Mendelian randomization reveals the causal relationship between cholecystectomy and ischemic stroke

Abstract

Background

The association between cholecystectomy and ischemic stroke remains unclear. Therefore, this study aimed to investigate the potential association of genetically predicted liability to cholecystectomy with ischemic stroke and its subtypes using Mendelian randomization (MR).

Methods

The inverse variance weighted (IVW) method was used as the primary analytical method, with weighted median and MR-Egger regression as complementary methods. Overall ischemic stroke was defined as the primary outcome, whereas ischemic stroke subtypes were considered secondary outcomes. A Bonferroni-corrected significance threshold was applied to the three secondary outcomes. Heterogeneity, horizontal pleiotropy, and the influence of individual variants were assessed using sensitivity analyses.

Results

Genetically predicted liability to cholecystectomy was not associated with overall ischemic stroke in the primary IVW analysis (OR = 0.57, 95% CI 0.23–1.37, P = 0.208). In the secondary analyses, the inverse association with small-vessel ischemic stroke remained significant after Bonferroni correction (OR = 0.10, 95% CI 0.02–0.57, P = 0.009). The inverse association with large-artery atherosclerotic ischemic stroke was nominally significant but did not meet the corrected threshold (OR = 0.09, 95% CI 0.01–0.71, P = 0.023), whereas no association was observed for cardioembolic ischemic stroke.

Conclusion

Genetically predicted liability to cholecystectomy was not significantly associated with overall ischemic stroke. The secondary analyses supported an inverse association with small-vessel ischemic stroke and suggested a possible inverse association with large-artery atherosclerotic ischemic stroke. No significant association was observed for cardioembolic ischemic stroke.

Supplementary Information

The online version contains supplementary material available at 10.1186/s12883-026-05246-0.

Introduction

Cholecystectomy is used as the standard treatment for gallbladder disease [1]. Between 2000 and 2019, 1,194,786 cases of cholecystectomy were reported in the England alone [2]. However, the changes in patients’ systemic functioning after cholecystectomy are still uncertain, and studies with relevant long-term follow-up are still lacking. And most of the current studies have been limited to postoperative complications associated with surgery, such as bile duct injury, and bile leakage [3]. Few studies have explored the dimension of whether cholecystectomy is associated with an increased risk of other diseases. Clarifying the association between liability to cholecystectomy and subsequent cerebrovascular outcomes may improve our understanding of the long-term health characteristics of patients who undergo this procedure.

Cerebrovascular accidents (CVAs) are the second largest contributor to global mortality [4]. The most recent definition of stroke by the American Stroke Association is an episode of acute neurologic dysfunction presumed to be caused by ischemia or hemorrhage that lasts ≥ 24 h or until death [5]. Strokes are divided into two main categories, ischemic and hemorrhagic, with ischemic strokes accounting for the vast majority of strokes [6]. In 2019, the global prevalence of ischemic stroke reached 77.19 million, with 3.29 million deaths [7]. Multiple risk factors have been shown to precipitate ischemic stroke, including infectious events, hypertension, and hyperlipidemia [8–10]. After an ischemic stroke, most patients experience impaired motor function, brain dysfunction, difficulty swallowing, and even paralysis [11–14]. This will seriously affect the patient’s future quality of life.

A clinical study tentatively confirmed an association between cholecystectomy and ischemic stroke, with follow-up after undergoing cholecystectomy showing fewer strokes compared to a control group that did not undergo cholecystectomy. However, the causal relationship is not yet clear [15]. Therefore, it is necessary to further explore the relationship between genetic susceptibility to cholecystectomy and the risk of ischemic stroke and its subtypes.

Mendelian randomization (MR) is an emerging method of causal inference that uses genetic variants strongly associated with a hypothesized exposure as instrumental variables (IVs), usually single nucleotide polymorphisms (SNPs) [16, 17]. The advantage of this approach over traditional observational epidemiological studies is that it rejects bias caused by confounders or reverse causation, and therefore it can be regarded as a natural randomized controlled trial (RCT) [18]. Therefore, this study used a two-sample MR design to investigate the associations of genetically predicted liability to cholecystectomy with overall ischemic stroke and its major etiological subtypes.

