Examining complex cancer etiologies within the Korean population: A high-throughput multivariable Mendelian randomization study

Examining complex cancer etiologies within the Korean population: A high-throughput multivariable Mendelian randomization study | Research Square window.SnipcartSettings = { analytics: { enabled: false } }; (function() { var accessVector = localStorage.getItem('access_vector') || ''; window.dataLayer = window.dataLayer || []; if (accessVector) { window.dataLayer.push({ user: { profile: { profileInfo: { snid: accessVector } } } }); } })(); (function(w,d,s,l,i){w[l]=w[l]||[];w[l].push({'gtm.start':new Date().getTime(),event:'gtm.js'});var f=d.getElementsByTagName(s)[0],j=d.createElement(s),dl=l!='dataLayer'?'&l='+l:'';j.async=true;j.src='https://www.googletagmanager.com/gtm.js?id='+i+dl;f.parentNode.insertBefore(j,f);})(window,document,'script','dataLayer','GTM-K279D39R'); Browse Preprints In Review Journals COVID-19 Preprints AJE Video Bytes Research Tools Research Promotion AJE Professional Editing AJE Rubriq About Preprint Platform In Review Editorial Policies Our Team Advisory Board Help Center Sign In Submit a Preprint Cite Share Download PDF Research Article Examining complex cancer etiologies within the Korean population: A high-throughput multivariable Mendelian randomization study Keum Ji Jung, Wes Spiller, Dae Sub Song, Jong Won Shin, Kyoungho Lee, and 1 more This is a preprint; it has not been peer reviewed by a journal. https://doi.org/ 10.21203/rs.3.rs-4249634/v1 This work is licensed under a CC BY 4.0 License Status: Posted Version 1 posted You are reading this latest preprint version Abstract Background Despite an extensive body of observational research related to risk factors for cancer incidence, it is unclear whether the estimated associations are causal or a result of unmeasured confoundingfactors. To consider this possibility, this study explored a range of candidate epidemiological factors associated with the onset of cancer within a Mendelian randomization framework. Methods Multivariate Mendelian randomization (MVMR) analyses were conducted using data from the Korean Cancer Prevention Study-II Biobank and the Korean Genome Epidemiologic Study. Analyses were performed to investigate 13 cancer-related risk factors and 13 types of cancer. Initially, univariate Mendelian randomization analyses were performed for each factor, estimating its association with cancer. Subsequently, a set of factors was explored using MVMR. Results By analyzing factors related to the onset of cancer, it was determined that smoking is associated with lung cancer, while hepatitis B surface antigen (HBsAg) positivity is significantly linked to gastric cancer, liver cancer, and cervical cancer. PSA levels are estimated to be causally related to prostate cancer, while bilirubin has emerged as a novel factor showing a positive association with lung cancer. To confirm the causal effect between HBsAg and cancer, a MVMR was conducted, controlling for bilirubin and gamma-glutamyl transferase. The results indicated a positive association between HBsAg and cervical cancer, liver cancer, and lung cancer. Conversely, breast cancer and pancreatic cancer showed a negative association. In the case of breast cancer, individuals with HBsAg at the age of over 50 years exhibited a significantly lower risk, with an odds ratio of 0.87 ( P = 3.07 × 10 -16 ). Conclusions Smoking status, HBsAg, and PSA levels replicated findings from previous studies suggesting causal relationships. However, bilirubin and HBsAg demonstrated positive causal associations with some cancers, while HBsAg exhibited negative associations with other cancers. Further research is warranted to explore the cancer-specific causality of HBsAg. Cancer Hepatitis B antigen Bilirubin Gamma-glutamyl transferase Figures Figure 1 Figure 2 Figure 3 Background Through past observational studies, various research reports on factors related to cancer development have been published. The two most significant factors exhibiting the strongest associations with cancer incidence are smoking and lung cancer[ 1 ] and hepatitis B surface antigen (HBsAg) and liver cancer[ 2 ]. In the case of smoking, investigating its relevance not only to lung cancer but also to other types of cancer is meaningful. Similarly, it is crucial to explore how HBsAg influences cancers beyond its association with liver cancer. The association between identified risk factors and cancer incidence, as revealed through previous observational studies, faced limitations in drawing causal conclusions due to confounding variables, reverse causation, measurement errors, and other reasons. To overcome such limitations, Mendelian randomization (MR)[ 3 ] has been proposed as a method to estimate effects without residual confounding. MR is an epidemiological approach that uses genetic variants as a tool to strengthen causal inference in the association between an exposure and an outcome[ 4 ]. One of the key advantages of MR is its ability to minimize confounding since genetic variants are randomly assigned at conception and are therefore unrelated to environmental and self-adopted factors that typically act as confounders. In MR, germline genetic variation is used as a proxy variable for modifiable exposures concerning causal inference about outcomes[ 5 , 6 ]. This design also helps to mitigate the issue of reverse causality, as the fixed alleles used in MR analyses are not influenced by the onset or progression of disease. The fundamental principle of MR is that differences between individuals due to genetic variations are not influenced by the outcome. Confounding or reverse causation biases that distort observational findings. Therefore, the natural 'randomization' of alternate alleles (mutation forms) can be likened to the random allocation of treatment conducted in randomized controlled trials. This means experiencing an average exposure difference without being different in confounding factors associated with alternate genetic variants[ 4 , 5 ]. Cancer is a multifactorial disease, and different risk factors exist for each cancer type. Therefore, it is not appropriate to discuss causality using only one factor. Using a multivariable Mendelian randomization (MVMR) method that takes into account the genetic correlation of several variables that can influence the main risk factor, we aimed to determine the causal impact on cancer. This MR aimed to investigate the causal relationships between various exposure factors and types of cancer among Koreans using novel data from two large-scale biobanks. Methods Study population and data sources Korean Cancer Prevention Study (KCPS)-II Biobank. The Korean Cancer Prevention Study (KCPS)-II Biobank is a prospective population-based cohort comprising adults recruited from 18 health examination centers across South Korea. Detailed descriptions of the KCPS-II have been previously described[ 7 ]. The cohort consisted of 156,701 participants (94,840 men and 61,861 women) who underwent routine health assessments between 2004 and 2013, provided blood samples, and provided informed consent for long-term follow-up. At baseline, participants provided information on sociodemographic factors, alcohol consumption, smoking habits, diet, exercise, and past medical history through a questionnaire. This study included 153,971 participants who completed genetic analysis out of the 156,701 total participants (IRB no. 4-2023-1722). Korean Genome Epidemiology Study (KoGES) Biobank The Korean Genome Epidemiologic Study (KoGES) Biobank established encompasses a cohort built from 2004 to 2013, incorporating community cohorts in the Ansan, Anseong, and Urban cohorts and Rural cohorts, totaling approximately 210,000 individuals. The KoGES participants were men and women aged 40 and above. The KoGES comprises 195,544 participants (69,579 men and 125,965 women) who provided blood samples and gave informed consent for long-term follow-up. This study included 195,544 individuals for whom genetic testing data were available. Detailed descriptions of KoGES have been previously published[ 8 ]. Exposure and Outcome This study selected 13 exposure factors related to cancer incidence from the KCPS-II Biobank and conducted genome-wide association study (GWAS) analyses for each, incorporating relevant genetic factors into the research. Alcohol consumption, serum bilirubin, body mass index (BMI), waist circumference, smoking amount, carcinoembryonic antigen (CEA), fasting blood sugar (FBS), thyroid-stimulating hormone (TSH), prostate-specific antigen (PSA), aspartate aminotransferase (AST), alanine transferase (ALT), gamma-glutamyl transferase (GGT), and hepatitis B surface antigen (HBsAg), which are known risk factors for cancer site, were selected as exposure factors for analysis. Additionally, a comprehensive set of cancers and 13 specific cancer types were chosen from the KoGES, and GWAS analyses were performed for each, investigating the associated genetic factors. The main outcome of this study was cancer incidence. The cancer incidence information about the subjects was confirmed by linking the cancer registration data, a collection of diagnosis records from hospitals. Information on the cancer site, cancer diagnosis date, and histological type was collected in connection with the national cancer registration data. Until December 31, 2020, the cancer registration data of the follow-up National Cancer Center and the cause of death data of the National Statistical Office were used. Statistical analysis Univariable Mendelian randomization This study initially conducted univariable MR analyses using 13 risk factors from the KCPS-II BioBank and 13 cancer types from the KoGES. MR employs genetic variants, commonly single nucleotide polymorphisms (SNPs), as instrumental variables to assess causal effects free from the influence of unmeasured confounding. This methodology is applicable when a specific genetic variant, designated a candidate (IV1), has a robust association with targeted exposure. Additionally, MR assumes the absence of confounders affecting both the genetic variant and the outcome (IV2) and independence from the outcome when considering the exposure (IV3)[ 9 ]. Assuming the assumptions were met, beta values and standard errors (SE) were estimated using the inverse-variance weighting (IVW) method. To derive a Wald ratio estimate for a single genetic instrument, one divides the association between the genetic variant and the outcome by the association between the genetic variant and the exposure[ 10 , 11 ]. In cases involving multiple SNPs, these ratio estimates are frequently amalgamated using a fixed-effects meta-analysis approach[ 10 , 12 ]. Furthermore, considering the presence of pleiotropy, beta values and SEs were estimated using the MR‒Egger method. To determine the existence of pleiotropy, an intercept test was conducted. Multivariable Mendelian randomization (MVMR) The instrumental variable may exhibit pleiotropy, encompassing multiple exposures. To address this, we conducted MR with multiple instrumental variable analysis by creating genetic variables to control for pleiotropy in explaining various exposures[ 13 ]. A new forward selection algorithm has been introduced to help determine the most appropriate MVMR model based on instrument strength[ 14 ]. The goal is to include the maximum number of exposures that have conditional F-statistics between exposure and genetic variants surpassing a threshold value of 10. Initially, all potential exposure pairs are assessed regarding their conditional F-statistics. Subsequently, the average F-statistic for each pair is computed, taking into account the inverse of the absolute difference between the F-statistics within a specific pair. This weighted calculation of the average F-statistic for each pair aims to address situations where two exposures collectively display a high mean F-statistic, primarily influenced by one of the exposures. Initially, we computed weighted conditional F-statistics for pairs of exposures (13 in total) and selected the final three variables. This selection process, including the methodology, is illustrated in Fig. 2 . To achieve this, we presented the F values for each factor included in the model. Analyses were performed using the TwoSampleMR and MVMR R packages. Analyses were conducted in R (version 4.0.3). Results Table 1 shows the general characteristics of the two main cohorts used in this study. The mean age of KoGES participants is more than 10 