Hepatocellular carcinoma (HCC), the most common form of liver cancer, is among the highly prevalent types of malignancy all over the world and ranks the third cause of cancer-related death 1. In recent years, the mortality and occurrence rate of HCC has been reported to be increased 2. Because of the rapid development and metastasis, the prognosis of patients with HCC is poor 3, HCC patients usually diagnose at an advanced stage 4, and the 5-year survival rate is as low as 30% 5, 6. With the rapid development of gene sequencing technology, we have a deeper understanding of the molecular pathogenesis of HCC 7, 8. High-throughput analysis of a large number of samples shows that accumulated data can be used to identify key biomarkers related to HCC progression 9. However, the number of biomarkers associated with HCC prognosis is still limited. Therefore, it's an urgent need to explore a novel method to guide clinical treatments and improve the prognosis of HCC.The peroxisome is a monolayer membrane organelle, which exists in all kinds of eukaryotic cells and mediates varieties of biological processes. Peroxisome has been considered an important site for the generation and removal of free radicals in cells 10. The balance mechanism of free radical generation and scavenging in the peroxisome is essential to maintain the normal function of cells 11, 12. When the peroxisome undergoes oxidative stress, free radicals will be produced, which are closely associated with the occurrence and development of human tumors, and tumor progression may be influenced in multiple aspects 13-15. Moreover, reduced or disappeared peroxisome was found in lung cancer and liver cancer cells 16, and changes in the oxidative stress pathway of peroxisomes in tumor cells play a very important role in mediating the development of liver tumors 13. Therefore, an in-depth understanding of the peroxisome process in HCC could provide an important solution for the development of a new treatment method. In recent years, gene chips and high-throughput sequencing technology have made great progress, which implies that the genetic signature of the peroxisome can be used to predict the overall survival (OS) of HCC.In this present study, a 9-gene risk model of HCC with a good performance in prognostic prediction was established and it was validated by the external validation cohort. The model was further evaluated under various clinical settings including survival, clinical-pathological characteristics, immune infiltration, immune pathways, immune checkpoints, multidrug resistance-related genes and chemotherapy. Besides, the expression of prognostic genes between HCC and adjacent normal tissues was validated in an independent sample cohort. The results showed that the model could be used as an independent prognostic evaluation index for HCC patients and could provide a new perspective for individual treatment for HCC patients.
Materials and Methods
Data collection
Transcriptome profiling data and related clinical data were downloaded from the Cancer Genome Atlas Liver Hepatocellular Carcinoma dataset (TCGA-LIHC, https://portal.gdc.cancer.gov/) as the train cohort. The International Cancer Genome Consortium Liver Hepatocellular Carcinoma dataset Japan (ICGC-LIRI-JP, https://dcc.icgc.org/) was regarded as the validation cohort. The downloaded profiles all comply with the TCGA and ICGC data access rules. The inclusion standard was that the patients pathologically diagnosed as hepatocellular carcinoma. The exclusion criteria were: 1) patients with hepatocellular metastasis and cholangiocarcinoma; 2) patients with co-existing cancers of other tissues; 3) Patients whose HCC samples lacked RNA sequencing data; 4) patients who lack time and status to survive; and 5) the follow-up with 0 day. A peroxisome-related gene set including 104 genes was retrieved from the Molecular Signatures Database (MsigDB, https://www.gsea-msigdb.org/gsea/index.jsp), (.
Identification of differentially expressed genes (DEGs) between HCC and adjacent non-tumorous tissues
The R package “limma” was used to identify DEGs between 365 HCC and adjacent non-tumorous tissues with the false discovery rate (FDR) < 0.05. Protein-protein interaction (PPI) networks of differently expressed peroxisome-related genes were constructed by using the STRING database.
