Establishment of a prognostic signature for hepatocellular carcinoma using disulfidptosis-related lncRNAs

doi:10.21203/rs.3.rs-3203009/v1

Download PDF

Article

Establishment of a prognostic signature for hepatocellular carcinoma using disulfidptosis-related lncRNAs

https://doi.org/10.21203/rs.3.rs-3203009/v1

This work is licensed under a CC BY 4.0 License

Version 1

posted

You are reading this latest preprint version

Background

Disulfidptosis is a recently discovered form of programmed cell death that may be a new direction in tumor treatment. Long non-coding RNAs (lncRNAs) play an important role in the development and progression of hepatocellular carcinoma (HCC). However, how disulfidptosis-related lncRNAs (DRLs) are involved in regulating HCC is not yet understood. This study aimed to establish a prognostic signature for DRLs and analyze their clinical value in patients with HCC.

Method

RNA sequencing, mutation, and clinically relevant data were collected from the Cancer Genome Atlas database (TCGA). Univariate Cox analysis, least absolute shrinkage and selection operator (LASSO) analysis, and multivariate Cox analysis were conducted to evaluate DRLs. On the basis of these analyses, a prognostic signature was developed. Subsequently, we validated the accuracy of this prognostic signature using receiver operating characteristic (ROC) curves, C-index, survival curve, nomogram, and principal component analysis (PCA). Finally, we performed Gene Ontology (GO) and Kyoto Encyclopedia of Genes and Genomes (KEGG) analyses, gene set enrichment analysis (GSEA), tumor mutation burden (TMB) analysis, immune-related analysis, tumor immune dysfunction and exclusion (TIDE) analysis, and half-maximal inhibitory concentration (IC₅₀) predictions.

Results

A prognostic signature consisting of MKLN1-AS, TMCC1-AS1, AL603839.2, AC245060.7 and AL049840.3 was developed. This prognostic signature demonstrated reliable predictive capability for estimating the survival time of patients with HCC. We observed notable differences between the high- and low-risk groups in terms of immune cell population, immune function, TIDE, and IC₅₀.

Conclusions

A new prognostic signature was developed based on the five DRLs to predict the prognosis of patients with HCC, which may be helpful for individualized therapeutic strategies.

Biological sciences/Cancer

Biological sciences/Immunology

hepatocellular carcinoma

lncRNA

disulfidptosis

prognosis

bioinformatics analysis

Hepatocellular carcinoma (HCC) is a common and serious malignancy that is a major health problem worldwide[1]. Despite the many treatment options available, the prognosis of hepatocellular carcinoma remains relatively uncertain, largely attributable to tumor heterogeneity [2]. Therefore, the lack of reliable prognostic signatures to clarify the prognosis and guide the individualization of treatment for patients with hepatocellular carcinoma is a clinical challenge.

Disulfidptosis is a novel form of cell death that differs from the currently known programmed cell death, including apoptosis, autophagy, pyroptosis, ferroptosis, and cuproptosis[3]. Disulfidptosis is a rapid form of death caused by disulfide stress resulting from the accumulation of excessive intracellular cystine. This form of programmed cell death usually occurs under conditions of glucose starvation[4]. The discovery of disulfidptosis provides new insights into the mechanisms of tumorigenesis and promises to be a target for intervention in tumor treatment [5]. However, there is limited literature available on the association between disulfidptosis and HCC, making it difficult to draw definitive conclusions.

Long non-coding RNAs (lncRNAs) are a class of non-coding RNA molecules more than 200 nucleotides in length[6, 7]. In contrast to traditional non-coding RNAs such as rRNA and tRNA, lncRNAs are not translated into proteins after transcription[8, 9]. Nevertheless, they perform important regulatory functions within the cell and are involved in a variety of biological processes including cell differentiation, immune response, chromosomal imbalance, and disease development. Aberrant expression of some lncRNAs is associated with a variety of diseases such as cancer, cardiovascular disease, and neurological disorders[10]. Therefore, the study of lncRNAs is important to understand gene regulatory networks and disease mechanisms.

Currently, the role of disulfidptosis and its associated lncRNAs in HCC remains unclear. In this study, we used The Cancer Genome Atlas (TCGA) database to construct a prognostic signature for HCC with disulfidptosis-related lncRNAs (DRLs) using bioinformatics approaches. The identification of a prognostic signature for HCC provides value for prognostic assessment and individualized treatment.

2.1 Datasets and patients

Transcriptome data, which included 374 HCC tissues and 50 normal tissues, and clinical data of 365 HCC patients were downloaded from TCGA database (https://portal.gdc.cancer.gov). Language Perl (version Strawberry—perl-5.32.1.1; https://www.perl.org (accessed on May 15, 2023)) was used to extract RNA sequencing data from the standardized FPKM format[11]. The clinical data included age, sex, grade, stage, tumor lymph node metastasis, survival time, and survival status. Based on earlier literature, the disulfidptosis-related genes (GYS1, NDUFS1, OXSM, LRPPRC, NDUFA11, NUBPL, NCKAP1, RPN1, SLC3A2, and SLC7A11) were retrieved[4]. The clinical and sample information of 365 patients was randomly divided into training and validation sets according to a ratio of 1:1 using the caret package. The basic information on the training and validation sets is presented in Table 1. LncRNAs and mRNAs were isolated by classifying the downloaded transcriptome data using Perl software. Since clinical information for patients participating in this study was obtained from TCGA database and TCGA publication guidelines were strictly adhered to, ethics committee approval was not required.