Methods

This study is reported according to the guidelines “Enhancing reporting of observational epidemiological studies by using Mendelian randomization methods: the STROBE-MR statement“ [19]. There are three main assumptions that must be fulfilled in MR studies: (1) the correlation assumption: there is a strong correlation between the IVs and the exposure factor; (2) the independence assumption: the IVs are not associated with potential risk factors for the outcome; (3) the exclusivity assumption: the IVs affect the outcome only through the exposure [20].

Study design

We conducted a two-sample MR study to investigate the association between genetically predicted liability to cholecystectomy and overall ischemic stroke, which was designated as the primary outcome. Small-vessel, large-artery atherosclerotic, and cardioembolic ischemic stroke were analyzed as secondary outcomes to explore potential subtype-specific associations. The study design is summarized in Fig. 1.

Fig. 1.

Schematic diagram of MR

Data source

Summary data from the largest genome-wide association study (GWAS) was obtained from the public website “Open GWAS” (https://gwas.mrcieu.ac.uk/) which has been produced by the MRC Integrated Epidemiology Unit (IEU) at the University of Bristol. The GWAS ID of cholecystectomy is “ukb-b-6235”, which includes 18,319 cases and 444,614 controls. Ischemic stroke cases are defined using standard diagnostic criteria based on clinical and imaging findings and are further classified into subtypes, including small-vessel ischemic stroke, large-artery atherosclerotic ischemic stroke, and cardioembolic ischemic stroke, using the ORG 10 172 test of the Treatment of Acute Stroke (TOAST) criteria [21, 22]. Potential sample overlap was assessed by reviewing the original publications and cohort information for the exposure and outcome GWAS datasets. The exposure GWAS was derived entirely from the UK Biobank, whereas the outcome GWAS datasets were obtained from the MEGASTROKE consortium. The MEGASTROKE meta-analysis included 29 studies, none of which included the UK Biobank as a contributing cohort [21]. Therefore, no known cohort-level overlap was identified between the exposure and outcome datasets. However, participant-level overlap could not be completely excluded because individual-level identifiers were unavailable. In addition, all GWAS datasets included in this study were restricted to individuals of European ancestry, thereby reducing potential bias related to population stratification. All GWAS summary statistics used in this study were obtained from publicly accessible databases. Therefore, no additional permission was required for their use. Table 1 shows more relevant information about the exposure and outcome datasets.

Table 1.

Details of the GWAS summary dataset for exposure and outcome

Item	GWAS ID	Population	Cohort	Cases	Controls	Total sample size	Number of SNPs	Year
Cholecystectomy	ukb-b-6235	European	UK Biobank	18,319	444,614	462,933	9,851,867	2018
Ischemic stroke	ebi-a-GCST006908	European	MEGASTROKE	34,217	406,111	440,328	8,296,492	2018
Small-vessel ischemic stroke	ebi-a-GCST005841	European	MEGASTROKE	5,386	192,662	198,048	6,150,261	2018
Large-artery atherosclerotic ischemic stroke	ebi-a-GCST005840	European	MEGASTROKE	4,373	146,392	150,765	7,992,739	2018
Cardioembolic ischemic stroke	ebi-a-GCST006910	European	MEGASTROKE	7,193	204,570	211,763	8,271,294	2018

SNP single nucleotide polymorphism

The selection of IVs

To select suitable IVs, SNPs associated with cholecystectomy at genome-wide significance (P < 5 × 10⁻⁸) were first identified. Linkage disequilibrium (LD) clumping was then performed using an R² threshold of 0.001 and a window size of 10,000 kb to obtain independent variants [23]. Exposure and outcome data were harmonised to align the effect alleles. For palindromic SNPs, strand orientation was inferred using effect-allele frequency information. Palindromic variants with a minor allele frequency greater than 0.42 were considered ambiguous and excluded. Variants with incompatible allele information were also excluded. Instrument strength was assessed using the per-variant F-statistic, calculated as F = (β_exposure/SE_exposure)², where β_exposure and SE_exposure represent the estimated SNP effect on cholecystectomy and its standard error, respectively. An F-statistic greater than 10 was considered indicative of a sufficiently strong instrument [24]. After LD clumping, 45 independent exposure-associated SNPs were retained as candidate IVs, with F-statistics ranging from 30.25 to 2903.36 and a median value of 47.66, suggesting that weak instrument bias was unlikely. Detailed information on the IVs is provided in Supplementary Table S1.

