A novel defined ferroptosis-related gene signature for predicting the prognosis of Osteosarcoma

doi:10.21203/rs.3.rs-3561530/v1

Download PDF

Research Article

A novel defined ferroptosis-related gene signature for predicting the prognosis of Osteosarcoma

https://doi.org/10.21203/rs.3.rs-3561530/v1

This work is licensed under a CC BY 4.0 License

Version 1

posted

You are reading this latest preprint version

Background

Osteosarcoma is a malignant tumor that is harmful to adolescents and has a poor prognosis. There is no effective treatment at now. This study aimed to construct a prognosis based on ferroptosis-related genes to provide guidance for the treatment of osteosarcoma.

Methods

In this study, we identified three ferroptosis-regulating genes that were distinctly expressed in osteosarcoma and normal tissues. Osteosarcoma patients from three datasets were classified using consensus clustering of these three ferroptosis-regulated genes, and gene differences among the three subtypes were analyzed in one step, and these differentially expressed genes were analyzed by univariate regression (DEGs), the ssGSEA method was used to assess the tumor microenvironment (TME), and the principal component analysis algorithm was used to quantitatively construct a predictive model. Finally, we use qRT-PCR to detect gene expression to validate the model.

Results

We identified three distinct ferroptosis genotypes, and the differences between the three subtypes were used to build a prognostic model, which we termed the "Ferroptosis-Score". The score can predict the prognosis and overall survival time of patients with osteosarcoma. The survival of patients with high scores is higher than that of patients with low scores (P < 0.05). At the same time, it is observed that immune cell infiltration in patients with high scores is significantly higher than that in patients with low scores.

Conclusion

We constructed a Ferroptosis-Score scoring system, the higher the score, the better the prognosis of the patient. And found that cluster C may activates ferroptosis through autophagy and JAK-STAT signaling pathway, so the patients of cluster C have better prognosis.

Osteosarcoma

Ferroptosis

prognosis

Immune microenvironment

Subtype analysis

Osteosarcoma (OS) is the most common primary malignancy which originate from mesenchymal stem cells, with a high incidence in adolescents and children.¹ OS is characterized by rapid invasive growth, early lung metastasis and high recurrence rate.² Existing studies have shown that the overall 5-year survival rate of OS patients is between 65% and 75%. Nearly 1/5 of osteosarcoma are metastatic by the time they are diagnosed, and the survival rate of metastatic osteosarcoma patients is less than 20%.^{3, 4} In nearly 30 years, the survival rate of patients with OS has plateaued.⁵ Although there are currently treatment options such as adjuvant chemotherapy and surgical treatment that improve patient outcomes, the fact that these patients develop resistance to chemotherapy is a significant challenge. Therefore, it is imperative to discover specific diagnostic and prognostic biomarkers for OS.

As an essential element of life, iron plays crucial roles in various biological processes, including DNA synthesis, energy transport, erythropoiesis, as well as cellular signaling.^{6, 7} Ferroptosis is an iron dependent non apoptotic programmed cell death that, unlike apoptosis, necrosis or autophagy,⁸ is primarily an oxidative cell death that results from the accumulation of iron mediated lipid peroxidation.⁹ In addition, cells undergoing ferroptosis display characteristic morphological changes such as reduced mitochondrial volume, increased mitochondrial membrane density, reduced/disappeared mitochondrial cristae and outer membrane rupture.⁹ Accumulating evidence suggests that multiple human diseases related to iron metabolism or ROS, including cancer, are involved in the process of ferroptosis. ^10-12Interestingly, ferroptosis plays dual roles in tumor promotion and suppression in different models. Iron absorption by cancer cells can promote tumor initiation and progression, and on the other hand, ferroptosis inducers such as the anticancer drugs sorafenib, sulfasalazine, and artesunate have been shown to potently kill cancer cells¹³ and can mediate the release of interferons γ (IFNγ) from CD8+ T cells to exert tumor suppressive effects.¹⁴

In this study, we obtained gene expression and clinical information for 175 samples from TARGET(n=85)、GSE21257（n=53）、GSE39055（n=37）.Acquiring 259 ferroptosis related gene (FRG) from FerrDb database（http://www.zhounan.org/ferrdb/）. We analyzed the association between osteosarcoma and ferroptosis, screened the key regulators related to the overall survival (OS) of osteosarcoma patients, then classified based on these regulators and analyzed the association between each classification and m6A modification genes, as well as the prognostic and roles, to further evaluate the role of ferroptosis and m6A modification in the prognosis of osteosarcoma.