Table 1

Clinical characteristics of the validation set and training set for HCC [n(%)].
Covariates	Type	Total	Validation	Training	P value
Age	<=65	227(62.19%)	112(61.54%)	115(62.84%)	0.8817
Age	> 65	138(37.81%)	70(38.46%)	68(37.16%)
Gender	FEMALE	119(32.6%)	51(28.02%)	68(37.16%)	0.0801
Gender	MALE	246(67.4%)	131(71.98%)	115(62.84%)
Grade	G1	55(15.07%)	32(17.58%)	23(12.57%)	0.4087
Grade	G2	175(47.95%)	86(47.25%)	89(48.63%)
Grade	G3	118(32.33%)	58(31.87%)	60(32.79%)
Grade	G4	12(3.29%)	4(2.2%)	8(4.37%)
Grade	unknown	5(1.37%)	2(1.1%)	3(1.64%)
Stage	Stage I	170(46.58%)	94(51.65%)	76(41.53%)	0.0078
Stage	Stage II	84(23.01%)	47(25.82%)	37(20.22%)
Stage	Stage III	83(22.74%)	28(15.38%)	55(30.05%)
Stage	Stage IV	4(1.1%)	2(1.1%)	2(1.09%)
Stage	unknown	24(6.58%)	11(6.04%)	13(7.1%)
T	T1	180(49.32%)	99(54.4%)	81(44.26%)	0.0044
T	T2	91(24.93%)	51(28.02%)	40(21.86%)
T	T3	78(21.37%)	28(15.38%)	50(27.32%)
T	T4	13(3.56%)	3(1.65%)	10(5.46%)
T	unknown	3(0.82%)	1(0.55%)	2(1.09%)
M	M0	263(72.05%)	124(68.13%)	139(75.96%)	1
M	M1	3(0.82%)	1(0.55%)	2(1.09%)
M	unknown	99(27.12%)	57(31.32%)	42(22.95%)
N	N0	248(67.95%)	119(65.38%)	129(70.49%)	0.6829
N	N1	4(1.1%)	1(0.55%)	3(1.64%)
N	unknown	113(30.96%)	62(34.07%)	51(27.87%)

2.2 Identification of DRLs

The association coefficient between lncRNAs and disulfidptosis-related genes was calculated based on RNA expression levels. Co-expression analysis was performed using Spearman’s correlation coefficient to identify lncRNAs related to disulfidptosis-related genes[12]. The screening criteria were an absolute value > 0.4, p-value < 0.001, and the Sankey relationship between disulfidptosis-related genes and DRLs was visualized using the ggplot2 and ggalluvial packages of R software[13].

2.3 Establishment of the DRLs prognostic signature

Initially, prognostic DRLs were acquired using univariate Cox regression, and the results were visualized to create a forest map. Subsequently, to mitigate data overfitting, we utilized least absolute shrinkage and selection operator (LASSO) regression analysis to obtain a total of five DRLs. [14]. Finally, the outcomes of the Lasso screening process were incorporated into a multivariate Cox regression model to evaluate the contribution of each lncRNA as a prognostic factor in the overall survival (OS) of HCC, and the risk score was calculated. Risk score = (Expi × bi). (Exp: expression level of the model gene; bi: model gene coefficient). The survival, glmnet, survminer, and timeROC packages were used in this procedure.

2.4 Assessment of the DRLs prognostic signature

The included participants were categorized into high- and low-risk groups based on the median risk score for the training set. Survival analysis was used to compare the OS in the training set, validation set, and among all sets, and the progression-free survival (PFS) of all participants. Survival, survminer, and pheatmap packages were used for the survival analysis. Risk scores and other clinical features were then incorporated into univariate and multivariate Cox regression analyses to analyze whether the risk score was a prognostic factor independent of other clinical indicators. The survival, survminer and timeROC packaged in R software were used to plot the receiver operating characteristic (ROC) curves at 1, 3 and 5 years and ROC of the risk scores and other clinical features, while the R software survival, rms, and pec packages are used to plot the C-index curve to assess the predictive power of the risk score[15].

2.5 Nomogram and principal component analysis (PCA)

First, we combined the risk score with the clinicopathological traits of patients with HCC to create a nomogram using the rms package. Next, the nomogram was used to predict the 1-year, 3-year, and 5-year OS rates of HCC patients, and the calibration curves for 1, 3, and 5-year was drawn to assess the discrepancy between the predicted and actual results. A calibration curve was constructed to confirm the predictive power of the nomogram model. PCA was used to validate the risk signature and visualize the results using the R software scatterplot3D package. Moreover, patients were classified into different clinical variables to determine whether the prognostic signature was suitable for patients with different clinical variables.