Statistical analysis

Inverse variance weighting (IVW) was used as the main analytical method [25]. IVW requires all genetic variation to be valid instruments [26]. To ensure the reliability of the results, we also used two additional analytical methods, including MR-Egger regression and weighted median. The weighted median method allows for less than 50% invalid IVs, whereas MR-Egger, because it is not susceptible to directional pleiotropy, can still provide valid estimates when all IVs are invalid [27, 28]. We compared the results of the IVW method with those of the weighted median and MR-Egger methods. Given the different assumptions underlying these methods, complete agreement was not expected; however, concordance in effect direction was considered to strengthen the credibility of the findings.

Heterogeneity, pleiotropy, and sensitivity analysis

Heterogeneity in the IVW model was assessed by the Cochran’s Q test. The Cochran’s Q test P < 0.05 suggests the existence of significant heterogeneity. Nevertheless, the presence of heterogeneity does not imply that the IVW model is certainly invalid. If significant heterogeneity is observed, the random-effects model is adopted; otherwise, the fixed-effect model is used [29]. To assess the horizontal pleiotropy of results, we used the MR-Egger test [30, 31]. Once outliers were detected by the MR-PRESSO method, they were eliminated. Asymmetric funnel plots are also usually indicative of potential pleiotropy. Leave-one-out analyses were performed to assess whether the removal of individual SNPs had a significant effect on the results.

Analysis tools and statistical significance

All analyses were performed in R software (version 4.4.1), using the two-sample MR and MR-PRESSO software package [32]. All statistical tests were two-sided. A P-value < 0.05 was considered statistically significant for the primary outcome. For the three secondary outcomes, a Bonferroni-corrected significance threshold of P < 0.017 (0.05/3) was applied to the primary IVW estimates.

Functional annotation of instrumental variants

To explore the potential biological significance of the candidate IVs, functional annotation was performed after LD clumping using the Ensembl Variant Effect Predictor based on the GRCh37 genome assembly [33]. Information on variant consequences, mapped genes, transcript biotypes, and predicted functional effects was extracted.

Results

Primary MR analysis of overall ischemic stroke

In the primary analysis, genetically predicted liability to cholecystectomy was not significantly associated with overall ischemic stroke in the IVW analysis (OR = 0.57, 95% CI 0.23–1.37, P = 0.208; Table 2). The weighted median method suggested an inverse association, whereas the MR-Egger estimate did not reach statistical significance. Significant heterogeneity was detected by Cochran’s Q test (Q = 65.289, P = 0.007; Table 3); therefore, the random-effects IVW estimate was used for interpretation. No evidence of directional horizontal pleiotropy was observed based on the MR-Egger intercept test (P = 0.277; Table 3). MR-PRESSO identified rs11887534 as an outlier. After removal of this variant, the point estimate changed direction; however, the IVW estimate remained statistically nonsignificant (OR = 1.85, 95% CI 0.63–5.41, P = 0.261). Therefore, the primary analysis provided no evidence of an association between genetically predicted liability to cholecystectomy and overall ischemic stroke.

Table 2.