Establishment and validation of a peroxisome-related prognostic model
We firstly carried out univariate Cox regression analysis to find prognostic genes among the DEGs and identify the prognostic value of the DEGs for OS. Subsequently, the least absolute shrinkage selection operator (LASSO) Cox regression analysis was further performed to narrow down the number of candidate genes. The risk score was calculated by the following equation: Score =e sum (each gene's expression × corresponding coefficient). And all HCC patients were divided into high- and low-risk groups on account of the median value of risk score. To validate the reliability of this classification, we performed principal component analysis (PCA) and t-distributed stochastic neighbor embedding (t-SNE) to display the distribution of different groups using the "Rtsne" R package. The Kaplan-Meier curve analysis (log-rank test) was used to evaluate the survival difference between two groups via the "Survival" R package. The time-dependent receiver operating characteristic (ROC) curves were plotted, and the area under the curve (AUC) values were calculated with the application of the “SurvivalROC” R package to assess the predictive ability. Then, we further validated the predictive capacity of this model in the ICGC cohort.
Identification of Independent prognostic factors for OS in HCC
Univariate and multivariate Cox regression analyses were performed to confirm whether the model could be independent of other clinical parameters (including gender, age, histological grade, and tumor stage) in predicting OS of HCC patients.
Evaluation of immune status in different risk groups
We calculated the relative infiltrations of immune cell types and immune functional pathways by single sample gene set enrichment analysis (ssGSEA), analyzing tumor microenvironment (TME) in different risk groups. Six immune subtypes were defined to measure immune infiltrates in immune infiltration 17. The association between the risk score and immune infiltration subtypes in TME was analyzed by analysis of variance (ANOVA). Tumor stemness features extracted from HCC transcriptomic were measured by stemness score based on RNA methylation (RNAss) 18, and Wilcoxon test was carried out to analyze the RNAss in different risk groups.
Immune checkpoint genes, cell cycle-related genes, multidrug resistance genes expression in different risk groups
Wilcoxon test was carried out to analyze the expression of immune checkpoints genes, cell cycle-related genes and multidrug resistance genes in different risk groups, respectively. Pearson correlation was used to analyze the correlation between the risk scores and the genes.
GO and KEGG Functional Enrichment Analyses
To explore main biological functions and signaling pathways in different risk groups, Gene Ontology (GO) and Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway enrichment analyses were conducted through gene set enrichment analysis (GSEA) in GSEA software 4.1. FDR < 0.05 was regarded as statistically significant.
Anti-tumor drugs sensitivity analysis
The NCI-60 database was accessed through the CellMiner interface, which contains 60 different cancer cell lines from 9 different types of tumors (https://discover.nci.nih.gov/cellminer). Pearson correlation was used to analyze the correlation between the prognostic gene expression and the 263 anti-tumor drugs proved by the FDA or obtained from clinical trials. The 263 drugs were showed in .
qRT-PCR analysis
The real-time quantitative-polymerase chain reaction (qRT-PCR) assays were used to validate the mRNA expression levels of the 9 prognostic genes screened above in 20 paired samples of HCC and adjacent non-tumorous tissues. The samples were from the First Affiliated Hospital of Wenzhou Medical University. And this study was approved by the Review of Ethics Committee in Clinical Research of the First Affiliated Hospital of Wenzhou Medical University. Written informed consents were obtained from all patients for the use of the biospecimens for research purposes. Total RNA was extracted with the Trizol reagent following the manufacturer's instructions (Servicebio). Then cDNA was synthesized using reverse transcriptase that provided by Thermo. RT-PCR analysis was performed using FastStart Universal SYBR Green Master (Roche) by ABI StepOne (Applied Biosystems). The primers used are shown in .
Immunohistochemical analysis
The protein expression of prognostic peroxisome-related genes between HCC and adjacent non-tumorous tissues was evaluated in 10 paired samples by the Immunohistochemistry (IHC) method. Samples were fixed with 10% formalin, dehydrated with alcohol, xylene transparentized, embedded with paraffin immersion wax, and sectioned with paraffin microtome. The sections were dewaxed with conventional xylene and dehydrated with gradient alcohol. After the antigen was repaired, the endogenous peroxidase was blocked by H2O2 (3%) for 25 minutes, and the nonspecific bindings were blocked by 10% rabbit serum or 3% bovine serum albumin for 30 minutes. IHC staining was carried out for the protein expression of ABCC5, BCL10, FDPS, ITGB1BP1, MSH2, PABPC1, PRDX1, SLC25A19, and YWHAH using specific primary antibodies at 4 °C overnight, followed by staining with species-specific secondary antibodies labeled with horseradish peroxidase. The information of the antibodies was provided in . After the slices were stained with diaminobenzidine and counterstained with hematoxylin, the sample was dehydrated, transparent, sealed, observed, and photographed.
