{"paper_id":"0f8e16bb-9134-405d-85d0-745e6b3c0b72","body_text":"Endometriosis is a common condition in premenopausal women characterized by chronic\npelvic inflammation causing pain and subfertility ( 1 ), and has an estimated heritability of 51% ( 2 ). The International Endogene Consortium (IEC) performed the\nlargest endometriosis GWAS to date in 3194 surgically confirmed cases (including 1364\nmoderate–severe—Stage B—cases) and 7060 controls of European\nancestry, with replication in a further 2392 cases and 2271 controls ( 3 ). One genome-wide significant locus was\nobserved in an intergenic region on chromosome 7p15.2 (rs12700667), primarily associated\nwith Stage B disease ( P  = 1.5 × 10 −9 ,\nOR = 1.38, 95% CI 1.24–1.53). A second locus near\n WNT4  (rs7521902) was found after meta-analysis with published\nresults from a Japanese GWAS of 1423 cases and 1318 controls ( 4 ); a genome-wide meta-analysis confirmed the two loci and\nfound a further five ( 5 ).\nRs12700667 on 7p15.2 also marked 1 of 16 reported genome-wide significant loci\nassociated with waist-to-hip ratio adjusted for BMI (WHRadjBMI) in an independent GWAS\nmeta-analysis by the GIANT Consortium involving 77 167 individuals of European ancestry\nwith replication in a further 113 636 individuals (rs1055144: discovery\n P  = 1.5 × 10 −8 ; meta-analysis\n P  = 1.0 × 10 −24 ;\n r 2  = 0.5 with rs12700667 in 1000G pilot CEU data)\n( 6 , 7 ). This was surprising, as prospective epidemiological studies have\nsuggested consistently that reduced BMI—a measure of overall adiposity—is\nassociated with increased risk of endometriosis, but there is relatively limited\nevidence for an association with WHRadjBMI—a measure of fat distribution ( 8 , 9 ). We conducted a logistic regression analysis in the IEC dataset of rs1055144\non endometriosis disease status, conditioning on rs12700667, which demonstrated that the\nSNPs reflected the same association signal (unpublished data; conditional\n P  = 0.65).\nThe epidemiological evidence of an association between endometriosis and BMI, together\nwith the observed GWAS locus in common between endometriosis and WHRadjBMI, led us to\nconduct a systematic investigation of overlap in association signals between the IEC\nendometriosis GWAS and GIANT Consortium WHRadjBMI ( N  = 77 167)\n( 6 , 7 ) and BMI ( N  = 123 865) ( 7 , 10 ) meta-GWAS\ndatasets through genetic enrichment analyses.\n\nUsing independent, imputed (1000 Genomes pilot reference panel) GWAS datasets of\nendometriosis (IEC; 3194 cases including 1364 Stage B cases, 7060 controls), BMI\n(GIANT; 123 865 individuals) and WHRadjBMI (GIANT: 77 167 individuals), we first\nconsidered loci genome-wide significantly associated with endometriosis, BMI or\nWHRadjBMI in each of the individual GWAS. The two genome-wide significant\nendometriosis loci (intergenic 7p15.2 and  WNT4 ) had significantly\nlower  P -values of association than expected by chance in the\nWHRadjBMI GWAS (Table  1 :\nrs12700667,  P  = 4.4 × 10 −5  and\nrs7521902,  P  = 1.3 × 10 −3 ; binomial\n P  = 1.0 × 10 −4 ), while 2 of the\n16 genome-wide significant WHRadjBMI loci (intergenic 7p15.2 and\n GRB14 ) had  P  < 0.01 in the endometriosis\nGWAS (binomial  P  = 0.011). No enrichment between genome-wide\nsignificantly associated loci was observed for endometriosis versus BMI ( Supplementary Material, Table S1 : rs12700667,  P \n= 0.27 and rs7521902,  P  = 0.92).  Table 1. Association results of published IEC genome-wide significant endometriosis\nloci ( 3 ) in the GIANT WHRadjBMI\nGWAS, and of WHRadjBMI loci ( 6 , 7 ) in endometriosis\nGWAS (lookup results are shown in bold) GWAS SNP (proxy;\n r 2 ) Ch Location (B36) RAF (allele) Status Endometriosis all cases Endometriosis Stage B only Overall WHRadjBMI Female-limited WHRadjBMI Nearest gene P -value c OR (95% CI) P -value c OR (95% CI) P -value d Effect (SE) P -value e Effect (SE) Endometriosis rs12700667 7 25 868 164 0.74 (A) G 5.1 × 10 −7 1.21 (1.12–1.31) 3.3 × 10 −8 1.36 (1.23–1.50) 4.4 × 10 −5 − 0.023 (0.005) 3.3 × 10 −8 − 0.023 (0.005) Intergenic Endometriosis rs7521902 1 22 363 311 0.25 (A) G 8.9 × 10 −5 1.16 (1.08–1.25) 7.5 × 10 −5 1.26 (1.14–1.39) 1.3 × 10 −3 − 0.020 (0.006) 6.1 × 10 −3 − 0.023 (0.009) WNT4 WHRadjBMI rs1055144 a 7 25 837 634 0.19 (T) G 3.7 × 10 −5 0.84 (0.77–0.91) 3.1 × 10 −4 0.78 (0.70–0.88) 1.5 × 10 −8 0.034 (0.006) 2.3 × 10 −6 0.039 (0.008) Intergenic WHRadjBMI rs10195252 2 165 221 337 0.41 (C) G 9.8 × 10 −3 0.92 (0.85–0.98) 0.56 0.92 (0.84–1.00) 3.2 × 10 −10 −0.031 (0.005) 6.3 × 10 −15 −0.053 (0.007) GRB14 Female WHRadjBMI rs4684854 3 12 463 882 0.43 (C) I (0.98) 0.07 1.06 (0.99–1.14) 0.14 1.07 (0.98–1.17) 1.0 × 10 −4 0.019 (0.005) 2.3 × 10 −8 0.039 (0.007) PPARG WHRadjBMI rs718314 12 26 344 550 0.24 (G) G 0.11 1.06 (0.99–1.15) 0.054 1.10 (0.99–1.22) 2.4 × 10 −8 0.031 (0.005) 8.2 × 10 −10 0.047 (0.008) ITPR2-SSPN WHRadjBMI rs6861681 5 173 362 458 0.32 (A) I (0.96) 0.15 0.95 (0.86–1.04) 0.11 0.93 (0.85–1.00) 1.4 × 10 −6 0.026 (0.005) 2.1 × 10 −4 0.027 (0.007) CPEB4 WHRadjBMI rs6795735 3 64 680 405 0.41 (T) G 0.21 1.04 (0.98–1.12) 0.32 1.04 (0.96–1.14) 2.5 × 10 −7 −0.025 (0.005) 7.8 × 10 −7 −0.033 (0.007) ADAMTS9 WHRadjBMI rs2820446  (rs4846567,  r 2  =\n1) b 1 21 974 881 0.71 (C) I (0.99) 0.31 1.04 (0.97–1.12) 0.22 1.06 (0.97–1.17) 5.1 × 10 −12 0.037 (0.005) 8.5 × 10 −18 0.064 (0.007) LYPLAL1 WHRadjBMI rs498778  (rs6784615,  r 2  =\n1) b 3 52 453 893 0.93 (T) I (0.95) 0.32 1.08 (0.93–1.24) 0.25 1.06 (0.89–1.27) 4.6 × 10 −5 0.055 (0.010) 1.1 × 10 −3 0.063 (0.019) NISCH-STAB1 WHRadjBMI rs1294421 6 6 743 149 0.39 (T) I (0.96) 0.37 1.03 (0.94–1.10) 0.28 1.03 (0.94–1.13) 6.3 × 10 −9 −0.029 (0.005) 3.4 × 10 −8 −0.038 (0.007) LY86 WHRadjBMI rs9491696 6 127 452 639 0.51 (C) I (0.99) 0.43 0.97 (0.91–1.03) 0.64 0.98 (0.90–1.06) 2.1 × 10 −14 −0.037 (0.005) 3.4 × 10 −8 −0.038 (0.007) RSPO3 WHRadjBMI rs1443512 12 52 628 951 0.22 (A) G 0.62 1.02 (0.94–1.10) 0.63 0.97 (0.88–1.08) 3.3 × 10 −8 0.031 (0.005) 1.4 × 10 −9 0.048 (0.008) HOXC13 WHRadjBMI rs984222 1 119 305 366 0.39 (C) I (0.99) 0.69 0.99 (0.93–1.05) 0.31 0.95 (0.87–1.04) 3.8 × 10 −14 −0.037 (0.005) 1.2 × 10 −7 −0.036 (0.007) TBX15-WARS2 WHRadjBMI rs4823006 22 29 451 671 0.57 (A) I (0.97) 0.72 1.01 (0.95–1.08) 0.82 1.01 (0.92–1.11) 4.7 × 10 −10 0.030 (0.005) 6.9 × 10 −8 0.037 (0.007) ZNRF3 Female WHRadjBMI rs10478424 5 118 816 619 0.79 (A) I (0.97) 0.80 1.01 (0.93–1.10) 0.56 1.03 (0.93–1.15) 1.6 × 10 −4 0.023 (0.006) 1.0 × 10 −5 0.037 (0.009) HSD17B4 WHRadjBMI rs1011731 1 170 613 171 0.44 (G) G 0.81 0.99 (0.93–1.05) 0.77 1.01 (0.93–1.11) 1.7 × 10 −10 0.031 (0.005) 2.1 × 10 −5 0.028 (0.007) DNM3-PIGC WHRadjBMI rs6905288 6 43 866 851 0.56 (A) I (0.80) 0.66 0.98 (0.91–1.05) 0.66 0.99 (0.90–1.08) 4.2 × 10 −10 0.033 (0.005) 7.7 × 10 −13 0.052 (0.007) VEGFA a Logistic regression analysis in the IEC GWAS shows that rs1055144 marks\nthe same locus as rs12700667 (conditional  P  =\n0.65;  r 2  = 0.8). b SNP was not genotyped in the endometriosis GWAS dataset; result shown is\nof proxy SNP. c Results are based on an updated GWAS performed using genotype data\nimputed up to 1000 Genomes pilot reference panel (B36, June 2010). d Results are from the GIANT WHRadjBMI discovery GWAS dataset\n( N  = 77 167); 3 of the 14 WHRadjBMI loci have\n P  > 5.0 × 10 −8 ,\nhowever, they reached genome-wide significance combined with replication\nanalyses in up to a further 113 636 individuals ( 6 ). e Results from the GIANT WHRadjBMI discovery female-limited GWAS dataset\n( N  = 42 969); one of the two female-limited\nWHRadjBMI loci have  P  > 5.0 ×\n10 −8 , however, they reached genome-wide significance\ncombined with replication analyses in up to a further 71 295 individuals\n( 7 ).\nAssociation results of published IEC genome-wide significant endometriosis\nloci ( 3 ) in the GIANT WHRadjBMI\nGWAS, and of WHRadjBMI loci ( 6 , 7 ) in endometriosis\nGWAS (lookup results are shown in bold)\nLogistic regression analysis in the IEC GWAS shows that rs1055144 marks\nthe same locus as rs12700667 (conditional  P  =\n0.65;  r 2  = 0.8).\nSNP was not genotyped in the endometriosis GWAS dataset; result shown is\nof proxy SNP.\nResults are based on an updated GWAS performed using genotype data\nimputed up to 1000 Genomes pilot reference panel (B36, June 2010).\nResults are from the GIANT WHRadjBMI discovery GWAS dataset\n( N  = 77 167); 3 of the 14 WHRadjBMI loci have\n P  > 5.0 × 10 −8 ,\nhowever, they reached genome-wide significance combined with replication\nanalyses in up to a further 113 636 individuals ( 6 ).\nResults from the GIANT WHRadjBMI discovery female-limited GWAS dataset\n( N  = 42 969); one of the two female-limited\nWHRadjBMI loci have  P  > 5.0 ×\n10 −8 , however, they reached genome-wide significance\ncombined with replication analyses in up to a further 71 295 individuals\n( 7 ).\nTo investigate whether statistical enrichment extended beyond genome-wide significant\nloci, we investigated the most significant ( P  < 1 ×\n10 −3 ) independent ( r 2  < 0.2)\nendometriosis GWAS signals for enrichment of WHRadjBMI or BMI signals with\n P  < 0.05 and vice versa (number of lookup SNPs per\ndataset:  n  = 717 to 748; see  Supplementary Material, Methods ). We observed statistically\nsignificant enrichment between variants associated with endometriosis (particularly\nStage B) and WHRadjBMI (all endometriosis versus WHRadjBMI:  P \n= 3.7 × 10 −3 ; Stage B endometriosis versus WHRadjBMI:\n P  = 4.5 × 10 −4 ), but not between\nendometriosis and BMI (all endometriosis versus BMI:  P  =\n0.79; Stage B endometriosis versus BMI:  P  = 0.85)\n(Fig.  1 ;  Supplementary Material, Table S2 ). Results were similar when using\nfemale-limited WHRadjBMI ( N  = 42 969 women) and BMI\n( N  = 73 137 women) GWAS summary statistics ( 7 ); to optimize power, in the remainder of\nthe paper we therefore focus on sex-combined WHRadjBMI and BMI datasets ( Supplementary Material, Fig. S1 ). Empirical testing of statistical\nenrichment through permutation (see  Supplementary Material, Methods ) provided near-identical results\n(Fig.  1 ;  Supplementary Material, Fig. S1 ).  Figure 1. Genetic enrichment analyses between endometriosis, BMI and WHRadjBMI GWAS\ndatasets, using independent ( r 2  < 0.2)\nSNPs. The panels show (i) The proportion of SNPs nominally associated\n( P  < 0.05) with WHRadjBMI ( A ) or BMI\n( B ) by significance of overall and Stage B endometriosis\nassociation ( P  < 1.0 × 10 −3 \nversus  P  ≥ 1 × 10 −3 ); (ii)\nThe proportion of SNPs nominally associated ( P  <\n0.05) with overall and Stage B endometriosis by significance of WHRadjBMI\n(C) and BMI (D) association ( P  < 1.0 ×\n10 −3  versus  P  ≥ 1 ×\n10 −3 ).  P -values of\n χ 2  tests assessing statistical\ndifference between proportions are shown above each set of bars, and\n95% confidence intervals of the proportions are given on each bar.\nFor differences with  P chisq  < 0.2,\nempirical  P -values are given in brackets (see  Supplementary Material, Methods ).\nGenetic enrichment analyses between endometriosis, BMI and WHRadjBMI GWAS\ndatasets, using independent ( r 2  < 0.2)\nSNPs. The panels show (i) The proportion of SNPs nominally associated\n( P  < 0.05) with WHRadjBMI ( A ) or BMI\n( B ) by significance of overall and Stage B endometriosis\nassociation ( P  < 1.0 × 10 −3 \nversus  P  ≥ 1 × 10 −3 ); (ii)\nThe proportion of SNPs nominally associated ( P  <\n0.05) with overall and Stage B endometriosis by significance of WHRadjBMI\n(C) and BMI (D) association ( P  < 1.0 ×\n10 −3  versus  P  ≥ 1 ×\n10 −3 ).  P -values of\n χ 2  tests assessing statistical\ndifference between proportions are shown above each set of bars, and\n95% confidence intervals of the proportions are given on each bar.\nFor differences with  P chisq  < 0.2,\nempirical  P -values are given in brackets (see  Supplementary Material, Methods ).