Editor's evaluation: Smoking, alcohol consumption, and 24 gastrointestinal diseases: Mendelian randomization analysis

Abstract

Article Figures and data Abstract Editor's evaluation eLife digest Introduction Materials and methods Results Discussion Data availability References Decision letter Author response Article and author information Metrics Abstract Background: Whether the positive associations of smoking and alcohol consumption with gastrointestinal diseases are causal is uncertain. We conducted this Mendelian randomization (MR) to comprehensively examine associations of smoking and alcohol consumption with common gastrointestinal diseases. Methods: Genetic variants associated with smoking initiation and alcohol consumption at the genome-wide significance level were selected as instrumental variables. Genetic associations with 24 gastrointestinal diseases were obtained from the UK Biobank, FinnGen study, and other large consortia. Univariable and multivariable MR analyses were conducted to estimate the overall and independent MR associations after mutual adjustment for genetic liability to smoking and alcohol consumption. Results: Genetic predisposition to smoking initiation was associated with increased risk of 20 of 24 gastrointestinal diseases, including 7 upper gastrointestinal diseases (gastroesophageal reflux, esophageal cancer, gastric ulcer, duodenal ulcer, acute gastritis, chronic gastritis, and gastric cancer), 4 lower gastrointestinal diseases (irritable bowel syndrome, diverticular disease, Crohn’s disease, and ulcerative colitis), 8 hepatobiliary and pancreatic diseases (non-alcoholic fatty liver disease, alcoholic liver disease, cirrhosis, liver cancer, cholecystitis, cholelithiasis, and acute and chronic pancreatitis), and acute appendicitis. Fifteen out of 20 associations persisted after adjusting for genetically predicted alcohol consumption. Genetically predicted higher alcohol consumption was associated with increased risk of duodenal ulcer, alcoholic liver disease, cirrhosis, and chronic pancreatitis; however, the association for duodenal ulcer did not remain statistically significant after adjustment for genetic predisposition to smoking initiation. Conclusions: This study provides MR evidence supporting causal associations of smoking with a broad range of gastrointestinal diseases, whereas alcohol consumption was associated with only a few gastrointestinal diseases. Funding: The Natural Science Fund for Distinguished Young Scholars of Zhejiang Province; National Natural Science Foundation of China; Key Project of Research and Development Plan of Hunan Province; the Swedish Heart Lung Foundation; the Swedish Research Council; the Swedish Cancer Society. Editor's evaluation This is a valuable article that is methodologically convincing and provides evidence, through Mendelian Randomisation, that genetic predisposition to smoking and alcohol consumption influences the risk to develop different gastrointestinal diseases. The findings largely corroborate the findings from observational studies, especially for the effects of smoking. The major strength of the paper is the use of the largest possible genetic datasets for both the exposures and outcomes, which makes the findings more robust. https://doi.org/10.7554/eLife.84051.sa0 Decision letter eLife's review process eLife digest People who smoke cigarettes or drink large amounts of alcohol are more likely to develop disorders with their digestive system. But it is difficult to prove that heavy drinking or smoking is the primary cause of these gastrointestinal diseases. For example, it is possible that having a digestive disorder makes people more likely to take up these habits to reduce pain or discomfort caused by the illness (an effect known as reverse causation). The association may also be the result of confounding factors, such as age or diet, which contribute to digestive problems as well as the health outcomes of smoking and drinking. Additionally, many people who smoke also drink alcohol and vice versa, making it challenging to determine if one or both behaviors contribute to the disease. One solution is to employ Mendelian randomization which uses genetics to determine if two variables are linked. Using