Exploring Potential Drug Targets for Pancreatic Cancer Based on Mendelian Randomization

doi:10.21203/rs.3.rs-4183057/v1

Download PDF

Research Article

Exploring Potential Drug Targets for Pancreatic Cancer Based on Mendelian Randomization

https://doi.org/10.21203/rs.3.rs-4183057/v1

This work is licensed under a CC BY 4.0 License

You are reading this latest preprint version

Objective

Genome-wide association studies (GWAS) provide a rich resource for identifying risk factors and biomarkers associated with cancer susceptibility. This study aims to use Mendelian randomization (MR) analyses within the proteome and transcriptome to explore potential protein markers and therapeutic targets for pancreatic cancer.

Methods

Exposure data were derived from expression quantitative trait loci (eQTL) data from GTEx V8 and the eQTLGen Consortium, covering 838 and 31,684 participants, as well as protein quantitative trait loci (pQTL) data for 3,703 proteins with a sample size of 27,698 participants. The pancreatic cancer GWAS dataset was obtained from the FinnGen Consortium, including 1,626 pancreatic cancer patients and 314,193 controls. The inverse variance weighted(IVW) and Wald ratio were the main analytical methods to assess the causal relationship between the proteome/transcriptome and pancreatic cancer. Cochran's Q test and MR-Egger intercept were used to evaluate heterogeneity and pleiotropy. Gene ontology (GO) analysis, Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway enrichment analysis, and protein-protein interaction networks revealed functional characteristics and biological relevance.

Results

A total of 16,059 mRNAs and 1,608 proteins were included in the study. MR analysis using pQTL and eQTL data showed that 88 proteins and 811 mRNAs were causally related to pancreatic cancer based on the IVW and Wald ratio methods. Among these 88 proteins and 811 mRNAs, eight genes overlapped, including HAGH, FGF2, DTD2, IDUA, and CD248, demonstrating consistent causal effects with pancreatic cancer at both the protein and mRNA levels. However, IRF3, PILRA, and AMY2B showed inconsistent effects on pancreatic cancer at the protein and mRNA levels. GO analysis highlighted processes related to cellular transport, and KEGG pathway analysis suggested involvement in metabolic pathways and signaling pathways.

Conclusion

This study identified key proteins and mRNAs associated with pancreatic cancer, enhancing our understanding of the disease's molecular mechanisms and providing insights for future research and therapeutic development.

expression quantitative trait loci

protein quantitative trait loci

mendelian randomization

drug targets

pancreatic cancer

Pancreatic cancer is a highly malignant tumor. According to the GLOBOCAN 2020 global cancer statistics report, there were approximately 496,000 new cases of pancreatic cancer worldwide in 2020, ranking 14th among malignant tumors in incidence. Correspondingly, it caused about 466,000 deaths annually, ranking 7th among malignant tumors^[1]. It is estimated that by 2050, the global incidence of pancreatic cancer will increase to 18.6 cases per 100,000 people, with an annual growth rate of 1.1%^[2]. Pancreatic cancer is a highly lethal cancer, and its challenges in early diagnosis and treatment make it a very challenging health issue.

In recent years, there have been advances in the treatment of pancreatic cancer, including surgery, chemotherapy, immunotherapy, and targeted therapy. However, the overall survival rates for pancreatic cancer globally remain very limited. According to data from the American Cancer Society in 2023, the 5-year relative survival rate for pancreatic cancer in the United States is 12.5%^[3]. While adjuvant chemotherapy after surgery can delay the recurrence of pancreatic ductal adenocarcinoma, around 80% of patients still experience recurrence after 14 months^[4], with a total 5-year survival rate of less than 30%^[5]. For patients who have lost the opportunity for surgery but have a Karnofsky Performance Status (KPS) score of ≥ 70 and metastatic pancreatic cancer, standard treatment includes combination chemotherapy. However, chemotherapy resistance is inevitable, and the prognosis is poor, with limited options for second-line chemotherapy after disease progression following first-line chemotherapy. In recent years, breakthroughs have been made in immunotherapy and targeted drugs for various solid tumors^[6–8]. However, for pancreatic cancer, known as an "immunologically cold" tumor, the efficacy of immune checkpoint inhibitors as a single agent is limited due to the tumor microenvironment being in an immune-suppressed state, lacking effective predictive biomarkers to select patients who would benefit from immunotherapy or predict treatment efficacy^{[9, 10]}. KRAS, CDKN2A, TP53, and SMAD4 are pancreatic cancer's main oncogenic driver genes^[11]. With over 90% of pancreatic ductal adenocarcinoma patients having a KRAS mutation, KRAS has become a critical target for the treatment of pancreatic cancer patients^[12]. Currently, clinical trials of targeted drugs in pancreatic cancer have not achieved ideal results, and further research is needed to find more effective targeted drugs.

