Identification of a novel cuproptosis-related pattern and its tumor microenvironment infiltration characteristics in pancreatic cancer

doi:10.21203/rs.3.rs-3215150/v1

Download PDF

Research Article

Identification of a novel cuproptosis-related pattern and its tumor microenvironment infiltration characteristics in pancreatic cancer

https://doi.org/10.21203/rs.3.rs-3215150/v1

This work is licensed under a CC BY 4.0 License

Version 1

posted

You are reading this latest preprint version

Purpose

Cuproptosis is a novel mechanism of copper-dependent cell death mechanism that can regulate the progression, immune response, and prognosis of tumors. However, the potential roles of cuproptosis-related genes (CRGs) in the clinical outcomes, tumor microenvironment (TME), and immunotherapy of pancreatic cancer (PC) remain unclear.

Methods

We comprehensively evaluated the CRG patterns in PC samples from two GEO datasets and TCGA based on 19 CRGs. LASSO and multivariate Cox regression were used to construct the cuproptosis model, and a nomogram was constructed to predict the sensitivity of anticancer drugs.

Results

Methylation regulated the expression level of eight CRGs in PC. Three distinct cuproptosis-related patterns with different biological processes and prognoses were developed. The immune infiltration features of the three cuproptosis patterns were immune-excluded, immune-inflamed, and immune-desert phenotypes, respectively. Higher expression of CRGs indicated a poor prognosis of PC. Based on the cuproptosis phenotype associated signature genes, we constructed a cuproptosis score to study the cuproptosis modification pattern of the individual sample. Univariate and multivariate Cox regression analyses confirmed the cuproptosis risk score model is an independent prognosis biomarker. The high-risk group was characterized by poor prognosis, high expression of CRGs, high frequency of mutation and immune activation, and immunotherapy advantage. The sensitivity of anticancer drugs was significantly different between the high- and low-risk score groups. CRG pattern associated with the clinical-pathological features, TME, and prognosis of PC.

Conclusions

The newly developed cuproptosis model could guide the design of individualized treatment strategies and facilitate accurate prognosis prediction for PC.

Pancreatic cancer

cuproptosis

tumor microenvironment

immunotherapy

Pancreatic cancer (PC) is one of the most common deadliest tumors and has a slightly increasing annual incidence(Siegel et al. 2022), however, the rate of pancreatic cancer is higher among Chinese individuals when compared to white individuals(Yin et al. 2022). At present, the underlying genetic differences contributing to this disparity remain unclear. Due to its poor prognosis, PC is considered the sixth leading cause of cancer-related death in China, with the 8th highest incidence among Chinese males and the 12th among females(China Anti-Cancer Association Committee of Pancreatic Cancer. 2023). The 5-year overall survival rate of PC is less than 10%, while the median overall survival time is less than 6 months(Su et al. 2022). The reasons for the poor prognosis of PC are multifactorial and include universal late diagnosis, rapid tumor progression, lack of reliable biomarkers to detect PC, and effective medicine to provide long-term benefits for PC(Hingorani. 2023). Thus, it is crucial to develop reliable and novel biomarkers to estimate the prognostic and drug sensitivity prediction for further treatment.

Copper is involved in the biological processes of various diseases, including Alzheimer’s disease, Wilson disease, blood diseases, and cancers(Chen et al. 2020). Copper plays an important role in the occurrence and development of carcinogenesis by promoting cell proliferation, metastasis, and angiogenesis(Sha et al. 2022). Gul et al found that dysregulated copper transmembrane transport resulted in intracellular copper accumulation and subsequent cytotoxicity and cell death(Gul et al. 2020). Recently, a previously unknown mechanism of copper-induced cell death regulation has been revealed, termed cuproptosis. Cuproptosis differs from other cell death mechanisms such as autophagy, apoptosis, pyroptosis, and ferroptosis. Cuproptosis is caused by the binding of copper to fatty acylation components, resulting in the generation of acylating protein, iron-containing sulfur cluster protein loss, and abnormal fatty aggregates, which influence mitochondrial respiration and the TCA cycle(Tsvetkov et al. 2022). Some studies have indicated that the levels of copper in the tumor tissues or serum of patients with various types of cancer, including colorectal cancer, breast cancer, and lung cancer, are higher than those in healthy people(Aubert et al. 2020;Feng et al. 2020;Michniewicz et al. 2021;Wang et al. 2021). However, few studies have elucidated the effect of the cuproptosis mechanism on PC(Guan et al. 2022; Huang et al. 2022).Therefore, it may be valuable to explore the role of cuproptosis in PC.

Recent studies have highlighted the vital role of the tumor microenvironment (TME) in novel drug treatment, such as the anti-tumor immune response that is initiated by anti-CTLA4 treatment(Risom et al. 2022;Yamamoto et al. 2020). The TME comprises stromal components and cancer cells, including infiltrating immune cells, healthy epithelial cells, fibroblasts, and extracellular matrix (ECM). Previous studies have proven that the TME plays vital roles in cancer development, promotion, suppression, immune escape, and the response to immunotherapy(Jiang et al. 2022;Vitale et al. 2021). PC is characterized by immunosuppressive cell infiltration, the absence of penetrating affecting T cells, and a heavy mechanical barrier, all of which result in a high risk of metastasis, recurrence, and insensitivity to immunotherapy(Wood et al. 2022). Currently, many novel medicines targeting the tumor microenvironment burden (TMB), including multi-kinase inhibitors, and immune checkpoints have revealed exciting efficacy in some tumor patterns(Bockorny et al. 2022;Wang et al. 2023). However, few of these drugs have satisfactory effects on PC due to deficient knowledge of the TME(Balachandran et al. 2019). Consequently, there remains an urgent need to conduct a deeper analysis of the TME infiltration landscapes to guide treatment and predict the outcome of PC.

Here, we integrated the gene expression profiles of PC samples to assess the cuproptosis-related gene (CRG) patterns and explore the closed relationship between CRG patterns and TME infiltrating characteristics. A novel cuproptosis risk model was constructed to accurately predict the prognosis of patients with PC and their sensitivity to anticancer drugs.

