{"paper_id":"016c1af8-73a8-4a5d-9888-4dd05ec8e1df","body_text":"1 \n \nGenetic analyses of gynecological disease identify genetic relationships between uterine \nfibroids and endometrial cancer, and a novel endometrial cancer genetic risk region at \nthe WNT4 1p36.12 locus \n \nAuthors \nPik Fang Kho1, 2, Sally Mortlock3, Endometrial Cancer Association Consortium, International \nEndometriosis Genetics Consortium, Peter A.W. Rogers 4, Dale R. Nyholt 2, Grant W. \nMontgomery3, Amanda B. Spurdle1, Dylan M. Glubb1*, Tracy A. O’Mara1*.   \n \nAffiliations \n1Department of Genetics and Computational Biology, QIMR Berghofer Medical Research \nInstitute, Brisbane, Queensland, Australia. \n2School of Biomedical Science, Faculty of Health, Queensland University of Technology, \nBrisbane, Queensland, Australia. \n3The Institute for Molecular Bioscience, The University of Queensland, Brisbane, \nQueensland, Australia. \n4Department of Obstetrics and Gynaecology, Gynaecology Research Centre, Royal Women’s \nHospital, University of Melbourne, Parkville, Victoria, Australia. \n*These authors contributed equally to the work. \n \nCorresponding Author \nDr Tracy O’Mara, PhD, Molecular Cancer Epidemiology Group, QIMR Berghofer Medical \nResearch Institute, 300 Herston Road, Brisbane QLD Australia 4006. Phone: +61 7 3362 \n0389. Email: Tracy.OMara@qimrberghofer.edu.au \n \n  \n . CC-BY 4.0 International licenseIt is made available under a \n is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review)\nThe copyright holder for this preprint this version posted April 1, 2021. ; https://doi.org/10.1101/2020.11.09.20228114doi: medRxiv preprint \nNOTE: This preprint reports new research that has not been certified by peer review and should not be used to guide clinical practice.\n\n2 \n \nDeclarations \nFunding  \nPFK is supported by an Australian Government Research Training Program PhD Scholarship \nand QIMR Berghofer Postgraduate Top-Up Scholarship, TAO’M, GWM and ABS are \nsupported by NHMRC Investigator Fellowships (APP1173170, GNT1177194 and \nAPP1177524).  \nThis work was supported by National Health and Medical Research Council (NHMRC) \nProject Grants (APP1109286, GNT1026033, GNT1105321 and GNT1147846). Funding \nsources had no role in study design, data curation and analysis, data interpretation, report \nwriting and submission for publication. \n \nConflict of interest/Competing interests \nThe authors declare no potential conflicts of interest. \n \nEthics approval \nThis work used summary-level GWAS meta-analysis results, and thus ethical approval was \nnot required.  \n \nConsent to participate \nNot applicable \n \nConsent for publication \nNot applicable \n \nAvailability of data and material (data transparency) \nSummary-level GWAS meta-analysis results for PCOS, uterine fibroids and endometrial \ncancer that support the findings of this study are available at the NHGRI-EBI GWAS Catalog \n(https://www.ebi.ac.uk/gwas/downloads/summary-statistics\n). Other data generated during this \nstudy are included in this article and its supplementary information files or are available on \nreasonable request.  \n \nCode availability \nNot applicable \n . CC-BY 4.0 International licenseIt is made available under a \n is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review)\nThe copyright holder for this preprint this version posted April 1, 2021. ; https://doi.org/10.1101/2020.11.09.20228114doi: medRxiv preprint \n\n3 \n \nAbstract \nEndometriosis, polycystic ovary syndrome (PCOS) and uterine fibroids have been proposed \nas endometrial cancer risk factors; however, disentangling their relationships with \nendometrial cancer is complicated due to shared risk factors and comorbidities. Using \ngenome-wide association study (GWAS) data, we explored the relationships between these \nnon-cancerous gynecological diseases and endometrial cancer risk by assessing genetic \ncorrelation, causal relationships and shared risk loci. We found significant genetic correlation \nbetween endometrial cancer and PCOS, and uterine fibroids. Adjustment for genetically \npredicted body mass index (a risk factor for PCOS, uterine fibroids and endometrial cancer) \nsubstantially attenuated the genetic correlation between endometrial cancer and PCOS but did \nnot affect the correlation with uterine fibroids. Mendelian randomization analyses provided \nevidence of a causal relationship between only uterine fibroids and endometrial cancer. Gene-\nbased analyses revealed risk regions shared between endometrial cancer and endometriosis, \nand uterine fibroids. Multi-trait GWAS analysis of endometrial cancer and the genetically \ncorrelated gynecological diseases identified a novel genome-wide significant endometrial \ncancer risk locus at 1p36.12, which replicated in an independent endometrial cancer dataset. \nInterrogation of functional genomic data at 1p36.12 revealed biologically relevant genes, \nincluding WNT4 which is necessary for the development of the female reproductive system. \nIn summary, our study provides genetic evidence for a causal relationship between uterine \nfibroids and endometrial cancer. It further provides evidence that the comorbidity of \nendometrial cancer, PCOS and uterine fibroids may partly be due to shared genetic \narchitecture. Notably, this shared architecture has revealed a novel genome-wide risk locus \nfor endometrial cancer. \n \nKeywords \nEndometrial cancer, endometriosis, polycystic ovary syndrome, uterine fibroids, genetic \ncorrelation, Mendelian randomization. \n \n  \n . CC-BY 4.0 International licenseIt is made available under a \n is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review)\nThe copyright holder for this preprint this version posted April 1, 2021. ; https://doi.org/10.1101/2020.11.09.20228114doi: medRxiv preprint \n\n4 \n \nIntroduction \nEndometriosis, polycystic ovary syndrome (PCOS) and uterine fibroids are three common \nnon-cancerous gynecological diseases affecting 10-15% (Parasar et al. 2017), 6-9% (Azziz et \nal. 2011) and 5-69% (Stewart et al. 2017) of women of reproductive age, respectively; \nhowever, their prevalence is likely underestimated because of under diagnosis (Agarwal et al. \n2019; De La Cruz and Buchanan 2017). Although these non-cancerous gynecological \ndiseases primarily affect premenopausal women and endometrial cancer is largely a \npostmenopausal malignancy, many risk factors are shared with endometrial cancer (e.g. \nchronic estrogen exposure, inflammation, insulin resistance and obesity (Harris and Terry \n2016; Li et al. 2019; Wise et al. 2016)), suggesting some shared biological relationship.  \n \nA number of studies have used observational data to assess associations between the three \nnon-cancerous gynecological diseases and endometrial cancer risk, the findings of which \nhave been heterogeneous (Harris and Terry 2016; Johnatty et al. 2020; Li et al. 2019; Wise et \nal. 2016). Indeed, the use of observational studies to evaluate these associations could be \nconfounded by: (i) the failure to adequately account for potential confounders that are \nassociated with risk of endometrial cancer and/or gynecological disease e.g. oral \ncontraceptive use; (ii) the reliance of disease status classification on self-reported data which \nis subject to misclassification bias from asymptomatic undiagnosed cases; (iii) misdiagnosis \nof early stage endometrial cancer as uterine fibroids due to shared clinical presentation (Wise \net al. 2016); (iv) detection bias in cohort studies as a result of an increased surveillance for \nendometrial cancer among patients with non-cancerous gynecological diseases; and (v) the \ncomorbidity of non-cancerous gynecological diseases (Choi et al. 2017; Johnatty et al. 2020; \nMatalliotaki et al. 2018; Nagai et al. 2015; Uimari et al. 2011; Wise et al. 2007). Thus, it \nremains difficult to determine from observational studies the precise nature of the \nrelationships between endometrial cancer and these non-cancerous gynecological diseases.  \n \nGenome-wide association study (GWAS) data have demonstrated genetic correlation \nbetween endometrial cancer and endometriosis (Masuda et al. 2020; Painter et al. 2018), and \nuterine fibroids (Masuda et al. 2020), which may partly explain the comorbidities of these \ndiseases; whether these comorbidities are due to causal relationships or shared genetic \netiology remains to be explained. In this study, we have used a variety of approaches to \nanalyze GWAS data and elucidate relationships between endometrial cancer and non-\n . CC-BY 4.0 International licenseIt is made available under a \n is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review)\nThe copyright holder for this preprint this version posted April 1, 2021. ; https://doi.org/10.1101/2020.11.09.20228114doi: medRxiv preprint \n\n5 \n \ncancerous gynecological disease (summarized in Supplementary Figure 1). Firstly, we have \nperformed genetic correlation analysis, using the largest currently available datasets to clarify \nthe shared genetic risk between the non-cancerous gynecological diseases and endometrial \ncancer. As inherited genetic variants are less influenced by confounding inherent in \nobservational studies, we have performed genetic causal inference analyses using \ngynecological disease-associated variants to investigate causal relationships. It is possible \nthat these diseases may not be genetically correlated or causally related to endometrial cancer \nbut share overlapping genetic risk regions. To assess this possibility, we have performed \ngene-based association analyses. Lastly, we have performed multi-trait GWAS, leveraging \ngenetic correlation between endometrial cancer and non-cancerous gynecological diseases to \ndiscover novel GWAS risk loci.  \n \nMaterials and Methods \nGWAS data \nGWAS summary data publicly available for PCOS (Day et al. 2018) \n(https://doi.org/10.17863/CAM.27720) and uterine fibroids (Gallagher et al. 2019) \n(ftp://ftp.ebi.ac.uk/pub/databases/gwas/summary_statistics/GallagherCS_31649266_GCST00\n9158), and via collaboration for endometriosis (Sapkota et al. 2017), were used for all \nanalyses except Mendelian randomization. For PCOS and uterine fibroids, GWAS summary \ndata from the 23andMe, Inc., cohort had been excluded because of restrictions related to data \nsharing agreements (Day et al. 2018; Gallagher et al. 2019). For Mendelian randomization \nanalyses, risk estimates and respective standard errors of genome-wide significant variants \nwere accessed from the largest published GWAS for each disease (Day et al. 2018; Gallagher \net al. 2019; Rahmioglu et al. 2018). Details of studies and sample sizes used in each analysis \nare shown in Supplementary Table 1.  Detailed descriptions of the quality control \nprocedures and GWAS analysis can be found in the corresponding publications.  \n \nGWAS summary data for endometrial cancer were available from O'Mara et al. (2018). As \nthe GWAS for endometrial cancer (O'Mara et al. 2018), endometriosis (Rahmioglu et al. \n2018), and uterine fibroids (Gallagher et al. 2019) included participants from the UK Biobank \n(https://www.ukbiobank.ac.uk/), we re-analyzed the endometrial cancer dataset, excluding \nthese participants to avoid sample overlap bias in the two sample Mendelian randomization \nanalysis. This also allowed us to use the UK Biobank endometrial cancer dataset as part of \n . CC-BY 4.0 International licenseIt is made available under a \n is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review)\nThe copyright holder for this preprint this version posted April 1, 2021. ; https://doi.org/10.1101/2020.11.09.20228114doi: medRxiv preprint \n\n6 \n \nthe replication set to confirm multi-trait GWAS results. This revised endometrial cancer \nGWAS meta-analysis consisted of 12,270 cases and 46,126 controls of European descent. \nGenetic variants with minor allele frequency (MAF) < 1% and imputation information score \n< 0.4 were excluded, leaving ~9 million genetic variants. The revised endometrial cancer \nGWAS data were used only in Mendelian randomization and multi-trait GWAS analyses, \nwhile the published endometrial cancer GWAS data (O'Mara et al. 2018) were used in all \nother analyses. Prior to genetic correlation analysis, genetic variants in the extended human \nmajor histocompatibility complex region (26–34 Mb on chromosome 6) were removed due to \nthe complex linkage disequilibrium (LD) structure in this region. \n \nGenetic correlation between endometrial cancer and non-cancerous gynecological diseases \n \nWe used LD Score regression (Bulik-Sullivan et al. 2015) to estimate the genetic correlation \nbetween endometrial cancer and each non-cancerous gynecological disease. Genetic \ncorrelation analyses were restricted to common HapMap3 variants (MAF > 0.01). To reduce \nbias from potential residual confounding in genetic correlation analyses, including bias from \nunknown sample overlap, we used the estimated genetic covariance intercept, obtained \nwithout constraint. Genetic correlation values range from -1 to 1; positive values indicated \nthat shared genetic variants have concordant effects across the genome, whereas negative \nvalues indicated divergent effects. \n \nObesity is a major risk factor for endometrial cancer, and is prevalent amongst women with \nPCOS and uterine fibroids (Ilaria and Marci 2018; Sam 2007). For genetic correlation \nanalysis between endometrial cancer and PCOS or uterine fibroids, we thus additionally \ncorrected for the effect of obesity, as measured by genetically predicted BMI. PCOS and \nuterine fibroids GWAS were conditioned using summary data from a large GWAS of BMI \n(Yengo et al. 2018) in GCTA-mtCOJO analysis (Zhu et al. 2018) before performing LD score \nregression analysis. \n \nGenetic causal inference tests\n \nWe performed two-sample Mendelian randomization analysis to explore potential causal \nrelationships between non-cancerous gynecological diseases and endometrial cancer. \nIndependent (LD r\n2 < 0.01) genetic variants associated with the non-cancerous gynecological \ndiseases at genome wide significance (P < 5 × 10-8) were used as genetic instruments. The list \nof genetic instruments and the respective risk association estimates were extracted from the \n . CC-BY 4.0 International licenseIt is made available under a \n is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review)\nThe copyright holder for this preprint this version posted April 1, 2021. ; https://doi.org/10.1101/2020.11.09.20228114doi: medRxiv preprint \n\n7 \n \nlargest GWAS of endometriosis (Rahmioglu et al. 2018), PCOS (Day et al. 2018) and uterine \nfibroids (Gallagher et al. 2019). We excluded independent genetic variants with ambiguous \nalleles and intermediate frequencies (i.e., variants with A/T or C/G alleles and minor allele \nfrequency of more than 0.42), leaving 26 variants as genetic instruments for endometriosis, \n14 for PCOS and 23 for uterine fibroids. \n \n \nAs the three non-cancerous gynecological diseases mostly affect premenopausal women and \nendometrial cancer primarily affects postmenopausal women, we performed a unidirectional \nMendelian randomization analysis, assessing the effect of genetic predisposition to non-\ncancerous gynecological disease on endometrial cancer risk. We used inverse variance \nweighted (IVW) analysis as the primary analysis by regressing the genetic variant-\nendometrial cancer association on the genetic variant-non-cancerous gynecological disease \nassociation, weighted by inverse of their variance. This method has the most power to detect \nassociations although it has a strong assumption of no heterogeneity (potentially resulting \nfrom pleiotropy) amongst genetic variants (Hemani et al. 2018); thus, this method assumes all \ngenetic variants for the exposure of interest have a proportional effect on outcome risk.  \n \nWe also performed several sensitivity analyses for Mendelian randomization that are more \nrobust to heterogeneity amongst genetic variants: MR-Egger, weighted median, and weighted \nmode analysis. MR-Egger analysis regresses the genetic variant-outcome association on the \ngenetic variant-exposure association, without constraining the regression intercept (Bowden \net al. 2015). If the MR-Egger regression intercept is non-zero, it provides evidence that \ndirectional horizontal pleiotropy amongst genetic variants is driving the causal estimates (i.e. \ngenetic variation influences the outcome through a pathway other than the exposure and \nindicates that the ratio of genetic variants with positive and negative pleiotropic effects is not \nbalanced). The MR-Egger regression slope represents a valid effect estimate after adjustment \nfor pleiotropic effects, provided the Instrument Strength Independent of Direct Effect \n(InSIDE) assumption is met (i.e. the association of a genetic variant with the exposure of \ninterest is independent from its direct effect on outcome) (Bowden et al. 2015). We also \nperformed weighted median (Bowden et al. 2016) and weighted mode (Hartwig et al. 2017) \nanalyses, which are more robust to violation of the InSIDE assumption. Weighted median \nanalysis relies on the assumption that more than 50% of the weights come from valid genetic \ninstruments (Bowden et al. 2016), while weighted mode analysis relies on the assumption \nthat most of the weights come from valid genetic instruments (Hartwig et al. 2017). \n . CC-BY 4.0 International licenseIt is made available under a \n is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review)\nThe copyright holder for this preprint this version posted April 1, 2021. ; https://doi.org/10.1101/2020.11.09.20228114doi: medRxiv preprint \n\n8 \n \nCochran’s Q statistic was used to assess the heterogeneity in the effects of variants (a \npotential indicator of horizontal pleiotropy) (Bowden et al. 2018), and leave-one-out analysis \nwas used to assess whether a single variant drives the causal association (Hemani et al. 2018).  \n \nTwo-sample Mendelian randomization analysis was performed using the “TwoSampleMR” \n(Hemani et al. 2018) package in R. Unless stated otherwise, results with a Bonferroni-\ncorrected p-value for testing the effects of the three non-cancerous gynecological diseases (P \n< 0.05/3 = 0.017) on endometrial cancer risk were considered statistically significant. \n \nGene-based association analysis\n \nTo identify genetic risk regions shared between the non-cancerous gynecological diseases \nand endometrial cancer, we performed gene-based analysis using the fast and flexible set-\nbased association test (fastBAT) (Bakshi et al. 2016). fastBAT was used to perform an \nenrichment analysis on GWAS risk variants, located within 50kb of gene regions, for the \nnon-cancerous gynecological cancers and endometrial cancer. A random sample of 10,000 \nunrelated participants from the UK Biobank was used as the reference panel in these \nanalyses. We applied a false discovery rate (FDR) < 0.05 for the gene-based analysis, and \nadjacent risk-associated genes were considered a single risk region if within 1 Mb of each \nother.  \n \nMulti-trait analysis of GWAS (MTAG)  \nMTAG (Turley et al. 2018) was used to improve endometrial cancer risk loci discovery \nthrough joint analysis of endometrial cancer and non-cancerous gynecological diseases that \nshowed evidence of genetic correlation with endometrial cancer (i.e. PCOS and uterine \nfibroids). GWAS summary statistics were used as input and bivariate LD score regression \nwas used to account for sample overlap. Using pre-computed LD scores for Europeans, \nMTAG analysis was performed on common variants (MAF > 0.01). Alleles of genetic \nvariants were aligned across GWAS, and only variants present across included studies were \nassessed by MTAG. The final number of included variants for MTAG was 4,734,443. \nSummary statistics were produced for each trait where effect sizes and standard error \nestimates could be interpreted as the output from a single-trait GWAS.   \n \nReplication of novel endometrial cancer GWAS risk loci \n . CC-BY 4.0 International licenseIt is made available under a \n is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review)\nThe copyright holder for this preprint this version posted April 1, 2021. ; https://doi.org/10.1101/2020.11.09.20228114doi: medRxiv preprint \n\n9 \n \nMTAG assumes that the variance-covariance matrix across traits is homogenous across the \ngenome, but it is likely some variants are null for one trait and not null for another trait(s). \nViolation of this assumption could increase false positive discovery in MTAG (discussed in \n(Turley et al. 2018)). To address this issue, we tested the replication of novel genome-wide \nsignificant endometrial cancer risk variants from MTAG in an independent GWAS meta-\nanalysis using data from the Finnish Biobank Study (FinnGen; https://www.finngen.fi/en) \nand the UK Biobank. Endometrial cancer GWAS summary statistics for 566 cases and \n75,822 controls were downloaded directly from FinnGen (data freeze 2; http://r2.finngen.fi/). \nQuality control procedures for the FinnGen GWAS data are described in \nhttps://finngen.gitbook.io/documentation/\n. For UK Biobank, we performed an endometrial \ncancer GWAS using genotype and phenotype data obtained under the application number \n25331. Endometrial cancer cases were defined based on ICD10 code (C54) in the data fields \nof 40006, 41270 and 41202. Controls were selected randomly from unrelated women \nparticipants ( π /i1 < 0.1) with no history of any cancers. GWAS was performed on 1,866 \ncases and 18,660 controls using REGENIE (Mbatchou et al. 2020) to implement a logistic \nmixed model, adjusting for genotyping array and the top 10 principal components. A genetic \nrelationship matrix was included in the model as a random effect to account for cryptic \nrelatedness and population stratification. As recommended by REGENIE, we excluded \ngenetic variants with MAF < 0.01, minor allele count below 100, genotype missingness \nabove 10% and variants which deviated from Hardy-Weinberg equilibrium (P value < 1×10\n-\n15). After quality control exclusion, a total of 9,789,172 SNPs remained in the GWAS \nanalysis. \n \nTo create a replication set, the FinnGen and UK Biobank GWAS results were meta-analysed \nby a fixed-effect inverse variance weighted model using “meta” software in R. Novel \nendometrial cancer genome-wide significant variants identified by MTAG were considered to \nhave replicated if they had the same effect direction, and a P-value < 0.05 for association in \nthe replication set.  \n \nIdentification of candidate target genes at the 1p36.12 endometrial cancer risk locus \nWe used previously generated promoter-associated HiChIP chromatin looping data from \nendometrial (one immortalized and three tumor) cell lines (O'Mara et al. 2019) to explore \npotential regulatory interactions between credible causal risk variants and gene promoters at \n . CC-BY 4.0 International licenseIt is made available under a \n is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review)\nThe copyright holder for this preprint this version posted April 1, 2021. ; https://doi.org/10.1101/2020.11.09.20228114doi: medRxiv preprint \n\n10 \n \nthe 1p36.12 locus. Credible causal risk variants were defined as variants with P-value for \nassociation within two orders of magnitude of the lead variant P-value. We also explored the \ncandidate target genes through overlap of credible causal variants with lead cis-eQTLs from \nGTEx v8 and the Blood eQTL Browser data (Munz et al. 2020).  \n \nResults \nWe found endometrial cancer to be significantly genetically correlated with PCOS (r\nG = 0.36, \nse = 0.12) and uterine fibroids ( rG = 0.24, se = 0.09) but not with endometriosis ( Table 1). \nAfter adjusting for genetically predicted BMI, the genetic correlation between PCOS and \nendometrial cancer was no longer statistically significant, indicating that the initial genetic \ncorrelation was, at least partly, mediated by genetically predicted BMI (Table 1). In contrast, \nthere was no material difference in the genetic correlation between uterine fibroids and \nendometrial cancer after adjusting for genetically predicted BMI ( Table 1), consistent with a \nprevious observation of no significant differences in BMI for endometrial cancer cases with \nor without uterine fibroids (Johnatty et al. 2020).  \n \nIVW Mendelian randomization analysis for the effects of genetic predisposition to the non-\ncancerous gynecological diseases on endometrial cancer provided evidence only for uterine \nfibroids (Table 2, Figure 1). Although sensitivity analyses were not statistically significant, \nthe directionality of the associations between uterine fibroids and endometrial cancer were \nconsistent with the IVW result ( Table 2, Figure 1). The MR-Egger intercept did not \nsignificantly differ from zero (Table 2) providing no evidence for confounding by directional \nhorizontal pleiotropy amongst genetic instruments. However, Cochran’s Q statistics indicated \nevidence of heterogeneity between causal estimates based on individual variants (Cochran’s \nQ statistics = 42.1, degrees of freedom = 22, P = 6×10\n-3), suggesting that some variants may \nbe associated with endometrial cancer risk through pathways other than uterine fibroids. \nLeave-one-out analysis showed that no single variant was driving the causal association \nrevealed by the IVW analysis (Supplementary Figure 2).  \n \nWhile genetic correlation analysis assesses the average genetic concordance across the \ngenome for two traits, it does not reveal common genomic regions that harbor trait-associated \nvariation. Further, a lack of evidence for genetic correlation may reflect opposing pleiotropic \neffects across the genome. Thus, we performed gene-based analyses to identify common risk \nregions across endometrial cancer and the non-cancerous gynecological diseases. The initial \n . CC-BY 4.0 International licenseIt is made available under a \n is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review)\nThe copyright holder for this preprint this version posted April 1, 2021. ; https://doi.org/10.1101/2020.11.09.20228114doi: medRxiv preprint \n\n11 \n \nanalysis revealed 24 genetic regions associated with endometrial cancer risk, 28 regions with \nendometriosis risk and 41 regions with uterine fibroids ( Supplementary Table 2 ). No \nassociations with PCOS passed FDR < 0.05, potentially reflecting a lack of power due to the \nsmall sample size of this cohort. We found four genetic risk regions (3q21.3, 9p21.3, 15q15.1 \nand 17q21.32), containing seven shared candidate susceptibility genes, were shared between \nendometriosis and endometrial cancer (Table 3). Three of these regions (9p21.3, 15q15.1 and \n17q21.32) have independently been associated with the risks of endometrial cancer (O'Mara \net al. 2018) and endometriosis through GWAS (Rahmioglu et al. 2018). The LD of lead risk \nvariants at each gene was compared and only one region (17q21.32) demonstrated evidence \nof a shared genetic risk signal ( r\n2 > 0.9; Table 3).  Additionally, we found two genetic risk \nregions (5p15.33 and 11p13), containing five shared candidate susceptibility genes, were \nshared between uterine fibroids and endometrial cancer ( Table 3 ). 5p15.33 has been \nassociated with uterine fibroids risk through GWAS (Gallagher et al. 2019) while 11p13 has \nindependently associated with uterine fibroids and endometrial cancer risk in GWAS \n(Gallagher et al. 2019; O'Mara et al. 2018). The LD of lead risk variants at each gene was \ncompared but there was no strong genetic correlation at either 5p15.33 or 11p13 ( r\n2 ≤  0.4; \nTable 3), suggesting that the genetic risk signals may be independent.  \n \nIncorporation of the two gynecological diseases genetically correlated with endometrial \ncancer (uterine fibroids and PCOS) in MTAG revealed ten genome-wide significant risk loci \nfor endometrial cancer (Table 4, Figure 2). We observed an inflation of median test statistics \nin the MTAG result ( λ = 1.06), which was likely due to a polygenic signal (LD score \nregression intercept = 0.98, se = 0.01) rather than population stratification. Two of the risk \nloci (5p15.33 and 1p36.12) were novel endometrial cancer genome-wide risk loci. We \nassessed both these risk loci in an independent endometrial cancer dataset and found that only \nthe association at the 1p36.12 locus replicated (Table 4).  \n \nTo identify candidate target genes at the replicated novel endometrial cancer GWAS risk \nlocus (1p36.12), we intersected candidate causal variants with promoter-associated chromatin \nloops from four endometrial (immortalized and tumor) cell lines (O'Mara et al. 2019). We \nidentified six candidate target genes through chromatin looping, including WNT4 for which a \ncandidate causal risk variant was revealed as a lead eQTL in lung tissue ( Supplementary \nTables 3 and 4; Figure 3). Additionally, we identified CDC42 as a candidate target gene \n . CC-BY 4.0 International licenseIt is made available under a \n is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review)\nThe copyright holder for this preprint this version posted April 1, 2021. ; https://doi.org/10.1101/2020.11.09.20228114doi: medRxiv preprint \n\n12 \n \nthrough a candidate causal risk variant located in a chromatin looping anchor at its promoter \n(Supplementary Table 3; Figure 3 ). Furthermore, candidate causal risk variants were lead \neQTLs for CDC42  in peripheral blood (Westra et al. 2013) ( Supplementary Table 4) , \nproviding additional evidence for regulatory targeting.  \n \nDiscussion \nUsing large-scale genome-wide datasets, we observed evidence of positive genetic \ncorrelation between endometrial cancer and PCOS, and uterine fibroids, but not \nendometriosis. The observed genetic correlation between endometrial cancer and PCOS was \nat least partly mediated by genetically predicted BMI, consistent with the role of BMI as a \nrisk factor for both PCOS and endometrial cancer. Mendelian randomization analysis \nprovided evidence for a causal relationship only between genetic predisposition to uterine \nfibroids and endometrial cancer risk. Gene-based analyses revealed several genetic risk \nregions shared between endometrial cancer and endometriosis, and uterine fibroids. This \nincluded one apparent joint genetic risk signal, for endometrial cancer and endometriosis at \n17q21.32. Multi-trait GWAS analysis, including endometrial cancer and the genetically \ncorrelated gynecological diseases identified two novel genome-wide significant risk loci for \nendometrial cancer, one of which (1p36.12) replicated in an independent endometrial cancer \ndataset. Lastly, functional analyses highlighted CDC42 and WNT4 as candidate target genes \nat the 1p36.12 endometrial cancer risk locus. \n \nTwo previous studies have reported a positive genetic correlation between endometriosis and \nendometrial cancer (Masuda et al. 2020; Painter et al. 2018), but we found no evidence for \nsuch genetic correlation. This discrepancy may be related to: i) the smaller sample sets used \nby the prior studies; ii) the ethnicity studied (Masuda et al. (2020) analyzed a Japanese \npopulation); or iii) the different genetic correlation analysis approaches used. For example, \nunlike Painter et al. (2018), we used an unconstrained LD score regression intercept to \naccount for potential residual confounding, resulting in a conservative estimate of genetic \ncorrelation. Indeed, we found the estimated genetic covariance intercept to be significantly \ndifferent from zero, suggesting the presence of bias from population stratification and/or \nsample overlap. The null results from the genetic causal inference analyses of endometriosis \nand endometrial cancer are concordant with observational studies that observed no \nassociations after controlling for ascertainment bias by excluding recent endometriosis \ndiagnosis (Melin et al. 2007; Olson et al. 2002; Rowlands et al. 2011).  \n . CC-BY 4.0 International licenseIt is made available under a \n is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review)\nThe copyright holder for this preprint this version posted April 1, 2021. ; https://doi.org/10.1101/2020.11.09.20228114doi: medRxiv preprint \n\n13 \n \n \nAlthough there was minimal evidence for genetic correlation or a causal relationship between \nendometriosis and endometrial cancer, we identified four shared genetic risk regions, three of \nwhich (9p21.3, 15q15.1 and 17q21.32) have independently been associated with risk of both \ndiseases through GWAS (O'Mara et al. 2018; Rahmioglu et al. 2018). Notably, the shared \ncandidate susceptibility genes at 9p21.3 (C DKN2B-AS1), 15q15.1 ( BMF) and 17q21.32 \n(CBX1, MIR1203, SKAP1 and SNX11) have been previously identified as candidate target \ngenes at endometrial cancer GWAS risk loci through promoter-associated chromatin looping \nstudies (O'Mara et al. 2019). The remaining shared endometriosis and endometrial cancer risk \nregion at 3q21.3 has not been independently identified by GWAS for either disease and may \nrepresent a novel GWAS risk locus for both in future studies. Indeed, this region was recently \nreported as an endometrial cancer risk region in a cross-cancer GWAS meta-analysis of \nendometrial, breast, ovarian and prostate cancer (Kar et al. 2020).  \n \nWe found PCOS and endometrial cancer to be genetically correlated but no association was \nobserved in genetic causal inference analyses, concordant with observational studies that \naccount for the effect of obesity (Fearnley et al. 2010; Zucchetto et al. 2009). These findings \nare consistent with our observation of substantial attenuation in genetic correlation between \nPCOS and endometrial cancer after adjusting for genetic components of BMI.  \n \nWe detected evidence of positive genetic correlation between uterine fibroids and \nendometrial cancer risk, consistent with observational studies (Fortuny et al. 2009; Rowlands \net al. 2011; Wise et al. 2016). IVW Mendelian randomization analysis provided evidence of a \ncausal relationship between genetic predisposition to uterine fibroids and endometrial cancer \nrisk. However, Cochran’s Q statistics showed evidence that variants used in the IVW analysis \nhad heterogeneous effects, suggesting that not all variants that increase uterine fibroids risk \nare expected to increase endometrial cancer risk. Although results from subsequent sensitivity \nanalyses that are robust to the presence of varying levels of pleiotropy were not statistically \nsignificant, they showed concordant effect directions with IVW result. It is important to note \nthat the Mendelian randomization sensitivity analyses have lower power to detect causal \nrelationships compared with IVW analysis. As genetic causal inference tests rely on the \nstatistical power of GWAS used, future larger GWAS are required to provide more accurate \ncausal estimates and thus greater confidence with regards to the nature of the relationship \nbetween uterine fibroids and endometrial cancer. \n . CC-BY 4.0 International licenseIt is made available under a \n is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review)\nThe copyright holder for this preprint this version posted April 1, 2021. ; https://doi.org/10.1101/2020.11.09.20228114doi: medRxiv preprint \n\n14 \n \n \nGene-based analysis revealed two genetic risk regions (5p15.33 and 11p13) that were shared \nby endometrial cancer and uterine fibroids. The 11p13 shared risk region has been associated \nwith the risks of uterine fibroids (Gallagher et al. 2019) and endometrial cancer (O'Mara et al. \n2018) in GWAS. From the gene-based analysis, we identified WT1 and WT1-AS as candidate \nsusceptibility genes for both uterine fibroids and endometrial cancer at 11p13.  Consistent \nwith this finding, we had previously established both genes  as candidate targets of \nendometrial cancer risk GWAS variation through promoter-associated chromatin looping \nstudies (O'Mara et al. 2019) and WT1 had also been identified through chromatin looping as a \ncandidate target of uterine fibroids risk variants (Rafnar et al. 2018). WT1 encodes a \ntranscription factor that is essential for urogenital development (reviewed by Roberts (2005)) \nand in the GTEx database of tissue gene expression it is most highly expressed in the uterus \n(https://gtexportal.org/home/\n). These observations suggest that alteration of uterine WT1 \nexpression by endometrial cancer and uterine fibroids genetic risk variation may affect \nsusceptibility to these diseases. \n \nThe 5p15.33 region was found to associate with endometrial cancer risk through both the \ngene-based analysis and the multi-trait GWAS. However, the multi-trait GWAS association \ndid not replicate in the independent endometrial cancer dataset, with discordant effect \ndirections and non-overlapping confidence intervals. Previously, this region has associated \nwith uterine fibroids risk in a GWAS (Gallagher et al. 2019), with endometrial cancer risk in \na candidate locus study (Carvajal-Carmona et al. 2015) and in a cross-cancer GWAS meta-\nanalysis of endometrial cancer and ovarian cancer (Glubb et al. 2021). The gene-based \nanalysis at this region revealed three candidate risk genes that were shared between uterine \nfibroids and endometrial cancer. The most biologically relevant of these genes is TERT, \nwhich encodes telomerase reverse transcriptase and maintains chromosomal stability by \nelongating the telomere (Rubtsova et al. 2012). Relevantly, chromosomes in uterine fibroids \n(Bonatz et al. 1998; Rogalla et al. 1995) and in endometrial tumors (reviewed by Alnafakh et \nal. (2019)) have been shown to have shorter telomere length. Indeed, a recent Mendelian \nrandomization study found genetically predicted telomere length to be strongly associated \nwith endometrial cancer risk (Telomeres Mendelian Randomization et al. 2017).   \n \nThe novel 1p36.12 endometrial cancer risk locus, revealed by the multi-trait GWAS, \nreplicated in the independent endometrial cancer GWAS dataset. Genetic variation at this \n . CC-BY 4.0 International licenseIt is made available under a \n is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review)\nThe copyright holder for this preprint this version posted April 1, 2021. ; https://doi.org/10.1101/2020.11.09.20228114doi: medRxiv preprint \n\n15 \n \nregion has associated with traits that are genetically correlated or causally related to \nendometrial cancer (e.g. heel bone mineral density (Morris et al. 2019), body mass index \n(Pulit et al. 2019), diabetes (Vujkovic et al. 2020), age at menarche (Kichaev et al. 2019) and \novarian cancer (Kuchenbaecker et al. 2015)). Furthermore, genetic variation at 1p36.12 has \nassociated with endometriosis and the lead endometrial cancer risk variant from the multi-\ntrait GWAS also represents a GWAS risk signal for pelvic organ prolapse (Olafsdottir et al. \n2020). Promoter-associated chromatin looping data from endometrial cell lines highlighted \nseven candidate target genes, two of which (CDC42 and WNT4) were supported by candidate \ncausal risk variants that represent eQTLs for these genes. The lead candidate causal risk \nvariant at 1p36.12 (rs3820282) and two candidate causal variants (rs61768001 & \nrs12037376) have previously been associated with expression of CDC42  in blood and a long \nnon-coding RNA (LINC00339) in blood and the endometrium (Mortlock et al. 2020). Semi-\nquantitative chromatin looping analysis in an endometrial cancer cell line demonstrated \nevidence of an interaction between a region containing rs3820282 and a ~15 kb region \ncontaining the promoter of LINC00339 (Powell et al. 2016). However, the quantitative \nchromatin looping data from the HiChIP analysis of the normal immortalized and tumoral \nendometrial cell lines (O'Mara et al. 2019)), which also has much greater resolution  (Lareau \nand Aryee 2018), did not provide evidence for a physical interaction between LINC00339 and \ncandidate causal endometrial cancer risk variants.  \n \nLINC00339, CDC42 and WNT4 all have biological function relevant to endometrial cancer. \nLINC00339 has been found to promote oncogenesis in several different cancer types (Gao et \nal. 2020; Ye et al. 2020; Zhao et al. 2020), although not specifically endometrial cancer. \nCDC42 encodes a small GTPase of the Rho-subfamily that regulates cell cycle, cell-cell \nadhesion, cell migration and cancer progression (Qadir et al. 2015). Notably, CDC42 binds to \nPAK6 (encoded by an endometrial cancer GWAS risk candidate target gene (O'Mara et al. \n2019)) and this complex, which localizes to cell-cell adhesions, is correlated with epithelial \ncolony escape (Morse et al. 2016). WNT4 encodes a protein that activates WNT/ β -catenin \nsignaling and appears to be crucial for the development of the female reproductive system, \nincluding the uterus (reviewed in (Biason-Lauber and Konrad 2008)). Moreover, genes \nbelonging to the WNT/ β -catenin pathway are frequently mutated in cancer, including the \ngene encoding β -catenin which is mutated in ~25% of endometrial tumors (Kandoth et al. \n2013). As with CDC42, there are also links between WNT4 and other genes located at \nendometrial cancer GWAS risk loci, such as WT1 and RSPO1 (O'Mara et al. 2018). For \n . CC-BY 4.0 International licenseIt is made available under a \n is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review)\nThe copyright holder for this preprint this version posted April 1, 2021. ; https://doi.org/10.1101/2020.11.09.20228114doi: medRxiv preprint \n\n16 \n \nexample, in the ovary there is evidence of WNT4 regulation by proteins encoded by both of \nthese genes (Biason-Lauber 2012; Gao et al. 2014) and RSPO protein activity potentiates \nWNT signaling (Bugter et al. 2021). There also appears to be some connection between \nCDC42 and WNT4: both genes have been found to be differentially expressed in the \nendometrium during the menstrual cycle (Powell et al. 2016).  \n \nTo reduce confounding inherent in the comorbidity observational studies of endometrial \ncancer and gynecological disease, prospective studies with long follow-up, large sample sizes \nand case identification using surgical confirmation would ideally be performed. Nevertheless, \nour study has demonstrated the utility of genetic causal inference analysis as a cost-effective \nalternative approach for unravelling relationships while reducing bias from unmeasured \nconfounding. However, a limitation of our study is that the sample size of PCOS GWAS (the \nlargest publicly available) was relatively small, reducing power to identify shared genetic risk \nregions or a causal relationship between PCOS and endometrial cancer. Consequently, these \nanalyses should be revisited when more genome-wide significant variants are revealed in \nfuture PCOS GWAS.  \n \nIn conclusion, our study has provided insights into the comorbidity of non-cancerous \ngynecological diseases and endometrial cancer by revealing shared genetic risk architecture, a \npotential causal relationship between uterine fibroids and endometrial cancer, and shared \ncandidate risk regions and genes. Furthermore, our study has leveraged this shared genetic \narchitecture to identify a novel risk locus for endometrial cancer, uncovering biologically \nrelevant candidate target genes and furthering our understanding of endometrial cancer \netiology.  \n \nAcknowledgements \nThis work was conducted using the UK Biobank Resource (application number 25331). We \nthank the research participants and employees of 23andMe for making this work possible. We \nthank the participants and investigators of FinnGen study. We thank the many individuals \nwho participated in the Endometrial Cancer Association Consortium and the International \nEndometriosis Genetics Consortium studies, and the numerous institutions and their staff who \nsupported recruitment. A full list of consortium members and acknowledgements can be \nfound in the Supplementary Note. \n . CC-BY 4.0 International licenseIt is made available under a \n is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review)\nThe copyright holder for this preprint this version posted April 1, 2021. ; https://doi.org/10.1101/2020.11.09.20228114doi: medRxiv preprint \n\n17 \n \n \nReferences \nAgarwal SK et al. (2019) Clinical diagnosis of endometriosis: a call to action Am J Obstet \nGynecol 220:354 e351-354 e312 doi:10.1016/j.ajog.2018.12.039 \nAlnafakh RAA, Adishesh M, Button L, Saretzki G, Hapangama DK (2019) Telomerase and \nTelomeres in Endometrial Cancer Front Oncol 9:344 doi:10.3389/fonc.2019.00344 \nAzziz R, Dumesic DA, Goodarzi MO (2011) Polycystic ovary syndrome: an ancient \ndisorder? Fertility and sterility 95:1544-1548 doi:10.1016/j.fertnstert.2010.09.032 \nBakshi A, Zhu Z, Vinkhuyzen AA, Hill WD, McRae AF, Visscher PM, Yang J (2016) Fast \nset-based association analysis using summary data from GWAS identifies novel gene \nloci for human complex traits Sci Rep 6:32894 doi:10.1038/srep32894 \nBiason-Lauber A (2012) WNT4, RSPO1, and FOXL2 in sex development Semin Reprod \nMed 30:387-395 doi:10.1055/s-0032-1324722 \nBiason-Lauber A, Konrad D (2008) WNT4 and sex development Sex Dev 2:210-218 \ndoi:10.1159/000152037 \nBonatz G, Frahm SO, Andreas S, Heidorn K, Jonat W, Parwaresch R (1998) Telomere \nshortening in uterine leiomyomas American journal of obstetrics and gynecology \n179:591-596 doi:10.1016/s0002-9378(98)70050-x \nBowden J, Davey Smith G, Burgess S (2015) Mendelian randomization with invalid \ninstruments: effect estimation and bias detection through Egger regression Int J \nEpidemiol 44:512-525 doi:10.1093/ije/dyv080 \nBowden J, Davey Smith G, Haycock PC, Burgess S (2016) Consistent Estimation in \nMendelian Randomization with Some Invalid Instruments Using a Weighted Median \nEstimator Genet Epidemiol 40:304-314 doi:10.1002/gepi.21965 \nBowden J, Hemani G, Davey Smith G (2018) Invited Commentary: Detecting Individual and \nGlobal Horizontal Pleiotropy in Mendelian Randomization-A Job for the Humble \nHeterogeneity Statistic? Am J Epidemiol 187:2681-2685 doi:10.1093/aje/kwy185 \nBugter JM, Fenderico N, Maurice MM (2021) Mutations and mechanisms of WNT pathway \ntumour suppressors in cancer Nat Rev Cancer 21:5-21 doi:10.1038/s41568-020-\n00307-z \nBulik-Sullivan B et al. (2015) An atlas of genetic correlations across human diseases and \ntraits Nat Genet 47:1236-1241 doi:10.1038/ng.3406 \nCarvajal-Carmona LG et al. (2015) Candidate locus analysis of the TERT-CLPTM1L cancer \nrisk region on chromosome 5p15 identifies multiple independent variants associated \nwith endometrial cancer risk Hum Genet 134:231-245 doi:10.1007/s00439-014-1515-\n4 \nChoi EJ, Cho SB, Lee SR, Lim YM, Jeong K, Moon HS, Chung H (2017) Comorbidity of \ngynecological and non-gynecological diseases with adenomyosis and endometriosis \nObstetrics & gynecology science 60:579-586 doi:10.5468/ogs.2017.60.6.579 \nConsortium GT (2013) The Genotype-Tissue Expression (GTEx) project Nat Genet 45:580-\n585 doi:10.1038/ng.2653 \nDay F et al. (2018) Large-scale genome-wide meta-analysis of polycystic ovary syndrome \nsuggests shared genetic architecture for different diagnosis criteria PLoS Genet \n14:e1007813 doi:10.1371/journal.pgen.1007813 \nDe La Cruz MS, Buchanan EM (2017) Uterine Fibroids: Diagnosis and Treatment Am Fam \nPhysician 95:100-107 \nFearnley EJ, Marquart L, Spurdle AB, Weinstein P, Webb PM, Australian Ovarian Cancer \nStudy G, Australian National Endometrial Cancer Study G (2010) Polycystic ovary \n . CC-BY 4.0 International licenseIt is made available under a \n is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review)\nThe copyright holder for this preprint this version posted April 1, 2021. ; https://doi.org/10.1101/2020.11.09.20228114doi: medRxiv preprint \n\n18 \n \nsyndrome increases the risk of endometrial cancer in women aged less than 50 years: \nan Australian case-control study Cancer causes & control : CCC 21:2303-2308 \ndoi:10.1007/s10552-010-9658-7 \nFortuny J et al. (2009) Risk of endometrial cancer in relation to medical conditions and \nmedication use Cancer epidemiology, biomarkers & prevention : a publication of the \nAmerican Association for Cancer Research, cosponsored by the American Society of \nPreventive Oncology 18:1448-1456 doi:10.1158/1055-9965.EPI-08-0936 \nGallagher CS et al. (2019) Genome-wide association and epidemiological analyses reveal \ncommon genetic origins between uterine leiomyomata and endometriosis Nat \nCommun 10:4857 doi:10.1038/s41467-019-12536-4 \nGao F, Zhang J, Wang X, Yang J, Chen D, Huff V, Liu YX (2014) Wt1 functions in ovarian \nfollicle development by regulating granulosa cell differentiation Hum Mol Genet \n23:333-341 doi:10.1093/hmg/ddt423 \nGao J, Dai C, Yu X, Yin XB, Zhou F (2020) Long noncoding RNA LINC00324 exerts \nprotumorigenic effects on liver cancer stem cells by upregulating fas ligand via PU \nbox binding protein FASEB J 34:5800-5817 doi:10.1096/fj.201902705RR \nGlubb DM et al. (2021) Cross-Cancer Genome-Wide Association Study of Endometrial \nCancer and Epithelial Ovarian Cancer Identifies Genetic Risk Regions Associated \nwith Risk of Both Cancers Cancer epidemiology, biomarkers & prevention : a \npublication of the American Association for Cancer Research, cosponsored by the \nAmerican Society of Preventive Oncology 30:217-228 doi:10.1158/1055-9965.EPI-\n20-0739 \nHarris HR, Terry KL (2016) Polycystic ovary syndrome and risk of endometrial, ovarian, and \nbreast cancer: a systematic review Fertil Res Pract 2:14 doi:10.1186/s40738-016-\n0029-2 \nHartwig FP, Davey Smith G, Bowden J (2017) Robust inference in summary data Mendelian \nrandomization via the zero modal pleiotropy assumption Int J Epidemiol 46:1985-\n1998 doi:10.1093/ije/dyx102 \nHemani G et al. (2018) The MR-Base platform supports systematic causal inference across \nthe human phenome Elife 7 doi:10.7554/eLife.34408 \nIlaria S, Marci R (2018) From obesity to uterine fibroids: an intricate network Curr Med Res \nOpin 34:1877-1879 doi:10.1080/03007995.2018.1505606 \nJohnatty SE et al. (2020) Co-existence of leiomyomas, adenomyosis and endometriosis in \nwomen with endometrial cancer Scientific reports 10:3621 doi:10.1038/s41598-020-\n59916-1 \nKandoth C et al. (2013) Integrated genomic characterization of endometrial carcinoma Nature \n497:67-73 doi:10.1038/nature12113 \nKar SP et al. (2020) Combining genome-wide studies of breast, prostate, ovarian and \nendometrial cancers maps cross-cancer susceptibility loci and identifies new genetic \nassociations bioRxiv \nKichaev G et al. (2019) Leveraging Polygenic Functional Enrichment to Improve GWAS \nPower Am J Hum Genet 104:65-75 doi:10.1016/j.ajhg.2018.11.008 \nKuchenbaecker KB et al. (2015) Identification of six new susceptibility loci for invasive \nepithelial ovarian cancer Nat Genet 47:164-171 doi:10.1038/ng.3185 \nLareau CA, Aryee MJ (2018) hichipper: a preprocessing pipeline for calling DNA loops from \nHiChIP data Nat Methods 15:155-156 doi:10.1038/nmeth.4583 \nLi J et al. (2019) Impact of endometriosis on risk of ovarian, endometrial and cervical \ncancers: a meta-analysis Arch Gynecol Obstet 299:35-46 doi:10.1007/s00404-018-\n4968-1 \n . CC-BY 4.0 International licenseIt is made available under a \n is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review)\nThe copyright holder for this preprint this version posted April 1, 2021. ; https://doi.org/10.1101/2020.11.09.20228114doi: medRxiv preprint \n\n19 \n \nMasuda T et al. (2020) GWAS of five gynecologic diseases and cross-trait analysis in \nJapanese Eur J Hum Genet 28:95-107 doi:10.1038/s41431-019-0495-1 \nMatalliotaki C et al. (2018) Co-existence of benign gynecological tumors with endometriosis \nin a group of 1,000 women Oncology letters 15:1529-1532 doi:10.3892/ol.2017.7449 \nMbatchou J et al. (2020) Computationally efficient whole genome regression for quantitative \nand binary traits bioRxiv \nMelin A, Sparen P, Bergqvist A (2007) The risk of cancer