Exploring the genetic link between gastroesophageal reflux disease and pancreatic cancer: insights from Mendelian randomization

Abstract

Background

Gastroesophageal reflux disease (GERD) is increasingly recognized for its associations with extragastric diseases, yet its potential role in pancreatic cancer (PC) etiology remains underexplored. This study investigates the genetic causal relationship between GERD and PC using Mendelian randomization (MR), a method designed to reduce confounding factors.

Methods

A two-sample MR analysis was conducted using genome-wide association studies (GWAS) data. The inverse variance weighted (IVW) method was applied, with additional sensitivity analyses performed to evaluate pleiotropy and heterogeneity.

Results

The IVW analysis demonstrated a significant genetic association between the genetic signature predisposing to GERD and an increased risk of PC (OR: 1.36, 95% CI: 1.04–1.80, P = 0.03). There was no evidence of pleiotropy (P = 0.71) or heterogeneity (P = 0.94).

Conclusions

Our study provides robust genetic evidence supporting that the genetic predisposition to GERD is associated with an increased risk of PC. These findings emphasize the necessity of integrating GERD into PC risk assessments and encourage further research to elucidate the underlying biological mechanisms. This insight holds potential to inform strategies for early detection, prevention, and personalized management of PC in GERD patients.

Supplementary Information

The online version contains supplementary material available at 10.1186/s12885-025-14128-6.

Introduction

Pancreatic cancer (PC) is a highly aggressive malignancy characterized by a poor prognosis, largely due to its asymptomatic nature in the early stages and the subsequent late diagnosis. This makes PC one of the leading causes of cancer-related mortality worldwide [1]. Although advancements in treatment have been made, survival rates remain dismal, emphasizing the need for early detection and prevention strategies. As the global incidence of PC continues to rise, there is an urgent need to identify modifiable risk factors that could facilitate earlier detection and improve patient outcomes [2].

Gastroesophageal reflux disease (GERD) is a common gastrointestinal disorder that has been primarily associated with esophageal complications, including Barrett’s esophagus and esophageal adenocarcinoma [3]. The pathophysiological mechanisms of GERD, such as chronic inflammation, repeated exposure of the esophageal lining to acidic and bile contents, and subsequent DNA damage, are well documented as contributors to carcinogenesis in the esophagus [4–7]. These mechanisms suggest that GERD could potentially influence the risk of other cancers within the gastrointestinal tract, including pancreatic cancer [8, 9]. Chronic inflammation is a recognized factor in carcinogenesis, and the potential for GERD to contribute to an inflammatory environment that may extend beyond the esophagus provides a biological rationale for investigating a link to PC [4, 8, 10].

However, the current understanding of the GERD-PC relationship remains limited. Despite the plausible biological mechanisms, the relationship between GERD and PC remains underexplored. Existing studies on this potential association are limited and have yielded inconsistent results, possibly due to challenges in controlling for confounding factors and establishing a clear temporal relationship between GERD and PC [8, 11–15]. This inconsistency highlights a significant gap in the literature regarding whether GERD contributes to an increased risk of PC. Therefore, there is a need for further investigation using robust methodologies that can provide more definitive evidence. Mendelian randomization (MR) offers a powerful approach to addressing these challenges [16]. By utilizing genetic variants as instrumental variables, MR can infer causality between exposure factors and disease outcomes while minimizing the biases of confounding and reverse causation that commonly affect observational studies. In this study, we employed a two-sample MR approach, utilizing genome-wide association study (GWAS) data, to investigate the potential causal relationship between GERD and PC [17].

This study aims to explicitly fill the existing gap in the literature by providing novel insights into the genetic underpinnings of the GERD-PC relationship. Understanding whether GERD causally influences PC risk could have significant implications for the early detection and prevention of PC, ultimately guiding clinical practice and public health strategies. Specifically, we aim to investigate the genetic causal relationship between GERD and PC using Mendelian Randomization.