In recent years, whole-genome association studies have provided new opportunities for developing targeted drugs. Combining genome-wide association study (GWAS) data with genomic data from pancreatic cancer patients, key gene variations closely related to tumor occurrence, development, and treatment can be identified, thereby determining potential drug targets. GWAS data analysis can help identify molecular biomarkers and pathways closely associated with the occurrence and development of pancreatic cancer, providing a more precise targeting for personalized treatment. This personalized treatment strategy can assist doctors in selecting the most suitable drugs based on the patient's genetic characteristics and reduce the likelihood of adverse reactions. Furthermore, GWAS data can provide new clues and targets for drug development, accelerating the discovery and development of new drugs.

Mendelian randomization (MR) is the "randomized, double-blind trial created by nature." Its core idea is to use genetic variations associated with exposure to infer causality between exposure and outcome^[13]. In drug development, regulating protein levels in the human body to influence disease characteristics is a key issue. Recently, whole-genome association studies have identified single nucleotide polymorphisms (SNPs) associated with protein and mRNA expression levels, providing protein quantitative trait loci (pQTL) and expression quantitative trait loci (QTL), which offer tools for studying the causal mechanisms of disease risk of drug targets.

MR analysis has been applied in various fields, such as cardiovascular diseases, mental disorders, and tumors, demonstrating this analytical method's empirical validation and reliability^[14–16]. MR studies have been widely used to explore the causal relationships between various modifiable exposure factors and pancreatic cancer, such as gut microbiota, obesity, dietary habits, and emotions^[17–19]. However, MR studies integrating eQTL and pQTL data with pancreatic cancer data are still lacking. Therefore, conducting a dual-sample drug target MR analysis based on eQTL and pQTL data for pancreatic cancer, identifying a series of potentially causally associated transcript or protein expressions with the onset of pancreatic cancer, exploring the potential pathways in which the selected genes participate in the pathogenesis of pancreatic cancer, will help uncover new risk genes and disease mechanisms of pancreatic cancer, providing promising genetic therapeutic targets for pancreatic cancer.

Study Design

MR analysis should satisfy the following three assumptions ^[20]: (1) the selected instrumental variables are strongly associated with the exposure (pQTL and eQTL); (2) instrumental variables are not related to any potential confounding factors; (3) instrumental variables only influence the outcome through the exposure and not through any other pathways, there is no genetic pleiotropy. The overall study design is illustrated in Fig. 1.

Data Source

eQTL and pQTL data were obtained from the study by Zheng et al. and selected as instrumental variables ^[21]. The eQTL data were sourced from the GTEx V8 and eQTLGen consortium ^{[22, 23]}. The GTEx V8 database comprises 838 donors and 15,201 samples, covering 49 tissues or cell types. The eQTLGen consortium collected blood samples from 31,684 individuals across 37 datasets. The pQTL dataset was derived from six aggregated datasets ^{[24, 25]}, specifically including 3,301 healthy participants from the INTERVAL study, involving 50,000 blood donors from 25 centers in England^[26]; 1,000 individuals from Germany^[27]; 3,394 European participants ^[28]; 7,333 participants from the Birmingham Cardiovascular Risk Study^[29]; 5,457 individuals aged 65 and older from Iceland ^[30]; and 7,213 European American participants from the Atherosclerosis Risk in Communities study^[25].

Pancreatic cancer GWAS summary statistics data were obtained from the FinnGen consortium R10 version (https://r10.finngen.fi/) and used as the outcome instrument. In this study, "Malignant neoplasm of pancreas (excluding all tumors)" was employed as the phenotype. The GWAS data sample consisted of 315,819 individuals, including 1,626 cases of pancreatic cancer and 314,193 controls. The exposed and outcome groups were predominantly composed of individuals of European descent to avoid potential biases due to ethnic differences, with little to no apparent sample overlap between the exposed and outcome groups.

This study involves reanalyzing previously collected and published data, so institutional review board ethical approval is not required.

Instrumental Variable SNP Selection

By filtering SNPs in the pQTL and eQTL data with P < 5x10^− 8 and calculating F values > 10, SNPs were selected to meet the strong correlation with pancreatic cancer. The instrumental variables in the eQTL data exhibit minimal linkage disequilibrium (LD r² < 0.001). For the pQTL instrumental variables, out of 5,326 instrumental variables, 272 (5.1%) show low to moderate linkage disequilibrium (r² between 0.2 and 0.6).