1.Data Collection

The RNA expression data of healthy tissues were collected from the genotype-tissue expression database. (GTEx)1 The fragments per kilobase of transcript per million mapped reads (FPKM) values in the expression data were converted using log2(x + 0.001). Public gene-expression data and full clinical annotation were searched in Gene-Expression Omnibus (GEO) and the TARGET database. Patients without survival information were removed from further evaluation. In total, 3 eligible OS cohorts (GSE21257、GSE39055 and TARGET-osteosarcoma ) were gathered in this study for further analysis. GSE21257、GSE39055 and TARGET data sets were merged and batch-to-batch variation was removed using “sva” package. The FPKM values were used to indicate gene expression levels.

2.Differentially Expressed FRGs (DE-FRGs)

We extracted 259 ferroptosis -related genes (FRGs) from Ferrdb (http://www.zhounan.org/ferrdb/). |LogFC| > 1 and adjusted p < 0.05 as the threshold, DE-FRGs between osteosarcoma and healthy tissues were identified using “limma” package. Besides, the venn online tool was utilized to draw the Venn diagram (http://jvenn.toulouse.inra.fr/app/example.html); “limma” package was used to generate Kapplan-Meier curves; Prognostic network diagram was performed by “igraph”, “psych”, “reshape2” and “RColorBrewer” packages

3.Consensus Clustering

Consensus clustering was applied to identify distinct ferroptosis-related patterns relating to the expression of ferroptosis regulators by the k-means method. The number of clusters and their stability were determined by the consensus clustering algorithm using the “ConsensuClusterPlus” package. We performed 1,000 times repetitions to guarantee the stability of our classification.

4.Principal component analysis (PCA)

PCA was a kind of mathematical algorithm which could decrease the dimensionality of the data as well as retain most of the variation in the dataset.¹⁵ And “limma” and “ggplot2” packages were utilized to perform PCA.

5.Gene Set Variation Analysis (GSVA)

GSVA was a kind of nonparametric and unsupervised method which was utilized to estimate the variation in pathway and biological process. In order to explore the differences in biological processes and pathways among different patterns, we performed GSVA by “GSVA” package (adjusted p < 0.05 was considered statistically significance).

6.TME Cell Infiltration

We used the ssGSEA (single-sample gene-set enrichment analysis) algorithm to quantify the relative abundance of each cell infiltration in the OS TME. The gene set for marking each TME infiltration immune cell type was obtained from the study of Charoentong,¹⁶ which stored various human immune cell subtypes including activated CD8 T cell, activated dendritic cell, macrophage, natural killer T cell, regulatory T cell and so on. The enrichment scores calculated by ssGSEA analysis were utilized to represent the relative abundance of each TME infiltrating cell in each sample.

7.Enrichment Analyses

Gene ontology (GO) and kyoto encyclopedia of genes and genomes (KEGG) analyses were performed were performed by “clusterProfiler”, “org.Hs.eg.db”, “enrichplot” and “ggplot2” packages.

8.Univariate and multivariate Cox Regression Analyses

Univariate and multivariate Cox regression analyses were performed by “survival” package. P < 0.05 was considered statistically significant and could be considered as an independent prognostic factor.

9.Establishment of a PS-Score Scoring Model

PCA was utilized to establish the PS-Score Scoring Model. We used principal component 1 and 2 to calculate PS-Score. The advantage of PCA was to concentrate the score on the largest set of highly correlated (or unrelated) gene blocks in the set, and to down-weight the genes contributions which were not tracked with other set members. We define the PS-Score using a method similar to GGI

PS-Score =\(\sum (\text{P}\text{C}1\text{i}+\text{P}\text{C}2\text{i})\) (1)

where i was expression of 10 DEGs.

10.Analyses of PS-Score Scoring Model

The “surv-cutpoint” function was utilized to dichotomize PS-Score, and the OS patients were divided into high and low PS-Score group. Subsequently, survival analyses were performed by “survminer” and “survival” packages; mulberry figure was generated using “ggalluvial”, “ggplot2” and “dplyr” packages; “limma” and “ggpubr” packages were utilized to explore correlations between PS-Score and ferroptosis-cluster or gene-cluster; univariate and multivariate Cox regression analyses were performed by “survival” package; “plyr”, “ggplot2”, “ggpubr”, “survival” and “survminer” packages were utilized to explore the correlations between PS-Score and three clinical traits (age, gender and fustat).