\nThe choice of significance thresholds in the discovery and lookup datasets was based\non a balance between applying a sufficiently stringent significance threshold in the\ndiscovery dataset that would minimize the proportion of false-positive association\nsignals, while still having sufficient numbers of loci in each of the phenotypic\nassociation strata to investigate statistical enrichment, and allow the pursuit of\nmeaningful biological pathway analyses subsequently. We considered the effect of\ndifferent significance thresholds, for both discovery and lookup, which confirmed\nresults showing enrichment of association signals between endometriosis and WHRadjBMI\n( Supplementary Material, Table S3 ), but no enrichment between\nendometriosis and BMI ( Supplementary Material, Table S4 ).\nTo investigate potential genome-wide sharing of loci between endometriosis and\nWHRadjBMI or BMI, we performed polygenic prediction analyses ( 11 ) evaluating whether the aggregate effect of many\nvariants of small effect in the WHRadjBMI and BMI GWAS could predict endometriosis\nstatus in the IEC GWAS (see  Supplementary Material, Methods ). There was no significant association\nbetween the WHRadjBMI- or BMI-derived profile scores (overall or female limited) and\nall/Stage B endometriosis ( Supplementary Material, Tables S5–S8 ), suggesting no evidence\nfor a directionally consistent  en masse , genome-wide, shared common\ngenetic component.\nWe next investigated the variants with most significant evidence for association with\nboth endometriosis ( P  < 1 × 10 −3 )\nand WHRadjBMI ( P  < 0.05) for predominance in direction of\nphenotypic effects ( Supplementary Material, Tables S9 and S10 \n and Fig. S2 ). No statistically significant directional consistency was\nobserved for these variants ( P  > 0.47), nor for the 17 loci\n(Table  1 ) that were\ngenome-wide significantly associated with either trait (Fig.  2 ,  P  > 0.44).\nIntergenic 7p15.2 and  WNT4  showed discordant directions of effect,\nwhile the effect of  GRB14  was concordant (Fig.  2 ). This could suggest the presence of\nmultiple biological pathways through which the variants influence the two phenotypes.\nWe next set out to investigate the common biology suggested by genetic variants\nassociated with both endometriosis and WHRadjBMI.  Figure 2. Directions of effect of 17 independent SNPs genome-wide significantly\nassociated with all ( A ) or Stage B ( B )\nendometriosis, or WHRadjBMI. Intergenic 7p15.2,  WNT4 , and\n GRB14  are shown in red. Linear regression\n R 2  and  P -values used to test\nfor significant directionality of effects ( 35 ) are shown.\nDirections of effect of 17 independent SNPs genome-wide significantly\nassociated with all ( A ) or Stage B ( B )\nendometriosis, or WHRadjBMI. Intergenic 7p15.2,  WNT4 , and\n GRB14  are shown in red. Linear regression\n R 2  and  P -values used to test\nfor significant directionality of effects ( 35 ) are shown.\nOur analysis showing significant enrichment between SNPs associated with all or Stage\nB endometriosis ( P  < 1 × 10 −3 ) and\nWHRadjBMI ( P  < 0.05) shown in Figure  1  involved 1284 independent\n( r 2  > 0.2) loci. We explored the biological\nfunction of the loci most strongly associated with WHRadjBMI, at nominal\n P  < 0.005 ( n  = 16,\nTable  2 ; see  Supplementary Material, Tables S11 and S12  for all variants associated\nat  P  < 0.05). Two novel loci, rs560584 near\n KIFAP3  (all endometriosis) and rs11619804 in\n CAB39L  (Stage B endometriosis), were significantly associated\nwith WHRadjBMI after Bonferroni correction allowing for 1284 independent tests\n( P  < 3.89 × 10 −5 ).  Table 2. Results of the top all/Stage B endometriosis loci ( P \n< 1 × 10 −3 ) associated with WHRadjBMI at\n P  < 0.005 SNP Chr Position (B36) RAF (allele) Endometriosis Overall WHRadjBMI Female-limited WHRadjBMI Nearest loci P -value OR (95% CI) P -value Effect SE P -value Effect SE (distance) All cases  rs560584 1 168 357 136 0.41 (T) 1.4 × 10 −4 1.14 (1.07–1.22) 1.4 × 10 −5 −0.021 0.005 1.1 × 10 −3 −0.022 0.677 KIFAP3  (46 632)  rs12700667 7 25 868 164 0.74 (A) 5.1 × 10 −7 1.22 (1.13–1.32) 4.4 × 10 −5 −0.023 0.005 3.4 × 10 −4 −0.028 0.284 NFE2L3  (2 90 221)  rs2921188 3 12 387 115 0.64 (A) 5.9 × 10 −4 1.13 (1.05–1.21) 1.1 × 10 −3 0.017 0.005 1.8 × 10 −4 0.026 0.054 PPARG  (0)  rs1250248 2 215 995 338 0.27 (A) 1.6 × 10 −5 1.17 (1.09–1.26) 1.0 × 10 −3 0.018 0.005 9.9 × 10 −4 0.025 0.242 FN1  (0)  rs2630787 3 21 847 339 0.52 (C) 9.2 × 10 −4 1.12 (1.05–1.19) 1.9 × 10 −3 −0.015 0.004 0.38 −0.006 0.030 ZNF659  (79 518)  rs1430788 2 67 721 916 0.31 (C) 9.3 × 10 −5 1.15 (1.07–1.23) 2.7 × 10 −3 0.016 0.005 3.1 × 10 −3 0.022 0.330 ETAA1  (230 878)  rs906721 3 184 687 691 0.41 (A) 6.1 × 10 −5 1.16 (1.08–1.24) 4.2 × 10 −3 0.015 0.005 1.7 × 10 −3 0.023 0.140 KLHL6  (322)  rs1868894 4 187 606 728 0.80 (C) 2.3 × 10 −4 1.16 (1.07–1.26) 4.9 × 10 −3 −0.018 0.006 0.13 −0.013 0.524 MTNR1A  (85 075)  rs3820282 1 22 340 802 0.16 (T) 3.3 × 10 −7 1.26 (1.15–1.37) 5.0 × 10 −3 −0.019 0.007 0.09 −0.016 0.749 WNT4  (0) Stage B cases  rs11619804 13 49 888 131 0.53 (C) 4.8 × 10 −4 1.17 (1.07–1.28) 1.1 × 10 −5 0.022 0.005 2.2 × 10 −2 0.016 0.022 CAB39L  (0)  rs12700667 7 25 868 164 0.74 (A) 3.3 × 10 −9 1.36 (1.23–1.50) 4.4 × 10 −5 −0.023 0.005 3.4 × 10 −4 −0.028 0.284 NFE2L3  (290 221)  rs2782659 6 45 794 768 0.33 (G) 4.2 × 10 −4 1.18 (1.08–1.30) 9.2 × 10 −5 0.020 0.005 1.7 × 10 −4 0.027 0.108 RUNX2  (167 970)  rs6556301 5 176 460 183 0.63 (G) 7.4 × 10 −4 1.17 (1.07–1.28) 1.9 × 10 −4 −0.021 0.005 7.8 × 10 −3 −0.021 0.845 FGFR4  (2450)  rs1250248 2 215 995 338 0.27 (A) 2.9 × 10 −8 1.32 (1.19–1.45) 1.2 × 10 −3 0.018 0.005 9.9 × 10 −4 0.025 0.242 FN1  (0)  rs4131816 1 161 662 648 0.85 (T) 5.4 × 10 −4 1.24 (1.10–1.41) 1.5 × 10 −3 0.022 0.007 0.25 0.011 0.072 NUF2  (70 470)  rs9912335 17 77 552 948 0.69 (T) 3.1 × 10 −4 1.19 (1.08–1.31) 3.5 × 10 −3 −0.021 0.007 0.10 −0.016 0.454 ASPSCR1  (0)  rs10878362 12 64 703 760 0.69 (C) 4.9 × 10 −4 1.19 (1.08–1.31) 3.6 × 10 −3 0.015 0.005 3.1 × 10 −3 