A causal relationship between chronic pancreatitis and cardiovascular disease: A two-sample Mendelian randomization analysis.

This 2-sample MR study is the 1st comprehensive analysis of possible causal relationships between CP and various CVD events. It therefore serves as a valuable reference for the screening and prevention of cardiovascular risk in patients with CP. MR analysis minimizes any potential confounding biases and reverse causation, while ensuring the stability of results through sensitivity analysis. Regarding horizontal pleiotropy, though the MR-Egger intercept test showed no significant directional bias, we confirmed that none of the selected SNPs were located at IL6R or C-reactive protein loci – key shared risk regions for inflammation and CVD. Strict quality control (LD clumping R 2  = 0.001; F statistic >10) further mitigated pleiotropic risks. [ 21 , 24 ] This study is based on a European population and may not be applicable to other populations. Genetic heterogeneity across ethnicities may lead to divergent effect estimates in non-European cohorts, highlighting the need for multiethnic validation in future research. Additionally, the null results for CES and SVS do not undermine the causal link between CP and ischemic stroke; instead, these findings reflect subtype-specific causality – CP modulates large artery atherosclerosis (the basis of LAS) but not the cardiac/microvascular pathways underlying CES/SVS. [ 33 ] Our study employed summary data from GWAS rather than individual-level data, and hence it was not possible to carry out stratified analysis based on factors such as age and gender.

In conclusion, this MR study suggests that alcohol-induced CP is significantly and causally associated with an increased risk of myocardial infarction and LAS. Moreover, our study suggests that CP may be causally associated with stroke and ischemic stroke.

Chronic pancreatitis (CP) is a multifactorial fibroinflammatory syndrome caused by repeated bouts of pancreatic inflammation that lead to extensive fibrotic tissue substitution. [ 1 ] This results in enduring pain, deficiencies in exocrine and endocrine pancreatic functions, diminished quality of life, and reduced life expectancy. Longer-term complications include pancreatic cancer and diabetes mellitus. [ 2 ] Early diagnosis of CP was formerly quite difficult, but the advent of new diagnostic techniques has made early diagnosis increasingly accurate. Epidemiological studies have shown an increasing incidence of CP in all countries, with an associated increase in the health management and social costs of this condition. [ 3 ] The relationship between CP and several chronic diseases has also attracted the attention of researchers. Cardiovascular disease (CVD) is a significant global health burden, the leading cause of mortality worldwide, and a major contributor to disability. [ 4 , 5 ] Limited observational evidence suggests that individuals with CP have an elevated risk of CVD compared to the general population. Results from multiple cohort studies indicate that CP constitutes a risk factor for cerebrovascular disease. [ 6 , 7 ] The relationship between CP and different types of cerebrovascular disease was further analyzed by Wong et al, who found significant associations of CP with ischemic stroke and hemorrhagic stroke. [ 8 ] However, the conclusion regarding the association between CP and CVD is not unanimous, [ 7 ] with the factors of inflammation, oxidative stress, exocrine insufficiency and diabetes also thought to be associated with CVD. [ 9 – 12 ] In summary, the research evidence regarding the correlation between CP and CVD is quite limited, and observational studies are unable to establish causation. Further verification is therefore required to establish a causal relationship between CP and the risk of CVD. Mendelian randomization (MR) analysis has been extensively employed to estimate causal relationships between risk factors and disease outcomes. It does this by utilizing genetic variants such as single-nucleotide polymorphisms (SNPs) as instrumental variables (IVs). In contrast to observational studies, MR offers an alternative estimation technique. With MR analysis, SNPs are randomly assigned during conception and are unaffected by confounding environmental factors. This approach can bolster causal inference by mitigating significant confounding bias and avoiding reverse causality. The aim of this study was therefore to explore the causal relationship between CP and CVD using a 2-sample MR analysis. This should provide a reference for long-term prognosis screening and management of CP.

Conceptualization: Shien Shen, Liang Zhao, Xiaona Shao, Jianwei Shen. Data curation: Shien Shen, Yuenan Zhu. Formal analysis: Shien Shen. Funding acquisition: Liang Zhao, Xiaona Shao, Jianwei Shen. Investigation: Shien Shen, Jieqiong Lin, Nuonan Yang, Yuning Huang, Ruiwei Shen. Methodology: Shien Shen, Jieqiong Lin, Nuonan Yang, Yuning Huang, Ruiwei Shen. Software: Shien Shen, Yuenan Zhu. Supervision: Liang Zhao, Xiaona Shao, Jianwei Shen. Validation: Jieqiong Lin, Nuonan Yang, Yuning Huang, Ruiwei Shen. Writing – original draft: Shien Shen, Yuenan Zhu. Writing – review & editing: Shien Shen, Yuenan Zhu, Jieqiong Lin, Liang Zhao, Xiaona Shao, Nuonan Yang, Yuning Huang, Ruiwei Shen, Jianwei Shen.