Mendelian Randomization

This study primarily employed inverse variance weighted (IVW) and Wald ratio to evaluate the relationship between plasma proteins, mRNA, and pancreatic cancer. MR-Egger and Weighted Median methods were used to complement and test the robustness of MR results because these three calculation methods have different assumptions about the effectiveness of instrumental variables, with IVW demonstrating higher efficiency than the other two MR methods. IVW is a weighted linear regression model that assumes all genetic variants are valid. It calculates the causal effect estimate of individual instrumental variables using the ratio method and combines each estimate through weighted linear regression to obtain the overall causal effect estimate^[31]. In cases where only a single instrumental variable is available, the Wald ratio is used to calculate the MR estimate for each SNP. MR-Egger regression analysis can identify and control for bias caused by horizontal pleiotropy. Even if all SNPs are invalid, as long as the association of a single SNP with the exposure is independent of the corresponding horizontal pleiotropy, MR-Egger regression analysis can provide an effective estimate where the slope represents the causal effect estimate of exposure on the outcome^[32]. The main difference between MR-Egger and IVW is considering the intercept term in regression. The Weighted Median method utilizes the intermediate effects of all available genetic variants, obtaining estimates by inversely weighting each SNP's correlation with the outcome and requiring at least 50% of the weight to be contributed by effective genetic variants^[33]. If the estimates from these three methods are consistent in direction, it suggests the reliability of the MR analysis. In inconsistent directions, a reassessment of the MR analysis is needed after narrowing down significant P-values^[34].

Sensitivity Analysis

Sensitivity analysis is one of the methods used to assess the reliability of MR study results. This study utilized Cochran’s Q test and MR-Egger regression analysis. Cochran’s Q test was employed to quantify the heterogeneity in the magnitude of effects generated by the selected genetic instrumental variables, with P < 0.05 indicating the presence of heterogeneity. When the intercept of the MR-Egger regression is non-zero and P < 0.05, it suggests the presence of instrumental variable pleiotropy^[35].

Enrichment Analysis

In order to investigate the functional characteristics and biological relevance of genes, we conducted Gene Ontology (GO) and Kyoto Encyclopedia of Genes and Genomes (KEGG) enrichment analyses. GO analysis compares gene sets with predefined functional categories, such as cellular components (CC), molecular functions (MF), and biological processes (BP), to identify enriched functional categories, thus aiding in understanding the main functional modules and biological processes involved in the gene set^[36]. KEGG enrichment analysis compares gene sets with metabolic pathways, signaling pathways, and other pathways in the KEGG database to determine whether specific pathways are enriched in certain diseases or biological processes. This helps reveal the association between gene sets and specific pathways, furthering our understanding of their roles in disease development or biological processes^[37]. Systematic analysis of the functional and pathway enrichment information of gene sets can provide insights into the functions and interactions of gene sets in biological processes and diseases, offering important clues and guidance for subsequent experimental design and mechanistic studies. The Database for Annotation, Visualization, and Integrated Discovery (DAVID) is a powerful online tool for functional annotation and analysis of gene and protein data ^[38]. In this study, we utilized DAVID for GO and KEGG enrichment analysis.

Protein-Protein Interaction Network Construction

We used the online tool STRING to construct a Protein-Protein Interaction (PPI) network analysis of significant MR results to explore potential interactions among the identified genes. We imported the STRING result file into Cytoscape for visualization. Subsequently, we used the cytoHubba plugin to identify key genes in the network further to investigate the importance of these genes in biological processes.

Statistical Analysis

All MR analysis and enrichment analysis visualizations were performed using R software (version 4.2.1). The MR analysis utilized the TwoSampleMR package (version 0.5.6). Regarding statistical significance, a two-sided p-value less than 0.05 was considered significant.

Instrument Variable Selection

A total of 39,630 eQTL instrumental variables from 16,059 transcriptomes and 5,326 pQTL instrumental variables from 1,608 plasma proteins were included in the MR analysis. For the pQTL instrumental variables, the selected SNP numbers ranged from 1 to 3, with F values exceeding 10 (ranging from 10 to 8146). In the eQTL instrumental variables, the selected SNP numbers ranged from 1 to 581, with F statistics greater than 10 (ranging from 29 to 850), indicating an effective reduction of bias caused by weak instrumental variables. Please refer to Supplementary Table S1-2 for detailed information on the instrumental variables.

MR Results

Using volcano plots, we visualized significant MR results for the causal relationships between pQTL, eQTL, and pancreatic cancer (Fig. 2, Supplementary Table S3-4). IVW and Wald ratio in MR analysis identified 88 proteins causally associated with pancreatic cancer. Among these, an increase in the abundance of 44 proteins was significantly associated with increased pancreatic cancer risk. In comparison, an increase in the abundance of another 44 proteins was significantly associated with decreased pancreatic cancer risk (Fig. 2A).

Through MR analysis of tissue or cell mRNA expression eQTLs, we determined causal relationships between 811 genes and pancreatic cancer risk, with 407 showing positive causal relationships and 404 showing negative causal relationships (Fig. 2B).

Comparing the MR results in forest plot for overlapping genes (Fig. 3), we found that genes such as IRF3, HAGH, FGF2, PILRA, DTD2, IDUA, CD248, and AMY2B exhibit significant causal effects on pancreatic cancer at both transcriptional and protein levels in blood, tissue, and cells. Specifically (Fig. 4), HAGH, FGF2, DTD2, IDUA, and CD248 show consistent directional effects on pancreatic cancer between protein and mRNA levels.