0.022 0.204 HMGA2  (57 421)  rs2807357 1 22 364 571 0.64 (A) 9.7 × 10 −4 1.16 (1.06–1.27) 3.7 × 10 −3 −0.015 0.005 1.0 × 10 −3 −0.024 0.081 WNT4  (22 373)  rs906721 3 184 687 691 0.41 (A) 1.4 × 10 −4 1.20 (1.09–1.32) 4.2 × 10 −3 0.015 0.005 1.7 × 10 −3 0.023 0.140 KLHL6  (322)  rs12267660 10 4 419 530 0.85 (G) 7.9 × 10 −4 1.24 (1.09–1.40) 4.6 × 10 −3 0.02 0.007 8.0 × 10 −3 0.030 0.133 CR749391  (191 913)  rs11685481 2 67 590 253 0.15 (C) 8.4 × 10 −4 1.23 (1.09–1.38) 4.8 × 10 −3 0.018 0.006 1.1 × 10 −2 0.022 0.451 ETAA1  (99 215)\nResults of the top all/Stage B endometriosis loci ( P \n< 1 × 10 −3 ) associated with WHRadjBMI at\n P  < 0.005\nThe endometriosis risk allele T of rs560584 (OR = 1.14 (1.07–1.22),\n P  = 1.42 × 10 −4 ) was associated\nwith lower WHRadjBMI ( β  = −0.021,\n P  = 1.47 × 10 −5 ), and located in\nan intergenic region 46 kb downstream of  KIFAP3 \n( Kinesin-associated protein 3 ). Together with\n KIF3A  and  KIF3B ,  KIFAP3  forms a\nkinesin motor complex, KIF3, that mediates cellular transport of N-cadherin and\nβ-catenins ( 12 ), which are\ninvolved in cell adhesion, the  Wnt  canonical pathway and cell cycle\nprogression ( 13 ). The\n Wnt/β -catenin signalling pathway acts as a molecular\nswitch for adipogenesis ( 14 ) and has\nmultiple suggested roles in endometriosis through sex hormone homeostasis regulation\n( 15 ), its role in development of\nfemale reproductive organs ( 16 ),\nmolecular mechanisms of infertility ( 17 )\nand mediation of fibrogenesis ( 18 ).\nThe Stage B endometriosis risk allele C of rs11619804 (OR = 1.17\n(1.07–1.28);  P  = 4.88 ×\n10 −4 ), located in  CAB39L  (Calcium-Binding Protein\n39-Like), was associated with increased WHRadjBMI ( β  =\n0.022,  P  = 1.06 × 10 −5 ;\nTable  2 ). The function of\nthis gene is not well characterized but the encoded protein interacts with a serine\nthreonine kinase ( STK11 ) that functions as a tumour suppressor\n( 19 ).\nRs12700667 in the intergenic region 7p15.2 remained among the strongest associated\nshared signals, with the endometriosis risk allele A associated with reduced\nWHRadjBMI ( β  = −0.023,  P \n= 4.4 × 10 −5 ). The association maps to an intergenic\nhigh LD region of 48 kb ( r 2  > 0.8) of unknown\nfunctionality. Further interesting nearby loci include the miRNA\n hsa-mir-148a , with a purported role in\n Wnt/β -catenin signalling ( 14 );  NFE2L3  (nuclear factor\n(erythroid-derived 2)-like 3), a transcription factor suggested to be involved in\ncell differentiation, inflammation and carcinogenesis ( 20 ). The  WNT  signalling pathway was\nfurther highlighted by the nominal association of two independent\n( r 2  = 0.06) endometriosis risk variants near\n WNT4  (wingless-type MMTV integration site family), rs3820282\n(genic) and rs2807357 (22.4 kb downstream), with reduced WHRadjBMI\n( β  = −0.019,  P  = 5.0\n× 10 −3 ;  β  = −0.015,\n P  = 3.7 × 10 −3 ;\nTable  2 ). Of note is that all\nshared variants implicated in  WNT  signalling (in/near intergenic\n7p15.2,  WNT4 ,  KIFAP3 ) showed\nconsistent—discordant—phenotypic directions of effect.\nRisk variant rs10195252, 34.6 kb downstream of  GRB14  (growth factor\nreceptor-bound protein 14) was the third locus with significant evidence for\nassociation with both overall (not Stage B) endometriosis and WHRadjBMI\n(Table  1 ).\n GRB14  has an inhibitory effect on insulin receptor signalling\n( 21 ), may have a role in signalling\npathways that regulate growth and metabolism and has been shown to interact with\nfibroblast growth factor receptors ( 22 ).\nThis shared variant is also in high LD ( r 2  = 0.93\nand = 0.87, respectively) with a type 2 diabetes risk variant rs13389219\n( 23 ) and fasting insulin risk variant\nrs6717858 ( 24 ).\nOther loci of interest include rs2921188 in  PPARG  and rs6556301 near\n FGFR4  (Table  2 ) .  The endometriosis risk allele A of rs2921188 (OR\n= 1.13, 95% CI: 1.05–1.21),  P  = 5.9\n× 10 −4 ) in  PPARG  (peroxisome\nproliferator-activated receptor gamma) is associated with increased WHRadjBMI\n( β  = 0.017;  P  = 1.1\n× 10 −3 ).  PPARG  is a nuclear hormone\nreceptor that regulates fatty acid storage and glucose metabolism. Synthetic ligands,\nsuch as insulin sensitizing drugs, target  PPARG  in treatment of\ndiabetes to lower serum glucose levels ( 25 ) and are also documented to have anti-inflammatory, anti-angiogenic and\nanti-proliferative effects on endometrium, with baboon models suggesting a role in\ntargeting endometriotic disease ( 26 ).\nStage B endometriosis risk allele G of rs6556301 near  FGFR4 \n( fibroblast growth factor receptor , OR = 1.17\n[1.07–1.28],  P  = 7.4 × 10 −4 )\nis associated with reduced WHRadjBMI ( β  =\n−0.021,  P  = 1.9 × 10 −4 ).\n FGFR4  interacts with fibroblast growth factors, which have roles\nin angiogenesis, wound healing and cell migration ( 27 ).\nWe investigated the potential impact of the described 16 genes (Table  2 ) shared between endometriosis and\nWHRadjBMI on transcriptional function using three public expression data resources:\n(i) the Mammalian Gene Expression Uterus database (MGEx-Udb) ( 28 ) containing published information on transcriptional\nactivity of specific genes in human endometrial tissue from individuals with and\nwithout endometriosis; (ii) the MuTHER study which collected expression and eQTL data\nfrom 776 abdominal fat tissues ( 29 ); and\n(iii) the MOLOBB dataset of differential expression levels between abdominal and\ngluteal fat from 49 individuals ( 30 ).