Dataset and preprocessing

The RNA-sequencing and clinical data of 178 PC cases and 171 healthy cases were downloaded from The Cancer Genome Atlas (TCGA) and Genotype-Tissue Expression (GTEx) databases. The PC gene expression profiles and clinical characteristics of GSE57495 (n = 63) and GSE62452 (n = 69) were downloaded from the Gene Expression Omnibus (GEO) database. Patients with PC who lacked survival information were removed from the analysis. The tumor samples of the GEO and TCGA were merged, and the “sva” R package was employed to rectify the batch effects. Nineteen CRGs (NFE2L2, NLRP3, SLC31A1, FDX1, ATP7B, ATP7A, LIAS, LIPT1, LIPT2, DLD, DLAT, MTF1, DHA1, PDHB, GLS, GCSH, CDKN2A, DBT, DLST) were acquired from previous research(Huang et al. 2022;Sha et al. 2022;Tsvetkov et al. 2022). The copy number variation was obtained from UCSC Xena (https://xenabrowser.net).

Differential expression analysis and PPI network construction

We used the “limma” and “reshape2” packages to analyze the differences in CRG expression between PC and healthy samples. A protein-protein interaction (PPI) network of 19 CRGs was generated using the STRING database (https://string-db.org/). A medium confidence of 0.4 was set for PPI analysis.

Mutation and correlation analysis

Waterfall plots of CRGs and two risk score groups in PC were generated using the “maftools” R package to identify tumor mutations. The CNV landscape of 19 CRGs on 23 chromosomes was plotted by the R package “RCircos.” A correlation plot based on Pearson’s correlation analysis between CRGs was conducted using the “Corrplot” package, and the UALCAN database (http://ualcan.path.uab.edu) was used to perform CRG methylation analysis.

Unsupervised consensus clustering analysis of CRGs

Unsupervised consensus clustering algorithms were applied to classify patients with PC into distinct molecular patterns using the “ConsensusClusterPlus” R package, based on the expression of CRGs. Repetitions of 1,000 times were conducted to ensure the stability of classification. For clustering, a cumulative distribution function (CDF) was performed to determine the optimal number of subtypes. Principal component analysis (PCA) was employed to confirm the existence of distinct subgroups.

Functional enrichment analysis

The R package “GSVA” were used to perform GSVA analysis to determine the biological function differences between distinct subtypes. The HALLMARK gene set “h.all.v2022.1.Hs.symbols.gmt,” obtained from the MSigDB database (https://www.gsea-msigdb.org/gsea/index.jsp), was used to run the GSVA analysis(Liberzon et al. 2015). The R package “clusterProfiler” was employed to conduct gene ontology (GO) and Kyoto Encyclopedia of Genes and Genomes (KEGG) analyses.

Cuproptosis risk model construction

A total of 306 patients with PC were randomly allocated into training and testing groups at a ratio of 1:1. The Differential Expressed Genes (DEGs) between three cuproptosis subtypes were compared in pairs three times with a Log2(fold change) > 0.585 and adjusted P value < 0.05, and the intersecting genes were used to create cuproptosis gene signatures. Next, univariate significant prognostic correlation genes were acquired through univariate Cox regression analysis. Based on the univariate significant genes, LASSO Cox regression analysis was conducted to minimize the least absolute shrinkage between signatures. Multivariate Cox regression was conducted to screen out the best risk gene and to obtain the correlation coefficient. The cuproptosis model was constructed as follows: cuproptosis score \(= {\sum }_{i = 1}^{n}\text{e}\text{x}\text{p}\left(\text{X}\text{i}\right)\times \text{c}\text{o}\text{e}\text{f}\left(\text{X}\text{i}\right)\), where exp(Xi) represents the expression level and coef(Xi) represents the coefficient.

According to the median risk score, all patients with PC were classified into high- and low-risk groups. The specificity and sensitivity of this risk model were evaluated with time-dependent receiver operating characteristics (ROC).

Prognosis analysis of PC

The “survival” and “survminer” packages were used to plot and analyze the Kaplan–Meier survival. Patients with PC were classified into subtypes, including the cuproptosisCluster subtype, geneCluster subtype, CuproptosisScore subtype, and TMB subtype.

Assessment of the TME

The CIBERSORT algorithm was employed to analyze the fractions of 22 human immune cell phenotypes in each patient with PC. The sum of the proportion of all calculated immune cell phenotypes in every tissue was equal to 1. Single-sample gene set enrichment analysis (ssGSEA) was adopted to analyze the immune function and relative immune cell infiltration in the PC TME(Rooney et al. 2015). Spearman’s analysis was used to investigate the differential expression of the immune checkpoint blockade (ICB) gene between two CuproptosisScore groups and the results are displayed as boxplots.

Drug sensitivity analysis

To calculate the drug sensitivity of each sample, 198 drugs, abstracted from the genomics of drug sensitivity in cancer v2 (GDSC) (https://www.cancerrxgene.org/) data source, were calculated by the “oncoPredict” R package(Maeser et al. 2021). P values < 0.05 were considered to indicate a significant difference.

Construction of nomograms

Nomograms, including clinical characteristics and risk subtypes, were constructed to predict the 1-, 2-, and 3-year OS quantitatively. Calibration plots describe the prognostic value between the predicted and actual value. The nomogram and calibration plot analysis used the “rms” R package.

Statistical analysis

All data analyses were conducted by R software (4.2.1 version). The continuous data are described as (x ± s). For two group tests, normally distributed variable data were calculated with Student’s t-tests, while nonnormally distributed variable data were analyzed using the Wilcoxon test. For more than two groups, the Kruskal-Wallis test was employed. P values < 0.05 were considered to indicate statistical significance.