2.6 Enrichment function analysis

The limma package in R was used to identify differentially expressed genes (DEGs) in different risk groups based on the criteria |logFC| > 1 and adjusted p < 0.05. The identified DEGs were then subjected to Gene Ontology (GO) function and Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway enrichment analyses using R software packages, such as clusterProfiler, org.Hs.eg.db, enrichplot, ggplot2, circlize, RColorBrewer, dplyr, ggpubr, and ComplexHeatmap. GO enrichment analysis allows for the annotation and classification of genes based on their involvement in biological processes (BP), molecular functions (MF), and cellular components (CC)[16]. KEGG pathway enrichment analysis helps elucidate the potential signaling pathways associated with HCC in different risk groups and sheds light on the biological processes related to this disease[17]. The org.Hs.eg.db, limma, clusterProfiler, and enrichplot packages were used for gene set enrichment analysis (GSEA) to further identify enrichment pathways in different risk groups.

2.7 Immune-related function analysis

The ESTIMATE algorithm was used to assess tumor purity in patients with HCC[18]. The association between the prognostic signature and TME was evaluated by calculating immune, stromal, and ESTIMATE scores. Additionally, immune cell infiltration in tumor samples was estimated using the CIBERSORT method, and the relative abundance of immune cells was computed to illustrate the relationship between the risk score and immune status. GSEA was performed to investigate expression differences in immune-related functional gene sets between the high- and low-risk groups. This analysis used various R packages, including estimate, limma, reshape2, ggpubr, GSVA, and GSEABase.

2.8 Tumor mutation burden (TMB) and tumor immune dysfunction and exclusion (TIDE) analysis

To process the somatic mutation data and visualize the results, we utilized the maftools package. Additionally, we assessed the clinical significance of TMB in patients with HCC by generating Kaplan-Meier (KM) curves, which visually demonstrated the impact of TMB on the survival time of patients with HCC. To download the tumor immune dysfunction and rejection file from the TIDE website (http://tide.dfci.harvard.edu, (accessed on 15 May 2023)), use the limma and ggpubr packages were used to explore different immune escape mechanisms in different risk groups[14].

2.9 Prediction of anti-HCC drug response

Gene expression and drug sensitivity data for drug response prediction were downloaded from the Genomics of Drug Sensitivity in Cancer website (GDSC, https://www.cancerrxgene.org, (accessed on May 15, 2023)). The half-maximal inhibitory concentrations (IC₅₀) of four conventional anti-HCC drugs, including sorafenib, cisplatin, paclitaxel, and gemcitabine, were calculated for each TCGA-LIHC project sample using the R packages limma, oncoPredict, and parallel. The differences in IC₅₀ between the high- and low-risk groups were compared.

2.10 Statistical analysis

All data were analyzed using the R software (version 4.3.0). The statistical significance was determined as follows: *p < 0.05, **p < 0.01, and ***p < 0.001. Spearman’s correlation was used to evaluate the associations between numeric variables. Survival curves were generated using the KM assay, and the significance of statistical differences was determined using the log-rank test.

3.1 The expression of prognosis-related expression of lncRNAs in co-expression with disulfidptosis-related genes in HCC

A flowchart of the study is shown in Fig. 1. In total, 16,876 lncRNAs were identified in the TCGA-LIHC database. A series of 10 disulfidptosis-related genes (GYS1, NDUFS1, OXSM, LRPPRC, NDUFA11, NUBPL, NCKAP1, RPN1, SLC3A2 and SLC7A11) were identified based on previous literature, and a disulfidptosis-associated lncRNA co-expression network was constructed to identify disulfidptosis-associated lncRNAs. As a result, 746 disulfidptosis-associated lncRNAs (|cor|>0.4, p-value < 0.001) were detected by screening (Fig. 2a). A total of 65 lncRNAs associated with the prognosis of patients with HCC were initially screened using univariate Cox regression analysis, and forest plots were mapped (Fig. 2b). The initially screened variables were further screened and analyzed by LASSO regression to reduce data overfitting, and the Lasso regression coefficient profiles were plotted (Fig. 2c and d). Multivariate Cox regression analysis was performed to construct a risk model and evaluate the contribution of each lncRNA as a prognostic factor in the OS of patients with HCC. Finally, the prognostic signature of HCC was constructed using the five key DRLs, which included MKLN1-AS, TMCC1-AS1, AL603839.2, AC245060.7, and AL049840.3. Riskscore = 0.58515791763298*MKLN1-AS + 0.521536701804474*TMCC1-AS1 + 0.580776185231736*AL603839.2 + 1.05328540222151*AC245060.7 + 0.594960826047964*AL049840.3. The relationship between the five DRLs and disulfidptosis-related genes is shown in Fig. 2e.

3.2 Predicting the prognosis of HCC patients with the prognostic signature

Based on the median risk score, trained patients were assigned to high-risk and low-risk groups for survival analysis. The results showed that OS was significantly better in low-risk patients than in patients in high-risk group (P < 0.001; Fig. 3d). Mortality in patients with HCC increased with increasing risk score (Fig. 3a and b). The same phenomenon was observed in the test group and all patients (Fig. 3e, f, h, i, j, and l). MKLN1-AS, TMCC1-AS1, AL603839.2, AC245060.7 and AL049840.3 expressed more in the high-risk group than in the low-risk group for training, testing, and whole sets (Fig. 3c, g, and k), suggesting that these DRLs could be poor prognostic predictors. In addition, the PFS was significantly better in low-risk patients than in high-risk group among all patients. (P < 0.001; Fig. 3m). Simultaneously, the probability of survival and clinical characteristics of HCC patients were compared according to age, gender and TNM stage. High-risk patients had shorter OS than low-risk patients, regardless of the clinical variables (Fig. 4). Based on the results obtained, our prognostic signature can be applied to different clinical variables.

