==== Front
Sci Rep
Sci Rep
Scientific Reports
2045-2322
Nature Publishing Group UK London

37836
10.1038/s41598-023-37836-0
Article
Telomere-related prognostic biomarkers for survival assessments in pancreatic cancer
Chen Shengyang chenshengyang2006@126.com

1
Hu Shuiquan 1
Zhou Baizhong 1
Cheng Bingbing 1
Tong Hao 1
Su Dongchao 1
Li Xiaoyong 1
Chen Yanjun 1
Zhang Genhao 2
1 grid.207374.5 0000 0001 2189 3846 Department of Hepatobiliary Pancreatic Surgery, Zhengzhou University Fifth Affiliated Hospital, Kangfu Front Street 3#, Zhengzhou, 450052 China
2 grid.412633.1 0000 0004 1799 0733 Department of Blood Transfusion, Zhengzhou University First Affiliated Hospital, Zhengzhou, China
30 6 2023
30 6 2023
2023
13 105862 2 2023
28 6 2023
© The Author(s) 2023
https://creativecommons.org/licenses/by/4.0/ Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/.
Human telomeres are linked to genetic instability and a higher risk of developing cancer. Therefore, to improve the dismal prognosis of pancreatic cancer patients, a thorough investigation of the association between telomere-related genes and pancreatic cancer is required. Combat from the R package “SVA” was performed to correct the batch effects between the TCGA-PAAD and GTEx datasets. After differentially expressed genes (DEGs) were assessed, we constructed a prognostic risk model through univariate Cox regression, LASSO-Cox regression, and multivariate Cox regression analysis. Data from the ICGC, GSE62452, GSE71729, and GSE78229 cohorts were used as test cohorts for validating the prognostic signature. The major impact of the signature on the tumor microenvironment and its response to immune checkpoint drugs was also evaluated. Finally, PAAD tissue microarrays were fabricated and immunohistochemistry was performed to explore the expression of this signature in clinical samples. After calculating 502 telomere-associated DEGs, we constructed a three-gene prognostic signature (DSG2, LDHA, and RACGAP1) that can be effectively applied to the prognostic classification of pancreatic cancer patients in multiple datasets, including TCGA, ICGC, GSE62452, GSE71729, and GSE78229 cohorts. In addition, we have screened a variety of tumor-sensitive drugs targeting this signature. Finally, we also found that protein levels of DSG2, LDHA, and RACGAP1 were upregulated in pancreatic cancer tissues compared to normal tissues by immunohistochemistry analysis. We established and validated a telomere gene-related prognostic signature for pancreatic cancer and confirmed the upregulation of DSG2, LDHA, and RACGAP1 expression in clinical samples, which may provide new ideas for individualized immunotherapy.

Subject terms

Cancer microenvironment
Gastrointestinal cancer
Tumour biomarkers
issue-copyright-statement© Springer Nature Limited 2023
==== Body
pmcIntroduction

The purpose of telomeres, which are highly conserved nuclear protein structures made up of TTAGGG repeating base sequences and found at the ends of eukaryotic linear chromosomes, is to prevent the DNA damage response (DDR) from being activated and to ensure genomic integrity1,2. In fact, the length of telomeres becomes shorter each time a somatic cell divides because it is impossible for DNA polymerase to completely replicate the ends of linear chromosomes. When the length of telomeres shortens to a certain limit, the cell promotes spontaneous apoptotic or senescence mechanisms3. Although most human somatic cells do not display telomerase activity due to telomerase reverse transcriptase (TERT) gene suppression, it does not apply to cancer cells. Cancer cells' progressive telomere shortening activation can be avoided by activating the telomere maintenance mechanism (TMM)4, which consists of two main forms, telomerase activation and Alternative-lengthening of Telomere (ALT). Although telomerase-positive cancer cells can maintain their telomere length and achieve unlimited proliferation capacity by activating telomerase, their telomeres are usually dramatically shorter than those of normal cells, mainly due to the late activation of telomerase during tumorigenesis; as for ALT-dependent cancers, although telomeres are longer, there is greater heterogeneity5. Although ALT is relatively rare in cancer6, it is observed with high frequency in pancreatic neuroendocrine carcinoma7. Pancreatic neuroendocrine carcinoma accounts for only 1–2% of pancreatic tumors, but the unique genetic profile that distinguishes it from gland-dependent pancreatic tumors warrants further in-depth analysis, which will help to improve the evaluation of risk assessment. Due to the small number, pancreatic neuroendocrine tumors were not included in this study. Activated telomerase keeps telomeres in equilibrium over time and reduces genomic instability to tolerable levels, thereby enabling the near-immortality of tumor cells. In most types of tumors in the TCGA database, longer telomeres are observed in normal adjacent tissues than in telomerase-positive tumors8. Therefore, the study of the correlation between tumors and telomeres has been a hot topic.

Telomeres play a two-sided role in the development of tumors. On the one hand, telomere shortening can inhibit cell proliferation and hinder tumor growth, while on the other hand, telomere shortening can also promote cancer development by causing widespread genomic instability. Although telomerase-activated cancers can lengthen telomeres, most tumors have shorter telomere lengths than normal tissue. Reduced DKC1 expression induces telomere-associated senescence and apoptosis in lung adenocarcinoma9. In addition, telomere shortening was a potential risk factor for pancreatic cancer10. These all suggest a potential use of telomere length as a predictive biomarker for tumors.

Considering that previous studies have focused on telomere length in pancreatic cancer and its role in pancreatic cancer prognosis, there is still a lack of relevant studies on the relationship between telomere-related genes and pancreatic cancer prognosis. In the present study, we constructed and validated a prognostic signature of pancreatic cancer based on the telomere-related genes, and also evaluated the underlying value of this signature in the selection of therapeutic drugs.