11.Prediction of immunotherapeutic response

We also attempted to explore prediction of immunotherapeutic response of PS-score. Hence, we performed expression levels of 19 immune checkpoints in high-risk and low-risk groups. And the above process was performed using “limma”, “reshape2”, “ggplot2” and “ggpubr” packages.

12.Quantitative real-time polymerase chain reaction(qRT-PCR)

Total RNA was isolated from cells using Trizol reagent(Takara, Japan). Reverse transcription was carried out with the PrimeScrip RT reagent Kit with gDNA Eraser (Takara, Japan). TB Green Premix Ex Taq kit (Takara, Japan) was used to perform real-time PCR, and we used the CFX96 Real-Time PCR Detection System (Bio-Rad, USA).The primer sequences are as follows( Table 1)

Table 1

The primer sequence of genes.
Gene	Forward Primer	Reverse Primer
HSPB1	ACGGTCAAGACCAAGGATGG	AGCGTGTATTTCCGCGTGA
FH	GGAGGTGTGACAGAACGCAT	CATCTGCTGCCTTCATTATTGC
EGFR	AGGCACGAGTAACAAGCTCAC	ATGAGGACATAACCAGCCACC
IMMT	CGGGCCTGTCAGTTATCGG	CAATGGACGGAGGACAAACTT
ALG5	GCAGCCCTCGTACTGATTTCC	TGTTTGGTAGGTGAGTCCCAT
DEXI	TACGTGGGCCTGTTCTTCG	CTCCGAGTCAGCATCGTAAAG
FAM104A	CCGGAAGTAGTGGTGGTACG	CTCCGCGAAGAGAGGGAAC
GDI2	ACCGGCCTGACGGAATGTA	CCGTAGTAAGGGTTTCGATCCA
CTSO	GCCTTCCGGGAAAGTCTTAATAG	TCCAGTCAAATCTTAACGGCAAA
FDFT1	CCACCCCGAAGAGTTCTACAA	TGCGACTGGTCTGATTGAGATA
CHORDC1	CCTTGCTGTGCTACAACCG	CGGAACACCTGGGTGGTATG
GAPDH	GGTGGTCTCCTCTGACTTCAACA	GTTGCTGTAGCCAAATTCGTTGT

13.Statistical Analysis and Cut-Off Value

R studio (version: 4.0.3) was used to analyze data, which was considered to be significant when p < 0.05.

1. Identification of ferroptosis related mRNAs.

The RNA sequences and clinical information of osteosarcoma obtained from three datasets, TARGET database (n = 85) and GSE21257 (n = 53), GSE39055 (n = 37) in the EGO database. We obtained 259 ferroptosis related genes from the FerrDb database (supplementary Table 1). We compared the RNA data of osteosarcoma with the gene data of ferroptosis, by the Venn diagram we can see a total of 108 ferroptosis related genes were present in the three datasets simultaneously (Fig. 1A and Supplementary Table 2). Based on |LogFC| > 1, adjpvalue < 0.05, these 108 ferroptosis related genes were differentially analyzed in the datasets of GTEx and TARGET. Through the results of heat map and volcano plot (Fig. 1B and 1C), 75 ferroptosis related genes with significant difference were obtained. Survival analysis of these 75 ferroptosis associated genes revealed that 27 genes were associated with overall survival (OS) of patients (Fig. 1D) and were statistically significant (P < 0.05).

2. Prognostic assessment of ferroptosis related mRNAs

The PPI network plot explained there is a tight connection among these 75 ferroptosis related genes (Fig. 2A). After univariate Cox regressive analysis of the 75 ferroptosis associated genes (Fig. 2B), three genes with significant differences and statistical significance were obtained, HSPB1, FH and EGFR. Through Venn diagram (Fig. 2C), it’s easy to find that these three genes were associated with the OS of patients. Therefore, we supposed that these three genes were not only related to ferroptosis, but also associated with the OS of patients. Perhaps they can serve as independent prognostic factors for osteosarcoma patients.