\nBased on the limited available evidence in the MGEx-Udb database, two genes are\ntranscribed in endometrial tissue of women with endometriosis but dormant in those\nwithout endometriosis:  PPARG  and  FGFR4  ( Supplementary Material, Table S13 ). Of the 16 genes, 15 had probes\npresent within 1 Mb either side of the SNP in the MuTHER database; however, none\nshowed significant association with nearby transcripts in abdominal fat tissue\n( Supplementary Material, Table S14 ). The MOLOBB study data showed\n cis -eQTL evidence for differential expression of two genes;\n KIFAP3  (rs560584; fold change = 0.14, adjusted\n P  = 0.04) ( Supplementary Material, Table S15 ). Additional transcriptional\nevidence relevant to the intergenic 7p15.2 locus includes the presence of an\nexpression QTL associated with a transcript of unknown function,\n AA553656 , in subcutaneous abdominal fat tissue ( 6 ), and the differential expression of\nnearby  hsa-miR-148a  between gluteal and abdominal fat tissue samples\n( 31 ).\nTo identify potential common biological pathways involved in the aetiology of\nendometriosis and the variability of fat distribution, we conducted pathway analyses\nusing genes with evidence for enrichment between the traits using (i) the PANTHER\ndatabase ( 32 ) and (ii) GRAIL ( 33 ). For the PANTHER analysis, we selected\nthe 91 and 108 genes located in a 1 Mb interval surrounding each independent SNP\nassociated with all endometriosis ( P  < 1.0 ×\n10 −3 ) and WHRadjBMI ( P  < 0.05), and\nStage B endometriosis ( P  < 1.0 ×\n10 −3 ) and WHRadjBMI ( P  < 0.05),\nrespectively (see  Supplementary Material, Methods ). This excluded intergenic loci\nwithout a gene within 1 Mb, such as our top shared locus at 7p15.2. We tested whether\nthe two sets of genes showed significant overrepresentation of a particular pathway,\nfor each of 176 curated pathways and 241 biological processes. The top enriched\npathways were ‘developmental processes’ (all endometriosis:\n P  = 1.2 × 10 −5 ; Stage B:\n P  = 1.25 × 10 −4 ),\n‘ WNT  signalling’ (all endometriosis:\n P  = 1.07 × 10 −4 ),\n‘gonadotropin-releasing hormone receptor’ (all endometriosis:\n P  = 1.48 × 10 −3 ),\n‘cadherin signalling’ (Stage B:  P  = 6.42\n× 10 −4 ), ‘FGF signalling’ (Stage B:\n P  = 2.96 × 10 −3 ) and\n‘TGF-beta signalling’ (Stage B:  P  = 1.48\n× 10 −3 ) pathways ( Supplementary Material, Tables S16 and S17 ). Bonferroni correction for\nthe number of pathways tested (see  Supplementary Material, Methods ) rendered ‘ WNT \nsignalling’, ‘developmental processes’, ‘cellular\nprocesses’ and ‘cell communication’ significantly enriched;\nhowever, this adjustment is conservative, as exemplified by ‘cadherin\nsignalling’ genes being a subset of those in the ‘ WNT \nsignalling’ pathway. Sensitivity analyses exploring the effect of different\nendometriosis association thresholds on pathway analyses showed very consistent\nresults for threshold  P  < 1.0 ×\n10 −2 , with the same top three enriched\npathways— WNT  signalling, Cadherin signalling and\nGonadotropin-releasing hormone receptor pathway. No meaningful pathway analyses could\nbe conducted on the limited number of genes passing association threshold\n P  < 1 × 10 −4  ( Supplementary Material, Table S18 ).\nWe used GRAIL ( 33 ) to search for\nconnectivity between the 91 and 108 genes all/Stage B endometriosis and\nWHRadjBMI-associated genes and specific keywords from the published literature that\ndescribe potential functional connections. We identified 17 genes with nominal\nsignificance ( P  < 0.05) for potential functional connectivity\nfor ‘all’ endometriosis and WHRadjBMI and six genes for Stage B\nendometriosis and WHRadjBMI ( Supplementary Material, Fig. S3 \n  and Tables S19 and S20 ). The keywords associated with these\nconnections included ‘cadherin’, ‘differentiation’,\n‘development’ and ‘insulin’ for ‘all’ endo,\nand ‘development’ and ‘embryos’ for Stage B\nendometriosis, marking again developmental processes and cadherin signalling as\nbiological pathways shared in the origins of endometriosis and fat distribution.\n\nIn this study, we have investigated the overlap in genetic association signals from the\nlargest GWA studies to date of endometriosis, overall adiposity (BMI) and fat\ndistribution (WHRadjBMI). Our results demonstrated that there is a shared genetic basis\nbetween endometriosis and fat distribution that extends over and above the single\ngenome-wide significant locus that has been reported in GWAS of the separate traits. Our\nanalyses highlight novel loci in/near  KIFAP3  and\n CAB39L , which together with intergenic 7p15.2,  WNT4 \nand  GRB14 , showed significant evidence of trait association sharing.\nThe strength of evidence of enrichment was similar for overall versus female-limited\nWHRadjBMI loci, which may be unexpected, given that endometriosis is a female condition.\nHowever, the lack of a stronger enrichment between female-specific WHRadjBMI GWAS\nresults and endometriosis, compared with all WHRadjBMI results should be considered\nagainst the effects of a reduced sample size used for female-specific WHRadjBMI analyses\non power of association detection.\nThe enrichment of associated variants was generally stronger when the endometriosis\ncases were restricted to moderate–severe (Stage B) disease, despite the smaller\nsample size. Indeed, the association of the top intergenic GWAS locus on 7p15.2, also\ngenome-wide significantly associated with WHRadjBMI, is limited to Stage B\nendometriosis. Stage B—or ASRM Stages III/IV disease ( 34 )—is typically characterized by ovarian\n(endometrioma) or deep infiltrating (rectovaginal) lesions, which were shown to have a\nsubstantially greater underlying genetic contribution than milder, peritoneal disease\n(ASRM Stage I/II) ( 3 ). The particular\nenrichment between WHRadjBMI and Stages III/IV endometriosis is intriguing, and another\nreason for further functional work to concentrate on this endometriosis sub-type. There\nare, however, specific loci that show enrichment of association with WHRadjBMI and\noverall endometriosis, the analysis of which therefore remains of interest. An example\nis  GRB14 , which did not show significant association with Stage B\ndisease, displayed a concordant direction of effect between endometriosis and WHRadjBMI,\nand the biological function of which also seems to suggest an entirely different\ncontribution to the origins of both phenotypes than the 7p15.2 and  WNT4 \nloci.