A schematic representation of the research design is presented in Figure 1 . The focus of this MR study was to investigate the causal impact of genetically-predicted CP on the susceptibility to CVD. Our principal analysis was designed to fulfill the following criteria for valid genetic variants: reliable and robust association with the exposure, independent of any risk factor-outcome confounders, and influence the outcome only via the exposure. The workflow of Mendelian randomization analysis reveals causality between CP and CVD. It underscores 3 fundamental assumptions of a valid genetic instrument for MR analysis. The dotted line and the “×” symbolize variables that become unreliable once correlated with the outcome or potential confounders. CP = chronic pancreatitis, CVD = cardiovascular disease, IVW = inverse variance weighted, MR = Mendelian randomization, SNPs = single-nucleotide polymorphisms. Databases for publicly available genome-wide association studies (GWAS) were searched to obtain eligible datasets of exposure and outcomes in the Integrative Epidemiology Unit open GWAS ( https://gwas.mrcieu.ac.uk ). No further ethical approvals were necessary. Given that population confounding can result in biased estimates, we restricted the genetic background of the MR study population to individuals of European descent. A summary of the data sources for CP and for each type of CVD is presented in Table 1 . Detailed information on included traits in this study. SNP = single-nucleotide polymorphism. The summary statistics for CP were obtained from 2 publicly available GWAS datasets (GWAS ID: finn-b-ALCOPANCCHRON and finn-b-K11_CHRONPANC) conducted by Finn Gen. The dataset finn-b-ALCOPANCCHRON included 218,792 Europeans (977 cases and 217,815 controls) with 16,380,466 SNPs. The dataset finn-b-K11_CHRONPANC included 196,881 Europeans (1737 cases and 195,144 controls) with 16,380,413 SNPs. A detailed description of the data sources for each type of CVD is presented in Table 1 . The 11 types of CVD trait were coronary artery disease, angina pectoris, myocardial infarction, heart failure, stroke, intracerebral hemorrhage, venous thromboembolism, ischemic stroke, and 3 subtypes of stroke: large artery atherosclerosis stroke (LAS), small-vessel stroke (SVS), cardioembolic stroke (CES). The GWAS database was searched for SNP selection based on the above assumptions. To mitigate linkage disequilibrium (LD), all SNPs were clumped within a stringent clump window (proportion of variance [ R 2 ] = 0.001, kb = 10,000). When the threshold was set as P  < 5 × 10 −8 , the minimum requirement for studies with at least 10 eligible IVs could not be met. [ 13 ] SNPs with a relatively relaxed and previously used instrument threshold of P  < 5 × 10 −6 were selected to ensure a more comprehensive outcome. [ 14 ] In cases where genetic instruments linked to the exposure factor were absent in the outcome GWAS dataset, proxy SNPs exhibiting a high level of LD ( R 2  > 0.8) in tandem with the summary statistics from the outcome GWAS dataset were adopted for the causal estimation process. If no proxy SNP was found with this framework, the unavailable SNP instrument was abandoned. To address potential weak instrument bias, SNPs with an F statistic >10 were deemed sufficiently robust for our analysis. [ 15 ] Furthermore, the R 2 explained by the SNPs was utilized to validate the strength of the instruments pertaining to the exposure factors. The F statistic and R 2 were computed using established methodologies as a means to assess the potency of the eligible SNPs. [ 16 ] Details of the SNPs are listed in Table S1 , Supplemental Digital Content 1. Two-sample MR analysis was performed using R software (version 4.3.1, R Foundation for Statistical Computing) with TwoSampleMR (version 0.5.10) and mr.raps (version 0.2) packages. The classic inverse variance weighted model (IVW) was employed in the primary MR analyses. [ 17 ] In the absence of directional pleiotropy, the IVW method can provide a stable and accurate causal assessment by leveraging a meta-analytic approach to combine Wald estimates for each IV. [ 18 , 19 ] However, this approach may be influenced by horizontal pleiotropy. [ 20 ] To address this, we employed multiple MR analyses as complementary methods, such as the weighted median and MR-Egger, to assess the robustness and reliability of the MR estimates. [ 21 , 22 ] The weighted median method can provide a robust result when >50% of the weights originate from invalid IVs, and reduce the type I error to calculate a more accurate causal association in the presence of horizontal pleiotropy. [ 21 ] The MR-Egger method provides a relatively robust estimate independently of the validity of IVs, and an adjusted result in the presence of horizontal pleiotropy through the regression slope and intercept. [ 22 ] However, both the weighted median and MR-Egger methods have reduced power compared to the IVW method, as evidenced by wider confidence intervals (CIs). [ 23 ] Therefore, in the present study they were both considered to be supplementary methods. The Bonferroni correction was applied to avoid false-positive results due to multiple testing. Hence, statistical significance was considered to be 0.05/(2 exposure × 11 outcome) = 0.0027. A P value of <.05 but greater than the Bonferroni corrected statistical significance value was defined as being suggestive of a causal association. The heterogeneity between IVs was assessed by Cochrane Q statistic. Significant heterogeneity was indicated by P  < .05, in which case a random-effects model was employed in the subsequent analyses. Otherwise, a fixed-effect model was employed. [ 24 ] For the robustness of results, the values from the 2 models were calculated separately. SNPs showing significant heterogeneity were excluded, and the MR estimates were then reassessed. The MR-Egger method operates under the assumption of no intercept term in the model, thus effectively considering the intercept to be zero. If P  > .05 in the MR-Egger intercept test, this could provide evidence for the absence of pleiotropic bias. [ 22 ] The leave-one-out sensitivity test was used to assess the stability of the MR results by systematically excluding 1 IV at a time. [ 25 ] MR results were reported as the OR (odds ratio) together with the corresponding 95% CI, thereby reflecting the risk associated with unit changes in exposure.