3. Typing and their biological characterization based on ferroptosis associated genes

We combined the RNA data and clinical information of the three datasets into a new cohort. According to the expression amounts of HSPB1, FH and EGFR, we classified the patients into three clusters by the unsupervised clustering method (Fig. 3A). Principal component analysis (PCA) revealed distinct distribution patterns among the three clusters (Fig. 3B). We next used survival analysis to evaluate their characteristics. The Kaplan – Meier curve showed that Cluster C had a significantly higher survival advantage than Clusters A and B, with Cluster A having the worst prognosis (Fig. 3C). After that we combined the clinical information of these three clusters and found that HSPB1, FH and EGFR had obvious differences among the three clusters. Meanwhile, patients in alive were mainly enriched in Cluster B and C; dead's patients were mainly enriched in Cluster A (Fig. 3D), which was also consistent with our survival analysis results. To explore the differences in biological behaviors among these three clusters, we performed GSVA enrichment analysis (Fig. 3E), which revealed that Cluster A was mainly enriched on metabolism related pathways, including cysteine and methionine metabolism, citric acid pathway, porphyrin and chlorophyll metabolism and histidine metabolism. Brent R et al. reported that ferroptosis is not only related to metabolic reactions and redox, but also to tumors, thus Cluster A was also significantly enriched in tumor related pathways. Interestingly, Cluster B was mainly enriched in pathways related to the nervous system, such as neuroactive ligand receptor interaction and olfactory signal transduction. In addition, enrichment analysis of Cluster C showed that it had a strong association with autophagy and complement system (Fig. 3F), and also activated the JAK-STAT signaling pathway. Next, we analyzed the infiltration of 23 types of cellular immunity (Fig. 3G) and found that Th17 cells, activated B cells, activated CD8 T cells and mature B cells showed the highest expression and infiltration in cluster C, giving Cluster C a better survival advantage, which was consistent with the results of our survival analysis. Finally, we further analyzed the activities of 13 immune related pathways between the three clusters (Fig. 3H), and the results showed that chemokine (CCR), immune checkpoint, APC co inhibition and T cell co inhibition were highest in Cluster C, lowest in Cluster A and intermediate in Cluster B.

4. Different ferroptosis-related gene clusters showed diverse clinical features and TME cell infiltration

Next, we performed differential analysis of genes related to the ferroptosis phenotype among the three clusters, resulting in 47 genes with significant differences between Cluster A and Cluster B, 180 genes between Cluster B and Cluster C, 618 genes between Cluster A and Cluster C (adjpvalue < 0.001, Supplementary Table 3). By Venn diagram, we finally obtained 26 genes with significant differences (Fig. 4A). Univariate cox regressive analysis of these 26 genes yielded 10 genes with independent risk factors: HSPB1, FH, IMMT, ALG5, DEXI, FAM104A, GDI2, CTSO, FDFT1, CHORDC1 (Fig. 4B). Based on the expression amounts of these 10 genes, by the unsupervised clustering method, we classified the patients into three geneclusters (Fig. 4C), and principal component analysis (PCA) showed that the three geneclusters had different distribution patterns among patients (Fig. 4D). After that we performed survival analysis for OS among the three geneclusters (Fig. 4E), found that geneclusterC had the best prognosis with better survival advantage, geneclusterB is the second, geneclusterA had the worst survival prognosis. Likewise, combined clinical information of these three clusters and found that these 10 genes have distinct differences. In addition, patients in alive are mainly enriched in ferroptosisCluster B and C; Dead patients were mainly enriched in ferroptosisCluster A (Fig. 4F), which is also consistent with our survival analysis results.

After GSVA enrichment analysis (Fig. 4G-H), it was found that Cluster A was mainly related to metabolic, tumor and cell cycle pathways, Cluster B was mainly enriched in metabolic and immune related pathways, Cluster C was enriched in nervous system and metabolism, and was related to autophagy, it also activated JAK-STAT signaling pathway and calcium signaling pathway. Through the immune infiltration analysis (Fig. 4I), we found that Th17 cells, regulatory T cells, activated B cells and immune B cells had the highest expression and infiltration in Cluster C, which made Cluster C has a better survival advantage, which was consistent with the results of survival analysis. Finally, we further analyzed the activities of 13 immune related pathways among the three clusters (Fig. 4J). The results showed that APC co-inhibition, chemokine (CCR), immune checkpoint and T cell co-inhibition scored the highest in Cluster C and the lowest in Cluster A.

5. High Ferroptosis-Score Identified the Alleviation of Clinical Characteristics and Immune Pathway Activation。

Based on the expression of these 10 genes, we use survival prognosis analysis to score each patient in the data set, which we call Ferroptosis-Score (Supplementary Table 4), and then divide the patients into high and low groups according to their optimal cutoff = 1.059876. Through the survival analysis of the two groups of patients (Fig. 5A), the higher the score of patients, the better the survival prognosis. Later, in order to explore which pathways Ferroptosis-Score is related to, we conducted a GSVA enrichment analysis (Fig. 5B), found that patients in the high group are mainly related to metabolism, immunity, inflammation and nervous system, while patients in the low group are mainly related to metabolism and tumor related pathways.