\nThe limited available eQTL data showed significant evidence for differential expression\nof  KIFAP3  between different fat depots. The variants with most evidence\nfor enrichment between the traits, in/near intergenic 7p15.2,  KIFAP3 \nand  WNT4 , were all implicated in  WNT  signalling and\nhad consistent—discordant—directions of effect, with endometriosis risk\nalleles associated with a decreased WHRadjBMI. Indeed, biological pathway analyses\nshowed significant evidence for the involvement of developmental processes and\n WNT  signalling in endometriosis aetiology and regulation of fat\ndistribution, a potential pleiotropic connection that has not been reported to date.\nThe relatively limited epidemiological evidence of phenotypic correlation between\nendometriosis and WHRadjBMI ( 8 , 9 ) is consistent with the absence of strong\ndirectional consistency of phenotypic effects of genetic variants underlying both traits\nat a genome-wide level. Most studies of genetic pleiotropy between traits to date have\nfocused on genome-wide directional consistency between epidemiologically or clinically\n(postulated) correlated traits, such as different metabolic traits ( 6 , 35 ) or psychiatric conditions ( 36 ). However, genome-wide consistency in directionality of phenotypic effects\nwould most likely apply to traits that share a large proportion of causality, and that\nepidemiologically lie on the same causal pathway(s) and are thus more likely to be\nexamples of mediated (genetic variants influencing one phenotype indirectly through\nassociation with a second phenotype) rather than biological (genetic variants exerting a\ndirect biological influence on more than one phenotype) pleiotropy ( 37 ). Thus, our results of genetic enrichment\nbetween endometriosis and WHRadjBMI demonstrate an example of the biological complexity\nof aetiological associations between complex traits, and suggest that the underlying\nshared loci are potentially biologically pleiotropic, given the absence of phenotypic\ncorrelation between endometriosis and WHRadjBMI and absence of  en masse \ndirectional consistency of shared genetic variants on the phenotypes ( 37 , 38 ). It also demonstrates more generally how potential perturbation of a\ncausal pathway through, for example, drug treatment targeting one trait could have\nunexpected effects on another, even when there is no clear evidence that these traits\nare associated clinically or epidemiologically—a problem often encountered in\ndrug development. Systematic exploration of biological pleiotropy of genetic variants\nmarking potential drug targets may help in highlighting the potential of such unwanted\nor unexpected effects.\nWhile the observed genetic enrichment between endometriosis and WHRadjBMI presents new\navenues for exploring common biology, the total absence of any genetic enrichment\nbetween endometriosis and BMI (within the limits of power presented by these large\ndatasets) is intriguing given the consistent, prospective, observational epidemiological\nevidence of phenotypic association between reduced BMI and endometriosis risk ( 8 ). Our analyses represent an adaptation of\nMendelian randomization analyses ( 39 , 40 ), in which genetic variants robustly\nassociated with BMI in the largest GWAS analyses to date ( 10 ) are investigated for association with endometriosis. The\ntotal lack of genetic enrichment suggests that reduced BMI is not causally related to\nendometriosis risk. Rather, it suggests that the observed phenotypic association ( 8 ) is either driven by shared environmental\nfactors, or is due to confounding factors related to BMI affecting, for example,\ndiagnostic opportunity for endometriosis.\nThese novel findings present an entirely new opportunity for functional targeted\nfollow-up of pleiotropic loci between endometriosis and WHRadjBMI in relevant disease\ntissues such as endometrium and fat tissue, cellular systems, animal models and further\ncross-trait comparisons, to uncover their biological functions and to assess how studies\nin the fat distribution research field can inform research into endometriosis\npathogenesis, biomarker identification and drug target discovery and validation.\n\nThis GWAS included 3194 surgically confirmed endometriosis cases and 7060 controls\nfrom Australia and the UK. Disease severity of the endometriosis cases was\nassessed retrospectively from surgical records using the rAFS classification\nsystem and grouped into two phenotypes: Stage A (Stage I or II disease or some\novarian disease with a few adhesions;  n  = 1686) or Stage B\n(Stage III or IV disease;  n  = 1364). We previously showed\nan increased genetic loading among 1364 cases with Stage B endometriosis compared\nwith 1666 with Stage A disease ( 3 ),\nwhich led to two GWA analyses, including (i) 3194 ‘all’\nendometriosis case and (ii) 1364 Stage B cases (Table  3 ). The genotyped data were imputed up\nto 1000 Genomes pilot reference panel (B36, June 2010) and the GWAS was performed\nagain, using a missing data likelihood in a logistic regression model including a\ncovariate representing the Australian and the UK strata, with the imputed data\n( N  = 12.5 million SNPs). The enrichment analysis we\npresent is from this set of results.  Table 3. Summary description of the GWAS used in the genetic enrichment\nanalysis GWAS Consortium Sample size No. of SNPs (million) References Endometriosis—all cases IEC 3194 cases, 7060 controls ∼12.5 Painter  et al . ( 3 ) Endometriosis—Stage B cases IEC 1363 cases, 7060 controls ∼12.5 Painter  et al . ( 3 ) WHRadjBMI GIANT 77 167 ∼2.85 Heid  et al . ( 6 ) Female-limited WHRadjBMI GIANT 42 969 ∼2.85 Randall  et al . ( 7 ) BMI GIANT 123 865 ∼2.85 Speliotes  et al . ( 10 ) Female-limited BMI GIANT 73 137 ∼2.85 Randall  et al . ( 7 ) IEC, International Endogene Consortium; GIANT, Genetic Investigation\nof Anthropometric Traits Consortium; BMI, body mass index adjusted for\nage; WHRadjBMI, waist to hip ratio adjusted for BMI and age.