Statistical analyses
All statistical analyses were performed in R software (Version 4.0.2). Gene expression levels between HCC and adjacent non-tumorous tissues were analyzed by the Wilcoxon test. Differences in proportions were assessed by the Chi-squared test. Kaplan-Meier analysis was used in different groups for OS. Univariate and multivariate Cox regression analyses were used to screen independent predictors for OS. Mann-Whitney test was used to compare the ssGSEA scores of immune cells or the activity immune pathways in different risk groups. Spearman correlation was carried out to analyze the associations between the risk score and immune checkpoint-related genes, cell cycle-related genes, and multidrug resistance-related genes. Pearson correlation was utilized to analyze the relationship of the prognostic gene expression to drug sensitivity. A two-tailed P < 0.05 was considered to be statistically significant.
Results
Figure shows the process of establishing the gene signature and the prognostic model of this study. According to the criteria for the inclusion and exclusion of HCC patients, we ultimately identified and included 365 patients from TCGA dataset and 231 patients from ICGC dataset. The detailed clinical characteristics are summarized in Table .
Identification of DEGs related to peroxisome in the TCGA cohort
More than half of the 104 peroxisome-related genes (71/104, 68.3%) were differentially expressed between tumor and adjacent non-tumorous tissues. Univariate Cox regression analysis showed that 29 DEGs were related to OS (Figure . Figure implies that 28 genes in the TCGA cohort had significantly prognostic relevance (excluded FABP6 gene that was not expressed in most samples). The interactions among these candidate genes were shown in PPI and correlation network (Figure .
Establishment of the prognostic model in the TCGA cohort
LASSO Cox regression analysis was performed to establish a prognostic model based on the above-mentioned 28 DEGs. Subsequently, a 9-gene signature was identified, consisting of ABCC5, BCL10, FDPS, ITGB1BP1, MSH2, PABPC1, PRDX1, SLC25A19, and YWHAH (. The risk score was calculated as follows: score = 0.238 * expression level of ABCC5+ 0.144 * expression level of BCL10 + 0.022 * expression level of FDPS + 0.137 * expression level of ITGB1BP1 + 0.026 * expression level of MSH2 + 0.011 * expression level of PABPC1 + 0.185 * expression level of PRDX1 + 0.031 * expression level of SLC25A19 +0.016 * expression level of YWHAH. The median risk score of the TCGA cohort served as the unified cut-off for dividing HCC patients into high- and low-risk groups (Figure . Moreover, we observed that the high-risk group had a significantly higher percentage of HCC patients with worse clinicopathological characteristics, such as an advanced tumor stage and a later histological grade (Table . As shown in Figure , HCC patients in the high-risk group presented a shorter survival time and more occurrences of death. PCA and t-SNE analyses showed that patients in the two groups were distributed in two different directions (Figure . Conformably, the Kaplan-Meier curve indicated that the prognosis of the high-risk group had a significantly shorter OS than that of the low-risk group (P<0.001) (Figure . Time-dependent ROC curves of the prognostic model at 1, 2, and 3 years were 0.760, 0.679, and 0.656, respectively, indicating a good predictive performance (Figure . According to the optimal cut-off expression value of each prognostic gene, survival analysis showed that the high expression levels of the 9 genes were all correlated to the poor prognosis (P<0.05) (. Moreover, implies that the expression of each prognostic gene in HCC was distinctly higher than that in adjacent non-tumorous tissues (P<0.001).