3.3 The prognostic signature could be a robust prognostic factor to predict clinical outcomes for HCC patients

The risk score was an independent prognostic factor for patients with HCC based on univariate and multivariate Cox regression analyses (Fig. 5a and b). The area under the ROC curve (AUC) was 0.767, 0.685, and 0.629 for 1, 3, and 5 years, respectively (Fig. 5c). ROC curves were plotted for the risk score and different clinical characteristics, and it was found that the AUC for the risk score was higher than the AUC scores for TMN, age, and gender, indicating a relatively accurate predictive power for this risk score (Fig. 5d). We also found that the C-index values of the risk score were higher than those of other clinical characteristics such as age, gender, and disease stage (Fig. 5e).

3.4 Nomogram and PCA

To enhance the prediction of patient survival rates, we developed a nomogram and used it for plotting. Furthermore, we assessed the accuracy of the predictions by generating a calibration plot that demonstrated concordance between the predicted and actual probabilities. The calibration plot served as evidence that the predicted probabilities obtained from the nomogram aligned well with the observed probabilities, reinforcing the reliability of our predictive model (Fig. 6a and b). To investigate the differences between the high- and low-risk groups in terms of the four distinct expression profiles, we employed PCA. The results of this analysis revealed that these risk lncRNAs exhibited consistent patterns, indicating their potential utility in constructing a reliable signature. (Fig. 6c-f).

3.5 GO, KEGG and GSEA for the high- and low-risk groups

First, DEGs were defined based on the mean expression of high- and low-risk samples (p < 0.05, |log2 (fold-change) |≥1). To explore the biological properties of DEGs, we performed GO and KEGG enrichment analyses using the R package clusterprofiler. As indicated by biological process (BP) terminology, DEGs contribute significantly to "organelle fission", "nuclear division, and "mitotic cell cycle phase transition". In the cellular component (CC) domain, "chromosomal region", apical part of cell" and "condensed chromosome" were significantly enriched. DEGs were enriched in molecular functions (MFs) associated with "tubulin binding", "ATP hydrolysis activity, and "microtubule binding" (Fig. 7a and b). Additionally, KEGG analysis showed that DEGs were enriched in the "Cell cycle", "Cellular senescence" and "Protein digestion and absorption" in both low and high-risk groups (Fig. 7c and d). In the high-risk group, moreover, "Cell cycle" and "Dilated cardiomyopathy" were activated (Fig. 7e). In low-risk patients, there was a significant enrichment in " beta-alanine metabolism" and " fatty acid metabolism" (Fig. 7f).

3.6 Estimate the difference in the immune microenvironment landscape between the high- and low-risk groups

Based on the results obtained using the ESTIMATE algorithm, stromal, immune, and ESTIMATE scores were calculated for each patient in the study. By comparing the scores between the two groups (low- and high-risk), they observed certain discrepancies. As shown in Fig. 8a, the low-risk group had significantly higher stromal and ESTIMATE scores than the high-risk group. This suggests that low-risk patients have a higher abundance of stromal and immune cells in their tumor microenvironment. Although the low-risk patients showed elevated immune scores, the differences between the two groups were not statistically significant. This indicates that immune activity may not be a distinguishing factor between low- and high-risk patients. In Fig. 8b, the researchers performed immune-related function comparisons between the two groups. They found that low-risk HCC patients exhibited higher levels of cytolytic activity, dendritic cells (DCs), neutrophils, natural killer (NK) cells, and type II interferon (IFN) response than the high-risk group. In contrast, macrophages, major histocompatibility complex (MHC) class I expression, and regulatory T cells (Tregs) were higher in the high-risk group. These findings suggest a more favorable immune profile in low-risk patients, characterized by increased anti-tumor immune responses. To further investigate the landscape of immune cell infiltration, researchers employed the CIBERSORT algorithm. This algorithm provided information about the relative percentages of different immune cell types in the tumor samples of all HCC patients (Fig. 8d). Specifically, they observed a significant reduction in the infiltration rate of M0 macrophages in the low-risk group compared with that in the high-risk group (Fig. 8c). This indicates that the composition of immune cells within the tumor microenvironment varies between low- and high-risk HCC patients. These findings collectively suggest that low-risk HCC patients have a more favorable tumor microenvironment, characterized by higher stromal and immune scores, increased cytolytic activity, elevated levels of DCs, neutrophils, NK cells, and type II IFN responses, as well as reduced infiltration of M0 macrophages. However, macrophages, MHC class I expression, and Tregs were more abundant in the high-risk group.