Differential expression and genetic variation of CRGs in pancreatic cancer

Nineteen genes related to cuproptosis were included in this study. We first summarized the frequency of copy number variation (CNV) and somatic mutation of the 19 CRGs in 179 PC samples. Thirty-seven PC samples showed somatic mutations with a 21.39% incidence (Fig. 1A). As shown in Fig. 1A, CDKN2A experienced the highest mutation rate with a frequency of 17%. CNV analysis showed a copy number loss in ten CRGs and a gain in seven genes, while LIPT1 and PDHA1 showed an equal CNV (Fig. 1B). The location of the CNV alterations on the chromosomes is shown in Fig. 1C. A total of 179 PAAD patients from TCGA and 167 healthy tissues from GTEx were used to analyze the mRNA differential expression of 19 CRGs to further study the influence of CNV and somatic mutation. The results showed that the expression of NFE2L2, NLRP3, ATP7B, ATP7A, SLC31A1, FDX1, DLAT, DBT, DLST, CDKN2A, and MTF1 was upregulated, while that of LIAS, LIPT1, GLS, PDHB, GCSH, and PDHA1 was downregulated in PC tissues (Fig. 1D). These results were somewhat consistent with CNV change, which indicated that the expression of 19 CRGs may be influenced by CNV. We next employed a correlation of each CRG and the results indicated a strong positive correlation between PDHA1 and GCSH (Fig. 1E). To further study the cross-talk between CRGs, we constructed PPI analysis by STRING (Fig. 1F). Methylation analysis through the UALCAN database showed that the promoter methylation level of CDKN2A, ATP7B, LIPT2, FDX1, LIAS, GLS, and NFE2L2 was upregulated, while DLD was downregulated (Fig. S2).

Identification of CRG clusters in PC

A total of 310 patients with PC from TCGA-PAAD, GSE62452, and GSE57495 were enrolled for further analyses. The prognostic roles of 17 CRGs in patients with PC were analyzed by univariate Cox regression (Fig. S3A). The Kaplan–Meier curves revealed that patients with PC with lower expression of DTB, ATP7A, PHB, DALT, DLST, and NFE2L2 had a longer overall survival time compared to those with higher expression of these CRGs. In contrast, patients with PC with higher expression of DLD, LIPT1, and LIAS showed longer overall survival times (Fig. S3J–K). We constructed a network map for 17 CRGs to demonstrate the overall landscape of gene interaction, relationships, and the prognostic impact on patients with PC (Fig. 2A). To further clarify samples with distinct expression signatures of CRGs in PC, a consensus clustering algorithm was developed to group the patients with PC based on the expression levels of 17 CRGs. The results showed that K = 3 had the best clustering results from K = 2 to K = 9 (Fig. 2B–D). The PC samples were categorized according to three different genes into CRGcluster A (n = 86), CRGcluster B (n = 66), and CRGcluster C (n = 154) (Fig. 2F). CRGcluster A demonstrated high expression of SLC31A1, MTF1, NFE2L2, ATP7A, DLAT, and DBT, while low expression of DBT, DLAT, ATP7A, MTF1 was found in CRGcluster B (Fig. 2E). Next, the relationship between the clinical characteristic and three clusters was explored and is shown in Fig. 2F. A significant difference in OS between the three CRGclusters was found through Kaplan–Meier analysis and CRGcluster B had the best OS for PC, followed by CRGclusters C and A. The results of principal component analysis (PCA) revealed that the PC sample could be divided into clusters A, B, and C, based on the median value of the cuproptosis score (Fig. 2G).

TME characteristic in distinct CRGclusters

Next, GSVS was employed with hallmark gene sets to study the biological functions in three CRGclusters. CRGcluster A was significantly enriched in the tumorigenesis-related pathway and stromal pathways, including the TGF β, PI3K-AKT-MTOR, and NOTCH pathways. CRGcluster B was enriched in coagulation and pancreatic β cell pathways (Fig. 3A–C). ssGSEA analysis showed that CRGcluster A was enriched in innate immune cells and immunocompetent cells, including Th1 cells, regulatory T cells, Mast cells, dendritic cells, neutrophils, Gamma delta T cells, immature B cells, natural killer T cells, natural killer cells, and active CD4 T cells compared to CRGclusters B and C. CRGcluster B was enriched in CD56dim natural killer cells and monocytes (Fig. 3D).

Generation of CRG signatures in PC

We next investigated the underlying biological characteristics of the three cuproptosis subgroups. Eighteen intersected DEGs were identified using the “limma” package and were compared between the three subgroups in pairs three times (Fig. 4A). GO and KEGG enrichment analyses were conducted using “ClusterProfler” packages based on these DEGs. GO analysis showed that these DEGs participated in the cellular component of the glutamatergic synapse, the external side of the plasma membrane, and the cell-cell junction. Molecular function (MF) was significantly related to receptor ligand activity, signaling receptor activator activity, and growth factor receptor binding, and biological process (BP) was largely involved in the cytokine-mediated signaling pathway, and negative chemotaxis (Fig. 4B). The KEGG pathway results revealed that the DEGs were involved in cytokine-cytokine receptor interaction, platinum drug resistance, Toll-like receptor signaling pathway, viral protein interaction with cytokine and cytokine receptor, and apoptosis, indicating the vital roles of cuproptosis in the immune regulation of PC (Fig. 4C).

To further analyze the important roles, univariate Cox regression was performed to identify the relationship between the 18 cluster-related DEGs and the prognosis in PC (Fig. S4A). A total of 17 CRGs were enrolled with significant prognostic P values < 0.05. Consensus clustering analysis was implemented and used to classify all patients with PC into two subtypes, termed genecluster A and genecluster B (Fig. 4D). The PCA results further confirmed significant differences between geneclusters A and B (Fig. S4G). Kaplan–Meier analysis revealed that the OS in genecluster B patients with PC was higher than that in genecluster A (Fig. 4E). According to the heatmap transcriptomic profile, genecluster A exhibited increased expression levels of BIRC3, GBP9, CXCL10, ECT2, and ZNF770 (Fig. 4F). Additionally, differential gene expression related to CRGs between geneclusters A and B were analyzed. Figure 4G shows that the gene expression of 13 CRGs, including NFE2L2, ATP7B, FDX1, and DLD, were significantly higher in cluster A.

