The Role of Inflammatory Biomarkers in Mediating the Effect of Inflammatory Bowel Disease on nonmalignant Digestive System Diseases: A Multivariable Mendelian Randomized Study

Background While observation studies have shown a positive correlation between inflammatory bowel disease (IBD) and the risk of nonmalignant digestive system diseases, a definitive causal relationship has not yet been clearly established. Methods Mendelian randomization (MR) was employed to investigate the potential causal association between genetic susceptibility to IBD and nonmalignant gastrointestinal diseases. Genetic variants were extracted as instrumental variables (IVs) from a genome-wide association study (GWAS) meta-analysis, which included 12,194 cases of Crohn's disease (CD) and 28,072 control cases of European ancestry. The GWAS for ulcerative colitis (UC) included 12,366 UC and 33,609 control cases of European ancestry. All IVs reached genome-wide significance (GWAS p value <5 × 10−8). Summary-level data for acute pancreatitis (AP), irritable bowel syndrome (IBS), gastroesophageal reflux disease, cholelithiasis, and CeD (celiac disease) were obtained from the GWAS meta-analysis and the FinnGen dataset. Summary-level data on relevant inflammatory factors were provided by the International Genetic Consortium. Univariate MR analysis was conducted using inverse variance weighting as the primary method for estimating causal effects. Multivariate MR analyses were also performed to detect possible mediators. Results Genetic susceptibility to UC was associated with an increased risk of AP (OR = 1.08; 95% CI = 1.03–1.13; p=0.002) and IBS odds ratio (OR] = 1.07; 95% confidence interval (CI] = 1.03–1.11; (p < 0.001). In terms of potential mediators, interleukin 6 (IL-6) had a driving effect on the association between UC and AP. There was no apparent evidence of increased risk with CD. Meanwhile, genetic susceptibility to CD increases the risk of CeD (OR = 1.14; 95% CI = 1.03–1.25; p=0.01). Conclusions The evidence suggests that UC is associated with an elevated risk of AP and IBS, and IL-6 may be responsible in AP. CD is associated with an increased risk of developing CeD. Implementing a proactive monitoring program for assessing the risk of gastrointestinal diseases in UC patients, particularly those with elevated IL-6 levels, may be of interest. In addition, the presence of AP and IBS may indicate the presence of UC. Preventing CeD is an essential consideration in the therapeutic management of patients with CD.


Background
Nonmalignant digestive system diseases (NMDSD) encompass a broad category of noncancerous gastrointestinal, hepatobiliary, and pancreatic disorders, including conditions such as irritable bowel syndrome (IBS), gastroesophageal refux disease (GERD), acute pancreatitis (AP), celiac disease (CeD), and cholelithiasis [1].NMDSD imposes a substantial medical and economic burden, and given the severe clinical and societal consequences they entail, it is imperative to elucidate their associated risk factors to facilitate timely treatment and prevention [1].
Several investigations have indicated that individuals with IBD are more susceptible to NMDSD [9], likely owing to the sustained infammatory state and immune activation associated with IBD [10][11][12].A meta-analysis, for instance, found a signifcantly higher prevalence of IBS in patients with IBD compared to controls [13].Meanwhile, a metaanalysis showed that patients with IBD had a 3.96-fold increased risk of CED compared to the general population [14].Furthermore, IBD has been linked to an increased risk of cholelithiasis in a cohort study [15].Moreover, a multicenter study revealed that individuals with IBD were more prone to developing AP [16].However, it is worth noting that many agents used to treat IBD, such as salicylates, azathioprine, 6-mercaptopurine, and glucocorticosteroids, can also induce NMDSD.Te evidence pertaining to the relationship between IBD and the risk of NMDSD is still contradictory and limited.Tus, it is crucial to establish IBD and NMDSD as having a causal relationship.
Mendelian randomization, an emerging technique in medical research, has been used to assess causal relationships between specifc exposures and outcomes by utilizing genetic variations as instrumental variables (IVs) [17].MR is highly efective in mitigating common confounding and reverse causation biases encountered in traditional observational studies [18].As genetic variants are randomly inherited by ofspring during meiosis and conception, MR operates on principles akin to randomized clinical trials (RCTs), which are often considered the gold standard for establishing causal relationships.Consequently, MR has emerged as a pivotal epidemiological method for inferring causality when RCTs are not feasible.In this context, our study aimed to achieve two primary objectives: (1) in this context, our study aimed to achieve two primary objectives; (2) explore the potential links between IBD and clinical infammatory markers, including interleukin-6 (Il-6), Creactive protein (CRP), and tumor necrosis factor-α (TNF-α), employing univariate MR (UVMR).In addition, utilizing multivariate MR (MVMR), we sought to examine whether these infammatory markers acted as potential mediators in the association between IBD and NMDSD.

