Morning chronotype and digestive tract cancers: Mendelian randomization study

Morning chronotype has been associated with a reduced risk of prostate and breast cancer. However, few studies have examined whether chronotype is associated with digestive tract cancer risk. We conducted a Mendelian randomization (MR) study to assess the associations of chronotype with major digestive tract cancers. A total of 317 independent genetic variants associated with chronotype at the genome-wide significance level (P < 5 × 10 –8) were used as instrumental variables from a genome-wide meta-analysis of 449 734 individuals. Summary-level data on overall and six digestive tract cancers, including esophageal, stomach, liver, biliary tract, pancreatic and colorectal cancers, were obtained from the UK Biobank (11 952 cases) and FinnGen (7638 cases) study. Genetic liability to morning chronotype was associated with reduced risk of overall digestive tract cancer and cancers of stomach, biliary tract and colorectum in UK Biobank. The associations for the overall digestive tract, stomach and colorectal cancers were directionally replicated in FinnGen. In the meta-analysis of the two sources, genetic liability to morning chronotype was associated with a decreased risk of overall digestive tract cancer (odds ratio [OR] 0.94, 95% confidence interval [CI]: 0.90-0.98), stomach cancer (OR 0.84, 95% CI: 0.73-0.97) and colorectal cancer (OR 0.92, 95% CI: 0.87-0.98), but not with the other studied cancers. The associations were consistent in multivariable MR analysis with adjustment for genetically predicted sleep duration, short sleep, insomnia and body mass index. The study provided MR evidence of inverse associations of morning chronotype with digestive tract cancer, particularly stomach and colorectal cancers.

Grant/Award Number: 2018-00123; Swedish Heart-Lung Foundation (Hjärt-Lungfonden), Grant/Award Number: 20210351 other studied cancers. The associations were consistent in multivariable MR analysis with adjustment for genetically predicted sleep duration, short sleep, insomnia and body mass index. The study provided MR evidence of inverse associations of morning chronotype with digestive tract cancer, particularly stomach and colorectal cancers.

K E Y W O R D S
chronotype, colorectal cancer, digestive system cancer, gastric cancer, Mendelian randomization What's new?
Cancers of the digestive tract are associated with various traditional risk factors, including smoking and obesity. A possible novel risk factor may be chronotype, an individual's natural tendency to sleep at a particular time. In our study, the authors assessed potential associations between digestive tract cancers and chronotype using Mendelian randomization analysis of UK Biobank and FinnGen study data. Overall risk of digestive tract cancer, stomach cancer and colorectal cancer was found to be inversely associated with genetic liability to morning chronotype. Analyses further indicate that the associations are independent of sleep duration, short sleep, insomnia and body mass index.

| INTRODUCTION
Digestive tract cancer with 4.8 million new cases (26% of global cancer incidence) caused 3.4 million premature deaths (35% of cancer-related mortality) in 2018 worldwide, which poses a large global disease burden. 1 Previous studies identified several risk factors for digestive tract cancer, including obesity, cigarette smoking, alcohol consumption and hepatitis B virus infection. [1][2][3][4][5] Primary and secondary prevention strategies targeting at above risk factors may play an important role in preventing and controlling gastrointestinal malignancies. 1 However, except for the above traditional risk factors, novel factors are scarcely studied, such as chronotype, possibly influencing the risk of gastrointestinal cancers.
Chronotype is the natural propensity for the individual to sleep at a particular time. Two extreme types of chronotype, termed morningness (an early bird) and eveningness (a night owl), means having an advanced and delayed sleep period, respectively. Evening chronotype associates with less physical activity 6 and unhealthy dietary habits 7 and thus could potentially have adverse health effects and increase the risk of cancer. 8 A matched case-control study revealed that patients with gastroenteropancreatic neuroendocrine tumors had more commonly an evening chronotype compared to healthy controls. 9 Another study using metagenomic sequencing analysis found that morning chronotype was associated with a decreased abundance of Alistipes and an elevated abundance of Lachnospira. 10 These two gut microbiome genera may play a role in the development of cancers in the gastrointestinal tract, such as colorectal cancer. 11,12 Even though a few studies linked chronotype to digestive tract cancer, no studies examined this association.
Mendelian randomization (MR) is an epidemiological approach that can strengthen the causal inference by utilizing genetic variants as instrumental variables for an exposure. 13 The design has two merits. First, it can minimize confounding effects since genetic variants are randomly assorted at conception and therefore not correlated with environmental or self-adopted factors that are usually confounders in the association between the exposure and outcome. 13 Second, the approach can diminish reverse causality because the onset and progression of disease cannot modify the germline genotype. 13 Previous MR studies revealed inverse associations of morningness with the risk of prostate and breast cancers. [14][15][16] Nevertheless, no studies were conducted to examine the associations between chronotype and digestive tract cancers. Here, we conducted an MR investigation to explore the causality of these associations.

