According to epidemiological studies, hepatocellular carcinoma (HCC) is considered the seventh most common malignancy and the second most common cause of cancer-related death (1). Many factors, including chronic infection with HBV/HCV, alcohol abuse, long-term obesity or exposure to aflatoxin, have been reported to be related to the progression of HCC (2). Considering that the early monitoring methods for HCC are still limited and that the tumor easily metastasizes and has a poor prognosis, it is necessary to develop new detection methods and identify new therapeutic targets for HCC.

Ferroptosis is a unique form of regulated cell death associated with iron metabolism, which is different from apoptosis, necrosis, and autophagy (3). Although the detailed mechanism underlying the role of ferroptosis in tumors is still unclear, several studies have reported that ferroptosis is involved in various cancers, including breast cancer (4–6), pancreatic cancer (7), ovarian cancer (8), and HCC (9–12). Compared to normal nontumor cells, cancer cells have a higher level of iron, which indicates the potential of ferroptosis inducers in new antitumor strategies (13, 14). For example, the triterpene saponin ardisiacrispin B and epunctanone exert cytotoxic effects on cancer cells with multiple drug resistance partly via ferroptosis (15, 16). In addition to ferroptosis-inducing agents, an increasing number of ferroptosis-related genes have been identified and found to be involved in the progression of cancers by serving as mediators of ferroptosis-related pathways. In HCC, CISD1 and a polymorphism of the TP53 gene (the S47 variant) are reported to negatively regulate ferroptosis, which proves that ferroptosis-related genes play a role in tumor progression (17, 18). In addition, several ferroptosis-related genes, including MI1G, NRF2, and Rb, were found to protect HCC cells from sorafenib-induced ferroptosis (19–21).

Based on the existing findings, we have noticed that ferroptosis plays a pivotal role in the progression of HCC; however, the specific function of ferroptosis-related long noncoding RNAs (lncRNAs) in HCC has not been fully elucidated. lncRNAs are a class of noncoding transcripts more than 200 nucleotides in length (22). It has been proved that lncRNAs serve as pivotal players in posttranscriptional regulatory mechanisms that target mRNA splicing, stability, or translation, the scope of which is still expanding (23). Dynamic alterations in the expression and mutation of lncRNAs are closely associated with tumorigenesis, tumor progression, metastasis, and cancer immunity indicating the emerging roles of lncRNAs as new biomarkers and therapeutic targets for cancer treatment strategies (24–26). Therefore, investigating lncRNAs related to ferroptosis and HCC is essential to our understanding of the mechanisms of tumor development. Recently, a model containing 3 ferroptosis-related lncRNAs was reported; however, it exhibited low predictive power for HCC with an area under the curve (AUC)=0.7 (27).

In this study, we constructed a new ferroptosis-related signature including mRNAs and lncRNAs by both weighted gene co-expression analysis (WGCNA) and least absolute shrinkage and selection operator (LASSO) regression analysis. We evaluated the predictive value of the ferroptosis-related signature and investigated the differential immune response in a variety of ways, including tumor microenvironment (TME) analysis and single-sample gene set enrichment analysis (ssGSEA). Furthermore, functional enrichment analysis was performed to clarify the biological functions of these differentially expressed genes (DEGs). Based on the DEG results, several small molecule drugs targeting ferroptosis-related signature components were identified via the L1000FWD database, and SIB-1893, geldanamycin and PD-184352 were visualized by PubChem.

In this study, we explored the role of the ferroptosis-related signature, which includes mRNAs and lncRNAs, in HCC. A prognostic model including 10 ferroptosis-related mRNAs/lncRNAs was first constructed and tested in the TCGA-LIHC dataset. Furthermore, immune analysis, including analysis with various bioinformatics tools, indicated obvious differences in the TME and immune cell infiltration between the low-risk and high-risk groups, especially in terms of checkpoint genes and m6A-associated genes. Functional analysis revealed that many tumor-related pathways were enriched. These findings strongly implied the great potential roles of ferroptosis in HCC.

The rapid progression of cancer is accompanied by the transformation and acceleration of a variety of metabolic pathways, which often means that a large number of metabolic byproducts, such as oxygen free radicals, accumulate in tumor cells to activate the oxidative stress pathway. Although tumor cells should have been vulnerable to ferroptosis, it has been found that cancer cells can acquire their resistance to ferroptosis through alteration of gene expression (36, 37). However, reactivation of the ferroptosis-related pathway in tumor cells may provide new therapeutic targets for tumor therapy. Although current therapeutic strategies, including surgical intervention, tumor-targeted drugs, immunotherapeutic agents or antiviral treatment, have improved OS in HCC patients in an inspiring way, their applications are still limited by tumor heterogeneity (38) and the development of drug resistance (39, 40).