Methods

Study design and data sources

This study utilized a two-sample MR approach to investigate the potential causal relationship between GERD and PC. The two-sample MR method was chosen due to its ability to enhance the robustness and precision of causal estimates by leveraging independent data from distinct GWAS. Compared to a single-sample MR approach, this method minimizes biases related to sample overlap and confounding, which may be present in single-sample MR studies. Furthermore, the two-sample MR design allows for better control of pleiotropy and reduces the risk of bias associated with weak instrument variables. In this study, GERD was treated as the exposure variable, and PC as the outcome variable. Instrumental variables (IVs) were selected based on their strong association with GERD, ensuring that the MR analysis met the three core assumptions (illustrated in Fig. 1):Relevance: The selected IVs are strongly linked to GERD, with P values less than 5 × 10^-8 and F-statistics greater than 10 (Fig. 1, A). Independence: The IVs are not associated with potential confounding factors (Fig. 1, B). Exclusion Restriction: The IVs influence PC solely through their effect on GERD, without involvement in other pathways (Fig. 1, C). The GWAS summary data for GERD were obtained from the UK Biobank, which included 129,080 GERD patients and 473,524 control subjects, all of European descent [18]. The GWAS data for PC were sourced from the MRC IEU OpenGWAS project, comprising 1,196 nonhereditary PC patients and 475,049 controls [19].

Fig. 1.

Key assumptions of Mendelian randomization. A: Demonstrates the relevance of the IV, showing a strong correlation between the IV and the exposure (GERD).B: Ensures the independence of the IV, indicating that the IV is not associated with any confounding factors.C: Confirms the exclusion restriction, asserting that the IV influences the outcome (PC) solely through the exposure (GERD)

Instrumental variable selection

Selection of IVs is critical to the validity of MR analysis. Initially, SNPs significantly associated with GERD (P < 5 × 10^-8) were identified from the GWAS data. To ensure the statistical strength of the IVs, SNPs with an F-statistic less than 10 were excluded. The F-statistic was calculated using the formula F = R^2 × (N-2)/(1-R^2), where R^2 represents the proportion of variance in GERD explained by each SNP.

To address potential linkage disequilibrium (LD) among SNPs, a clumping procedure was employed (r^2 < 0.001, physical window = 10,000 kb) to ensure the independence of IVs. Furthermore, SNPs associated with known confounders, such as smoking and obesity, were excluded using PhenoScanner V2 [20]. The final step involved harmonizing the exposure and outcome datasets by removing palindromic SNPs with intermediate allele frequencies, ensuring consistency in the effect alleles between GERD and PC.

.

Statistical analysis

Multiple Mendelian randomization (MR) methods were utilized to investigate the genetic correlation between GERD and PC. These methods included MR‒Egger regression, weighted median, inverse variance weighted (IVW), simple mode, and weighted mode methods. The IVW method, which assumes that all SNPs used in the analysis are valid instrumental variables (IVs), was selected as the primary analytical approach due to its potential for providing the most precise estimates when this assumption holds true. MR‒Egger, in contrast, can provide consistent estimates even when some IVs are invalid but is sensitive to horizontal pleiotropy. The weighted median approach offers robustness as it provides accurate estimates even if up to 50% of the instruments are invalid. Simple and weighted mode methods further serve to identify valid instruments based on clustering IVs with similar effects, offering protection against pleiotropy.

To ascertain the robustness of the results, several tests were conducted. Cochran’s Q test was applied to detect heterogeneity across the SNPs, which would indicate variability beyond that expected under the IV assumptions, while funnel plots were used to examine the symmetry of the analysis, serving as a visual check for potential bias. Additionally, the MR‒Egger intercept test and the MR pleiotropy residual sum and outlier (MR-PRESSO) global test were employed to identify pleiotropy and outliers. Notably, MR-PRESSO also allows for the recalculation of estimates after excluding outliers, thus improving the reliability of the results. A leave-one-out sensitivity analysis was also conducted to evaluate the influence of individual SNPs on the overall results. This involved sequentially removing SNPs to observe the cumulative effect of the remaining SNPs, thereby determining whether the overall estimate was disproportionately influenced by any single SNP.

All the statistical analyses were performed using R software (version 4.3.1) and the TwoSampleMR package. A P-value of less than 0.05 was considered to indicate statistical significance.

Results

Instrumental variable selection

During the IV selection process, 80 single-nucleotide polymorphisms (SNPs) were initially identified due to their strong correlation with GERD, as evidenced by P values less than 5 × 10^-8 and F-statistics exceeding 10. These SNPs were further confirmed to be mutually independent, characterized by r² values less than 0.001 and physical distances less than 10,000 kb. To refine this selection, the PhenoScanner tool was utilized to eliminate SNPs that might be confounded by associations with other conditions, such as smoking or obesity, which could influence the risk of PC. After this refinement, 76 SNPs were finalized as instrumental variables for the subsequent analysis, as detailed in Supplementary Table 1. To enhance the clarity and accessibility of the SNP selection process, a flow diagram (Fig. 2) was included to visually represent the steps involved in selecting and refining the IVs. This diagram outlines the criteria used for SNP selection, the application of LD clumping, and the exclusion of confounding SNPs, providing a clear overview of the rigorous process employed to ensure the validity of the IVs.

