu J, Lu C, Xu F, Ge D (2022) Development of a ferroptosis-based model to predict prognosis, tumor microenvironment, and drug response for lung adenocarcinoma with weighted genes co-expression network analysis. Front Pharmacol 13:1072589 [DOI] [PMC free article] [PubMed] [Google Scholar]
  6. Cohen PA, Jhingran A, Oaknin A, Denny L (2019) Cervical cancer. Lancet 393(10167):169–182 [DOI] [PubMed] [Google Scholar]
  7. Cristescu R, Mogg R, Ayers M, Albright A, Murphy E, Yearley J, Sher X, Liu XQ, Lu H, Nebozhyn M, Zhang C, Lunceford JK, Joe A, Cheng J, Webber AL, Ibrahim N, Plimack ER, Ott PA, Seiwert TY, Ribas A, McClanahan TK, Tomassini JE, Loboda A, Kaufman D (2018) Pan-tumor genomic biomarkers for PD-1 checkpoint blockade-based immunotherapy. Science 362(6411):eaar3593 [DOI] [PMC free article] [PubMed] [Google Scholar]
  8. Direito I, Paulino J, Vigia E, Brito MA, Soveral G (2017) Differential expression of aquaporin-3 and aquaporin-5 in pancreatic ductal adenocarcinoma. J Surg Oncol 115(8):980–996 [DOI] [PubMed] [Google Scholar]
  9. Dixon SJ, Lemberg KM, Lamprecht MR, Skouta R, Zaitsev EM, Gleason CE, Patel DN, Bauer AJ, Cantley AM, Yang WS, Morrison B 3rd, Stockwell BR (2012) Ferroptosis: an iron-dependent form of nonapoptotic cell death. Cell 149(5):1060–1072 [DOI] [PMC free article] [PubMed] [Google Scholar]
  10. Eroglu Z, Zaretsky JM, Hu-Lieskovan S, Kim DW, Algazi A, Johnson DB, Liniker E, Ben K, Munhoz R, Rapisuwon S, Gherardini PF, Chmielowski B, Wang X, Shintaku IP, Wei C, Sosman JA, Joseph RW, Postow MA, Carlino MS, Hwu WJ, Scolyer RA, Messina J, Cochran AJ, Long GV, Ribas A (2018) High response rate to PD-1 blockade in desmoplastic melanomas. Nature 553(7688):347–350 [DOI] [PMC free article] [PubMed] [Google Scholar]
  11. Finotello F, Trajanoski Z (2018) Quantifying tumor-infiltrating immune cells from transcriptomics data. Cancer Immunol Immunother 67(7):1031–1040 [DOI] [PMC free article] [PubMed] [Google Scholar]
  12. Gao W, Wang X, Zhou Y, Wang X, Yu Y (2022) Autophagy, ferroptosis, pyroptosis, and necroptosis in tumor immunotherapy. Signal Transduct Target Ther 7(1):196 [DOI] [PMC free article] [PubMed] [Google Scholar]
  13. Gorvel L, Olive D (2023) Tumor associated macrophage in HPV(+) tumors: Between immunosuppression and inflammation. Semin Immunol 65:101671 [DOI] [PubMed] [Google Scholar]
  14. Guo X, Sun T, Yang M, Li Z, Li Z, Gao Y (2013) Prognostic value of combined aquaporin 3 and aquaporin 5 overexpression in hepatocellular carcinoma. Biomed Res Int 2013:206525 [DOI] [PMC free article] [PubMed] [Google Scholar]
  15. Hangauer MJ, Viswanathan VS, Ryan MJ, Bole D, Eaton JK, Matov A, Galeas J, Dhruv HD, Berens ME, Schreiber SL, McCormick F, McManus MT (2017) Drug-tolerant persister cancer cells are vulnerable to GPX4 inhibition. Nature 551(7679):247–250 [DOI] [PMC free article] [PubMed] [Google Scholar]
  16. Hassannia B, Vandenabeele P, Vanden Berghe T (2019) Targeting ferroptosis to iron out cancer. Cancer Cell 35(6):830–849 [DOI] [PubMed] [Google Scholar]
