HPV-ferroptosis related genes as biomarkers to predict the prognosis of cervical cancer

Abstract

Background

Ferroptosis can be used as a powerful predictor of cancer prognosis. HPV persistent infection is the main cause of cervical cancer, so it is very important to improve the prognosis of patients. Therefore, it is necessary to explore the value of HPV-ferroptosis related genes as prognostic biomarkers of cervical cancer patients.

Methods

In this study, differentially expressed HPV-ferroptosis related genes were obtained from GSE7410, HPV gene set crossed with iron death genes. Five HPV-ferroptosis related genes with prognostic features were finally identified: CYBB, VEGFA, CKB, EFNA1 and HELLS. Multifactorial Cox regression was applied to establish and validate the prognostic model, and drug susceptibility and immune infiltration analyses were also performed.

Results

The prognostic model was validated in the training set (TCGA) and validation set (GSE44001). Kaplan–Meier curves reveal significant differences in overall survival (OS) between high-risk and low-risk groups. Receiver operating characteristic (ROC) curve reflects the stability and accuracy of the prognostic model established in this study. In terms of immune function, T cell costimulation was better in the low-risk group than in the high-risk group (P < 0.01). The therapeutic effects of cisplatin, paclitaxel, docetaxel and cyclophosphamide, commonly used chemotherapy drugs for cervical cancer, are better in the high-risk group than in the low-risk group (P < 0.001).

Conclusion

HPV-ferroptosis related gene prognostic model not only has good stability and accuracy in predicting the prognosis of cervical cancer patients, but also has certain guiding value for clinicians in terms of drug sensitivity and immune microenvironment.

Introduction

Cervical cancer (CC) currently ranks 4th in incidence of female malignant tumors [1]. Globally, there are about 500,000 new cases per year, accounting for 5% of new cases of female malignant tumors, of which more than 80% are in developing countries. Meanwhile, more than 260,000 women in low- and middle-income countries die of cervical cancer every year [2]. In China, the number of new cases reach 131,500 per year, and the number of deaths from cervical cancer are about 53,000 per year, accounting for about 18.4% of all female malignant tumor deaths [3]. Human papillomavirus (HPV) is a highly heterogeneous DNA virus that causes intraepithelial neoplasia in the skin and mucous membranes, and persistent HPV infection can lead to precancerous cervical lesions, which may eventually develop into cancer [4]. There are more than 100 types of HPV genes identified. Low-risk HPV mainly includes HPV6 and HPV11; high-risk HPV mainly includes HPV16, HPV18, HPV31, HPV33, HPV35, HPV45, HPV52, HPV58, HPV59 and other types, among which HPV16 and HPV18, two viruses that are highly pathogenic and the most common [5]. HPV plays an important role in the pathogenesis of CC: it affects apoptosis, cell cycle, cell adhesion, and DNA repair mechanisms within host cells, and also activates the immune response [6, 7]. Integration of HPV viruses into the host genome tends to take place in the genomic region of transcription within the genome, a mechanism that is utilized by viruses to increase the expression of certain viral products, including the E6 and E7 viral oncogenes [8, 9]. In addition, the integration of HPV viruses is closely linked to the development of CC [10]. Although cervical cancer can significantly improve the prognosis of patients with early screening, early diagnosis and early treatment, some patients still develop recurrence and metastasis thus triggering a poorer prognosis. Therefore, new tumor biomarkers are needed for a systematic and comprehensive assessment of their prognosis.

