Establishment of a prognostic signature for pancreatic cancer using immune checkpoint-related lncRNAs

doi:10.21203/rs.3.rs-4057958/v1

Objective: This study aimed to develop a prognostic signature for pancreatic cancer (PC) using immune checkpoint-related long non-coding RNAs (lncRNAs) and evaluate their clinical significance in PC patients.

Methods: Transcriptome data and clinical information of PC patients were obtained from the Cancer Genome Atlas. Immune checkpoint-related genes were analyzed to identify immune checkpoint-related lncRNAs. Univariate Cox analysis, Lasso analysis, and multivariate Cox analysis were performed to select relevant lncRNAs and construct a prognostic signature. Independent prognostic analyses were conducted using univariate and multivariate Cox analyses. The performance of the prognostic signature was assessed using receiver operating characteristic curves, C-index, survival curves, and nomogram. Additionally, gene enrichment analyses, immune-related function analyses, tumor mutation burden analysis, and tumor immune dysfunction and exclusion analysis were conducted.

Results: A total of 83 immune checkpoint-related lncRNAs were identified. A prognostic signature consisting of two checkpoint-related lncRNAs (LINC02245 and AL008729.2) was developed. The prognostic signature exhibited reliable predictive capability, with area under the curves of 0.660, 0.666, and 0.692 at 1, 3, and 5 years, respectively. The prognostic signature was identified as an independent prognostic factor. Significant differences in immune cell populations, immune function, and tumor mutation burden were observed between the high- and low-risk groups.

Conclusion: This study successfully established a prognostic signature for PC based on two checkpoint-related lncRNAs. The signature demonstrated strong predictive capability for PC prognosis and served as an independent prognostic factor. These findings enhance our understanding of PC and may have implications for personalized therapeutic strategies.

pancreatic cancer

lncRNA

immune checkpoint

prognosis

bioinformatics analysis

Pancreatic cancer (PC) is one of the leading malignancies globally in terms of both incidence and mortality rates. According to the latest statistical data, there are hundreds of thousands of new cases of PC reported globally each year. The mortality rate associated with PC is also alarmingly high, causing the loss of hundreds of thousands of lives annually[1]. In fact, PC ranks among the highest in terms of fatality rates compared to other types of cancers. The high mortality rate of PC can be attributed to several factors, including the difficulty of early diagnosis, its aggressive growth, and the high tendency for metastasis. Predicting the prognosis of PC patients poses significant challenges due to the complex etiological factors and high heterogeneity observed in this disease[2]. Additionally, the limited treatment options currently available for PC underscore the importance of developing new prognostic models to improve patient outcomes.[3]

Immune checkpoints play a crucial role in regulating the immune system by balancing stimulatory and inhibitory pathways. They are essential for maintaining self-tolerance and controlling the type, strength, and duration of immune responses. However, tumors can exploit these checkpoints to evade immune recognition and suppress immune responses[4]. To counteract this, immune checkpoint blockade therapies have emerged as a promising approach for cancer treatment. These therapies involve the use of antibodies that target specific molecules involved in immune checkpoint pathways[5]. Among the most extensively studied molecules are CTLA-4, PD-1, and PD-L1. Therapies targeting CTLA-4 and the PD-1/PD-L1 axis have shown significant clinical benefits and have been approved by the US Food and Drug Administration for the treatment of various cancers[6]. Immune checkpoint blockade therapies have revolutionized cancer immunotherapy and hold great potential for improving patient outcomes. They represent a major breakthrough in the field and continue to be actively researched and developed for the treatment of different types of cancer[7].

Long non-coding RNAs (lncRNAs) are a class of RNA molecules that are longer than 200 nucleotides and do not encode proteins. Unlike traditional non-coding RNAs such as rRNA and tRNA, lncRNAs do not undergo translation into proteins. However, they play important regulatory roles within cells and are involved in various biological processes, including cell differentiation, immune response, chromosomal imbalances, and disease pathogenesis[8]. In recent years, there has been increasing recognition of the functional significance of lncRNAs in cellular processes and disease mechanisms. Despite not encoding proteins, lncRNAs exert their influence through various mechanisms. They can act as molecular scaffolds, decoys, or guides for chromatin modifiers, transcription factors, and other regulatory molecules. By interacting with proteins, DNA, and other RNAs, lncRNAs can modulate gene expression and contribute to the complex regulatory networks that govern cellular functions[9; 10]. Although numerous studies have reported the impact of lncRNAs on the development of PC, there have been no studies to date investigating the use of immune checkpoint-related lncRNAs as molecular markers for predicting the prognosis of PC patients[11]. Therefore, the association between immune checkpoint-related lncRNAs and the prognosis of PC patients remains unclear. This highlights the critical need to explore these lncRNAs as potential molecular markers for prognostic prediction in PC.

To address this issue, we initially downloaded transcriptomic data and relevant clinical data of PC from public databases. We then identified lncRNA data associated with immune checkpoint-related genes. Using univariate Cox regression analysis and multivariate Cox regression analysis, we constructed a prognostic multi-lncRNA risk signature. This risk signature has the potential to provide valuable prognostic assessment and individualized treatment options for PC patients.