  17. Hellmann MD, Callahan MK, Awad MM, Calvo E, Ascierto PA, Atmaca A, Rizvi NA, Hirsch FR, Selvaggi G, Szustakowski JD, Sasson A, Golhar R, Vitazka P, Chang H, Geese WJ, Antonia SJ (2019) Tumor mutational burden and efficacy of nivolumab monotherapy and in combination with ipilimumab in small-cell lung cancer. Cancer Cell 35(2):329 [DOI] [PubMed] [Google Scholar]
  18. Imbert C, Montfort A, Fraisse M, Marcheteau E, Gilhodes J, Martin E, Bertrand F, Marcellin M, Burlet-Schiltz O, Peredo AG, Garcia V, Carpentier S, Tartare-Deckert S, Brousset P, Rochaix P, Puisset F, Filleron T, Meyer N, Lamant L, Levade T, Ségui B, Andrieu-Abadie N, Colacios C (2020) Resistance of melanoma to immune checkpoint inhibitors is overcome by targeting the sphingosine kinase-1. Nat Commun 11(1):437 [DOI] [PMC free article] [PubMed] [Google Scholar]
  19. Kim SW, Chun M, Ryu HS, Chang SJ, Kong TW, Lee EJ, Lee YH, Oh YT (2017) Salvage radiotherapy with or without concurrent chemotherapy for pelvic recurrence after hysterectomy alone for early-stage uterine cervical cancer. Strahlenther Onkol 193(7):534–542 [DOI] [PubMed] [Google Scholar]
  20. Kodach LL, Peppelenbosch MP (2021) Targeting the myeloid-derived suppressor cell compartment for inducing responsiveness to immune checkpoint blockade is best limited to specific subtypes of gastric cancers. Gastroenterology 161(2):727 [DOI] [PubMed] [Google Scholar]
  21. Kuang F, Liu J, Xie Y, Tang D, Kang R (2021) MGST1 is a redox-sensitive repressor of ferroptosis in pancreatic cancer cells. Cell Chem Biol 28(6):765-775.e765 [DOI] [PubMed] [Google Scholar]
  22. Li D, Li Y (2020) The interaction between ferroptosis and lipid metabolism in cancer. Signal Transduct Target Ther 5(1):108 [DOI] [PMC free article] [PubMed] [Google Scholar]
  23. Li A, Lu D, Zhang Y, Li J, Fang Y, Li F, Sun J (2013) Critical role of aquaporin-3 in epidermal growth factor-induced migration of colorectal carcinoma cells and its clinical significance. Oncol Rep 29(2):535–540 [DOI] [PubMed] [Google Scholar]
  24. Liao P, Wang W, Wang W, Kryczek I, Li X, Bian Y, Sell A, Wei S, Grove S, Johnson JK, Kennedy PD, Gijón M, Shah YM, Zou W (2022) CD8(+) T cells and fatty acids orchestrate tumor ferroptosis and immunity via ACSL4. Cancer Cell 40(4):365-378.e366 [DOI] [PMC free article] [PubMed] [Google Scholar]
  25. Machida Y, Ueda Y, Shimasaki M, Sato K, Sagawa M, Katsuda S, Sakuma T (2011) Relationship of aquaporin 1, 3, and 5 expression in lung cancer cells to cellular differentiation, invasive growth, and metastasis potential. Hum Pathol 42(5):669–678 [DOI] [PubMed] [Google Scholar]
  26. Manolakos A, Ochoa I, Venkat K, Goldsmith AJ, Gevaert O (2014) CaMoDi: a new method for cancer module discovery. BMC Genom 15(Suppl 10):S8 [DOI] [PMC free article] [PubMed] [Google Scholar]
  27. Manz DH, Blanchette NL, Paul BT, Torti FM, Torti SV (2016) Iron and cancer: recent insights. Ann N Y Acad Sci 1368(1):149–161 [DOI] [PMC free article] [PubMed] [Google Scholar]