Fig. 2.

Flowchart of SNP selection process for instrumental variables. This flowchart illustrates the systematic process used to select single-nucleotide polymorphisms (SNPs) as IVs for the Mendelian randomization analysis. The steps include initial SNP selection based on a strong association with GERD (P < 5 × 10⁻⁸, F > 10), ensuring independence between SNPs (r² < 0.001, distance < 10,000 kb), and exclusion of confounding SNPs using PhenoScanner. The final selection resulted in 76 SNPs, which were used in the subsequent analysis

Mendelian randomization analysis

In the analysis investigating the genetic link between GERD and pancreatic cancer (PC) using the IVW method as the primary approach, a statistically significant association was identified between GERD and an increased risk of PC, with an odds ratio (OR) of 1.36 (95% CI: 1.04 to 1.80, P = 0.03).

This finding suggests that the genetic signature predisposing to GERD moderately elevates the risk of PC (Table 1). Other Mendelian randomization (MR) methods were also employed in the analysis. Although these additional methods did not yield statistically significant results, they consistently indicated a positive correlation between GERD and PC, as reflected by odds ratios greater than 1 (Table 1).

Table 1.

Results of the Mendelian randomization analysis using different methods

Exposure	Outcome	Method	SNP	OR	OR95%CI		P value
GERD	PC	MR Egger	76	1.83	0.37	9.09	0.46
GERD	PC	Weighted median	76	1.28	0.86	1.91	0.22
GERD	PC	IVW	76	1.36	1.04	1.80	0.03
GERD	PC	Simple mode	76	1.27	0.49	3.26	0.62
GERD	PC	Weighted mode	76	1.18	0.47	2.97	0.73

This table summarizes the outcomes of the Mendelian randomization analysis, comparing the genetic correlation between GERD and PC across five distinct methods: MR‒Egger regression, weighted median, IVW method, simple mode, and weighted mode. The odds ratios (OR), 95% confidence intervals (CI), and P values are provided for each method, highlighting the statistically significant association identified by the IVW method (OR = 1.36, 95% CI: 1.04 to 1.80, P = 0.03)

To aid in the interpretation of these results, Fig. 3A presents a forest plot delineating the estimated causal relationships between GERD and the risk of PC, as inferred from each instrumental variable (IV) utilized in the analysis. The aggregate estimate derived through the IVW method is distinctly emphasized. Additionally, Fig. 3B illustrates a scatter plot showing the relationship between genetic variants linked to GERD and their respective impacts on the risk of PC. The slope of the line represents the overall estimated causal effect, as determined by the MR analysis.

Fig. 3.

Causal Relationship between GERD and PC based on mendelian randomization analysis. (A) Forest plot showing the estimated causal relationships between GERD and PC as inferred from each IV used in the analysis. The overall estimate from the IVW method is emphasized. (B) Scatter plot depicting the relationship between genetic variants associated with GERD and their respective impacts on PC risk. The slope of the line represents the overall estimated causal effect as determined by the MR analysis

Heterogeneity and sensitivity analysis

Heterogeneity, or variability in causal estimates from different SNPs, was assessed using Cochran’s Q test (P = 0.94), indicating no significant heterogeneity (Table 2). Figure 4 (Supplementary Material) presents the funnel plot analysis, which shows no evidence of bias or asymmetry.

Table 2.

Heterogeneity analysis of the Mendelian randomization study investigating GERD and PC

Exposure	Outcome	Method	Q	Q_df	P value
GERD	PC	MR Egger	56.51	74	0.93
GERD	PC	IVW	56.65	75	0.94

This table presents the outcomes of the heterogeneity analysis conducted using Cochran’s Q test. The Q statistic, degrees of freedom (df), and P values for both the MR‒Egger regression and IVW methods are provided. The lack of significant heterogeneity (IVW Q = 56.65, P = 0.94) suggests consistent causal estimates across the SNPs used as instruments

Pleiotropy was evaluated via MR‒Egger regression and the MR-PRESSO global test. Both analyses showed no substantial horizontal pleiotropy (MR-Egger intercept P = 0.71), confirming that the IVs do not likely influence the outcome through alternative pathways (Table 3).