years older than that of KCPS-II participants. Of the 147 390 KoGES participants used as the outcome data set, 15,062 were affected by cancer, of which incidence rate per 100, 000-person year for prostate cancer (males), breast cancer (females), thyroid cancer, and gastric cancer was more than 100 (Table 2 ). Table 1 General characteristics of Korean biobanks KCPS-II KoGES Men (N = 93 118 ) Women (N = 60 853) Men (N = 53 348) Women (N = 94 042) Mean (SD) Mean (SD) Age, year 42.2 (10.0) 41.0 (11.3) 54.6 (9.1) 53.4 (8.4) Alcohol amount, g/day 22.7 (29.9) 5.8 (13.8) 24.8 (37.5) 6.3 (14.3) Smoking amount, cig./day 11.38 (9.9) 0.7 (3.4) 12.9 (11.5) 0.4 (2.3) Body mass index, kg/m 2 24.4 (2.9) 22.2 (3.1) 24.3 (2.8) 23.8 (3) Waist circumference, cm 84.8 (7.8) 74.2 (8.5) 85.6 (7.6) 79.2 (8.5) CEA, ng/ml 1.9 (2.1) 1.6 (1.9) - - TSH, uU/ml 1.9 (2.7) 2.2 (5.0) 2.0 (4.1) 2.3 (3.8) PSA, ng/ml 1.0 (1.1) - - FBS, ng/ml 93.5 (20.8) 88.3 (16.3) 99.2 (24.2) 93.3 (19.6) AST, u/l 25.5 (20.0) 20.0 (11.3) 26.5 (19.0) 22.9 (19.9) ALT, u/l 30.5 (28.1) 17.3 (17.3) 27.4 (22.4) 20.2 (21.7) GGT, IU/L 47.4 (61.8) 19.1 (20.4) 49.1 (70.2) 22.3 (23.8) Bilirubin, mg/dl 0.9 (0.4) 0.7 (0.3) 0.8 (0.3) 0.7 (0.3) HBsAg (%) 4.4 2.9 - - KCPS-II, Korean Cancer Prevention Study-II; KoGES, Korean Genome Epidemiologic Study; CEA, Carcinoembryonic antigen; TSH, Thyroid Stimulating Hormone; PSA, Prostate Specific Antigen; FBS, Fasting blood sugar; AST, aspartate aminotransferase; ALT, alanine transferase; GGT, Gamma-glutamyl transferase; HBsAg, hepatitis B surface antigen Table 2 Site-specific cancer incidence in Korean Genome Epidemiologic Study Biobank Number of participants Sum of person year Number of incident cancer Incidence per 100,000 PY All cancer 147 390 1 655 481.3 15 062 909.8 Stomach cancer 147 390 1 734 455.6 2 033 117.2 Colorectal cancer 147 390 1 736 811.6 1 652 95.1 Liver cancer 147 390 1 742 497.5 872 50.0 Gallbladder and bile duct cancer 147 390 1 745 626.9 418 23.9 Pancreas cancer 147 390 1 745 992.6 385 22.1 Lung cancer 147 390 1 739 296.6 1 553 89.3 Breast cancer 94 042 1 098 578.0 1 477 134.4 Cervix cancer 94 042 1 106 004.2 146 13.2 Prostate cancer 53 348 635 647.3 1 007 158.4 Thyroid cancer 147 390 1 732 641.6 2 213 127.7 Kidney cancer 147 390 1 746 168.6 317 18.2 Bladder cancer 147 390 1 746 587.6 248 14.2 Non-Hodgkin’s Lymphoma 147 390 1 746 552.6 265 15.2 Supplementary Fig. 1-A presents the univariable (MR analysis results for overall cancer, gastric cancer, colorectal cancer, and liver cancer). TSH exhibited a significant negative correlation with overall cancer incidence. In gastric cancer, HBsAg was the only factor showing a significant positive association. In liver cancer, HBsAg demonstrated a highly significant positive correlation, along with a significant positive association observed for AST. Supplementary Fig. 1-B focuses on gallbladder and other biliary tract cancers, pancreatic cancer, lung cancer, and breast cancer. No significant factors were identified for biliary tract or pancreatic cancers. However, current smoking status exhibited a highly significant positive association with lung cancer. Subsequently, although bilirubin had a wide 95% confidence interval, it showed a remarkably significant positive association. Conversely, breast cancer showed a negative association with HBsAg. As shown in Supplementary Fig. 1-C, which shows renal cancer, cervical cancer, prostate cancer, and thyroid cancer, renal cancer was significantly positively associated with CEA and smoking status. Cervical cancer exhibited a positive association with HBsAg status. Despite a broad confidence interval, prostate cancer was significantly positively associated with PSA. Thyroid cancer was significantly positively associated with CEA and smoking. Supplementary Fig. 1-D depicts univariable MR analyses for 13 exposures concerning bladder cancer and non-Hodgkin's lymphoma. For bladder cancer patients, an inverse association with alcohol consumption was observed. Conversely, non-Hodgkin's lymphoma did not exhibit any significant associations. Figure 1 shows the weighted conditional F-statistics, including differing pairs of exposures. Hence, to estimate effects using MVMR, it was necessary to utilize a subset of exposures. Employing the aforementioned forward selection approach, all conceivable pairs of exposures were assessed, highlighting HBsAg and GGT as the pair with the highest combined conditional F-statistics. Subsequently, in a stepwise manner, additional exposures were considered, leading to the inclusion of bilirubin in a three-exposure MVMR model. Additionally, it displays a heatmap demonstrating the conditional F-statistics for each exposure, portraying the intensity of the weighted mean conditional F-statistic through a gradient. Figure 2 displays the results of MVMR analysis for 13 exposures and 13 cancer types. Specifically, this study presents the MVMR model outcomes controlling for bilirubin and GGT to examine the causal effect of HBsAg. HBsAg was positively associated with lung cancer, liver cancer, and cervical cancer. Conversely, negative associations were identified for pancreatic cancer, breast cancer, and thyroid cancer. Figure 3 shows the analysis results stratified by age group for breast cancer patients. In the case of female subjects aged 50 and above, those with HBsAg exhibited a significantly lower risk, with an odds ratio of 0.87 (0.84–0.90) and P= \(3.07\times {10}^{-16}\) . However, no significant association was observed between HBsAg and breast cancer in females under the age of 50. Discussion This study performed the analysis by independently selecting exposure and outcome datasets using a two-sample MR method. The final estimates were obtained through MVMR, controlling for important pleiotropic variables. By initially exploring topics known for their causality or significant impact on outcome risks due to genetically determined exposures, we aimed to establish a basis for the overall validity of the entire dataset. The first consideration is the relationship between smoking quantity and lung cancer incidence. In this study, smoking quantity exhibited a significant positive causal relationship with lung cancer according to the crude two-sample MR analysis. This finding serves as evidence validating the credibility of results from previous observational studies on the association between smoking and lung cancer[ 15 – 17 ]. Second, the relationship between HBsAg and liver cancer was examined. Compared with the HBsAg-negative group, the HBsAg-positive group showed a significantly positive causal relationship with the risk of liver cancer, as observed in both univariable two-sample MR and MVMR analyses[ 18 ]. Finally, the causal relationship between prostate-specific antigen (PSA) and prostate cancer is a topic that may be subject to debate[ 19 , 20 ]. In this study, PSA exhibited a consistently significant positive causal relationship in univariable two-sample MR. By clearly establishing causal relationships for these three exposures and three cancer types, this study provides a basis for the reliability and validity of the data. Moving forward, the findings newly revealed in this study are discussed. A Review on Cancer Incidence and Its Positive Association with HBsAg Positivity HBsAg showed a positive association not only with liver cancer but also with lung cancer and cervical cancer in this study. Consistent with our findings, a Mendelian randomization (MR) study conducted in East Asian populations in 2022 reported significant positive associations for liver cancer (OR 1.27, p = 0.0001), lung cancer (OR 1.12, p = 0.0002), gastric cancer (OR 1.09, p = 0.0002), and cervical cancer (OR 1.57, p = 0.0001)[ 18 ]. It is believed that HBV protein X, a transcriptional coactivator, plays a crucial role in initiating tumorigenesis by modulating key regulators of apoptosis and interfering with DNA repair pathways and tumor suppressor genes. However, further functional tests are required to elucidate the potential mechanism of action. We observed that in most cancer patients, the expression of the HBX and anti-HBc proteins were greater in cancer tissue specimens than in healthy tissue specimens. It is possible that HBV may be harbored in nonliver cells, potentially inducing local inflammation. Chronic inflammation induced by HBV infection might play a role in the development of cancer; the viral oncogenic HBX protein may play a direct role in the development of cancer. However, we found a weaker association between HBV infection and nonliver cancer than between HBV infection and HCC. Although we confirmed HBV replication and expression in stomach cancer and pancreatic cancer, the immunohistochemistry results suggested that HBX was a pro[ 21 – 23 ]. However, further functional tests are needed to explain the potential mechanism of action. The biological mechanism linking chronic HBV infection with extrahepatic cancers has not yet been fully illustrated. However, it is thought that HBV protein X, a transcriptional coactivator, plays a crucial role in initiating tumorigenesis by modulating key regulators of apoptosis and interfering with DNA repair pathways and tumor suppressor genes[ 22 ]. Moreover, Song et al. reported greater HBV protein X expression in the cancerous part of the tissue specimen than in the healthy part of the same specimen, supporting the oncogenic role of HBV protein X[ 23 ]. Furthermore, some studies have detected HBV DNA in some extrahepatic tissues, including gastric, kidney, gallbladder and pancreatic tissues, suggesting that HBV can initiate and promote tumorigenesis outside the liver[ 18 ]. A Review of HBsAg Positivity and Reduced Cancer Incidence The association between HBsAg and cancer is controversial. HBsAg has shown an inverse relationship with breast cancer, pancreatic cancer, and thyroid cancer. In a Mendelian randomization study conducted in East Asian populations in 2022, no significant association was found between HBsAg and pancreatic cancer, and there have been no studies on breast cancer and thyroid cancer[ 18 ]. When analyzing breast cancer by dividing it into age groups, a greater inverse association was observed in women aged 50 and older, particularly after menopause. Unfortunately, specific references or information on recent studies supporting these results was not available. However, we can propose several hypotheses. First, in relation to immunomodulation, we suggest that HBsAg may stimulate the immune system, leading to enhanced immune surveillance and clearance of cancer cells. This immunomodulatory effect could contribute to a decreased risk of breast cancer in women who test positive for HBsAg. Additionally, we hypothesize that the hepatitis B virus has a direct impact on breast cancer cells. It is postulated that the virus may inhibit the growth or proliferation of breast cancer cells via various mechanisms. Such antiproliferative effects could contribute to the observed decreased risk. Finally, the relationship between HBsAg and breast cancer risk may be influenced by population-specific factors such as genetic variations, environmental exposures, or lifestyle patterns. We speculate that Korean women may exhibit varying susceptibilities to the risk of HBsAg-associated breast cancer. This content suggests the need for further, more specific research. In this study, HBsAg inhibited the occurrence of pancreatic cancer. This result is presumed to be statistically significant due to the large number of participants in the study. However, it can be confirmed that HBsAg does not increase the risk of pancreatic cancer. According to a large-scale cohort study conducted by Abe et al. (2016) involving a total of 20,360 subjects (324,394 person-years) from the Japan Public Health Center (JPHC)-based prospective cohort study cohort II, which was followed for an average of 16 years, 116 patients with newly diagnosed pancreatic cancer were confirmed[ 24 ]. Among the HBsAg-positive participants, there were no patients with pancreatic cancer. Therefore, the JPHC study did not observe a statistically significant association between hepatitis B and the risk of pancreatic cancer[ 24 ]. Additionally, a recent study in Taiwan reported no results for any antibody variants[ 25 ]. In contrast, a previous meta-analysis primarily based on case‒control studies conducted in other Asian countries suggested that HBsAg increases the risk of pancreatic cancer[ 26 ]. According to a prospective cohort study involving over 500,000 adults in 10 regions of