\nSummary description of the GWAS used in the genetic enrichment\nanalysis\nIEC, International Endogene Consortium; GIANT, Genetic Investigation\nof Anthropometric Traits Consortium; BMI, body mass index adjusted for\nage; WHRadjBMI, waist to hip ratio adjusted for BMI and age.\nA total of 77 167 subjects of European ancestry informative of body fat\ndistribution measurement WHR from 32 GWAS were included ( 6 ). The genotype data were imputed up to HapMap 2 CEU\nreference panel. The associations of 2.85 million SNPs with WHR were examined\nin a fixed-effects meta-analysis, after inverse normal transformation of WHR\nand adjusting for BMI and age within each study in an additive genetic model;\nanalyses were conducted for males and females combined ( 6 ) and limited to females only ( 7 ) (Table  3 ).\nA total of 123 865 subjects with overall adiposity measurement BMI from 46 GWAS\nwere included ( 10 ). The genotype\ndata were imputed up to HapMap two CEU reference panels. The associations of\n2.85 million SNPs with BMI were tested in an inverse-variance meta-analysis,\nafter inverse normally transformation of BMI and adjusting for age and other\nappropriate covariates in an additive genetic model within each study; analyses\nwere conducted for males and females combined ( 10 ) and limited to females only ( 7 ) (Table  3 ).\nWith one test of association conducted for each SNP, the GWAS analyses produced a\ngenome-wide distribution of  P -values of individual SNP associations.\nPrior to testing enrichment: (i) the overlap of SNPs present in endometriosis GWAS\nversus WHRadjBMI and BMI GWAS was taken, (ii) all SNPs with MAF ≤ 0.01 were\nremoved, (iii) all SNPs with A/T and C/G base pairs were removed, (iv) correlated\nSNPs ( r 2  > 0.2) were removed as previously\nreported ( 41 ) by taking the most\nsignificantly associated SNP and eliminating all SNPs that have a HapMap CEU pairwise\ncorrelation coefficient ( r 2 ) > 0.2 with that SNP,\nthen processing to the next strongly associated SNP remaining. This resulted in 173\n157 independent SNPs in endometriosis versus WHRadjBMI and 173 223 in endometriosis\nversus BMI enrichment analyses.\nThe independent SNPs in the tails ( P  < 1 ×\n10 −3 ) of the association results distribution of the two\nendometriosis GWAS (all endometriosis and ‘Stage B’ cases) were\ninvestigated for enrichment of WHRadjBMI or BMI low  P -value\n( P  < 0.05) association signals; in reversal, SNPs in the\ntails of WHRadjBMI and BMI GWAS ( P  < 1 ×\n10 −3 ) were investigated for evidence of nominal association\n( P  < 0.05) in the two endometriosis GWAS. The threshold of\n P  < 1 × 10 −3  corresponded to the\npoint at which endometriosis GWAS results started to deviate from the null\ndistribution (evidence for association) in the overall and Stage B endometriosis\nQ–Q plots ( Supplementary Material, Fig. S4 ). Enrichment was assessed in R by\nmeans of Pearson's  χ 2  tests with\nYates' continuity correction, testing for the difference in proportion of SNPs\nwith association  P  < 0.05 in the lookup dataset according to\nassociation in the discovery dataset ( P  < 1 ×\n10 −3  versus  P  ≥ 1 ×\n10 −3 ). To test for consistency in directionality of phenotypic\neffects of the SNPs with evidence of enrichment, linear regression analysis was\nperformed on the effect ( β ) of each SNP for WHRadjBMI as\npredictor variable and the effect ( β ) of endometriosis risk\nas the outcome variable ( 35 ). In\naddition, a two-sided binomial test was performed with null hypothesis\n P  = 0.50.\nFor those results that showed nominally significant ( P  <\n0.10) evidence for enrichment in  χ 2  tests of\ncontingency tables, we performed permutation-based analyses to obtain empirical\nestimates of significance of enrichment. We (i) randomly picked the same number of\nindependent SNPs ‘associated’ with the discovery trait at\n P  < 1 × 10 −3  (e.g. the number of\nSNPs associated with all endometriosis at  P  < 1 ×\n10 −3  was  n  = 717) from the WHRadjBMI\ndataset; (ii) counted how many of the randomly selected SNPs had\n P -values of association with WHRadjBMI <0.05; (iii) repeated\nSteps (i) and (ii) 10 000 times; (iv) determined the number of instances among the 10\n000 draws in which the number of SNPs associated at  P  < 0.05\nwith WHRadjBMI was greater or equal to the number we observed in our original\nanalysis (e.g. ≥52/717). For example, for overall endometriosis and overall\nWHRadjBMI, we observed this in 26/10 000 instances, corresponding to a\n P -value of 2.6 × 10 −3 , which was very\nsimilar to the  P -value obtained from the\n χ 2  test ( P  = 3.7\n× 10 −3 ).\nThe independent SNPs in both WHRadjBMI and endometriosis datasets were used to\nconduct a polygenic prediction analysis ( 11 ). The aim of this analysis was to evaluate the aggregate effects of\nmany SNPs of small effect and assess whether subsets of SNPs selected in this manner\nfrom one disease/trait GWAS predict disease/trait status in another, thus providing a\nmeasure of a common polygenic component with concordant directions of effect\nunderlying the traits. Briefly, subsets of SNPs were selected from the WHRadjBMI GWAS\ndata based on their association with WHRadjBMI using increasingly liberal thresholds,\nthat is,  P  < 0.01,  P  < 0.05,\n P  < 0.1,  P  < 0.2,\n P  < 0.3,  P  < 0.4,\n P  < 0.5 and  P  < 0.75. Using these\nthresholds, we defined sets of allele-specific scores in the WHRadjBMI dataset to\ngenerate risk profile scores for individuals in the endometriosis dataset. For each\nindividual in the endometriosis dataset, we calculated the number of score alleles\nthey possessed, each weighted by their effect size ( β -value)\nof association in the WHRadjBMI dataset. To assess whether the aggregate scores were\nassociated with endometriosis risk, we tested for a higher mean score in cases\ncompared with controls. Logistic regression was used to assess the relationship\nbetween endometriosis disease status and aggregate risk score.