| Outcome data sources
Genetic associations with six digestive tract cancers (esophageal, stomach, liver, biliary tract, pancreatic and colorectal cancers) and overall digestive tract cancer were obtained from the UK Biobank study and the FinnGen study. 17 The UK Biobank study is an ongoing cohort study that was initiated by recruiting about 500 000 adults between 2006 and 2010. Our study was based on data from a total of up to 367 542 individuals followed up until April 2022 after removal of those with non-Caucasian ethnicity (to reduce population stratification bias), sex mismatch, excess heterozygosity and low genotype call rate, and those related by third degree or higher. We defined 11 952 digestive tract cancer cases including 1339 esophageal, 1086 stomach, 503 liver, 656 biliary tract, 1414 pancreatic and 7543 colorectal cancer patients. These cases were diagnosed by using codes of International Classification of Diseases (9th and 10th revisions) and selfreported information verified by interview with a nurse from national registries (Table S1). For the FinnGen study, we used data from the R6 release that includes 309 154 Finnish individuals after excluding participants with ambiguous gender, high genotype missingness (>5%), excess heterozygosity (±4 SD) and non-Finnish ancestry. 17 A total of 7638 patients with digestive tract cancers were ascertained, including 358 esophageal, 889 stomach, 442 liver, 157 biliary tract, 881 pancreatic and 4401 colorectal cancer patients. Similarly, cancer cases were defined by codes of International Classification of Diseases (8th, 9th and 10th revisions) with information from nationwide registries (Table S2). The associations were adjusted for age, sex and 10 genetic principal components in both data sources.

| Genetic instrument selection
Single-nucleotide polymorphisms (SNPs) associated with chronotype at the genome-wide significance level (P < 5 Â 10 À8 ) were extracted from a genome-wide meta-analysis of 449 734 individuals of European-ancestry from the UK Biobank and 248 098 individuals of European-ancestry from 23andMe. 18 Linkage disequilibrium was estimated among selected SNPs based on the 1000 Genomes European reference panel. 19 SNPs in linkage disequilibrium (r 2 > .01) were excluded and the SNP with the smallest P value for the genetic association with chronotype was retained. After clumping, 317 SNPs in autosomal chromosomes were selected as instrumental variables (Table S3). The genetic score of these SNPs showed no associations with sleep duration and quality as well as insomnia. 18 In the main analysis, we used the estimates (beta and corresponding SE coefficients) of the associations between SNPs and chronotype from the metaanalysis of UK Biobank and 23andMe to increase the power. In addition, we used the genetic association estimates obtained from only 23andMe to minimize the sample overlap in the sensitivity analysis (Table S3). Genetic associations were scaled to one point increase in chronotype category (À2 for definitely evening, À1 for more evening than morning, 0 for unknown, 1 for more morning than evening and 2 for definitely morning) and estimated by a linear mixed model with adjustment for age, sex, study center, genotyping array and genetic principal components. 18

| Sleep features and body mass index data sources
Summary-level data on sleep duration and short sleep (<7 hours/day) were obtained from a genome-wide association analysis of 446 118 individuals of European ancestry (106 192 short sleep cases compared to 305 742 controls with 7-to 8-hour sleep duration). 20 Genetic associations were adjusted for age, sex, 10 principal components of ancestry, genotyping array and genetic correlation matrix. 20 Summary-level data on insomnia were extracted from a genome-wide association study of 453 379 individuals of European ancestry (345 022 cases with any insomnia symptoms and 108 357 controls). 21 Genetic associations were adjusted for age, sex, 10 principal components of ancestry and genotyping array. 21 Genetic associations with body mass index (BMI) were extracted from the genome-wide analysis in the Genetic Investigation of ANthropometric Traits consortium including 806 834 individuals with adjustment for age, sex and genetic principal components. 22