However, IRF3, PILRA, and AMY2B show some differences in their effects on pancreatic cancer between protein and mRNA levels. For instance, an increase in protein levels of PILRA and AMY2B is associated with increased pancreatic cancer risk. In contrast, a corresponding increase in mRNA levels is associated with decreased pancreatic cancer risk. This inconsistency may reflect the complexity of the regulatory mechanisms of genes at transcriptional and protein levels, requiring further research to elucidate their roles in the pathogenesis of pancreatic cancer.

Sensitivity Analysis

Cochran's Q test P-values quantify the degree of heterogeneity between genes. If the P-value is < 0.05, it indicates the presence of heterogeneity. According to Cochran's Q test results, the MR analysis of the 88 genes in pQTL did not show heterogeneity (all P-values > 0.05). In the eQTL analysis, except for EPHB4 showing heterogeneity (P = 0.013), heterogeneity did not affect the rest of the genes. Because the random-effects IVW method was used in this experiment, heterogeneity could be accepted.

MR-Egger regression test is used to detect horizontal pleiotropy, as the presence of horizontal pleiotropy may make MR results less robust, thus affecting conclusions. If the test for horizontal pleiotropy is significant, it indicates that the instrumental variable is related to the exposure and other confounding factors. However, in this study, MR-Egger regression indicates the absence of horizontal pleiotropy (P > 0.05).

Tables 1 and 2 show the sensitivity results of pQTL and eQTL intersection genes. Supplementary Table S5-8 shows detailed sensitivity analysis results.

Table 1

Sensitivity analysis of MR analysis on the causal effects of intersecting genes of pQTL on pancreatic cancer.
Intersection genes	Cochran’s Q test		MR-egger
Intersection genes	Q	P -value	intercept	P- value
IRF3	0.167	0.920	0.037	0.789
HAGH	1.957	0.581	-0.486	0.375
FGF2	/	/	/	/
PILRA	0.002	0.999	0.002	0.983
DTD2	1.370	0.504	0.112	0.481
IDUA	10.716	0.635	-0.018	0.636
CD248	/	/	/	/
AMY2B	10.996	0.529	-0.066	0.169

Table 2

Sensitivity analysis of MR analysis on the causal effects of intersecting genes of eQTL on pancreatic cancer.
Intersection genes	Cochran Q test		MR-egger
Intersection genes	Q	P value	intercept	P value
IRF3	0.391	0.532	/	/
HAGH	5.164	0.523	-0.005	0.866
FGF2	/	/	/	/
PILRA	1.464	0.691	0.048	0.444
DTD2	/	/	/	/
IDUA	0.061	0.970	0.004	0.963
CD248	2.107	0.349	-0.093	0.531
AMY2B	0.098	0.992	0.016	0.910

GO and KEGG Enrichment Analysis

Based on the results of GO and KEGG enrichment analysis using DAVID software for the 88 genes with causal pQTL and 811 genes with eQTL, it was found that the pQTL were enriched in 296 biological processes (BP), 34 cellular components (CC), 43 molecular functions (MF), and 14 pathways. Similarly, the eQTL was enriched in 184 BP, 50 CC, 37 MF, and 13 pathways (Supplementary Table S9-10).

As shown in Fig. 5A, the 88 genes of pQTL are mainly enriched in BP, such as cell adhesion, cellular metabolism, and response to stimuli in the BP category. In the CC category, they are mainly enriched in the extracellular matrix, endoplasmic reticulum, Golgi apparatus, and cytoplasmic vesicles. In the MF category, they are mainly enriched in binding to various substances, such as receptors, calcium ions, metal ions, protein complexes, macromolecular complexes, and cell adhesion molecules. Four significantly enriched (P < 0.05) KEGG pathways were identified, involving metabolic pathways, nucleotide metabolism, pancreatic secretion, purine metabolism, and the PI3K-Akt signaling pathway.

Figure 5B displays that the 811 genes of eQTL are primarily involved in BP, such as intracellular and extracellular protein transport, localization, and cellular metabolism. The CC category mainly involves the Golgi apparatus, endoplasmic reticulum, mitochondria, and cellular vesicles. Regarding MF, these genes are mainly associated with binding substances such as protein kinases, ions, phospholipids, nucleotides, and enzymes. The KEGG enrichment analysis significantly enriched metabolism and signaling pathways, including metabolic pathways, amino sugar and nucleotide sugar metabolism, ketone body metabolism, MAPK signaling, and Wnt signaling pathways.

PPI Network Analysis

The genes corresponding to the 88 significant proteins and the 811 significant genes were separately input into the STRING database to construct PPI networks, and the PPI network results were saved in TSV text format. Subsequently, Cytoscape was used to visualize these networks and identify the top 8 core genes.