We investigated the associations of alcohol-induced CP and CP with coronary artery disease, angina pectoris, myocardial infarction, heart failure, stroke, intracerebral hemorrhage, venous thromboembolism, stroke, and 3 subtypes of stroke (LAS, CES, and SVS). The primary results were obtained with the IVW method. Figure 2 presents the forest plots for associations between CP and CVD. For alcohol-induced CP, the primary results indicated significant associations with myocardial infarction (OR = 1.001, 95% CI: 1.000–1.002, P  = .0010) and with LAS (OR = 1.067, 95% CI: 1.030–1.105, P  = .0003). For CP, the primary analysis suggested significant associations with stroke (OR = 1.040, 95% CI: 1.015–1.065, P  = .0015) and with ischemic stroke (OR = 1.041, 95% CI: 1.005–1.078, P  = .0024). No significant associations between CP and any ischemic stroke subtype were observed in the primary analysis. Forest plot of Mendelian randomization results. (A) Mendelian randomization associations of alcohol-induced CP with CVD and (B) Mendelian randomization associations of CP with CVD. CI = confidence interval, CP = chronic pancreatitis, CVD = cardiovascular disease, IVW = inverse variance weighted, MR = Mendelian randomization, OR = odds ratio. Cochrane Q test revealed no evidence of heterogeneity, and hence a fixed-effects model was adopted for the primary MR analysis (Table 2 ). No directional pleiotropy was found in the MR-Egger regression (Table 2 ). Sensitivity analysis of MR analysis. IVW = inverse variance weighted, MR = Mendelian randomization. The MR-Egger analysis revealed a significant association between alcohol-induced CP and coronary artery disease (OR = 1.049, 95% CI: 0.991–1.110, P  = .047). However, this result was considered unreliable due to the wide CI. The weighted median showed a significant association between CP and stroke (OR = 1.056, 95% CI: 1.011–1.102, P  = .0134). This result was replicated with the IVW method. In line with the primary MR analysis, the leave-one-out analysis indicated the main causal associations were not directly driven by SNPs (Fig. 3 ). Leave-one-out sensitivity tests. Calculate the MR results of the remaining IVs after removing the IVs 1 by 1. (A) MR effect size for alcohol-induced CP on large artery atherosclerosis stroke; (B) MR effect size for alcohol-induced CP on myocardial infarction; (C) MR effect size for CP on stroke; and (D) MR effect size for CP on ischemic stroke. CP = chronic pancreatitis, IVs = instrumental variables, MR = Mendelian randomization.

