Gallbladder disease and colorectal cancer: a two-sample bidirectional Mendelian randomization study.

Introduction
Colorectal cancer (CRC) is one of the most common malignant tumors worldwide in terms of both incidence and mortality. In 2020, there were more than 1.9 million new cases and approximately 930,000 deaths, and its disease burden is projected to further increase by 2040, posing a serious threat to public health [1, 2]. The development of CRC is influenced by a combination of factors, including genetic susceptibility, dietary patterns, obesity, metabolic syndrome, and the intestinal microbiota, among which a considerable proportion of risk factors are modifiable. Therefore, identifying new potential risk factors is of great importance for formulating precise prevention strategies [2]. Gallbladder diseases (including cholelithiasis, cholecystitis, and gallbladder polyps) are very common in the general population, and cholecystectomy, as the main treatment for these conditions, is currently one of the most frequently performed abdominal surgeries in clinical practice [3, 4]. With the continuing rise in the prevalence of gallbladder disease and the volume of cholecystectomy, the potential association between these conditions and intestinal tumors, particularly CRC, has attracted increasing attention.
In recent years, multiple observational epidemiological studies have suggested a possible association between gallbladder disease and colorectal adenomas as well as CRC [3]. Meta-analyses based on large cohort and case–control studies have found that patients with gallbladder disease have an increased risk of colorectal precancerous lesions and CRC, and some studies have indicated that the risk may be more pronounced in the proximal colon or within specific postoperative time windows [3–5]. However, other studies have reported that, after fully accounting for the latency period and rigorously controlling for confounding factors, there is no significant association between cholecystectomy and the overall incidence of CRC, and the previously observed association may even disappear [6–8]. The inconsistency of findings across studies may be closely related to differences in study population characteristics, definitions of exposure and outcome, follow-up duration, and the extent of adjustment for confounders such as diet, obesity, and metabolic abnormalities. At the same time, gallbladder disease itself may more likely reflect the presence of underlying risk factors such as metabolic syndrome. Based on traditional observational studies, it is difficult to distinguish whether gallbladder disease plays the role of a “cause” or a “marker” in CRC development, and it is also challenging to completely rule out residual confounding and reverse causation.
From the perspective of potential mechanisms, gallbladder disease and cholecystectomy can alter the storage and excretion pattern of bile, leading to significant changes in the composition and exposure duration of bile acids in the lumen of the small intestine and colon, thereby affecting gut microbiota homeostasis, mucosal barrier function, and the local immune microenvironment [9, 10]. Multiple basic and clinical studies have shown that the accumulation of secondary bile acids mediated by the gut microbiota can promote the progression of the adenoma–carcinoma sequence in the colorectum by inducing oxidative stress, DNA damage, and multiple pro-inflammatory and pro-proliferative signaling pathways, whereas certain microbiota-derived secondary bile acids may exhibit antitumor effects under specific conditions, giving the bile acid–gut microbiota–tumor axis a “double-edged sword” character [9, 11–13]. Against this background, Mendelian randomization (MR), by using genetic variants associated with an exposure as instrumental variables and under the core assumptions of relevance, independence, and exclusion restriction, can mitigate confounding and reverse causation within a framework analogous to a randomized controlled trial, and has been widely applied to explore potentially modifiable risk factors for CRC [14–16]. Leveraging summary statistics from genome-wide association studies (GWAS) provided by large-scale prospective cohorts such as the UK Biobank and open databases such as IEU OpenGWAS, the present study adopted a two-sample, bidirectional MR design to systematically assess, at the genetic level, whether there is a bidirectional causal relationship between gallbladder disease and CRC, thereby providing more robust evidence for the formulation of primary prevention strategies for CRC in patients with gallbladder disease [17, 18].