and the role of parity among \nwomen with endometriosis Hum Reprod 22:3021-3026 doi:10.1093/humrep/dem209 \nMorris JA et al. (2019) An atlas of genetic influences on osteoporosis in humans and mice \nNat Genet 51:258-266 doi:10.1038/s41588-018-0302-x \nMorse EM, Sun X, Olberding JR, Ha BH, Boggon TJ, Calderwood DA (2016) PAK6 targets \nto cell-cell adhesions through its N-terminus in a Cdc42-dependent manner to drive \nepithelial colony escape J Cell Sci 129:380-393 doi:10.1242/jcs.177493 \nMortlock S et al. (2020) Tissue specific regulation of transcription in endometrium and \nassociation with disease Hum Reprod 35:377-393 doi:10.1093/humrep/dez279 \nMunz M, Wohlers I, Simon E, Reinberger T, Busch H, Schaefer AS, Erdmann J (2020) \nQtlizer: comprehensive QTL annotation of GWAS results Sci Rep 10:20417 \ndoi:10.1038/s41598-020-75770-7 \nNagai K et al. (2015) Disease history and risk of comorbidity in women's life course: a \ncomprehensive analysis of the Japan Nurses' Health Study baseline survey BMJ open \n5:e006360 doi:10.1136/bmjopen-2014-006360 \nO'Mara TA et al. (2018) Identification of nine new susceptibility loci for endometrial cancer \nNat Commun 9:3166 doi:10.1038/s41467-018-05427-7 \nO'Mara TA, Spurdle AB, Glubb DM, Endometrial Cancer Association C (2019) Analysis of \nPromoter-Associated Chromatin Interactions Reveals Biologically Relevant \nCandidate Target Genes at Endometrial Cancer Risk Loci Cancers (Basel) 11 \ndoi:10.3390/cancers11101440 \nOlafsdottir T et al. (2020) Genome-wide association identifies seven loci for pelvic organ \nprolapse in Iceland and the UK Biobank Commun Biol 3:129 doi:10.1038/s42003-\n020-0857-9 \nOlson JE, Cerhan JR, Janney CA, Anderson KE, Vachon CM, Sellers TA (2002) \nPostmenopausal cancer risk after self-reported endometriosis diagnosis in the Iowa \nWomen's Health Study Cancer 94:1612-1618 \nPainter JN et al. (2018) Genetic overlap between endometriosis and endometrial cancer: \nevidence from cross-disease genetic correlation and GWAS meta-analyses Cancer \nmedicine doi:10.1002/cam4.1445 \nParasar P, Ozcan P, Terry KL (2017) Endometriosis: Epidemiology, Diagnosis and Clinical \nManagement Curr Obstet Gynecol Rep 6:34-41 doi:10.1007/s13669-017-0187-1 \nPowell JE et al. (2016) Endometriosis risk alleles at 1p36.12 act through inverse regulation of \nCDC42 and LINC00339 Hum Mol Genet 25:5046-5058 doi:10.1093/hmg/ddw320 \nPulit SL et al. (2019) Meta-analysis of genome-wide association studies for body fat \ndistribution in 694 649 individuals of European ancestry Hum Mol Genet 28:166-174 \ndoi:10.1093/hmg/ddy327 \nQadir MI, Parveen A, Ali M (2015) Cdc42: Role in Cancer Management Chem Biol Drug \nDes 86:432-439 doi:10.1111/cbdd.12556 \nRafnar T et al. (2018) Variants associating with uterine leiomyoma highlight genetic \nbackground shared by various cancers and hormone-related traits Nat Commun \n9:3636 doi:10.1038/s41467-018-05428-6 \n . CC-BY 4.0 International licenseIt is made available under a \n is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review)\nThe copyright holder for this preprint this version posted April 1, 2021. ; https://doi.org/10.1101/2020.11.09.20228114doi: medRxiv preprint \n\n20 \n \nRahmioglu N et al. (2018) Large-scale genome-wide association meta-analysis of \nendometriosis reveals 13 novel loci and genetically-associated comorbidity with other \npain conditions bioRxiv:406967 doi:10.1101/406967 \nRoberts SG (2005) Transcriptional regulation by WT1 in development Curr Opin Genet Dev \n15:542-547 doi:10.1016/j.gde.2005.08.004 \nRogalla P, Rohen C, Hennig Y, Deichert U, Bonk U, Bullerdiek J (1995) Telomere repeat \nfragment sizes do not limit the growth potential of uterine leiomyomas Biochem \nBiophys Res Commun 211:175-182 doi:10.1006/bbrc.1995.1793 \nRowlands IJ, Nagle CM, Spurdle AB, Webb PM, Australian National Endometrial Cancer \nStudy G, Australian Ovarian Cancer Study G (2011) Gynecological conditions and \nthe risk of endometrial cancer Gynecologic oncology 123:537-541 \ndoi:10.1016/j.ygyno.2011.08.022 \nRubtsova MP, Vasilkova DP, Malyavko AN, Naraikina YV, Zvereva MI, Dontsova OA \n(2012) Telomere lengthening and other functions of telomerase Acta Naturae 4:44-61 \nSam S (2007) Obesity and Polycystic Ovary Syndrome Obes Manag 3:69-73 \ndoi:10.1089/obe.2007.0019 \nSapkota Y et al. (2017) Meta-analysis identifies five novel loci associated with endometriosis \nhighlighting key genes involved in hormone metabolism Nat Commun 8:15539 \ndoi:10.1038/ncomms15539 \nStewart EA, Cookson CL, Gandolfo RA, Schulze-Rath R (2017) Epidemiology of uterine \nfibroids: a systematic review BJOG 124:1501-1512 doi:10.1111/1471-0528.14640 \nTelomeres Mendelian Randomization C et al. (2017) Association Between Telomere Length \nand Risk of Cancer and Non-Neoplastic Diseases: A Mendelian Randomization Study \nJAMA Oncol 3:636-651 doi:10.1001/jamaoncol.2016.5945 \nTurley P et al. (2018) Multi-trait analysis of genome-wide association summary statistics \nusing MTAG Nat Genet 50:229-237 doi:10.1038/s41588-017-0009-4 \nUimari O, Jarvela I, Ryynanen M (2011) Do symptomatic endometriosis and uterine fibroids \nappear together? Journal of human reproductive sciences 4:34-38 doi:10.4103/0974-\n1208.82358 \nVujkovic M et al. (2020) Discovery of 318 new risk loci for type 2 diabetes and related \nvascular outcomes among 1.4 million participants in a multi-ancestry meta-analysis \nNat Genet 52:680-691 doi:10.1038/s41588-020-0637-y \nWestra HJ et al. (2013) Systematic identification of trans eQTLs as putative drivers of known \ndisease associations Nat Genet 45:1238-1243 doi:10.1038/ng.2756 \nWise LA, Palmer JR, Stewart EA, Rosenberg L (2007) Polycystic ovary syndrome and risk \nof uterine leiomyomata Fertility and sterility 87:1108-1115 \ndoi:10.1016/j.fertnstert.2006.11.012 \nWise LA, Sponholtz TR, Rosenberg L, Adams-Campbell LL, Kuohung W, LaValley MP, \nPalmer JR (2016) History of uterine leiomyoma and risk of endometrial cancer in \nblack women Cancer causes & control : CCC 27:545-552 doi:10.1007/s10552-016-\n0728-3 \nYe H et al. (2020) The SP1-Induced Long Noncoding RNA, LINC00339, Promotes \nTumorigenesis in Colorectal Cancer via the miR-378a-3p/MED19 Axis Onco Targets \nTher 13:11711-11724 doi:10.2147/OTT.S277254 \nYengo L et al. (2018) Meta-analysis of genome-wide association studies for height and body \nmass index in approximately 700000 individuals of European ancestry Hum Mol \nGenet 27:3641-3649 doi:10.1093/hmg/ddy271 \nZhao H, Xiao H, Lu Y, Liu S, Wang C (2020) Long noncoding RNA LINC00339 promotes \nthe oncogenicity of gastric cancer by regulating SRY-box 9 expression via sponging \nof microRNA-539 Cell Cycle 19:1143-1157 doi:10.1080/15384101.2020.1749404 \n . CC-BY 4.0 International licenseIt is made available under a \n is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review)\nThe copyright holder for this preprint this version posted April 1, 2021. ; https://doi.org/10.1101/2020.11.09.20228114doi: medRxiv preprint \n\n21 \n \nZhu Z et al. (2018) Causal associations between risk factors and common diseases inferred \nfrom GWAS summary data Nat Commun 9:224 doi:10.1038/s41467-017-02317-2 \nZucchetto A et al. (2009) Hormone-related factors and gynecological conditions in relation to \nendometrial cancer risk Eur J Cancer Prev 18:316-321 \n . CC-BY 4.0 International licenseIt is made available under a \n is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review)\nThe copyright holder for this preprint this version posted April 1, 2021. ; https://doi.org/10.1101/2020.11.09.20228114doi: medRxiv preprint \n\n22 \n \nTables \nTable 1. Genetic correlation between non-cancerous gynecological diseases and endometrial cancer  \nNon-cancerous gynecological disease Covariate rG SE P-value \nEndometriosis - -0.02 0.09 0.83 \nPCOS - 0.36 0.12 1.6×10 -3 \nPCOS BMI 0.19 0.14 0.17 \nUterine fibroids - 0.24 0.09 5.4×10-3 \nUterine fibroids BMI 0.23 0.10 0.02 \n rG: genetic correlation, SE: standard error. Results passing Bonferroni correction (P < 0.017) are bolded \n . CC-BY 4.0 International licenseIt is made available under a \n is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review)\nThe copyright holder for this preprint this version posted April 1, 2021. ; https://doi.org/10.1101/2020.11.09.20228114doi: medRxiv preprint \n\n23 \n \nTable 2. Genetic causal inference results for effects of non-cancerous gynecological diseases on endometrial cancer  \nGynecological disease Genetic causal inference analysis Beta SE P-value \nEndometriosis IVW 0.09 0.09 0.34 \nMR-Egger 0.44 0.30 0.15 \nMR-Egger (intercept) -0.03 0.03 0.23 \nWeighted median 0.11 0.08 0.15 \nWeighted mode 0.13 0.11 0.26 \nPCOS IVW -0.05 0.04 0.26 \nMR-Egger 