3.7 Mutational landscape for HCC and TMB together with TIDE analysis

To analyze the changes in somatic mutations in the high- and low-risk groups, we downloaded somatic mutation data from TCGA database. The 15 most highly mutated genes were TP53, CTNNB1, TTN, MUC16, PCLO, ALB, RYR2, APOB, XIRP2, LRP1B, CSMD3, OBSCN, ABCA13, HMCN1, and FLG (Fig. 9a and b). TP53 and LRP1B mutations were more common in the high-risk group than in the low-risk group (Fig. 9a and b). However, from an overall perspective, the low- and high-risk groups did not show a significant difference in TMB (Fig. 9c). The TIDE tool was used to predict the response to cancer immunotherapy. Furthermore, by comparing TIDE scores between groups with high- and low-risk scores, it was observed that high-risk patients with HCC exhibited a substantial infiltration of dysfunctional and immune-excluded T cells in their tumor immune microenvironment (Fig. 9d). This finding implies that, in contrast to low-risk patients, high-risk patients have reduced sensitivity to immune checkpoint blockade therapy.

The survival analysis illustrated in Fig. 9e demonstrates that patients with HCC who have high TMB experience significantly shorter survival times than those with low TMB, as depicted by Kaplan-Meier curves. Furthermore, among HCC patients, those classified as high-risk with elevated TMB exhibited the shortest survival time, whereas low-risk patients with low TMB showed the most favorable prognosis, as depicted in Fig. 9f.

3.8 Prediction of chemotherapy and targeted therapy response in different risk groups

Sorafenib, cisplatin, paclitaxel, and gemcitabine are commonly used anti-HCC agents. The IC₅₀ values of these drugs in patients with different risk signatures were analyzed to investigate the possible efficacy of these drugs in patients with HCC. Figures 10a-d show the IC₅₀ values of sorafenib, cisplatin, paclitaxel, and gemcitabine in the low-risk and high-risk groups. The results showed that patients with HCC in the high-risk group had higher IC₅₀ values for sorafenib, cisplatin, and gemcitabine, suggesting that patients in the low-risk group are more likely to benefit from these conventional drugs. In addition, patients with HCC in the low-risk group had a higher IC₅₀ for paclitaxel, suggesting that patients in the high-risk group may benefit more from this drug. What stands out in this section is the ability of the risk signature to identify HCC patients who may benefit from conventional anti-HCC drugs.

In this study, HCC data were obtained from the TCGA website. lncRNAs associated with a process called disulfidptosis have been identified. Several analyses, including univariate Cox analysis, Lasso analysis, and multivariate COX analysis, were conducted on these DRLs. As a result, a prognostic signature comprising 5 lncRNAs was established. These lncRNAs include MKLN1-AS, TMCC1-AS1, AL603839.2, AC245060.7, and AL049840.3. Both univariate and multivariate COX analyses demonstrated that the risk score based on these DRLs could serve as independent prognostic indicators for patients with HCC. Additional evaluation measures, such as ROC curves, C-index, survival curves, nomograms, and PCA results, were employed to assess the prognostic performance of the prognostic signature. Overall, these analyses indicate that the established prognostic signature exhibited strong predictive capabilities for determining the prognosis of patients with HCC.

Few studies have investigated disulfidptosis in HCC. In a previous study, a risk score for disulfidptosis-related genes was established to predict prognosis and guide clinical treatment[19]. Although there are many studies on the use of lncRNAs as molecular biomarkers to predict the prognosis of HCC patients, no study has systematically used DRLs as molecular biomarkers to predict the prognosis of HCC patients[20–23]. To our knowledge, this is the first study of the prognostic signature of DRLs to predict prognosis in patients with HCC. In this study, a prognostic signature with DRLs was developed and validated using data mining.

There are few studies on the five DRLs in HCC. Aberrant MKLN1-AS expression is associated with HCC development and progression. Several studies have shown that upregulation of MKLN1-AS is associated with the proliferation, migration, and invasion of HCC[24]. In addition, the expression level of MKLN1-AS may be associated with the prognosis and prediction of HCC and may serve as a potential biomarker[25]. Overexpression of MKLN1-AS enhanced the stability of YAP1 mRNA, which is required for the oncogenic activity of MKLN1-AS. However, the exact mechanism of action of MKLN1-AS in HCC is unclear, and further studies are needed to gain insights into its function and related pathways.

Studies have shown that abnormal expression of TMCC1-AS1 is associated with the development and progression of HCC. Several studies have shown that the expression level of TMCC1-AS1 is significantly increased in HCC tissues. The high expression of TMCC1-AS1 may be associated with malignant behaviors such as cell proliferation, invasion, and metastasis in HCC[26]. The exact mechanism of action and function of TMCC1-AS1 are not fully understood. Further studies may help reveal the regulatory network and molecular pathways of TMCC1-AS1 in HCC. In addition, the expression level of TMCC1-AS1 may also serve as a potential prognostic and predictive indicator for hepatocellular carcinoma, providing new targets and strategies for the diagnosis and treatment of this disease.