The prognostic model in our study integrated 10 ferroptosis-related components, including 2 genes (SLC1A5 and SLC7A11) and 8 lncRNAs (AC245297.3, MYLK-AS1, NRAV, SREBF2-AS1, AL031985.3, ZFPM2-AS1, AC015908.3, MSC-AS1). Solute carrier family 1 member 5 (SLC1A5) is reported to be a driver gene of ferroptosis that mediates the uptake of glutamine, a conditionally essential amino acid in rapidly proliferating tumor cells (41, 42). In erastin- and RSL3-induced ferroptosis, glutamine importation and metabolism induce lipid ROS generation and promote cell death (43). Suppression of SLC1A5 by miR-137 or the small molecular inhibitor GPNA strongly inhibits glutaminolysis to cause ferroptotic cell death (44). As SLC1A5-mediated glutamine transport plays a key role in tumor cell metabolism, proliferation, and ferroptosis, blocking SLC1A5 has been shown to successfully prevent tumor cell proliferation in melanoma (45), non-small-cell lung cancer (46, 47), prostate cancer (48) and acute myeloid leukemia (49). Solute carrier family 7 member 11 (SLC7A11) is a pivotal protein component of system Xc- that is responsible for maintaining redox homeostasis by importing cystine, where it is then reduced to cysteine and used to synthesize the major antioxidant GSH (50, 51). Numerous experiments have demonstrated its high expression in various cancers and multiple effects on cancer growth, invasion, metastasis and unfavorable prognosis (52–60). It has been demonstrated that SLC7A11 confers resistance to ferroptosis in cancer cells by importing cystine for the synthesis of GSH and indirectly relieving lipid ROS stress by activating the essential enzyme GPX4 to reduce lipid hydroperoxides (50, 61). Moreover, SLC7A11 has been demonstrated to be involved in the resistance to anticancer treatments, which was supported by experiments in which therapeutic resistance was reversed by directly targeting SLC7A11 (59, 60, 62–66). Therefore, targeting SLC7A11 exhibits good potential for the treatment of cancer, and several drugs targeting SLC7A11 are being prepared for clinical testing (54–58).

As the factor with the highest positive correlation coefficient in the prognostic model, lncRNA MSC-AS1 has been reported to be involved in HCC (67), lung adenocarcinoma (68), laryngeal cancer (69) and kidney renal clear cell carcinoma (70). It acts sponging miR-33b-5p to upregulate GPAM (68). Another study indicated that it activates the Wnt/β-catenin pathway to regulate tumor proliferation and migration via miR-3924/WNT5A (70). MYLK-AS1 was reported to be associated with tumor progression and angiogenesis in HCC. The mechanism involves targeting the miR-424-5p/E2F7 axis and activating the VEGFR-2 signaling pathway (71) or stimulating the EGFR/HER2-ERK1/2 signaling pathway (72). LncRNA NRAV has been reported to be involved in the antiviral immune response (73, 74). In addition, several independent bioinformatics analyses have demonstrated that it is a valuable clinical prognostic biomarker in HCC (75, 76) and lower-grade glioma (77). ZFPM2-AS1 is reported to be involved in numerous tumors, including lung adenocarcinoma (78), renal cell cancer (79), gastric carcinoma (80) and HCC (81). Mechanistically, it is reported to attenuate the p53 pathway by stabilizing MIF (80) or regulating miR-139/GDF10 (81).

Few studies have investigated lncRNA MSC-AS1, AL031985.3, and AC245297.3. However, they are indicated by different bioinformatics analyses to have prognostic value [MSC-AS1 (82, 83), AL031985.3 (84–86) in HCC, and AC245297.3 (87, 88) in breast cancer]. Therefore, these factors may be involved in multiple mechanisms in HCC and breast cancer, and their specific regulation mechanisms still further study. In particular, MSC-AS1 was the only protective lncRNA in the model and had a high correlation coefficient. There are no studies on lncRNA SREBF2-AS1 at present, so it needs further study. Notably, although several identified ferroptosis-related lncRNAs have not been well studied, in our study, the lncRNAs were closely co-expressed with many ferroptosis-related genes ( Figure 3C ).