Methods

Acquisition of telomere-relevant genes and publicly available data

A total of 2086 telomere-relevant genes were downloaded from the TelNet database11. RNA sequencing (RNA-seq) data of The Cancer Genome Atlas-PAAD (TCGA-PAAD) cohort and the GTEx database obtained from the UCSC Xena project (https://xenabrowser.net) were combined as a discovery cohort for screening differentially expressed genes (DEGs) and prognostic model construction. After normalizing the RNA-seq data using a log2 (FPKM + 1) transformation, Combat from the R package “SVA” was performed to correct the batch effects between the two datasets. Samples with no complete survival data or survival time of less than 1 month were excluded. DEGs were screened out using an adjusted P value of less than 0.05 and |logFC| ≥ 1 as cutoff criteria. Data of the PACA-CN cohort was obtained from the ICGC database (https://dcc.icgc.org/), and GSE62452, GSE71729, and GSE78229 cohorts from the gene expression omnibus (GEO) database (https://www.ncbi.nlm.nih.gov/geo/) served as the test cohort.

Prognostic model construction and validation

The telomere-relevant DEGs were assessed to construct a prognostic risk model through univariate Cox regression, LASSO-Cox regression, and multivariate Cox regression as we previously reported12. Risk score = ∑iCoefficient (mRNAi)*Expression (mRNAi). We first divided the pancreatic cancer patients in the five datasets mentioned above into two subgroups according to a range of ideal risk scores determined by the R “survminer” package and then performed a Kaplan–Meier survival check to see if there was a difference in survival rates between the two groups. The ROC check was used to assess the prognostic effect of the model across the five datasets. Cox risk regression models were used to assess whether the prognostic model was an independent prognostic factor for patients with pancreatic cancer.

Characterization of functional analysis

To investigate substantially changed HALLMARK, GO, and KEGG items, the Metascape database was used13,14.

Immune cell infiltration and immune checkpoint gene analysis

Based on the expression patterns of multiple immune and stromal cell genes in tumor samples from pancreatic cancer patients, ESTIMATE was used to estimate the stromal and immune fractions of the tumor tissue15. A higher immune score represents a higher proportion of immune cell infiltration, a higher stromal score represents a higher proportion of stromal cell infiltration, and a higher ESTIMATE score represents a lower purity of tumor cells. Additionally, the abundance ratio of tumor-infiltrating immune cells (TIICs) was measured using the CIBERSORT16, xCELL17, MCPcounter18, and TIMER19 databases to more thoroughly examine the distribution variations of TIICs in the pancreatic cancer TME. Last but not least, we evaluated the variations in gene expression across distinct immune checkpoint subgroups for PD1, PD-L1, CD276, CTLA4, LAG3, CXCR4, IL1A, IL6, TGFB1, TNFRSF4, TNFRSF9, and PD-L2.

Potential therapeutic drug sensitivity analysis

We first downloaded NCI-60 compound activity data and RNA-seq expression profiles from the CallMiner database20 and selected Food and Drug Administration (FDA)-approved drugs or drugs from clinical trials for subsequent analysis. Based on the differential expression of the three genes in the prognostic model, we assessed the scores of each sample and analyzed the correlation between drugs and scores. Drugs were deemed tumor-sensitive if their adjusted P values were less than 0.001 and their Pearson correlation coefficients were greater than 0.3. Then, we contrasted the variations in half maximum inhibitory concentration (IC50) of tumor-sensitive medicines amongst various scores groupings. The results were declared statistically significant when the P value was less than 0.05.

Immunohistochemistry (IHC) evaluation

A validation cohort was created using 30 samples of pancreatic cancer and associated paracancerous tissues that had been validated by normal pathological investigation from our pathology department. All samples have comprehensive clinical information. Following serial sectioning of the tissue wax blocks, sample sections were collected and kept in a 4 °C freezer for later use. We subsequently performed IHC assays on formalin-fixed de-paraffinized sections using DSG2 (Abcam, ab150372), LDHA (Abcam, ab52488), and RACGAP1 (Abcam, ab134972) antibodies and took images with microscopy as described previously21.

Statistical analysis

Continuously distributed numerical data were evaluated using the grouped t-test or Mann–Withney–Wilcoxon test, whilst the qualitative data were examined using the chi-square test. Using the survival and survminer package, Kaplan–Meier survival analysis was done to determine overall survival rates between groups. Using the ‘timeROC’ package, receiver operating characteristic (ROC) curves were created to determine the area under the curve (AUC) values at 1, 2, and 3 years. To determine if the signature was an independent prognostic predictor, univariate and multivariate Cox regression analyses were carried out. Based on the findings, a nomogram was created using the RMS package. The Pearson or Spearman correlation test was used to do the correlation study. The results were declared statistically significant when the P value was less than 0.05.

Ethics approval and consent to participate

This study was approved by the Ethics Committees of Zhengzhou University. Written informed consent was obtained from all patients. All methods were performed following the relevant guidelines and regulations.

Results

Identification of telomere-related DEGs and roles of telomere-related genes

After batch effect removal, we screened out 6329 DEGs between the 171 normal pancreatic tissues and 178 pancreatic cancer tissues (Fig. 1A). Then, 502 telomere-related DEGs were identified (Fig. 1B). In addition, these telomere-associated DEGs could distinguish well between normal and pancreatic cancer tissue samples (Supplementary Fig. S1), suggesting that telomere-related genes contributed to tumor heterogeneity and may have a key role in the tumorigenesis of pancreatic cancer. According to the results of Gene Set Enrichment Analysis (GSEA), these genes are mainly enriched in biological processes related to the cell cycle and telomere homeostasis (Fig. 1C). Overall, we obtained 502 telomere-associated DEGs for subsequent analysis.Figure 1 Identification of telomere-related DEGs and establishment of telomere-related subtypes (A) DEGs between the 171 normal pancreatic tissues and 178 pancreatic cancer tissues. (B) 502 overlapping genes were identified as telomere-related DEGs. (C) GSEA enrichment results of these telomere-related DEGs. (D) 72 of the 502 telomere-related DEGs were calculated as prognostic genes. (E) LASSO-Cox regression analysis. (F) The scores of patients and their distribution. (G) Patients with higher risk ratings exhibited noticeably lower survival outcomes. (H) ROC analysis.

