Cuproptosis-related genes establishment score system to assess the prognosis and immune microenvironment in cholangiocarcinoma

doi:10.21203/rs.3.rs-2964180/v1

Download PDF

Article

Cuproptosis-related genes establishment score system to assess the prognosis and immune microenvironment in cholangiocarcinoma

https://doi.org/10.21203/rs.3.rs-2964180/v1

This work is licensed under a CC BY 4.0 License

Version 1

posted

You are reading this latest preprint version

Background

Cuproptosis is a novel type of mediated cell death strongly associated with the progression of several cancers and has been implicated as a potential therapeutic target. However, the role of cuproptosis in cholangiocarcinoma (CCA) for prognostic prediction, subgroup classification, and therapeutic strategies remains largely unknown.

Methods

A systematic analysis was conducted among 146 cuproptosis-related genes (CRGs) and clinical information based on independent mRNA and protein datasets to elucidate the potential mechanisms and prognostic prediction value of CRGs. A ten-CRG prediction model was constructed, and its effects on CCA prognosis were significantly connected to poor patient survival. Additionally, the expression patterns of our model included genes that were validated with several CCA cancer cell lines and a normal biliary epithelial cell line.

Results

First, a ten-CRG signature (ADAM9, ADAM17, ALB, AQP1, CDK1, MT2A, PAM, SOD3, STEAP3 and TMPRSS6) displayed excellent predictive performance for the overall survival of CCA. The low-cuproptosis group had a significantly better prognosis than the high-cuproptosis group with transcriptome and protein cohorts. Second, compared with the high-risk and low-risk groups, the two groups displayed distinct tumor microenvironments, reduced proportions of endothelial cells and increased levels of cancer-associated fibroblasts based on CIBERSORTx and EPIC analyses. Third, patients’ sensitivities to chemotherapeutic drugs and immune checkpoints revealed distinctive differences between the two groups. Finally, in replicating the expression patterns of the ten genes, these results were validated with qRT‒PCR results validating the abnormal expression pattern of the target genes in CCA.

Conclusions

Collectively, we established and verified an effective prognostic model that could separate CCA patients into two heterogeneous cuproptosis subtypes based on the molecular or protein characteristics of ten CRGs. These findings may provide potential benefits for unveiling molecular characteristics, and defining subgroups could improve the early diagnosis and individualized treatment of CCA patients.

Biological sciences/Cancer

Health sciences/Gastroenterology

Health sciences/Oncology

cholangiocarcinoma

cuproptosis

prognostic biomarker

tumor microenvironment

comprehensive analysis

CCA is a rare and lethal disease accounting for nearly 3% of digestive neoplasias and 15% of primary liver cancers ^1–3. However, the global incidence and mortality rates of this fatal disease have increased in recent decades ^4–6. Due to the absence of obvious clinical symptoms or characteristics at the early stages, most CCA patients are diagnosed at advanced stages with limited therapeutic options, resulting in a disappointing prognosis with a median overall survival (OS) duration of less than 18 months ^7–10. Despite advances in curative therapeutics for CCA, surgery is not suitable for patients at advanced stages, and the operative resection rate is no more than 30%, with a high recurrence rate after resection ^3,9. Although individualized medicine and precise therapies have been improved for various tumors in recent decades, the characteristics of the biological mechanisms in CCA remain largely unknown.

Previous studies have indicated that different forms of cell death are related to distinctive clinical outcomes of cancer. Necroptosis and pyroptosis have been reported to play an important role in increasing antitumor functions, such as restraining tumorigenesis and metastasis ^11–13. In addition, therapy-resistant cancer cells are vulnerable to detecting ferroptosis signals, especially in the mesenchymal state, and are prone to metastasis ¹⁴. Badgley et al. showed that a genetically modified ferroptosis sensitivity mouse model dramatically decreased the formation and progression of pancreatic cancer ¹⁵. Recently, cuproptosis was shown to be a novel regulated form of cell death that depends on excess intracellular Cu ions, which induces accumulation of lipoylated dihydrolipoamide S-acetyltransferase and disruption of the mitochondrial tricarboxylic acid (TCA) cycle, resulting in proteotoxic stress ¹⁶. Interestingly, the process of cuproptosis in various cancer cell lines was not repressed when necroptosis, pyroptosis and ferroptosis were inhibited via genetic and pharmacological methods ^16,17. Previous studies have reported that increased Cu ion concentrations were pervasively identified in tumor tissue and serum in numerous kinds of cancers ^18,19. Compared with healthy cell lines, cancer cell lines revealed a higher requirement for Cu ionophores to maintain propagation and metastasis ²⁰. Although cuproptosis and CRGs are involved in the activation of cancer cell proliferation- and metastasis-related pathways, few studies have implicated the therapeutic functions and molecular characterization of cuproptosis in the progression of CCA. Hence, illuminating the key mechanisms of cuproptosis in oncology and its potential association with clinical outcomes may facilitate the treatment and prognostic prediction of CCA patients.

In the current study, we aimed to comprehensively investigate the prognostic and therapeutic values of CRGs in CCA. The expression patterns of CRGs in tumor tissues and adjacent normal specimens were examined with The Cancer Genome Atlas (TCGA) and two independent datasets from the Gene Expression Omnibus (GEO). Additionally, based upon the expression data of CRGs and the clinical information of CCA patients from the national Omics Data Encydopedia (NGDC) dataset and two other independent datasets, we constructed and validated a ten-CRG signature to stratify CCA patients into two heterogeneous cuproptosis subtypes with specific prognostic outcomes. Notably, these two groups presented discrepancy functional and biological processes, immune microenvironment and drug sensitivity. Furthermore, independent cohort analysis and qRT‒PCR were performed to verify the expression patterns of the genes involved in the prognostic model. Our findings presented the prognostic value of a novel ten-CRG prediction model and provided novel insights into the potential mechanism as well as therapeutic opportunities in CCA.

Data and resources

The count-based gene expression profiles and associated clinical information of the TCGA-CCA cohort (n = 45), which included 9 adjacent normal tissue samples and 36 tumor tissue samples, were downloaded from the UCSC Xena browser (https://gdc.xenahubs.net) ²¹. Two high-throughput sequencing datasets, GSE107943 and GSE76297, were collected from GEO as validation datasets (https://www.ncbi.nlm.nih.gov/geo/). GSE107943 contains 30 CCA tissue samples and 27 adjacent normal tissue samples, and GSE76297 includes paired CCA tumor tissues and adjacent controls from 91 CCA patients. However, overall survival (OS) information was not reported in the GSE76297 dataset. To develop and replicate the accuracy and sensitivity of the novel prognostic model, the transcriptomic and proteomic profiles and associated clinical information of 255 CCA patients were acquired from the NGDC dataset (https://www.biosino.org/node/project/detail/OEP001105). To ameliorate potential false-positive results, lowly expressed genes with average counts less than 5 were excluded from downstream analyses.

