Genome-wide en ichmen analysis be ween
endome iosis and obesi y- ela ed ai s
e eals no el suscep ibili y loci
Nilu e Rahmioglu1, S ua Macg ego 2, Alexande W. D ong1,A
˚sa K. Hedman1,5, Holly R. Ha is6,7,
Joshua C. Randall8, Inga P okopenko1,9,10, The In e na ional Endogene Conso ium (IEC),
The GIANT Conso ium, Dale R. Nyhol 3, And ew P. Mo is1,11,
{
, G an W. Mon gome y4,
{
,
S acey A. Missme 6,
{
, Cecilia M. Lindg en1,12,
{
and K ina T. Zonde an1,13,∗,
{
1
Wellcome T us Cen e o Human Gene ics, Uni e si y o Ox o d, Ox o d OX3 7BN, UK,
2
S a is ical Gene ics,
3
Neu ogene ics,
4
Molecula Epidemiology, QIMR Be gho e Medical Resea ch Ins i u e, B isbane, QLD 4029,
Aus alia,
5
Depa men o Medical Sciences, Molecula Epidemiology and Science o Li e Labo a o y, Uppsala
Uni e si y, Uppsala, Sweden,
6
Depa men o Obs e ics, Gynecology and Rep oduc i e Biology, B igham and
Women’s Hospi al and Ha a d Medical School, 75 F ancis S ee , Bos on, MA 02115, USA,
7
Uni o Nu i ional
Epidemiology, Ins i u e o En i onmen al Medicine, Ka olinska Ins i u e , PO Box 210, SE-171 77 S ockholm,
Sweden,
8
Wellcome T us Sange Ins i u e, Hinx on, Camb idge CB10 1SA, UK,
9
Depa men o Genomics o
Common Disease, Impe ial College London, London W12 0NN, UK,
10
Ox o d Cen e o Diabe es, Endoc inology
and Me abolism, Uni e si y o Ox o d, Ox o d OX3 7LJ, UK,
11
Depa men o Bios a is ics, Uni e si y o
Li e pool, Duncan Building, Daulby S ee , Li e pool L69 3GA, UK,
12
B oad Ins i u e o he Massachuse s Ins i u e
o Technology and Ha a d Uni e si y, Camb idge 02142 MA, USA and
13
Nu ield Depa men o Obs e ics
and Gynaecology & Endome iosis CaRe Cen e, Uni e si y o Ox o d, John Radcli e Hospi al, Ox o d OX3 9DU,
UK
Recei ed Ap il 4, 2014; Re ised and Accep ed Oc obe 6, 2014
Endome iosis is a ch onic in lamma o y condi ion in women ha esul s in pel ic pain and sub e ili y, and has
been associa ed wi h dec eased body mass index (BMI). Gene ic a ian s con ibu ing o he he i able
componen ha e s a ed o eme ge om genome-wide associa ion s udies (GWAS), al hough he majo i y
emain unknown. Unexpec edly, we obse ed an in e genic locus on 7p15.2 ha was genome-wide signi ican ly
associa ed wi h bo h endome iosis and a dis ibu ion (wais - o-hip a io adjus ed o BMI; WHRadjBMI) in an
independen me a-GWAS o Eu opean ances y indi iduals. This led us o in es iga e he po en ial o e lap in
gene ic a ian s unde lying he ae iology o endome iosis, WHRadjBMI and BMI using GWAS da a. Ou ana-
lyses demons a ed signi ican en ichmen o common a ian s be ween a dis ibu ion and endome iosis
(P53.7 310
23
), which was s onge when we es ic ed he in es iga ion o mo e se e e (S age B) cases
(P54.5 310
24
). Howe e , no gene ic en ichmen was obse ed be ween endome iosis and BMI (P50.79).
In addi ion o 7p15.2, we iden i y ou mo e a ian s wi h s a is ically signi ican e idence o in ol emen in
bo h endome iosis and WHRadjBMI (in/nea KIFAP3,CAB39L,WNT4,GRB14); wo o hese, KIFAP3 and
CAB39L, a e no el associa ions o bo h ai s. KIFAP3,WNT4 and 7p15.2 a e associa ed wi h he WNT signalling
pa hway; o mal pa hway analysis con i med a s a is ically signi ican (P56.41 310
24
) o e ep esen a ion o
sha ed associa ions in de elopmen al p ocesses/WNT signalling be ween he wo ai s. Ou esul s
†
These au ho s join ly di ec ed his wo k.
∗
To whom co espondence should be add essed a : Wellcome T us Cen e o Human Gene ics/Nu ield Depa men o Obs e ics & Gynaecology,
Uni e si y o Ox o d, Roose el D i e, Ox o d OX3 7BN, UK. Tel: +44 1865 287627; Email: [email p o ec ed]
#The Au ho 2014. Published by Ox o d Uni e si y P ess.
This is an Open Access a icle dis ibu ed unde he e ms o he C ea i e Commons A ibu ion License (h p://c ea i ecommons.o g/licenses/by/4.0/),
which pe mi s un es ic ed euse, dis ibu ion, and ep oduc ion in any medium, p o ided he o iginal wo k is p ope ly ci ed.