this statistical approach, Yuan, Chen, Ruan et al. investigated if people who display genetic variants that predispose someone to becoming a smoker or drinker are at greater risk of developing certain digestive disorders. This reduces the possibility of confounding and reverse causation, as any association between genetic variants will have been present since birth, and will have not been impacted by external factors. Yuan, Chen, Ruan et al. used data from two studies that had collected the genetic and health information of thousands of people living in the United Kingdom or Finland. The analyses revealed that genetic variants associated with cigarette smoking increase the risk of 20 of the 24 gastrointestinal diseases investigated. This risk persisted for most of the disorders, even after adjusting for genes linked with alcohol consumption. Further analysis showed that genetic variants linked to heavy drinking increase the risk of duodenal ulcer, alcoholic liver disease, cirrhosis, and chronic pancreatitis. However, accounting for smoking-linked genes eliminated the relationship with duodenal ulcer. These findings suggest that smoking has detrimental effects on gastrointestinal health. Reducing the number of people who start smoking or encouraging smokers to quit may help prevent digestive diseases. Even though there were fewer associations between heavy alcohol consumption and gastrointestinal illness, further studies are needed to investigate this relationship in more depth. Introduction Tobacco smoking and alcohol consumption are leading causes of the global burden of disease and are major contributors to premature mortality (GBD 2016 Alcohol Collaborators, 2018; GBD 2016 Alcohol Collaborators, 2020). Gastrointestinal diseases account for considerable health care use and expenditures, and a holistic approach to lifestyle interventions may result in more health gains and less economic burdens (Peery et al., 2022). Population-based studies have identified tobacco smoking as a risk factor for several gastrointestinal diseases, including gastroesophageal reflux disease (Eusebi et al., 2018), esophageal cancer (Fund WCR and Research AIfC, 2007), Crohn’s disease (Piovani et al., 2019), liver cancer (McGee et al., 2019), and pancreatitis (Yadav and Whitcomb, 2010). Evidence on the association between tobacco smoking and risk of other gastrointestinal diseases is limited and inconsistent. Like smoking, heavy alcohol consumption has been associated with increased risk of several gastrointestinal outcomes, including gastritis (Bujanda, 2000), gastric cancer (Laszkowska et al., 2021), colorectal cancer (McNabb et al., 2020), cirrhosis (Simpson et al., 2019), liver cancer (McGee et al., 2019), and pancreatitis (Yadav and Whitcomb, 2010). However, whether these associations are all causal remains unestablished, since most of the evidence was obtained from observational studies where the results may be biased by reverse causality and confounding. Of note, even though reverse causality may not be an issue in the studies for any of studied gastroenterological outcomes, it might exist for certain gastroenterological diseases causing pain, which smoker patients may try to increase smoking dose to mitigate via an intake of higher levels of nicotine. In addition, as smoking and alcohol consumption are phenotypically and genetically correlated (Roberts et al., 2020; Liu et al., 2019), the independent impacts of smoking and alcohol consumption on gastrointestinal diseases are unclear. Establishing the causal association of tobacco smoking and alcohol consumption with gastrointestinal diseases is crucial, as this provides further evidence for subsequent recommending public policies and clinical interventions. Mendelian randomization (MR) is an epidemiological approach that utilizes genetic variants as an instrument to strengthen the causal inference in an exposure-outcome association (Davey Smith and Hemani, 2014). MR is by nature not prone to confounding since genetic variants are randomly assorted at conception and thus unrelated to environmental and self-adopted factors that usually act as confounders. Additionally, this method can minimize reverse causality since fixed alleles are unaffected by the onset