Identification of survival-related genes and construction of a prognostic signature for CCA

A total of 146 CRGs were selected from previous articles and the Molecular Signature Database (MsigDB) v7.0 dataset (http://www.gsea-msigdb.org/gsea/msigdb/) to assess the expression patterns between tumor and adjacent tissue ^16,18,22,23. We performed differential expression gene (DEG) analyses using the limma package (v. 3.48.3) ²⁴. Genes with P values less than < 0.05 and log2-transformed fold changes with concordant direction in the GSE76297, GSE107943 and TCGA-CCA datasets were defined as candidate CRGs to perform subsequent bioinformatics analyses. In addition, expression data from OEP001105 were used to identify the association between the abovementioned CRGs and OS. A univariate Cox proportional hazards regression model was conducted to screen prognosis-related genes through the R package glmnet (v 4.1.2) ²⁵. P values less than 0.05 were considered statistically significant. The intersecting CRGs in both DEG analysis and univariate Cox regression were selected for further analyses. To decipher the potential biological functions of the overlapping genes, protein‒protein interaction network analysis with selected targets was performed using the STRING dataset (v.10.5) with default parameters (https://string-db.org).

A LASSO-penalized Cox regression model with 10-fold cross-validation was performed to develop the model with the best predictive performance. According to the linear combination of the regression coefficients from LASSO, a risk score was calculated for each patient. The patients were divided into high- and low-risk groups based on the median value of all risk scores. Kaplan–Meier analysis (K-M) and time-dependent receiver operating characteristic (ROC) curves were employed to assess the sensitivity and specificity of the constructed model. A P value less than 0.05 in the K-M analysis was considered significant. To test the prognostic risk score and assess its predictive precision and sensitivity in different populations, 30 patients in GSE107943 and 36 patients in the TCGA-CCA dataset were classified into low- and high-risk groups based on the median risk score. The K-M and ROC analyses were conducted by using the R packages timeROC (v 0.4), survival (v 3.2–11), survminer (v 0.4.9), and survivalROC (v 1.0.3) in both the training and validation datasets. Additionally, the relationships between the expression levels of the genes in the prognostic signature, including mRNAs, and the OS rate were analyzed in the datasets. The K-M survival curves were drawn with a modified drawing function ‘ggsurvplot’ in survminer (v 0.4.9).

Pathway enrichment analysis

To illuminate the potential pathologies of CCA, the R package ClusterProfiler (v. 3.14.3) was applied to perform Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway enrichment analysis ²⁶. Items with false discovery rate (FDR) P values less than 0.05 were considered significantly enriched pathways.

Assessment of immune cell infiltration and stromal cells

The CIBERSORTx algorithm was applied to evaluate the relative abundance of 22 types of tumor-infiltrating immune cells based on gene expression in the OEP001105 dataset ^27,28. The category of immune cell types included naive B cells, memory B cells, plasma cells, resting memory CD4⁺ T cells, activated memory CD4⁺ T cells, naive memory CD4⁺ T cells, CD8⁺ T cells, follicular helper T cells, regulatory T cells, gamma delta T cells, resting natural killer cells, activated NK cells, monocytes, M0-M2 macrophages, resting mast cells, activated MCs, resting dendritic cells, activated dendritic cells, eosinophils, and neutrophils. Because of the limited number of samples in the separate datasets, we combined the two datasets to identify the immune-related cell proportions. In addition, the proportions of cancer-associated fibroblasts (CAFs) and endothelial cells (ECs) were calculated using the online EPIC algorithm (https://gfellerlab.shinyapps.io/EPIC_1-1/) ²⁹.

Drug sensitivity differences in the risk score classification

Drug sensitivity to chemotherapeutic drugs and novel therapeutic drugs for CCA patients was evaluated according to the Genomics of Drug Sensitivity in Cancer (GDSC) database (https://www.cancerrxgene.org/). The half-maximal inhibitory concentration (IC50) of each drug for CCA was predicted by the R package OncoPredict (v. 0.2) with the RNA-seq expression matrix ³⁰. The Wilcoxon test was applied to calculate drug sensitivity differences between the low-risk and high-risk groups in the training and validation cohorts.

Independent prognostic value analysis

Univariate and multivariate Cox regression analyses of the risk score of the 10-CRG prediction model and conventional clinical features, such as tumor size diameter, carbohydrate antigen 19 − 9 (CA19-9), carcinoembryonic antigen (CEA), tumor stage, liver fluke infection, sex and age, were applied to evaluate independent risk factors for CCA patients. Only p values less than 0.05 were considered statistically significant.

Single-cell RNA sequencing analysis of the CCA tumor microenvironment

We performed single-cell analysis with 7 CCA patients extracted from the GSE189903 dataset to evaluate the expression level of ten CRG-related markers in T cells, B cells, hepatocytes, cholangiocytes, tumor-associated macrophages (TAMs), cancer-associated fibroblasts (CAFs) and tumor-associated endothelial cells (TECs) ³¹. The Seurat package (v. 4.3.0) were applied to remove the genes expressed in at least 3 cells and cells with more than 300 genes.

Cell culture and reverse transcription-quantitative PCR

Human intrahepatic biliary epithelial cells (HIBEC; BeNa Culture Collection Co. Ltd. China) and human cholangiocarcinoma cell lines (HuCCT1, HCCC9810 and RBE; HaoKE Boitechnology Co. Ltd). China) were cultured in RPMI-1640 containing 10% fetal bovine serum, 1% penicillin G, and streptomycin solution under standard conditions (37 ℃ in 5% CO2). These cell lines were negative for mycoplasma contamination.

The aberrant expression levels of the genes included in the prediction model were validated by using qRT‒PCR. Reverse transcription reactions were performed with the PrimeScript™ II 1st Strand cDNA Synthesis Kit (TaKaRa) following the manufacturer’s protocol. qRT‒PCR was performed with UltraSYBR Mixture (CWBIO). The primers are listed in Table S1. The transcriptome expression levels were assessed relative to the expression of β-actin by the 2 -^ΔΔCt method.