Materials and methods

Study design
This study adopted a two-sample, bidirectional Mendelian randomization (MR) design to evaluate the causal relationship between gallbladder disease and colorectal cancer (CRC). In the forward analysis, gallbladder disease was treated as the exposure and CRC as the outcome; in the reverse analysis, the genetic liability to CRC was treated as the exposure and gallbladder disease as the outcome (Fig. 1). MR inference relies on three core assumptions (Fig. 2): (i) the relevance assumption: the selected single nucleotide polymorphisms (SNPs) are strongly associated with the exposure; (ii) the independence assumption: the instrumental variables are independent of confounders of the exposure–outcome relationship; and (iii) the exclusion restriction assumption: the instrumental variables influence the outcome only through the exposure and not via alternative pathways [19].

Data sources
All data used in this study were obtained from the IEU Open GWAS project based on the UK Biobank. To avoid bias due to sample overlap, we selected two independent datasets from different sources. Summary statistics for gallbladder disease were derived from a genome-wide association study (GWAS) published in 2018 (ukb-d-K11_GALLBILPANC), including a total of 361,194 participants, of whom 13,922 were cases and 347,272 were controls, with 13,586,589 SNPs. Summary statistics for CRC were obtained from a GWAS published in 2021 (ebi-a-GCST90018808), comprising 470,002 participants, including 6,581 cases and 463,421 controls, with 24,182,361 SNPs. Both datasets were based on individuals of European ancestry, thereby minimizing concerns about population stratification. As all data used in this study were publicly available and de-identified summary-level data, no additional ethical approval was required. Details are presented in Table 1.

Selection of SNPs
SNP selection was performed using the TwoSampleMR package in R, under the premise that the core assumptions of MR analysis were satisfied. First, SNPs that were significantly associated with the exposure (gallbladder disease) were screened using a genome-wide significance threshold (P = 5 × 10⁻⁸). The clumping parameters were set to r² = 0.001 and kb = 10,000 to remove SNPs with r² > 0.001, thereby eliminating the influence of linkage disequilibrium. Next, exposure and outcome data were harmonized by aligning effect alleles, and palindromic SNPs with intermediate allele frequencies were excluded. MR-PRESSO was then applied to identify and remove outlier SNPs. Finally, the F statistic was calculated to ensure the absence of weak instruments (F > 10). The F statistic was calculated as F = beta²/se².

Statistical analysis
All analyses were performed using R version 4.4.1, with the “TwoSampleMR” package (version 0.6.8) employed for data import, instrument selection, MR analyses, and sensitivity analyses.

MR analyses
In this study, the inverse variance weighted (IVW) method was used as the primary MR analysis, supplemented by MR-Egger regression, weighted median, simple mode, and weighted mode methods [20]. The causal effect of gallbladder disease on CRC was estimated using odds ratios (ORs) and their 95% confidence intervals (CIs). Previous research has shown that, in the absence of statistically significant heterogeneity and horizontal pleiotropy (P > 0.05), IVW provides the most accurate causal estimates in MR analyses, whereas the other methods can offer additional validation and support for the IVW results.

Sensitivity analyses
To further assess the robustness of the MR findings, the following sensitivity analyses were conducted: (1) MR-Egger intercept analysis was used to evaluate horizontal pleiotropy, which is an important indicator of the reliability of MR results; if horizontal pleiotropy is present (P < 0.05), the results may be considered unreliable; (2) Cochran’s Q test was used to assess heterogeneity, and was performed separately for the IVW and MR-Egger methods; a P value > 0.05 indicates no evidence of heterogeneity; (3) a leave-one-out analysis was conducted by sequentially excluding each SNP to determine whether any single SNP had a disproportionate influence on the causal estimates; (4) funnel plots were generated and visually inspected for symmetry to evaluate potential bias when using individual SNPs as instrumental variables.

Results

Gallbladder disease and colorectal cancer (forward MR analysis)

Instrumental variables
A total of 19 SNPs that were significantly associated with the exposure and mutually independent were included. No SNPs were lost during harmonization with the outcome dataset, and no SNPs were excluded due to being palindromic or incompatible. Thus, 19 SNPs were ultimately retained for analysis (Table 2). All SNPs had F statistics greater than 10, indicating no evidence of weak instrument bias. MR-PRESSO analysis did not identify any outlier SNPs.