2.1 Preparation of transcriptomic data, mutation data, and clinically relevant information

Transcriptome data, mutation data and clinical data of PC patients were downloaded from The Cancer Genome Atlas (TCGA) database (https://portal.gdc.cancer.gov, accessed on 15 May 2023). The transcriptome data was extracted using Perl language (version Strawberry-perl-5.32.1.1) in TPM format and has been normalized. The clinical data includes age, grade, stage, T stage, M stage, N stage, survival status, and survival time. In accordance with previous literature, we retrieved the immune checkpoint-related genes (BTN2A2, BTNL3, BTNL9, CEACAM1, IDO1, TDO2, and others)[12]. Finally, the clinical and sample information of 173 patients were randomly divided into training and validation sets in a 1:1 ratio. Table 1 provides an overview of the basic information for the training and validation sets. The downloaded transcriptome data was classified and separated into lncRNA and mRNA using Perl software. Since the clinical information of the patients involved in this study is obtained from the TCGA database and strictly adheres to the TCGA publication guidelines, ethical committee approval is not required.

2.2 Identification of immune checkpoint‑related lncRNAs

The association coefficient between lncRNAs and immune checkpoint-related genes was calculated using the “limma” package in R software. The criteria for identifying immune checkpoint-related lncRNAs were set as |cor|>0.6 and p<0.001. To visualize the relationship between immune checkpoint-related genes and immune checkpoint-related lncRNAs, a Sankey diagram was created using the ggplot2 and ggalluvial packages in R software.

2.3 Establishment of the immune checkpoint‑related lncRNAs prognostic signature

The prognostic signature of immune checkpoint-related lncRNAs was developed using the training set, and its validation was performed on both the validation set and the entire dataset. Initially, prognostic immune checkpoint-related lncRNAs were identified using univariate Cox regression analysis, and the results were visualized using a forest plot. To address data overfitting, we employed Least Absolute Shrinkage and Selection Operator (Lasso) regression analysis to select a total of 4 immune checkpoint-related lncRNAs. Subsequently, the selected lncRNAs from the Lasso screening process were incorporated into a multivariate Cox regression model to identify the essential lncRNAs required for constructing the signature. A risk score was then calculated using the following formula: risk score = (Expi × bi), where Expi represents the expression level of the respective lncRNA and bi corresponds to the coefficient of the lncRNA in the model. To perform these analyses, several R packages, including survival, caret, glmnet, survminer, and timeROC, were utilized.

2.4 Assessment of the immune checkpoint-related lncRNAs prognostic signature

The participants in the study were divided into high-risk and low-risk groups based on the median risk score calculated from the training set. Survival analysis was performed to compare overall survival (OS) among the training set, validation set, and the entire dataset, as well as progression-free survival (PFS) for all participants. The survival, survminer, and pheatmap packages in R were used for the survival analysis. Furthermore, univariate and multivariate Cox regression analyses were conducted to assess whether the risk score independently served as a prognostic factor, regardless of other clinical indicators. The risk scores and other clinical features were included in these analyses. To evaluate the predictive power of the risk score, receiver operating characteristic (ROC) curves were plotted at 1, 3, and 5 years using the survival, survminer, and timeROC packages in R. Additionally, the R packages survival, rms, and pec were utilized to generate the C-index curve, which provides an evaluation of the accuracy of the risk score.

2.5 Nomogram

To integrate the risk score with the clinicopathological traits of PC patients, we created a nomogram using the rms package. The nomogram was used to predict the 1-year, 3-year, and 5-year OS rates of PC patients. Calibration curves were plotted for each time point to assess the accuracy of the predictions by comparing them with the actual observed results. Additionally, we evaluated the suitability of the risk signature across different subgroups by categorizing patients based on different clinical variables. This analysis aimed to determine the robustness and generalizability of the risk signature in diverse clinical scenarios.

2.6 Enrichment function analysis

To identify differentially expressed genes (DEGs) in different risk groups, we used the limma package in R with the criteria of |logFC|>1 and adjusted p<0.05. The resulting DEGs were then subjected to enrichment analysis using various R software packages, including clusterProfiler, org.Hs.eg.db, enrichplot, ggplot2, circlize, RColorBrewer, dplyr, ggpubr, and ComplexHeatmap. Gene Ontology (GO) enrichment analysis classified genes based on their involvement in biological processes (BP), molecular functions (MF), and cellular components (CC)[13]. Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway analysis was performed to reveal the pathways associated with high- and low-risk groups[14]. Furthermore, gene set enrichment analysis (GSEA) was conducted using the org.Hs.eg.db, limma, clusterProfiler, and enrichplot packages to further identify enriched pathways in different risk groups[15].

2.7 Analysis tumor immune microenvironment (TME) in different risk groups

We investigated the correlation between the risk score and the TME by calculating immune, stromal, and ESTIMATE scores[16]. Additionally, we assessed immune cell infiltration in tumor samples using the CIBERSORT method to determine the relative abundances of immune cells. This analysis allowed us to explore the relationship between the risk score and the immune status of the tumor[17]. Furthermore, we analyzed immune-related functions between the high-risk and low-risk groups using various R packages, including estimate, limma, reshape2, ggpubr, GSVA, and GSEABase. These analyses provided insights into the differences in immune-related functions between the two risk groups.