Construction of the cuproptosis model

The 306 patients with PC were randomly separated into a training set (153 samples) set and a testing set (153 samples) to explore the prognostic role of CRG-related DEGs. LASSO regression was used to reduce the individual complexity and heterogeneity of patients with PC, and six significant genes (TOP2A, FLRT3, IGF2BP3, PCSK1N, EREG, CXCL10) were identified (Figs. 5A, 5B). We then performed a multivariate Cox regression analysis of six significant genes and identified FLRT3, PCSK1N, EREG, and CXCL10 as being the most important genes. The cuproptosis model was established based on the following formula: Risk score = Exp (FLRT3) × (0.3043) + Exp (PCSK1N) × (–0.1031) + Exp (EREG) × (0.1063) + Exp (CXCL10) × (0.1368). All set files were combined by the training group and testing group files. The alluvial diagram shows the distribution of CRG scores with three CRGclusters, two geneclusters, two CuproptosisScore groups, and the PC patient’s survival status. The patients in geneCluster B were all in the low-risk group (Fig. 5C), and significant differences in expression were found between the three CRGclusters and the two geneclusters in the cuproptosis score (Figs. 5D, 5E). The boxplot result revealed that the expression of 11 CRGs differed between the high and low-risk groups (Fig. 5F).

Validation of the prognostic value of the predicted risk model

Kaplan–Meier survival curves were used to analyze the prognosis in the total set, testing set, and training set. The Kaplan–Meier curve revealed that patients with low-risk scores had a significantly better survival rate than those in the higher-risk score group in the total, testing, and training sets (P < 0.05) (Fig. 6A–C). Additionally, the time ROC curves were used to evaluate the specificity and sensitivity, and the AUC values showed acceptable 1-, 3-, and 5-year OS in three groups. The AUC values for the total, testing, and training sets were 0.654, 0.609, and 0.694, at 1 year, 0.724, 0.703,0.761, at 3 years, and 0.754,0.702,0.892, at 5 years, respectively (Fig. 6D–F). These results indicated a promising predictive performance in the survival of patients with PC. The distribution of risk score and survival status is shown in Fig. 7A–D. Both in training and testing sets, the death rate increased and survival time decreased with an increasing risk score, indicating a poor survival status in a higher-risk group. The heatmap indicates that the expression of FLRT 3, EREG, and CXCL10 in the high-risk group was higher than that in the low-risk group (Fig. 7G–I).

Building a cuproptosis‑predicting nomogram for PC

We conducted univariate and multivariate Cox regression to estimate whether the cuproptosis score is an independent prognostic biomarker for PC. Univariate Cox regression analysis indicated that the risk score was an independent risk factor of OS in both TCGA and GEO cohorts (Fig. 8A&E). Multivariate Cox regression also showed that the risk score was an independent prognostic biomarker for PC (Figs. 8B&F). The predictive nomogram was employed to predict the 1-, 2-, and 3-year OS for PC in both TCGA and GEO (Figs. 8C&G). The calibration curves indicated a relative link between predicted and observed OS (Figs. 8D&I).

TME between the two risk groups

ssGSEA analysis showed a significant difference in the immune infiltration of iDCs, Th1, Th2, and Treg cells found between the high- and low-risk groups (Fig. 9A). Nine immune functions showed a remarkable difference between these two groups (Fig. 9B). Correlation analysis between immune cells and four risk genes showed that EREG significantly correlated with M1 macrophages (Fig. 9C). The estimation of the ICB gene between the two risk score groups showed that the expression of most ICG genes, including PDCD1LG2 (PD-L2), CD274 (PD-L1), and CTLA4, were higher in the high-risk group. The cuproptosis risk score revealed a positive correlation with resting mast cells, M1 macrophages, CD4 memory-activated T cells, and neutrophils, and a negative correlation with CD8 T cells, naive B cells, and Tregs cells (Fig. S5).

Mutation and drug susceptibility analysis

We next explored the differences in tumor somatic mutation in the high- and low-risk groups. As shown in Figs. 10A–B, the top five mutated genes were CDKN2A, SMAD4, TP53, KRAS, and TTN in both groups, and their frequency of mutation all exceeded 10%. Additionally, Kaplan–Meier analysis showed that the OS in the high TMB group was lower than that in the low TMB group (Fig. 10C). The combination of risk score with TMB also indicated that the high-risk score plus high-TMB group exhibited a worse OS (Fig. 10F). We next conducted drug predictive analysis to estimate the sensitivity of 189 drugs between the high- and low-risk groups using the R package “oncoPredict.” The results demonstrated that 50 drugs were significantly different between the two groups. For instance, the low-risk group showed greater sensitivity to tamoxifen, oxaliplatin, and irinotecan (Fig. 10E–G), whereas the high-risk group showed a lower IC50 for afatinib, bortezomib, and sapitinib (Fig. 10H–J).

PC is a highly aggressive cancer with an extremely high mortality rate, which is largely ascribed to difficulties in achieving early diagnosis and the lack of effective measures(Zhou et al. 2017). Given the nonspecific clinical signs and lack of reliable and sensitive biomarkers for the early detection and treatment of PC, alternative treatment modalities and a reliable signature to predict prognosis are urgently needed. Cuproptosis, a newly detected mechanism of cooper-related regulated cell death, has proved to be a reliable biomarker in hepatocellular carcinoma, breast cancer, and colorectal cancer(Sha et al. 2022;Wang et al. 2022;Zhu et al. 2022).

In the present study, we evaluated 19 CRGs and investigated their somatic mutation, CNV, differential expression level, PPI network, methylation analysis, and biological pathway in PC. The higher frequency mutation rate of CDKN2A and downregulated CNV level further confirmed that at the time of malignancy, CNV is the main type of somatic mutation in the DNA sequence(Zhang et al. 2022). These results also suggested that CNV may interfere with the expression of 19 cuproptosis genes. The methylation level of the 19 CRGs was differentially expressed, with seven CRGs showing upregulated methylation and only DLD displaying downregulated methylation. The results indicated that methylation can regulate the gene expression level of cuproptosis in PC. Correlation analysis of the 19 CRGs indicated that the expression of most showed a significant correlation.