Ferroptosis is an iron-dependent programmed cell death caused by the accumulation of lipid-based reactive oxygen species [11]. Recently, relevant studies have elucidated that ferroptosis can play multiple roles in bioregulatory and signaling pathways, leading to tumorigenesis and development [12, 13]. Jiang et al. [14] found that ferroptosis activated by ACSL4 reduced cervical cancer tumor size and decreased Hela cell activity. In addition, FBXW7 [15], G6PD [16] and TP53 [17] promoted ferroptosis in tumor cells; whereas CISD2 [18], GPX4 [19] and SLC7A11 [20] acted as inhibitory factors to prevent ferroptosis.Wang et al. [21] demonstrated that the enhancement of ferroptosis specific lipid peroxidation could rely on the help of CD8 + T cells to achieve the antitumor efficacy of immunotherapy. Therefore, the perspective of the relationship between ferroptosis and immune cell infiltration may provide new insights into the efficacy of immunotherapy.

Currently oncogenic viruses EB, HBV have been reported to be associated with ferroptosis [47, 48]. HPV, however, has been rarely reported, and this time, bioinformatics was used to initially explore the application of the whole gene of HPV and the ferroptosis gene in the prognostic value, immune infiltration, and drug sensitivity of cervical cancer. The aim of this study is to provide a feasible new strategy for clinicians to predict the prognosis and immune microenvironmental profile of cervical cancer patients.

Materials and methods

Sample and data collection

The research plan is illustrated in Fig. 1A. RNA transcriptome data and clinical information were acquired from the GEO and TCGA databases (http://www.ncbi.nlm.nih.gov/projects/geo/, https://portal.gdc.cancer.gov/). All data were transformed using log2 to ensure normalization. HPV genes were obtained from Table S4 studied by Iden Met al.A total of 170 HPV genes were integrated into the human genome [22] and12 HPV genes previously reported by our group as popular loci for integration into the human genome [46]. Four databases, FerrDb (http://www.zhounan.org/ferrdb/), NCBI-gene (https://www.ncbi.nlm.nih.gov/gene), MSigDB (http://www.gsea-msigdb.org/gsea/msigdb/), and Genecard (https://www.genecards.org/), provided a total of 416 ferroptosis genes.

Fig. 1.

Research program

Differential expression and functional enrichment analysis

For the TCGA-CESC and GEO data, the R package edgeR conducted differential analysis on normal and cancer samples. The threshold was set at |Log(FC)|> 1, p adj < 0.05, and intersections of the differential genes with the HPV genes and the ferroptosis genes were taken (the criterion for filtering overlapping genes was that they appeared in at least 2 datasets and one dataset was the ferroptosis gene set). Differentially expressed HPV-ferroptosis related genes were analyzed by GO enrichment analysis and KEGG pathway enrichment analysis using the R package clusterProfiler (version 3.14.3).

Prognostic model establishment and prognostic analysis

The R package (glmnet version 4.1.1) executed LASSO regression on the differentially expressed HPV-ferroptosis related genes to filter out redundant factors. Subsequently, univariate/multivariate Cox regression analysis determined the prognostic genes and constructed a Risk score prognostic model (The majority of literature calculates Risk score based on the weighting of the product of gene expression and its coefficients. This study employed multivariate Cox regression to develop a model in which Risk score was determined as the weighting of the product of gene expression and its coefficients.). High-risk groups (n = 153) and low-risk groups (n = 153) were categorized according to the median of the Risk score. The R packages survival ROC (version 1.0.3) and rms (version 6.2.0) analyzed one-year, three-year, five-year survival prognoses and prognostic risk performance.

$$riskscore=\sum _{i=1}^n\beta i\times \text{Exp}(i)$$

Clinicopathological features and immune infiltration analysis

The correlation between the survival Risk score, constructed by HPV-ferroptosis related genes, and clinicopathological characteristics was assessed. At the same time, immune infiltration analysis was performed using ssGSEA algorithm to obtain the difference between different immune cells and immune functions in high and low Risk score groups.

Statistical analysis

Differential analysis of normal and cancer samples was conducted using the R package edgeR, with a threshold of |Log(FC)|> 1 and p adj < 0.05. The R package implemented LASSO regression and univariate/multivariate Cox regression analysis on differentially expressed FRGs. The R packages survminer (version 0.4.9) and survival ROC (version 1.0.3) performed KM and ROC curve analysis to predict the survival prognosis of patients with cervical cancer. Correlation analysis of survival Risk scores constructed from HPV-ferroptosis related genes and clinicopathological characteristics utilized univariate and multivariate Cox regression analysis. The test level was α = 0.05, and a difference was considered statistically significant with P < 0.05.