MR analysis results
The IVW analysis indicated that gallbladder disease did not increase the risk of CRC [OR = 4.683, 95% CI (0.610, 35.974), P = 0.138]. Similarly, the results from MR-Egger regression, weighted median, simple mode, and weighted mode methods were all consistent (P > 0.05), suggesting no causal relationship between gallbladder disease and CRC (P > 0.05). Detailed results are shown in Table 3; Fig. 3, and Figure S1.

Sensitivity analyses
The heterogeneity assessment using Cochran’s Q test showed that for the IVW method, Q = 32.758 and P = 0.018, and for the MR-Egger method, Q = 32.325 and P = 0.014, indicating the presence of heterogeneity in both analyses. Despite this heterogeneity, the overall robustness of the results was not substantially affected. The MR-Egger intercept test yielded an intercept of 0.004 with P = 0.639, indicating no evidence of horizontal pleiotropy (Tables 3 and 4). The leave-one-out analysis demonstrated that the exclusion of any single SNP did not materially alter the causal estimates (Fig. 4). The funnel plot appeared approximately symmetric overall, suggesting minimal bias when individual SNPs were used as instrumental variables (Fig. 5).

Colorectal cancer and gallbladder disease (reverse MR analysis)

Instrumental variables
In the reverse analysis, CRC was treated as the exposure and gallbladder disease as the outcome. The same selection thresholds as in the forward analysis were applied, and SNPs in linkage disequilibrium were removed. A total of 30 SNPs significantly associated with the exposure were initially identified. During harmonization with the outcome dataset, no SNPs were lost, and no SNPs were excluded due to being palindromic or incompatible; thus, 30 SNPs were ultimately included in the analysis (Table 5). All F statistics were greater than 10, ranging from 29.576 to 1,578.256, with a mean of 72.250, indicating no evidence of weak instrument bias. MR-PRESSO identified three outlier SNPs (rs11874392, rs4546885, rs7722513).

MR analysis results
The IVW analysis showed that CRC did not increase the risk of gallbladder disease [OR = 1.000, 95% CI (0.998, 1.003), P = 0.630]. Similarly, the results from MR-Egger regression, weighted median, simple mode, and weighted mode methods were all consistent (P > 0.05), suggesting no causal relationship between CRC and gallbladder disease (P > 0.05). Detailed results are presented in Table 3; Fig. 6, and Figure S2.

Sensitivity analyses
Heterogeneity testing showed that for the IVW method, Q = 29.702 and P = 0.280, and for MR-Egger regression, Q = 25.576 and P = 0.430, indicating no significant heterogeneity in this study. The MR-Egger intercept was − 0.000 with P = 0.056, suggesting no evidence of horizontal pleiotropy (Tables 3 and 4). The leave-one-out analysis, in which each SNP was systematically excluded in turn, demonstrated that no single SNP exerted a substantial influence on the causal estimates between CRC and gallbladder disease (Fig. 7). The funnel plot was generally symmetric, indicating minimal bias when individual SNPs were used as instrumental variables (Fig. 8).