In order to explore the further relationship between Ferroptosis-Score and immunity, we analyzed the cellular immune infiltration and the activity of immune related pathways between patients in high and low groups. Results (Fig. 5C) showed that MDSC and Th17 cells were highly expressed and had significant differences. In immune pathway analysis (Fig. 5D), chemokine (CCR), APC co-inhibition and immune checkpoint were highly expressed in high groups. Compared with the low group, the expression of LGALS9, LAIR1, CD40 and CD86 in the high group was significantly higher, while the expression of CD28 was lower. By analyzing the changes of immune checkpoint expression between high and low Ferroptosis-Score groups (Fig. 5E), compared with low groups, LGALS9, LAIR1 and CD86 were highly expressed in high groups. We further analyzed the relationship between Ferroptosis-Score and m6A genes (Fig. 5F) found that IGFBP3 and ALKBH5 genes were highly expressed in the low risk group with statistical significance.

6. Ferroptosis-Score as an independent prognostic factor in osteosarcoma

In order to further verify the relationship between Ferroptosis-Score and survival, we used clinical correlation analysis (Fig. 6A) to find that Ferroptosis-Score is only related to the survival status of patients, and the alive group has higher Ferroptosis-Score. This is also consistent with the results of previous survival analysis. Ferroptosis-Score had no significant relationship with age and gender of patients. According to the age and gender of the patients, we divided them into two subgroups. Through survival analysis, we found that in the two subgroups, the prognosis of the high expression group was better than that of the low expression group. This result further proves that the higher the score, the better the prognosis of patients. On the contrary, the lower the score, the lower the prognosis of patients. This conclusion is not affected by the age and gender of patients. In order to prove whether Ferroptosis-Score can be used as an independent prognostic factor for osteosarcoma, we performed univariate cox regressive analysis (Fig. 6C) and multivariate regressive analysis (Fig. 6D). The results confirmed that Ferroptosis-Score can be used as an independent prognostic factor for osteosarcoma and is a favorable factor (HR < 1).

7. Linkage of FerroptosisCluste, GeneCluster and Ferroptosis-Score

Firstly, we explored the relationship between FerroptosisCluster and Ferroptosis-Score, found that there was a significant difference between FerroptosisCluster A, B and C (Fig. 7A), and FerroptosisCluster C had the highest Ferroptosis-Score, followed by B, and A had the lowest score. We have previously proved that C has the best prognosis, followed by B, and A has the worst prognosis. This proves the accuracy and scientific nature of Ferroptosis-Score. Then, we continued to explore the relationship between GeneCluster and Ferroptosis-Score (Fig. 7B), found that A in on GeneB scored the lowest Ferroptosis-Score. We have previously proved that C has the best prognosis, followed by B, and A has the worst prognosis. This proves the accuracy and scientific nature of Ferroptosis-Score once again. Finally, we use the Sangui diagram (Fig. 7C) to visualize the association among FerropositisCluster, GeneCluster, Ferropositis-Score, age, gender and Fustat. We find a tight association between FerropositisCluster A-GeneCluster A-low Ferropositis-Score-Alive. FerroptosisclusterC-GeneClusterC-high Ferroptosis-Score-Dead is also closely related.

We tested the expression of each gene in osteosarcoma cell lines MG63 and U2OS to validate our scoring system. We found that ALG5, DEXI, EGFR, and IMMT were highly expressed in both cell lines. CTSO, FDFT1, and HSPB1 were underexpressed in MG63 and U2OS(Fig. 7D)..

Osteosarcoma is a common primary malignant tumor with high invasiveness, and its high drug dependence is a bottleneck limiting its treatment.¹⁷ Recent studies have shown that drugs alter the tumor microenvironment have great potential in overcoming the drug resistance of inducing apoptosis by inducing cancer cell ferroptosis.¹⁸ However, most of the current studies focused on the correlation between a single FRG type and osteosarcoma, and the overall TME infiltrate characteristics mediated by multiple FRG types have not been fully recognized. Clarifying the correlation will help to enhance our understanding of the TME antitumor immune response and guide more effective therapeutic strategies.