Establishment of telomere-related subtypes

According to the univariate Cox regression analysis, 72 of the 502 telomere-related DEGs were calculated as prognostic genes (Fig. 1D). DEGs with increased expression in tumors are risk genes while DEGs with down-regulated expression in tumors are protective genes. These prognostic DEGs were explored through the LASSO-Cox regression and multivariate Cox proportional regression (Fig. 1E). Three genes (DSG2, LDHA, and RACGAP1) were finally screened for the construction of prognostic models. Score = (0.3045195 × DSG2) + (0.4734231 × RACGAP1) + (0.4349522 × LDHA). The scores of patients and their distribution were shown in Fig. 1F. We found that the score correlated with most telomere-related genes (Supplementary Table S1). Patients with higher risk ratings exhibited noticeably lower survival outcomes when the optimal risk score threshold was employed to classify patients into high- or low-risk groups (Fig. 1G). With AUCs at 1-, 2-, and 3-year of 0.718, 0.679, and 0.651, this classifier displayed high predictive performance (Fig. 1H). In addition, patient death (Fig. 2A), advanced grades (Fig. 2B), and relapses (Fig. 2C) were all linked to higher scores, while gender, age, and TNM stage were not associated with scores (Fig. 2D–F). Finally, univariable and multivariate Cox regression analysis showed that this signature may be employed as a standalone prognostic factor for individuals with pancreatic cancer (univariable HR = 2.718, 95%CI 1.822–4.056, P = 9.7e−07; multivariate HR = 2.445, 95%CI 1.459–4.097, P = 0.0006). Overall, we constructed a three-telomere gene-related model that can be used to accurately predict patient prognosis.Figure 2 Patient death (A), advanced grades (B), and relapses (C) were all linked to higher risk scores, while gender (D), age (E), and TNM stage (F) were not associated with risk scores.

Verification of the signature in multiple additional datasets

We calculated risk scores for patients in the ICGC, GSE62452, GSE71729, and GSE78229 datasets using the same formula and validated this signature in these patients. Patients with greater risk scores were significantly related to having lower OS rates, according to Kaplan–Meier survival analysis (Fig. 3A), and according to ROC analysis, this signature exhibited good prognostic performance (Fig. 3B). Overall, the model we constructed also accurately predicted the prognosis of patients in the validation set.Figure 3 Verification of the signature in multiple additional datasets. (A) Kaplan–Meier survival analysis. (B) ROC analysis.

Characterization of genomic variation and functional analysis

According to the GSEA results, the major enrichment of differential genes between high and low-scoring subgroups for HALLMARK entries was “HALLMARK ALLOGRAFT REJECTION”, “HALLMARK E2F TARGETS”, “HALLMARK G2M CHECKPOINT”, “HALLMARK GLYCOLYSIS”, and “HALLMARK MITOTIC SPINDLE”, for GO entries was “GOBP ACTIVATION OF IMMUNE RESPONSE”, “GOBP ADAPTIVE IMMUNE RESPONSE”, “GOBP ANTIGEN RECEPTOR MEDIATED SIGNALING PATHWAY”, and “GOBP ATTACHMENT OF SPINDLE MICROTUBULES TO KINETOCHORE”, and for KEGG entries was “KEGG CHEMOKINE SIGNALING PATHWAY”, “KEGG CYTOKINE CYTOKINE RECEPTOR INTERACTION”, “KEGG NEUROACTIVE LIGAND RECEPTOR INTERACTION”, “KEGG CELL CYCLE”, and “KEGG CELL ADHESION MOLECULES CAMS” (Supplementary Fig. S2). Overall, telomere scores may be involved in the prognostic regulation of pancreatic cancer patients by influencing the cell cycle of tumor cells, the body's immune response, and chemotherapy-related biological processes. After analyzing the tumor mutational burden (TMB) value of each pancreatic cancer patient (Fig. 4A), we discovered that the TMB value was greater in patients in the high-risk group and that the risk score was strongly positively linked with the TMB value (Fig. 4B), suggesting that immunotherapy might be less effective in patients with higher scores. Additionally, we discovered that patients with lower scores had higher levels of the genes CTLA4, CD4, CXCR4, LAG3, PD-L2, TNFRSF4, TNFRSF9, and PD1 than patients with higher scores (Fig. 4C), and scores were strongly correlated with the expression levels of CTLA4, CXCR4, TNFRSF4, and PD1 (Fig. 4D). By activating immune checkpoints and inhibiting the presentation of antigens to T cells, tumor cells can escape surveillance and survive by blocking the process of antigen presentation in the tumor immune microenvironment, thereby inhibiting the immune function of T cells. The immune checkpoint inhibitor relieves this inhibition and allows the immune cells to be reactivated to work against the cancer cells. Due to the high expression of immune checkpoint genes in low-scoring patients, these ICIs may be more beneficial for patients with lower scores.Figure 4 Characterization of ICIs analysis. (A) The tumor mutational burden (TMB) value of each pancreatic cancer patient. (B) The TMB value was greater in patients in the high-risk group and the risk score was strongly positively linked with the TMB value. (C) Patients with lower scores had higher levels of the genes CTLA4, CD4, CXCR4, LL1A, TNFRSF4, and PD1 than patients with higher scores. (D) Scores were strongly correlated with the expression levels of CTLA4, CD4, CXCR4, and LL1A.

Correlation analysis of the signature and TME

Figure 5A demonstrates that in comparison to patients with lower risk scores, patients with higher risk scores had lower stromal, immune, and estimate scores. Additionally, as seen in Fig. 5B, according to the TIMER database, patients with higher risk scores had lower abundance levels of CD4 T cells when compared to patients with lower risk scores. As seen in Fig. 5C, according to the CIBERSORT, patients with higher risk scores had lower abundance levels of naïve B cell, CD8 T cell, resting CD4 T cell, monocyte, and M2 macrophage when compared to patients with lower risk scores. As seen in Fig. 5D, according to the MCPcounter database, patients with higher risk scores had lower abundance levels of T cell, CD8 T cell, B cell, monocyte, endothelial cell, myeloid dendritic cell, and macrophage monocyte when compared to patients with lower risk scores. As seen in Fig. 5E, according to the xCELL database, patients with higher risk scores had lower abundance levels of B cell, naïve CD4 T cell, CD8 T cell, central memory CD8 T cell, class-switched memory B cell, common lymphoid progenitor, myeloid dendritic cell, endothelial cell, cancer-associated fibroblast, hematopoietic stem cell, macrophage, macrophage M1, macrophage M2, memory B cell, monocyte, NK T cell, Th1 CD4 T cell, and activated myeloid dendritic cell when compared with patients with lower risk scores. When the four algorithms described above are considered together, patients with lower risk scores possessed a higher number of immune cell infiltrates in TME, as shown in Supplementary Table S2, CD4 T cell was the common differential immune cell.Figure 5 Correlation analysis of the signature and TME. (A) In comparison to patients with lower scores, patients with higher scores had lower stromal, immune, and estimate scores. Immune cell infiltration analysis by the TIMER (B), CIBERSORT (C), MCPcounter (D), and xCELL (E).