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

For the analysis of TMB, we processed the somatic mutation data and visualized the results using the ‘maftools’ package. Additionally, we assessed the clinical relevance of TMB in PC patients by generating Kaplan-Meier curves to illustrate its impact on survival time. To predict tumor immune escape and its potential influence on patients’ response to immunotherapy, we constructed a T cell dysfunction and exclusion signature using the tumor immune dysfunction and exclusion (TIDE) algorithm (http://tide.dfci.harvard.edu, accessed on 25 May 2023)[18]. We then used the limma and ggpubr packages to explore the differences in potential clinical response to immune checkpoint inhibitors among different risk groups.

2.9 Statistical analysis

All data were analyzed using R software (version 4.3.0). Statistical significance was assessed using the following thresholds: *p < 0.05, **p < 0.01, and ***p < 0.001. The Wilcoxon test was employed for co-expression analysis and differential analysis. The Spearman correlation was used to evaluate associations between numerical variables. Survival curves were generated using the Kaplan-Meier method, and statistical differences were determined using the log-rank test.

3.1 Immune checkpoint-related lncRNAs screening in PC patients

The study followed a specific flow chart, as depicted in Figure 1. Initially, the TCGA-PAAD database were utilized and a total of 16,876 lncRNAs were identified. A total of 83 immune checkpoint-related lncRNAs were screened when setting the criteria of p<0.001 and |cor|>0.6. The Sankey diagram of 79 immune checkpoint-related genes and immune checkpoint-related lncRNAs was presented in Figure 2.

3.2 Construction of immune checkpoint-related lncRNAs prognostic signature

To investigate the prognostic significance of these lncRNAs in PC, univariate Cox regression analysis was performed and identified 8 lncRNAs associated with PC patient prognosis in the training set. Forest plots were generated to illustrate these associations (Figure 3A). Subsequently, the initially identified lncRNAs underwent further screening using Lasso regression to reduce overfitting. The Lasso regression coefficient profiles were plotted (Figure 3B and C).

To construct a prognostic signature and evaluate the contribution of each lncRNA as prognostic factors in OS of PC patients, multivariate Cox regression analysis was performed. Finally, a prognostic signature for PC was developed using 2 immune checkpoint-related lncRNAs, namely LINC02245 and AL008729.2. The risk score formula for this signature was determined as follows: risk score = (-2.37949091418402) × LINC02245 + (-0.313015834836748) × AL008729.2. Additionally, Figure 3D illustrates the relationships between the 2 immune checkpoint-related lncRNAs and the immune checkpoint-related genes.

3.3 Predicting the prognosis of PC patients with the prognostic signature

The survival analysis revealed that patients in the low-risk group had significantly better OS compared to those in the high-risk group (p< 0.001; Figure 4D). The risk score was found to be positively associated with mortality in PC patients, as demonstrated by the increasing mortality rates with higher risk scores (Figure 4A and B). This trend was consistent when analyzing the validation set and all patients (Figure 4E, F, H, I, J, and L). In the low-risk group, LINC02245 and AL008729.2 exhibited higher expression levels compared to the high-risk group across the training, testing, and whole sets (Figure 4C, G, and K), indicating that these immunecheckpoint-related lncRNAs may serve as good prognostic predictors. Furthermore, PFS was significantly better in the low-risk group compared to the high-risk group among all patients (p<0.001; Figure 4M). These results suggest that the prognostic signature developed in the study can effectively stratify patients into different risk groups.

Additionally, the survival probability and clinical features of PC patients were compared various clinical variables, including age, gender, and TNM stage. Apart from high-stage patients, high-risk patients consistently exhibited shorter OS compared to low-risk patients (Figure 5). One possible reason is that there were too few patients with advanced stage PC included in the analysis, leading to a lack of credibility in the obtained results.

3.4 The risk score could be a robust prognostic factor to predict clinical outcomes for PC patients

The risk score demonstrated its independent prognostic value for PC patients, as determined through both univariate and multivariate Cox regression analyses (Figure 5A and B). The receiver operating characteristic (ROC) analysis revealed the area under the curve (AUC) to be 0.660, 0.666 and 0.692 for 1, 3, and 5 years, respectively, indicating its predictive accuracy over time (Figure 5C). Furthermore, ROC curves were generated to compare the predictive performance of the risk score with various clinical characteristics. It was observed that the AUC of the risk score surpassed those of TNM staging, age, and gender, implying a relatively higher accuracy in prediction (Figure 5D). Additionally, the C-index values of the risk score were higher than those associated with other clinical characteristics, including age, gender, and disease stage (Figure 5E). Overall, these findings indicate that the risk score possesses independent prognostic significance for PC patients and exhibits a more accurate predictive power compared to other clinical characteristics evaluated in the study.