Study Design.
In order to examine the potential causal link between IBD and prevalent NMDSD, including AP, IBS, GERD, and cholelithiasis, a UVMR analysis was conducted.
Te MR study followed a comprehensive procedure, as illustrated in Figure 1, guided by three fundamental assumptions: a. IVs are closely related to IBD; b.IVs must be independent of potential confounding factors; c.IVs should not be relevant to the NMDSD unless by way of IBD [19].
Our study proceeded in several steps.Initially, we examined the link between genetic susceptibility to IBD and NMDSD using UVMR analysis.To delve into potential mechanisms, we explored the association between IBD and clinically common infammatory biomarkers.Subsequently, we conducted MVMR to investigate the mediating role of infammatory factors in the relationship between genetic susceptibility to IBD and NMDSD.It is important to note that our study exclusively utilized publicly available summary-level data from genome-wide association studies (GWAS), obviating the need for additional ethical permissions.
We performed the analysis of this study according to MR guidelines [20].All analyses were conducted with R software (4.1.2),using the "TwoSampleMR," "MRPRESSO," and "Mr.Rap" packages.

Data Source.
We obtained summary data for each phenotype from various sources, as detailed in Table 1.Summary statistics for IBD were acquired from a GWAS meta-analysis involving 59,957 participants, primarily of European ancestry.Tis dataset included 12,194 CD and 12,366 UC patients [21].All cases in this study were diagnosed through recognized radiological, endoscopic, and histopathological evaluations that met the clinical diagnostic criteria for IBD.

Selection.
We only included single nucleotide polymorphisms (SNPs) that demonstrated genome-wide signifcance (p value <5 × 10 −8 ).SNPs with lower minor allele frequencies (MAF < 0.01) were removed.Further, linkage disequilibrium (LD, r 2 > 0.01) among the selected SNPs was assessed to screen out independent SNPs as IVs.We harmonized summary statistics of exposures and outcomes and excluded palindromic SNPs.Te beta and standard error (SE) coefcients of these SNPs were scaled by the logtransformed odds of IBD.We performed strength assessments for the screened SNPs to avoid weak instrumental 2 Canadian Journal of Gastroenterology and Hepatology bias.If F > 10, it indicated that the association between the IVs and IBD was sufciently robust, and the likelihood of weak instrumental bias afecting the results' reliability was minimized [28].

Statistical Analysis.
For UVMR analysis, the inverse variance weighting (IVW) method was used as the primary analysis method.Te choice of analytical model depended on the presence or absence of heterogeneity in the sensitivity analysis.If heterogeneity existed, the random-efects IVW method was selected, and if there was no heterogeneity, the fxed-efects IVW method was opted [29].
To bolster our causal inference, we also conducted supplementary analyses using various MR models, including the MR-robust adjusted profle score (RAPS) [30], weighted median analysis [31], maximum-likelihood estimation [32], MR-Egger method [33], and MR-pleiotropy residual sum and outlier (MR-PRESSO) [34].Although IVW provides the most accurate estimates, horizontal polymorphism and heterogeneity may afect its accuracy.MR-RAPS is an extension of the IVW method that can increase statistical validity and efectively address the issue of horizontal pleiotropy by including numerous weak IVs [30].In the weighted median analysis, over half of the weights were derived from efective SNPs [31].Te MR-Egger method detects horizontal pleiotropy by means of intercept tests [33].However, for most cases, the statistical power of this analysis is weak.Te MR-PRESSO method reduces the efect of horizontal pleiotropy by identifying and removing outliers [34].Each of these analytical approaches was selected based on its unique assumptions and ability to produce unbiased causal estimates.You can fnd more details on each method in Additional fle 1: Table S1.When examining the association between IBD and NMDSD, a causal relationship was considered signifcant at a threshold of p < 0.0125 (Bonferroni-corrected, 0.05 for 4 outcomes).
Given that horizontal pleiotropy can interfere with MR estimation to some extent, a sensitivity analysis was performed (Table 2).We evaluated heterogeneity using the Cochran's Q-test.Potential heterogeneity was present when p < 0.05.Te MR-Egger intercept was utilized to detect horizontal pleiotropy.Excessive horizontal pleiotropy indicates that this analysis violates the fundamental assumptions of MR analysis, making the results less reliable.To ensure that the results were not driven by strong correlations of individual SNPs, we performed a leave-one-out analysis (Additional fle 2: Figures S1-S3).
In consideration of the potential associations between infammatory factors and NMDSD, we conducted MVMR to explore whether the observed associations were infuenced by clinically common infammatory markers [35].Tis approach allows us to dissect the total efect of IBD on NMDSD, estimated through our UVMR, into two components: the direct efect and the indirect efect mediated by the infammatory markers.To calculate the proportion of the efect mediated by infammatory factors, we divided the direct efect by the total efect and subtracted it from 1. Notably, this method assumes that the medium is a continuous variable to eliminate any bias in the estimation of mediating efects [36].