Construction of a nomogram model

To test the efficacy of this signature's coefficient prediction, a nomogram model was created. The outcomes demonstrated that the nomogram with a concordance index (C-index) of 0.733 may assist in the creation of a precise quantitative technique for forecasting the 1-, 2-, and 3-year survival rates (Supplementary Fig. S3A). The projected and actual likelihood of 1-, 2-, and 3-year survival rates' overlap on the calibration curves revealed good agreement (Supplementary Fig. S3B). Overall, the nomogram can be used for clinician management of patients.

Potential therapeutic drug sensitivity analysis

We found 18 tumor-sensitive drugs (Supplementary Table S3), and the top 16 most significant tumor-sensitive drugs were shown in Fig. 6A. Among the 18 tumor-sensitive drugs, as seen in Fig. 6B, we also found that patients with higher scores had a lower IC50 for Actinomycin D, Vinblastine, Paclitaxel, Elliptinium Acetate, and Pipamperone, implying that patients with higher scores were more sensitive to these drugs.Figure 6 Potential therapeutic drug sensitivity analysis. (A) Top 16 most important tumor-sensitive drugs. (B) Differential analysis of the IC50 of tumor-sensitive drugs.

Expression levels of DSG2, LDHA, and RACGAP1

We first explored the expression levels of the three genes in the GEPIA database22. We found that all three genes were up-regulated in pancreatic cancer tissues compared to normal tissues (Supplementary Fig. S4A). We also found that high levels of DSG2, LDHA, and RACGAP1 predicted poorer overall (Supplementary Fig. S4B) and disease-free survival in pancreatic cancer patients (Supplementary Fig. S4C), although these three genes did not all show good performance in the four validation datasets (Supplementary Fig. S5A–D). We then explored the protein expression of DSG2, LDHA, and RACGAP1 in normal pancreatic and pancreatic cancer tissues in the human protein atlas database (HPA)23, and found that all of them showed differential expression (Supplementary Fig. S6). Finally, we examined the protein levels of DSG2, LDHA, and RACGAP1 in clinical samples using IHC techniques. According to Figs. 7A–C, pancreatic cancer tissues had higher protein levels of all three genes than normal tissues, which was in line with the bioinformatics study previously mentioned.Figure 7 Protein expression levels and percentage of positivity for the three genes in clinical samples. (A) DSG2. (B) LDHA. (C) RACGAP1.

Discussion

In the development of pancreatic cancer, telomeres are crucial24. Telomerase treatment was very useful for treating pancreatic cancer patients since telomere shortening and telomerase activation were common in human pancreatic cancer25–30. Because telomerase was repressed in the vast majority of human somatic cells, an immunotherapy that targeted tumor cell telomerase had no negative impact on normal cells. Furthermore, telomerase activity was only present in small subpopulations of normal human cells, such as stem/progenitor cells, activated lymphocytes, and other highly proliferative cells, but these cells had telomere lengths that were significantly longer than those of cancer cells, suggesting that tumor cells would respond more readily to telomerase inhibition therapy31. As a result, we may destroy telomeres in tumor cells before harm is done to telomerase-positive, healthy human cells. Notably, low residual levels of telomerase activity before telomerase function is eliminated may affect the efficacy of telomerase-targeted therapies5.

Desmoglein-2 (DSG2) is a cell adhesion protein belonging to the calmodulin superfamily that plays an important regulatory role in the functional development of cardiomyocytes32. DSG2 is also involved in the regulation of telomerase activity8. Although many studies have demonstrated that abnormal expression of DSG2 is associated with tumorigenesis33, its role in tumors is still controversial. The poor prognosis of patients with high expression of DSG2 in cutaneous squamous cell carcinoma and cervical cancer was correlated34–36. Downmodulation of DSG2 significantly inhibited the proliferation of colon cancer cells and the progression of non-small cell lung cancer37,38. In contrast, low expression of DSG2 was associated with shorter survival times in patients with gastric and prostate cancers39,40. In addition, reduced DSG2 expression also promotes lymph node metastasis in breast cancer41. Interestingly, although DSG2 was previously reported to be minimally expressed in pancreatic cancer42, in our study we found that DSG2 was highly expressed in pancreatic cancer tissues compared to normal pancreatic tissues and that high DSG2 expression was strongly associated with poor prognosis. This phenomenon may be caused by tumor heterogeneity in pancreatic cancer patients. Lactate dehydrogenase A (LDHA) is involved in the regulation of telomere signaling by PCA/BiFC (protein complementation assay/bimolecular fluorescence complementation) assay and co-precipitation experiments43. LDHA is also an important component of the glycolytic pathway and plays a key role in tumorigenesis44. Hypoxia-induced abnormal LDHA expression regulates the progression of gastric cancer45. Platinum complexes inhibit the metabolism and migration of triple-negative breast cancer cells by targeting LDHA46. Erythropoietin L-2HG can be produced by LDHA in hypoxic settings to maintain the proper balance of pancreatic cancer stem cell differentiation47. Reducing LDHA expression significantly increased the sensitivity of cancer cells to gemcitabine and inhibited tumor progression48. In our study, we found that LDHA can be used as a telomere-related gene to accurately predict the prognosis of pancreatic cancer patients. Rac GTPase-activating protein 1 (RACGAP1) is included in the TelNet database as a yeast homolog, but direct experimental validation is lacking, even though we have found RACGAP1 to be associated with telomere maintenance through GSEA (Supplementary Fig. S7). RACGAP1 is also a star molecule associated with poor tumor prognosis and is involved in the development and progression of many tumors. RACGAP1 can be involved in the regulation of breast cancer metastasis by targeting ECT249. By controlling CDC25C, RACGAP1 encourages the growth of cervical cancer cells50. RACGAP1 deletion has also been associated with mitotic mutations in esophageal squamous cell carcinoma51. When creating gene models for metastasis that forecast overall survival for patients with pancreatic ductal adenocarcinoma, RACGAP1 can be used as a predictor52. In our study, we found that RACGAP1 can be used as a telomere-related gene to accurately predict the prognosis of pancreatic cancer patients.