Table 3.

Evaluation of Pleiotropy in the Mendelian randomization analysis

Exposure	Outcome	Egger Intercept	SE	P value
GERD	PC	-0.01	0.03	0.71

This table details the assessment of pleiotropy within the MR analysis of GERD and PC. The MR‒Egger intercept, standard error (SE), and P value are reported. The non-significant P value (P = 0.71) indicates an absence of substantial horizontal pleiotropy, supporting the validity of the selected instrumental variables

Robustness was confirmed with a leave-one-out sensitivity analysis Fig. 5 (Supplementary Material). Results remained consistent, demonstrating that no single SNP unduly influenced the overall causal estimate, further supporting the reliability of the findings.

Discussion

In this study, we utilized a two-sample MR approach to investigate the potential causal relationship between GERD and the risk of PC. By leveraging genetic variants as IVs, we sought to minimize the biases inherent in traditional observational studies, such as confounding factors and reverse causation. Our findings suggest a possible causal link between GERD and an elevated risk of PC, contributing new insights into the complex interplay between these conditions.

Our results align with, and expand upon, previous research that has explored the relationship between GERD and other gastrointestinal cancers [21–25]. For instance, GERD is a well-established risk factor for esophageal adenocarcinoma, driven by chronic inflammation and acid and bile reflux [4, 8, 15, 21, 26–30]. Similar mechanisms may explain the association between GERD and PC, with persistent bile reflux potentially contributing to pancreatic inflammation and carcinogenesis [8, 31]. Previous observational studies suggested a link between prolonged proton pump inhibitors use for GERD and increased PC risk, but were limited by confounding factors [13, 32–37]. Our MR analysis, by addressing these limitations, provides more robust evidence suggesting a direct genetic link between GERD and PC.

The biological mechanisms by which the genetic signature predisposing to GERD might increase the risk of PC are multifaceted and warrant further exploration. One proposed mechanism involves bile reflux, where GERD induces the regurgitation of bile and other digestive fluids into the esophagus and stomach [15]. These agents may reach the pancreas, causing chronic inflammation, a well-established risk factor for cancer development [21, 38–43]. Chronic inflammation can lead to DNA damage and promote a pro-carcinogenic environment in pancreatic tissues [44–48]. Additionally, GERD-associated chronic inflammation could result in increased cellular proliferation, potentially extending to the pancreatic ducts and enhancing the risk of malignancy [47, 49–51]. Another plausible mechanism is the alteration of pancreatic digestive enzymes due to GERD, which could modify the pancreatic microenvironment and further increase susceptibility to cancer [52, 53]. Expanding upon these mechanisms presents a critical avenue for future research. Investigations should focus on elucidating the specific molecular pathways through which bile reflux and chronic inflammation facilitate carcinogenesis in pancreatic tissues. Additionally, studying the impact of GERD on the expression and activity of pancreatic enzymes may offer insights into how this disease modulates the pancreatic microenvironment, potentially revealing novel biomarkers or therapeutic targets. Such research could provide deeper insights into the pathophysiology of GERD-associated pancreatic carcinogenesis, ultimately paving the way for targeted interventions aimed at preventing or mitigating the progression of pancreatic cancer.

While our study offers important insights, several limitations must be acknowledged, as they could influence the findings and their interpretation [49]. First, the use of GWAS data from populations of predominantly European ancestry limits the generalizability of our results to other ethnic groups. This demographic restriction highlights the need for future studies to include more diverse populations, which would help verify the broader applicability of our findings [54]. Additionally, population stratification, which can lead to confounding due to differences in allele frequencies across populations, may also impact the validity of the genetic associations, particularly in studies involving ancestrally homogeneous groups. Future research should aim to incorporate diverse populations to reduce this bias. Second, although the MR framework effectively minimizes confounding and reverse causation, it is still subject to limitations. For example, residual pleiotropy, where genetic variants affect the outcome through pathways unrelated to the exposure, cannot be entirely ruled out, even though we applied MR‒Egger and MR-PRESSO tests to detect and correct for such biases. However, the potential for undetected pleiotropy remains, which may still affect our causal estimates despite the application of these sensitivity analyses. Additionally, residual confounding due to unmeasured or poorly measured environmental factors might also influence the results. Finally, the reliance on summary-level data from preexisting GWAS meta-analyses precluded more granular analyses specific to different age groups or regions [55]. These factors could potentially impact the accuracy and applicability of the causal estimates derived from our study. While COLOC analysis was not feasible due to the absence of colocalized causal variants, future studies should integrate multi-omics datasets (e.g., eQTLs, proteomics) to identify shared pathways between GERD and PC. The causal link between the genetic signature predisposing to GERD and PC risk highlights the need for future research into shared molecular pathways.