Statistical Analysis

All analyses and creation of graphs were performed in the R package Tableone (v. 0.13.0) or SPSS (v. 22.0). Statistically significant differences were evaluated by Wilcoxon or Student’s t tests. The log-rank test was carried out to evaluate differences in survival between low-risk and high-risk groups with these independent cohorts. The correlation matrix was constructed with R according to the Pearson correlation coefficient.

Identification of candidate cuproptosis-related genes

In the present study, the analysis processes were implemented according to the workflow shown in Fig. 1. In total, 146 CRGs were included for further analyses. Compared with tumor tissues and adjacent controls, 114, 53, and 98 cuproptosis-related genes were aberrantly expressed in the GSE107943, GSE76297 and TCGA-CCA cohorts, respectively. Among these targets, 47 significant CRGs overlapped in a consistent direction in all datasets (Fig. 2a, Supplementary Table 1). As shown in Fig. 2b, a univariate Cox regression analysis was conducted using 255 CCA patients from the OEP001105 cohort to inspect prognostic biomarkers. A total of 50 prognostic genes were detected to be significantly associated with OS (Supplementary Table 2). Among the cuproptosis-related DEGs and prognostic markers, 16 overlapping genes, 8 upregulated and 8 downregulated, were annotated as candidate CRGs for CCA (Fig. 2b). To gain better insight into the molecular mechanisms involved in CCA, we constructed a protein‒protein network and investigated the orchestration of the candidate CRGs for CCA (Fig. 2c). In addition, we evaluated the correlations among the candidate genes’ transcriptome expression profiles using a correlation matrix (Fig. 2d). Interestingly, these candidate CRGs exhibited strong correlations with each other, which indicated a close interaction and exerted their effects by producing costimulatory or inhibitory signals in biological processes.

Construction of the prognostic prediction model

To assess the joint effect of the prognostic significance of the 16 candidate CRGs on CCA patient survival, we constructed a multigene prognostic model by performing LASSO penalized regression analysis in the OEP001105 cohort. Consequently, LASSO modeling identified a ten-DEG-based signature with the minimum optional lambda value (Fig. 3a, b), which included ADAM Metallopeptidase Domain 9 (ADAM9), ADAM Metallopeptidase Domain 17 (ADAM17), albumin (ALB), aquaporin 1 (AQP1), cyclin-dependent kinase 1 (CDK1), metallothionein 2A (MT2A), peptidylglycine alpha-amidating monooxygenase (PAM), superoxide dismutase 3 (SOD3), STEAP3 metalloreductase (STEAP3) and transmembrane serine protease 6 (TMPRSS6). The risk score for CCA was calculated as follows: (6.83×10⁶ × expression level of ADAM9) + (1.21×10⁵ × expression level of ADAM17) + (-4.27 × 10⁷ × expression level of ALB) + (-1.43 × 10⁵ × expression level of AQP1) + (0.0898670 × expression level of CDK1) + (1.76 × 10⁵ × expression level of MT2A) + (1.6696174 × expression level of PAM) + (0.7596718 × expression level of SOD3) + (0.1891796 × expression level of STEAP3) + (0.1891796 × expression level of TMPRSS6).

All CCA patients were divided into high-risk and low-risk groups for the discovery (high-risk: n = 127, low-risk: n = 128) and validation datasets (GSE107943: high-risk: n = 15, low-risk: n = 15; TCGA-CCA: high-risk: n = 18, low-risk: n = 18) according to corresponding median values of the risk score. The expression patterns of the ten-CRG signature in the high- and low-risk groups of the OEP001105 dataset were visualized (Fig. 3c-e). K-M curves were conducted to compare survival conditions in the two risk groups. The K-M analysis indicated that the low-risk group had a better OS rate than the high-risk group in the training dataset (Fig. 3f, P < 0.0001). In addition, in line with the K-M analysis in the training dataset, the high-risk group was associated with a worse OS rate in the validation datasets (Fig. 3g, GSE107943: P = 0.0092; Fig. 3h, TCAG-CCA: P = 0.04). ROC curves were applied to assess the precision of the ten-CRG prognostic model in predicting patient survival. Remarkably, the ROC curve analysis in the discovery cohort showed that the areas under the ROC curve (AUC) values at 1, 2 and 3 years were up to 0.755, 0.76 and 0.743, respectively (Fig. 3i). In the validation dataset, the AUC values at 1, 2 and 3 years were 0.912, 0.786 and 0.767 for the GSE107943 dataset and 0.838, 0.845 and 0.794 for the TCGA-CCA dataset, respectively (Fig. 3j, k). Therefore, these results indicated that the CRG signature could predict CCA patients with high sensitivity and specificity.

Assessment of prognostic factors in CCA

To better understand the relevance and underlying mechanisms of the novel prognostic model in CCA, we applied univariate and multivariate Cox regression analyses to evaluate whether the ten-gene model was an independent prognostic factor of OS for cholangiocarcinoma patients. Associations between the clinicopathological features and the risk score of the ten-CRG scoring system were assessed in the OEP001105 cohort. The distribution of the value of the risk score, patient survival status, and survival time of CCA patients is visualized in Fig. 4a and 4b. To analyze the clinical characteristics of CCA, tumor size diameter as well as the expression levels of CA19-9 and CEA were significantly higher in the high-risk group than in the low-risk group (P = 0.015, P = 0.009, P = 0.007, Fig. 4c). Furthermore, compared with the early stage and advanced stage of low-risk CCA patients, the risk score was dramatically upregulated in the high-risk group (P < 0.0001, Fig. 4d). Additionally, to test whether the developed ten-CRG signature risk score was independent of other clinicopathological features of CCA, we performed univariate and multivariate Cox regression analyses for risk score, involving fluke infection, tumor size diameter, CA19-9, CEA, TNM stage, sex and age (Fig. 4e, f). Both univariate and multivariate Cox regression revealed that tumor size diameter and risk score were independent prognostic predictors for CCA.