A growing body of research indicates that TME contributes significantly to the development and spread of pancreatic cancer. TME contains a range of cell types that can be implicated in tumor development, growth, and migration, including immune cells, stromal cells, endothelial cells, complicated cytokine secretion, and fibroblasts53. When the outcomes of the four algorithms TIMER, CIBERSORT, XCELL, and MCPcounter were taken into account collectively, CD4+ T cells emerged as the distinct immune cells between the high- and low-risk subgroups, indicating that the telomere-related signature may modulate CD4+ T cell infiltration to worsen the prognosis of pancreatic cancer. Excessive infiltration of cancer-associated fibroblasts (CAFs) in TME is associated with poor survival in pancreatic cancer patients and can also induce the infamous clinical treatment resistance54. In addition, CAFs expressing MHC class II in TME have been reported to present antigens to CD4+ T cells and modulate immune responses in pancreatic tumors55. Considering that complex TME can induce resistance of pancreatic cancer cells to immune checkpoint inhibitors (ICIs) and thus affect the efficacy of immunotherapy in patients56, we also analyzed the variation in expression levels of multiple immune checkpoint genes among high- and low-risk subgroups. We discovered that patients with lower scores had higher levels of the genes CTLA4, CD4, CXCR4, TNFRSF4, and PD1 than patients with higher scores, and scores were strongly correlated with the expression levels of CTLA4, CXCR4, TNFRSF4, and PD1, suggesting that these ICIs may be more beneficial for patients with lower scores. Interestingly, Nobuhiko has reported that telomerase-specific cleavage adenovirus can be combined with ICIs to treat tumors30, so ICIs combined with specific cleavage adenovirus constructed based on telomere-related genes for improving the prognosis of pancreatic cancer patients will have a good prospect in the future.

Finally, we also analyzed the tumor-sensitive drugs targeting this signature through the CellMiner database. We identified a total of 18 tumor-sensitive drugs targeting this signature, with the IC50 for Actinomycin D, Vinblastine, Paclitaxel, Elliptinium Acetate, and Pipamperone being lower in patients with higher scores, implying that patients with higher scores are more sensitive to these drugs. Actinomycin-D (ActD) is a peptide antibiotic that can limit RNA and protein synthesis by inhibiting RNA polymerase II and can be used in conjunction with dimethylamine monophenol lactone to treat human pancreatic cancer cells57. A variety of Vinblastine-based bioactive drugs can inhibit the proliferation of tumor cells and shine in the treatment of pancreatic cancer58,59. Paclitaxel can be combined with gemcitabine to treat patients with metastatic pancreatic cancer and improve their overall survival60–62. As for Elliptinium Acetate and Pipamperone, their use in the treatment of pancreatic cancer patients has not yet been reported and deserves further attention in the future.

Our manuscript still has some highlights, even though other comparable publications have been published employing built characteristics to predict the prognosis of pancreatic cancer patients. First, we examined the patient's prognosis for pancreatic cancer for the first time using telomere-related genes. Second, we successfully confirmed the signature using several publicly available datasets. Finally, we examined clinical tissue samples to confirm the protein levels of the genes responsible for the composition of the signature. Of course, there are certain restrictions on our study. To assess this signature, future large multicenter randomized controlled investigations are required. In the future, further in vivo and in vitro research is required to further investigate the expression, prognostic predictive relevance, and particular mechanisms of these three genes in pancreatic cancer.

Conclusion

We established and validated telomere gene-related prognostic features in pancreatic cancer and confirmed the upregulation of DSG2, LDHA, and RACGAP1 expression in clinical samples, which may provide new ideas for individualized immunotherapy targeting telomere genes.

Supplementary Information

Supplementary Information.

Abbreviations

DEGs Differentially expressed genes

TME Tumor microenvironment

IHC Immunohistochemical staining

DDR DNA damage response

TERT Telomerase reverse transcriptase

TMM Telomere maintenance mechanism

ALT Alternative-lengthening of Telomere

RNA-seq RNA sequencing

GSEA Gene set enrichment analysis

FDA Food and Drug Administration

IC50 Half maximum inhibitory concentration

ROC Receiver operating characteristic curves

TMB Tumor mutational burden

CCLE Cancer cell line encyclopedia

HPA Human protein atlas

DSG2 Desmoglein-2

LDHA Lactate dehydrogenase A

RACGAP1 Rac GTPase-activating protein 1

ICIs Immune checkpoint inhibitors

ActD Actinomycin-D

Supplementary Information

The online version contains supplementary material available at 10.1038/s41598-023-37836-0.

Author contributions

C.S.Y. designed the study and wrote the manuscript. H.S.Q., Z.B.Z., C.B.B., T.H., S.D.C., and L.X.Y. downloaded and analyzed the data. C.Y.J. and Z.G.H. critically reviewed the manuscript. The final manuscript has been approved by all authors.

Data availability

The datasets used and/or analyzed during the current study (including TCGA-PAAD, GTEx, ICGC-PACA-CN, GSE62452, GSE71729, and GSE78229 cohorts) are available from the corresponding author upon reasonable request.

Competing interests

The authors declare no competing interests.