Discussion
Based on large-scale GWAS data, this study used a two-sample, bidirectional MR design to systematically evaluate, from a genetic perspective, the causal relationship between gallbladder disease and CRC. Previous epidemiological studies have yielded inconsistent conclusions regarding this association: early meta-analyses suggested a mildly increased risk of CRC, particularly proximal colon cancer, after cholecystectomy, and several more recent large cohort studies have also reported that cholecystectomy may be associated with the risk of tumors at multiple sites in the digestive tract [4, 21, 22]. However, updated systematic reviews and meta-analyses incorporating a greater number of high-quality cohorts have found no clear association between cholecystectomy and overall CRC incidence, with only a modestly increased risk observed in certain colonic segments [23]. Against this background, our results show that in both the forward MR analysis, with gallbladder disease as the exposure and CRC as the outcome, and the reverse MR analysis, with genetic liability to CRC as the exposure and gallbladder disease as the outcome, neither the IVW method nor multiple complementary methods detected a statistically significant causal effect. This suggests that gallbladder disease per se is unlikely to be a major causal driver of CRC. It should be noted that although the point estimate of the OR from the IVW method in the forward analysis was greater than 1, the 95% CI was extremely wide and P > 0.05, indicating that this apparent “risk increase” is more likely to reflect instability of the effect estimate rather than a true causal association.
From a mechanistic standpoint, remodeling of the bile acid pool and disruption of the intestinal micro-ecology following gallbladder disease and cholecystectomy have long been regarded as potential pro-carcinogenic factors for CRC. A large body of basic and translational research has demonstrated that secondary bile acids can promote the progression of the adenoma–carcinoma sequence in the colon epithelium by inducing oxidative stress, DNA damage, dysregulation of apoptosis, and alterations in the chronic inflammatory microenvironment [24, 25]. At the same time, changes in the concentration and composition of bile acids within the intestinal lumen may convert them from “physiological signaling molecules” into cytotoxic and pro-carcinogenic mediators [26]. In recent years, the bile acid–gut microbiota axis has been recognized as a key hub linking biliary diseases to intestinal tumors; by modulating microbial composition, metabolic products, and mucosal barrier function, it may participate in the initiation and progression of CRC [27, 28]. Under this “strong mechanistic hypothesis,” the failure of our MR analysis to demonstrate a causal effect of gallbladder disease on CRC at the genetic level suggests that bile acid– and microbiota-related pro-carcinogenic mechanisms are more likely mediated by other upstream risk factors, rather than CRC being directly driven by a genetically determined “gallbladder disease predisposition.”
Our findings also help to explain the coexistence of “positive” and “null” results in observational studies. Early evidence, predominantly from case–control studies, often suggested an increased risk of CRC after cholecystectomy, with particularly elevated risks in the proximal colon, whereas more recent high-quality cohorts and meta-analyses, after careful control for confounders such as timing of follow-up initiation, baseline comorbidities, and lifestyle factors, have mostly observed only weak associations or no significant overall relationship [22, 23]. By using genetically determined instrumental variables that approximate “random allocation,” MR reduces residual confounding and reverse causation at the design level [29]. Thus, the “negative” results obtained in our study are likely to be closer to the true long-term causal effect between gallbladder disease and CRC. Taken together with previous evidence, one may speculate that, on the one hand, gallbladder disease and cholecystectomy in clinical practice often coexist with established CRC risk factors such as obesity, type 2 diabetes, and high-fat diets, thereby predisposing traditional cohort analyses to residual confounding; on the other hand, intensified diagnostic evaluations before and after surgery may lead to detection bias, temporarily increasing the detection rate of CRC in the early postoperative period, while the true long-term risk does not actually increase [4, 23]. Our genetic findings support the latter explanation, namely that gallbladder-related conditions are more likely to serve as a “risk marker” of CRC rather than a causal factor.