\nThe mammalian gene expression uterus database (MGEx-Udb) is a manually curated\nuterus-specific database created using a meta-analysis approach from published\npapers ( 28 ) that provides lists of\ntranscribed and dormant genes for various normal, pathological (e.g.\nendometriosis, cervical cancer and endometrial cancer) and experimental (e.g.\ntreatment and knockout) conditions. Each gene's expression status is\nindicated by a reliability score, derived based on the consensus across multiple\nsamples and studies which highly variable ( http://resource.ibab.ac.in/MGEx-Udb/ ).\nThe MuTHER resource includes LCLs, skin and adipose tissue-derived simultaneously\nfrom a subset of well-phenotyped healthy female twins ( 29 ). Whole-genome expression profiling of the samples,\neach with either two or three technical replicates, was performed using the\nIllumina Human HT-12 V3 BeadChips (Illumina, Inc.) according to the protocol\nsupplied by the manufacturer. Log2 transformed expression signals were normalized\nseparately per tissue as follows: quantile normalization was performed across\ntechnical replicates of each individual followed by quantile normalization across\nall individuals.\nGenotyping was conducted using a combination of Illumina arrays (HumanHap300,\nHumanHap610Q, 1M-Duo and 1.2MDuo 1 M). Untyped HapMap2 SNPs were imputed using the\nIMPUTE software package (v2). In total, there were 776 samples with genotypes and\nexpression values in adipose tissue. Association between all SNPs (MAF >\n5%, IMPUTE info score > 0.8) within a gene or within 1 Mb of the\ngene transcription start or end site, and normalized expression values, were\nperformed with the GenABEL/ProbABEL packages ( 42 ) using polygenic linear models incorporating a\nkinship matrix (GenABEL) followed by the mm score test with imputed genotypes\n(ProbABEL). Age and experimental batch were included as cofactors in the analysis.\nBenjamini Hochberg corrected  P -values are reported.\nWe performed differential  cis -eQTL analysis to compare the\nexpression levels in gluteal and abdominal fat tissue from 49 individuals in the\nMolOBB dataset (24 with and 25 without metabolic syndrome—MetSyn) ( 30 ). We first checked for the presence\nof the SNP in the MolOBB genotype data and, in the case of absence, selected any\nproxies ( r 2  > 0.8) available. We then searched\nfor nearby genes (±500 kb) covered by the expression data using the\nbioconductor R package GenomicRanges ( 43 ) and tested for association at each pair using a linear model with\nthe expression level as an outcome and the SNP allelic dosage as a predictor,\nadjusting for age, gender and MetSyn case–control status. This analysis was\ncarried out for both abdominal and gluteal subcutaneous adipose tissue. To\ninvestigate whether genes were differentially expressed between the two tissues,\nwe applied a linear mixed model with tissue, MetSyn case–control status,\ngender and plate were as fixed effects, and subject as a random effect using\nMAANOVA ( 44 ), as previously described\nin Min  et al . ( 30 ).\nWe report the uncorrected and genome-wide FDR corrected  F s test\n P -values ( 30 ).\nWe conducted analyses using the PANTHER 8.1 database containing pathway\ninformation on 20 000 genes ( Homo sapiens ) ( 32 ). We selected independent SNPs, which had association\n P -values < 1 × 10 −3  in the\nendometriosis datasets and an association  P -value of <0.05\nin the WHRadjBMI dataset, resulting in (i) 91 SNPs for all endometriosis and\nWHRadjBMI and (ii) 108 SNPs for Stage B endometriosis and WHRadjBMI. Each SNP was\nmapped to the closest gene within 1 Mb; 88 of 91 and 103 of 108 genes were present\nin the PANTHER database, and these subsets were tested for correlation with 241\nbiological processes and 176 pathways classified in the database ( 32 ). For each biological\nprocess/pathway, the difference between the observed fraction of genes in that\npathway and the number expected by chance was tested using Fisher exact test. A\nBonferroni correction was used as a conservative method for adjusting for the\nmaximum number of biological processes ( n  = 278;\n P  = 1.80 × 10 −4 ) and pathways\n( n  = 78;  P  = 6.41 ×\n10 −4 ) tested.\n\nSupplementary Material is available at  HMG  online .\n\nThe endometriosis GWAS was supported by a grant from the  Wellcome Trust  ( WT084766/Z/08/Z ) and makes use of WTCCC2 control data generated by the\nWellcome Trust Case-Control Consortium. A full list of the investigators who contributed\nto the generation of these data is available from  http://www.wtccc.org.uk . Funding for the\nWTCCC project was provided by the  Wellcome\nTrust  under awards  076113 and\n085475 . The QIMR study was supported by grants from the  National Health and Medical Research Council (NHMRC) of\nAustralia  ( 241944, 339462, 389927,\n389875, 389891, 389892, 389938, 443036, 442915, 442981, 496610, 496739, 552485 and\n552498 ), the Cooperative Research Centre for Discovery of Genes for\nCommon Human Diseases (CRC), Cerylid Biosciences (Melbourne) and donations from N.\nHawkins and S. Hawkins. S.M. was supported by  NHMRC\nCareer Development Awards  ( 496674,\n613705 ). D.R.N. was supported by the  NHMRC Fellowship  ( 339462 and\n613674 ) and the  ARC Future\nFellowship  ( FT0991022 )\nschemes. A.P.M. was supported by a Wellcome Trust Senior Research Fellowship. G.W.M. was\nsupported by the  NHMRC Fellowships\nScheme  ( 339446, 619667 ).\nK.T.Z. was supported by a  Wellcome Trust Research\nCareer Development Fellowship  ( WT085235/Z/08/Z ). C.M.L. was supported by a  Wellcome Trust Research Career Development Fellow \n( 086596/Z/08/Z ). N.R. was supported by an\n MRC  grant ( MR/K011480/1 ). Funding to pay the Open Access publication\ncharges for this article was provided by the Wellcome Trust.","source_license":"public-domain-us","license_restricted":false}