3.5 Nomogram

To enhance the precision of predicting patient survival rates, we devised a nomogram and employed it to construct a graphical representation (Figure 6A). Furthermore, we assessed the dependability of our predictive model by generating a calibration plot, which gauged the concordance between the predicted probabilities and the actual probabilities. The calibration plot demonstrated that the predicted probabilities obtained from the nomogram were in good agreement with the observed probabilities, thereby reinforcing the credibility of our nomogram (Figure 6B).

3.6 Functional enrichment analysis

GO analyses revealed that in terms of BPs, DEGs were significantly associated with " production of molecular mediator of immune response", "immune response-regulating signaling pathway" and "immunoglobulin production". Regarding CCs, enrichment was observed in "external side of plasma membrane", "plasma membrane signaling receptor complex" and "T cell receptor complex". Moreover, DEGs exhibited MFs related to "antigen binding", "receptor ligand activity", and "peptide binding" (Figure 7A and B). In addition, KEGG analysis demonstrated that DEGs were enriched in pathways such as "cytokine-cytokine receptor interaction", "neuroactive ligand-receptor interaction " and "chemokine signaling pathway" (Figure 7C). Specifically, GSEA exhibited that the "chemokine signaling pathway" and "cytokine cytokine receptor interaction" were activated in the low-risk group (Figure 7D). Conversely, in high-risk patients, significant enrichment was not observed.

3.7 Estimate the difference of immune microenvironment landscape between the high- and low-risk groups

In Figure 9A, it was found that the low-risk group had significantly higher stromal scores, immune scores, and ESTIMATE scores compared to the high-risk group. This indicates that low-risk patients have a higher abundance of stromal, immune activity, and immune cells in their tumor microenvironment. Apart from the type II IFN response and MHC class I, the levels of the other 27 immune signature gene sets were generally lower in the high-risk group compared to the low-risk group (Figure 9B). To further investigate the landscape of immune cell infiltration, the researchers employed the CIBERSORT algorithm. This algorithm provided information about the relative percentage of different immune cell types in the tumor samples of all PC patients. The low-risk group exhibited a significant increase in B cells naive compared to the high-risk group, while a significant decrease in Macrophages M0 was observed in the low-risk group compared to the high-risk group (Figure 9C). In addition, the relative percentage of each immune cell type in tumor samples of all patients with different risk groups was showed (Figure 9D).

3.8 Mutational landscape for PC and TIDE analysis

To analyze the changes in somatic mutations between the high- and low-risk groups, the researchers obtained somatic mutation data from the TCGA database. They identified the 15 most highly mutated genes in PC, which were KRAS, TP53, SMAD4, CDKN2A, TTN, MUC16, RNF43, TNXB, RYR1, TGFBR2, HECW2, ARID1A, CACNA1B, RIMS2, and GLI3 (Figure 9A and B). The analysis revealed that mutations in KRAS, TP53, SMAD4, CDKN2A, and GLI3 were evidently more common in the high-risk group compared to the low-risk group (Figure 10A and B). Additionally, the high-risk group exhibited a higher TMB than the low-risk group (Figure 10C). TMB refers to the number of mutations present in the tumor genome and is considered an indicator of genomic instability and potential response to immunotherapy. To assess the potential response to cancer immunotherapy, the researchers used the TIDE tool, which predicts the likelihood of response to immune checkpoint blockade therapy. Interestingly, when comparing the TIDE scores between the high- and low-risk groups, no significant difference in TIDE was observed (Figure 10D). In summary, high-risk patients may be more sensitive to immune checkpoint blockade therapy than low-risk patients, despite similar TIDE levels.

Furthermore, survival analysis demonstrated that patients with PC who had high TMB experienced significantly shorter survival times compared to those with low TMB, as shown by the Kaplan-Meier curves (Figure 10E). Within the PC patient population, those classified as high-risk with elevated TMB exhibited the shortest survival time, while low-risk patients with low TMB had the most favorable prognosis (Figure 10F). These findings suggest that TMB and the risk signature can potentially serve as prognostic indicators for PC patients and may help guide treatment decisions. Overall, the analysis indicates that high-risk patients with specific somatic mutations, such as KRAS, TP53, SMAD4, CDKN2A, and GLI3, have increased sensitivity to immune checkpoint blockade therapy, despite similar TIDE levels. Additionally, TMB and the risk signature can provide valuable prognostic information for patients with PC.

Although there have been numerous studies on utilizing lncRNAs as molecular markers to predict the prognosis of PC patients, there is currently no systematic investigation on utilizing lncRNAs associated with immune checkpoint to predict the prognosis of PC patients[19; 20]. To the best of our knowledge, this study is the first exploration of using immune checkpoint related lncRNAs as molecular markers for predicting the prognosis of PC patients.