Compared to the other two groups, CRGcluster A showed a high expression level of most CRGs. In contrast, the prognosis of CRGcluster A was the poorest among the three groups, indicating that high expression of CRGs in patients with PC may lead to a poor overall survival rate, which is consistent with a previous study on hepatocellular carcinoma(Wang et al. 2022). GAVS enrichment analysis demonstrated that CRGcluster A was enriched in tumor-related pathways and stromal activation, including the PI3K-AKT-mTOR and TGFβ pathways. The present study also indicated a significant difference in infiltration in most TME immune cells. CRGcluster A was enriched in the most immune cell infiltration than the other two groups, such as natural killer T cells, natural killer cells, and active CD4 T cells. These results are consistent with previous research showing that CRGcluster A was enriched in immune cell and stromal pathways, corresponding to immune-excluded phenotypes. Due to immune cell infiltration and immune activation, CRGcluster B was classified into immune-inflamed phenotypes(Chen and Mellman. 2017). We next performed consensus clustering analysis on 18 DGEs combined with prognosis to classify patients with PC into two gene subtypes. Genecluster A experienced a high expression level of most of CRGs and DGEs, and a poorer prognosis, further confirming that higher expression of CRGs can generate a worse survival rate. Consequently, a comprehensive analysis of CRGs can help to construct a relatively accurate prognosis model to predict the survival rate of PC. Additionally, a greater understanding of the relationship between CRGs and the TME could help clinicians to assess the value of immunotherapy as a treatment modality.

Due to the individual heterogeneity of CRGs, we constructed a risk score system across training and testing groups to better identify the cuproptosis signature of the individual patient. The expression of 11 CRGs in the high-risk group was significantly higher than that in the low-risk group, and the cuproptosis risk score was associated with the prognosis of PC. Interestingly, CRGcluster A, which was characterized by immune-excluded phenotypes, showed a higher risk score compared to CRGclusters B and C, and the high-risk score subtypes experienced a poor OS. In contrast, CRGcluster B was regarded as an immune-inflamed phenotype, which exhibited a lower risk score and achieved a better prognosis. Taken together, our results correlated with colorectal cancer in that the immune-inflamed phenotype experienced a better survival rate, while the immune-excluded phenotype showed a lower survival rate(Wang et al. 2022). Univariate and multivariate Cox regression further confirmed that the cuproptosis risk score is an independent predictor of prognosis for patients with PC. Therefore, we can conclude that the cuproptosis risk score is a robust and reliable model for the estimation of individual samples and can be utilized to deduce the prognosis of PC and tumor immune subtypes(W. Li et al. 2022;Oliveri. 2022).

Increasing evidences have indicated that TME plays an important role in tumorigenesis and immunotherapy in PC(Bockorny et al. 2022;K. Li et al. 2022;Wood et al. 2022). The result of immune cell infiltration revealed that CuproptosisScore was negatively correlated with some anticancer immune cells, including naive B cells, CD8 + T cells, and Treg cells, which was consistent with the previous m5C and m6A risk score, which showed that higher CD8 + T cell infiltration was indicative of favorable prognosis(Xu et al. 2022;Yun et al. 2022). Therefore, we assumed that cuproptosis may be linked with the regulation of the TME and may contribute to tumor growth, progression, and even metastasis. Immunotherapy, especially ICB, such as PD-1, PDL-1, and CTLA4, has provided a novel perspective for cancer treatment(Balachandran et al. 2019;Reiss et al. 2022;Wang et al. 2023;Wood et al. 2022). Regarding the expression level of the ICB gene, the high-risk score group showed a high level of CD274 and CTLA4, suggesting that these patients were more amenable to anti-PD-L1 and anti-CTLA-4 therapy.

The difference between the two CuproptosisScore groups may be due to the crosstalk between immune and tumor cells(Wood et al. 2022). The effect of single-agent immunotherapy for PC remains unsatisfactory due to the complex tumor immune microenvironment of the PC, especially the large amount of immune-suppressor cells and stromal cells(Wood et al. 2022;Wu et al. 2019;Zhu et al. 2017). Fortunately, targeting the complex TIME and the combination of immunomodulatory therapy and the target drugs has reported to be a promising treatment choice for PC(K. Li et al. 2022;Padrón et al. 2022;Reiss et al. 2022). In summary, the CuproptosisScore was correlated with immunotherapy and may contribute to better therapeutic decision-making for PC.

The TMB has been demonstrated to play an important role in the diagnosis, treatment, and prognostic prediction of PC(Lawlor et al. 2021;Wu et al. 2019). Indeed, TMB analysis revealed that the high-risk group had a higher KRAS and TP53 mutation rate and a poor prognosis, suggesting that the high-risk group may respond better to immunotherapy(Di Federico et al. 2022;Lawlor et al. 2021). Moreover, the frequency of KRAS mutation is the highest in this study, which was consistent with previous studies(He et al. 2022). KRAS was regarded as an “untargetable” oncogene of PC for a long time, but now, the KRAS G12D inhibitor is increasingly being considered as a potential application for PC(He et al. 2022). Increasing evidence has suggested that chemotherapy drug resistance is a major challenge in the treatment of PC(Schuth et al. 2022;Springfeld et al. 2023). Therefore, we performed an anti-drug sensitivity analysis for different cuproptosis score groups to identify whether the cuproptosis risk model could predict drug candidates. We found that patients with PC with a high-risk score had a lower IC50 for afatinib, bortezomib, and sapitinib, while the low-risk group showed better sensitivity to commonly used chemotherapy agents, such as tamoxifen, oxaliplatin, and Irinotecan; this indicates that the cuproptosis model has value in improving the selection of anticancer agents and in predicting drug sensitivities in patients with PC.

However, our study had some limitations that warrant discussion. First, our study was a secondary analysis and the data were enrolled from TCGA and GEO public databases; therefore, in vivo and in vitro experiments should be conducted to verify our findings. Second, this was a retrospective study based on public data and may result in selection bias of samples. Therefore, well-designed prospective clinical studies are necessary to validate the clinical utility of the cuproptosis model for PC. Third, the risk gene defined in the present study needs further experimental work to confirm its authenticity and reliability.

In this study, we conducted a systematic analysis of CRG patterns, including the regulatory effects on different clinical-pathological features, the TME, and prognosis. We also constructed a new cuproptosis risk model to better predict the prognosis and anti-tumor drug sensitivity for patients with PC. This cuproptosis model could provide guidance for the design of individualized treatment strategies and accurate prognosis prediction for patients with PC.