To the best of our knowledge, this is the 1st study that leverages MR to comprehensively investigate associations between CP and CVD. The results show causal associations of alcohol-induced CP with myocardial infarction and LAS. Furthermore, the results suggest associations of CP with the occurrence of stroke and ischemic stroke. The statistically significant but numerically small OR (1.001) for alcohol-induced CP and myocardial infarction should be interpreted by the inherent nature of MR analysis: this approach quantifies the effect of genetically-predicted CP susceptibility, rather than clinically established CP. [ 26 ] Such modest genetic effects are common for complex traits like CVD. [ 27 ] While the individual-level impact is limited, the high population prevalence of alcohol consumption and alcohol-related CP renders this effect clinically meaningful at the population scale. CP is an inflammatory disease of the pancreas and is often linked with excessive alcohol consumption. [ 28 ] Evidence from both in vivo and in vitro studies indicates that the damaging effects of alcohol on the pancreas are due to direct toxicity caused by metabolites and byproducts of ethanol metabolism, such as reactive oxygen species. [ 29 ] Ethanol may contribute to pancreatitis development through multiple mechanisms. A recent study suggests it hinders digestive enzyme activation in the pancreas. This effect may arise from acinar cell sensitization to pathological stimuli or from cholecystokinin release by duodenal I-cells. [ 30 ] Notably, our study revealed subtype-specific causal effects of CP on stroke, with significant associations restricted to LAS but not CES or SVS. This specificity arises from 3 CP-associated pathways: elevated interleukin-6 (IL-6) from pancreatic inflammation accelerates large artery plaque formation [ 9 , 31 ] ; alcohol-induced dyslipidemia promotes large vessel atherosclerosis [ 29 , 32 ] ; and splanchnic hypoperfusion impairs endothelial function. [ 27 ] In contrast, CES and SVS are driven by cardiac emboli or microvascular lesions, which are independent of CP-related pathophysiology. [ 33 ] This finding underscores the need for targeted cardiovascular risk stratification in alcohol-induced CP patients. A previous study reported that alcohol-related CP was significantly associated with an increased risk of carotid artery atheroma. [ 34 ] In addition, Bang et al found that alcohol-induced CP was not associated with a higher risk of cancer or death compared to non-alcohol-induced CP. [ 7 ] Khan et al concluded that CP could increase the prevalence and OR of myocardial infarction. [ 32 ] However, few studies have investigated whether there is a causal association between alcohol-induced CP and CVD. To our knowledge, the present study is the 1st that utilizes the MR design to evaluate the causal relationship between alcohol-induced CP and CVD. Furthermore, results from the IVW method suggest that alcohol-induced CP might be associated with myocardial infarction and LAS. Several mechanisms may partially explain these associations. First, dyslipidemia mediates the link between alcohol-induced CP and atherosclerosis. Second, prolonged splanchnic hypoperfusion impairs vascular endothelial function. Third, subclinical inflammation exacerbates vascular damage. [ 27 , 32 ] Additional research is necessary to confirm these correlations and to gain further insights into the possible underlying mechanisms. The association between CP and CVD has been investigated through observational studies and systematic reviews. [ 6 , 7 , 27 ] A retrospective, population-based cohort study reported that CP was associated with an increased risk of subsequent cerebrovascular disease. [ 8 ] Moreover, the incidence rates of ischemic, hemorrhagic, and other types of cerebrovascular disease were found to be higher in the CP group compared to those in the normal group. Similarly, Sung et al reported that CP was associated with the risk of stroke, including ischemic and other types of stroke. [ 35 ] A cohort propensity-matched study from the USA found that CP patients had a higher risk of ischemic heart disease and cerebrovascular accidents. [ 6 ] Furthermore, patients with CP who also had ischemic heart disease exhibited higher rates of acute coronary syndrome (ACS), cardiac arrest, and mortality. A nationwide cohort analysis demonstrated that CP patients aged <40 years exhibited the highest risk of ACS, with CP emerging as an independent risk factor for ACS. [ 36 ] In the present study, we observed causal associations of CP with stroke and ischemic stroke. However, we did not find any significant associations between CP and 3 subtypes of ischemic stroke. Ischemic stroke is a complex disease. The specific subtypes of ischemic stroke are mainly determined by genetic factors, and their pathogenesis may also be different. [ 33 ] Further studies are necessary to elucidate the complex role of CP in stroke and its subtypes. Previous studies have mostly indicated that CP is associated with an increased risk of CVD, independent of other major risk factors. [ 11 ] Our study demonstrated a robust association between CP, myocardial infarction, and stroke. For associations reported in other studies, indirect links may be explained by underlying metabolic factors and potential mechanisms. Patients with CP typically experience pancreatic exocrine insufficiency and pancreatogenic diabetes (type 3c diabetes), [ 12 , 27 ] both of which might increase the risk of CVD. Furthermore, several essential mechanisms can influence the development of CVD, including inflammation, oxidative stress, and insulin resistance. [ 9 , 10 ] Previous research found elevated levels of IL-6 in patients with CP. [ 9 ] As a deteriorative adipokine, IL-6 has the potential to exacerbate arteriosclerosis, thereby contributing to CVD. [ 31 ] A recent MR study also reported that IL-6 was associated with incident CVD. [ 26 ] In addition, CP and CVD might share some common risk factors, such as chronic inflammation and dyslipidemia. Further research is necessary to comprehensively explore the inherent connections between CP and CVD. The results of such studies could provide novel insights into CVD prevention strategies.

We are grateful to investigators from the UK Biobank, the Finn Gen and the MRC IEU Open GWAS Project for providing publicly available data.