The results from both immune infiltration analysis and enrichment analysis indicate that a higher level of APCs and humoral immunity including B cells and the complement system in the high-risk group. Besides, results from both difference in APC co-stimulation and co-inhibition appears contradictory in the immune microenvironment. However, we found the results are somewhat similar to the researches about ferroptosis-related signature in other cancers (6, 89–91). Though the mechanisms illustrating tumor susceptibility to ferroptosis have been an intense area of research in past decades, the complex relationship between tumor immunity and ferroptosis remains elusive. Tumor-infiltrating lymphocytic B cells (TIL-B) have been reported to be a main component of TILs in ovarian and breast cancers, which may be correlated with improved survival (92, 93). However, tumor development is enhanced when B cells are present have been reported in several mouse models (94, 95). For pancreatic carcinoma, several researches have demonstrated that TIL-Bs in supporting both early and more advanced stages of pancreatic tumorigenesis by multiple mechanisms, including suppression of other immune cells (e.g., CD8+ T cells and macrophages) in the tumor microenvironment and promoting pancreatic cancer cell proliferation (96).

Based on the DEGs between high- and low-risk groups, several small molecule drugs targeting ferroptosis-related signature components were identified via the L1000FWD database which was mainly enriched in the glutamate receptor antagonist, HSP90 inhibitor, MEK inhibitor and so on. As described above, glutamate metabolism was involved in the SLC1A5-induced ferroptosis, which may be inhibited by glutamate receptor antagonist, such as SIB-1893. HSP90-associated chaperone-mediated autophagy has been demonstrated to obviously promote ferroptosis (97, 98), which could be reduced by geldanamycin. The activation of the Raf-MEK-ERK pathway plays an important role in the proliferation, differentiation, invasion and metastasis of cancer cells (99). PD-184352 was the first MEK inhibitor to enter the clinical trial. It was terminated in phase II clinical trial because of its poor solubility, low oral bioavailability and large individual differences (100). Though the application of MEK inhibitor is limited by clinical efficacy and drug resistance, the development of novel MEK inhibitor is still one of the promising directions of anti-cancer treatment (101–103).

The model exhibited several advantages. 1) The prognostic model showed excellent performance in the ROC curve analysis, with AUCs as high as 0.830 and 0.806 in the training and test groups, respectively. As depicted in the DCA, the ferroptosis-related signature had better predictive value than all pathological characteristics. The risk score was identified as an independent risk factor, with HR=3.038 in the training group and 2.824 in the test group. Therefore, the model showed excellent clinical prognostic value. 2) According to the risk score from the model, the HCC patients were divided into two distinct groups. Analysis of the TME and immune infiltration showed that there were numerous variables with statistically significant differences between the two groups, especially the expression of checkpoint genes. These differences were consistent with the differences observed in the GO and KEGG analysis. Therefore, immunotherapies targeting the 10 components of the ferroptosis-related signature are promising methods for anti-tumor treatment. 3) Considering that there are frequent updates in the field of ferroptosis research, a large number of studies and ferroptosis-related databases were assessed (12, 13, 28–30). As many as 267 candidate ferroptosis-related genes were included in the analysis. To our knowledge, this is the largest ferroptosis-related gene list used for bioinformatics analysis at present.

There are also some limitations to this study. Due to the limited knowledge of ferroptosis, most of the signature components in our research are involved not only in ferroptosis-related pathways but also in other pathways, such as immunity and autophagy. This may be more common for lncRNAs, which have nonspecific functions in biological processes. Therefore, it is difficult to evaluate the exact role of ferroptosis alone in HCC with our risk score. These results should be further validated in external HCC cohorts from multicenter research. In another attempt to investigate the role of ferroptosis-related in other cancers, we performed the survival analysis of SLC7A11, MSC-AS1 and MYLK-AS1 in gastrointestinal cancer including colon cancer, esophageal cancer, liver cancer, pancreatic cancer and stomach cancer ( Supplementary Figure ). Most of results showed that they are only play a specific role in liver cancer.

In conclusion, our research established a 10-component ferroptosis-related signature including mRNAs and lncRNAs for predicting the prognosis of HCC patients. The ferroptosis-related signature showed excellent performance in predicting clinical prognosis. The signature can be used to calculate the risk score, which accurately reflects the tumor environment and immune filtration of patients, thereby providing a reference for clinical treatment. Therefore, the ferroptosis-related signature is expected to be a new biomarker for both diagnosis and treatment decision making. Further investigation of the role and mechanism of the 10-component ferroptosis-related signature in the progression of HCC is still needed.

The original contributions presented in the study are included in the article/ Supplementary Material . Further inquiries can be directed to the corresponding authors.

Z-AC collected the papers and analyzed data, analyzed the conclusions, and drafted the manuscript. HT reviewed the data and conclusions. D-MY and YZ contributed to writing. C-JY and Z-JF presented the idea of this manuscript, supported the funding, analyzed the conclusions, drafted and revised the manuscript. All authors contributed to the article and approved the submitted version.

This study was funded by the Natural Science Foundation of Hebei Province (Grant No. H2020206337).

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.