External validation of the prognostic model in the ICGC cohort
To verify whether the prognostic model is robust, we used the independent cohort (ICGC) for external validation. Similar to the results of the TCGA cohort, HCC patients were stratified into a high-risk group and a low-risk group based on the median value from TCGA (Figure . And the high-risk group showed significantly poorer OS relative to the low-risk group (Figure . PCA and t-SNE analyses also indicated that patients in two groups were distributed in discrete directions (Figure . The Kaplan-Meier curve demonstrated that HCC patients in the high-risk group had a shorter survival time (Figure . In the ICGC cohort, time-dependent ROC curves at 1, 2, and 3 years were 0.648, 0.613, and 0.637, respectively (Figure .
Independent prognostic value of the 9-gene signature
Univariate and multivariate Cox regression analyses were performed to examine the predictive capability of the prognostic model. As shown in Figure , univariate Cox regression analysis revealed that the risk score and the tumor stage were significantly associated with OS of HCC (P < 0.001). Further, multivariate Cox regression analysis indicated that the risk score was an independent prognostic factor affecting long-term survival in the TGCA cohort. Figure shows that the AUC of the risk score, the stage and the risk score combined with the tumor stage at 3-year OS is 0.675, 0.652 and 0.722, respectively. The results mentioned above showed that the combined model had better prediction accuracy for OS. The same prognostic significance of the risk score was verified using the data obtained from the ICGC cohort (Figure .
Risk score and prognostic genes in different clinical characteristics groups
The risk score distributed in different clinical characteristics groups (including gender, age, histological grade, and tumor stage) in HCC patients based on the TCGA and ICGC data is shown in Figure , respectively. The risk score was higher in the high histological grade and advanced tumor stage in the TCGA cohort, and a similar conclusion was seen in the ICGC cohort (There were no data about the histological grade of LICH in the ICGC cohort). Furthermore, combining prognostic gene expression levels with clinical characteristics of HCC patients indicated that the expression of MSH2 had significant differences in clinical characteristics (.
Relationship between risk score and immune status
Immune cells' infiltration in different risk groups was analyzed to investigate the impact of risk score on TME. In the high-risk group, the infiltration of aDCs, iDCs, Macrophages, and Th2 cells was significantly increased in the TCGA and ICGC cohorts (P< 0.01), implying that high expression of peroxisome-related genes may promote the release of these immune cells (Figure . In terms of immune function, we observed that type II IFN response in the low-risk group was significantly stronger than that in the high-risk group (Figure .Subsequently, we estimated the risk score in different immune subtypes in HCC patients. Six immune subtypes defined to measure immune infiltrates in tumor immune response, numbered from lowest to the highest relative abundance of cytotoxic cells, were C1 (wound healing), C2 (INF-γ dominant), C3 (inflammatory), C4 (lymphocyte depleted), C5 (immunologically quiet), and C6 (TGF-β dominant) 19. No patient sample belonged to the C5 immune subtype in HCC. As shown in Figure , the high-risk score was significantly correlated with C1 and C2, and the low-risk score was correlated with C3, C4, and C6. And the high levels of all the prognostic genes were positively related to the C1 and C2, showing their promoting role in the process of HCC. Correlations of the 9 prognostic genes with the 6 immune types were shown in .Tumor cells can gradually lose a differentiated phenotype and acquire progenitor and stem-cell-like features in the cancer progression 20. Study suggests that an abundant population of tumor cells with stemness features may be a signal of poor prognosis in HCC patients 21. Then the correlation between the risk model and tumor stemness measured by RNAss was explored. As shown in Figure , RNAss was higher in the high-risk group than that in the low-risk group, and RNAss was positively with the risk score. What's more, we observed that prognostic genes had a positive association with RNAss except for ABCC5, BCL10, and YWHAH (P< 0.0001) (. These results revealed that the high expression of prognostic genes in HCC correlated to increased cancer cell stemness was consistent with the fact that increased expression of prognostic genes supported worse survival.Immune checkpoint-associated genes play an important role in the process of cancer 20. We further explored correlations of the expression levels of immune checkpoint-associated genes in different risk groups. We found that PD-L2 and Galectin9 were overexpressed in the high-risk group compared with the low-risk group in the TCGA cohort (Figure . Furthermore, the expression levels of PD-L2 and Galectin9 were all positively correlated with the risk score in the TCGA cohort (Figure , implying that the prognostic model conducted in this study can distinguish the expression levels of the immune checkpoint-associated genes. What's more, the similar conclusion was shown in the ICGC cohort (Figure .