China, there was no association between hepatitis B virus (HBV) infection and various cancer types, such as thyroid cancer, skin cancer, and leukemia[ 23 ]. Therefore, additional research is necessary for thyroid cancer treatment. Review of the positive association between serum bilirubin and lung cancer In a study conducted by Horsfall et al. (2020) utilizing the UK Biobank dataset, an inverse correlation between serum bilirubin levels and lung cancer was reported[ 27 ]. Serum bilirubin was indirectly implicated in its role as an antioxidant. The research team identified two genetic factors, rs887829 and rs4149056, that are associated with elevated serum bilirubin. They presented evidence of a decreased incidence of lung cancer based on the genotype, proposing these two SNPs as potent determinants explaining approximately 40% of the variance in serum bilirubin. However, it is essential to note that this investigation targeted a European cohort and revealed associations solely within the subset of smokers[ 27 ]. By conducting research utilizing the same UK Biobank dataset, a study categorized genetic variants into UGT1A1 SNPs and non-UGT1A1 SNPs. Interestingly, when exclusively considering non-UGT1A1 SNPs, it was found that genetically increased circulating bilirubin levels were associated with an elevated risk of lung cancer. Notably, this association was evident only in smokers and specifically for patients with squamous cell carcinoma. Consequently, direct comparison of these subgroup results with our findings may pose challenges, emphasizing the need for subsequent studies to address this aspect[ 28 ]. Our study has several strengths, notably leveraging the MR approach, which effectively mitigates certain confounding factors frequently encountered in epidemiological studies. Additionally, we employed multiple SNPs strongly linked to risk factors. Furthermore, we utilized a homogenous population, thereby reducing the typical heterogeneity observed in genetic studies involving individuals from populations of diverse ancestry. Additionally, this study used two large biobanks of a single ethnic group (Korean), providing significant statistical strength in identifying causal effects. A two-sample study design can maximize the effect of sample size. Sensitivity analyses such as MR‒Egger regression are required for Mendelian randomization studies. In the univariable MR setting, employing such sensitivity analyses ensures more conservative results, especially when weak instruments are used. Although consistent findings were observed when applying sensitivity analyses in the univariable MR setting, significant heterogeneity indicated potential violations of MR assumptions. This warrants further investigation using MVMR. This study has several limitations. One major limitation of the MR design is the presence of horizontal pleiotropy, which occurs when the genetic variants used in the analysis influence the outcomes through pathways unrelated to the exposure of interest. However, in this study, the potential bias due to horizontal pleiotropy is expected to be minimal. This is supported by the findings of the MR‒Egger analysis, which did not indicate any evidence of horizontal pleiotropy. Additionally, consistent results were obtained from a series of sensitivity analyses, further strengthening the robustness of the associations observed. Furthermore, the use of a multivariable MR analysis with mutual adjustment helped to control for potential confounding factors, providing additional evidence for the robustness of the findings. Another limitation is that the overall sample size was large, but the sample size for specific cancers was small. Our analysis included individuals of East Asian origin, who have a high prevalence of chronic HBV infection, so our results may not be generalizable to other ancestry populations. Conclusions In conclusion, this study revealed strong causal relationships between smoking status, HBsAg, and PSA and lung cancer, liver cancer, and prostate cancer, respectively. Notably, HBsAg demonstrated a causal association not only with liver cancer but also with gastric cancer and cervical cancer. However, the inverse relationship observed between HBsAg and breast cancer, as well as the association between serum bilirubin and lung cancer, suggests the need for additional research in these specific contexts. Abbreviations HBsAg: hepatitis B surface antigen MR: Mendelian randomization MVMR: multivariable Mendelian randomization GWAS: genome-wide association study BMI: body mass index CEA: carcinoembryonic antigen FBS: fasting blood sugar TSH: thyroid-stimulating hormone PSA: prostate-specific antigen AST: aspartate aminotransferase ALT: alanine transferase GGT: gamma-glutamyl transferase SNPs: single nucleotide polymorphisms Declarations Availability of data and materials The datasets used and/or analyzed during the current study are available from the corresponding author on reasonable request. Funding This research was supported by the Basic Science Research Program through the National Research Foundation of Korea funded by the Ministry of Education (RS-2023-00239122). Ethical Statement This study protocol was approved by the Institutional Review Board of the Severance Hospital (approval number: 4-2023-1722). Author Contributions Conceptualization: Jee SH., Jung KJ., W. Spiller, Lee K. Data curation: Jee SH., Jung KJ., Song DS., Lee K. Formal analysis: Jung KJ., Song DS., W. Spiller, Funding acquisition: Jee SH. Investigation: Jung KJ, W. Spiller, Shin JW, Song DS., Lee K, Jee SH Methodology: W. Spiller, Jung KJ, Jee SH Project administration: Jee SH., Lee K Supervision: Jee SH., Lee K Writing – original draft: Jung KJ, W. Spiller, Lee K, Jee SH, Writing – review & editing: Jung KJ, W. Spiller, Shin JW, Song DS., Lee K, Jee SH Conflicts of interest Conflict of interest relevant to this article was not reported. References Leiter, A., R.R. Veluswamy, and J.P. Wisnivesky, The global burden of lung cancer: current status and future trends. Nature Reviews Clinical Oncology, 2023. 20 (9): p. 624-639. Kaur, S.P., et al., Hepatocellular carcinoma in hepatitis B virus-infected patients and the role of hepatitis B surface antigen (HBsAg). Journal of clinical medicine, 2022. 11 (4): p. 1126. Skrivankova, V.W., et al., Strengthening the reporting of observational studies in epidemiology using mendelian randomisation (STROBE-MR): explanation and elaboration. bmj, 2021. 375 . Davey Smith, G. and G. Hemani, Mendelian randomization: genetic anchors for causal inference in epidemiological studies. Human molecular genetics, 2014. 23 (R1): p. R89-R98. Plotnikov, D. and J.A. Guggenheim, Mendelian randomisation and the goal of inferring causation from observational studies in the vision sciences. Ophthalmic and Physiological Optics, 2019. 39 (1): p. 11-25. Lawlor, D.A., et al., Mendelian randomization: using genes as instruments for making causal inferences in epidemiology. Statistics in medicine, 2008. 27 (8): p. 1133-1163. Jee, Y.H., et al., Cohort Profile: The Korean Cancer Prevention Study-II (KCPS-II) Biobank. International Journal of Epidemiology, 2017. 47 (2): p. 385-386f. Kim, Y., B.-G. Han, and K. Group, Cohort profile: the Korean genome and epidemiology study (KoGES) consortium. International journal of epidemiology, 2017. 46 (2): p. e20-e20. Von Hinke, S., et al., Genetic markers as instrumental variables. Journal of Health Economics, 2016. 45 : p. 131-148. Burgess, S., A. Butterworth, and S.G. Thompson, Mendelian randomization analysis with multiple genetic variants using summarized data. Genetic epidemiology, 2013. 37 (7): p. 658-665. Wald, A., The fitting of straight lines if both variables are subject to error. The annals of mathematical statistics, 1940. 11 (3): p. 284-300. Bowden, J., G. Davey Smith, and S. Burgess, Mendelian randomization with invalid instruments: effect estimation and bias detection through Egger regression. International journal of epidemiology, 2015. 44 (2): p. 512-525. Sanderson, E., et al., An examination of multivariable Mendelian randomization in the single-sample and two-sample summary data settings. International journal of epidemiology, 2019. 48 (3): p. 713-727. Sanderson, E., W. Spiller, and J. Bowden, Testing and correcting for weak and pleiotropic instruments in two-sample multivariable mendelian randomisation. BioRxiv, 2020: p. 2020.04. 02.021980. Larsson, S.C., et al., Smoking, alcohol consumption, and cancer: a mendelian randomisation study in UK Biobank and international genetic consortia participants. PLoS medicine, 2020. 17 (7): p. e1003178. Gandini, S., et al., Tobacco smoking and cancer: a meta ‐analysis. International journal of cancer, 2008. 122 (1): p. 155-164. Pirie, K., et al., The 21st century hazards of smoking and benefits of stopping: a prospective study of one million women in the UK. The Lancet, 2013. 381 (9861): p. 133-141. Kamiza, A.B., et al., Hepatitis B infection is causally associated with extrahepatic cancers: A Mendelian randomization study. EBioMedicine, 2022. 79 . Collette, L., T. Burzykowski, and F.H. Schröder, Prostate-specific antigen (PSA) alone is not an appropriate surrogate marker of long-term therapeutic benefit in prostate cancer trials. European Journal of Cancer, 2006. 42 (10): p. 1344-1350. Caram, M., T.A. Skolarus, and K.A. Cooney, Limitations of prostate-specific antigen testing after a prostate cancer diagnosis. European urology, 2016. 70 (2): p. 209-210. Lucifora, J. and U. Protzer, Attacking hepatitis B virus cccDNA–The holy grail to hepatitis B cure. Journal of hepatology, 2016. 64 (1): p. S41-S48. Levrero, M. and J. Zucman-Rossi, Mechanisms of HBV-induced hepatocellular carcinoma. Journal of hepatology, 2016. 64 (1): p. S84-S101. Song, C., et al., Associations between hepatitis B virus infection and risk of all cancer types. JAMA network open, 2019. 2 (6): p. e195718-e195718. Abe, S.K., et al., Hepatitis B and C Virus Infection and Risk of Pancreatic Cancer: A Population-Based Cohort Study (JPHC Study Cohort II). Cancer Epidemiology, Biomarkers & Prevention, 2016. 25 (3): p. 555-557. Chang, M.C., et al., Hepatitis B and C viruses are not risks for pancreatic adenocarcinoma. World J Gastroenterol, 2014. 20 (17): p. 5060-5. Xu, J.H., et al., Hepatitis B or C viral infection and risk of pancreatic cancer: a meta-analysis of observational studies. World J Gastroenterol, 2013. 19 (26): p. 4234-41. Horsfall, L.J., et al., Genetically raised serum bilirubin levels and lung cancer: a cohort study and Mendelian randomisation using UK Biobank. Thorax, 2020. 75 (11): p. 955-964. Seyed Khoei, N., et al., Genetically raised circulating bilirubin levels and risk of ten cancers: A mendelian randomization study. Cells, 2021. 