In Figs. 6A and 6B, the PPI network constructed from the 88 genes of pQTL is displayed, comprising 60 nodes and 80 edges. Further analysis using the cytoHubba plugin in Cytoscape software identified eight core genes with strong interactions, namely FN1, FGF2, KITLG, CBL, PRKCA, ASPN, COL6A1, and NT5E.

Additionally, the PPI network constructed from the 811 genes of pQTL consists of 501 nodes and 1282 edges, as shown in Figs. 6C and 6D. After analysis, the top 8 core genes with strong interactions were determined to be RPS19, RPS18, EIF4A1, PABPC1, EIF4G1, EIF3C, HSPA4, and H4C6.

This study represents the first attempt to investigate pancreatic cancer MR drug targets using multi-omics data on plasma protein and transcript levels. The study identified 88 significantly correlated proteins and 811 mRNAs. Additionally, a comprehensive method was proposed to identify pancreatic cancer-related proteins, mRNAs, and pathways, filling the knowledge gap in genetic variations left by previous GWAS studies.

Through pQTL MR analysis, we identified 88 genes associated with pancreatic cancer. Based on the MR effect values, ALDH3A1 was the most positively associated gene with pancreatic cancer, while CD248 was the most negatively associated gene. Similarly, eQTL MR analysis also identified 811 genes associated with pancreatic cancer, with TUSC2 exhibiting the largest positive effect and RDH14 showing the largest negative effect. Results from Cochran's Q test and MR-Egger regression demonstrated the stability and reliability of the study findings, indicating no heterogeneity or horizontal pleiotropy. Notably, ALDH3A1 has been identified as a biomarker for pancreatic cancer stem-like cells characterized by high tumorigenicity^[39], self-renewal, differentiation capabilities^[40], and involvement in chemotherapy resistance and invasive behavior^[41]. CD248, expressed by tumor-associated fibroblasts and pericytes, promotes tumor angiogenesis and growth^[42]. TUSC2 was initially discovered as a potential tumor suppressor gene, and its restoration of expression can facilitate tumor suppression, reduce cell proliferation, stemness, and tumor growth, and increase apoptosis^[43]. These previous studies support the effectiveness of this approach.

In this study, we also observed that IRF3, HAGH, FGF2, PILRA, DTD2, IDUA, CD248, and AMY2B showed causal risk associations with pancreatic cancer in both eQTL and pQTL analyses, consistent with previous research findings^{[44, 45]}. For instance, IRF3, an interferon regulatory factor, is critical in immune responses and tumor cell biology processes^{[46, 47]}. Low levels of IRF3 have been associated with significantly shortened overall survival in pancreatic cancer patients^[48]. FGF2 is a major factor leading to tissue fibrosis, which plays an important role in the progression of pancreatic cancer. Sakai et al. found that subcutaneous inoculation of BxPC-3 pancreatic cancer cells in mice and administration of FGF2 led to an increased stromal fibrotic area compared to mice inoculated with BxPC-3 alone. This fibrosis was characterized by increased collagen deposition in mice and was associated with increased monocyte/macrophage content in tumor tissue, promoting pancreatic cancer progression^[49].

The inconsistent results between protein and mRNA levels may be attributed to several factors. Firstly, differences in equipment, reagents, and statistical methods could lead to inaccuracies in the QTL data. Secondly, variations in sample size and the number of genes in the pQTL and eQTL datasets may also play a role. Additionally, these discrepancies could be due to post-transcriptional modifications, such as mRNA splicing and protein degradation^[50].

Tissue- and cell-specific gene expression has been shown to elucidate different biological molecular mechanisms. Evaluating significant MR results in the broader context of pathways, tissue, and cellular functions reveals that the genes associated with pancreatic cancer primarily involve functions related to cellular energy metabolism, intracellular and extracellular protein transport and localization, and binding of various macromolecular complexes and ions, among others. Enrichment analysis shows significant enrichment of these genes in metabolic pathways, the PI3K-Akt signaling pathway, the MAPK signaling pathway, and the Wnt signaling pathway. These findings provide new insights into the molecular mechanisms of pancreatic cancer pathology, potentially advancing the development of targeted therapeutic interventions. Pancreatic cancer cells exhibit extensive metabolic reprogramming, enabling them to survive in harsh microenvironments^[51]. The hallmark of pancreatic tumors is the hypoxic tumor microenvironment and the resulting acidic environment that promotes enhanced glycolytic metabolism. LaRue et al. found that tumor-associated macrophages promote pancreatic cancer fibrosis through collagen turnover-induced metabolic reprogramming. Macrophages with a tumor-associated phenotype accumulate intracellular free amino acids derived from collagen breakdown through MRC1-mediated collagen internalization, leading to increased arginine biosynthesis. This results in elevated intracellular arginine levels, inducing upregulation of inducible nitric oxide synthase and production of reactive nitrogen species. The reactive nitrogen species from internalized and degraded collagen promote a pro-fibrotic phenotype in pancreatic stellate cells, enhancing intratumoral collagen deposition ^[52].