In this study, we obtained transcriptomic and clinical data from the TCGA-PAAD project and identified a set of lncRNAs associated with immune checkpoint genes. We utilized various analytical approaches, including univariate Cox analysis, L analysis, and multivariate Cox analysis, to investigate the role of immune checkpoint-related lncRNAs. Consequently, we identified a prognostic signature comprising two immune checkpoint-related lncRNAs, namely LINC02245 and AL008729.2. Both univariate and multivariate Cox analyses confirmed the independent prognostic value of this signature in predicting the prognosis of PC patients. To assess the prognostic performance of the signature, we employed additional evaluation measures such as ROC curves, C index, survival curves, and nomograms. Collectively, our analyses demonstrated the robust predictive capabilities of the established prognostic signature for determining the prognosis of PC patients.

The specific roles and functions of LINC02245 and AL008729.2 in PC remain unclear. Further studies are needed to elucidate their biological roles, regulatory mechanisms, and their association with the development and progression of PC. Investigating the involvement of non-coding RNAs in PC has emerged as a current research focus. For non-coding RNAs such as LINC02245 and AL008729.2, researchers may explore their expression patterns, interaction networks, and potential biological functions in PC through transcriptomics, functional studies, and bioinformatics. These investigations aim to enhance our understanding of the underlying mechanisms of PC and provide novel insights and targets for its diagnosis and treatment.

Currently, immunotherapy strategies for PC mainly include immune checkpoint inhibitors and tumor vaccines. Immune checkpoint inhibitors, such as anti-PD-1 and anti-CTLA-4 antibodies, have been evaluated in clinical trials. Some early studies have shown that PD-1 inhibitors demonstrate some efficacy in a small subset of PC patients. Additionally, other investigational immunotherapy approaches being studied include CAR-T cell therapy and tumor-associated antigen vaccines[23].

Immunotherapy for PC is still in the research and development stage[24]. Compared to other types of cancer, PC has a relatively low response to immunotherapy. This may be due to the presence of multiple immune evasion mechanisms in the PC microenvironment, including increased immune inhibitory cells, overexpression of immune checkpoints, and a lack of tumor-associated antigens[25; 26]. However, some studies have shown that immunotherapy may provide certain benefits in specific subgroups of patients[27]. Researchers are actively studying these factors to identify reliable biomarkers that can help predict patient response to immunotherapy.

The presence of neoantigens, which are generated by somatic alterations, can trigger specific immune reactions against tumors. However, these immune reactions can be suppressed by immune checkpoints. One approach to identify individuals who may benefit from immune checkpoint blockade is by assessing potential neoantigens or surrogate markers such as TMB. TMB quantifies the number of non-synonymous coding mutations in the tumor’s exome, representing the total number of somatic mutations in the coding regions of the tumor’s genome. It has been proposed that highly mutated tumors can produce a large number of neoantigens, some of which may enhance T cell reactivity. Therefore, it is hypothesized that tumors with higher mutation levels are more likely to exhibit improved responses to immune checkpoint blockade therapy[28]. Although there is a correlation between high TMB and the effectiveness of immunotherapy, the use of TMB as a diagnostic predictor in clinical trials has not been widely adopted due to conflicting research findings. The efficacy of TMB as a biomarker for predicting responsiveness to immunotherapy may be influenced by variations in the methods used to evaluate and interpret TMB[29].

Tumors employ two distinct mechanisms of immune evasion: on one hand, certain immune inhibitory factors can impede T cell infiltration; on the other hand, some tumors, despite having high levels of cytotoxic T cell infiltration, exhibit functionally inactive T cells[30]. TIDE integrates the assessment of these two mechanisms to predict the immune evasion capability of tumors. The TIDE tool is utilized to forecast the response to cancer immunotherapy[31]. A higher TIDE score is associated with a poorer response to immune checkpoint inhibition therapy. By comparing TIDE scores between groups with high and low-risk groups, it was observed that no significant differences were observed in the TIDE scores.

There are several limitations that should be acknowledged in this study. Firstly, to increase the credibility of our findings, it would be beneficial to validate the prognostic signature using independent databases. It is important to note that the lncRNAs obtained from the TCGA database may not perfectly match those from other databases due to differences in chip platforms and recording methods. Although we attempted to consult other datasets such as SEER and ICGC, we were unable to find a dataset that contained both clinical data and the corresponding expression of the lncRNAs. Therefore, validating the effectiveness of our signature in additional datasets would greatly enhance the robustness of our results. Secondly, while we conducted enrichment analysis and made assumptions about the functions of the different risk groups, further research is needed to explore the underlying mechanisms. It is important to investigate the molecular pathways and biological processes through experimental studies to gain a deeper understanding of the findings. Thirdly, conducting in vitro experiments would provide valuable experimental evidence to support our findings. However, it is worth noting that acquiring fresh tissue samples with sufficient survival time, necessary for testing lncRNA expression, is challenging within a limited timeframe. Integrating the results of bioinformatics analysis with clinical data in future research would further enhance the understanding of the findings. Lastly, although the total sample size is substantial, obtaining a larger sample size would yield more robust and convincing results. Increasing the sample size would help to minimize potential biases and improve the statistical power of the study.

In brief, a novel prognostic signature has been created utilizing 2 immune checkpoint related lncRNAs to forecast the prognosis of PC patients. Additionally, this prognostic signature is highly relevant to the infiltration of immune cells and the immune response in patients with PC, enabling accurate prediction of effectiveness of immunotherapy. It offers targeted guidance for the treatment and management of PC patients.