The risk model predicted the infiltration of immune cells in CCA

To recognize the indicative role of prognostic signals in the tumor microenvironment (TME), we implemented CIBERSORT to evaluate the infiltration of 22 immune-related cells in CCA in the OEP001105 cohort (Fig. 5a). Then, the differential distribution of the detected immune-related cells between the high-risk and low-risk groups was examined by the Wilcoxon rank-sum test. Compared with those in the low-risk group, the proportions of activated memory CD4 T cells, follicular helper T cells, monocytes, M0 macrophages, activated mast cells and neutrophils were significantly increased in the high-risk group (P = 3.48 × 10^–03, P = 6.80 × 10^–04, P = 4.02 × 10^–02, P = 8.30 × 10^–04, P = 3.80 × 10^–07, P = 1.42 × 10^–03, respectively). Naive B cells, CD8 T cells, resting memory CD4 T cells, activated NK cells, M1 macrophages, and resting dendritic cells were significantly highly infiltrated in the low-risk group (P = 5.40 × 10^–04, P = 2.41 × 10^–02, P = 1.90 × 10^–06, P = 2.10 × 10^–06, P = 2.96 × 10^–03, P = 8.27 × 10^–03, respectively). In addition, as shown in Fig. 5b, based on the EPIC analysis, CAF scores were significantly higher and endothelial scores were decreased in the high-risk group compared with the low-risk group (P = 0.01, P = 0.003).

Potential therapeutic targeting and tertiary lymphoid structure analyses in CCA

To identify the drug sensitivity differences of chemotherapeutic drugs and novel inhibitors for CCA patients between the high-risk and low-risk groups, we investigated the expression patterns of hub genes for chemokines, immune checkpoints, and chemotherapeutic regimens in the field of CCA treatment. A total of 17, 8 and 7 hub targets for chemokines, immune checkpoints, and chemotherapeutic regimens were acquired from previous studies to evaluate the therapeutic effects between the two groups ^32,33. Among the 17 chemokine-specific markers, the expression levels of 5 genes (CCL17, CCL19, CCL21, CCL7, CCL8) were significantly decreased and those of 2 markers (CCL13, CCL24) were significantly increased in the high-risk group compared with the low-risk group (Fig. 5c). Moreover, immune checkpoint and chemotherapeutic regimen analyses showed that HAVCR2, IDO1, LAG3, PD-L1, CDK1, CLSPN, PARP1 and RAD51 were significantly higher in the high-risk group than in the low-risk group (Fig. 5d, e). Subsequently, to infer the potential functions of the hub genes, we performed correlation analysis between the immune-related infiltrating cells and the hub genes for chemokines, immune checkpoints, and chemotherapeutic regimens (Fig. 5f).

In addition, the mRNA expression matrix was used for estimation with OncoPredict (v. 0.2) ³⁰. Bonferroni correction was used to define significant drugs (i.e., P = 0.05/198 = 2.5 × 10^− 3), where 198 is the total number of drugs evaluated for the software. As a result, AGI-6780, AZD1208, AZD8055, cediranib, ibrutinib, ML323, OSI-027, PAK-5399, PCI-34051, PD325901, PFI3, picolinici-acid, rapamycin, SCH772984, selumetinib and Wnt-C59 were lower in the low-risk group than in the high-risk group, with P values ranging from 1.80 × 10^− 4 to 1.90 × 10^− 8, illuminating that the low-risk group was more prone to sensitivity to these drugs than the high-risk group (Fig. 6a-p).

Validation of the risk score at the protein expression level

To further replicate the sensitivity and accuracy of our prediction model, we performed a similar analysis with the target proteins. A total of 8,320 proteins were detected in the OEP001105 cohort, and eight of the ten-CRG signatures were found in this dataset (ADAM9, ADAM17, ALB, AQP1, CDK1, MT2A, PAM, SOD3 and STEAP3). The scoring system of CCA was conducted to assess the prognosis of CCA in this cohort dependent on these eight gene protein expression levels. Of note, based on the K-M analysis results, the survival probability was significantly lower in the high-risk group than in the low-risk group (Fig. 7a). The AUCs at 1, 2 and 3 years of survival were 0.713, 0.698 and 0.723, respectively (Fig. 7b). In addition, Pearson correlation between the biomarker of mRNA and protein expression levels was performed to test the relationship. Among the eight genes, the mRNA and protein expression levels were strongly correlated with each other (ADAM9: r = 0.77, P < 2.2 × 10^− 16; ADAM17: r = 0.32, P = 4.0 × 10^− 06; AQP1: r = 0.79, P < 2.2 × 10^− 16; CDK1: r = 0.80, P < 2.2 × 10^− 16; MT2A: r = 0.32, P = 4.9 × 10^− 06; PAM: r = 0.62, P < 2.2 × 10^− 16; SOD3: r = 0.53, P = 1.1 × 10^− 15; STEAP3: r = 0.75, P < 2.2 × 10^− 16; Fig. 7c). Furthermore, we performed differential expression protein analysis between the high-risk and low-risk groups via the limma package. A total of 131 significantly upregulated and 107 downregulated proteins were identified for subsequent analysis (Fig. 7d). ClusterProfiler was used to reveal the underlying differences in functions between the low-risk and high-risk groups based on KEGG pathways. Fourteen CCA-related pathways, such as the PI3K-Akt signaling pathway, chemical carcinogenesis DNA adducts and bile secretion, were significantly associated with the risk score after FDR correction (Fig. 7e). These results indicated that our ten-CRG indicator playing an effect on CCA development is closely related to the immune response and hepatobiliary functions.

Expression level of CRG-related genes in CCA

To explore the expression patterns of CRG-related genes in different types of infiltrating immune cells, a total of 47,261 cells from 7 CCA patients were classified into T cells, B cells, hepatocytes, cholangiocytes, TAMs, CAFs and TECs according to 22 markers (Fig. 8a, b). Our evidence revealed that MT2A, ADAM9, PAM and ADAM17 pervasively existed in all kinds of cells, and SOD3, AQP1 and ALB were highly expressed in CAFs, TECs and hepatocytes, respectively (Fig. 8c-l).

Validation of the prediction model including gene expression patterns

According to the qRT‒PCR results, compared with the normal cell line, the majority of the target genes were aberrantly expressed in at least one CCA cell line, except PAM and STEAP (P < 0.05, Fig. 9a-k). The expression levels of AQP1, MT2A and SOD3 were significantly upregulated in the three CCA cell lines. ADAM9 and ADAM17 were evaluated in RBE and HCCC9810 cell lines, and CDK1 was upregulated in HuCCT1 and HCCC9810 cells. In addition, TMPRSS6 and ALB were significantly decreased in the HuCCT1 and RBE cell lines and the HCCC9810 cell line, respectively. However, inconsistent results for TMPRSS6 and ALB were presented in the remaining cell lines.