The detailed roles and functions of AL603839.2, AC245060.7 and AL049840.3 in hepatocellular carcinoma are not yet clear. Further studies are required to gain insight into its biological role, regulatory mechanisms, and relationship with hepatocellular carcinogenesis and progression. Investigating the role of lncRNAs in hepatocellular carcinoma has become a current research trend. For non-coding RNAs such as AL603839.2, AC245060.7, and AL049840.3, scientists may explore their expression patterns, interaction networks, and potential biological functions in hepatocellular carcinoma through transcriptomics, functional studies, and bioinformatics to improve our understanding of the pathogenesis of hepatocellular carcinoma and provide new clues and targets for the diagnosis and treatment of this disease.

We then performed GSEA to explore the role of differential gene expression in tumor behavior across the risk groups. The analysis revealed that several biological processes and signaling pathways were significantly enriched in the high-risk group, including the cell cycle, dilated cardiomyopathy, ECM-receptor interaction, hematopoietic cell lineage, and neuroactive ligand receptor interaction pathways. In HCC, abnormal regulation of the cell cycle may lead to uncontrolled cell proliferation and division, thereby promoting tumor growth and progression[27]. This suggests that cell cycle regulation may be an important mechanism in the development and progression of hepatocellular carcinoma in high-risk groups. Furthermore, there was an enrichment of biological processes associated with extracellular matrix interactions in the high-risk group of HCC. This suggests that HCC development of hepatocellular carcinoma may be influenced by interactions between tumor cells and surrounding tissues, such as cell adhesion, migration, and infiltration[28]. These interactions may play an important role in the invasion and metastasis of hepatocellular carcinoma. In the low-risk group of HCC, GSEA analysis showed enrichment of the following metabolic pathways and functional modules: β-alanine metabolism, fatty acid metabolism, glycine, serine, and threonine metabolism, major bile acid biosynthesis, and proteasome. This implies that metabolic disorders play a crucial role in the development of liver cancer in the low-risk group, and that targeting disordered metabolic-related signaling pathways may be a future research direction for HCC in the low-risk group. These findings have the potential to be applied to individualized targeted anti-cancer therapies for different risk groups of patients with HCC.

As an increasing number of immune checkpoint inhibitors are approved for clinical use, the role of immunotherapy in HCC treatment is becoming increasingly significant[29]. The effectiveness of immunotherapy in HCC can be influenced by various factors, including TME, immune cell infiltration, TMB, and expression of immune checkpoint molecules[30]. Researchers are actively studying these factors to identify reliable biomarkers that can help predict patient responses to immunotherapy.

Immunological reactions specific to tumors can be triggered by neoantigens resulting from somatic alterations. However, immunological checkpoints can suppress these reactions. One way to identify individuals who might benefit from checkpoint blockade is by assessing potential neoantigens or surrogates such as TMB. TMB quantifies all non-synonymous coding mutations in the tumor exome, which refers to the total number of somatic mutations in each coding region of the tumor genome[31]. It has been suggested that highly mutated tumors can generate numerous neoantigens, some of which may enhance T cell reactivity. Therefore, it is hypothesized that malignancies with higher mutation levels are more likely to show improved responses to immune checkpoint blockade therapy[32]. Although there is a correlation between high TMB and immunotherapy effectiveness, the use of TMB as a diagnostic predictor in clinical trials has not been widely adopted because of conflicting findings. The efficacy of TMB as a biomarker for predicting responsiveness to immunotherapy may be influenced by variations in the methods used to evaluate and interpret TMB[33].

The TIDE tool was used to predict the response to cancer immunotherapy[34]. By comparing TIDE scores between groups with high- and low-risk scores, it was observed that high-risk patients with HCC exhibited a substantial infiltration of dysfunctional and immune-excluded T cells in their tumor immune microenvironment. This finding implies that, in contrast to low-risk patients, high-risk patients with HCC have a reduced sensitivity to immune checkpoint blockade.

This study had several limitations that should be acknowledged. First, to enhance the credibility of our findings, it would be beneficial to validate the risk signature by using independent databases. It is important to note that the lncRNAs obtained from TCGA database may not completely align with those from other databases because of variations in chip platforms and recording methods. We attempted to consult other datasets, such as SEER, ICGC, and TANRIC; however, we did not find a dataset that contained both clinical data and the corresponding expression of the lncRNAs. Therefore, validating the effectiveness of our signature using additional datasets would greatly strengthen our results. Second, although we conducted an enrichment analysis and made assumptions regarding the function of different risk groups, further research is needed to explore the underlying mechanisms. Third, conducting in vitro experiments would provide valuable evidence to support our findings. However, it is important to note that acquiring sufficient survival time for fresh tissue samples, which is necessary for lncRNA expression testing, within a short timeframe is challenging. Incorporating bioinformatics analysis results with clinical data in future research would further enhance our understanding of these findings. Lastly, while the total sample size of 424 was substantial, obtaining a larger sample size would yield more robust and convincing results. Increasing the sample size would help to minimize potential bias and enhance the statistical power of the study.

In summary, a new prognostic signature was developed based on the five DRLs to predict the prognosis of patients with HCC. Furthermore, they offer insights into the interplay between disulfidptosis, TME, and immunotherapy in the context of HCC. This knowledge can guide future research efforts and potentially lead to the development of targeted therapies that exploit disulfidptosis and harness the immune system to treat HCC more effectively.