Results
Tere was an association between susceptibility to UC and an increased risk of IBS and AP.Furthermore, interleukin-6 (IL-6) played a mediating role in these associations.Furthermore, genetic susceptibility to CD disease is associated with a higher risk of CeD.Te F-statistic for all IVs was greater than 10, suggesting that the reliability of the results was unlikely to have been afected by weak instrumental bias (additional fle 1: Table S2).Te summary data for all SNPs used for statistical analysis are shown in additional fle 1 Tables S3-S13.Canadian Journal of Gastroenterology and Hepatology Te causal estimates obtained from the diferent MR statistical models were generally consistent, except for the MR-Egger which produced causal estimates with lower precision.As described previously, in most cases, the MR-Egger method does not provide sufcient statistical power [33].No outliers were found by MR-PRESSO analysis of IBS and AP, further indicating the robustness of the results.Cochran's Q-test revealed no heterogeneity in the analysis of the association of UC with AP, IBS, and GERD; thus, the fxed-efects IVW method was used as the "gold standard" for this analysis.Heterogeneity was detected in the association of UC with cholelithiasis; thus, the random-efects model is considered as the gold standard (Figure 2).MR-Egger intercepts revealed weak horizontal pleiotropy.In addition, the leave-one-out method suggested that the results of causal estimation were not mediated by a single SNP (Additional fle 2: Figures S1-S3).Forest plots and funnel plots were used to visualize the heterogeneity analysis (Additional fle 2: Figures S1-S6).
In terms of the infammatory markers, genetic susceptibility to UC was associated with elevated levels of IL-6 and CRP (Figure 4).Te MR-PRESSO method detected the presence of outliers in the analysis of CRP, and the correlation was altered after removal of these outliers.Terefore, to ensure the reliability of the results, CRP was not included in the subsequent mediation analysis.We detected heterogeneity in the analysis of IL-6 but no horizontal pleiotropy (Table 2).After removal of the outliers identifed by the MR-PRESSO method, the correlations remained consistent.Moreover, multiple MR analysis models yielded consistent analytical results, indicating that this result was sufciently reliable.In addition, we did not fnd a correlation between genetic susceptibility to UC and TNF-α (Figure 4).Tere was no evidence that genetic susceptibility to CD associated with IL-6, CRP, and TNF levels (Figure 5).