CCA is a highly fatal digestive disease with poor diagnosis and prognosis. Molecular prognostic markers, as an effective and sensitive method, may be applied as a beneficial supplement to traditional clinicopathological parameters for bile duct cancer patients to increase prognostic prediction, early diagnostic precision, subgroup identification and personalized treatment. Due to the tumor heterogeneity of CCA, the molecular characterization and physiological characteristics of CCA are still unclear, and the effect of the predictive model is unsatisfactory. Numerous studies have indicated that cell death is one of the most important mechanisms engaged in tumors, and dysregulation of cell death-related genes is directly associated with prognosis in several types of cancers. As a novel type of programmed cell death, cuproptosis-related pathway signatures and genes have gained rising prominence in several kinds of cancer ^16,17,34. However, no study has investigated the prognostic and therapeutic values of CRGs for CCA patients. In this study, we systemically detected the classification potential, immune cell infiltration and prognostic values of CRGs for CCA patients with an integrative bioinformatic analysis and validation through several independent cohorts. However, most of the prognostic prediction models did not perform validation for the included targets. Overall, we have implemented complicated analyses for identifying cuproptosis-related subgroups and extending the knowledge about the potential mechanisms of CRGs, which suggest that CRGs may play essential roles and serve as novel targets for therapy.

In the current study, 146 cuproptosis-related genes were selected to evaluate the expression profiles between tumor and adjacent tissues. We identified 47 replicated DEGs overlapping in three independent cohorts of CCA, and 17 of them were significantly associated with OS. Furthermore, a novel ten-gene-based signature was constructed by LASSO Cox regression to predict the OS of CCA. Because the heterogeneity among different patients has been widely reported to influence the biological mechanisms, management, and pathological features of CCA patients ⁸, constructing a prognostic biomarker with high precision and sensitivity as well as validation in an independent cohort is urgently needed to improve prognosis ³⁵. In our study, the prognostic signature was established and validated in several CCA transcriptome and proteome sequencing datasets with high sensitivity and accuracy. Of note, limited studies have validated the association between predictive models and OS at the proteome level. Compared with the protein expression profile, the constructed prognostic prediction model via the transcriptome profile as a dynamic tool fluctuating with tumor progression to reflect the prognosis is much easier to affect by other factors. Strikingly, consistent with the analysis results at the mRNA level, the low-cuproptosis group separated by the novel prognostic model, which was constructed based on the protein level, had a better prognosis than the high-cuproptosis group. Furthermore, the expression levels of eight out of ten targets (ADAM9, ADAM17, ALB, AQP1, CDK1, MT2A, SOD3 and TMPRSS6) were verified in at least one CCA cell line compared with normal bile duct cells. However, ALB and TMPRSS6 present conflicting results, possibly triggered by heterogeneity among different CCA cell lines.

Four of those ten prognostic genes (ADAM17, ALB, AQP1 and CDK1) were previously reported to be associated with CCA prognosis. ALB, as one of the markers of hepatocytes and cholangiocytes, plays oncogenic roles in hepatocyte-specific processes, such as complement activation, detoxification, fatty acid catabolic process, and bile acid secretion ³⁶. CDK1 regulates the hub gene of cytokinesis-related genes, which cooperate in the process of cellular proliferation, cytoskeleton organization and cytokinesis ^37,38. AQP1 is a small hydrophobic integral transmembrane protein that is ubiquitously expressed in multiple human cancers. Increasing evidence indicates that AQP1 overexpression accelerates migration, extravasation, tumor angiogenesis and metastasis in vitro and in vivo^39,40. Julia and colleagues suggested that TNFR1-dependent tumor cell-induced endothelial cell death, tumor cell extravasation, TNF-induced necroptosis, and subsequent metastatic seeding rely on the activation of the membrane-bound metalloprotease ADAM17 ⁴¹. In addition, all of these targets serve as prognostic biomarkers for many other types of cancers, such as ovarian cancer, colorectal cancer, hepatocellular carcinoma, and kidney cancer ^42–45. However, the biological function of these genes in CCA is still unclear, and further investigation is needed to explore the potential mechanisms of these target genes in the future.

The tumor immune microenvironment is considered a key regulator of carcinogenesis, angiogenesis, and tumor growth and affects the efficacy of radiotherapy, chemotherapy, and immune checkpoint therapy ^46,47. Bile duct carcinoma is characterized by an intricate microenvironment in which the stroma is composed of tumor-associated endothelial cells, fibroblasts, macrophages, neutrophils, NK cells, and T cells ⁴⁸. According to the infiltration of immune cells in CCA, we detected more naive B cells, CD8 T cells, resting memory CD4 T cells, activated NK cells, M1 macrophages, and resting dendritic cells in the low-risk group. In addition, the risk score for CCA was positively correlated with activated memory CD4 T cells, follicular helper T cells, monocytes, M0 macrophages, activated mast cells and neutrophils in CCA patients. M0 macrophages are nonactivated macrophages that differentiate into M1/2 macrophages in the presence of specific conditions. M1 macrophage infiltration promotes an inflammatory microenvironment by increasing the expression of IL-1, IL-6 and IL-12, and M2 macrophage infiltration promotes inflammatory escape by increasing the expression of PD-L1 and IL-10 ^49,50. A previous study suggested that patients with a higher infiltration of CD8 + T cells and M1 macrophages in the TME have a better prognosis and high immune checkpoint gene expression than those infiltrated by M0/2 macrophages ^51,52. In our results, the microenvironment of the CCA low-risk group had significantly enriched M1 macrophage and CD8 + T-cell infiltration. Previous studies reported that the abundance of CAFs was a major TME with tumorigenic properties, which were positively correlated with tumor growth and poor prognosis ^53,54. CAFs regulate innate immunity by suppressing NK cell activation to promote immunosuppression ⁵⁵. Moreover, NK cell-induced antibody-dependent cellular cytotoxicity increased cancer cell death, and infusion of NK cells in CCA mouse xenograft models resulted in tumor regression ^56,57. Immune checkpoint blockade with monoclonal antibody therapies has been implemented as a novel plan for treating numerous malignancies with remarkable and durable response rates ⁵⁸, while clinical trials of immunotherapies in CCA have displayed less notable success ⁸. NK cells are currently widely targeted in immunotherapy for cancer patients without major histocompatibility complex (MHC) class I ⁵⁹. When the MHC I expression level is downregulated or silenced in cancer cells, NK cells are activated to remove tumor cells. Laura et al demonstrated that activated mast cells could induce biliary hyperplasia, hepatic fibrosis and vascular bed dysfunction, and knockout or inhibition of mast cells was likely to be a better therapy for patients with cholangiopathies ⁶⁰. Interestingly, based on EPIC analysis, this study detected that CAF scores were significantly increased and endothelial scores were reduced in the high-risk group. In summary, these results strongly highlight that it is necessary to explore the therapeutic value of the TME of M1 macrophages and NK cells in inhibiting CAF therapy for CCA.