GO and KEGG functional enrichment analysis
To further explore whether biological functions and pathways are correlated with the risk score signature, GO and KEGG were carried out in different groups in the TCGA cohort. We found that some cancer-related gene sets were significantly gathered in HCC patients with a high-risk score. As shown in Figure , significant functions associated with tumorigenesis were enriched in the high-risk group defined by the 9-gene signature with the FDR<0.05, including Cell cycle G1-S phase transition, Regulation of autophagy, Apoptotic mitochondrial changes, Focal adhesion, Cell cycle pathway, Erbb signaling pathway, Notch signaling pathway, P53 signaling pathway, Wnt signaling pathway and Pathways in cancer. The detailed information is displayed in . We found that cell cycle related pathways were upregulated in the high risk group, so we further looked into cell cycle related genes in different groups. As shown in Figure , CDK2, CDK4, Cyclin and CDC25A were highly expressed in the high risk group, and all of them were positively associated with the risk score in the TCGA cohort. The results in the ICGC cohort were consistent with the TCGA cohort (Figure .
Analysis of the correlation between the risk model and multidrug resistance-related genes and chemotherapeutics
To evaluate the model in the clinic for HCC treatment, we further explored correlations of the expression levels of multidrug resistance-related gens and the risk score in different risk groups. We found that the expression of MRP1, MRP3 and MRP5 were higher in the high-risk group than that in the low-risk group (Figure . Furthermore, the expression levels of MRP1, MRP3 and MRP5 were all positively correlated with the risk score (Figure , implying that the prognostic model conducted in this study can distinguish the expression levels of multidrug resistance-related genes. Based on the above findings, correlations of the risk score and sensitivity of antitumor drugs were performed and the results in Figure showed that the prognostic genes were negative correlatively to some chemotherapy drug sensitivity. Among them, Fluorouracil, Doxorubicin, lenvatinib and epirubicin can be used in the clinical treatment of liver cancer, and Lenvatinib is a first-line targeted drug for the treatment of HCC. The more specific details are presented in and . Those findings indicated that the model could act as a potential predictor for multidrug resistance-related genes and chemosensitivity.
Validation of the differential expression of prognostic genes between HCC and adjacent non-tumorous tissues
To validate the different mRNA and protein expression levels of the 9 prognostic genes (ABCC5, BCL10, FDPS, ITGB1BP1, MSH2, PABPC1, PRDX1, SLC25A19, and YWHAH) between HCC and adjacent non-tumorous tissues, qRT-PCR and IHC were performed. As shown in Figure all the prognostic genes were highly expressed in HCC tissues than that in adjacent non-tumorous tissues, suggesting the same conclusion shown in the TCGA cohort and . Moreover, we conducted comparisons of expression of prognostic genes between normal and tumor samples across TCGA cancer types in Figure . Compared with the expression levels in normal samples, ABCC5, FDPS, ITGB1BP1, MSH2, PABPC1, PRDX1, SLC25A19 and YWHAH showed significantly high expression levels in most cancer types, while BCL10 was expressed lowly in most cancer types except for CHOL, LIHC, ESCA and STAD. These findings indicated that the expression of most prognostic genes in other tumors was consistent with that of HCC, which could further study the application value of this model in other tumors. At the same time, it also showed a significant inter-tumor heterogeneity considering the expression levels of some prognostic genes.