The potential causal relationship between GERD and PC identified in this study has important clinical implications. If GERD indeed increases the risk of PC, as our findings suggest, this could influence specific screening and prevention strategies for PC. Clinicians may need to consider more rigorous monitoring of patients with GERD, particularly those with chronic or severe symptoms, to detect early signs of pancreatic abnormalities. For instance, this could include initiating earlier or more frequent imaging studies, such as endoscopic ultrasound or MRI, in high-risk patients. Additionally, patients with GERD who present with atypical or persistent gastrointestinal symptoms could benefit from a more tailored diagnostic approach to rule out early pancreatic changes. Furthermore, these findings underscore the importance of effective GERD management, not only to alleviate symptoms but also potentially to reduce the risk of developing PC. Future research should focus on validating these findings in more diverse populations and exploring the impact of GERD management strategies, such as PPI use, on PC risk. Integrating biomarkers and clinical data could also enhance our understanding of the biological mechanisms underlying the GERD-PC relationship, leading to more targeted and effective preventive measures. Incorporating genetic and epigenetic markers, as well as inflammatory profiles, may allow for a more personalized approach to patient risk stratification. Furthermore, longitudinal studies that follow GERD patients over time could provide valuable insights into the temporal relationship between GERD and PC development.

Conclusions

This Mendelian randomization study suggests a potential causal link between GERD and an increased risk of PC. These findings contribute to the growing body of evidence identifying GERD as a potential risk factor for pancreatic cancer, expanding the current understanding of PC etiology. Our results highlight the importance of recognizing GERD not only as a common digestive disorder but also as a condition with potential long-term oncogenic risks. Clinically, this underscores the need for heightened vigilance in GERD management, particularly in monitoring for pancreatic abnormalities. Future research should prioritize validation of these findings across more diverse populations, as well as investigating the specific biological mechanisms underlying this association, with the aim of informing more effective prevention and early detection strategies for PC.

Electronic supplementary material

Below is the link to the electronic supplementary material.

Abbreviations

BMI

Body mass index

CI

Confidence intervals

df

Degrees of freedom

GERD

Gastroesophageal reflux disease

GAWS

Genome-wide association studies

IVW

Inverse variance weighted

IVs

Instrumental variables

LD

Linkage disequilibrium

MR

Mendelian randomization

OR

Odds ratio

PC

Pancreatic Cancer

PPIs

Proton pump inhibitors

SNPs

Single nucleotide polymorphisms

Author contributions

CY significantly contributed to the conceptualization, methodology design, original draft writing, and manuscript review and editing. FG played a key role in methodology design, software operation, formal analysis, data curation, and manuscript review and editing. MP was primarily focused on software operation, formal analysis, and data curation. LC was actively engaged in the review and editing of the manuscript. WL and KW was instrumental in conceptualization, supervision, visual representation, and project administration.

Funding

This research is supported by the National Natural Science Foundation of China General Projects (81571740) (KW); Provincial Key Research and Development Program of Heilongjiang Province (GA21C001) (KW); Postdoctoral Special Scientific Research Grant of Heilongjiang Provincial Government (LBH-Q17104) (KW); Distinguished Young Scientist Funding of Harbin Medical University Affiliated Tumor Hospital (JCQN2019-02); and Key Project of the Climbing Funding of the National Cancer Center (NCC201808B019).

Data availability

No datasets were generated or analysed during the current study.

Declarations

Ethics approval and consent to participate

This research utilizes data that is publicly accessible. Every study included in this GWAS received approval from the appropriate institutional review boards. Moreover, informed consent was secured from either the participants directly or, where applicable, their caretakers, legal guardians, or designated representatives.

Consent for publication

This paper has received the careful reading and approval of all listed authors prior to submission. We collectively affirm that the content herein is a product of our original work, devoid of any unauthorized materials, and that all pertinent contributors and resources have been duly acknowledged and cited in the text.

Competing interests

The authors declare no competing interests.