  28. Minion LE, Tewari KS (2018) Cervical cancer–state of the science: From angiogenesis blockade to checkpoint inhibition. Gynecol Oncol 148(3):609–621 [DOI] [PMC free article] [PubMed] [Google Scholar]
  29. Morgenstern R, Zhang J, Johansson K (2011) Microsomal glutathione transferase 1: mechanism and functional roles. Drug Metab Rev 43(2):300–306 [DOI] [PubMed] [Google Scholar]
  30. Nagase H, Ghosh S (2008) Epigenetics: differential DNA methylation in mammalian somatic tissues. Febs J 275(8):1617–1623 [DOI] [PubMed] [Google Scholar]
  31. Newman AM, Liu CL, Green MR, Gentles AJ, Feng W, Xu Y, Hoang CD, Diehn M, Alizadeh AA (2015) Robust enumeration of cell subsets from tissue expression profiles. Nat Methods 12(5):453–457 [DOI] [PMC free article] [PubMed] [Google Scholar]
  32. Ooko E, Saeed ME, Kadioglu O, Sarvi S, Colak M, Elmasaoudi K, Janah R, Greten HJ, Efferth T (2015) Artemisinin derivatives induce iron-dependent cell death (ferroptosis) in tumor cells. Phytomedicine 22(11):1045–1054 [DOI] [PubMed] [Google Scholar]
  33. Qi X, Fu Y, Sheng J, Zhang M, Zhang M, Wang Y, Li G (2021) A novel ferroptosis-related gene signature for predicting outcomes in cervical cancer. Bioengineered 12(1):1813–1825 [DOI] [PMC free article] [PubMed] [Google Scholar]
  34. Qin W, He C, Jiang D, Gao Y, Chen Y, Su M, Yang Y, Yang Z, Cai H, Wang H (2022) Systematic construction and validation of a novel ferroptosis-related gene model for predicting prognosis in cervical cancer. J Immunol Res 2022:2148215 [DOI] [PMC free article] [PubMed] [Google Scholar]
  35. Reiter JG, Baretti M, Gerold JM, Makohon-Moore AP, Daud A, Iacobuzio-Donahue CA, Azad NS, Kinzler KW, Nowak MA, Vogelstein B (2019) An analysis of genetic heterogeneity in untreated cancers. Nat Rev Cancer 19(11):639–650 [DOI] [PMC free article] [PubMed] [Google Scholar]
  36. Small W Jr, Bacon MA, Bajaj A, Chuang LT, Fisher BJ, Harkenrider MM, Jhingran A, Kitchener HC, Mileshkin LR, Viswanathan AN, Gaffney DK (2017) Cervical cancer: a global health crisis. Cancer 123(13):2404–2412 [DOI] [PubMed] [Google Scholar]
  37. Sun X, Ou Z, Chen R, Niu X, Chen D, Kang R, Tang D (2016) Activation of the p62-Keap1-NRF2 pathway protects against ferroptosis in hepatocellular carcinoma cells. Hepatology 63(1):173–184 [DOI] [PMC free article] [PubMed] [Google Scholar]
  38. Sung H, Ferlay J, Siegel RL, Laversanne M, Soerjomataram I, Jemal A, Bray F (2021) Global cancer statistics 2020: GLOBOCAN estimates of incidence and mortality worldwide for 36 cancers in 185 countries. CA Cancer J Clin 71(3):209–249 [DOI] [PubMed] [Google Scholar]
  39. Wan S, Lei Y, Li M, Wu B (2022) A prognostic model for hepatocellular carcinoma patients based on signature ferroptosis-related genes. Hepatol Int 16(1):112–124 [DOI] [PubMed] [Google Scholar]