Acknowledgments

We thank TCGA database for making the data available.

Author contributions

In this research YCY, CZ, and XM made significant contributions to the research conception and design of the study and revised the manuscript accordingly. NX, ZZM, HL, and DD collected and analyzed the data and wrote the manuscript. LZ, PYC, and MLL provided critical revisions. All authors agree to be accountable for all aspects of the work, and questions relevant to accuracy or integrity were addressed and surveyed appropriately. The final manuscript has been read and approved by all the authors.

Funding

This study was supported by the school-level key projects of Bengbu Medical College (2021byzd109).

Availability of data and materials

Data openly available in a public repository. The datasets generated and analyzed during the current study are available in TCGA database (https://portal.gdc.cancer.gov/).

Ethics approval and consent to participate

Not applicable.

Consent for publication

Not applicable.

Competing interests

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

Sung H, Ferlay J, Siegel RL, Laversanne M, Soerjomataram I, Jemal A, Bray F: Global Cancer Statistics 2020: GLOBOCAN Estimates of Incidence and Mortality Worldwide for 36 Cancers in 185 Countries. CA: a cancer journal for clinicians 2021, 71(3):209–249.
Brown ZJ, Tsilimigras DI, Ruff SM, Mohseni A, Kamel IR, Cloyd JM, Pawlik TM: Management of Hepatocellular Carcinoma: A Review. JAMA surgery 2023, 158(4):410–420.
Machesky LM: Deadly actin collapse by disulfidptosis. Nature cell biology 2023, 25(3):375–376.
Liu X, Nie L: Actin cytoskeleton vulnerability to disulfide stress mediates disulfidptosis. 2023, 25(3):404–414.
Zheng P, Zhou C, Ding Y, Duan S: Disulfidptosis: a new target for metabolic cancer therapy. 2023, 42(1):103.
Hu G, Niu F, Humburg BA, Liao K, Bendi S, Callen S, Fox HS, Buch S: Molecular mechanisms of long noncoding RNAs and their role in disease pathogenesis. Oncotarget 2018, 9(26):18648–18663.
Quinn JJ, Chang HY: Unique features of long non-coding RNA biogenesis and function. Nature reviews Genetics 2016, 17(1):47–62.
Li Y, Egranov SD, Yang L, Lin C: Molecular mechanisms of long noncoding RNAs-mediated cancer metastasis. Genes, chromosomes & cancer 2019, 58(4):200–207.
Schmitt AM, Chang HY: Long Noncoding RNAs in Cancer Pathways. Cancer cell 2016, 29(4):452–463.
Liz J, Esteller M: lncRNAs and microRNAs with a role in cancer development. Biochimica et biophysica acta 2016, 1859(1):169–176.
Conesa A, Madrigal P, Tarazona S, Gomez-Cabrero D, Cervera A, McPherson A, Szcześniak MW, Gaffney DJ, Elo LL, Zhang X et al: A survey of best practices for RNA-seq data analysis. Genome biology 2016, 17:13.
Ritchie ME, Phipson B, Wu D, Hu Y, Law CW, Shi W, Smyth GK: limma powers differential expression analyses for RNA-sequencing and microarray studies. Nucleic acids research 2015, 43(7):e47.
Gustavsson EK, Zhang D, Reynolds RH, Garcia-Ruiz S, Ryten M: ggtranscript: an R package for the visualization and interpretation of transcript isoforms using ggplot2. Bioinformatics (Oxford, England) 2022, 38(15):3844–3846.
Blanche P, Dartigues JF, Jacqmin-Gadda H: Estimating and comparing time-dependent areas under receiver operating characteristic curves for censored event times with competing risks. Statistics in medicine 2013, 32(30):5381–5397.
Mogensen UB, Ishwaran H, Gerds TA: Evaluating Random Forests for Survival Analysis using Prediction Error Curves. Journal of statistical software 2012, 50(11):1–23.
Gu Z, Gu L, Eils R, Schlesner M, Brors B: circlize Implements and enhances circular visualization in R. Bioinformatics (Oxford, England) 2014, 30(19):2811–2812.
Yu G, Wang LG, Yan GR, He QY: DOSE: an R/Bioconductor package for disease ontology semantic and enrichment analysis. Bioinformatics (Oxford, England) 2015, 31(4):608–609.
Yoshihara K, Shahmoradgoli M, Martínez E, Vegesna R, Kim H, Torres-Garcia W, Treviño V, Shen H, Laird PW, Levine DA et al: Inferring tumour purity and stromal and immune cell admixture from expression data. Nature communications 2013, 4:2612.
Wang T, Guo K, Zhang D, Wang H, Yin J, Cui H, Wu W: Disulfidptosis classification of hepatocellular carcinoma reveals correlation with clinical prognosis and immune profile. International immunopharmacology 2023, 120:110368.
Jin C, Zhao JS, Huang XQ, Yang XZ, Niu FY, Lin JR, Ma L, Shi YX, Li XS, Jiang P et al: A somatic mutation-derived LncRNA signatures of genomic instability predicts the prognosis and tumor microenvironment immune characters in hepatocellular carcinoma. Hepatology international 2022, 16(5):1220–1233.
Wang T, Yang Y, Sun T, Qiu H, Wang J, Ding C, Lan R, He Q, Wang W: The Pyroptosis-Related Long Noncoding RNA Signature Predicts Prognosis and Indicates Immunotherapeutic Efficiency in Hepatocellular Carcinoma. Frontiers in cell and developmental biology 2022, 10:779269.
Xia W, Zeng C, Zheng Z, Huang C, Zhou Y, Bai L: Development and Validation of a Novel Mitochondrion and Ferroptosis-Related Long Non-Coding RNA Prognostic Signature in Hepatocellular Carcinoma. Hepatology international 2022, 10:844759.
Yang X, Mei M, Yang J, Guo J, Du F, Liu S: Ferroptosis-related long non-coding RNA signature predicts the prognosis of hepatocellular carcinoma. Aging 2022, 14(9):4069–4084.
Guo C, Zhou S, Yi W, Yang P, Li O, Liu J, Peng C: SOX9/MKLN1-AS Axis Induces Hepatocellular Carcinoma Proliferation and Epithelial-Mesenchymal Transition. 2022, 60(6):1914–1933.
Gao W, Chen X, Chi W, Xue M: Long non–coding RNA MKLN1–AS aggravates hepatocellular carcinoma progression by functioning as a molecular sponge for miR–654–3p, thereby promoting hepatoma–derived growth factor expression. International journal of molecular medicine 2020, 46(5):1743–1754.
Chen C, Su N, Li G, Shen Y, Duan X: Long non-coding RNA TMCC1-AS1 predicts poor prognosis and accelerates epithelial-mesenchymal transition in liver cancer. Oncology letters 2021, 22(5):773.
Ng CKY, Dazert E, Boldanova T, Coto-Llerena M, Nuciforo S, Ercan C, Suslov A, Meier MA, Bock T, Schmidt A et al: Integrative proteogenomic characterization of hepatocellular carcinoma across etiologies and stages. Nature communications 2022, 13(1):2436.
Liu P, Kong L, Liu Y, Li G, Xie J, Lu X: A key driver to promote HCC: Cellular crosstalk in tumor microenvironment. Frontiers in oncology 2023, 13:1135122.
Pinter M, Scheiner B, Pinato DJ: Immune checkpoint inhibitors in hepatocellular carcinoma: emerging challenges in clinical practice. The lancet Gastroenterology & hepatology 2023.
Donne R, Lujambio A: The liver cancer immune microenvironment: Therapeutic implications for hepatocellular carcinoma. Hepatology (Baltimore, Md) 2023, 77(5):1773–1796.
Rizzo A, Ricci AD: PD-L1, TMB, and other potential predictors of response to immunotherapy for hepatocellular carcinoma: how can they assist drug clinical trials? Expert opinion on investigational drugs 2022, 31(4):415–423.
Marabelle A, Fakih M, Lopez J, Shah M, Shapira-Frommer R, Nakagawa K, Chung HC, Kindler HL, Lopez-Martin JA, Miller WH, Jr. et al: Association of tumour mutational burden with outcomes in patients with advanced solid tumours treated with pembrolizumab: prospective biomarker analysis of the multicohort, open-label, phase 2 KEYNOTE-158 study. The Lancet Oncology 2020, 21(10):1353–1365.
Wang X, Qiao Z, Aramini B, Lin D, Li X, Fan J: Potential biomarkers for immunotherapy in non-small-cell lung cancer. Cancer metastasis reviews 2023.
Li X, Kang J, Yue J, Xu D, Liao C, Zhang H, Zhao J, Liu Q, Jiao J, Wang L et al: Identification and validation of immunogenic cell death-related score in uveal melanoma to improve prediction of prognosis and response to immunotherapy. Aging 2023, 15(9):3442–3464.