Author contributions

Guozhen Huang conducted the formal analysis and wrote the original draft. Zhipeng Liu and Dai Luo performed the project administration. Guozhen Huang, Guoqing Ouyang and Tingfeng Xu participated in software analysis. Guozhen Huang, Guoqing Ouyang and Jie Wei conducted data curation. Guozhen Huang contributed to writing, reviewing, and editing the article. Guandou Yuan and Songqing He contributed to writing, reviewing, and editing the article. Guoqing Ouyang, Guandou Yuan and Songqing He provided funding acquisition. All authors contributed to the article and approved the submitted version.

Funding information

This work was supported in part by the 111 Project (D17011); Guangxi science and technology base and talent project (GuikeAA21220002); Special project of central government guiding local science and technology development (ZY20198011); Key Laboratory Base of Liver Injury and Repair of the First Affiliated Hospital of Guangxi Medical University (YYZS2020001); Hubei Chen Xiaoping Technology Development Fund (CXPJJH1190000-2019321); the project of Liuzhou People’s Hospital (lryjj201904).

Data availability statement

This data can be found here: the TCGA-PAAD (https://portal.gdc. cancer.gov/) and GSE57495 and GSE62452 in GEO (https:// www.ncbi.nlm.nih.gov/geo/).

Acknowledgments

We thank LetPub (www.letpub.com) for its linguistic assistance during the preparation of this manuscript.

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

Ethics approval The data was downloaded from public databases and did not require ethical approval. However, we followed the regulations set by the public databases.