When it comes to a serious illness like pancreatic cancer, a deep understanding of its potential pathogenic mechanisms and risk factors is crucial. By integrating multi-omics data and applying causal inference methods, we can more accurately identify genes and biological processes associated with pancreatic cancer, providing important clues for disease prevention, diagnosis, and treatment. MR significantly reduces confounding and reverse causation bias, enabling effective causal inference. Using large-scale GWAS data increases the power of study design, and the consistency of results across multiple datasets provides strong support for result accuracy. Moreover, research findings indicate that genes such as IRF3, HAGH, FGF2, PILRA, DTD2, IDUA, CD248, and AMY2B show causal associations with the risk of pancreatic cancer, offering important clues for understanding the etiology of pancreatic cancer. Through more in-depth functional enrichment analysis and gene interaction network evaluation, we can further elucidate the mechanistic role of these potential risk genes in pancreatic cancer development.

The study primarily covers samples of European ancestry to minimize bias due to different genetic backgrounds. However, the study also has some limitations, such as the sample mainly consisting of individuals of European descent, necessitating further research and validation of results in different populations. Additionally, the gene transcription and plasma protein levels are not paired samples, which may not accurately reflect tissue-specific expression patterns. Despite MR analysis minimizing bias to the greatest extent, it is still susceptible to unmeasured factors. Therefore, further research requires larger genetic datasets and complex experiments to validate MR results and elucidate the impact mechanisms of these genes on pancreatic cancer risk. Future studies can integrate more omics data and environmental factors to understand pancreatic cancer's potential mechanisms comprehensively. By considering various biological factors beyond genetic data, researchers can delve into the complex interactions among genetic, environmental, and lifestyle factors contributing to the disease, thereby discovering new treatment possibilities and ultimately improving the prognosis of pancreatic cancer patients.

Data availability statement

The data used in this study were obtained from publicly available databases. Pancreatic cancer data can be found in the Finngen consortium(https://r10.finngen.fi/).

Ethics statement

All data used in this study were retrieved from publicly available data and, therefore, did not require additional medical ethics review consent or ethics approval or declaration.

Author contributions

Jumping huang designed the study and conducted the statistical analysis; yi dang participated in data interpretation; ya wang and peishan yao drafted the initial manuscript. peishan yao has direct access to and is responsible for verifying all data reported in the manuscript. All authors have read and approved the submitted version of the manuscript.

Conflict of interest

The authors declare that they have no conflict of interest.

Funding

No funding.

Acknowledgments

The authors acknowledge the efforts and selfless contributions of all researchers for whom GWAS summary data are publicly available.