For the drug sensitivity analyses, 16 kinds of therapeutic drugs, AGI-6780, AZD1208, AZD8055, cediranib, ibrutinib, ML323, OSI-027, PAK-5399, PCI-34051, PD325901, PFI3, picolinici-acid, rapamycin, SCH772984, selumetinib and Wnt-C59, revealed more sensitivity in the low-risk group than in the high-risk group. Selumetinib, an inhibitor of mitogen-activated protein and extracellular signal-regulated kinase, can be targeted to suppress tumor growth and reduce IL-6 expression in several different types of cancer ^61,62. Moreover, selumetinib has highly effective effects on inhibiting tumor growth by reducing tumor cell proliferation and accumulating apoptosis in a K-Ras-driven CCA mouse model ⁶³. In our results, the IC50 value of selumetinib in the low-cuproptosis group was significantly lower than that in the high-cuproptosis group, suggesting that selumetinib may be more effective for the low-risk group. Thus, the novel prognostic model probability can serve as a personalized treatment indicator for CCA patients.

Although comprehensive analyses were performed to investigate the potential mechanisms of the prognostic biomarkers for CCA, this study has certain limitations. First, immune cell infiltration in bile duct cancer was assessed on the basis of publicly available TCGA, GSE107943 and NGDC RNA-seq data. Further biological experiments need to be implemented in vivo and in vitro to validate the immunotherapeutic value and interaction among immune cells and CRGs in CCA. Second, even if half of the genes included in the prognostic model have been reported to influence different cancers, biological experiments should be performed to elucidate the molecular mechanisms of cuproptosis-related genes in the pathogenesis and prognosis of CCA.

In summary, our results identified and verified an accurate and sensitive ten-CRG signature to predict the OS of CCA patients. Mechanistically, cuproptosis plays a key role in CCA development and is closely related to immune cell infiltration, CAFs, drug sensitivity, and checkpoint and chemotherapy marker expression. These findings provide a novel prognostic signature to account for the OS of CCA patients and may benefit the timely diagnosis and individualized treatment of CCA.

CCA

Cholangiocarcinoma

TCGA

The Cancer Genome Atlas

GEO

Gene Expression Omnibus

NGDC

National Genomics Data Center

DEGs

Differentially expressed genes

overall survival

CA19-9

carbohydrate antigen 19 − 9

KEGG

Kyoto Encyclopedia of Genes and Genomes

K-M

Kaplan-Meier analysis

qRT-PCR

Quantitative real-time PCR

ROC

time-dependent receiver operating characteristic curves

Resting natural killer cells

MCs

GDSC:Resting mast cells

Genomics of Drug Sensitivity in Cancer

IC50

Half-maximal inhibitory concentration.

Author Contributions

QL, LZ, YB collected the data, performed the bioinformatic analysis and wrote the manuscript. ZH, JZ, YG, YC, DX and HJ were involved in data collection and reviewed the manuscript. JY, XZ and HS conceived the current study and reviewed the manuscript. All authors have reviewed the subsequent versions and read and approved the final manuscript.

Funding

This study was supported in part by the Zhejiang Province's 2023 Key R&D Plan Project (Grant No. 2023C03054); the Natural Science Foundation of Zhejiang Province (Grant No. LQ23H160047 and LGF21H310004); Zhejiang Medical and Health Science and Technology Plan (Grant No. WKJ-ZJ-2136, 2021ZH003, 2022RC056 and 2023RC229); Hangzhou Medical and Health Science and Technology Plan (Grant No. 202004A14, 2022WJC003 and A20200119); Construction Fund of Medical Key Disciplines of Hangzhou (OO20190001).

Data availability

The datasets used and/or analysed during the current study available from the corresponding author on reasonable request.

Acknowledgments

We acknowledge TCGA, GEO, and NGDC database for uploading their datasets.

Disclosure statement

The authors declare that they have no conflict of interest.