10 (2): p. 394. Additional Declarations No competing interests reported. Supplementary Files 20240411Additionalfile1.docx Cite Share Download PDF Status: Posted Version 1 posted You are reading this latest preprint version Research Square lets you share your work early, gain feedback from the community, and start making changes to your manuscript prior to peer review in a journal. As a division of Research Square Company, we’re committed to making research communication faster, fairer, and more useful. We do this by developing innovative software and high quality services for the global research community. Our growing team is made up of researchers and industry professionals working together to solve the most critical problems facing scientific publishing. Also discoverable on Platform About Our Team In Review Editorial Policies Advisory Board Help Center Resources Author Services Accessibility API Access RSS feed Manage Cookie Preferences © Research Square 2026 | ISSN 2693-5015 (online) Privacy Policy Terms of Service Do Not Sell My Personal Information {"props":{"pageProps":{"initialData":{"identity":"rs-4249634","acceptedTermsAndConditions":true,"allowDirectSubmit":true,"archivedVersions":[],"articleType":"Research Article","associatedPublications":[],"authors":[{"id":291670573,"identity":"7879edd9-02e4-4102-995b-2df8d0c41a74","order_by":0,"name":"Keum Ji Jung","email":"","orcid":"","institution":"Yonsei University","correspondingAuthor":false,"prefix":"","firstName":"Keum","middleName":"Ji","lastName":"Jung","suffix":""},{"id":291670574,"identity":"ef4dd007-0d94-40ba-a19e-28a379702014","order_by":1,"name":"Wes Spiller","email":"","orcid":"","institution":"UCB Pharma","correspondingAuthor":false,"prefix":"","firstName":"Wes","middleName":"","lastName":"Spiller","suffix":""},{"id":291670576,"identity":"c70e2f57-4240-4e06-b345-36b36a94fdbb","order_by":2,"name":"Dae Sub Song","email":"","orcid":"","institution":"National Institute of Health, Korea Disease Control and Prevention Agency","correspondingAuthor":false,"prefix":"","firstName":"Dae","middleName":"Sub","lastName":"Song","suffix":""},{"id":291670578,"identity":"59c81cd8-7c56-431f-8161-9065b2f8ccc0","order_by":3,"name":"Jong Won Shin","email":"","orcid":"","institution":"Yonsei University","correspondingAuthor":false,"prefix":"","firstName":"Jong","middleName":"Won","lastName":"Shin","suffix":""},{"id":291670579,"identity":"aaaee84e-cbc5-4df8-98e2-d291a4677619","order_by":4,"name":"Kyoungho Lee","email":"","orcid":"","institution":"National Institute of Health, Korea Disease Control and Prevention Agency","correspondingAuthor":false,"prefix":"","firstName":"Kyoungho","middleName":"","lastName":"Lee","suffix":""},{"id":291670581,"identity":"0cbf3ecd-2af4-4e8e-b04b-26437cc2f46d","order_by":5,"name":"Sun Ha Jee","email":"data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAAZAAAAAyAQMAAABI0h/eAAAABlBMVEX///8AAABVwtN+AAAACXBIWXMAAA7EAAAOxAGVKw4bAAAAuElEQVRIiWNgGAWjYBACxgYGBmYGBhsGBgkStaSRoAUEgFoOk6CFuf+M2eOCmvOJG263P2D4UUOMw2bkmBvPOHY7ccOdMwaMPceI0sJjJs3DBtRyI4eBgbeBGC1Ah0nz/DsH1JL+gPEvUVoacsykedsOALUkGDATZ8uMtDJp3r5k45k3cgwOyxDjF8P+w9ukeb7ZyfbdSH/48A0xIWYIdYojiD5AhAYGBnkobU+U6lEwCkbBKBiZAADZxTfTWP8n3gAAAABJRU5ErkJggg==","orcid":"","institution":"Yonsei University","correspondingAuthor":true,"prefix":"","firstName":"Sun","middleName":"Ha","lastName":"Jee","suffix":""}],"badges":[],"createdAt":"2024-04-11 01:29:12","currentVersionCode":1,"declarations":"","doi":"10.21203/rs.3.rs-4249634/v1","doiUrl":"https://doi.org/10.21203/rs.3.rs-4249634/v1","draftVersion":[],"editorialEvents":[],"editorialNote":"","failedWorkflow":false,"files":[{"id":55316282,"identity":"066fce9a-a125-49f1-9c16-8a88052e8c3b","added_by":"auto","created_at":"2024-04-25 15:45:54","extension":"png","order_by":1,"title":"Figure 1","display":"","copyAsset":false,"role":"figure","size":275054,"visible":true,"origin":"","legend":"\u003cp\u003eA heat-map showing weighted conditional F-statistics for pairs of exposures initially included in an MVMR model. (Values of 0 are included on the diagonal to indicate identical exposures or instances where one or more of the conditional F-statistics for a pair of exposures failed to exceed the minimum threshold of 10.)\u003c/p\u003e","description":"","filename":"image1.png","url":"https://assets-eu.researchsquare.com/files/rs-4249634/v1/45a872cf1ab271b1b792bb38.png"},{"id":55316280,"identity":"3587e332-76f5-43e6-b739-d733bde1cb0f","added_by":"auto","created_at":"2024-04-25 15:45:54","extension":"png","order_by":2,"title":"Figure 2","display":"","copyAsset":false,"role":"figure","size":41409,"visible":true,"origin":"","legend":"\u003cp\u003eA forest plot showing MVMR MR estimates. Estimates are grouped by outcome, representing the direct effect of each exposure including all exposures within each MVMR model. (Effect estimates on the odds ratio scale, with accompanying 95% confidence intervals are illustrated, with corresponding values shown to the right of the plot. Univariable MR estimates for each exposure are included for comparison)\u003c/p\u003e","description":"","filename":"image2.png","url":"https://assets-eu.researchsquare.com/files/rs-4249634/v1/dabbaadb655e509cb28da408.png"},{"id":55316279,"identity":"76ca3681-5989-4e04-a86b-f650b7308c21","added_by":"auto","created_at":"2024-04-25 15:45:54","extension":"png","order_by":3,"title":"Figure 3","display":"","copyAsset":false,"role":"figure","size":33290,"visible":true,"origin":"","legend":"\u003cp\u003eA forest plot showing MVMR MR estimates for breast cancer. Estimates are divided by age group (50 aged), representing the direct effect of each exposure including all exposures within each MVMR model. (Effect estimates on the odds ratio scale, with accompanying 95% confidence intervals are illustrated, with corresponding values shown to the right of the plot. Univariable MR estimates for each exposure are included for comparison.)\u003c/p\u003e","description":"","filename":"image3.png","url":"https://assets-eu.researchsquare.com/files/rs-4249634/v1/36c507b8638c45966fdb145a.png"},{"id":58631575,"identity":"b77981d4-26bc-4011-b58d-2ace7f82e899","added_by":"auto","created_at":"2024-06-19 05:45:31","extension":"pdf","order_by":0,"title":"","display":"","copyAsset":false,"role":"manuscript-pdf","size":889476,"visible":true,"origin":"","legend":"","description":"","filename":"manuscript.pdf","url":"https://assets-eu.researchsquare.com/files/rs-4249634/v1/6971521f-2b4a-452b-94e3-5cebfe733b06.pdf"},{"id":55316283,"identity":"41421519-fb3b-484e-82b0-bccd4a40caa3","added_by":"auto","created_at":"2024-04-25 15:45:54","extension":"docx","order_by":4,"title":"","display":"","copyAsset":false,"role":"supplement","size":464544,"visible":true,"origin":"","legend":"","description":"","filename":"20240411Additionalfile1.docx","url":"https://assets-eu.researchsquare.com/files/rs-4249634/v1/c19c0aafc8ccc274bfa8c524.docx"}],"financialInterests":"No competing interests reported.","formattedTitle":"Examining complex cancer etiologies within the Korean population: A high-throughput multivariable Mendelian randomization study","fulltext":[{"header":"Background","content":"\u003cp\u003eThrough past observational studies, various research reports on factors related to cancer development have been published. The two most significant factors exhibiting the strongest associations with cancer incidence are smoking and lung cancer[\u003cspan citationid=\"CR1\" class=\"CitationRef\"\u003e1\u003c/span\u003e] and hepatitis B surface antigen (HBsAg) and liver cancer[\u003cspan citationid=\"CR2\" class=\"CitationRef\"\u003e2\u003c/span\u003e]. In the case of smoking, investigating its relevance not only to lung cancer but also to other types of cancer is meaningful. Similarly, it is crucial to explore how HBsAg influences cancers beyond its association with liver cancer.\u003c/p\u003e \u003cp\u003eThe association between identified risk factors and cancer incidence, as revealed through previous observational studies, faced limitations in drawing causal conclusions due to confounding variables, reverse causation, measurement errors, and other reasons. To overcome such limitations, Mendelian randomization (MR)[\u003cspan citationid=\"CR3\" class=\"CitationRef\"\u003e3\u003c/span\u003e] has been proposed as a method to estimate effects without residual confounding. MR is an epidemiological approach that uses genetic variants as a tool to strengthen causal inference in the association between an exposure and an outcome[\u003cspan citationid=\"CR4\" class=\"CitationRef\"\u003e4\u003c/span\u003e]. One of the key advantages of MR is its ability to minimize confounding since genetic variants are randomly assigned at conception and are therefore unrelated to environmental and self-adopted factors that typically act as confounders. In MR, germline genetic variation is used as a proxy variable for modifiable exposures concerning causal inference about outcomes[\u003cspan citationid=\"CR5\" class=\"CitationRef\"\u003e5\u003c/span\u003e, \u003cspan citationid=\"CR6\" class=\"CitationRef\"\u003e6\u003c/span\u003e]. This design also helps to mitigate the issue of reverse causality, as the fixed alleles used in MR analyses are not influenced by the onset or progression of disease. The fundamental principle of MR is that differences between individuals due to genetic variations are not influenced by the outcome. Confounding or reverse causation biases that distort observational findings. Therefore, the natural 'randomization' of alternate alleles (mutation forms) can be likened to the random allocation of treatment conducted in randomized controlled trials. This means experiencing an average exposure difference without being different in confounding factors associated with alternate genetic variants[\u003cspan citationid=\"CR4\" class=\"CitationRef\"\u003e4\u003c/span\u003e, \u003cspan citationid=\"CR5\" class=\"CitationRef\"\u003e5\u003c/span\u003e]. Cancer is a multifactorial disease, and different risk factors exist for each cancer type. Therefore, it is not appropriate to discuss causality using only one factor. Using a multivariable Mendelian randomization (MVMR) method that takes into account the genetic correlation of several variables that can influence the main risk factor, we aimed to determine the causal impact on cancer.\u003c/p\u003e \u003cp\u003eThis MR aimed to investigate the causal relationships between various exposure factors and types of cancer among Koreans using novel data from two large-scale biobanks.\u003c/p\u003e"},{"header":"Methods","content":"\u003cdiv id=\"Sec3\" class=\"Section2\"\u003e \u003ch2\u003eStudy population and data sources\u003c/h2\u003e \u003cp\u003e \u003cb\u003eKorean Cancer Prevention Study (KCPS)-II Biobank.\u003c/b\u003e \u003c/p\u003e \u003cp\u003eThe Korean Cancer Prevention Study (KCPS)-II Biobank is a prospective population-based cohort comprising adults recruited from 18 health examination centers across South Korea. Detailed descriptions of the KCPS-II have been previously described[\u003cspan citationid=\"CR7\" class=\"CitationRef\"\u003e7\u003c/span\u003e]. The cohort consisted of 156,701 participants (94,840 men and 61,861 women) who underwent routine health assessments between 2004 and 2013, provided blood samples, and provided informed consent for long-term follow-up. At baseline, participants provided information on sociodemographic factors, alcohol consumption, smoking habits, diet, exercise, and past medical history through a questionnaire. This study included 153,971 participants who completed genetic analysis out of the 156,701 total participants (IRB no. 4-2023-1722).\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec4\" class=\"Section2\"\u003e \u003ch2\u003eKorean Genome Epidemiology Study (KoGES) Biobank\u003c/h2\u003e \u003cp\u003eThe Korean Genome Epidemiologic Study (KoGES) Biobank established encompasses a cohort built from 2004 to 2013, incorporating community cohorts in the Ansan, Anseong, and Urban cohorts and Rural cohorts, totaling approximately 210,000 individuals. The KoGES participants were men and women aged 40 and above. The KoGES comprises 195,544 participants (69,579 men and 125,965 women) who provided blood samples and gave informed consent for long-term follow-up. This study included 195,544 individuals for whom genetic testing data were available. Detailed descriptions of KoGES have been previously published[\u003cspan citationid=\"CR8\" class=\"CitationRef\"\u003e8\u003c/span\u003e].\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec5\" class=\"Section2\"\u003e \u003ch2\u003eExposure and Outcome\u003c/h2\u003e \u003cp\u003eThis study selected 13 exposure factors related to cancer incidence from the KCPS-II Biobank and conducted genome-wide association study (GWAS) analyses for each, incorporating relevant genetic factors into the research. Alcohol consumption, serum bilirubin, body mass index (BMI), waist circumference, smoking amount, carcinoembryonic antigen (CEA), fasting blood sugar (FBS), thyroid-stimulating hormone (TSH), prostate-specific antigen (PSA), aspartate aminotransferase (AST), alanine transferase (ALT), gamma-glutamyl transferase (GGT), and hepatitis B surface antigen (HBsAg), which are known risk factors for cancer site, were selected as exposure factors for analysis. Additionally, a comprehensive set of cancers and 13 specific cancer types were chosen from the KoGES, and GWAS analyses were performed for each, investigating the associated genetic factors.\u003c/p\u003e \u003cp\u003eThe main outcome of this study was cancer incidence. The cancer incidence information about the subjects was confirmed by linking the cancer registration data, a collection of diagnosis records from hospitals. Information on the cancer site, cancer diagnosis date, and histological type was collected in connection with the national cancer registration data. Until December 31, 2020, the cancer registration data of the follow-up National Cancer Center and the cause of death data of the National Statistical Office were used.