MR estimates for the association of genetically predicted liability to cholecystectomy with ischemic stroke and its subtypes

Outcome	No. of SNPs	Method	P-value	OR (95% CI)
Primary outcome
Ischemic stroke	41	MR-Egger	0.104	0.33 (0.09–1.22)
		Weighted median	< 0.001	0.14 (0.05–0.37)
		IVW	0.208	0.57 (0.23–1.37)
Secondary outcomes
Small-vessel ischemic stroke	32	MR-Egger	0.383	0.31 (0.02–4.2)
		Weighted median	0.042	0.14 (0.02–0.93)
		IVW	0.009	0.10 (0.02–0.57)
Large-artery atherosclerotic ischemic stroke	39	MR-Egger	0.028	0.02 (0.001–0.58)
		Weighted median	< 0.001	0.01 (0.001–0.14)
		IVW	0.023	0.09 (0.01–0.71)
Cardioembolic ischemic stroke	41	MR-Egger	0.110	0.15 (0.02–1.45)
		Weighted median	0.054	0.15 (0.02–1.04)
		IVW	0.706	0.74 (0.16–3.47)

SNP single nucleotide polymorphism, OR odds ratio, IVW inverse-variance weighted, MR-Egger Mendelian randomization-Egger

Table 3.

Results of a sensitivity analyses of the genetically predicted association of cholecystectomy with ischemic stroke and its subtypes

Outcome	Cochran’s Q	Q P-value		MR-Egger Intercept	Intercept P-value
Primary outcome
Ischemic stroke	65.289		0.007	0.004		0.277
Secondary outcomes
Small-vessel ischemic stroke	54.375		0.008	-0.008		0.269
Large-artery atherosclerotic ischemic stroke	60.607		0.011	0.01		0.280
Cardioembolic ischemic stroke	51.906		0.098	0.012		0.072

MR-Egger Mendelian randomization-Egger

The scatter plots reflected the change in the fitted IVW estimate after removal of rs11887534 (Fig. 2A and C). The forest plots showed variation among the SNP-specific estimates and a corresponding change in the pooled estimate after outlier removal (Supplementary Material 2). No marked directional asymmetry was apparent in the funnel plots (Fig. 2B and D). Leave-one-out analysis further indicated that rs11887534 had a notable influence on the pooled estimate, whereas no single remaining SNP appeared to dominate the result after its removal (Supplementary Material 2).

Fig. 2.

Scatter and funnel plots for the primary MR analysis of overall ischemic stroke. A Scatter plot before outlier removal; B Funnel plot before outlier removal; C Scatter plot after removal of rs11887534; D Funnel plot after removal of rs11887534

Secondary MR analyses of ischemic stroke subtypes

For small-vessel ischemic stroke, the IVW analysis indicated an inverse association with genetically predicted liability to cholecystectomy (OR = 0.10, 95% CI 0.02–0.57, P = 0.009; Table 2). This association remained significant after Bonferroni correction. Significant heterogeneity was detected (Q = 54.375, P = 0.008; Table 3), and the random-effects IVW estimate was therefore used. The weighted median estimate was directionally consistent with the IVW result, whereas the MR-Egger estimate was nonsignificant. No evidence of directional horizontal pleiotropy was detected (P = 0.269; Table 3). The scatter and forest plots showed an overall inverse trend despite variation among SNP-specific estimates, and the leave-one-out analysis showed that the direction of the estimate remained broadly stable after sequential SNP exclusion (Fig. 3A; Supplementary Material 2). No marked directional asymmetry was apparent in the funnel plot (Fig. 3B).

Fig. 3.

Scatter and funnel plots for the secondary MR analyses of ischemic stroke subtypes. A Scatter plot for small-vessel ischemic stroke; B Funnel plot for small-vessel ischemic stroke; C Scatter plot for large-artery atherosclerotic ischemic stroke before outlier removal; D Funnel plot for large-artery atherosclerotic ischemic stroke before outlier removal; E Scatter plot for large-artery atherosclerotic ischemic stroke after removal of rs2727270; F Funnel plot for large-artery atherosclerotic ischemic stroke after removal of rs2727270; G Scatter plot for cardioembolic ischemic stroke; H Funnel plot for cardioembolic ischemic stroke