Charbel, H. & Al-Kawas, F. H. Cholangiocarcinoma: epidemiology, risk factors, pathogenesis, and diagnosis. Curr Gastroenterol Rep 13, 182-187, doi:10.1007/s11894-011-0178-8 (2011).
Banales, J. M. et al. Expert consensus document: Cholangiocarcinoma: current knowledge and future perspectives consensus statement from the European Network for the Study of Cholangiocarcinoma (ENS-CCA). Nat Rev Gastroenterol Hepatol 13, 261-280, doi:10.1038/nrgastro.2016.51 (2016).
DeOliveira, M. L. et al. Cholangiocarcinoma: thirty-one-year experience with 564 patients at a single institution. Ann Surg 245, 755-762, doi:10.1097/01.sla.0000251366.62632.d3 (2007).
Razumilava, N. & Gores, G. J. Cholangiocarcinoma. Lancet 383, 2168-2179, doi:10.1016/S0140-6736(13)61903-0 (2014).
Everhart, J. E. & Ruhl, C. E. Burden of digestive diseases in the United States Part III: Liver, biliary tract, and pancreas. Gastroenterology 136, 1134-1144, doi:10.1053/j.gastro.2009.02.038 (2009).
Saha, S. K., Zhu, A. X., Fuchs, C. S. & Brooks, G. A. Forty-Year Trends in Cholangiocarcinoma Incidence in the U.S.: Intrahepatic Disease on the Rise. Oncologist 21, 594-599, doi:10.1634/theoncologist.2015-0446 (2016).
Khan, S. A. et al. Guidelines for the diagnosis and treatment of cholangiocarcinoma: consensus document. Gut 51 Suppl 6, VI1-9, doi:10.1136/gut.51.suppl_6.vi1 (2002).
Banales, J. M. et al. Cholangiocarcinoma 2020: the next horizon in mechanisms and management. Nat Rev Gastroenterol Hepatol 17, 557-588, doi:10.1038/s41575-020-0310-z (2020).
Spolverato, G. et al. Management and Outcomes of Patients with Recurrent Intrahepatic Cholangiocarcinoma Following Previous Curative-Intent Surgical Resection. Ann Surg Oncol 23, 235-243, doi:10.1245/s10434-015-4642-9 (2016).
Yang, J. et al. Endoscopic radiofrequency ablation plus a novel oral 5-fluorouracil compound versus radiofrequency ablation alone for unresectable extrahepatic cholangiocarcinoma. Gastrointest Endosc 92, 1204-1212 e1201, doi:10.1016/j.gie.2020.04.075 (2020).
Fang, Y. et al. Pyroptosis: A new frontier in cancer. Biomed Pharmacother 121, 109595, doi:10.1016/j.biopha.2019.109595 (2020).
Yu, J. et al. Cleavage of GSDME by caspase-3 determines lobaplatin-induced pyroptosis in colon cancer cells. Cell Death Dis 10, 193, doi:10.1038/s41419-019-1441-4 (2019).
Gong, Y. et al. The role of necroptosis in cancer biology and therapy. Mol Cancer 18, 100, doi:10.1186/s12943-019-1029-8 (2019).
Jiang, X., Stockwell, B. R. & Conrad, M. Ferroptosis: mechanisms, biology and role in disease. Nat Rev Mol Cell Biol 22, 266-282, doi:10.1038/s41580-020-00324-8 (2021).
Badgley, M. A. et al. Cysteine depletion induces pancreatic tumor ferroptosis in mice. Science 368, 85-89, doi:10.1126/science.aaw9872 (2020).
Tsvetkov, P. et al. Copper induces cell death by targeting lipoylated TCA cycle proteins. Science 375, 1254-1261, doi:10.1126/science.abf0529 (2022).
Tang, D., Chen, X. & Kroemer, G. Cuproptosis: a copper-triggered modality of mitochondrial cell death. Cell Res 32, 417-418, doi:10.1038/s41422-022-00653-7 (2022).
Chen, L., Min, J. & Wang, F. Copper homeostasis and cuproptosis in health and disease. Signal Transduct Target Ther 7, 378, doi:10.1038/s41392-022-01229-y (2022).
Saleh, S. A. K., Adly, H. M., Abdelkhaliq, A. A. & Nassir, A. M. Serum Levels of Selenium, Zinc, Copper, Manganese, and Iron in Prostate Cancer Patients. Curr Urol 14, 44-49, doi:10.1159/000499261 (2020).
Denoyer, D., Masaldan, S., La Fontaine, S. & Cater, M. A. Targeting copper in cancer therapy: 'Copper That Cancer'. Metallomics 7, 1459-1476, doi:10.1039/c5mt00149h (2015).
Blum, A., Wang, P. & Zenklusen, J. C. SnapShot: TCGA-Analyzed Tumors. Cell 173, 530, doi:10.1016/j.cell.2018.03.059 (2018).
Yuan, Y., Fu, M., Li, N. & Ye, M. Identification of immune infiltration and cuproptosis-related subgroups in Crohn's disease. Front Immunol 13, 1074271, doi:10.3389/fimmu.2022.1074271 (2022).
Liberzon, A. et al. Molecular signatures database (MSigDB) 3.0. Bioinformatics 27, 1739-1740, doi:10.1093/bioinformatics/btr260 (2011).
Ritchie, M. E. et al. limma powers differential expression analyses for RNA-sequencing and microarray studies. Nucleic Acids Res 43, e47, doi:10.1093/nar/gkv007 (2015).
Friedman, J., Hastie, T. & Tibshirani, R. Regularization Paths for Generalized Linear Models via Coordinate Descent. J Stat Softw 33, 1-22 (2010).
Yu, G., Wang, L. G., Han, Y. & He, Q. Y. clusterProfiler: an R package for comparing biological themes among gene clusters. OMICS 16, 284-287, doi:10.1089/omi.2011.0118 (2012).
Newman, A. M. et al. Robust enumeration of cell subsets from tissue expression profiles. Nat Methods 12, 453-457, doi:10.1038/nmeth.3337 (2015).
Newman, A. M. et al. Determining cell type abundance and expression from bulk tissues with digital cytometry. Nat Biotechnol 37, 773-782, doi:10.1038/s41587-019-0114-2 (2019).
Racle, J., de Jonge, K., Baumgaertner, P., Speiser, D. E. & Gfeller, D. Simultaneous enumeration of cancer and immune cell types from bulk tumor gene expression data. Elife 6, doi:10.7554/eLife.26476 (2017).
Maeser, D., Gruener, R. F. & Huang, R. S. oncoPredict: an R package for predicting in vivo or cancer patient drug response and biomarkers from cell line screening data. Brief Bioinform 22, doi:10.1093/bib/bbab260 (2021).
Ma, L. et al. Multiregional single-cell dissection of tumor and immune cells reveals stable lock-and-key features in liver cancer. Nat Commun 13, 7533, doi:10.1038/s41467-022-35291-5 (2022).
Yang, Z. et al. A new risk factor indicator for papillary thyroid cancer based on immune infiltration. Cell Death Dis 12, 51, doi:10.1038/s41419-020-03294-z (2021).
Ado, A. D. [Reflections on the letter by N. A. Kraevskii, A. V. Smol'iannikov, G. A. Chekareva and O. D. Mishnev to the editors of the journal, Arkhiv patologii (No. 3, 1978)]. Arkh Patol 40, 77-80 (1978).
Liu, H. Pan-cancer profiles of the cuproptosis gene set. Am J Cancer Res 12, 4074-4081 (2022).