SUNG H, FERLAY J, SIEGEL R L. Global Cancer Statistics 2020: GLOBOCAN Estimates of Incidence and Mortality Worldwide for 36 Cancers in 185 Countries [J]. 2021, 71(3): 209-49.
HU J X, ZHAO C F, CHEN W B, et al. Pancreatic cancer: A review of epidemiology, trend, and risk factors [J]. World journal of gastroenterology, 2021, 27(27): 4298-321.
SIEGEL R L, MILLER K D. Cancer statistics, 2023 [J]. 2023, 73(1): 17-48.
PARK W, CHAWLA A, O'REILLY E M. Pancreatic Cancer: A Review [J]. Jama, 2021, 326(9): 851-62.
STROBEL O, NEOPTOLEMOS J. Optimizing the outcomes of pancreatic cancer surgery [J]. 2019, 16(1): 11-26.
KO B, TAO K, BRENNAN L, et al. Evaluating the efficacy of combination and single-agent immunotherapies in real-world patterns of disease progression and survival of metastatic melanoma patients [J]. Melanoma research, 2024,
EMENS L A. Breast Cancer Immunotherapy: Facts and Hopes [J]. Clinical cancer research : an official journal of the American Association for Cancer Research, 2018, 24(3): 511-20.
ABUSHANAB A K, MUSTAFA M T, MOUSA M T, et al. Efficacy and safety of tislelizumab for malignant solid tumor: a systematic review and meta-analysis of phase III randomized trials [J]. Expert review of clinical pharmacology, 2023, 16(11): 1153-61.
ROYAL R E, LEVY C, TURNER K, et al. Phase 2 trial of single agent Ipilimumab (anti-CTLA-4) for locally advanced or metastatic pancreatic adenocarcinoma [J]. J Immunother, 2010, 33(8): 828-33.
KAMATH S D, KALYAN A, KIRCHER S, et al. Ipilimumab and Gemcitabine for Advanced Pancreatic Cancer: A Phase Ib Study [J]. Oncologist, 2020, 25(5): e808-e15.
KAMISAWA T, WOOD L D, ITOI T, et al. Pancreatic cancer [J]. Lancet (London, England), 2016, 388(10039): 73-85.
SMIT V T, BOOT A J, SMITS A M, et al. KRAS codon 12 mutations occur very frequently in pancreatic adenocarcinomas [J]. Nucleic acids research, 1988, 16(16): 7773-82.
SMITH G D, EBRAHIM S. 'Mendelian randomization': can genetic epidemiology contribute to understanding environmental determinants of disease? [J]. International journal of epidemiology, 2003, 32(1): 1-22.
JIANG M, YAN W, ZHANG Y, et al. Phosphodiesterase and psychiatric disorders: a two-sample Mendelian randomization study [J]. Journal of translational medicine, 2023, 21(1): 560.
YARMOLINSKY J, DíEZ-OBRERO V. Genetically proxied therapeutic inhibition of antihypertensive drug targets and risk of common cancers: A mendelian randomization analysis [J]. 2022, 19(2): e1003897.
MAZIDI M, WRIGHT N, YAO P, et al. Plasma Proteomics to Identify Drug Targets for Ischemic Heart Disease [J]. Journal of the American College of Cardiology, 2023, 82(20): 1906-20.
JIANG Z, MOU Y, WANG H, et al. Causal effect between gut microbiota and pancreatic cancer: a two-sample Mendelian randomization study [J]. BMC cancer, 2023, 23(1): 1091.
CARRERAS-TORRES R, JOHANSSON M, GABORIEAU V, et al. The Role of Obesity, Type 2 Diabetes, and Metabolic Factors in Pancreatic Cancer: A Mendelian Randomization Study [J]. Journal of the National Cancer Institute, 2017, 109(9):
YUAN S, CHEN J. Smoking, alcohol consumption, and 24 gastrointestinal diseases: Mendelian randomization analysis [J]. 2023, 12(
EMDIN C A, KHERA A V, KATHIRESAN S. Mendelian Randomization [J]. Jama, 2017, 318(19): 1925-6.
ZHENG J, ZHANG Y, ZHAO H, et al. Multi-ancestry Mendelian randomization of omics traits revealing drug targets of COVID-19 severity [J]. EBioMedicine, 2022, 81(104112.
VõSA U, CLARINGBOULD A, WESTRA H J, et al. Large-scale cis- and trans-eQTL analyses identify thousands of genetic loci and polygenic scores that regulate blood gene expression [J]. Nat Genet, 2021, 53(9): 1300-10.
The GTEx Consortium atlas of genetic regulatory effects across human tissues [J]. Science, 2020, 369(6509): 1318-30.
ZHENG J, HABERLAND V, BAIRD D, et al. Phenome-wide Mendelian randomization mapping the influence of the plasma proteome on complex diseases [J]. Nat Genet, 2020, 52(10): 1122-31.
ZHANG J, DUTTA D, KöTTGEN A, et al. Plasma proteome analyses in individuals of European and African ancestry identify cis-pQTLs and models for proteome-wide association studies [J]. Nat Genet, 2022, 54(5): 593-602.
SUN B B, MARANVILLE J C, PETERS J E, et al. Genomic atlas of the human plasma proteome [J]. Nature, 2018, 558(7708): 73-9.
SUHRE K, ARNOLD M, BHAGWAT A M, et al. Connecting genetic risk to disease end points through the human blood plasma proteome [J]. Nat Commun, 2017, 8(14357.
FOLKERSEN L, FAUMAN E, SABATER-LLEAL M, et al. Mapping of 79 loci for 83 plasma protein biomarkers in cardiovascular disease [J]. PLoS Genet, 2017, 13(4): e1006706.
YAO C, CHEN G, SONG C, et al. Genome-wide mapping of plasma protein QTLs identifies putatively causal genes and pathways for cardiovascular disease [J]. Nat Commun, 2018, 9(1): 3268.
EMILSSON V, ILKOV M, LAMB J R, et al. Co-regulatory networks of human serum proteins link genetics to disease [J]. Science, 2018, 361(6404): 769-73.