Results

Screening and functional analysis of HPV-ferroptosis related genes

GSE7410 obtained 2142 differential genes by differential gene analysis (Fig. 2A). 54 differentially expressed HPV-ferroptosis related genes were obtained by intersecting GSE7410, HPV gene set and ferroptosis genes (Fig. 2B). Finally, the 54 HPV-ferroptosis related genes were subjected to GO enrichment analysis and KEGG pathway enrichment analysis, in which GO enrichment analysis was performed by biological process (BP), cellular component (CC) and molecular function (MF). The analysis results showed that the gene set was significantly enriched in hypoxia response, blood particles, and trivalent iron binding (Fig. 2C); KEGG pathway enrichment analysis showed that the gene set was significantly enriched in MicroRNAs, HIF-1 signaling pathway and ferroptosis (Fig. 2D).

Fig. 2.

Overview of HPV-ferroptosis related genes characterization A GSE7410 differentially expressed genes, red represents up-regulated genes and blue represents down-regulated genes. B Wayne plots of differentially expressed genes, HPV genes, and ferroptosis genes. C GO enrichment analysis of HPV-ferroptosis related genes. D KEGG enrichment analysis of HPV-ferroptosis related genes

Establishment and prognostic analysis of HPV-ferroptosis related genes prognostic model in cervical cancer

TCGA-CESC data were used as the training set. Firstly LASSO regression was performed on 54 HPV-ferroptosis related genes to screen out redundant factors, and a total of 11HPV-ferroptosis related genes were obtained. Secondly single/multifactorial Cox regression analysis obtained 5 HPV-ferroptosis related genes with independent prognostic effects (Table 1). Finally prognostic modeling was established based on the above 5 genes. High and low groups were categorized according to the median Risk score. The scatter plot showed the survival time and the occurrence of death events in patients grouped according to high and low risk, with more death events in the high-risk group than in the low-risk group (Fig. 3A). In the training set, the KM curve suggests that the OS of patients in the low-risk group is significantly longer than that of patients in the high-risk group (Fig. 3B); the ROC curve analyzes the OS at one-, three-, and five- years (AUC values of 0.751, 0.747, and 0.801, respectively) (Fig. 3C). In summary, the model has good stability and accuracy in predicting patient prognosis.

Table 1.

Univariate and multivariate Cox regression analysis of HPV-ferroptosis related genes

Characteristics	Total(N)	Univariate analysis		Multivariate analysis
		Hazard ratio (95% CI)	P value	Hazard ratio (95% CI)	P value
TFR2	306	1.331 (0.837–2.119)	0.227	1.097 (0.640–1.880)	0.737
CYBB	306	1.696 (1.061–2.711)	0.027	2.072 (1.217–3.526)	0.007
VEGFA	306	0.463 (0.284–0.757)	0.002	0.361 (0.208–0.625)	< 0.001
TP63	306	1.002 (0.629–1.594)	0.995	0.948 (0.567–1.587)	0.840
CKB	306	1.797 (1.118–2.890)	0.016	1.791 (1.053–3.045)	0.032
G0S2	306	0.745 (0.468–1.187)	0.215	0.898 (0.552–1.459)	0.663
PGK1P2	306	1.299 (0.817–2.065)	0.269	1.256 (0.768–2.054)	0.364
HP	306	0.718 (0.450–1.147)	0.166	0.657 (0.407–1.060)	0.085
C11orf86	306	0.949 (0.597–1.508)	0.824	1.520 (0.899–2.570)	0.119
EFNA1	306	0.497 (0.309–0.800)	0.004	0.579 (0.354–0.946)	0.029
HELLS	306	1.911 (1.188–3.074)	0.008	2.011 (1.115–3.628)	0.020

Bold italic and italic values indicate the statistical differences

Fig. 3.