and progression of disease. Previous MR studies have examined the associations of smoking and alcohol consumption with several gastrointestinal diseases (Yuan and Larsson, 2022a; Larsson et al., 2020; Yuan and Larsson, 2022b; Yuan et al., 2022c; Chen et al., 2022; Yuan et al., 2021). Nevertheless, whether smoking and alcohol consumption exert influence on a wide range of gastrointestinal outcomes has not been investigated in a comprehensive way. A thorough investigation on the gastrointestinal consequences of smoking and alcohol drinking is of great importance to develop non-pharmacological interventions on gastrointestinal diseases. Here, we conducted an MR investigation of the associations of smoking and alcohol consumption with the risk of common gastrointestinal diseases to fill in above knowledge gaps. Materials and methods Figure 1 shows the study design overview. The study was based on publicly available genome-wide association studies (GWAS), and the detailed information on used studies was presented in Supplementary file 1A. The genetic associations were estimated using data from the UK Biobank study (Sudlow et al., 2015), the FinnGen study (Kurki et al., 2022; https://www.finngen.fi/en), and several large consortia. The summary effect estimates were combined using meta-analysis for each gastrointestinal disease from different data resources. Included studies had been approved by corresponding institutional review boards and ethical committees, and consent forms had been signed by all participants. Figure 1 Download asset Open asset Overview of the present study design. GERA, Genetic Epidemiology Research on Aging; IIBDGC, the International Inflammatory Bowel Disease Genetics Consortium; MR, Mendelian randomization; MR-PRESSO, Mendelian randomization pleiotropy residual sum and outlier; SNP, single nucleotide polymorphism. Instrumental variable selection A total of 378 and 99 single nucleotide polymorphisms (SNPs) associated with smoking initiation (a binary phenotype indicating whether an individual had ever being a regular smoker, 1,232,091 individuals of European descent) and alcohol consumption (log-transformed drinks per week, 941,280 individuals of European descent) at the genome-wide significance threshold (p<5 × 10–8) were identified by the GWAS and Sequencing Consortium of Alcohol and Nicotine use (GSCAN) study (Liu et al., 2019). These SNPs explained approximately 2.3 and 0.3% of the variation in smoking initiation and alcohol consumption, respectively (Liu et al., 2019). SNPs in linkage disequilibrium (defined as r2 >0.01 or clump distance <10,000 kb) and with the weaker associations with the exposure were removed, leaving 314 independent SNPs as instrumental variables for smoking initiation and 84 for alcohol consumption. Smoking initiation and alcohol consumption shared two index genetic variants, which were rs1713676 and rs11692435. Considering partial sample overlap (around 30%) between the GSCAN study with full data and the UK Biobank study (Liu et al., 2019), we performed sensitivity analyses for smoking initiation and alcohol consumption using summary statistics data from the analysis excluding the UK Biobank and 23andMe. For a supplementary analysis of smoking behavior, we used 126 SNPs associated with a lifetime smoking index that considered smoking duration, heaviness, and cessation (Wootton et al., 2020). The set of genetic instruments captured around 0.36% of the variance in lifetime smoking (Wootton et al., 2020). We also conducted a sensitivity analysis using rs1229984 in ADH1B gene that encodes alcohol dehydrogenase 1B enzyme as the genetic instrument for alcohol consumption to minimize pleiotropy. Detailed information on used SNPs is presented in Supplementary file 1B. Gastrointestinal disease data sources Genetic associations with 24 gastrointestinal diseases were obtained from the UK Biobank study (Sudlow et al., 2015), the FinnGen study (Kurki et al., 2022), and two large consortia, including the International Inflammatory Bowel Disease Genetics Consortium (IIBDGC) (Liu et al., 2015) and Genetic Epidemiology Research on Aging (GERA) (Guindo-Martínez et al., 2021). Included outcomes were classified into four major categories according to the disease onset