For large-artery atherosclerotic ischemic stroke, the random-effects IVW analysis indicated a nominally significant inverse association (OR = 0.09, 95% CI 0.01–0.71, P = 0.023), which did not meet the Bonferroni-corrected threshold. Significant heterogeneity was detected among the SNP-specific estimates (Q = 60.607, P = 0.011; Table 3). The weighted median and MR-Egger estimates were directionally consistent with the IVW result. After removal of rs2727270, the inverse direction persisted (OR = 0.12, 95% CI 0.02–0.82, P = 0.030), but the association still did not meet the corrected significance threshold. The scatter and forest plots consistently showed an inverse trend before and after outlier removal, with no substantial change in the overall direction of effect (Fig. 3C and E; Supplementary Figure S1E and F). The funnel and leave-one-out analyses were broadly consistent with these findings (Fig. 3D and F; Supplementary Material 2). No evidence of directional horizontal pleiotropy was detected (P = 0.280; Table 3).

For cardioembolic ischemic stroke, no significant association was observed in the IVW analysis (OR = 0.74, 95% CI 0.16–3.47, P = 0.706; Table 2). Neither significant heterogeneity (Q = 51.906, P = 0.098) nor directional horizontal pleiotropy (P = 0.072) was detected (Table 3). The scatter and forest plots showed no clear evidence of a consistent association, and the leave-one-out analysis did not materially alter the overall estimate (Fig. 3G; Supplementary Material 2). The funnel plot showed no marked directional asymmetry (Fig. 3H).

Functional annotation of candidate IVs

Most variants were located in noncoding regions, including 24 intronic variants and seven intergenic variants, whereas six variants were annotated as missense variants. Several variants mapped to genes potentially involved in lipid metabolism, hepatobiliary transport, or inflammatory regulation, including ABCG5/ABCG8, FADS2, APOH, HNF1A, HNF4A, ATP8B1, SULT2A1, SERPINA1, LITAF and ATG16L2. Notably, rs11887534 and rs2727270, which were identified as outliers in the analyses of overall ischemic stroke and large-artery atherosclerotic ischemic stroke, mapped to the ABCG5/ABCG8 region and FADS2, respectively. Detailed annotation results are presented in Supplementary Table S1.

Discussion

This study investigated the associations of genetically predicted liability to cholecystectomy with overall ischemic stroke and three major etiological subtypes. The primary IVW analysis provided no evidence of an association with overall ischemic stroke, although the sensitivity analysis suggested that the overall estimate might be unstable. Among the secondary outcomes, the inverse association with small-vessel ischemic stroke remained significant after correction for multiple testing. For large-artery atherosclerotic ischemic stroke, inverse estimates were observed consistently across the MR methods, but the primary IVW result was only nominally significant and did not meet the Bonferroni-corrected threshold. No significant association was observed for cardioembolic ischemic stroke.

To the best of our knowledge, this is the first MR study to evaluate the associations of genetically predicted liability to cholecystectomy with overall ischemic stroke and its major etiological subtypes. By examining these associations from a genetic perspective, this study provides evidence complementary to that obtained from conventional observational studies.

There is still no consensus in recent articles on whether cholecystectomy affects ischemic stroke risk. For instance, a cohort study by Wei et al. published in 2019 concluded that cholecystectomy reduces the risk of ischemic stroke [15]. This is consistent with the conclusions we have reached. An article published in 2022 by Park’s team used a large population-based cohort study to show that cholecystectomy reduces the risk of cerebral infarction in people with gallbladder stone-associated infections, a predisposing factor for cerebral infarction in the case of ischemic stroke [34]. However, it seems that some articles hold the opposite view. For example, an observational study published in 2023 by Zhang’s team found no association between patients with gallbladder stones who underwent cholecystectomy and a reduced risk of stroke [35]. Chavez-Tapia et al. published a study in 2012 showing a higher prevalence of cardiovascular disease risk factors in patients undergoing cholecystectomy [36]. These discrepancies may be related to differences in study populations, surgical indications, severity of gallstone disease, outcome definitions and residual confounding. Moreover, most previous studies evaluated overall stroke or cerebral infarction without distinguishing etiological subtypes, which may have obscured potentially subtype-specific associations.