Data Availability

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

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.

Funding

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

Author Contributions

YCY, CZ, and XM contributed to the research conception and study design, and revised the manuscript. NX, ZZM, JYH, and DD collected and analyzed the data, and wrote the manuscript. LZ, PYC, and MLL provided critical revisions. All authors have agreed to be accountable for all aspects of the work and have addressed and resolved any questions regarding accuracy or integrity. The final manuscript has been read and approved by all authors.

Acknowledgments

We would like to thank the TCGA databases for making the data available.

H. Sung, J. Ferlay, R.L. Siegel, M. Laversanne, I. Soerjomataram, A. Jemal, and F. Bray, Global Cancer Statistics 2020: GLOBOCAN Estimates of Incidence and Mortality Worldwide for 36 Cancers in 185 Countries. CA: a cancer journal for clinicians 71 (2021) 209-249.
Z.H. Bao, X.B. Hou, H.L. Li, Y.F. Mao, and W.R. Wang, The mechanism and progress of ferroptosis in pancreatic cancer. Acta histochemica 124 (2022) 151919.
S. Shi, C.J. Qiu, X.B. Wang, Z.R. Zheng, C.Z. Yang, K. Lin, K. Zhang, M. Tu, K.R. Jiang, and W.T. Gao, Development and validation of a ferroptosis-related prognostic model in pancreatic cancer. Briefings in bioinformatics 39 (2021) 1507-1522.
T.F. Greten, A. Villanueva, F. Korangy, B. Ruf, M. Yarchoan, L. Ma, E. Ruppin, and X.W. Wang, Biomarkers for immunotherapy of hepatocellular carcinoma. Nature reviews. Clinical oncology 20 (2023) 780-798.
N. Guillot, B. Romeo, S.S. Manesh, G. Milano, P. Brest, L. Zitvogel, P. Hofman, and B. Mograbi, Manipulating the gut and tumor microbiota for immune checkpoint inhibitor therapy: from dream to reality. Trends in molecular medicine 29 (2023) 897-911.
S.L. Topalian, F.S. Hodi, J.R. Brahmer, S.N. Gettinger, D.C. Smith, D.F. McDermott, J.D. Powderly, R.D. Carvajal, J.A. Sosman, M.B. Atkins, P.D. Leming, D.R. Spigel, S.J. Antonia, L. Horn, C.G. Drake, D.M. Pardoll, L. Chen, W.H. Sharfman, R.A. Anders, J.M. Taube, T.L. McMiller, H. Xu, A.J. Korman, M. Jure-Kunkel, S. Agrawal, D. McDonald, G.D. Kollia, A. Gupta, J.M. Wigginton, and M. Sznol, Safety, activity, and immune correlates of anti-PD-1 antibody in cancer. The New England journal of medicine 366 (2012) 2443-54.
J.R. Brahmer, S.S. Tykodi, L.Q. Chow, W.J. Hwu, S.L. Topalian, P. Hwu, C.G. Drake, L.H. Camacho, J. Kauh, K. Odunsi, H.C. Pitot, O. Hamid, S. Bhatia, R. Martins, K. Eaton, S. Chen, T.M. Salay, S. Alaparthy, J.F. Grosso, A.J. Korman, S.M. Parker, S. Agrawal, S.M. Goldberg, D.M. Pardoll, A. Gupta, and J.M. Wigginton, Safety and activity of anti-PD-L1 antibody in patients with advanced cancer. The New England journal of medicine 366 (2012) 2455-65.
J.J. Quinn, and H.Y. Chang, Unique features of long non-coding RNA biogenesis and function. Nature reviews. Genetics 17 (2016) 47-62.
G. Hu, F. Niu, B.A. Humburg, K. Liao, S. Bendi, S. Callen, H.S. Fox, and S. Buch, Molecular mechanisms of long noncoding RNAs and their role in disease pathogenesis. Oncotarget 9 (2018) 18648-18663.
Y. Li, S.D. Egranov, L. Yang, and C. Lin, Molecular mechanisms of long noncoding RNAs-mediated cancer metastasis. Genes, chromosomes & cancer 58 (2019) 200-207.
W. Bin, C. Yuan, Y. Qie, and S. Dang, Long non-coding RNAs and pancreatic cancer: A multifaceted view. Biomedicine & pharmacotherapy = Biomedecine & pharmacotherapie 167 (2023) 115601.
F.F. Hu, C.J. Liu, L.L. Liu, Q. Zhang, and A.Y. Guo, Expression profile of immune checkpoint genes and their roles in predicting immunotherapy response. Briefings in bioinformatics 22 (2021).
J.A. Blake, K.R. Christie, M.E. Dolan, H.J. Drabkin, D.P. Hill, L. Ni, D. Sitnikov, S. Burgess, T. Buza, C. Gresham, F. McCarthy, L. Pillai, H. Wang, S. Carbon, H. Dietze, S.E. Lewis, C.J. Mungall, M.C. Munoz-Torres, M. Feuermann, P. Gaudet, S. Basu, R.L. Chisholm, R.J. Dodson, P. Fey, H. Mi, P.D. Thomas, A. Muruganujan, S. Poudel, J.C. Hu, S.A. Aleksander, B.K. McIntosh, D.P. Renfro, D.A. Siegele, H. Attrill, N.H. Brown, S. Tweedie, J. Lomax, D. Osumi-Sutherland, H. Parkinson, P. Roncaglia, R.C. Lovering, P.J. Talmud, S.E. Humphries, P. Denny, N.H. Campbell, R.E. Foulger, M.C. Chibucos, M.G. Giglio, H.Y. Chang, R. Finn, M. Fraser, A. Mitchell, G. Nuka, S. Pesseat, A. Sangrador, M. Scheremetjew, S.Y. Young, R. Stephan, M.A. Harris, S.G. Oliver, K. Rutherford, V. Wood, J. Bahler, A. Lock, P.J. Kersey, M.D. McDowall, D.M. Staines, M. Dwinell, M. Shimoyama, S. Laulederkind, G.T. Hayman, S.J. Wang, V. Petri, P. D'Eustachio, L. Matthews, R. Balakrishnan, G. Binkley, J.M. Cherry, M.C. Costanzo, J. Demeter, S.S. Dwight, S.R. Engel, B.C. Hitz, D.O. Inglis, P. Lloyd, S.R. Miyasato, K. Paskov, G. Roe, M. Simison, R.S. Nash, M.S. Skrzypek, S. Weng, E.D. Wong, T.Z. Berardini, D. Li, E. Huala, J. Argasinska, C. Arighi, A. Auchincloss, K. Axelsen, et al., Gene Ontology Consortium: going forward. Nucleic acids research 43 (2015) D1049-D1056.