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec6\" class=\"Section2\"\u003e \u003ch2\u003eStatistical analysis\u003c/h2\u003e \u003cdiv id=\"Sec7\" class=\"Section3\"\u003e \u003ch2\u003eUnivariable Mendelian randomization\u003c/h2\u003e \u003cp\u003eThis study initially conducted univariable MR analyses using 13 risk factors from the KCPS-II BioBank and 13 cancer types from the KoGES. MR employs genetic variants, commonly single nucleotide polymorphisms (SNPs), as instrumental variables to assess causal effects free from the influence of unmeasured confounding. This methodology is applicable when a specific genetic variant, designated a candidate (IV1), has a robust association with targeted exposure. Additionally, MR assumes the absence of confounders affecting both the genetic variant and the outcome (IV2) and independence from the outcome when considering the exposure (IV3)[\u003cspan citationid=\"CR9\" class=\"CitationRef\"\u003e9\u003c/span\u003e]. Assuming the assumptions were met, beta values and standard errors (SE) were estimated using the inverse-variance weighting (IVW) method.\u003c/p\u003e \u003cp\u003eTo derive a Wald ratio estimate for a single genetic instrument, one divides the association between the genetic variant and the outcome by the association between the genetic variant and the exposure[\u003cspan citationid=\"CR10\" class=\"CitationRef\"\u003e10\u003c/span\u003e, \u003cspan citationid=\"CR11\" class=\"CitationRef\"\u003e11\u003c/span\u003e].\u003c/p\u003e \u003cp\u003eIn cases involving multiple SNPs, these ratio estimates are frequently amalgamated using a fixed-effects meta-analysis approach[\u003cspan citationid=\"CR10\" class=\"CitationRef\"\u003e10\u003c/span\u003e, \u003cspan citationid=\"CR12\" class=\"CitationRef\"\u003e12\u003c/span\u003e].\u003c/p\u003e \u003cp\u003eFurthermore, considering the presence of pleiotropy, beta values and SEs were estimated using the MR‒Egger method. To determine the existence of pleiotropy, an intercept test was conducted.\u003c/p\u003e \u003c/div\u003e \u003c/div\u003e \u003cdiv id=\"Sec8\" class=\"Section2\"\u003e \u003ch2\u003eMultivariable Mendelian randomization (MVMR)\u003c/h2\u003e \u003cp\u003eThe instrumental variable may exhibit pleiotropy, encompassing multiple exposures. To address this, we conducted MR with multiple instrumental variable analysis by creating genetic variables to control for pleiotropy in explaining various exposures[\u003cspan citationid=\"CR13\" class=\"CitationRef\"\u003e13\u003c/span\u003e]. A new forward selection algorithm has been introduced to help determine the most appropriate MVMR model based on instrument strength[\u003cspan citationid=\"CR14\" class=\"CitationRef\"\u003e14\u003c/span\u003e]. The goal is to include the maximum number of exposures that have conditional F-statistics between exposure and genetic variants surpassing a threshold value of 10. Initially, all potential exposure pairs are assessed regarding their conditional F-statistics. Subsequently, the average F-statistic for each pair is computed, taking into account the inverse of the absolute difference between the F-statistics within a specific pair. This weighted calculation of the average F-statistic for each pair aims to address situations where two exposures collectively display a high mean F-statistic, primarily influenced by one of the exposures. Initially, we computed weighted conditional F-statistics for pairs of exposures (13 in total) and selected the final three variables. This selection process, including the methodology, is illustrated in Fig.\u0026nbsp;\u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e2\u003c/span\u003e. To achieve this, we presented the F values for each factor included in the model. Analyses were performed using the \u003cem\u003eTwoSampleMR and MVMR\u003c/em\u003e R packages. Analyses were conducted in R (version 4.0.3).\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003c/div\u003e"},{"header":"Results","content":"\u003cp\u003eTable\u0026nbsp;\u003cspan refid=\"Tab1\" class=\"InternalRef\"\u003e1\u003c/span\u003e shows the general characteristics of the two main cohorts used in this study. The mean age of KoGES participants is more than 10 years older than that of KCPS-II participants. Of the 147 390 KoGES participants used as the outcome data set, 15,062 were affected by cancer, of which incidence rate per 100, 000-person year for prostate cancer (males), breast cancer (females), thyroid cancer, and gastric cancer was more than 100 (Table\u0026nbsp;\u003cspan refid=\"Tab2\" class=\"InternalRef\"\u003e2\u003c/span\u003e).\u003c/p\u003e \u003cp\u003e \u003cdiv class=\"gridtable\"\u003e\u003ctable float=\"Yes\" id=\"Tab1\" border=\"1\"\u003e \u003ccaption language=\"En\"\u003e \u003cdiv class=\"CaptionNumber\"\u003eTable 1\u003c/div\u003e \u003cdiv class=\"CaptionContent\"\u003e \u003cp\u003eGeneral characteristics of Korean biobanks\u003c/p\u003e \u003c/div\u003e \u003c/caption\u003e \u003ccolgroup cols=\"5\"\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c1\" colnum=\"1\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c2\" colnum=\"2\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c3\" colnum=\"3\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c4\" colnum=\"4\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c5\" colnum=\"5\"\u003e\u003c/div\u003e \u003cthead\u003e \u003ctr\u003e \u003cth align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/th\u003e \u003cth align=\"left\" colspan=\"2\" nameend=\"c3\" namest=\"c2\"\u003e \u003cp\u003eKCPS-II\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colspan=\"2\" nameend=\"c5\" namest=\"c4\"\u003e \u003cp\u003eKoGES\u003c/p\u003e \u003c/th\u003e \u003c/tr\u003e \u003c/thead\u003e \u003ctbody\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eMen\u003c/p\u003e \u003cp\u003e(N\u0026thinsp;=\u0026thinsp;93 118 )\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003eWomen\u003c/p\u003e \u003cp\u003e(N\u0026thinsp;=\u0026thinsp;60 853)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003eMen\u003c/p\u003e \u003cp\u003e(N\u0026thinsp;=\u0026thinsp;53 348)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003eWomen\u003c/p\u003e \u003cp\u003e(N\u0026thinsp;=\u0026thinsp;94 042)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colspan=\"2\" nameend=\"c3\" namest=\"c2\"\u003e \u003cp\u003eMean (SD)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colspan=\"2\" nameend=\"c5\" namest=\"c4\"\u003e \u003cp\u003eMean (SD)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eAge, year\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e42.2 (10.0)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e41.0 (11.3)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e54.6 (9.1)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e53.4 (8.4)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eAlcohol amount, g/day\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e22.7 (29.9)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e5.8 (13.8)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e24.8 (37.5)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e6.3 (14.3)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eSmoking amount, cig./day\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e11.38 (9.9)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e0.7 (3.4)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e12.9 (11.5)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e0.4 (2.3)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eBody mass index, kg/m\u003csup\u003e2\u003c/sup\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e24.4 (2.9)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e22.2 (3.1)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e24.3 (2.8)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e23.8 (3)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eWaist circumference, cm\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e84.8 (7.8)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e74.2 (8.5)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e85.6 (7.6)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e79.2 (8.5)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eCEA, ng/ml\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e1.9 (2.1)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e1.6 (1.9)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e-\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e-\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eTSH, uU/ml\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e1.9 (2.7)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e2.2 (5.0)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e2.0 (4.1)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e2.3 (3.8)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003ePSA, ng/ml\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e1.0 (1.1)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e-\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e-\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eFBS, ng/ml\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e93.5 (20.8)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e88.3 (16.3)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e99.2 (24.2)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e93.3 (19.6)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eAST, u/l\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e25.5 (20.0)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e20.0 (11.3)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e26.5 (19.0)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e22.9 (19.9)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eALT, u/l\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e30.5 (28.1)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e17.3 (17.3)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e27.4 (22.4)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e20.2 (21.7)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eGGT, IU/L\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e47.4 (61.8)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e19.1 (20.4)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e49.1 (70.2)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e22.3 (23.8)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eBilirubin, mg/dl\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e0.9 (0.4)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e0.7 (0.3)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e0.8 (0.3)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e0.7 (0.3)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eHBsAg (%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e4.4\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e2.9\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e-\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e-\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003c/tbody\u003e \u003c/colgroup\u003e \u003ctfoot\u003e \u003ctr\u003e\u003ctd colspan=\"5\"\u003eKCPS-II, Korean Cancer Prevention Study-II; KoGES, Korean Genome Epidemiologic Study; CEA, Carcinoembryonic antigen; TSH, Thyroid Stimulating Hormone; PSA, Prostate Specific Antigen; FBS, Fasting blood sugar; AST, aspartate aminotransferase; ALT, alanine transferase; GGT, Gamma-glutamyl transferase; HBsAg, hepatitis B surface antigen\u003c/td\u003e\u003c/tr\u003e \u003c/tfoot\u003e \u003c/table\u003e\u003c/div\u003e \u003c/p\u003e \u003cp\u003e \u003cdiv class=\"gridtable\"\u003e\u003ctable float=\"Yes\" id=\"Tab2\" border=\"1\"\u003e \u003ccaption language=\"En\"\u003e \u003cdiv class=\"CaptionNumber\"\u003eTable 2\u003c/div\u003e \u003cdiv class=\"CaptionContent\"\u003e \u003cp\u003eSite-specific cancer incidence in Korean Genome Epidemiologic Study Biobank\u003c/p\u003e \u003c/div\u003e \u003c/caption\u003e \u003ccolgroup cols=\"5\"\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c1\" colnum=\"1\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c2\" colnum=\"2\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c3\" colnum=\"3\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c4\" colnum=\"4\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c5\" colnum=\"5\"\u003e\u003c/div\u003e \u003cthead\u003e \u003ctr\u003e \u003cth align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/th\u003e \u003cth align=\"left\" colname=\"c2\"\u003e \u003cp\u003eNumber of participants\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c3\"\u003e \u003cp\u003eSum of person year\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c4\"\u003e \u003cp\u003eNumber of incident cancer\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c5\"\u003e \u003cp\u003eIncidence per 100,000 