No competing interests reported.

Download PDF

Version 1

posted

You are reading this latest preprint version

Establishment of a prognostic signature for hepatocellular carcinoma using disulfidptosis-related lncRNAs

Status:

Version 1

Abstract

Background

Method

Results

Conclusions

Figures

1 Introduction

2 Materials and methods

2.1 Datasets and patients

2.2 Identification of DRLs

2.3 Establishment of the DRLs prognostic signature

2.4 Assessment of the DRLs prognostic signature

2.5 Nomogram and principal component analysis (PCA)

2.6 Enrichment function analysis

2.7 Immune-related function analysis

2.8 Tumor mutation burden (TMB) and tumor immune dysfunction and exclusion (TIDE) analysis

2.9 Prediction of anti-HCC drug response

2.10 Statistical analysis

3 Results

3.1 The expression of prognosis-related expression of lncRNAs in co-expression with disulfidptosis-related genes in HCC

3.2 Predicting the prognosis of HCC patients with the prognostic signature

3.4 Nomogram and PCA

3.5 GO, KEGG and GSEA for the high- and low-risk groups

3.6 Estimate the difference in the immune microenvironment landscape between the high- and low-risk groups

3.7 Mutational landscape for HCC and TMB together with TIDE analysis

3.8 Prediction of chemotherapy and targeted therapy response in different risk groups

4 Discussion

Declarations

References

Additional Declarations

Status:

Version 1