Establishment and assessment of HPV-ferroptosis related genes prognostic modeling. A Curve scatter plots and cumulative scatter plots of the risk of survival death events for the assessment of survival model efficacy in the training set. B Kaplan–Meier curves showing significant differences in OS between the high-risk group and the low-risk group in the training set. C Time-dependent ROC curves for prediction of one-, three-, and five-year survival. D One-, three-, and five-year column line plots used to predict OS for cervical cancer. E Calibration curves showing concordance between predicted and observed one-, three-, and five-year OS. F Decision curve analysis of the prognostic model's net gain at one-, three-, and five-year in the training set

Prognostic analysis of clinicopathological features by HPV-ferroptosis related genes prognostic model in cervical cancer

Clinicopathological features of the study were examined using univariate/multivariate Cox regression analyses. According to univariate analysis, clinical stage, TNM stage, and Risk score were significant factors predicting patient prognosis. In contrast, multivariate analysis revealed that Risk score was the sole independent factor predicting patient prognosis (Table 2). To ensure accurate patient prognosis, a nomogram incorporating various clinicopathologic parameters was generated (Fig. 3D). Additionally, DCA and calibration curves (Fig. 3E-F) demonstrated the model's role in assessing patient outcomes. In summary, the model could be employed as a novel and powerful tool for predicting patient prognosis.

Table 2.

Univariate/multivariate Cox regression analysis of clinicopathological characteristics of TCGA-CESC

Characteristics	Total(N)	Univariate analysis		Multivariate analysis
		Hazard ratio (95% CI)	P value	Hazard ratio (95% CI)	P value
Age(> 50 vs ≤ 50)	306	1.017 (0.999–1.035)	0.060	1.004 (0.982–1.027)	0.711
T stage (T3&T4 vs T1&T2)	243	1.776 (0.396–7.969)	0.453	2.764 (0.499–15.304)	0.244
N stage(N1 vs N0)	195	0.773 (0.364–1.639)	0.502	0.832 (0.370–1.872)	0.657
M stage(M1 vs M0)	127	1.415 (0.835–2.397)	0.197	1.093 (0.571–2.094)	0.788
Clinical stage (Stage III&Stage IV vs Stage I&Stage II)	299	4.448 (2.393–8.266)	< 0.001	3.655 (1.637–8.160)	0.002
Histologic grade (G3&G4 vs G1&G2)	274	3.806 (0.822–17.625)	0.087	2.449 (0.485–12.358)	0.278
Risk group (High vs Low)	306	1.836 (1.373–2.455)	< 0.001	1.610 (1.145–2.265)	0.006

Bold italic and italic values indicate the statistical differences

Validation and prognostic efficacy analysis of prognostic model of HPV-ferroptosis related genes in cervical cancer

To validate the model's applicability, the GSE44001 dataset was used as the validation set. A prognostic model was developed based on the six aforementioned genes, which were divided into high and low groups according to median Risk score. Scatter plots of survival outcomes and survival time indicated that the high-risk group had more fatalities than the low-risk group (Fig. 4A). The validation set's KM curve showed that the OS of the low-risk group was longer than that of the high-risk group (P < 0.001, Fig. 4B). ROC curves were employed to analyze the OS at 1 year, 3 years, and 5 years, with AUC values of 0.629, 0.630, and 0.59, respectively (Fig. 4C). The study results also indicated that the model was stable and accurate in the validation set. Univariate/multivariate Cox regression analyses were then employed with the dataset to further validate the model's clinicopathologic characteristics. According to univariate analysis results, Risk score and IB2 were the primary prognostic factors. In multivariate regression analysis, Risk score was considered an independent predictor of the study's outcome (Table 3). A nomogram was also used to evaluate the model's value in assessing patient prognosis in the GSE44001 dataset (Fig. 4D). DCA and calibration curves (Fig. 4E, F) also demonstrated that the model had a consistent effect on patient prognosis assessment. In conclusion, the model's practicality and suitability for various datasets render it an ideal choice for determining cervical cancer patients' prognosis.