From a methodological perspective, this study has several strengths in MR design and statistical analysis. First, we drew on large-scale GWAS summary statistics provided by platforms such as MR-Base/OpenGWAS, selecting exposure and outcome datasets that were independent in origin and relatively homogeneous in ancestry, thereby reducing bias due to sample overlap and population stratification to some extent [29]. All included instrumental SNPs had F statistics clearly greater than 10, indicating no obvious weak instrument bias. Second, IVW was used as the primary analysis method, and was complemented by MR-Egger, weighted median, and simple/weighted mode estimators to address scenarios in which some instruments might be invalid [30–32]. In our sensitivity analyses, we used the MR-Egger intercept to test for horizontal pleiotropy and applied MR-PRESSO to detect and remove outlier SNPs, thereby mitigating the impact of pleiotropy on causal estimates from a statistical perspective [33, 34]. Even in the forward analysis, where some heterogeneity was present, the direction and magnitude of effect estimates were highly consistent across MR methods, which further strengthened the robustness of our findings [30, 35]. In addition, this study closely followed recent MR methodological reviews and guidelines regarding two-sample MR, heterogeneity assessment, and the combined use of robust estimators, which contributes to improving the credibility of causal inference [29, 35, 36].
Naturally, several limitations of this study warrant cautious interpretation. First, the exposure was defined as a composite “gallbladder disease” phenotype derived from the UK Biobank, making it difficult to distinguish among cholelithiasis, cholecystitis, gallbladder polyps, and whether cholecystectomy had been performed, and precluding assessment of disease severity and duration. This may mask potential differences in CRC risk among specific subtypes of gallbladder disease [22, 23]. Second, we used GWAS summary statistics for overall CRC and were therefore unable to perform stratified analyses by anatomical subsites such as proximal/distal colon and rectum, even though previous observational studies have suggested that the association between gallbladder-related conditions and CRC risk may be stronger for proximal colon cancer [4, 21]. Thus, we cannot rule out the possibility that small effects exist in specific anatomical or molecular subtypes of CRC that our study was underpowered to detect. Third, our study population consisted mainly of individuals of European ancestry, and further work is needed to determine whether these findings can be generalized to East Asian and other populations [29, 36]. Fourth, although we evaluated and partially corrected for horizontal pleiotropy using methods such as MR-Egger and MR-PRESSO, complex pleiotropic structures may still remain in the presence of multiple potential biological pathways [30, 33, 34, 36]. Therefore, the “negative” findings of this study are best regarded as a strong complement to the existing observational evidence, rather than a definitive conclusion applicable to all populations and all subtypes of gallbladder disease.
At the clinical and public health levels, our results suggest that genetic susceptibility to gallbladder disease or cholecystectomy alone is insufficient to classify individuals as being at high risk for CRC, supporting current guideline recommendations that CRC screening strategies should be determined according to the general population risk profile and coexisting risk factors, rather than intensified solely on the basis of a history of gallbladder disease or cholecystectomy [22, 23]. Nevertheless, given the key role of the bile acid–gut microbiota axis in CRC development and progression and the frequent coexistence of gallbladder disease with obesity, dysregulation of glucose and lipid metabolism, and unhealthy lifestyles, comprehensive management of metabolic risk factors in this patient group should still be emphasized in clinical practice [24, 27, 28]. This includes weight control, dietary modification, and the rational use of statins and antidiabetic agents, with the aim of reducing the risk of CRC and other metabolically related cancers at a systemic level. In the future, it will be necessary to integrate more refined phenotypes, large multi-ethnic GWAS datasets, and multi-omics analyses to further dissect the causal chain along the gallbladder–bile acid–gut microbiota–CRC axis and to evaluate the feasibility of targeting bile acid metabolism or the gut microbiota as strategies for CRC prevention.

Conclusion
In this two sample bidirectional Mendelian randomization study does not support a causal relationship between gallbladder disease and colorectal cancer in either direction, suggesting that gallbladder disease itself is unlikely to be a major genetic driver of colorectal carcinogenesis. These findings indicate that a history of gallbladder disease or cholecystectomy alone should not be considered sufficient to classify individuals as high risk for colorectal cancer, and screening strategies should instead continue to be guided by established risk factors and overall risk profiles. Nonetheless, because gallbladder disease frequently coexists with obesity, metabolic abnormalities, and adverse lifestyle patterns, comprehensive management of metabolic risk factors remains essential to reduce the broader burden of colorectal and other related cancers. Future studies incorporating refined phenotyping, diverse populations, and multi-omics approaches are warranted to further elucidate the complex interactions among gallbladder pathology, bile acid metabolism, the gut microbiota, and colorectal cancer.

Electronic Supplementary Material
Below is the link to the electronic supplementary material.