M. Kanehisa, and S. Goto, KEGG: kyoto encyclopedia of genes and genomes. Nucleic acids research 28 (2000) 27-30.
M.V. Kuleshov, M.R. Jones, A.D. Rouillard, N.F. Fernandez, Q. Duan, Z. Wang, S. Koplev, S.L. Jenkins, K.M. Jagodnik, A. Lachmann, M.G. McDermott, C.D. Monteiro, G.W. Gundersen, and A. Ma'ayan, Enrichr: a comprehensive gene set enrichment analysis web server 2016 update. Nucleic acids research 44 (2016) W90-W97.
K. Yoshihara, M. Shahmoradgoli, E. Martínez, R. Vegesna, H. Kim, W. Torres-Garcia, V. Treviño, H. Shen, P.W. Laird, D.A. Levine, S.L. Carter, G. Getz, K. Stemke-Hale, G.B. Mills, and R.G. Verhaak, Inferring tumour purity and stromal and immune cell admixture from expression data. Nature communications 4 (2013) 2612.
A.M. Newman, C.L. Liu, M.R. Green, A.J. Gentles, W. Feng, Y. Xu, C.D. Hoang, M. Diehn, and A.A. Alizadeh, Robust enumeration of cell subsets from tissue expression profiles. Nature Methods 12 (2015) 453-+.
P. Jiang, S. Gu, D. Pan, J. Fu, A. Sahu, X. Hu, Z. Li, N. Traugh, X. Bu, B. Li, J. Liu, G.J. Freeman, M.A. Brown, K.W. Wucherpfennig, and X.S. Liu, Signatures of T cell dysfunction and exclusion predict cancer immunotherapy response. Nature medicine 24 (2018) 1550-1558.
Y. Hu, J. Tang, F. Xu, J. Chen, Z. Zeng, S. Han, F. Wang, D. Wang, M. Huang, Y. Zhao, Y. Huang, W. Zhuo, and G. Zhao, A reciprocal feedback between N6-methyladenosine reader YTHDF3 and lncRNA DICER1-AS1 promotes glycolysis of pancreatic cancer through inhibiting maturation of miR-5586-5p. Journal of Experimental & Clinical Cancer Research 41 (2022).
R. Tang, Z. Wu, Z. Rong, J. Xu, W. Wang, B. Zhang, X. Yu, and S. Shi, Ferroptosis-related lncRNA pairs to predict the clinical outcome and molecular characteristics of pancreatic ductal adenocarcinoma. Briefings in bioinformatics 23 (2022).
C.K.Y. Ng, E. Dazert, T. Boldanova, M. Coto-Llerena, S. Nuciforo, C. Ercan, A. Suslov, M.A. Meier, T. Bock, A. Schmidt, S. Ketterer, X. Wang, S. Wieland, M.S. Matter, M. Colombi, S. Piscuoglio, L.M. Terracciano, M.N. Hall, and M.H. Heim, Integrative proteogenomic characterization of hepatocellular carcinoma across etiologies and stages. Nature communications 13 (2022) 2436.
P. Liu, L. Kong, Y. Liu, G. Li, J. Xie, and X. Lu, A key driver to promote HCC: Cellular crosstalk in tumor microenvironment. Frontiers in oncology 13 (2023) 1135122.
L. Ma, E. Ruppin, X.W. Wang, M.N. Kronig, and M. Wehrli, "Hurdles race for CAR T-cell therapy in digestive tract cancer". Nature reviews. Clinical oncology 320 (2023) 100-119.
D. Schizas, N. Charalampakis, C. Kole, P. Economopoulou, E. Koustas, E. Gkotsis, D. Ziogas, A. Psyrri, and M.V. Karamouzis, Immunotherapy for pancreatic cancer: A 2020 update. Cancer Treatment Reviews 86 (2020).
P. Sarantis, E. Koustas, A. Papadimitropoulou, A.G. Papavassiliou, and M.V. Karamouzis, Pancreatic ductal adenocarcinoma: Treatment hurdles, tumor microenvironment and immunotherapy. World Journal of Gastrointestinal Oncology 12 (2020).
A.H. Morrison, K.T. Byrne, and R.H. Vonderheide, Immunotherapy and Prevention of Pancreatic Cancer. Trends in Cancer 4 (2018) 418-428.
V.P. Balachandran, G.L. Beatty, and S.K. Dougan, Broadening the Impact of Immunotherapy to Pancreatic Cancer: Challenges and Opportunities. Gastroenterology 156 (2019) 2056-2072.
A. Marabelle, M. Fakih, J. Lopez, M. Shah, R. Shapira-Frommer, K. Nakagawa, H.C. Chung, H.L. Kindler, J.A. Lopez-Martin, W.H. Miller, Jr., A. Italiano, S. Kao, S.A. Piha-Paul, J.P. Delord, R.R. McWilliams, D.A. Fabrizio, D. Aurora-Garg, L. Xu, F. Jin, K. Norwood, and Y.J. Bang, Association of tumour mutational burden with outcomes in patients with advanced solid tumours treated with pembrolizumab: prospective biomarker analysis of the multicohort, open-label, phase 2 KEYNOTE-158 study. The Lancet. Oncology 21 (2020) 1353-1365.
L. Fancello, S. Gandini, P.G. Pelicci, and L. Mazzarella, Tumor mutational burden quantification from targeted gene panels: major advancements and challenges. Journal for Immunotherapy of Cancer 7 (2019).
J. Wu, L. Li, H. Zhang, Y. Zhao, H. Zhang, S. Wu, and B. Xu, A risk model developed based on tumor microenvironment predicts overall survival and associates with tumor immunity of patients with lung adenocarcinoma. Oncogene 40 (2021) 4413-4424.
X. Li, J. Kang, J. Yue, D. Xu, C. Liao, H. Zhang, J. Zhao, Q. Liu, J. Jiao, L. Wang, and G. Li, Identification and validation of immunogenic cell death-related score in uveal melanoma to improve prediction of prognosis and response to immunotherapy. Aging 15 (2023) 3442-3464.