Fig. 4.

Validation of HPV-ferroptosis related genes prognostic model. A Scatterplot of curves assessing the efficacy of the survival model in the validation set and the cumulative scatterplot of the risk of surviving a death event. B Kaplan–Meier curves showing the significant difference in OS between the high-risk group and the low-risk group in the GSE44001. C Time-dependent ROC curves were used for the prediction of one-, three-, and five-year survival. D One-, three-, and five-year column line plots used to predict OS in cervical cancer. E Calibration curves showing concordance between predicted and observed one-, three-, and five-year OS. F Decision curve analysis validating the centralized prognostic model at one-, three-, and five-year net gain

Table 3.

Univariate and multivariate Cox regression analysis of clinicopathological characteristics of GSE44001 dataset

Characteristics	Total(N)	Univariate analysis		Multivariate analysis
		Hazard ratio (95% CI)	P value	Hazard ratio (95% CI)	P value
Stage	300
IB1	217	Reference
IA2	13	0.000 (0.000–Inf)	0.996	0.000 (0.000–Inf)	0.996
IB2	28	3.953 (1.807–8.651)	< 0.001	3.617 (1.642–7.964)	0.001
IIA	42	2.106 (0.932–4.758)	0.073	2.168 (0.960–4.897)	0.063
Risk group (High vs Low)	300	2.718 (1.439–5.134)	0.002	2.442 (1.318–4.524)	0.005

Bold italic and italic values indicate the statistical differences

Drug sensitivity analysis of HPV-ferroptosis related genes prognostic models

In order to understand the sensitivity of HPV-ferroptosis related genes prognostic model in different chemotherapeutic agents, we selected four commonly used chemotherapeutic agents (cisplatin, paclitaxel, docetaxel and cyclophosphamide) for cervical cancer. They were divided into high and low groups according to the median of Risk score score, and by analyzing them, we found that there were very obvious and significant differences between the above four chemotherapeutic agents in the signature high and low risk groups (P < 0.001, Fig. 5A). To further understand the correlation between the gene expression high and low groups included in signature and different chemotherapeutic drugs analysis revealed that none of these 5 genes had significant correlation with the above 4 chemotherapeutic drugs (Fig. 5B–F). In summary, signature can predict drug sensitivity, but the genes included in signature have no significant correlation with the four chemotherapeutic drugs commonly used for cervical cancer.

Fig. 5.

HPV-ferroptosis related genes prognostic modeling drug sensitivity. A Differences between cisplatin, paclitaxel, docetaxel, and cyclophosphamide in the high and low Risk groups. B Differences between cisplatin, paclitaxel, docetaxel, and cyclophosphamide in the high and low CKB groups.C Differences between cisplatin, paclitaxel, docetaxel, and cyclophosphamide in the high and low CYBB groups. D Differences between cisplatin, paclitaxel, docetaxel and cyclophosphamide in the high and low EFNA1 groups. E cisplatin, paclitaxel, docetaxel and cyclophosphamide in the high and low HELLS groups. F Cisplatin, paclitaxel, docetaxel and cyclophosphamide in the high and low VEGFA groups

Prognostic modeling of HPV-ferroptosis related genes and immune microenvironment correlation analysis