The subtype-specific pattern observed in the present study may have a biological basis. Alterations in lipid metabolism and inflammatory status represent plausible mechanisms, particularly for small-vessel and large-artery atherosclerotic ischemic stroke. Patients with gallstone disease commonly present with hypertriglyceridemia, hypercholesterolemia, and reduced high-density lipoprotein cholesterol levels, which are also established risk factors for atherosclerotic cardiovascular disease and ischemic stroke [37–40]. Previous studies have reported reductions in circulating triglyceride and cholesterol levels after cholecystectomy, potentially resulting from changes in the enterohepatic circulation and metabolism of bile acids [41]. Consistent with this explanation, functional annotation showed that several instrumental variants mapped to genes involved in cholesterol and fatty-acid metabolism or hepatobiliary transport, including ABCG5/ABCG8, FADS2, APOH, HNF1A, HNF4A, ATP8B1 and SULT2A1 [42–44]. These metabolic changes may partly account for the inverse associations observed for small-vessel and large-artery atherosclerotic ischemic stroke.

Inflammation may represent another relevant pathway. Gallstone disease is frequently accompanied by biliary stasis, recurrent infection and local or systemic inflammatory responses [45]. Chronic inflammation contributes to the initiation and progression of atherosclerosis, while inflammatory biomarkers such as lipoprotein-associated phospholipase A2 and high-sensitivity C-reactive protein have been associated with atherogenesis and stroke risk [46–48]. In line with this mechanism, several instrumental variants mapped to genes potentially involved in inflammatory or immune regulation, including SERPINA1, LITAF and ATG16L2 [49, 50]. Together, alterations in lipid metabolism, bile acid homeostasis and inflammatory activity may provide a biologically plausible explanation for the observed subtype-specific estimates.

By contrast, cardioembolic ischemic stroke is primarily related to atrial fibrillation and other cardiac sources of embolism rather than to atherosclerotic mechanisms. A Korean cohort study found no significant association between cholecystectomy and atrial fibrillation, and another clinical study reported no significant long-term association between cholecystectomy and coronary heart disease [51–53]. The distinct pathophysiology of cardioembolic stroke may therefore partly explain the absence of a significant association in the present MR analysis.

Significant heterogeneity was observed in the analyses of overall ischemic stroke, small-vessel ischemic stroke, and large-artery atherosclerotic ischemic stroke, indicating variation among the SNP-specific estimates. This heterogeneity may reflect differences in the biological pathways represented by the genetic instruments, instrument strength, and potential horizontal pleiotropy [31]. Therefore, MR-PRESSO was used to identify potential outlier variants, and the MR-Egger intercept test was used to assess directional horizontal pleiotropy. For overall ischemic stroke, MR-PRESSO identified rs11887534 as an outlier. Before removal of this variant, the IVW estimate was OR = 0.57 (95% CI 0.23–1.37; P = 0.208); after its removal, the estimate changed to OR = 1.85 (95% CI 0.63–5.41; P = 0.261). Although removal of rs11887534 resulted in a substantial change in the point estimate, both the original and outlier-corrected IVW estimates remained statistically nonsignificant. Thus, removal of this variant did not alter the primary statistical conclusion, although the exact magnitude and direction of the association remained uncertain in the presence of significant heterogeneity. The significant weighted median estimate may reflect differences in the assumptions and weighting schemes of the MR methods. Because IVW was the prespecified primary method and neither the original nor the outlier-corrected IVW analysis provided a significant result, the weighted median finding alone was considered insufficient to establish an association with overall ischemic stroke. Overall, the analyses provided no evidence of an association with overall ischemic stroke, while highlighting the need to interpret this result together with the heterogeneity and sensitivity analyses. For large-artery atherosclerotic ischemic stroke, the inverse direction persisted after removal of rs2727270, although the IVW result remained nominally significant. The MR-Egger intercept tests did not indicate significant directional horizontal pleiotropy.