Ferroptosis is a newly identified non-apoptotic cell death that is closely associated with the pathophysiological processes of multiple diseases.¹⁹ In this study, we determined the expression levels of 75 ferroptosis related genes in osteosarcoma and normal tissues from the GTEx, TARGET and GEO databases, and found significant differential expression of these genes. Then by survival analysis, 25 of these 75 genes expression had a great association with the prognostic survival of osteosarcoma patients. Meanwhile, we performed univariate Cox regressive analysis and identified the three genes, HSPB1, FH and EGFR to be associated with prognosis. Xiaofang Sun et al found that HSPB1 is a negative regulator of ferroptotic cancer cell death and it has important effects on ferroptotic cancer therapy.²⁰ While FH inactivation increases succinylation of cellular proteins, sensitizing cells to ferroptosis of various FINs.²¹ Jiazheng Zhao et al. found and validated that the EGFR gene is an independent prognostic factor in chemotherapy-resistant osteosarcoma and is associated with ferroptosis.²²

On the basis of the expression of these three genes, we classified all patients into three categories by consensus clustering analysis. Several studies have shown that tumor-infiltrating lymphocytes have superior therapeutic prospects, and tumor-infiltrating immune cells are positively correlated with the prognosis of clinical patients.^{23, 24} These three categories of patients have markedly different distribution patterns and immune cell infiltration profiles. Cluster A had the worst survival prognosis with minimal immune cell infiltration; Cluster B had a better survival prognosis relative to cluster A, as immune cell infiltration increased; Cluster C had the best survival prognosis with the highest immune cell infiltration. Some studies confirmed that autophagy promoted ferroptosis by degrading ferritin involved in iron metabolism in cancer cells.^{25, 26} Whereas to prevent ferroptosis, cancer cells activate their antioxidant capacity through multiple mechanisms, such as increased demand for cysteine to synthesize the GSH.²⁷ Immune regulators driven by JAK-STAT signaling contribute to the coordination of effective antitumor immune surveillance for antitumor effects.²⁸ A study by Fettah Erdogan found that JAK-STAT can be used as a therapeutic target for tumors and has a certain protective effect on patients.²⁹ In our study, cluster C was strongly correlated with autophagy and JAK-STAT signaling pathway. Therefore, we speculate that cluster C may promote tumor ferroptosis through autophagy, leading to better prognosis of patients and can also activate the JAK-STAT signaling pathway which increasing the degree of immune infiltration of tumors and leads to better patient outcomes in cluster C.

Furthermore, it was also demonstrated that the mRNA differences between these three different clusters and the patient survival and immune related biological pathways were significantly correlated in our study. We identified 10 ferroptosis phenotype associated genes with independent risk factors: HSPB1, FH, IMMT, ALG5, DEXI, FAM104A, GDI2, CTSO, FDFT1, CHORDC1. Current studies have shown that these genes are associated with the prognosis of multiple cancers and can be used as independent risk factors for tumors.^30–36

Consensus clustering with this 10 gene expression signature identified three genomic subtypes with similar results to the ferroptosis phenotype cluster. Similarly, using survival analysis, GSVA enrichment analysis, and immune cell infiltrate signature analysis, we demonstrated that overall TME infiltration mediated by multiple FRG types was strongly associated with overall survival in osteosarcoma patients again. Considering the individual heterogeneity and complexity among each patient, a scoring system was constructed to quantify the ferroptosis gene signature in osteosarcoma patients based on these 10 ferroptosis phenotype related genes. In our scoring, the higher the score, the better the prognosis. And we then verified that the riskscore score was higher in the Alive group and lower in the Dead group, which further verified the conclusion that the higher the score, the better the prognosis. Next, in order to verify whether our signature is still effective in different subgroups, we did a subgroup survival analysis and found that our signature was effective in all four subgroups, which further proved the applicability and effectiveness of our signature. Our signature, regardless of age or gender, is applicable to all types of patients. We also used the Ferroptosis-Score to evaluate two types of patients in previous and found that the Ferroptosis-Score discriminated groups of samples well, and the results were also consistent. Furthermore, we detected the expression of HSPB1, FH, EGFR, IMMT, ALG5, DEXI, FAM104A, GDI2, CTSO, FDFT1, and CHORDC1 in two osteosarcoma cell lines, MG63 and U2OS. The four genes ALG5, DEXI,, and IMMT are highly expressed in osteosarcoma cell lines, CTSO, FDFT1, and EGFR HSPB1 genes are lowly expressed. There is no significant difference in the expression of other genes. Most genes are differentially expressed in normal and tumor cell lines, further demonstrating the scientificity of our signature. Thus, Ferroptosis-Score may serve as an independent prognostic biomarker for predicting patient survival.