Publisher's note

Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
==== Refs
References

1. Moyzis RK Buckingham JM Cram LS Dani M Deaven LL Jones MD Meyne J Ratliff RL Wu JR A highly conserved repetitive DNA sequence, (TTAGGG)n, present at the telomeres of human chromosomes Proc. Natl. Acad. Sci. U.S.A. 1988 85 18 6622 6626 10.1073/pnas.85.18.6622 3413114
2. Li SC Jia ZK Yang JJ Ning XH Telomere-related gene risk model for prognosis and drug treatment efficiency prediction in kidney cancer Front. Immunol. 2022 13 975057 10.3389/fimmu.2022.975057 36189312
3. d'Adda di Fagagna F Reaper PM Clay-Farrace L Fiegler H Carr P Von Zglinicki T Saretzki G Carter NP Jackson SP A DNA damage checkpoint response in telomere-initiated senescence Nature 2003 426 6963 194 198 10.1038/nature02118 14608368
4. De Vitis M Berardinelli F Sgura A Telomere length maintenance in cancer: At the crossroad between telomerase and alternative lengthening of telomeres (ALT) Int. J. Mol. Sci. 2018 19 2 606 10.3390/ijms19020606 29463031
5. Shay JW Wright WE Telomeres and telomerase: Three decades of progress Nat. Rev. Genet. 2019 20 5 299 309 10.1038/s41576-019-0099-1 30760854
6. Heaphy CM Subhawong AP Hong SM Goggins MG Montgomery EA Gabrielson E Netto GJ Epstein JI Lotan TL Westra WH Prevalence of the alternative lengthening of telomeres telomere maintenance mechanism in human cancer subtypes Am. J. Pathol. 2011 179 4 1608 1615 10.1016/j.ajpath.2011.06.018 21888887
7. Heaphy CM de Wilde RF Jiao Y Klein AP Edil BH Shi C Bettegowda C Rodriguez FJ Eberhart CG Hebbar S Altered telomeres in tumors with ATRX and DAXX mutations Science (New York, NY) 2011 333 6041 425 10.1126/science.1207313
8. Barthel FP Wei W Tang M Martinez-Ledesma E Hu X Amin SB Akdemir KC Seth S Song X Wang Q Systematic analysis of telomere length and somatic alterations in 31 cancer types Nat. Genet. 2017 49 3 349 357 10.1038/ng.3781 28135248
9. Kan G Wang Z Sheng C Yao C Mao Y Chen S Inhibition of DKC1 induces telomere-related senescence and apoptosis in lung adenocarcinoma J. Transl. Med. 2021 19 1 161 10.1186/s12967-021-02827-0 33879171
10. Campa D Matarazzi M Greenhalf W Bijlsma M Saum KU Pasquali C van Laarhoven H Szentesi A Federici F Vodicka P Genetic determinants of telomere length and risk of pancreatic cancer: A PANDoRA study Int. J. Cancer 2019 144 6 1275 1283 10.1002/ijc.31928 30325019
11. Braun DM Chung I Kepper N Deeg KI Rippe K TelNet—A database for human and yeast genes involved in telomere maintenance BMC Genet. 2018 19 1 32 10.1186/s12863-018-0617-8 29776332
12. Zhang G Platelet-related molecular subtype to predict prognosis in hepatocellular carcinoma J. Hepatocell. Carcinoma 2022 9 423 436 10.2147/JHC.S363200 35615530
13. Zhou Y Zhou B Pache L Chang M Khodabakhshi AH Tanaseichuk O Benner C Chanda SK Metascape provides a biologist-oriented resource for the analysis of systems-level datasets Nat. Commun. 2019 10 1 1523 10.1038/s41467-019-09234-6 30944313
14. Kanehisa M Furumichi M Sato Y Ishiguro-Watanabe M Tanabe M KEGG: Integrating viruses and cellular organisms Nucleic Acids Res. 2021 49 D1 D545 d551 10.1093/nar/gkaa970 33125081
15. Yoshihara K Shahmoradgoli M Martínez E Vegesna R Kim H Torres-Garcia W Treviño V Shen H Laird PW Levine DA Inferring tumour purity and stromal and immune cell admixture from expression data Nat. Commun. 2013 4 2612 10.1038/ncomms3612 24113773
16. Gentles AJ Newman AM Liu CL Bratman SV Feng W Kim D Nair VS Xu Y Khuong A Hoang CD The prognostic landscape of genes and infiltrating immune cells across human cancers Nat. Med. 2015 21 8 938 945 10.1038/nm.3909 26193342
17. Aran D Hu Z Butte AJ xCell: Digitally portraying the tissue cellular heterogeneity landscape Genome Biol. 2017 18 1 220 10.1186/s13059-017-1349-1 29141660
18. Zhu K Xiaoqiang L Deng W Wang G Fu B Development and validation of a novel lipid metabolism-related gene prognostic signature and candidate drugs for patients with bladder cancer Lipids Health Dis. 2021 20 1 146 10.1186/s12944-021-01554-1 34706720
19. Li T Fan J Wang B Traugh N Chen Q Liu JS Li B Liu XS TIMER: A web server for comprehensive analysis of tumor-infiltrating immune cells Cancer Res. 2017 77 21 e108 e110 10.1158/0008-5472.CAN-17-0307 29092952
20. Reinhold WC Sunshine M Liu H Varma S Kohn KW Morris J Doroshow J Pommier Y Cell miner: A web-based suite of genomic and pharmacologic tools to explore transcript and drug patterns in the NCI-60 cell line set Cancer Res. 2012 72 14 3499 3511 10.1158/0008-5472.CAN-12-1370 22802077
21. Zhang G Regulatory T-cells-related signature for identifying a prognostic subtype of hepatocellular carcinoma with an exhausted tumor microenvironment Front. Immunol. 2022 13 975762 10.3389/fimmu.2022.975762 36189226
22. Tang Z Li C Kang B Gao G Li C Zhang Z GEPIA: A web server for cancer and normal gene expression profiling and interactive analyses Nucleic Acids Res. 2017 45 W1 W98 w102 10.1093/nar/gkx247 28407145