To study the correlation between HPV-ferroptosis related genes prognostic model and immune microenvironment. We first used the ssGSEA algorithm to assess the expression of 16 immune cells in the TCGA-CESC dataset in the split high and low risk groups (Fig. 6A). B cells, CD8 T cells, mast cells, plasma cell-like dendritic cells, tumor-infiltrating lymphocytes, dendritic cells, immature dendritic cells, natural killer cells, and helper T cells were affected by the Risk score. Among them, the expression differences between B cells, CD8 T cells, mast cells, plasma cell-like dendritic cells, and tumor-infiltrating lymphocytes were large between the high and low Risk groups (P < 0.001); dendritic cells also showed some differences between them (P < 0.01); and the expression differences between immature dendritic cells, natural killer cells, and helper T cells were not as obvious as those of other immune cells (P < 0.05). In addition,we evaluated the differences among 13 immune functions(Fig. 6B).check-point,Cytolytic_activity,Inflammation-promoting,T_cell_co-stimulation,Type_II_IFN_Reponse were affected by the Risk score. score scores. Among them, between high and low Risk groups in Check-point, Cytolytic_activity, and Inflammation-promoting were not as significant as other immune functions (P < 0.05); while in T_cell_co-stimulation, Type_II_IFN_Reponse showed some differences. Reponse showed some differences (P < 0.01).

Fig. 6.

A Differences of 16 immune cells in the groups with different expression levels of Risk score score. B Differences of 13 immune functions in the groups with different expression levels of Risk score score

Discussion

CC has made tremendous progress in prevention, screening and treatment, but cervical cancer outcomes have not improved significantly [23]. Patients with cervical cancer who develop metastasis or recurrence have a five-year OS of only 17% [24].Integration of HPV genes into the host cell genome, referred to as HPV gene insertion, not only destroys the virus' own genome, but also alters the structure of the host cell genome, a series of alterations that may lead to the development of cancer. For example, HPV integration disrupts the open reading (ORF) of the viral E2 gene, leading to the suppression of E2 gene expression, which results in the loss of expression regulation of two important HPV oncogenes, E6 and E7, and ultimately inhibits the functions of p53 and pRB, which results in the uncontrolled apoptosis of the host cells, inducing unlimited proliferation of the cells and a shift to malignancy [25]. Das et al. [26] proposed a new approach for the development of cancer based on the analysis of the clinical results of radical radiotherapy in cervical cancer patients. Radiotherapy clinical outcome analysis suggested that patients with high survival rate of radical radiotherapy for cervical cancer had a higher chance of HPV in free state, which led to HPV integration reducing the postoperative survival rate of cervical cancer patients. Iron death, as a type of programmed death, is currently an important topic of intensive research in the field of tumorigenesis and therapy. Several studies have shown that ferroptosis related biomarkers are powerful predictors of cancer prognosis and anti-tumor efficacy [27–29]. Based on the above, we believe that a comprehensive evaluation is necessary to illustrate the prognostic role of HPV-ferroptosis related genes in cervical cancer.

In this study, we focused on the prognostic impact of the HPV-ferroptosis related genes prognostic model, as well as to understand the relationship between the HPV-ferroptosis related genes prognostic model and the immune microenvironment, in order to further determine whether the model is a potential biomarker for prognosis. First, we analyzed the intersection of the GSE7410 dataset of differentially expressed genes, the HPV gene set and the iron death gene set. Notably, due to the small number of genes obtained from the intersection of the three datasets, we analyzed the intersection of each dataset and obtained a total of 54 HPV-ferroptosis related genes. Then, the 54 HPV-ferroptosis related genes were functionally analyzed, and the results indicated that these genes were associated with iron death, hypoxia response and HIF-1 signaling pathway. Univariate/multivariate Cox regression analysis was used to identify HPV-ferroptosis related genes with prognostic features and to establish a prognostic model for HPV-ferroptosis related genes. Subsequently, we analyzed the relationship between sensitivity to different chemotherapeutic drugs and the immune microenvironment of HPV-ferroptosis related genes prognostic models separately, and found that signature could predict drug sensitivity. Secondly we used ssGSEA algorithm to investigate the variability among different immune cells as well as immune functions. We found statistical significance in both different immune cells and immune functions.