In summary, we genotyped according to ferroptosis phenotype related genes in osteosarcoma and analyzed the characteristics of their immune microenvironment, establishing a survival prognostic scoring system that may provide guidance for future clinical work. Unfortunately, our study is only an online-database-based bioinformatics analysis study, more clinical osteosarcoma samples are needed to validate our conclusions later.

Our study finally calculated the Ferroptosis-Score, and the higher the score, the better the prognosis of the patient. On the contrary, the lower the score, the worse the prognosis. Assuming that we admitted a patient with osteosarcoma, we can judge the prognosis of the patient by calculating the score, which can provide a great referential for the clinical diagnosis and treatment of patients with osteosarcoma.

Acknowledgments

We sincerely thank FerrDb for gifting us the ferroptosis death data. We are indebted to all the patients who participated in studies in TARGET and GEO databases for their data assistance during the preparation of this manuscript.

Author contribution

Conceptualization: CP, XC; methodology: CP, GH, WZ; formal analysis and investigation: GH, WZ, ZC, YL, HP; writing — original draft preparation: CP, GH, WZ; writing — review and editing: CP, GH, TW; funding acquisition: XC; resources: WZ, TW; supervision: XC, TW. All authors reviewed and approved the final manuscript.

Funding

This study was supported by the Thousand Talents Program of Jiangxi Province (Grant number: JXSQ2019201026). Key Intervertebral Disc Laboratory of Jiangxi Province (Grant number: 20202BCD42018). National Natural Science Foundation of China (Grant number: 82060403).

Data Availability

The data used to support the results of this study are available from TARGET database (https://www.ocg.cancer.gov/programs/target) and the Gene Expression Omnibus (GEO) database (https://www.ncbi.nlm.nih.gov/geo/) reference number [GSE21257 and

GSE39055]and FerrDb (http://www.zhounan.org/ferrdb/current/).

Competing interests

The authors declare no competing interests.

Ethics approval and consent to participate

Research carried out here were in compliance with the Helsinki Declaration. The protocol of this study was approved by the Ethics Committee of The Second Affiliated Hospital of Nanchang University.

Consent for publication

Not applicable.