Aubert, L., Nandagopal, N., Steinhart, Z., Lavoie, G., Nourreddine, S., Berman, J., Saba-El-Leil, M. K., Papadopoli, D., Lin, S., Hart, T., Macleod, G., Topisirovic, I., Gaboury, L., Fahrni, C. J., Schramek, D., Meloche, S., Angers, S., Roux, P. P. (2020). Copper bioavailability is a KRAS-specific vulnerability in colorectal cancer. Nat Commun, 11(1), 3701. doi: 10.1038/s41467-020-17549-y
Balachandran, V. P., Beatty, G. L., Dougan, S. K. (2019). Broadening the Impact of Immunotherapy to Pancreatic Cancer: Challenges and Opportunities. Gastroenterology, 156(7), 2056–2072. doi: 10.1053/j.gastro.2018.12.038
Bockorny, B., Grossman, J. E., Hidalgo, M. (2022). Facts and Hopes in Immunotherapy of Pancreatic Cancer. Clin Cancer Res, 28(21), 4606–4617. doi: 10.1158/1078-0432.CCR-21-3452
Cancer, C. A. A. C. (2023). China Anti-Cancer Association guidelines for the holistic integrative management of pancreatic cancer (abridged version). Chin J Clin Oncol, 50(10), 487–496. doi: 10.12354/j.issn.1000-8179
Chen, D. S., Mellman, I. (2017). Elements of cancer immunity and the cancer–immune set point. Nature, 541(7637), 321–330. doi: 10.1038/nature21349
Chen, J., Jiang, Y., Shi, H., Peng, Y., Fan, X., Li, C. (2020). The molecular mechanisms of copper metabolism and its roles in human diseases. Pflugers Arch, 472(10), 1415–1429. doi: 10.1007/s00424-020-02412-2
Di Federico, A., Mosca, M., Pagani, R., Carloni, R., Frega, G., De Giglio, A., Rizzo, A., Ricci, D., Tavolari, S., Di Marco, M., Palloni, A., Brandi, G. (2022). Immunotherapy in Pancreatic Cancer: Why Do We Keep Failing? A Focus on Tumor Immune Microenvironment, Predictive Biomarkers and Treatment Outcomes. Cancers, 14(10), 2429. doi: 10.3390/cancers14102429
Feng, Y., Zeng, J., Ma, Q., Zhang, S., Tang, J., Feng, J. (2020). Serum copper and zinc levels in breast cancer: A meta-analysis. J Trace Elem Med Biol, 62, 126629. doi: 10.1016/j.jtemb.2020.126629
Guan, X., Lu, N., Zhang, J. (2022). The combined prognostic model of copper-dependent to predict the prognosis of pancreatic cancer. Front Genet, 13, 978988. doi: 10.3389/fgene.2022.978988
Gul, N. S., Khan, T., Chen, M., Huang, K., Hou, C., Choudhary, M. I., Liang, H., Chen, Z. (2020). New copper complexes inducing bimodal death through apoptosis and autophagy in A549 cancer cells. J Inorg Biochem, 213, 111260. doi: 10.1016/j.jinorgbio.2020.111260
He, Q., Liu, Z., Wang, J. (2022). Targeting KRAS in PDAC: A New Way to Cure It? Cancers, 14(20), 4982. doi: 10.3390/cancers14204982
Hingorani, S. R. (2023). Epithelial and stromal co-evolution and complicity in pancreatic cancer. Nat Rev Cancer, 23(2), 57–77. doi: 10.1038/s41568-022-00530-w
Huang, X., Zhou, S., Tóth, J., Hajdu, A. (2022). Cuproptosis-related gene index: A predictor for pancreatic cancer prognosis, immunotherapy efficacy, and chemosensitivity. Front Immunol, 13, 978865. doi: 10.3389/fimmu.2022.978865
Jiang, T., Yang, T., Chen, Y., Miao, Y., Xu, Y., Jiang, H., Yang, M., Mao, C. (2022). Emulating interactions between microorganisms and tumor microenvironment to develop cancer theranostics. Theranostics, 12(6), 2833–2859. doi: 10.7150/thno.70719
Lawlor, R. T., Mattiolo, P., Mafficini, A., Hong, S., Piredda, M. L., Taormina, S. V., Malleo, G., Marchegiani, G., Pea, A., Salvia, R., Kryklyva, V., Shin, J. I., Brosens, L. A., Milella, M., Scarpa, A., Luchini, C. (2021). Tumor Mutational Burden as a Potential Biomarker for Immunotherapy in Pancreatic Cancer: Systematic Review and Still-Open Questions. Cancers, 13(13), 3119. doi: 10.3390/cancers13133119
Li, K., Tandurella, J. A., Gai, J., Zhu, Q., Lim, S. J., Thomas, D. L., Xia, T., Mo, G., Mitchell, J. T., Montagne, J., Lyman, M., Danilova, L. V., Zimmerman, J. W., Kinny-Köster, B., Zhang, T., Chen, L., Blair, A. B., Heumann, T., Parkinson, R., Durham, J. N., Narang, A. K., Anders, R. A., Wolfgang, C. L., Laheru, D. A., He, J., Osipov, A., Thompson, E. D., Wang, H., Fertig, E. J., Jaffee, E. M., Zheng, L. (2022). Multi-omic analyses of changes in the tumor microenvironment of pancreatic adenocarcinoma following neoadjuvant treatment with anti-PD-1 therapy. Cancer Cell, 40(11), 1374–1391. doi: 10.1016/j.ccell.2022.10.001
Li, W., Zhang, X., Chen, Y., Pang, D. (2022). Identification of cuproptosis-related patterns and construction of a scoring system for predicting prognosis, tumor microenvironment-infiltration characteristics, and immunotherapy efficacy in breast cancer. Front Oncol, 12, 966511. doi: 10.3389/fonc.2022.966511
Liberzon, A., Birger, C., Thorvaldsdóttir, H., Ghandi, M., Mesirov, J. P., Tamayo, P. (2015). The Molecular Signatures Database Hallmark Gene Set Collection. Cell Syst, 1(6), 417–425. doi: 10.1016/j.cels.2015.12.004
Maeser, D., Gruener, R. F., Huang, R. S. (2021). oncoPredict: an R package for predictingin vivo or cancer patient drug response and biomarkers from cell line screening data. Brief Bioinform, 22(6), b260. doi: 10.1093/bib/bbab260
Michniewicz, F., Saletta, F., Rouaen, J. R. C., Hewavisenti, R. V., Mercatelli, D., Cirillo, G., Giorgi, F. M., Trahair, T., Ziegler, D., Vittorio, O. (2021). Copper: An Intracellular Achilles’ Heel Allowing the Targeting of Epigenetics, Kinase Pathways, and Cell Metabolism in Cancer Therapeutics. ChemMedChem, 16(15), 2315–2329. doi: 10.1002/cmdc.202100172
Oliveri, V. (2022). Selective Targeting of Cancer Cells by Copper Ionophores: An Overview. Front Mol Biosci, 9, 841814. doi: 10.3389/fmolb.2022.841814
Padrón, L. J., Maurer, D. M., O Hara, M. H., O Reilly, E. M., Wolff, R. A., Wainberg, Z. A., Ko, A. H., Fisher, G., Rahma, O., Lyman, J. P., Cabanski, C. R., Yu, J. X., Pfeiffer, S. M., Spasic, M., Xu, J., Gherardini, P. F., Karakunnel, J., Mick, R., Alanio, C., Byrne, K. T., Hollmann, T. J., Moore, J. S., Jones, D. D., Tognetti, M., Chen, R. O., Yang, X., Salvador, L., Wherry, E. J., Dugan, U., O Donnell-Tormey, J., Butterfield, L. H., Hubbard-Lucey, V. M., Ibrahim, R., Fairchild, J., Bucktrout, S., LaVallee, T. M., Vonderheide, R. H. (2022). Sotigalimab and/or nivolumab with chemotherapy in first-line metastatic pancreatic cancer: clinical and immunologic analyses from the randomized phase 2 PRINCE trial. Nat Med, 28(6), 1167–1177. doi: 10.1038/s41591-022-01829-9
Reiss, K. A., Mick, R., Teitelbaum, U., O'Hara, M., Schneider, C., Massa, R., Karasic, T., Tondon, R., Onyiah, C., Gosselin, M. K., Donze, A., Domchek, S. M., Vonderheide, R. H. (2022). Niraparib plus nivolumab or niraparib plus ipilimumab in patients with platinum-sensitive advanced pancreatic cancer: a randomised, phase 1b/2 trial
. Lancet Oncol, 23(8), 1009–1020. doi: 10.1016/S1470-2045(22)00369-2