Discussion
Our results show CD may increase the risk of CeD.Furthermore, this study is the frst to examine the causal relationship between IBD and specifc types of NMDSD.Our study revealed that UC is associated with a higher risk of AP and IBS.In addition, genetic susceptibility to UC was associated with elevated IL-6 levels.Further MVMR analysis showed that IL-6 had an important driving efect on the association between UC and AP.Tese results shed light on the potential mechanisms underlying the increased risk of AP in individuals with UC and the role of IL-6 in this relationship.Recent years have seen a growing volume of observational studies linking IBD to the risk of NMDSD [37].For example, a Swedish cohort study showed that people with CD were at a 4.36% higher risk of CeD than normal people.[38].Moreover, one observational study found patients with pancreatitis had a 10-fold increased risk of IBD compared with controls [39].A meta-analysis showed that the pooled prevalence of IBS in patients with IBD was signifcantly higher than in the general population [13].However, these studies could not clearly establish causality because the relationships remained largely associative rather than causal.
Our results on UC can be attributed to several potential mechanisms.Firstly, patients aficted by UC often exhibit visceral hypersensitivity, a phenomenon that persists even during periods of remission [40][41][42].Tis heightened visceral sensitivity is recognized as a fundamental pathogenic mechanism underlying IBS [43].Notably, this phenomenon may be linked to an increased release of proinfammatory cytokines [44].Furthermore, individuals with UC tend to have elevated levels of circulating markers, such as IL-6 [45].
According to a study conducted on guinea pigs, it functions as a neuromodulator, potentially exacerbating bowel symptoms in cases of IBS.Moreover, inhibition of the IL-6 receptors can ameliorate IBS-like symptoms by decreasing the expression of the T-type calcium channel Ca v 3.2, which plays a central role in visceral pain [46].In addition, the diminished quality of life often experienced by individuals with UC could conceivably contribute to the development of IBS [47,48].
Recent evidence has shown that there is a robust link between AP and IL-6, which exhibits high expression in UC [49,50].AP is a sudden infammation of the pancreatic tissue, caused by the activation of several infammatory mediators [51].Earlier reports have substantiated that human IL-6 exacerbates cerulein-induced AP in murine models.Furthermore, the signal transducer and activator of transcription-3 (STAT-3) assume a pivotal role in triggering the infammatory response [52].Te application of anti-IL-6 antibodies has shown promise in mitigating severe AP by inhibiting IL-6 and quelling the activation of STAT-3 [51,53].In addition, previous data suggests that IL-6 trans-signaling may play an essential role in AP-related acute lung injury in mouse models [54].In a crosssectional investigation, pancreatic duct abnormalities were found in approximately 16% of cases with UC [55].Tese abnormalities might contribute to the pathogenesis and pain of AP [56] 6 Canadian Journal of Gastroenterology and Hepatology in patients with CD but not in patients with UC or general population [57].Elevated ASCA titers may respond to increased intestinal permeability, and the occurrence of CeD is strongly associated with increased intestinal mucosal permeability defects.Besides, diet cannot be ignored in the development of CeD.In a cohort study, a high-infammatory-potential dietary pattern was shown to be associated with CD but not with UC [58].Tis highinfammatory potential diet in CD patients may increase the risk of CeD.Tis study substantiates the proposition that infammation plays a pivotal role in driving the susceptibility to IB) and AP in individuals with UC, with IL-6 emerging as a mediating factor in these connections.Our fndings advocate for vigilant monitoring of UC in patients diagnosed with AP, particularly those exhibiting heightened IL-6 levels.Such an approach could facilitate the early detection and diagnosis of UC.Furthermore, our results signify the importance of proactively conducting endoscopic examinations in individuals diagnosed with IBS.Tis strategy is paramount in the realm of precision medicine, given the considerable divergence in treatment approaches for UC and IBS in clinical practice [59,60].Although UC remains without a cure, early intervention holds the potential to decelerate disease progression and alleviate symptoms [61].By doing so, we can alleviate the healthcare burden on society and empower patients through timely detection and intervention.
Tis MR study evaluating the association between genetic susceptibility to IBD and NMDSD boasts several signifcant strengths.Firstly, MR analysis's inherent characteristics have efectively mitigated the infuence of confounding variables and reverse causality on the study outcomes.Meanwhile, multiple UVMR analysis models and sensitivity analyses were used to assess whether this study violated MR assumptions.Secondly, the summary-level statistics obtained from large-scale GWASs have substantially bolstered their statistical robustness.Furthermore, the acquisition of data pertaining to exposures, mediators, and outcomes from nonoverlapping GWASs has further Canadian Journal of Gastroenterology and Hepatology minimized potential biases [62].Finally, the application of mediated MR analysis has notably reduced the impact of unobservable confounding factors, a common challenge in traditional mediated analyses [63].Trough the use of MVMR analysis, this study efectively unraveled the causal infuence of specifc infammatory factors on the connection between genetic susceptibility to UC and NMDSD.Nonetheless, it is essential to acknowledge the limitations of this study.First, the potential biological role of the SNPs included in this study remains incompletely elucidated.Terefore, horizontal pleiotropy cannot be completely eliminated.Fortunately, the MVMR models based on diferent assumptions yielded consistent conclusions, and sensitivity analyses did not detect horizontal pleiotropy.Second, this MR analysis was performed based on data from participants of European ancestry.Consequently, the generalizability of the results to other ethnic populations might be limited.Tird, due to sample size limitations, we examined only three common infammatory factors.

Conclusions
We found that people with CD are more likely to sufer from CeD.Furthermore, our research highlights a substantial connection between genetic susceptibility to UC and an elevated risk of AP and IBS.IL-6 appears to be a key driver of the association between UC and AP.Tese results underscore the importance of proactive monitoring of UC patients for gastrointestinal diseases, particularly in cases where IL-6 levels are elevated.Additionally, mitigating infammation could serve as an efective preventive measure against gastrointestinal disorders in individuals with UC.Te presence of these disorders may also serve as an indicator of underlying UC.

Data Availability
Te datasets used and/or analysed during the current study available from the corresponding author on reasonable request.

Ethical Approval
All studies included in cited genome-wide association studies had been approved by a relevant review board.

Consent
Not applicable.

Figure 1 :
Figure 1: Main design of this study.

Table 2 :
Pleiotropy and heterogeneity test of the UC/CD IVs from nonmalignant digestive system diseases GWAS.

Table 3 :
MVMR analysis of the efect of ulcerative colitis on acute pancreatitis and irritable bowel syndrome.