0.07 0.21 0.75 \nMR-Egger (intercept) -0.02 0.03 0.57 \nWeighted median -0.08 0.06 0.21 \nWeighted mode -0.19 0.12 0.13 \nUterine fibroids IVW 0.17 0.07 0.01 \nMR-Egger 0.15 0.15 0.32 \nMR-Egger (intercept) 0.00 0.01 0.91 \nWeighted median 0.07 0.07 0.35 \nWeighted mode 0.01 0.12 0.96 \n \n \n \n . CC-BY 4.0 International licenseIt is made available under a \n is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review)\nThe copyright holder for this preprint this version posted April 1, 2021. ; https://doi.org/10.1101/2020.11.09.20228114doi: medRxiv preprint \n\n24 \n \nTable 3. Shared candidate endometrial cancer and non-cancerous gynecological diseases risk regions \nRegion Gene TopSNPEC \n(p-value) \nTopSNPGyne \n(p-value) \nLD between \nTopSNPs (r2)* \nfastBATEC \np-value \nfastBATEC \nFDR \nfastBATGyne \np-value \nfastBATGyne  \nFDR \nEndometrial cancer and endometriosis \n3q21.3 RUVBL1 rs872267 \n(9.95×10-7) \nrs4857864 \n(5.35×10\n-6) \n0.47 8.49×10 -6 8.01×10 -3 8.26×10 -5 0.043 \n9p21.3 CDKN2B-AS1 rs568447 \n(2.68×10\n-6) \nrs6475610 \n(1.73×10\n-9) \n1.0×10-4 7.26×10 -5 0.036 9.02×10 -9 3.55×10 -5 \n15q15.1 BMF rs28371998 \n(6.81×10\n-9) \nrs7183386 \n(7.23×10\n-6) \n0.36 1.64×10 -7 5.74×10 -4 4.93×10 -6 9.35×10 -3 \n17q21.32 CBX1 rs7225865 \n(1.34×10\n-8) \nrs10445377 \n(3.20×10\n-7) \n0.99 3.35×10 -6 4.32×10 -3 2.63×10 -5 0.025 \n17q21.32 MIR1203 rs4794505 \n(9.50×10\n-9) \nrs10445377 \n(3.20×10\n-7) \n0.99 5.88×10 -7 1.44×10 -3 4.93×10 -6 9.35×10 -3 \n17q21.32 SKAP1 rs882380 \n(4.66×10\n-9) \nrs10445377 \n(3.20×10\n-7) \n0.96 5.80×10 -6 5.93×10 -3 7.93×10 -6 0.013 \n17q21.32 SNX11 rs17681336 \n(1.17×10\n-8) \nrs10445377 \n(3.20×10\n-7) \n0.99 1.55×10 -6 3.17×10 -3 4.51×10 -6 9.35×10 -3 \nEndometrial cancer and uterine fibroids \n5p15.33 CLPTM1L rs2736100 \n(5.17×10\n-6) \nrs72709458 \n(6.07×10\n-15) \n0.28 3.93×10 -7 1.20×10 -3 1.00×10 -9 1.53×10 -6 \n . CC-BY 4.0 International licenseIt is made available under a \n is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review)\nThe copyright holder for this preprint this version posted April 1, 2021. ; https://doi.org/10.1101/2020.11.09.20228114doi: medRxiv preprint \n\n25 \n \nRegion Gene TopSNPEC \n(p-value) \nTopSNPGyne \n(p-value) \nLD between \nTopSNPs (r2)* \nfastBATEC \np-value \nfastBATEC \nFDR \nfastBATGyne \np-value \nfastBATGyne  \nFDR \n5p15.33 MIR4457 rs2736100 \n(5.17×10-6) \nrs72709458 \n(6.07×10\n-15) \n0.28 1.50×10 -6 3.17×10 -3 8.01×10 -11 1.63×10 -7 \n5p15.33 TERT rs2736100 \n(5.17×10-6) \nrs72709458 \n(6.07×10\n-15) \n0.28 1.72×10 -6 3.25×10 -3 1.50×10 -11 4.06×10 -8 \n11p13 WT1 rs10835920 \n(1.33×10-8) \nrs11031731 \n(2.04×10\n-21) \n0.20 1.86×10 -5 0.015 2.64×10 -16 3.22×10 -12 \n11p13 WT1-AS rs10835920 \n(1.33×10\n-8) \nrs11031762 \n(9.95×10\n-14) \n0.40 1.91×10 -6 3.34×10 -3 1.04×10 -12 4.24×10 -9 \nTopSNP: lead variant for gene from fastBAT analysis; EC: endometrial cancer; Gyne: non-cancerous gynecological diseases; LD: li nkage disequilibrium; \nFDR: false discovery rate. \n*LD was estimated using the EUR 1000Genomes reference panel. \n . CC-BY 4.0 International licenseIt is made available under a \n is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review)\nThe copyright holder for this preprint this version posted April 1, 2021. ; https://doi.org/10.1101/2020.11.09.20228114doi: medRxiv preprint \n\n26 \n \nTable 4. Genome-wide significant endometrial cancer risk loci identified using MTAG \nRegion SNP EA OA EAF Original GWAS MTAG Replication \nBeta SE P-value Beta SE P-value Beta P-value \nKnown risk loci \n6p22.3 rs1740828 A G 0.49 -0.135 0.018 2.47×10 -14 -0.111 0.015 3.23×10 -13 - - \n8q24.21 rs10089519 # G A 0.67 0.093 0.017 3.89×10 -8 0.096 0.015 7.61×10 -11 - - \n8q24.21 rs72724795 # G T 0.12 0.150 0.024 3.40×10 -10 0.126 0.021 1.53×10 -9 - - \n11p13 rs10835917 C T 0.36 0.089 0.016 3.99×10 -8 0.084 0.014 2.09×10 -9 - - \n12q24.12 rs3184504 C T 0.52 0.099 0.016 3.59×10 -10 0.100 0.014 5.61×10 -13 - - \n13q22.1 rs7981863 T C 0.27 -0.154 0.018 9.78×10 -18 -0.103 0.016 4.28×10 -11 - - \n15q21.2 rs12595627 C T 0.67 0.120 0.017 1.35×10 -12 0.085 0.015 6.73×10 -9 - - \n17q12 rs11263761 A G 0.54 0.141 0.016 2.29×10 -18 0.095 0.014 1.49×10 -11 - - \nNovel risk loci* \n1p36.12 rs3820282 T C 0.17 0.078 0.021 2.66×10 -4 0.134 0.018 2.74×10 -13 0.12 4.36×10 -3 \n5p15.33 rs7713218 G A 0.52 -0.077 0.016 3.33×10 -6 -0.09 0.014 3.63×10 -10 0.02 0.62 \nEA: effect allele; OA: other allele; EAF: effect allele frequency; SE: standard error; Beta, Se and P-values were estimated fro m MTAG analysis of \nendometrial cancer, PCOS and uterine fibroids.  \n#LD between rs10089519 and rs72724795 is 0.02. \n*Located >1Mb from known endometrial cancer risk loci. \n \n \n . CC-BY 4.0 International licenseIt is made available under a \n is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review)\nThe copyright holder for this preprint this version posted April 1, 2021. ; https://doi.org/10.1101/2020.11.09.20228114doi: medRxiv preprint \n\n27 \n \nFigure legends \n \nFigure 1.  Association between genetic predisposition to non-cancerous gynecological \ndiseases and endometrial cancer, obtained from two-sample Mendelian randomization \nanalysis. The boxes represent the risk of endometrial cancer (beta) per standard deviation \nincrement in genetic predisposition to non-cancerous gynecological disease. Error bars \nrepresent 95% confidence intervals.  \n \n \n \n \n \n \n \n \n \n . CC-BY 4.0 International licenseIt is made available under a \n is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review)\nThe copyright holder for this preprint this version posted April 1, 2021. ; https://doi.org/10.1101/2020.11.09.20228114doi: medRxiv preprint \n\n \nFigure 2. Manhattan plot of MTAG result for endometrial cancer risk. Known endometrial \ncancer GWAS risk loci are marked in black, and novel genome-wide significant risk loci that \nare located >1Mb from known endometrial cancer risk loci in red. \n \n . CC-BY 4.0 International licenseIt is made available under a \n is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review)\nThe copyright holder for this preprint this version posted April 1, 2021. ; https://doi.org/10.1101/2020.11.09.20228114doi: medRxiv preprint \n\n \nFigure 3.  The upper panel depicts a regional association plot for the 1p36.12 novel \nendometrial cancer risk locus. Genetic variants at the locus are plotted by their genomic \nposition (hg19) and MTAG -log 10(P) for association with endometrial cancer risk is on the \nleft y-axis. Recombination rate (cM/Mb) is on the right y-axis and plotted as blue lines. The \ncolor of the circles indicates the level of linkage disequilibrium between each variant and the \nlead variant, rs3820282 (purple diamond), from the 1000 Genomes 2014 EUR reference \npanel (see legend, inset). The lower panel shows promoter-associated chromatin looping at \n1p36.12 identified from HiChIP analysis of the ARK-1 endometrial cancer cell line. \nPromoter-associated loops that intersect with candidate causal variants (shown as red vertical \nlines) are shown as purple arcs. \n \n \n \n \n \n . CC-BY 4.0 International licenseIt is made available under a \n is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review)\nThe copyright holder for this preprint this version posted April 1, 2021. ; https://doi.org/10.1101/2020.11.09.20228114doi: medRxiv preprint \n\n \nSupplementary Figure \nSupplementary Figure 1. Study design to explore the relationships between non- cancerous gynecological diseases and en\ncancer. Genetic analyses including genome-wide genetic analysis, gene-based analysis, Mendelian randomization analysis, multi- tra\nof GWAS and GWAS functional analysis were used in this study.  \n30 \n \nendometrial \ntrait analysis \n . CC-BY 4.0 International licenseIt is made available under a \n is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review)\nThe copyright holder for this preprint this version posted April 1, 2021. ; https://doi.org/10.1101/2020.11.09.20228114doi: medRxiv preprint \n\n31 \n \n \nSupplementary Figure 2 . Leave-one-out sensitivity analysis plot for association between \ngenetic predisposition to uterine fibroids and endometrial cancer. Each black dot in the forest \nplot represents the IVW estimates after excluding the corresponding variant. The plot \nhighlighted in red represents the IVW estimate for all variants.  \n . CC-BY 4.0 International licenseIt is made available under a \n is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review)\nThe copyright holder for this preprint this version posted April 1, 2021. ; https://doi.org/10.1101/2020.11.09.20228114doi: medRxiv preprint","source_license":"CC0","license_restricted":false}