23. Colwill K Gräslund S A roadmap to generate renewable protein binders to the human proteome Nat. Methods 2011 8 7 551 558 10.1038/nmeth.1607 21572409
24. Hamada T Yuan C Bao Y Zhang M Khalaf N Babic A Morales-Oyarvide V Cochrane BB Gaziano JM Giovannucci EL Prediagnostic leukocyte telomere length and pancreatic cancer survival Cancer Epidemiol. Biomark. Prev. 2019 28 11 1868 1875 10.1158/1055-9965.EPI-19-0577
25. Wack S Rejiba S Parmentier C Aprahamian M Hajri A Telomerase transcriptional targeting of inducible Bax/TRAIL gene therapy improves gemcitabine treatment of pancreatic cancer Mol. Ther. J. Am. Soc. Gene Ther. 2008 16 2 252 260 10.1038/sj.mt.6300340
26. Kumari A Srinivasan R Wig JD Effect of c-MYC and E2F1 gene silencing and of 5-azacytidine treatment on telomerase activity in pancreatic cancer-derived cell lines Pancreatol. Off. J. Int. Assoc. Pancreatol. IAP 2009 9 4 360 368
27. Walter K Rodriguez-Aznar E Ferreira MSV Frappart PO Dittrich T Tiwary K Meessen S Lerma L Daiss N Schulte LA Telomerase and pluripotency factors jointly regulate stemness in pancreatic cancer stem cells Cancers 2021 13 13 3145 10.3390/cancers13133145 34201898
28. Zisuh AV Han TQ Zhan SD Expression of telomerase and its significance in the diagnosis of pancreatic cancer Indian J. Med. Res. 2012 135 1 26 30 10.4103/0971-5916.93420 22382179
29. Joseph I Tressler R Bassett E Harley C Buseman CM Pattamatta P Wright WE Shay JW Go NF The telomerase inhibitor imetelstat depletes cancer stem cells in breast and pancreatic cancer cell lines Cancer Res. 2010 70 22 9494 9504 10.1158/0008-5472.CAN-10-0233 21062983
30. Kanaya N Kuroda S Kakiuchi Y Kumon K Tsumura T Hashimoto M Morihiro T Kubota T Aoyama K Kikuchi S Immune modulation by telomerase-specific oncolytic adenovirus synergistically enhances antitumor efficacy with anti-PD1 antibody Mol. Ther. J. Am. Soc. Gene Ther. 2020 28 3 794 804 10.1016/j.ymthe.2020.01.003
31. Yuan X Dai M Xu D Telomere-related markers for cancer Curr. Top. Med. Chem. 2020 20 6 410 432 10.2174/1568026620666200106145340 31903880
32. Hermida A Fressart V Hidden-Lucet F Donal E Probst V Deharo JC Chevalier P Klug D Mansencal N Delacretaz E High risk of heart failure associated with desmoglein-2 mutations compared to plakophilin-2 mutations in arrhythmogenic right ventricular cardiomyopathy/dysplasia Eur. J. Heart Fail. 2019 21 6 792 800 10.1002/ejhf.1423 30790397
33. Zhou G Yang L Gray A Srivastava AK Li C Zhang G Cui T The role of desmosomes in carcinogenesis OncoTargets Ther. 2017 10 4059 4063 10.2147/OTT.S136367
34. Qin S Liao Y Du Q Wang W Huang J Liu P Shang C Liu T Xia M Yao S DSG2 expression is correlated with poor prognosis and promotes early-stage cervical cancer Cancer Cell Int. 2020 20 206 10.1186/s12935-020-01292-x 32514251
35. Kurzen H Münzing I Hartschuh W Expression of desmosomal proteins in squamous cell carcinomas of the skin J. Cutan. Pathol. 2003 30 10 621 630 10.1034/j.1600-0560.2003.00122.x 14744087
36. Saaber F Chen Y Cui T Yang L Mireskandari M Petersen I Expression of desmogleins 1–3 and their clinical impacts on human lung cancer Pathol. Res. Pract. 2015 211 3 208 213 10.1016/j.prp.2014.10.008 25468811
37. Kamekura R Kolegraff KN Nava P Hilgarth RS Feng M Parkos CA Nusrat A Loss of the desmosomal cadherin desmoglein-2 suppresses colon cancer cell proliferation through EGFR signaling Oncogene 2014 33 36 4531 4536 10.1038/onc.2013.442 24166502
38. Cai F Zhu Q Miao Y Shen S Su X Shi Y Desmoglein-2 is overexpressed in non-small cell lung cancer tissues and its knockdown suppresses NSCLC growth by regulation of p27 and CDK2 J. Cancer Res. Clin. Oncol. 2017 143 1 59 69 10.1007/s00432-016-2250-0 27629878
39. Yashiro M Nishioka N Hirakawa K Decreased expression of the adhesion molecule desmoglein-2 is associated with diffuse-type gastric carcinoma Eur. J. Cancer (Oxford, England: 1990) 2006 42 14 2397 2403 10.1016/j.ejca.2006.03.024
40. Barber AG Castillo-Martin M Bonal DM Rybicki BA Christiano AM Cordon-Cardo C Characterization of desmoglein expression in the normal prostatic gland. Desmoglein 2 is an independent prognostic factor for aggressive prostate cancer PLoS ONE 2014 9 6 e98786 10.1371/journal.pone.0098786 24896103
41. Davies E Cochrane R Hiscox S Jiang W Sweetland H Mansel R The role of desmoglein 2 and E-cadherin in the invasion and motility of human breast cancer cells Int. J. Oncol. 1997 11 2 415 419 21528229
42. Ramani VC Hennings L Haun RS Desmoglein 2 is a substrate of kallikrein 7 in pancreatic cancer BMC Cancer 2008 8 373 10.1186/1471-2407-8-373 19091121
43. Lee OH Kim H He Q Baek HJ Yang D Chen LY Liang J Chae HK Safari A Liu D Genome-wide YFP fluorescence complementation screen identifies new regulators for telomere signaling in human cells Mol. Cell. Proteom. MCP 2011 10 2 S1 S11 10.1074/mcp.M110.001628