  40. Wang J, Feng L, Zhu Z, Zheng M, Wang D, Chen Z, Sun H (2015) Aquaporins as diagnostic and therapeutic targets in cancer: how far we are? J Transl Med 13:96 [DOI] [PMC free article] [PubMed] [Google Scholar]
  41. Woloszynska-Read A, Mhawech-Fauceglia P, Yu J, Odunsi K, Karpf AR (2008) Intertumor and intratumor NY-ESO-1 expression heterogeneity is associated with promoter-specific and global DNA methylation status in ovarian cancer. Clin Cancer Res 14(11):3283–3290 [DOI] [PMC free article] [PubMed] [Google Scholar]
  42. Xu T, Ding W, Ji X, Ao X, Liu Y, Yu W, Wang J (2019) Molecular mechanisms of ferroptosis and its role in cancer therapy. J Cell Mol Med 23(8):4900–4912 [DOI] [PMC free article] [PubMed] [Google Scholar]
  43. Yamaguchi H, Hsu JL, Chen CT, Wang YN, Hsu MC, Chang SS, Du Y, Ko HW, Herbst R, Hung MC (2013) Caspase-independent cell death is involved in the negative effect of EGF receptor inhibitors on cisplatin in non-small cell lung cancer cells. Clin Cancer Res 19(4):845–854 [DOI] [PMC free article] [PubMed] [Google Scholar]
  44. Yang Q, Al-Hendy A (2022) The regulatory functions and the mechanisms of long non-coding RNAs in cervical cancer. Cells 11(7):1149 [DOI] [PMC free article] [PubMed] [Google Scholar]
  45. Yang C, Li J, Guo Y, Gan D, Zhang C, Wang R, Hua L, Zhu L, Ma P, Shi J, Li S, Su H (2022a) Role of TFRC as a novel prognostic biomarker and in immunotherapy for pancreatic carcinoma. Front Mol Biosci 9:756895 [DOI] [PMC free article] [PubMed] [Google Scholar]
  46. Yang L, Chen Z, Liu Y, Wang X, Li J, Ye Q (2022b) Immunization combined with ferroptosis related genes to construct a new prognostic model for head and neck squamous cell carcinoma. Cancers (basel) 14(17):4099 [DOI] [PMC free article] [PubMed] [Google Scholar]
  47. Yee PP, Wei Y, Kim SY, Lu T, Chih SY, Lawson C, Tang M, Liu Z, Anderson B, Thamburaj K, Young MM, Aregawi DG, Glantz MJ, Zacharia BE, Specht CS, Wang HG, Li W (2020) Neutrophil-induced ferroptosis promotes tumor necrosis in glioblastoma progression. Nat Commun 11(1):5424 [DOI] [PMC free article] [PubMed] [Google Scholar]
  48. Yoshihara K, Shahmoradgoli M, Martínez E, Vegesna R, Kim H, Torres-Garcia W, Treviño V, Shen H, Laird PW, Levine DA, Carter SL, Getz G, Stemke-Hale K, Mills GB, Verhaak RG (2013) Inferring tumour purity and stromal and immune cell admixture from expression data. Nat Commun 4:2612 [DOI] [PMC free article] [PubMed] [Google Scholar]
  49. Yuan Z, Zhu H, Zeng P, Yang S, Sun S, Yang C, Liu J, Zhou X (2020) Testing and controlling for horizontal pleiotropy with probabilistic Mendelian randomization in transcriptome-wide association studies. Nat Commun 11(1):3861 [DOI] [PMC free article] [PubMed] [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Data Availability Statement

The data that support the findings of this study are available on request from the corresponding author Ping Liu, upon reasonable request.

Articles from Journal of Cancer Research and Clinical Oncology are provided here courtesy of Springer

RESOURCES