Table 1. Clinical characteristics of the validation set and training set for PC [n(%)].

Covariates	Type	Total	Test	Train	P value
Age	<=65	94(52.81%)	51(57.3%)	43(48.31%)	0.2933
	>65	84(47.19%)	38(42.7%)	46(51.69%)
Gender	FEMALE	80(44.94%)	40(44.94%)	40(44.94%)	1
	MALE	98(55.06%)	49(55.06%)	49(55.06%)
Grade	G1	31(17.42%)	15(16.85%)	16(17.98%)	0.7903
	G2	95(53.37%)	45(50.56%)	50(56.18%)
	G3	48(26.97%)	27(30.34%)	21(23.6%)
	G4	2(1.12%)	1(1.12%)	1(1.12%)
	unknow	2(1.12%)	1(1.12%)	1(1.12%)
Stage	Stage I	21(11.8%)	10(11.24%)	11(12.36%)	0.376
	Stage II	147(82.58%)	75(84.27%)	72(80.9%)
	Stage III	3(1.69%)	0(0%)	3(3.37%)
	Stage IV	4(2.25%)	2(2.25%)	2(2.25%)
	unknow	3(1.69%)	2(2.25%)	1(1.12%)
T	T1	7(3.93%)	4(4.49%)	3(3.37%)	0.3455
	T2	24(13.48%)	11(12.36%)	13(14.61%)
	T3	142(79.78%)	72(80.9%)	70(78.65%)
	T4	3(1.69%)	0(0%)	3(3.37%)
	unknow	2(1.12%)	2(2.25%)	0(0%)
M	M0	80(44.94%)	39(43.82%)	41(46.07%)	1
	M1	4(2.25%)	2(2.25%)	2(2.25%)
	unknow	94(52.81%)	48(53.93%)	46(51.69%)
N	N0	49(27.53%)	27(30.34%)	22(24.72%)	0.5305
	N1	124(69.66%)	60(67.42%)	64(71.91%)
	unknow	5(2.81%)	2(2.25%)	3(3.37%)

No competing interests reported.

Establishment of a prognostic signature for pancreatic cancer using immune checkpoint-related lncRNAs

Status:

Version 1

Abstract

Figures

1 Introduction

2 Materials and methods

3 Results

4 Discussion

Declarations

References

Table

Additional Declarations

Status:

Version 1