BURGESS S, BUTTERWORTH A, THOMPSON S G. Mendelian randomization analysis with multiple genetic variants using summarized data [J]. Genetic epidemiology, 2013, 37(7): 658-65.
BURGESS S, THOMPSON S G. Interpreting findings from Mendelian randomization using the MR-Egger method [J]. European journal of epidemiology, 2017, 32(5): 377-89.
BOWDEN J, DAVEY SMITH G, HAYCOCK P C, et al. Consistent Estimation in Mendelian Randomization with Some Invalid Instruments Using a Weighted Median Estimator [J]. Genetic epidemiology, 2016, 40(4): 304-14.
CHEN X, KONG J, DIAO X, et al. Depression and prostate cancer risk: A Mendelian randomization study [J]. 2020, 9(23): 9160-7.
BOWDEN J, DAVEY SMITH G, BURGESS S. Mendelian randomization with invalid instruments: effect estimation and bias detection through Egger regression [J]. International journal of epidemiology, 2015, 44(2): 512-25.
ASHBURNER M, BALL C A, BLAKE J A, et al. Gene ontology: tool for the unification of biology. The Gene Ontology Consortium [J]. Nature genetics, 2000, 25(1): 25-9.
OGATA H, GOTO S, SATO K, et al. KEGG: Kyoto Encyclopedia of Genes and Genomes [J]. Nucleic acids research, 1999, 27(1): 29-34.
DENNIS G, JR., SHERMAN B T, HOSACK D A, et al. DAVID: Database for Annotation, Visualization, and Integrated Discovery [J]. Genome biology, 2003, 4(5): P3.
MARCATO P, DEAN C A, GIACOMANTONIO C A, et al. Aldehyde dehydrogenase: its role as a cancer stem cell marker comes down to the specific isoform [J]. Cell cycle (Georgetown, Tex), 2011, 10(9): 1378-84.
LI C, HEIDT D G, DALERBA P, et al. Identification of pancreatic cancer stem cells [J]. Cancer research, 2007, 67(3): 1030-7.
DUONG H Q, HWANG J S, KIM H J, et al. Aldehyde dehydrogenase 1A1 confers intrinsic and acquired resistance to gemcitabine in human pancreatic adenocarcinoma MIA PaCa-2 cells [J]. International journal of oncology, 2012, 41(3): 855-61.
HONG C L, YU I S. CD248 Regulates Wnt Signaling in Pericytes to Promote Angiogenesis and Tumor Growth in Lung Cancer [J]. 2022, 82(20): 3734-50.
ARRIGO A, REGUA A T. Tumor Suppressor Candidate 2 (TUSC2): Discovery, Functions, and Cancer Therapy [J]. 2023, 15(9):
REN D, QIN G, ZHAO J, et al. Metformin activates the STING/IRF3/IFN-β pathway by inhibiting AKT phosphorylation in pancreatic cancer [J]. American journal of cancer research, 2020, 10(9): 2851-64.
CHEN X, ZHAO H, CHEN C, et al. The HPA/SDC1 axis promotes invasion and metastasis of pancreatic cancer cells by activating EMT via FGF2 upregulation [J]. Oncology letters, 2020, 19(1): 211-20.
LIU W, DENG L, XU A, et al. Identifying a novel IRF3/circUHRF1/miR-1306-5p/ARL4C axis in pancreatic ductal adenocarcinoma progression [J]. Cell cycle (Georgetown, Tex), 2022, 21(4): 392-405.
BERNARDO A R, COSGAYA J M, ARANDA A, et al. Pro-apoptotic signaling induced by Retinoic acid and dsRNA is under the control of Interferon Regulatory Factor-3 in breast cancer cells [J]. 2017, 22(7): 920-32.
KOU Y Q, YANG Y P, PAN Z J, et al. Prognostic-Related Biomarkers in Pancreatic Ductal Adenocarcinoma Correlating with Immune Infiltrates Based on Proteomics [J]. Medical science monitor : international medical journal of experimental and clinical research, 2023, 29(e938785.
SAKAI S, IWATA C, TANAKA H Y, et al. Increased fibrosis and impaired intratumoral accumulation of macromolecules in a murine model of pancreatic cancer co-administered with FGF-2 [J]. Journal of controlled release : official journal of the Controlled Release Society, 2016, 230(109-15.
CORBETT A H. Post-transcriptional regulation of gene expression and human disease [J]. Current opinion in cell biology, 2018, 52(96-104.
VERNUCCI E, ABREGO J, GUNDA V, et al. Metabolic Alterations in Pancreatic Cancer Progression [J]. 2019, 12(1):
LARUE M M, PARKER S. Metabolic reprogramming of tumor-associated macrophages by collagen turnover promotes fibrosis in pancreatic cancer [J]. 2022, 119(16): e2119168119.

No competing interests reported.

SupplementaryTables.xlsx

Download PDF

Editorial decision: Revision requested
29 Mar, 2024
Editor assigned by journal
28 Mar, 2024
Submission checks completed at journal
28 Mar, 2024
First submitted to journal
28 Mar, 2024

You are reading this latest preprint version

Exploring Potential Drug Targets for Pancreatic Cancer Based on Mendelian Randomization

Status:

Version 1

Abstract

Objective

Methods

Results

Conclusion

Figures

Introduction

Materials and Methods

Study Design

Data Source

Instrumental Variable SNP Selection

Mendelian Randomization

Sensitivity Analysis

Enrichment Analysis

Protein-Protein Interaction Network Construction

Statistical Analysis

Instrument Variable Selection

MR Results

Sensitivity Analysis

GO and KEGG Enrichment Analysis

PPI Network Analysis

Discussion

Declarations

References

Additional Declarations

Supplementary Files

Status:

Version 1