site: (1) upper gastrointestinal diseases (gastroesophageal reflux disease, esophageal cancer, gastric ulcer, acute gastritis, chronic gastritis, and gastric cancer); (2) lower gastrointestinal diseases (irritable bowel disease, celiac disease, diverticular disease, Crohn’s disease, ulcerative colitis, and colorectal cancer); (3) hepatobiliary and pancreatic diseases (non-alcoholic fatty liver disease, alcoholic liver disease, cirrhosis, liver cancer, cholangitis, cholecystitis, cholelithiasis, acute pancreatitis, chronic pancreatitis, and pancreatic cancer); and (4) other (acute appendicitis). The UK Biobank study is a large multicenter cohort study of 500,000 participants recruited in the United Kingdom between 2006 and 2010 (Sudlow et al., 2015). We used the summary statistics of European ancestry from GWAS conducted by Lee lab, where the gastrointestinal outcomes were defined by codes of the International Classification of Diseases 9th Revision (ICD-9) and ICD-10 (Zhou et al., 2020). Genetic associations were estimated by logistic regression with adjustment for sex, birth year, and the first four genetic principal components. For the FinnGen study, we used summary-level data on the genetic associations with gastrointestinal diseases from the last publicly available R7 data release (Kurki et al., 2022). The FinnGen study is a nationwide genetic study where genetic and electronic health record data were collected. The gastrointestinal diseases were ascertained by the codes of the ICD-8, ICD-9, and ICD-10. Genome-wide association analyses were adjusted for sex, age, genetic components, and genotyping batch. Summary-level genetic data on Crohn’s disease (5956 cases and 14,927 controls) and ulcerative colitis (6968 cases and 20,464 controls) were additionally obtained from the IIBDGC (Liu et al., 2015), and data on irritable bowel syndrome (3117 cases and 53,520 controls) were obtained from the GERA (Guindo-Martínez et al., 2021). Detailed diagnostic codes are listed in Supplementary file 1C. Statistical analysis Data were harmonized to omit ambiguous SNPs with non-concordant alleles and palindromic SNPs with ambiguous minor allele frequency (>0.42 and <0.58) were removed from the analysis. The primary MR analyses were performed by the multiplicative random-effects inverse-variance weighted (IVW) method, which provides the most precise estimates though assuming that all SNPs are valid instruments. The analysis of rs1229984 for alcohol consumption was conducted by the Wald method. Estimates for each association from different sources were combined using fixed-effects meta-analysis, and the heterogeneity of the associations from different data sources was evaluated by the I2 statistic. Heterogeneity among SNPs’ estimates in each association was assessed by Cochran’s Q value. Multivariable MR analyses were conducted to mutually adjust for smoking initiation and alcohol consumption. To detect potential unbalanced pleiotropy (horizontal pleiotropy) and examine the consistency of the associations, three sensitivity analyses including the weighted median (Yavorska and Burgess, 2017), MR-Egger (Burgess and Thompson, 2017), and MR pleiotropy residual sum and outlier (MR-PRESSO) (Verbanck et al., 2018) analyses were performed. The weighted median method can provide consistent estimates when more than 50% of the weight comes from valid instrument variants (Yavorska and Burgess, 2017). The MR-Egger intercept test can detect unmeasured pleiotropy, and MR-Egger regression can generate estimates after accounting for horizontal pleiotropy albeit with less precision (Burgess and Thompson, 2017). The MR-PRESSO method can identify SNP outliers and provide results identical to that from IVW after removal of outliers (Verbanck et al., 2018). The F-statistic was estimated to quantify instrument strength, and an F-statistic >10 suggested a sufficiently strong instrument. Power analysis was performed using an online tool (Brion et al., 2013). The Benjamini-Hochberg correction that controls the false discovery rate was applied to correct for multiple testing. The association with a nominal p-value <0.05 but Benjamini-Hochberg adjusted p-value >0.05 was regarded suggestive, and the association with a Benjamini-Hochberg