Ferroptosis is now considered a form of immunogenic cell death characterized by the release of damage-associated molecular patterns (DAMP) from dead tumor cells [30–32]. The analysis revealed that B cells, CD8 T cells, T_cell_co-stimulation, and Type_II_IFN_Reponse showed higher abundance in the low-risk group compared to the high-risk group exhibiting higher immune scores. The above relevant results demonstrate to some extent the relevance of HPV-ferroptosis related genes prognostic models to immune infiltration in cervical cancer. It has been shown that malignant cells within the tumor affect the functions of CD8 T cells such as transport and differentiation by inhibiting their metabolic capacity [33], and glucose depletion is associated with infiltration of CD8 T cells and reduced anti-tumor function [34–36]. CD8 T cell subpopulations respond differently to low-glucose environments [37]. In particular, central memory T cells and naive T cells are involved in fatty acid synthesis, oxidative phosphorylation, and reduced glutamine catabolism in a low-glucose environment, which increases the level of fatty acid species and limits CD8 T cell function [37, 38]. Tumor cells outcompete immune cells in methionine uptake by upregulating the transporter protein SLC43A2, which impairs CD8 + T cell antitumor immunity by decreasing the level of SAM required to epigenetically maintain STAT5 expression [39]. The tumor microenvironment for CD8 + T cells determines their antitumor potential by affecting tumor-infiltrating lymphocytes (TIL). In conjunction with the above studies we hypothesize that it is the effect of the low glucose environment that leads to a reduction in the anti-tumor capacity of CD8 T cells in the high-risk group, leading to further development of tumor cells in the human body.Marta Canel et al. [40] showed that T-cell co-stimulation of the expression of the ligand CD80 by cancer cells sensitized murine tumors to the FAK inhibitor, and demonstrated that CD80 was activated by CD8 + T cells originating from solid epithelial carcinomas and some hematologic malignancies expressed by human cancer cells. In the absence of CD80, targeting alternative T cell co-stimulatory receptors, specifically OX-40 and 4-1BB in combination with FAK, was found to drive enhanced anti-tumor immunity and even complete ablation of murine tumors. Based on the above findings, combined with the fact that T cell co-stimulation expression was higher in the low-risk group than in the high-risk group in this study, we speculate that the combination of T cell co-stimulatory ligands with FAK in immunotherapy can lead to enhanced anti-tumor immunity and thus to the treatment of tumors. When B cells negatively regulate tumor growth, the presence of CD20 B-cell tumor-infiltrating lymphocytes in ovarian, non-small lung, and cervical cancers is associated with increased survival and decreased recurrence rates [41–43]. The results of these studies suggest that tumor-infiltrating lymphocytes are associated with tumor growth and tumor-infiltrating B cells are associated with a favorable prognosis.The underlying mechanism of B cell antitumor immunity may involve the secretion of effector cytokines, such as IFN-γ, by B cells, which may polarize T cells to a Th1 or Th2 response or promote a T-cell response through their role as antigen-presenting cells [44]. Further studies on how B cells negatively regulate cancer may be important for the development of effective anticancer therapies. Another example of the protective role of B cells in cancer is the ability of CpG-activated B cells to kill tumor cells through a TRAIL / Apo-2L-dependent mechanism, thus suggesting a novel role for B cell-mediated direct cytotoxicity in cancer. Further support for the cytotoxic potential of B cells was reported in B chronic lymphocytic leukemia, in which IL-21 and CpG-treated leukemia cells produced high levels of granzyme B and induced apoptosis in bystander B chronic lymphocytic leukemia cells [45]. The results of this study reveal new ways to induce the cytotoxic potential of B cells for the treatment of B-cell malignancies. The SIGNATURE established in our study is higher in the low risk group than in the high risk group in B cells, and in combination with the above study we hypothesize that negative regulation of tumor growth and thus tumor treatment is achieved.