Lai, M. et al. Identification and Validation of Two Heterogeneous Molecular Subtypes and a Prognosis Predictive Model for Hepatocellular Carcinoma Based on Pyroptosis. Oxid Med Cell Longev 2022, 8346816, doi:10.1155/2022/8346816 (2022).
Song, G. et al. Single-cell transcriptomic analysis suggests two molecularly subtypes of intrahepatic cholangiocarcinoma. Nat Commun 13, 1642, doi:10.1038/s41467-022-29164-0 (2022).
Wheatley, S. P. et al. CDK1 inactivation regulates anaphase spindle dynamics and cytokinesis in vivo. J Cell Biol 138, 385-393, doi:10.1083/jcb.138.2.385 (1997).
Kettenbach, A. N. et al. Quantitative phosphoproteomics identifies substrates and functional modules of Aurora and Polo-like kinase activities in mitotic cells. Sci Signal 4, rs5, doi:10.1126/scisignal.2001497 (2011).
Jiang, Y. Aquaporin-1 activity of plasma membrane affects HT20 colon cancer cell migration. IUBMB Life 61, 1001-1009, doi:10.1002/iub.243 (2009).
Tomita, Y. et al. Role of Aquaporin 1 Signalling in Cancer Development and Progression. Int J Mol Sci 18, doi:10.3390/ijms18020299 (2017).
Bolik, J. et al. Inhibition of ADAM17 impairs endothelial cell necroptosis and blocks metastasis. J Exp Med 219, doi:10.1084/jem.20201039 (2022).
Morrissey, J. J., London, A. N., Luo, J. & Kharasch, E. D. Urinary biomarkers for the early diagnosis of kidney cancer. Mayo Clin Proc 85, 413-421, doi:10.4065/mcp.2009.0709 (2010).
Zou, Y. et al. CDK1, CCNB1, and CCNB2 are Prognostic Biomarkers and Correlated with Immune Infiltration in Hepatocellular Carcinoma. Med Sci Monit 26, e925289, doi:10.12659/MSM.925289 (2020).
Sikora-Skrabaka, M., Walkiewicz, K. W., Nowakowska-Zajdel, E., Waniczek, D. & Strzelczyk, J. K. ADAM10 and ADAM17 as Biomarkers Linked to Inflammation, Metabolic Disorders and Colorectal Cancer. Curr Issues Mol Biol 44, 4517-4527, doi:10.3390/cimb44100309 (2022).
Whitehead, M., Taylor-Robinson, D. & Barr, B. Poverty, health, and covid-19. BMJ 372, n376, doi:10.1136/bmj.n376 (2021).
Anderson, N. M. & Simon, M. C. The tumor microenvironment. Curr Biol 30, R921-R925, doi:10.1016/j.cub.2020.06.081 (2020).
Kumagai, S. et al. An Oncogenic Alteration Creates a Microenvironment that Promotes Tumor Progression by Conferring a Metabolic Advantage to Regulatory T Cells. Immunity 53, 187-203 e188, doi:10.1016/j.immuni.2020.06.016 (2020).
Tamma, R. et al. Inflammatory cells infiltrate and angiogenesis in locally advanced and metastatic cholangiocarcinoma. Eur J Clin Invest 49, e13087, doi:10.1111/eci.13087 (2019).
Dufresne, A. et al. Specific immune landscapes and immune checkpoint expressions in histotypes and molecular subtypes of sarcoma. Oncoimmunology 9, 1792036, doi:10.1080/2162402X.2020.1792036 (2020).
Tamura, R., Tanaka, T., Yamamoto, Y., Akasaki, Y. & Sasaki, H. Dual role of macrophage in tumor immunity. Immunotherapy 10, 899-909, doi:10.2217/imt-2018-0006 (2018).
Subramanian, A. et al. Gene set enrichment analysis: a knowledge-based approach for interpreting genome-wide expression profiles. Proc Natl Acad Sci U S A 102, 15545-15550, doi:10.1073/pnas.0506580102 (2005).
Zhu, N. & Hou, J. Assessing immune infiltration and the tumor microenvironment for the diagnosis and prognosis of sarcoma. Cancer Cell Int 20, 577, doi:10.1186/s12935-020-01672-3 (2020).
Kalluri, R. The biology and function of fibroblasts in cancer. Nat Rev Cancer 16, 582-598, doi:10.1038/nrc.2016.73 (2016).
Chuaysri, C. et al. Alpha-smooth muscle actin-positive fibroblasts promote biliary cell proliferation and correlate with poor survival in cholangiocarcinoma. Oncol Rep 21, 957-969, doi:10.3892/or_00000309 (2009).
Ziani, L., Chouaib, S. & Thiery, J. Alteration of the Antitumor Immune Response by Cancer-Associated Fibroblasts. Front Immunol 9, 414, doi:10.3389/fimmu.2018.00414 (2018).
Morisaki, T. et al. Combining cetuximab with killer lymphocytes synergistically inhibits human cholangiocarcinoma cells in vitro. Anticancer Res 32, 2249-2256 (2012).
Jung, I. H. et al. In Vivo Study of Natural Killer (NK) Cell Cytotoxicity Against Cholangiocarcinoma in a Nude Mouse Model. In Vivo 32, 771-781, doi:10.21873/invivo.11307 (2018).
Wei, S. C., Duffy, C. R. & Allison, J. P. Fundamental Mechanisms of Immune Checkpoint Blockade Therapy. Cancer Discov 8, 1069-1086, doi:10.1158/2159-8290.CD-18-0367 (2018).
Purdy, A. K. & Campbell, K. S. Natural killer cells and cancer: regulation by the killer cell Ig-like receptors (KIR). Cancer Biol Ther 8, 2211-2220, doi:10.4161/cbt.8.23.10455 (2009).
Hargrove, L. et al. Bile duct ligation-induced biliary hyperplasia, hepatic injury, and fibrosis are reduced in mast cell-deficient Kit(W-sh) mice. Hepatology 65, 1991-2004, doi:10.1002/hep.29079 (2017).
Au, E. D., Desai, A. P., Koniaris, L. G. & Zimmers, T. A. The MEK-Inhibitor Selumetinib Attenuates Tumor Growth and Reduces IL-6 Expression but Does Not Protect against Muscle Wasting in Lewis Lung Cancer Cachexia. Front Physiol 7, 682, doi:10.3389/fphys.2016.00682 (2016).
Prado, C. M. et al. Skeletal muscle anabolism is a side effect of therapy with the MEK inhibitor: selumetinib in patients with cholangiocarcinoma. Br J Cancer 106, 1583-1586, doi:10.1038/bjc.2012.144 (2012).
Dong, M. et al. Efficacy of MEK inhibition in a K-Ras-driven cholangiocarcinoma preclinical model. Cell Death Dis 9, 31, doi:10.1038/s41419-017-0183-4 (2018).

No competing interests reported.

CHOLSupplementaryTable.xlsx

Download PDF

Version 1

posted

You are reading this latest preprint version

Cuproptosis-related genes establishment score system to assess the prognosis and immune microenvironment in cholangiocarcinoma

Status:

Version 1

Abstract

Background

Methods

Results

Conclusions

Figures

Introduction

Materials and Methods

Statistical Analysis

Results

Discussion

Abbreviations

Declarations

References

Additional Declarations

Supplementary Files

Status:

Version 1