adjusted p-value <0.05 was deemed significant. All analyses were two-sided and performed using the TwoSampleMR (Hemani et al., 2018), MendelianRandomization (Yavorska and Burgess, 2017), and MRPRESSO R packages (Verbanck et al., 2018) in R software 4.1.2. Results The F-statistic for each genetic variant was above 10, suggesting a good strength of used genetic instruments (Supplementary file 1B). Most associations were well powered (Supplementary file 1D). For smoking initiation, there was 80% power to detect the smallest odds ratio (OR) ranging from 1.08 to 1.40 for included outcomes. Although power was lower for alcohol consumption, it was adequate to detect a moderate effect size for most common gastrointestinal diseases. Smoking and gastrointestinal diseases Genetic predisposition to smoking initiation was associated with 20 of the 24 studied gastrointestinal diseases, and all these associations remained after multiple comparison correction (Table 1 and Supplementary file 1E). In detail, genetic liability to smoking initiation was positively associated with seven upper gastrointestinal diseases: gastroesophageal reflux (OR, 1.28; 95% confidence interval [CI], 1.20–1.37; p=4.09 × 10−14), esophageal cancer (OR, 1.67; 95% CI, 1.24–2.25; p=6.84 × 10−4), gastric ulcer (OR, 1.54; 95% CI, 1.37–1.72; p=3.83 × 10−14), duodenal ulcer (OR, 1.53; 95% CI, 1.34–1.75; p=8.47 × 10−10), acute gastritis (OR, 1.29; 95% CI, 1.09–1.53; p=0.003), chronic gastritis (OR, 1.33; 95% CI, 1.18–1.49; p=1.55 × 10–6), and gastric cancer (OR, 1.42; 95% CI, 1.13–1.79; p=0.003); genetic liability to smoking initiation was positively associated with four lower gastrointestinal diseases: irritable bowel syndrome (OR, 1.22; 95% CI, 1.12–1.32; p=3.50 × 10−6), diverticular disease (OR, 1.25; 95% CI, 1.18–1.33; p=5.23 × 10−14), Crohn’s disease (OR, 1.25; 95% CI, 1.11–1.40; p=3.03 × 10−4), and ulcerative colitis (OR, 1.15; 95% CI, 1.04–1.26; p=0.004); genetic liability to smoking initiation was positively associated with eight hepatobiliary and pancreatic diseases: non-alcoholic fatty liver disease (OR, 1.49; 95% CI, 1.26–1.76; p=3.82 × 10−6), alcoholic liver disease (OR, 1.99; 95% CI, 1.65–2.41; p=1.49 × 10−12), cirrhosis (OR, 1.68; 95% CI, 1.40–2.02; p=3.39 × 10−8), liver cancer (OR, 1.57; 95% CI, 1.13–2.17; p=0.007), cholecystitis (OR, 1.47; 95% CI, 1.29–1.68; p=4.71 × 10−9), cholelithiasis (OR, 1.20; 95% CI, 1.13–1.27; p=5.75 × 10−9), acute pancreatitis (OR, 1.39; 95% CI, 1.23–1.56; p=6.71 × 10−8), and chronic pancreatitis (OR, 1.38; 95% CI, 1.17–1.64; p=1.79 × 10−4); genetic liability to smoking initiation was positively associated with acute appendicitis (OR, 1.15; 95% CI, 1.08–1.23; p=1.27 × 10−5). Results were consistent in sensitivity analyses. An indication of horizontal pleiotropy was observed in the analysis of esophageal cancer in the FinnGen study (p for MR-Egger intercept <0.05, Supplementary file 1F). Although MR-PRESSO detected one to three outliers, the associations persisted and remained significant after removal of these out-lying SNPs (Supplementary file 1F). When using the genetic variants for smoking initiation based on data without the UK Biobank and 23andMe studies, the associations attenuated slightly albeit remained significant after multiple comparisons (Supplementary file 1L and Supplementary file 1G). All associations were replicated in the supplementary analysis of the lifetime smoking index (Supplementary file 1G). After correcting for multiple testing, genetically predicted lifetime smoking index was significantly associated with 17 of 24 gastrointestinal diseases, where the patterns of associations were generally similar to the analysis for smoking initiation (Supplementary file 1M and Supplementary file 1G). In distinction to the analysis of smoking initiation, genetically predicted lifetime smoking index was not significantly associated with acute gastritis, gastric cancer, Crohn’s disease, and ulcerative colitis, whereas genetically predicted lifetime smoking index was significantly associated with pancreatic cancer (OR, 2.09; 95% CI, 1.30–3.36). Table 1 Associations of genetic predisposition to smoking initiation with 24 gastrointestinal diseases in univariable and multivariable Mendelian