J Ritter and S S Bielack， Osteosarcoma. Ann Oncol. 2010;vii320-325
G Ottaviani and N Jaffe， The epidemiology of osteosarcoma. Cancer Treat Res. 2009;3-13
J S Whelan and L E Davis， Osteosarcoma, Chondrosarcoma, and Chordoma. Journal of clinical oncology : official journal of the American Society of Clinical Oncology. 2018;188-193
S Miwa, T Shirai, N Yamamoto, et al.， Current and Emerging Targets in Immunotherapy for Osteosarcoma. J Oncol. 2019;7035045
S S Bielack, B Kempf-Bielack, G Delling, et al.， Prognostic factors in high-grade osteosarcoma of the extremities or trunk: an analysis of 1,702 patients treated on neoadjuvant cooperative osteosarcoma study group protocols. Journal of clinical oncology : official journal of the American Society of Clinical Oncology. 2002;776-790
B T Paul, D H Manz, F M Torti, et al.， Mitochondria and Iron: current questions. Expert Rev Hematol. 2017;65-79
M U Muckenthaler, S Rivella, M W Hentze, et al.， A Red Carpet for Iron Metabolism. Cell. 2017;344-361
S J Dixon, K M Lemberg, M R Lamprecht, et al.， Ferroptosis: an iron-dependent form of nonapoptotic cell death. Cell. 2012;1060-1072
Y Xie, W Hou, X Song, et al.， Ferroptosis: process and function. Cell Death Differ. 2016;369-379
L Galluzzi, I Vitale, S A Aaronson, et al.， Molecular mechanisms of cell death: recommendations of the Nomenclature Committee on Cell Death 2018. Cell Death Differ. 2018;486-541
B R Stockwell, J P Friedmann Angeli, H Bayir, et al.， Ferroptosis: A Regulated Cell Death Nexus Linking Metabolism, Redox Biology, and Disease. Cell. 2017;273-285
X Fang, H Wang, D Han, et al.， Ferroptosis as a target for protection against cardiomyopathy. Proc Natl Acad Sci U S A. 2019;2672-2680
Y Liu, C Duan, R Dai, et al.， Ferroptosis-mediated Crosstalk in the Tumor Microenvironment Implicated in Cancer Progression and Therapy. Frontiers in cell and developmental biology. 2021;739392
B R Stockwell and X Jiang， A Physiological Function for Ferroptosis in Tumor Suppression by the Immune System. Cell Metab. 2019;14-15
M Ringnér， What is principal component analysis? Nature biotechnology. 2008;303-304
P Charoentong, F Finotello, M Angelova, et al.， Pan-cancer Immunogenomic Analyses Reveal Genotype-Immunophenotype Relationships and Predictors of Response to Checkpoint Blockade. Cell reports. 2017;248-262
R Belayneh, M S Fourman, S Bhogal, et al.， Update on Osteosarcoma. Current oncology reports. 2021;71
B Li, L Yang, X Peng, et al.， Emerging mechanisms and applications of ferroptosis in the treatment of resistant cancers. Biomedicine & pharmacotherapy = Biomedecine & pharmacotherapie. 2020;110710
J Li, F Cao, H L Yin, et al.， Ferroptosis: past, present and future. Cell death & disease. 2020;88
X Sun, Z Ou, M Xie, et al.， HSPB1 as a novel regulator of ferroptotic cancer cell death. Oncogene. 2015;5617-5625
M J Kerins, J Milligan, J A Wohlschlegel, et al.， Fumarate hydratase inactivation in hereditary leiomyomatosis and renal cell cancer is synthetic lethal with ferroptosis induction. Cancer science. 2018;2757-2766
J Zhao, Y Zhao, X Ma, et al.， Outstanding prognostic value of novel ferroptosis-related genes in chemoresistance osteosarcoma patients. Scientific reports. 2022;5029
S Wang, J Sun, K Chen, et al.， Perspectives of tumor-infiltrating lymphocyte treatment in solid tumors. BMC medicine. 2021;140
X Liu, S Wu, Y Yang, et al.， The prognostic landscape of tumor-infiltrating immune cell and immunomodulators in lung cancer. Biomedicine & pharmacotherapy = Biomedecine & pharmacotherapie. 2017;55-61
W Hou, Y Xie, X Song, et al.， Autophagy promotes ferroptosis by degradation of ferritin. Autophagy. 2016;1425-1428
S Ma, R F Dielschneider, E S Henson, et al.， Ferroptosis and autophagy induced cell death occur independently after siramesine and lapatinib treatment in breast cancer cells. PloS one. 2017;e0182921
F Ursini and M Maiorino， Lipid peroxidation and ferroptosis: The role of GSH and GPx4. Free radical biology & medicine. 2020;175-185
K L Owen, N K Brockwell and B S Parker， JAK-STAT Signaling: A Double-Edged Sword of Immune Regulation and Cancer Progression. Cancers. 2019;
F Erdogan, T B Radu, A Orlova, et al.， JAK-STAT core cancer pathway: An integrative cancer interactome analysis. Journal of cellular and molecular medicine. 2022;2049-2062
Y Hiyoshi, Y Sato, M Ichinoe, et al.， Prognostic significance of IMMT expression in surgically-resected lung adenocarcinoma. Thoracic cancer. 2019;2142-2151
H Lu, P Cunnea, K Nixon, et al.， Discovery of a biomarker candidate for surgical stratification in high-grade serous ovarian cancer. British journal of cancer. 2021;1286-1293
C J Walker, K Mrózek, H G Ozer, et al.， Gene expression signature predicts relapse in adult patients with cytogenetically normal acute myeloid leukemia. Blood advances. 2021;1474-1482
W Zhang, Z Liu, S Xia, et al.， GDI2 is a novel diagnostic and prognostic biomarker in hepatocellular carcinoma. Aging. 2021;25304-25324
D Liu, M F Ho, D J Schaid, et al.， Breast cancer chemoprevention pharmacogenomics: Deep sequencing and functional genomics of the ZNF423 and CTSO genes. NPJ breast cancer. 2017;30
R Huang, C Zhang, X Wang, et al.， Identification of FDFT1 as a potential biomarker associated with ferroptosis in ccRCC. Cancer medicine. 2022;
F Urabe, N Kosaka, Y Sawa, et al.， miR-26a regulates extracellular vesicle secretion from prostate cancer cells via targeting SHC4, PFDN4, and CHORDC1. Science advances. 2020;eaay3051.

No competing interests reported.

Download PDF

Version 1

posted

You are reading this latest preprint version

A novel defined ferroptosis-related gene signature for predicting the prognosis of Osteosarcoma

Status:

Version 1

Abstract

Background

Methods

Results

Conclusion

Figures

Introduction

Materials and Methods

Results

Discussion

Conclusions

Declarations

References

Additional Declarations

Supplementary Files

Status:

Version 1