Discussion
HCC is a fatal disease with high incidence and poor prognosis 22, 23. A large number of HCC patients are diagnosed at an advanced stage, which means that effective treatments are mostly lost 24. The development of high-throughput sequencing provides an opportunity to identify biomarkers that predict prognosis for HCC and to enhance treatment to improve clinical outcomes for HCC 25, 26. Therefore, it is necessary to identify key molecular markers that can affect the prognosis of HCC, so as to better achieve individualized survival prediction with better accuracy.In the present study, gene expression data were retrieved from TCGA and ICGC databases to classify prognostic DEGs between HCC and adjacent non-tumorous tissues. Then, a prognostic risk model including nine genes, which were selected by univariate Cox and LASSO Cox regression, was established. The risk model consisted of ABCC5, BCL10, FDPS, ITGB1BP1, MSH2, PABPC1, PRDX1, SLC25A19, and YWHAH was effective and stable to predict the prognosis of HCC patients through external validations.Among the 9 prognostic genes, ABCC5 is aberrantly upregulated in several human malignancies 27-29 and is responsible for the multidrug resistance phenotype causing HCC treatment failure based on the drug efflux pumps 30-32. BCL10, a key participator in the regulation of DNA double-strand breaks repair 33, is commonly involved in promoting the growth and invasion of cancer cells 34-36. Overexpression of FDPS leads to activation of oncogenic signaling and changes in the prenylation of small GTPases 37. ITGB1 functions as an oncogene in different types of human cancers 38, 39 and the overexpression of ITGB1 is found to promote the growth and metastasis of HCC cells 40. As far as we know, there is no report about ITGB1BP and HCC. The team of Hinrichsen pointed that up-regulation of MSH2 is positively related to the occurrence and metastasis of HCC 41. Previous research strongly showed that PABPC1 plays a role in HCC and can accelerate cell proliferation 42. High expression of PRDX1 in HCC tissues corresponds to adverse clinical outcomes, and the mechanism may be related to promoting tumor angiogenesis and regulating cell migration and invasion 43-45. A recent report showed that SLC25A19 is up-regulated in 43 breast cancer specimens, indicating its roles in breast cancer 46. The members of YWHAH are reported to be overexpressed in the cancerous area of various malignancies and they contribute to the carcinogenesis, including HCC 47-49. Given the importance of the 9 peroxisome-related genes in kinds of cancer types, these genes might be potential prognostic biomarkers for HCC patients. However, the specific molecular mechanism of the 9 genes in HCC needs further exploration.Owing to the development of microarray and next-generation sequencing technologies, many multigene prognostic models have been developed to predict survival for HCC patients 50, 51. However, this is the first study about a prognostic model in HCC patients constructed using multiple peroxisome-related genes. The novel prognostic model was further evaluated under various clinical settings including survival. Based on the risk score signature, survival analysis displayed a significant difference in OS between high- and low-risk groups, and high-risk patients always encountered death earlier than low-risk patients. The ROC curves suggested that our prognostic model had good accuracy, and the AUC values of 1-, 2-, and 3-year showed the good predictive value of the model in both short- and long-term follow-ups. The findings above indicated that the prognostic model we constructed was capable of general application. Besides, the risk score was an independent prognostic indicator in HCC in both training set and testing set and the ability of prognostic prediction was further enhanced when the risk score combined with the tumor stage.Furthermore, the immune status was different between the two groups. It is noteworthy that the high-risk group had a higher infiltration proportion of aDCs, iDCs, macrophages, and Th2 cells, as well as higher expression levels of immune checkpoints (PD-L2 and Galectin9). On the contrary, the patients in the high-risk group had lower activity of type II IFN response, which is described as a key to activate cell-mediated immune responses to control intracellular pathogens 52, suggesting the low anti-tumor immune response in the high-risk group. In terms of these immune cells, immune checkpoints, and multidrug resistance-related genes, the function of tumor infiltration has also been reported to be related to survival in HCC 