randomization analyses. DiseaseTotal casesTotal controlsUVMRMVMR adjusted for alcohol consumptionOR (95% CI)p ValueI2 (95% CI)OR (95% CI)p ValueUpper gastrointestinal diseasesGastroesophageal reflux34,135634,6291.28 (1.20, 1.37)4.09 × 10-14*46.241.65 (1.35, 2.02)1.38 × 10-6*Esophageal cancer1130702,1161.67 (1.24, 2.25)6.84 × 10-4*22.684.78 (2.10, 10.90)1.97 × 10-4*Gastric ulcer8651666,8791.54 (1.37, 1.72)3.83 × 10-14*44.961.95 (1.40, 2.71)7.31 × 10-5*Duodenal ulcer5713666,8791.53 (1.34, 1.75)8.47 × 10-10*0.001.64 (1.07, 2.52)0.024Acute gastritis3048643,4781.29 (1.09, 1.53)0.003*0.001.54 (0.91, 2.62)0.106Chronic gastritis7975643,4781.33 (1.18, 1.49)1.55 × 10-6*77.041.33 (0.96, 1.86)0.091Gastric cancer1608701,4721.42 (1.13, 1.79)0.003*0.002.29 (1.14, 4.59)0.020Lower gastrointestinal diseasesIrritable bowel disease15,718641,4891.22 (1.12, 1.32)3.50 × 10-6*11.841.43 (1.10, 1.85)0.008*Celiac disease4808631,7000.82 (0.66, 1.02)0.0710.000.87 (0.53, 1.43)0.590Diverticular disease50,065587,9691.25 (1.18, 1.33)5.23 × 10-14*67.291.56 (1.30, 1.87)1.41 × 10-6*Crohn’s disease10,846645,7181.25 (1.11, 1.40)3.03 × 10-4*0.001.48 (1.01, 2.16)0.042Ulcerative colitis16,770651,2551.15 (1.04, 1.26)0.004*0.000.94 (0.71, 1.25)0.677Colorectal cancer9519686,9531.03 (0.92, 1.14)0.63229.941.03 (0.76, 1.39)0.841Hepatobiliary and pancreatic diseasesNon-alcoholic fatty liver disease3242707,6311.49 (1.26, 1.76)3.82 × 10-6*0.002.11 (1.15, 3.88)0.016*Alcoholic liver disease2955680,3691.99 (1.65, 2.41)1.49 × 10-12*92.682.26 (1.26, 4.03)0.006Cirrhosis5904706,2001.68 (1.40, 2.02)3.39 × 10-8*0.001.92 (1.06, 3.47)0.032Liver cancer714702,0081.57 (1.13, 2.17)0.007*0.001.96 (0.73, 5.25)0.183Cholangitis1708664,7491.02 (0.80, 1.29)0.8920.001.31 (0.61, 2.84)0.489Cholecystitis5893664,7491.47 (1.29, 1.68)4.71 × 10-9*84.722.38 (1.57, 3.60)4.14 × 10-5*Cholelithiasis42,510664,7491.20 (1.13, 1.27)5.75 × 10-9*0.001.33 (1.02, 1.73)0.035Acute pancreatitis6634679,7131.39 (1.23, 1.56)6.71 × 10–8*79.711.55 (1.04, 2.31)0.031Chronic pancreatitis3173679,7131.38 (1.17, 1.64)1.79 × 10–4*0.001.27 (0.74, 2.16)0.384Pancreatic cancer1643701,4721.00 (0.79, 1.26)0.99967.212.08 (1.06, 4.10)0.034OtherAcute appendicitis25,361690,1491.15 (1.08, 1.23)1.27 × 10–5*0.001.15 (0.92, 1.44)0.221 * Significant association after multiple testing. UVMR, univariable Mendelian randomization; MVMR, multivariable Mendelian randomization; OR, odds ratio; CI, confidence interval. *Significant association after multiple testing. In multivariable MR analysis adjusted for genetically predicted alcohol consumption, the associations between genetically predicted smoking initiation and gastrointestinal diseases were consistent with that from univariable MR analysis (Table 1 and Supplementary file 1H). However, the associations became stronger with wider CIs, in particular the associations for gastrointestinal reflux, esophageal cancer, gastric ulcer, irritable bowel syndrome, diverticular disease, non-alcoholic fatty liver disease, alcoholic liver disease, and cholecystitis (Table 1). In addition, the association for pancreatic cancer became suggestive significant from null. Alcohol consumption and gastrointestinal diseases Genetically predicted alcohol consumption was nominally positively associated with esophageal cancer (OR, 2.86; 95% CI, 1.18–6.91; p=0.020), duodenal ulcer (OR, 1.92; 95% CI, 1.23–3.00; p=0.004), alcoholic liver disease (OR, 14.35; 95% CI, 7.69–26.81; p=6.32 × 10−17), cirrhosis (OR, 2.96; 95% CI, 1.50–5.85; p=0.002), and chronic pancreatitis (OR, 2.96; 95% CI, 1.80–4.89; p=2.13 × 10−5), and nominally inversely associated with irritable bowel disease (OR, 0.73; 95% CI 0.57–0.93; p=0.012) (Table 2). After Benjamini-Hochberg correction, the associations for duodenal ulcer, alcoholic liver disease, cirrhosis, and chronic pancreatitis remained (Supplementary file 1E). Results were consistent in sensitivity analyses, and no horizontal pleiotropy was detected (Supplementary file 1I). One outlier was detected in the analysis of duodenal ulcer in the FinnGen study, and the association slightly changed after removal of this outlier (Supplementary file 1I). Results were consistent in the sensitivity analysis, where the genetic associations with alcohol consumption were