44. Huo N Cong R Sun ZJ Li WC Zhu X Xue CY Chen Z Ma LY Chu Z Han YC STAT3/LINC00671 axis regulates papillary thyroid tumor growth and metastasis via LDHA-mediated glycolysis Cell Death Dis. 2021 12 9 799 10.1038/s41419-021-04081-0 34404767
45. Wang XH Jiang ZH Yang HM Zhang Y Xu LH Hypoxia-induced FOXO4/LDHA axis modulates gastric cancer cell glycolysis and progression Clin. Transl. Med. 2021 11 1 e279 10.1002/ctm2.279 33463054
46. Bai X Ali A Lv Z Wang N Zhao X Hao H Zhang Y Rahman FU Platinum complexes inhibit HER-2 enriched and triple-negative breast cancer cells metabolism to suppress growth, stemness and migration by targeting PKM/LDHA and CCND1/BCL2/ATG3 signaling pathways Eur. J. Med. Chem. 2021 224 113689 10.1016/j.ejmech.2021.113689 34293698
47. Gupta VK Sharma NS Durden B Garrido VT Kesh K Edwards D Wang D Myer C Mateo-Victoriano B Kollala SS Hypoxia-driven oncometabolite L-2HG maintains stemness-differentiation balance and facilitates immune evasion in pancreatic cancer Cancer Res. 2021 81 15 4001 4013 10.1158/0008-5472.CAN-20-2562 33990397
48. Dai S Peng Y Zhu Y Xu D Zhu F Xu W Chen Q Zhu X Liu T Hou C Glycolysis promotes the progression of pancreatic cancer and reduces cancer cell sensitivity to gemcitabine Biomed. Pharmacother. 2020 121 109521 10.1016/j.biopha.2019.109521 31689601
49. Ren K Zhou D Wang M Li E Hou C Su Y Zou Q Zhou P Liu X RACGAP1 modulates ECT2-Dependent mitochondrial quality control to drive breast cancer metastasis Exp. Cell Res. 2021 400 1 112493 10.1016/j.yexcr.2021.112493 33485843
50. Ruan X Jiang J RACGAP1 promotes proliferation and cell cycle progression by regulating CDC25C in cervical cancer cells Tissue Cell 2022 76 101804 10.1016/j.tice.2022.101804 35489195
51. Zhao W Wang M Wang C Liu Y Liu H Luo S RACGAP1 is transcriptionally regulated by E2F3, and its depletion leads to mitotic catastrophe in esophageal squamous cell carcinoma Ann. Transl. Med. 2020 8 15 950 10.21037/atm-20-2901 32953750
52. Wu M Li X Liu R Yuan H Liu W Liu Z Development and validation of a metastasis-related gene signature for predicting the overall survival in patients with pancreatic ductal adenocarcinoma J. Cancer 2020 11 21 6299 6318 10.7150/jca.47629 33033514
53. Hinshaw DC Shevde LA The tumor microenvironment innately modulates cancer progression Cancer Res. 2019 79 18 4557 4566 10.1158/0008-5472.CAN-18-3962 31350295
54. Koikawa K Kibe S Suizu F Sekino N Kim N Manz TD Pinch BJ Akshinthala D Verma A Gaglia G Targeting Pin1 renders pancreatic cancer eradicable by synergizing with immunochemotherapy Cell 2021 184 18 4753 4771.e4727 10.1016/j.cell.2021.07.020 34388391
55. Elyada E Bolisetty M Laise P Flynn WF Courtois ET Burkhart RA Teinor JA Belleau P Biffi G Lucito MS Cross-species single-cell analysis of pancreatic ductal adenocarcinoma reveals antigen-presenting cancer-associated fibroblasts Cancer Discov. 2019 9 8 1102 1123 10.1158/2159-8290.CD-19-0094 31197017
56. Carbone C Piro G Agostini A Delfino P De Sanctis F Nasca V Spallotta F Sette C Martini M Ugel S Intratumoral injection of TLR9 agonist promotes an immunopermissive microenvironment transition and causes cooperative antitumor activity in combination with anti-PD1 in pancreatic cancer J. Immunother. Cancer 2021 9 9 e002876 10.1136/jitc-2021-002876 34479922
57. Lamture G Crooks PA Borrelli MJ Actinomycin-D and dimethylamino-parthenolide synergism in treating human pancreatic cancer cells Drug Dev. Res. 2018 79 6 287 294 10.1002/ddr.21441 30295945
58. Shah VM Dorrell C Al-Fatease A Allen-Petersen BL Woo Y Bortnyak Y Gheewala R Sheppard BC Sears RC Alani AW Microfluidics formulated liposomes of hypoxia activated prodrug for treatment of pancreatic cancer Pharmaceutics 2022 14 4 713 10.3390/pharmaceutics14040713 35456547
59. Albukhaty S Al-Musawi S Abdul Mahdi S Sulaiman GM Alwahibi MS Dewir YH Soliman DA Rizwana H Investigation of dextran-coated superparamagnetic nanoparticles for targeted vinblastine controlled release, delivery, apoptosis induction, and gene expression in pancreatic cancer cells Molecules (Basel, Switzerland) 2020 25 20 4721 10.3390/molecules25204721 33076247
60. Von Hoff DD Ervin T Arena FP Chiorean EG Infante J Moore M Seay T Tjulandin SA Ma WW Saleh MN Increased survival in pancreatic cancer with nab-paclitaxel plus gemcitabine N. Engl. J. Med. 2013 369 18 1691 1703 10.1056/NEJMoa1304369 24131140
61. Kunzmann V Siveke JT Algül H Goekkurt E Siegler G Martens U Waldschmidt D Pelzer U Fuchs M Kullmann F Nab-paclitaxel plus gemcitabine versus nab-paclitaxel plus gemcitabine followed by FOLFIRINOX induction chemotherapy in locally advanced pancreatic cancer (NEOLAP-AIO-PAK-0113): A multicentre, randomised, phase 2 trial Lancet Gastroenterol. Hepatol. 2021 6 2 128 138 10.1016/S2468-1253(20)30330-7 33338442
62. Goldstein D El-Maraghi RH Hammel P Heinemann V Kunzmann V Sastre J Scheithauer W Siena S Tabernero J Teixeira L Nab-paclitaxel plus gemcitabine for metastatic pancreatic cancer: Long-term survival from a phase III trial J. Natl. Cancer Inst. 2015 107 2 1 10 10.1093/jnci/dju413