Risom, T., Glass, D. R., Averbukh, I., Liu, C. C., Baranski, A., Kagel, A., McCaffrey, E. F., Greenwald, N. F., Rivero-Gutiérrez, B., Strand, S. H., Varma, S., Kong, A., Keren, L., Srivastava, S., Zhu, C., Khair, Z., Veis, D. J., Deschryver, K., Vennam, S., Maley, C., Hwang, E. S., Marks, J. R., Bendall, S. C., Colditz, G. A., West, R. B., Angelo, M. (2022). Transition to invasive breast cancer is associated with progressive changes in the structure and composition of tumor stroma. Cell, 185(2), 299–310. doi: 10.1016/j.cell.2021.12.023
Rooney, M. S., Shukla, S. A., Wu, C. J., Getz, G., Hacohen, N. (2015). Molecular and Genetic Properties of Tumors Associated with Local Immune Cytolytic Activity. Cell, 160(1–2), 48–61. doi: 10.1016/j.cell.2014.12.033
Schuth, S., Le Blanc, S., Krieger, T. G., Jabs, J., Schenk, M., Giese, N. A., Büchler, M. W., Eils, R., Conrad, C., Strobel, O. (2022). Patient-specific modeling of stroma-mediated chemoresistance of pancreatic cancer using a three-dimensional organoid-fibroblast co-culture system. J Exp Clin Cancer Res, 41(1), 312. doi: 10.1186/s13046-022-02519-7
Sha, S., Si, L., Wu, X., Chen, Y., Xiong, H., Xu, Y., Liu, W., Mei, H., Wang, T., Li, M. (2022). Prognostic analysis of cuproptosis-related gene in triple-negative breast cancer. Front Immunol, 13, 922780. doi: 10.3389/fimmu.2022.922780
Siegel, R. L., Miller, K. D., Fuchs, H. E., Jemal, A. (2022). Cancer statistics, 2022. CA Cancer J Clin, 72(1), 7–33. doi: 10.3322/caac.21708
Springfeld, C., Ferrone, C. R., Katz, M. H. G., Philip, P. A., Hong, T. S., Hackert, T., Büchler, M. W., Neoptolemos, J. (2023). Neoadjuvant therapy for pancreatic cancer. Nat Rev Clin Oncol, 20(5), 318–337. doi: 10.1038/s41571-023-00746-1
Su, Y., Qi, R., Li, L., Wang, X., Li, S., Zhao, X., Hou, R., Ma, W., Liu, D., Zheng, J., Shi, M. (2022). An immune-related gene prognostic risk index for pancreatic adenocarcinoma. Front Immunol, 13, 945878. doi: 10.3389/fimmu.2022.945878
Tsvetkov, P., Coy, S., Petrova, B., Dreishpoon, M., Verma, A., Abdusamad, M., Rossen, J., Joesch-Cohen, L., Humeidi, R., Spangler, R. D., Eaton, J. K., Frenkel, E., Kocak, M., Corsello, S. M., Lutsenko, S., Kanarek, N., Santagata, S., Golub, T. R. (2022). Copper induces cell death by targeting lipoylated TCA cycle proteins. Science, 375(6586), 1254–1261. doi: 10.1126/science.abf0529
Vitale, I., Shema, E., Loi, S., Galluzzi, L. (2021). Intratumoral heterogeneity in cancer progression and response to immunotherapy. Nat Med, 27(2), 212–224. doi: 10.1038/s41591-021-01233-9
Wang, S. J., Dougan, S. K., Dougan, M. (2023). Immune mechanisms of toxicity from checkpoint inhibitors. Trends Cancer, 9(7), 543–553. doi: 10.1016/j.trecan.2023.04.002
Wang, W., Wang, X., Luo, J., Chen, X., Ma, K., He, H., Li, W., C, J. (2021). Serum Copper Level and the Copper-to-Zinc Ratio Could Be Useful in the Prediction of Lung Cancer and Its Prognosis: A Case-Control Study in Northeast China. Nutr Cancer, 73(10), 1908–1915. doi: 10.1080/01635581.2020.1817957
Wang, Y., Zhang, Y., Wang, L., Zhang, N., Xu, W., Zhou, J., Zhao, Y., Zhu, W., Zhang, T., Wang, L. (2022). Development and experimental verification of a prognosis model for cuproptosis-related subtypes in HCC. Hepatol Int, 16(6), 1435–1447. doi: 10.1007/s12072-022-10381-0
Wood, L. D., Canto, M. I., Jaffee, E. M., Simeone, D. M. (2022). Pancreatic Cancer: Pathogenesis, Screening, Diagnosis, and Treatment. Gastroenterology, 163(2), 386–402. doi: 10.1053/j.gastro.2022.03.056
Wu, A. A., Jaffee, E., Lee, V. (2019). Current Status of Immunotherapies for Treating Pancreatic Cancer. Curr Oncol Rep, 21(7), 60. doi: 10.1007/s11912-019-0811-5
Xu, H., Yin, L., Xu, Q., Xiang, J., Xu, R. (2022). N6-methyladenosine methylation modification patterns reveal immune profiling in pancreatic adenocarcinoma. Cancer Cell Int, 22(1), 199. doi: 10.1186/s12935-022-02614-x
Yamamoto, K., Venida, A., Yano, J., Biancur, D. E., Kakiuchi, M., Gupta, S., Sohn, A. S. W., Mukhopadhyay, S., Lin, E. Y., Parker, S. J., Banh, R. S., Paulo, J. A., Wen, K. W., Debnath, J., Kim, G. E., Mancias, J. D., Fearon, D. T., Perera, R. M., Kimmelman, A. C. (2020). Autophagy promotes immune evasion of pancreatic cancer by degrading MHC-I. Nature, 581(7806), 100–105. doi: 10.1038/s41586-020-2229-5
Yun, D., Yang, Z., Zhang, S., Yang, H., Liu, D., Grützmann, R., Pilarsky, C., Britzen-Laurent, N. (2022). An m5C methylation regulator-associated signature predicts prognosis and therapy response in pancreatic cancer. Front Cell Dev Biol, 10, 975684. doi: 10.3389/fcell.2022.975684
Zhang, C., Huang, R., Xi, X. (2022). Cuproptosis patterns in papillary renal cell carcinoma are characterized by distinct tumor microenvironment infiltration landscapes. Front Mol Biosci, 9, 910928. doi: 10.3389/fmolb.2022.910928
Zhou, B., Xu, J., Cheng, Y., Gao, J., Hu, S., Wang, L., Zhan, H. (2017). Early detection of pancreatic cancer: Where are we now and where are we going? Int J Cancer, 141(2), 231–241. doi: 10.1002/ijc.30670
Zhu, L., Staley, C., Kooby, D., El-Rays, B., Mao, H., Yang, L. (2017). Current status of biomarker and targeted nanoparticle development: The precision oncology approach for pancreatic cancer therapy. Cancer Lett, 388, 139–148. doi: 10.1016/j.canlet.2016.11.030
Zhu, Z., Zhao, Q., Song, W., Weng, J., Li, S., Guo, T., Zhu, C., Xu, Y. (2022). A novel cuproptosis-related molecular pattern and its tumor microenvironment characterization in colorectal cancer. Front Immunol, 13, 940774. doi: 10.3389/fimmu.2022.940774

No competing interests reported.

Download PDF

Version 1

posted

You are reading this latest preprint version

Identification of a novel cuproptosis-related pattern and its tumor microenvironment infiltration characteristics in pancreatic cancer

Status:

Version 1

Abstract

Purpose

Methods

Results

Conclusions

Figures

Background

Materials and Methods

Dataset and preprocessing

Differential expression analysis and PPI network construction

Mutation and correlation analysis

Unsupervised consensus clustering analysis of CRGs

Functional enrichment analysis

Cuproptosis risk model construction

Prognosis analysis of PC

Assessment of the TME

Drug sensitivity analysis

Construction of nomograms

Statistical analysis

Results

Differential expression and genetic variation of CRGs in pancreatic cancer

Identification of CRG clusters in PC

TME characteristic in distinct CRGclusters

Generation of CRG signatures in PC

Construction of the cuproptosis model

Validation of the prognostic value of the predicted risk model

Building a cuproptosis‑predicting nomogram for PC

TME between the two risk groups

Mutation and drug susceptibility analysis

Discussion

Conclusions

Declarations

References

Additional Declarations

Supplementary Files

Status:

Version 1