53. ANOVA showed that the more aggressive subtypes of immune infiltration (C1 and C2) were observed in the high-risk group, suggesting a positive correlation of prognostic gene expression with poor prognosis. Collectively, the evidence for the association between risk scores and immune highlighted the importance of the peroxisome-related gene signature in prognostic prediction and treatment for HCC patients. Compared with other prognostic models of HCC, our analysis provides clear additional evidence that the risk score model based on the prognostic genes is directly linked to immune infiltration, immune pathways and immune checkpoints. However, further research, including in vivo and in vitro validation, as well as clinical trials, is needed to evaluate the correlation between peroxisome-related genes and immune regulation more accurately in HCC.Cancer stem cells (CSCs) have been classified as a small subset of tumor cells with the characteristics to influence self-renewal and differentiation, which makes it hard to eliminate the tumor 54. CSCs have been identified in numerous solid tumors, such as breast cancer, colon cancer, and HCC 55-57. We further analyzed the relationship of the expression of peroxisome-related genes with RNAss in HCC. Interestingly, the findings were consistent with the fact that the expression of stem cells in various tumors correlates inversely with the outcome 58. This strongly suggested that the model we established had the ability to identify the score of CSCs and provided the possibility that targeting those genes might inhibit the process of HCC via affecting CSCs.Accumulating evidence has indicated that chemotherapy of HCC is facing drug resistance, which leads to unsatisfactory therapeutic effect 59. The high expression of multidrug resistance-related genes and the decrease of drug sensitivity have drawn the eyes 60, 61. Thus, the further focus was on the prognostic model in the HCC chemotherapy. Consistently, the findings exhibited that MRP1, MRP3 and MRP5 were overexpressed in the high-risk group than that in low-risk group, and the lower drug sensitivity was observed in the higher expression levels of each prognostic gene. The results suggested that lower expression of prognostic genes benefited the treatment with chemotherapy of HCC patients.
Conclusion
This study demonstrated that a novel signature constructed by the 9 peroxisome-related genes to predict the prognosis for HCC patients by the TCGA and ICGC cohorts. And we evaluated the novel prognostic model under various clinical settings including survival and clinical-pathological characteristics, immune infiltration, immune pathways, immune checkpoints, multidrug resistance-related genes and chemotherapy. Last, an independent sample cohort was carried out to validate the mRNA and protein expression levels of the 9 peroxisome-related genes. To sum up, this prognostic model can accurately predict the OS of HCC, distinguish the immune status of HCC, and the risk score is related to anti-tumor drugs. It is essential to systematically explore the potential role of peroxisome-related prognostic genes in HCC progression, and provided new possibilities for HCC therapeutic intervention.Supplementary figures and tables.Click here for additional data file.
Table 1
Clinical characteristics of the HCC patients used in this study
TCGA-LIHC cohort
ICGC-LIRP-JI cohort
No. of patients
365
231
Age (median, range)
57 (16-90)
67 (31-89)
Gender
Female
119 (32.6%)
61 (26.4%)
Male
246 (67.4%)
170 (73.6%)
Grade
Grade 1
55 (15.1%)
NA
Grade 2
175 (47.9%)
NA
Grade 3
118 (32.3%)
NA
Grade 4
12 (3.3%)
NA
Unknown
5 (1.4%)
NA
Stage
I
170 (46.6%)
36 (15.6%)
II
84 (23.0%)
105 (45.5%)
III
83 (22.7%)
71 (30.7%)
IV
4 (1.1%)
19 (8.2%)
Unknown
24 (6.6%)
0 (0%)
Survival status
Alive
235 (64.4%)
189 (81.8%)
Deceased
130 (35.6%)
42 (18.2%)
Table 2
Baseline characteristics of the patients in different risk groups
Authors: Prasanna Ekambaram; Jia-Ying Lloyd Lee; Nathaniel E Hubel; Dong Hu; Saigopalakrishna Yerneni; Phil G Campbell; Netanya Pollock; Linda R Klei; Vincent J Concel; Phillip C Delekta; Arul M Chinnaiyan; Scott A Tomlins; Daniel R Rhodes; Nolan Priedigkeit; Adrian V Lee; Steffi Oesterreich; Linda M McAllister-Lucas; Peter C Lucas Journal: Cancer Res Date: 2017-12-19 Impact factor: 12.701
Authors: Anna Merlos-Suárez; Francisco M Barriga; Peter Jung; Mar Iglesias; María Virtudes Céspedes; David Rossell; Marta Sevillano; Xavier Hernando-Momblona; Victoria da Silva-Diz; Purificación Muñoz; Hans Clevers; Elena Sancho; Ramón Mangues; Eduard Batlle Journal: Cell Stem Cell Date: 2011-03-17 Impact factor: 24.633