PY\u003c/p\u003e \u003c/th\u003e \u003c/tr\u003e \u003c/thead\u003e \u003ctbody\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eAll cancer\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e147 390\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e1 655 481.3\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e15 062\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e909.8\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eStomach cancer\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e147 390\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e1 734 455.6\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e2 033\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e117.2\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eColorectal cancer\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e147 390\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e1 736 811.6\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e1 652\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e95.1\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eLiver cancer\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e147 390\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e1 742 497.5\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e872\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e50.0\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eGallbladder and bile duct cancer\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e147 390\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e1 745 626.9\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e418\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e23.9\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003ePancreas cancer\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e147 390\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e1 745 992.6\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e385\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e22.1\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eLung cancer\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e147 390\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e1 739 296.6\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e1 553\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e89.3\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eBreast cancer\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e94 042\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e1 098 578.0\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e1 477\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e134.4\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eCervix cancer\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e94 042\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e1 106 004.2\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e146\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e13.2\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eProstate cancer\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e53 348\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e635 647.3\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e1 007\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e158.4\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eThyroid cancer\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e147 390\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e1 732 641.6\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e2 213\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e127.7\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eKidney cancer\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e147 390\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e1 746 168.6\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e317\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e18.2\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eBladder cancer\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e147 390\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e1 746 587.6\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e248\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e14.2\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eNon-Hodgkin\u0026rsquo;s Lymphoma\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e147 390\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e1 746 552.6\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e265\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e15.2\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003c/tbody\u003e \u003c/colgroup\u003e \u003c/table\u003e\u003c/div\u003e \u003c/p\u003e \u003cp\u003eSupplementary Fig.\u0026nbsp;1-A presents the univariable (MR analysis results for overall cancer, gastric cancer, colorectal cancer, and liver cancer). TSH exhibited a significant negative correlation with overall cancer incidence. In gastric cancer, HBsAg was the only factor showing a significant positive association. In liver cancer, HBsAg demonstrated a highly significant positive correlation, along with a significant positive association observed for AST. Supplementary Fig.\u0026nbsp;1-B focuses on gallbladder and other biliary tract cancers, pancreatic cancer, lung cancer, and breast cancer. No significant factors were identified for biliary tract or pancreatic cancers. However, current smoking status exhibited a highly significant positive association with lung cancer. Subsequently, although bilirubin had a wide 95% confidence interval, it showed a remarkably significant positive association. Conversely, breast cancer showed a negative association with HBsAg. As shown in Supplementary Fig.\u0026nbsp;1-C, which shows renal cancer, cervical cancer, prostate cancer, and thyroid cancer, renal cancer was significantly positively associated with CEA and smoking status. Cervical cancer exhibited a positive association with HBsAg status. Despite a broad confidence interval, prostate cancer was significantly positively associated with PSA. Thyroid cancer was significantly positively associated with CEA and smoking. Supplementary Fig.\u0026nbsp;1-D depicts univariable MR analyses for 13 exposures concerning bladder cancer and non-Hodgkin's lymphoma. For bladder cancer patients, an inverse association with alcohol consumption was observed. Conversely, non-Hodgkin's lymphoma did not exhibit any significant associations.\u003c/p\u003e \u003cp\u003eFigure \u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e1\u003c/span\u003e shows the weighted conditional F-statistics, including differing pairs of exposures. Hence, to estimate effects using MVMR, it was necessary to utilize a subset of exposures. Employing the aforementioned forward selection approach, all conceivable pairs of exposures were assessed, highlighting HBsAg and GGT as the pair with the highest combined conditional F-statistics. Subsequently, in a stepwise manner, additional exposures were considered, leading to the inclusion of bilirubin in a three-exposure MVMR model. Additionally, it displays a heatmap demonstrating the conditional F-statistics for each exposure, portraying the intensity of the weighted mean conditional F-statistic through a gradient.\u003c/p\u003e \u003cp\u003eFigure \u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e2\u003c/span\u003e displays the results of MVMR analysis for 13 exposures and 13 cancer types. Specifically, this study presents the MVMR model outcomes controlling for bilirubin and GGT to examine the causal effect of HBsAg. HBsAg was positively associated with lung cancer, liver cancer, and cervical cancer. Conversely, negative associations were identified for pancreatic cancer, breast cancer, and thyroid cancer.\u003c/p\u003e \u003cp\u003eFigure \u003cspan refid=\"Fig3\" class=\"InternalRef\"\u003e3\u003c/span\u003e shows the analysis results stratified by age group for breast cancer patients. In the case of female subjects aged 50 and above, those with HBsAg exhibited a significantly lower risk, with an odds ratio of 0.87 (0.84\u0026ndash;0.90) and P=\u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\(3.07\\times {10}^{-16}\\)\u003c/span\u003e\u003c/span\u003e. However, no significant association was observed between HBsAg and breast cancer in females under the age of 50.\u003c/p\u003e "},{"header":"Discussion","content":"\u003cp\u003eThis study performed the analysis by independently selecting exposure and outcome datasets using a two-sample MR method. The final estimates were obtained through MVMR, controlling for important pleiotropic variables. By initially exploring topics known for their causality or significant impact on outcome risks due to genetically determined exposures, we aimed to establish a basis for the overall validity of the entire dataset.\u003c/p\u003e \u003cp\u003eThe first consideration is the relationship between smoking quantity and lung cancer incidence. In this study, smoking quantity exhibited a significant positive causal relationship with lung cancer according to the crude two-sample MR analysis. This finding serves as evidence validating the credibility of results from previous observational studies on the association between smoking and lung cancer[\u003cspan additionalcitationids=\"CR16\" citationid=\"CR15\" class=\"CitationRef\"\u003e15\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR17\" class=\"CitationRef\"\u003e17\u003c/span\u003e]. Second, the relationship between HBsAg and liver cancer was examined. Compared with the HBsAg-negative group, the HBsAg-positive group showed a significantly positive causal relationship with the risk of liver cancer, as observed in both univariable two-sample MR and MVMR analyses[\u003cspan citationid=\"CR18\" class=\"CitationRef\"\u003e18\u003c/span\u003e]. Finally, the causal relationship between prostate-specific antigen (PSA) and prostate cancer is a topic that may be subject to debate[\u003cspan citationid=\"CR19\" class=\"CitationRef\"\u003e19\u003c/span\u003e, \u003cspan citationid=\"CR20\" class=\"CitationRef\"\u003e20\u003c/span\u003e].\u003c/p\u003e \u003cp\u003eIn this study, PSA exhibited a consistently significant positive causal relationship in univariable two-sample MR. By clearly establishing causal relationships for these three exposures and three cancer types, this study provides a basis for the reliability and validity of the data. Moving forward, the findings newly revealed in this study are discussed.\u003c/p\u003e \u003cdiv id=\"Sec11\" class=\"Section2\"\u003e \u003ch2\u003eA Review on Cancer Incidence and Its Positive Association with HBsAg Positivity\u003c/h2\u003e \u003cp\u003eHBsAg showed a positive association not only with liver cancer but also with lung cancer and cervical cancer in this study. Consistent with our findings, a Mendelian randomization (MR) study conducted in East Asian populations in 2022 reported significant positive associations for liver cancer (OR 1.27, p\u0026thinsp;=\u0026thinsp;0.0001), lung cancer (OR 1.12, p\u0026thinsp;=\u0026thinsp;0.0002), gastric cancer (OR 1.09, p\u0026thinsp;=\u0026thinsp;0.0002), and cervical cancer (OR 1.57, p\u0026thinsp;=\u0026thinsp;0.0001)[\u003cspan citationid=\"CR18\" class=\"CitationRef\"\u003e18\u003c/span\u003e]. It is believed that HBV protein X, a transcriptional coactivator, plays a crucial role in initiating tumorigenesis by modulating key regulators of apoptosis and interfering with DNA repair pathways and tumor suppressor genes. However, further functional tests are required to elucidate the potential mechanism of action.\u003c/p\u003e \u003cp\u003eWe observed that in most cancer patients, the expression of the HBX and anti-HBc proteins were greater in cancer tissue specimens than in healthy tissue specimens. It is possible that HBV may be harbored in nonliver cells, potentially inducing local inflammation. Chronic inflammation induced by HBV infection might play a role in the development of cancer; the viral oncogenic HBX protein may play a direct role in the development of cancer. However, we found a weaker association between HBV infection and nonliver cancer than between HBV infection and HCC. Although we confirmed HBV replication and expression in stomach cancer and pancreatic cancer, the immunohistochemistry results suggested that HBX was a pro[\u003cspan additionalcitationids=\"CR22\" citationid=\"CR21\" class=\"CitationRef\"\u003e21\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR23\" class=\"CitationRef\"\u003e23\u003c/span\u003e]. However, further functional tests are needed to explain the potential mechanism of action. The biological mechanism linking chronic HBV infection with extrahepatic cancers has not yet been fully illustrated. However, it is thought that HBV protein X, a transcriptional coactivator, plays a crucial role in initiating tumorigenesis by modulating key regulators of apoptosis and interfering with DNA repair pathways and tumor suppressor genes[\u003cspan citationid=\"CR22\" class=\"CitationRef\"\u003e22\u003c/span\u003e].