In addition to this study, we note that the correlation between viral and iron death was validated. Su H et al. [47] showed that elevated SLC1A5 was an independent prognostic biomarker in patients with HBV-associated hepatocellular carcinoma (HCC). High levels of SLC1A5 were associated with poor overall survival, poor disease-specific survival, and tumor progression; SLC1A5-induced high expression of immune checkpoint genes in HBV-associated HCC may inhibit the therapeutic response in patients treated with ICIs. In addition Liu L et al. [48] showed that HBV regulates PCLAF aberrant selective splicing by inhibiting SRSF2. HBV reduces iron death through the SRSF2/PCLAF tv1 axis, causing sorafenib resistance. Thus, the SRSF2/PCLAF tv1 axis may be a prospective molecular therapeutic target for HBV-associated HCC and a predictor of sorafenib resistance.Inhibition of the SRSF2/PCLAF tv1 axis may be critical for the emergence of systemic chemotherapy resistance in HBV-associated hepatocellular carcinoma. From this, we can see that virus-ferroptosis related genes not only can have better prognostic guidance for different cancers, but also through further research we can discover their pathogenic or therapeutic mechanisms, which can bring more benefits for patients' prognosis.

Of course, there are two limitations of this study. One is that one of the cohorts included relatively few indicators in the clinical information, resulting in insufficient validation of some of the results; and the other is the use of retrospective data from public databases to construct and validate the prognostic model of HPV-ferroptosis related genes. In contrast using prospective data to assess its clinical utility would be more convincing. Based on the above two points, combined with the current situation that the mechanism of HPV-ferroptosis related genes in cervical cancer has not yet been determined, further exploration of the biological functions of HPV-ferroptosis related genes in cervical cancer in subsequent studies is essential.

In conclusion, we established the HPV-ferroptosis prognostic model through screening, which not only has a strong and accurate role in guiding prognosis. Meanwhile, in the immune microenvironment, immunotherapy in the low-risk group was superior to that in the high-risk group. In terms of drug sensitivity, chemotherapeutic drugs commonly used in cervical cancer also have good sensitivity in this signature. Therefore, our model can provide new ideas and insights for the clinic.

Conclusion

HPV-ferroptosis prognostic model through screening, which not only has a strong and accurate role in guiding prognosis. Meanwhile, in the immune microenvironment, immunotherapy in the low-risk group was superior to that in the high-risk group. In terms of drug sensitivity, chemotherapeutic drugs commonly used in cervical cancer also have good sensitivity in this signature. Therefore, our model can provide new ideas and insights for the clinic.

Author contributions

Songtao Han made substantial contributions to the study design, collection, and analysis of data and the writing of the article. Senyu Wang, Yuxia Li participated in drafting and revising the article critically for important intellectual content. YuJiao He and Jing Ma refined the data analysis. Songtao Han,Senyu Wang, Yuxia Li Yu, Jiao He, Jing Ma and Yangchun Feng were involved in revising the article. Songtao Han,Senyu Wang, Yuxia Li, YuJiao He and Jing Ma contributed to the study design and the final approval of the version to be published. All authors have read and approved the final manuscript.

Funding

This work was funded by National Natural Science Foundation of China (32360041), Tianchi Talent project of Xinjiang UyguR Autonomous Region (2023TCYCFYC), the Postdoctoral Fund of Affilated Tumor hospital of Xinjiang Medical University (2021),Talent support project of Affilated Tumor hospital of Xinjiang Medical University (2021). And funded by Xinjiang Uygur Autonomous Region Natural Science Foundation Youth Science Fund (2022D01C180).

Data availability

The datasets generated and/or analyzed during the current study are available in the GEO database (http://www.ncbi.nlm.nih.gov/projects/geo/)). FerrDb (http://www.zhounan.org/ferrdb/), NCBI-gene (https://www.ncbi.nlm.nih.gov/gene), MSigDB (http://www.gsea-msigdb.org /gsea/msigdb/), Genecard (https://www.genecards.org/) four databases to download the Ferroptosis gene set.

Declarations

Ethics approval and consent to participate

Not applicable.

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.