obtained from the genome-wide association analysis excluding the UK Biobank and 23andMe studies (Supplementary file 1N and Supplementary file 1G). The associations were directionally consistent albeit with wider CIs in the analysis, where alcohol consumption was instrumented by rs1229984 (Supplementary file 1J). The associations for alcoholic liver disease, cirrhosis, and chronic pancreatitis persisted after adjustment for genetic liability to smoking initiation and multiple testing correction (Table 2 and Supplementary file 1H). Table 2 Associations of genetically predicted alcohol consumption with 24 gastrointestinal diseases in univariable and multivariable Mendelian randomization analyses. DiseaseTotal casesTotal controlsUVMRMVMR adjusted for smoking initiationOR (95% CI)p ValueI2 (95% CI)OR (95% CI)p ValueUpper gastrointestinal diseasesGastroesophageal reflux34,135634,6290.99 (0.81, 1.21)0.89346.240.88 (0.72, 1.08)0.219Esophageal cancer1130702,1162.86 (1.18, 6.91)0.02022.681.28 (0.59, 2.82)0.533Gastric ulcer8651666,8791.30 (0.95, 1.77)0.09844.961.06 (0.77, 1.47)0.721Duodenal ulcer5713666,8791.92 (1.23, 3.00)0.004*0.001.54 (1.01, 2.34)0.045Acute gastritis3048643,4780.99 (0.58, 1.69)0.9600.000.88 (0.52, 1.48)0.621Chronic gastritis7975643,4781.33 (0.90, 1.95)0.14777.041.33 (0.93, 1.89)0.115Gastric cancer1608701,4721.57 (0.75, 3.30)0.2330.001.59 (0.79, 3.21)0.194Lower gastrointestinal diseasesIrritable bowel disease15,718641,4890.73 (0.57, 0.93)0.01211.840.74 (0.57, 0.97)0.027Celiac disease4808631,7000.69 (0.44, 1.07)0.0970.001.04 (0.64, 1.68)0.887Diverticular disease50,065587,9690.95 (0.79, 1.13)0.55367.290.94 (0.79, 1.13)0.527Crohn’s disease10,846645,7180.91 (0.62, 1.32)0.6130.000.74 (0.53, 1.05)0.088Ulcerative colitis16,770651,2551.11 (0.82, 1.50)0.5090.000.88 (0.67, 1.15)0.358Colorectal cancer9519686,9531.09 (0.76, 1.55)0.64929.941.28 (0.95, 1.72)0.098Hepatobiliary and pancreatic diseasesNon-alcoholic fatty liver disease3242707,6311.20 (0.63, 2.28)0.5740.000.99 (0.54, 1.79)0.962Alcoholic liver disease2955680,36914.35 (7.69, 26.81)6.32 × 10-17*92.689.60 (5.28, 17.46)1.25 × 10-13*Cirrhosis5904706,2002.96 (1.50, 5.85)0.002*0.002.41 (1.29, 4.52)0.006*Liver cancer714702,0081.16 (0.43, 3.11)0.7750.000.76 (0.29, 2.02)0.585Cholangitis1708664,7490.96 (0.44, 2.08)0.9120.000.72 (0.33, 1.55)0.397Cholecystitis5893664,7491.36 (0.91, 2.03)0.13284.720.96 (0.64, 1.45)0.862Cholelithiasis42,510664,7491.02 (0.75, 1.39)0.8780.001.03 (0.79, 1.35)0.801Acute pancreatitis6634679,7131.36 (0.91, 2.03)0.12879.711.17 (0.78, 1.75)0.456Chronic pancreatitis3173679,7132.96 (1.80, 4.89)2.13 × 10-5*0.003.24 (1.86, 5.64)3.18 × 10-5**Pancreatic cancer1643701,4720.63 (0.32, 1.26)0.19367.210.79 (0.40, 1.56)0.496OtherAcute appendicitis25,361690,1490.80 (0.63, 1.01)0.0630.000.77 (0.61, 0.97)0.024 * Significant association after multiple testing. UVMR, univariable Mendelian randomization; MVMR, multivariable Mendelian randomization; OR, odds ratio; CI, confidence interval. Discussion We conducted a comprehensive MR investigation to examine the causal role of smoking and alcohol consumption in 24 gastrointestinal diseases, and the result summary of this comprehensive analysis is shown in Figure 2 and Supplementary file 1K. We found robust associations between genetic predisposition to smoking and increased risk of 15 gastrointestinal outcomes independent of alcohol consumption, showing an extensive impact on gastrointestinal health. In contrast, genetically predicted alcohol consumption was robustly and predominantly associated with increased risk of liver and pancreatic diseases, including alcoholic liver disease, cirrhosis, and chronic pancreatitis after adjustment for smoking. Figure 2 Download asset Open asset Summary of associations of genetically predicted smoking initiation, lifetime smoking, and alcohol consumption with 24 gastrointestinal diseases. UVMR, univariable Mendelian randomization; MVMR, multivariable Mendelian randomization. The numbers in the box are the odds ratios for associations of exposure for gastrointestinal diseases. The association with a p-value <0.05 but Benjamini-Hochberg adjusted p-value >0.05 was regarded suggestive, and the association with a Benjamini-Hochberg adjusted p-value <0.05 was deeme