\u003c/p\u003e \u003cp\u003eMoreover, Song et al. reported greater HBV protein X expression in the cancerous part of the tissue specimen than in the healthy part of the same specimen, supporting the oncogenic role of HBV protein X[\u003cspan citationid=\"CR23\" class=\"CitationRef\"\u003e23\u003c/span\u003e]. Furthermore, some studies have detected HBV DNA in some extrahepatic tissues, including gastric, kidney, gallbladder and pancreatic tissues, suggesting that HBV can initiate and promote tumorigenesis outside the liver[\u003cspan citationid=\"CR18\" class=\"CitationRef\"\u003e18\u003c/span\u003e].\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec12\" class=\"Section2\"\u003e \u003ch2\u003eA Review of HBsAg Positivity and Reduced Cancer Incidence\u003c/h2\u003e \u003cp\u003eThe association between HBsAg and cancer is controversial. HBsAg has shown an inverse relationship with breast cancer, pancreatic cancer, and thyroid cancer. In a Mendelian randomization study conducted in East Asian populations in 2022, no significant association was found between HBsAg and pancreatic cancer, and there have been no studies on breast cancer and thyroid cancer[\u003cspan citationid=\"CR18\" class=\"CitationRef\"\u003e18\u003c/span\u003e]. When analyzing breast cancer by dividing it into age groups, a greater inverse association was observed in women aged 50 and older, particularly after menopause. Unfortunately, specific references or information on recent studies supporting these results was not available. However, we can propose several hypotheses. First, in relation to immunomodulation, we suggest that HBsAg may stimulate the immune system, leading to enhanced immune surveillance and clearance of cancer cells. This immunomodulatory effect could contribute to a decreased risk of breast cancer in women who test positive for HBsAg. Additionally, we hypothesize that the hepatitis B virus has a direct impact on breast cancer cells. It is postulated that the virus may inhibit the growth or proliferation of breast cancer cells via various mechanisms. Such antiproliferative effects could contribute to the observed decreased risk. Finally, the relationship between HBsAg and breast cancer risk may be influenced by population-specific factors such as genetic variations, environmental exposures, or lifestyle patterns. We speculate that Korean women may exhibit varying susceptibilities to the risk of HBsAg-associated breast cancer. This content suggests the need for further, more specific research.\u003c/p\u003e \u003cp\u003eIn this study, HBsAg inhibited the occurrence of pancreatic cancer. This result is presumed to be statistically significant due to the large number of participants in the study. However, it can be confirmed that HBsAg does not increase the risk of pancreatic cancer. According to a large-scale cohort study conducted by Abe et al. (2016) involving a total of 20,360 subjects (324,394 person-years) from the Japan Public Health Center (JPHC)-based prospective cohort study cohort II, which was followed for an average of 16 years, 116 patients with newly diagnosed pancreatic cancer were confirmed[\u003cspan citationid=\"CR24\" class=\"CitationRef\"\u003e24\u003c/span\u003e]. Among the HBsAg-positive participants, there were no patients with pancreatic cancer. Therefore, the JPHC study did not observe a statistically significant association between hepatitis B and the risk of pancreatic cancer[\u003cspan citationid=\"CR24\" class=\"CitationRef\"\u003e24\u003c/span\u003e]. Additionally, a recent study in Taiwan reported no results for any antibody variants[\u003cspan citationid=\"CR25\" class=\"CitationRef\"\u003e25\u003c/span\u003e]. In contrast, a previous meta-analysis primarily based on case‒control studies conducted in other Asian countries suggested that HBsAg increases the risk of pancreatic cancer[\u003cspan citationid=\"CR26\" class=\"CitationRef\"\u003e26\u003c/span\u003e].\u003c/p\u003e \u003cp\u003eAccording to a prospective cohort study involving over 500,000 adults in 10 regions of China, there was no association between hepatitis B virus (HBV) infection and various cancer types, such as thyroid cancer, skin cancer, and leukemia[\u003cspan citationid=\"CR23\" class=\"CitationRef\"\u003e23\u003c/span\u003e]. Therefore, additional research is necessary for thyroid cancer treatment.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec13\" class=\"Section2\"\u003e \u003ch2\u003eReview of the positive association between serum bilirubin and lung cancer\u003c/h2\u003e \u003cp\u003eIn a study conducted by Horsfall et al. (2020) utilizing the UK Biobank dataset, an inverse correlation between serum bilirubin levels and lung cancer was reported[\u003cspan citationid=\"CR27\" class=\"CitationRef\"\u003e27\u003c/span\u003e]. Serum bilirubin was indirectly implicated in its role as an antioxidant. The research team identified two genetic factors, rs887829 and rs4149056, that are associated with elevated serum bilirubin. They presented evidence of a decreased incidence of lung cancer based on the genotype, proposing these two SNPs as potent determinants explaining approximately 40% of the variance in serum bilirubin. However, it is essential to note that this investigation targeted a European cohort and revealed associations solely within the subset of smokers[\u003cspan citationid=\"CR27\" class=\"CitationRef\"\u003e27\u003c/span\u003e].\u003c/p\u003e \u003cp\u003eBy conducting research utilizing the same UK Biobank dataset, a study categorized genetic variants into UGT1A1 SNPs and non-UGT1A1 SNPs. Interestingly, when exclusively considering non-UGT1A1 SNPs, it was found that genetically increased circulating bilirubin levels were associated with an elevated risk of lung cancer. Notably, this association was evident only in smokers and specifically for patients with squamous cell carcinoma. Consequently, direct comparison of these subgroup results with our findings may pose challenges, emphasizing the need for subsequent studies to address this aspect[\u003cspan citationid=\"CR28\" class=\"CitationRef\"\u003e28\u003c/span\u003e].\u003c/p\u003e \u003cp\u003eOur study has several strengths, notably leveraging the MR approach, which effectively mitigates certain confounding factors frequently encountered in epidemiological studies. Additionally, we employed multiple SNPs strongly linked to risk factors. Furthermore, we utilized a homogenous population, thereby reducing the typical heterogeneity observed in genetic studies involving individuals from populations of diverse ancestry. Additionally, this study used two large biobanks of a single ethnic group (Korean), providing significant statistical strength in identifying causal effects. A two-sample study design can maximize the effect of sample size. Sensitivity analyses such as MR‒Egger regression are required for Mendelian randomization studies. In the univariable MR setting, employing such sensitivity analyses ensures more conservative results, especially when weak instruments are used. Although consistent findings were observed when applying sensitivity analyses in the univariable MR setting, significant heterogeneity indicated potential violations of MR assumptions. This warrants further investigation using MVMR. This study has several limitations. One major limitation of the MR design is the presence of horizontal pleiotropy, which occurs when the genetic variants used in the analysis influence the outcomes through pathways unrelated to the exposure of interest. However, in this study, the potential bias due to horizontal pleiotropy is expected to be minimal. This is supported by the findings of the MR‒Egger analysis, which did not indicate any evidence of horizontal pleiotropy. Additionally, consistent results were obtained from a series of sensitivity analyses, further strengthening the robustness of the associations observed. Furthermore, the use of a multivariable MR analysis with mutual adjustment helped to control for potential confounding factors, providing additional evidence for the robustness of the findings. Another limitation is that the overall sample size was large, but the sample size for specific cancers was small. Our analysis included individuals of East Asian origin, who have a high prevalence of chronic HBV infection, so our results may not be generalizable to other ancestry populations.\u003c/p\u003e \u003c/div\u003e"},{"header":"Conclusions","content":"\u003cp\u003eIn conclusion, this study revealed strong causal relationships between smoking status, HBsAg, and PSA and lung cancer, liver cancer, and prostate cancer, respectively. Notably, HBsAg demonstrated a causal association not only with liver cancer but also with gastric cancer and cervical cancer. However, the inverse relationship observed between HBsAg and breast cancer, as well as the association between serum bilirubin and lung cancer, suggests the need for additional research in these specific contexts.\u003c/p\u003e"},{"header":"Abbreviations","content":"\u003cp\u003eHBsAg: hepatitis B surface antigen\u003c/p\u003e\n\u003cp\u003eMR: Mendelian randomization\u003c/p\u003e\n\u003cp\u003eMVMR: multivariable Mendelian randomization\u003c/p\u003e\n\u003cp\u003eGWAS: genome-wide association study\u003c/p\u003e\n\u003cp\u003eBMI: body mass index\u003c/p\u003e\n\u003cp\u003eCEA: carcinoembryonic antigen\u003c/p\u003e\n\u003cp\u003eFBS: fasting blood sugar\u003c/p\u003e\n\u003cp\u003eTSH: thyroid-stimulating hormone\u003c/p\u003e\n\u003cp\u003ePSA: prostate-specific antigen\u003c/p\u003e\n\u003cp\u003eAST: aspartate aminotransferase\u003c/p\u003e\n\u003cp\u003eALT: alanine transferase\u003c/p\u003e\n\u003cp\u003eGGT: gamma-glutamyl transferase\u003c/p\u003e\n\u003cp\u003eSNPs: single nucleotide polymorphisms\u003c/p\u003e"},{"header":"Declarations","content":"\u003cp\u003e\u003cstrong\u003eAvailability of data and materials\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThe datasets used and/or analyzed during the current study are available from the corresponding author on reasonable request.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eFunding\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThis research was supported by the Basic Science Research Program through the National Research Foundation of Korea funded by the Ministry of Education (RS-2023-00239122).\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eEthical Statement\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThis study protocol was approved by the Institutional Review Board of the Severance Hospital (approval number: 4-2023-1722).\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eAuthor Contributions\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eConceptualization: Jee SH., Jung KJ., W. Spiller, Lee K.\u003c/p\u003e\n\u003cp\u003eData curation: Jee SH., Jung KJ., Song DS., Lee K.\u003c/p\u003e\n\u003cp\u003eFormal analysis: Jung KJ., Song DS., W. Spiller,\u003c/p\u003e\n\u003cp\u003eFunding acquisition: Jee SH.\u003c/p\u003e\n\u003cp\u003eInvestigation: Jung KJ, W. Spiller, Shin JW, Song DS., Lee K, Jee SH\u003c/p\u003e\n\u003cp\u003eMethodology: W. Spiller, Jung KJ, Jee SH\u003c/p\u003e\n\u003cp\u003eProject administration: Jee SH., Lee K\u003c/p\u003e\n\u003cp\u003eSupervision: Jee SH., Lee K\u003c/p\u003e\n\u003cp\u003eWriting \u0026ndash; original draft: Jung KJ, W. Spiller, Lee K, Jee SH,\u003c/p\u003e\n\u003cp\u003eWriting \u0026ndash; review \u0026amp; editing: Jung KJ, W. Spiller, Shin JW, Song DS., Lee K, Jee SH\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eConflicts of \u003c/strong\u003e\u003cstrong\u003einterest\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eConflict of interest relevant to this article was not reported.\u003c/p\u003e"},{"header":"References","content":"\u003col\u003e\n\u003cli\u003eLeiter, A., R.R. Veluswamy, and J.P. Wisnivesky, \u003cem\u003eThe global burden of lung cancer: current status and future trends.\u003c/em\u003e Nature Reviews Clinical Oncology, 2023. \u003cstrong\u003e20\u003c/strong\u003e(9): p. 624-639.