| Statistical analysis
We used the random-effects multiplicative inverse variance weighted method as the primary analysis to estimate the association between genetic liability to chronotype and the risk of digestive tract cancers.
Given that the analysis is sensitive to outliers and horizontal pleiotropy, four sensitivity analyses, including the weighted median, 23 MR-Egger, 24 MR-PRESSO 25 and contamination mixture 26 methods, were utilized to examine the consistency of the results and detect horizontal pleiotropy if any. We described assumptions and strengths of above methods in Table 1 (Figure 1). The associations for cancers of overall digestive tract, stomach and colorectum but not for biliary tract cancer were directionally consistent in the FinnGen study ( Figure 1). In the combined analysis of the two data sources, genetic liability to morning chrono-  (Table S4). Genetic liability to being a morning person was not associated with cancers of the esophagus, liver or pancreas ( Figure 1).
The associations remained consistent but with wider CIs in the sensitivity analyses ( however, the associations were stable after removal of the outlier ( As a novel health-related factor, chronotype has been studied in relation to cancer risk in a few studies in recent decades. Observational evidence is consistent on the association between morning chronotype and reduced risk of breast and prostate cancers. 29,30 These associations were further strengthened in MR studies. [14][15][16]31 Being an early bird was also associated with the lower risk of other hormone-related cancers, like ovarian and endometrial cancers, 32,33 and lung cancer. 34 However, little data were available on association between chronotype and digestive tract cancer. A case-control study including a total of 318 participants found that patients with gastroenteropancreatic neuroendocrine tumors had a higher percentage of having evening chronotype compared to the healthy controls. 9 That study further showed that a higher chronotype score indicating a higher adherence to morningness was also associated with lower risk of metastasis, grading G2 and progressive disease. 9  Abbreviations: CI, confidence interval; OR, odds ratio.
Note: The P intercept is the P for MR-Egger intercept test and a P intercept < .05 indicates horizontal pleiotropy. One outlier was identified in MR-PRESSO analysis of colorectal cancer in UK Biobank and that of biliary tract cancer in FinnGen.
BMI, the observed associations might still be biased by residual confounding, which hindered the examination on the causal impact of chronotype on gastroenterological cancer. Our MR results supported this observational study and strengthened the causality of the inverse association between morning chronotype and digestive tract cancer.
Notably, we further examined the associations with gastrointestinal tract cancers by site and found strong effects on stomach and colorectal cancers. We also observed an inverse association between morning chronotype and biliary tract cancer in UK Biobank, but not in the combined analysis, which was caused by the heterogeneous association observed in FinnGen. Thus, larger studies are needed to further explore the association between morning chronotype and biliary tract cancer.
Some underlying mechanisms in support of morningness and the reduced risk of digestive tract cancer have been proposed albeit scarcely examined. First, chronotype has been linked to the profile of gut microbiota that may influence the development of digestive tract cancer. 10,35 A detailed appraisal of mediation of gut microbiota in the association between chronotype and digestive tract cancer is of importance for clinical and therapeutic work. Second, genetic liability to morning chronotype has been associated with increased intake of foods known to constitute a healthy diet (eg, fresh fruit and bran cereals) and lower intake of unhealthy foods and beverages (eg, processed meat and beer). 36  confinement to the European population may limit the generalizability of our findings to other populations. Second, we may overlook the moderate associations due to inadequate power caused by a small number of cases for certain cancers even though we combined two data sources. Third, the interplay of chronotype and diet cannot be studied in our study based on summary-level statistic data. 44 In summary, this MR study found the inverse associations between morning chronotype and digestive tract cancers, in particular cancers of stomach and colorectum. These findings suggest a causal role of being an early bird instead of a night owl in lowering digestive tract cancer. The mechanism behind these associations needs further study.