The differences observed across the IVW, weighted median, and MR-Egger estimates may be explained by the distinct assumptions and statistical properties of these methods. IVW generally provides the most precise estimate but may be affected when some genetic instruments exhibit directional pleiotropy. The weighted median method can provide a consistent estimate when at least 50% of the total weight is derived from valid instruments [27]. In contrast, MR-Egger regression estimates an additional intercept to account for directional pleiotropy, which substantially reduces its statistical power and often results in wider confidence intervals. Consequently, MR-Egger estimates may differ from IVW estimates in magnitude and statistical significance because of limited precision, regression dilution or sensitivity to influential variants [28]. For small-vessel ischemic stroke, the MR-Egger estimate was directionally consistent with the IVW estimate but was less precise and did not reach statistical significance. For large-artery atherosclerotic ischemic stroke, the IVW, weighted median, and MR-Egger methods all yielded directionally consistent inverse estimates that reached nominal significance. However, because the primary IVW result did not meet the Bonferroni-corrected significance threshold, this finding should be interpreted with caution.

Several limitations should be acknowledged. First, the genetic instruments used in this study reflect liability to undergoing cholecystectomy. Cholecystectomy may be performed for various gallbladder conditions, including cholelithiasis, acute or chronic cholecystitis, gallbladder polyps, and an atrophic or chronically diseased gallbladder. Therefore, the present analysis cannot fully determine whether the observed inverse associations are attributable to an independent protective effect of cholecystectomy or to the underlying gallbladder conditions leading to surgery. Further studies are needed to distinguish these effects and clarify the independent role of cholecystectomy. Second, significant heterogeneity was observed in several analyses, some effect estimates were imprecise, and residual or balanced horizontal pleiotropy could not be completely excluded despite the sensitivity analyses. Third, the evaluation of multiple stroke outcomes may increase the risk of chance findings and false-positive results. To address this issue, a Bonferroni-corrected significance threshold was applied to the primary IVW analyses of the three secondary outcomes. Fourth, although no known cohort-level overlap was identified between the exposure and outcome GWAS datasets, participant-level overlap could not be completely excluded because individual-level identifiers were unavailable. Finally, all GWAS data were derived from individuals of European ancestry, which may limit the generalizability of the findings to other populations.

Despite these limitations, this study has several strengths. First, the use of genetic variants as IVs may reduce the influence of residual confounding and bias related to limited follow-up in conventional observational studies, because genetic variants are randomly allocated during meiosis. Second, germline genetic variants are fixed at conception and are generally not altered by subsequent disease processes, thereby reducing the possibility of reverse causation [54]. Third, the MR design enables potential associations to be investigated without direct intervention or long-term prospective follow-up, providing an efficient approach complementary to conventional observational studies. Finally, functional annotation of the instrumental variants provided additional information regarding the potential biological pathways represented by these variants.

In summary, this study provides a new perspective on the potential long-term systemic effects of cholecystectomy. Future studies should validate these findings in independent and ancestrally diverse populations and further investigate the underlying mechanisms, particularly those involving bile acid metabolism, lipid regulation, and inflammatory pathways. Integrating prospective cohort studies, functional experiments, and mediation analyses may help clarify the biological basis and clinical relevance of these subtype-specific associations.

Conclusion

In conclusion, this MR study found no evidence of an association between genetically predicted liability to undergoing cholecystectomy and either overall ischemic stroke or cardioembolic ischemic stroke. The inverse association with small-vessel ischemic stroke remained statistically significant after correction for multiple testing, whereas the inverse effect estimate for large-artery atherosclerotic ischemic stroke was nominally significant but did not meet the Bonferroni-corrected significance threshold. Future studies using larger GWAS datasets are needed to validate these findings, and longitudinal follow-up studies are warranted to further clarify the association between cholecystectomy and ischemic stroke.

Authors’ contributions

Y.X. wrote the manuscript, Z.Z. collected the data, and J.L.L. reviewed the manuscript.

Funding

This study was supported by Natural Science Foundation of Hebei Province (Grant No. H2025406008).

Data availability

All data used in this study were obtained from the publicly and freely accessible OpenGWAS database (https://gwas.mrcieu.ac.uk/); therefore, no additional permission was required. The analysis scripts used in this study are available from the corresponding author upon reasonable request.

Declarations

Ethics approval and consent to participate

Not applicable.

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.