\u003c/li\u003e\n\u003cli\u003eKaur, S.P., et al., \u003cem\u003eHepatocellular carcinoma in hepatitis B virus-infected patients and the role of hepatitis B surface antigen (HBsAg).\u003c/em\u003e Journal of clinical medicine, 2022. \u003cstrong\u003e11\u003c/strong\u003e(4): p. 1126.\u003c/li\u003e\n\u003cli\u003eSkrivankova, V.W., et al., \u003cem\u003eStrengthening the reporting of observational studies in epidemiology using mendelian randomisation (STROBE-MR): explanation and elaboration.\u003c/em\u003e bmj, 2021. \u003cstrong\u003e375\u003c/strong\u003e.\u003c/li\u003e\n\u003cli\u003eDavey Smith, G. and G. Hemani, \u003cem\u003eMendelian randomization: genetic anchors for causal inference in epidemiological studies.\u003c/em\u003e Human molecular genetics, 2014. \u003cstrong\u003e23\u003c/strong\u003e(R1): p. R89-R98.\u003c/li\u003e\n\u003cli\u003ePlotnikov, D. and J.A. Guggenheim, \u003cem\u003eMendelian randomisation and the goal of inferring causation from observational studies in the vision sciences.\u003c/em\u003e Ophthalmic and Physiological Optics, 2019. \u003cstrong\u003e39\u003c/strong\u003e(1): p. 11-25.\u003c/li\u003e\n\u003cli\u003eLawlor, D.A., et al., \u003cem\u003eMendelian randomization: using genes as instruments for making causal inferences in epidemiology.\u003c/em\u003e Statistics in medicine, 2008. \u003cstrong\u003e27\u003c/strong\u003e(8): p. 1133-1163.\u003c/li\u003e\n\u003cli\u003eJee, Y.H., et al., \u003cem\u003eCohort Profile: The Korean Cancer Prevention Study-II (KCPS-II) Biobank.\u003c/em\u003e International Journal of Epidemiology, 2017. \u003cstrong\u003e47\u003c/strong\u003e(2): p. 385-386f.\u003c/li\u003e\n\u003cli\u003eKim, Y., B.-G. Han, and K. Group, \u003cem\u003eCohort profile: the Korean genome and epidemiology study (KoGES) consortium.\u003c/em\u003e International journal of epidemiology, 2017. \u003cstrong\u003e46\u003c/strong\u003e(2): p. e20-e20.\u003c/li\u003e\n\u003cli\u003eVon Hinke, S., et al., \u003cem\u003eGenetic markers as instrumental variables.\u003c/em\u003e Journal of Health Economics, 2016. \u003cstrong\u003e45\u003c/strong\u003e: p. 131-148.\u003c/li\u003e\n\u003cli\u003eBurgess, S., A. Butterworth, and S.G. Thompson, \u003cem\u003eMendelian randomization analysis with multiple genetic variants using summarized data.\u003c/em\u003e Genetic epidemiology, 2013. \u003cstrong\u003e37\u003c/strong\u003e(7): p. 658-665.\u003c/li\u003e\n\u003cli\u003eWald, A., \u003cem\u003eThe fitting of straight lines if both variables are subject to error.\u003c/em\u003e The annals of mathematical statistics, 1940. \u003cstrong\u003e11\u003c/strong\u003e(3): p. 284-300.\u003c/li\u003e\n\u003cli\u003eBowden, J., G. Davey Smith, and S. Burgess, \u003cem\u003eMendelian randomization with invalid instruments: effect estimation and bias detection through Egger regression.\u003c/em\u003e International journal of epidemiology, 2015. \u003cstrong\u003e44\u003c/strong\u003e(2): p. 512-525.\u003c/li\u003e\n\u003cli\u003eSanderson, E., et al., \u003cem\u003eAn examination of multivariable Mendelian randomization in the single-sample and two-sample summary data settings.\u003c/em\u003e International journal of epidemiology, 2019. \u003cstrong\u003e48\u003c/strong\u003e(3): p. 713-727.\u003c/li\u003e\n\u003cli\u003eSanderson, E., W. Spiller, and J. Bowden, \u003cem\u003eTesting and correcting for weak and pleiotropic instruments in two-sample multivariable mendelian randomisation.\u003c/em\u003e BioRxiv, 2020: p. 2020.04. 02.021980.\u003c/li\u003e\n\u003cli\u003eLarsson, S.C., et al., \u003cem\u003eSmoking, alcohol consumption, and cancer: a mendelian randomisation study in UK Biobank and international genetic consortia participants.\u003c/em\u003e PLoS medicine, 2020. \u003cstrong\u003e17\u003c/strong\u003e(7): p. e1003178.\u003c/li\u003e\n\u003cli\u003eGandini, S., et al., \u003cem\u003eTobacco smoking and cancer: a meta\u003c/em\u003e\u003cem\u003e‐analysis.\u003c/em\u003e International journal of cancer, 2008. \u003cstrong\u003e122\u003c/strong\u003e(1): p. 155-164.\u003c/li\u003e\n\u003cli\u003ePirie, K., et al., \u003cem\u003eThe 21st century hazards of smoking and benefits of stopping: a prospective study of one million women in the UK.\u003c/em\u003e The Lancet, 2013. \u003cstrong\u003e381\u003c/strong\u003e(9861): p. 133-141.\u003c/li\u003e\n\u003cli\u003eKamiza, A.B., et al., \u003cem\u003eHepatitis B infection is causally associated with extrahepatic cancers: A Mendelian randomization study.\u003c/em\u003e EBioMedicine, 2022. \u003cstrong\u003e79\u003c/strong\u003e.\u003c/li\u003e\n\u003cli\u003eCollette, L., T. Burzykowski, and F.H. Schr\u0026ouml;der, \u003cem\u003eProstate-specific antigen (PSA) alone is not an appropriate surrogate marker of long-term therapeutic benefit in prostate cancer trials.\u003c/em\u003e European Journal of Cancer, 2006. \u003cstrong\u003e42\u003c/strong\u003e(10): p. 1344-1350.\u003c/li\u003e\n\u003cli\u003eCaram, M., T.A. Skolarus, and K.A. Cooney, \u003cem\u003eLimitations of prostate-specific antigen testing after a prostate cancer diagnosis.\u003c/em\u003e European urology, 2016. \u003cstrong\u003e70\u003c/strong\u003e(2): p. 209-210.\u003c/li\u003e\n\u003cli\u003eLucifora, J. and U. Protzer, \u003cem\u003eAttacking hepatitis B virus cccDNA\u0026ndash;The holy grail to hepatitis B cure.\u003c/em\u003e Journal of hepatology, 2016. \u003cstrong\u003e64\u003c/strong\u003e(1): p. S41-S48.\u003c/li\u003e\n\u003cli\u003eLevrero, M. and J. Zucman-Rossi, \u003cem\u003eMechanisms of HBV-induced hepatocellular carcinoma.\u003c/em\u003e Journal of hepatology, 2016. \u003cstrong\u003e64\u003c/strong\u003e(1): p. S84-S101.\u003c/li\u003e\n\u003cli\u003eSong, C., et al., \u003cem\u003eAssociations between hepatitis B virus infection and risk of all cancer types.\u003c/em\u003e JAMA network open, 2019. \u003cstrong\u003e2\u003c/strong\u003e(6): p. e195718-e195718.\u003c/li\u003e\n\u003cli\u003eAbe, S.K., et al., \u003cem\u003eHepatitis B and C Virus Infection and Risk of Pancreatic Cancer: A Population-Based Cohort Study (JPHC Study Cohort II).\u003c/em\u003e Cancer Epidemiology, Biomarkers \u0026amp; Prevention, 2016. \u003cstrong\u003e25\u003c/strong\u003e(3): p. 555-557.\u003c/li\u003e\n\u003cli\u003eChang, M.C., et al., \u003cem\u003eHepatitis B and C viruses are not risks for pancreatic adenocarcinoma.\u003c/em\u003e World J Gastroenterol, 2014. \u003cstrong\u003e20\u003c/strong\u003e(17): p. 5060-5.\u003c/li\u003e\n\u003cli\u003eXu, J.H., et al., \u003cem\u003eHepatitis B or C viral infection and risk of pancreatic cancer: a meta-analysis of observational studies.\u003c/em\u003e World J Gastroenterol, 2013. \u003cstrong\u003e19\u003c/strong\u003e(26): p. 4234-41.\u003c/li\u003e\n\u003cli\u003eHorsfall, L.J., et al., \u003cem\u003eGenetically raised serum bilirubin levels and lung cancer: a cohort study and Mendelian randomisation using UK Biobank.\u003c/em\u003e Thorax, 2020. \u003cstrong\u003e75\u003c/strong\u003e(11): p. 955-964.\u003c/li\u003e\n\u003cli\u003eSeyed Khoei, N., et al., \u003cem\u003eGenetically raised circulating bilirubin levels and risk of ten cancers: A mendelian randomization study.\u003c/em\u003e Cells, 2021. \u003cstrong\u003e10\u003c/strong\u003e(2): p. 394.\u003c/li\u003e\n\u003c/ol\u003e"}],"fulltextSource":"","fullText":"","funders":[],"hasAdminPriorityOnWorkflow":false,"hasManuscriptDocX":true,"hasOptedInToPreprint":true,"hasPassedJournalQc":"","hasAnyPriority":false,"hideJournal":true,"highlight":"","institution":"","isAcceptedByJournal":false,"isAuthorSuppliedPdf":false,"isDeskRejected":"","isHiddenFromSearch":false,"isInQc":false,"isInWorkflow":false,"isPdf":false,"isPdfUpToDate":true,"isWithdrawnOrRetracted":false,"journal":{"display":true,"email":"[email protected]","identity":"researchsquare","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":true,"externalIdentity":"","sideBox":"","snPcode":"","submissionUrl":"/submission","title":"Research Square","twitterHandle":"researchsquare","acdcEnabled":true,"dfaEnabled":false,"editorialSystem":"","reportingPortfolio":"","inReviewEnabled":false,"inReviewRevisionsEnabled":true},"keywords":"Cancer, Hepatitis B antigen, Bilirubin, Gamma-glutamyl transferase","lastPublishedDoi":"10.21203/rs.3.rs-4249634/v1","lastPublishedDoiUrl":"https://doi.org/10.21203/rs.3.rs-4249634/v1","license":{"name":"CC BY 4.0","url":"https://creativecommons.org/licenses/by/4.0/"},"manuscriptAbstract":"\u003cp\u003e\u003cstrong\u003eBackground\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eDespite an extensive body of observational research related to risk factors for cancer incidence, it is unclear whether the estimated associations are causal or a result of unmeasured confoundingfactors. To consider this possibility, this study explored a range of candidate epidemiological factors associated with the onset of cancer within a Mendelian randomization framework.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eMethods\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eMultivariate Mendelian randomization (MVMR) analyses were conducted using data from the Korean Cancer Prevention Study-II Biobank and the Korean Genome Epidemiologic Study. Analyses were performed to investigate 13 cancer-related risk factors and 13 types of cancer. Initially, univariate Mendelian randomization analyses were performed for each factor, estimating its association with cancer. Subsequently, a set of factors was explored using MVMR.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eResults\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eBy analyzing factors related to the onset of cancer, it was determined that smoking is associated with lung cancer, while hepatitis B surface antigen (HBsAg) positivity is significantly linked to gastric cancer, liver cancer, and cervical cancer. PSA levels are estimated to be causally related to prostate cancer, while bilirubin has emerged as a novel factor showing a positive association with lung cancer. To confirm the causal effect between HBsAg and cancer, a MVMR was conducted, controlling for bilirubin and gamma-glutamyl transferase. The results indicated a positive association between HBsAg and cervical cancer, liver cancer, and lung cancer. Conversely, breast cancer and pancreatic cancer showed a negative association. In the case of breast cancer, individuals with HBsAg at the age of over 50 years exhibited a significantly lower risk, with an odds ratio of 0.87 (\u003cem\u003eP \u003c/em\u003e= 3.07 × 10\u003csup\u003e-16\u003c/sup\u003e).\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eConclusions\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eSmoking status, HBsAg, and PSA levels replicated findings from previous studies suggesting causal relationships. However, bilirubin and HBsAg demonstrated positive causal associations with some cancers, while HBsAg exhibited negative associations with other cancers. Further research is warranted to explore the cancer-specific causality of HBsAg.\u003c/p\u003e","manuscriptTitle":"Examining complex cancer etiologies within the Korean population: A high-throughput multivariable Mendelian randomization study","msid":"","msnumber":"","nonDraftVersions":[{"code":1,"date":"2024-04-25 15:45:49","doi":"10.21203/rs.3.rs-4249634/v1","editorialEvents":[{"type":"communityComments","content":0}],"status":"published","journal":{"display":true,"email":"[email protected]","identity":"researchsquare","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":true,"externalIdentity":"","sideBox":"","snPcode":"","submissionUrl":"/submission","title":"Research Square","twitterHandle":"researchsquare","acdcEnabled":true,"dfaEnabled":false,"editorialSystem":"","reportingPortfolio":"","inReviewEnabled":false,"inReviewRevisionsEnabled":true}}],"origin":"","ownerIdentity":"a62effae-d9db-415a-b6d0-544c0205ccd7","owner":[],"postedDate":"April 25th, 2024","published":true,"recentEditorialEvents":[],"rejectedJournal":[],"revision":"","amendment":"","status":"posted","subjectAreas":[],"tags":[],"updatedAt":"2024-06-19T05:45:23+00:00","versionOfRecord":[],"versionCreatedAt":"2024-04-25 15:45:49","video":"","vorDoi":"","vorDoiUrl":"","workflowStages":[]},"version":"v1","identity":"rs-4249634","journalConfig":"researchsquare"},"__N_SSP":true},"page":"/article/[identity]/[[...version]]","query":{"redirect":"/article/rs-4249634","identity":"rs-4249634","version":["v1"]},"buildId":"qtupq5eGEP_6zYnWcrvyt","isFallback":false,"isExperimentalCompile":false,"dynamicIds":[84888],"gssp":true,"scriptLoader":[]}