Human Molecula Gene ics, 2015, Vol. 24, No. 4 1185–1199
doi:10.1093/hmg/ddu516
Ad ance Access published on Oc obe 8, 2014
a Tampe e Uni e si y Lib a y. Depa men o Heal h Sciences on Sep embe 27, 2016h p://hmg.ox o djou nals.o g/Downloaded om
demons a e an example o po en ial biological pleio opy ha was hi he o unknown, and ep esen an oppo -
uni y o unc ional ollow-up o loci and u he c oss-pheno ype compa isons o assess how a dis ibu ion
and endome iosis pa hogenesis esea ch ields can in o m each o he .
INTRODUCTION
Endome iosis is a common condi ion in p emenopausal women
cha ac e ized by ch onic pel ic in lamma ion causing pain and
sub e ili y (1), and has an es ima ed he i abili y o 51% (2).
The In e na ional Endogene Conso ium (IEC) pe o med
he la ges endome iosis GWAS o da e in 3194 su gically
con i med cases (including 1364 mode a e–se e e—S age
B—cases) and 7060 con ols o Eu opean ances y, wi h eplica-
ion in a u he 2392 cases and 2271 con ols (3). One genome-
wide signi ican locus was obse ed in an in e genic egion on
ch omosome 7p15.2 ( s12700667), p ima ily associa ed wi h
S age B disease (P¼1.5 ×10
29
,OR¼1.38, 95% CI 1.24–
1.53). A second locus nea WNT4 ( s7521902) was ound a e
me a-analysis wi h published esul s om a Japanese GWAS
o 1423 cases and 1318 con ols (4); a genome-wide me a-
analysis con i med he wo loci and ound a u he i e (5).
Rs12700667 on 7p15.2 also ma ked 1 o 16 epo ed genome-
wide signi ican loci associa ed wi h wais - o-hip a io adjus ed
o BMI(WHRadjBMI) in an independen GWASme a-analysis
by he GIANT Conso ium in ol ing 77 167 indi iduals o
Eu opean ances y wi h eplica ion in a u he 113 636 indi i-
duals ( s1055144: disco e y P¼1.5 ×10
28
; me a-analysis
P¼1.0 ×10
224
;
2
¼0.5 wi h s12700667 in 1000G pilo
CEU da a) (6,7). This was su p ising, as p ospec i e epidemio-
logical s udies ha e sugges ed consis en ly ha educed
BMI—a measu e o o e all adiposi y—is associa ed wi h
inc eased isk o endome iosis, bu he e is ela i ely limi ed
e idence o an associa ion wi h WHRadjBMI—a measu e o
a dis ibu ion (8,9). We conduc ed a logis ic eg ession analysis
in he IEC da ase o s1055144 on endome iosis disease
s a us, condi ioning on s12700667, which demons a ed ha
he SNPs e lec ed he same associa ion signal (unpublished
da a; condi ional P¼0.65).
The epidemiological e idence o an associa ion be ween
endome iosis and BMI, oge he wi h he obse ed GWAS
locus in common be ween endome iosis and WHRadjBMI,
led us o conduc a sys ema ic in es iga ion o o e lap in associ-
a ion signals be ween he IEC endome iosis GWAS and GIANT
Conso ium WHRadjBMI (N¼77 167) (6,7) and BMI (N¼
123 865) (7,10) me a-GWAS da ase s h ough gene ic en ich-
men analyses.
RESULTS
Gene ic en ichmen analysis o endome iosis wi h
o e all adiposi y and a dis ibu ion
Using independen , impu ed (1000 Genomes pilo e e ence
panel) GWAS da ase s o endome iosis (IEC; 3194 cases in-
cluding 1364 S age B cases, 7060 con ols), BMI (GIANT;
123 865 indi iduals) and WHRadjBMI (GIANT: 77 167 indi i-
duals), we i s conside ed loci genome-wide signi ican ly
associa ed wi h endome iosis, BMI o WHRadjBMI in each
o he indi idual GWAS. The wo genome-wide signi ican
endome iosis loci (in e genic 7p15.2 and WNT4) had signi i-
can ly lowe P- alues o associa ion han expec ed by chance
in he WHRadjBMI GWAS (Table 1: s12700667, P¼4.4 ×
10
25
and s7521902, P¼1.3 ×10
23
; binomial P¼1.0 ×
10
24
), while 2 o he 16 genome-wide signi ican WHRadjBMI
loci (in e genic 7p15.2 and GRB14) had P,0.01 in he
endome iosis GWAS (binomial P¼0.011). No en ichmen
be ween genome-wide signi ican ly associa ed loci was ob-
se ed o endome iosis e sus BMI (Supplemen a y Ma e ial,
Table S1: s12700667, P¼0.27 and s7521902, P¼0.92).
To in es iga e whe he s a is ical en ichmen ex ended beyond
genome-widesigni ican loci, wein es iga ed hemos signi ican
(P,1×10
23
) independen (
2
,0.2) endome iosis GWAS
signals o en ichmen o WHRadjBMI o BMI signals wi h
P,0.05 and ice e sa (numbe o lookup SNPs pe da ase :
n¼717 o 748; see Supplemen a y Ma e ial, Me hods). We
obse ed s a is ically signi ican en ichmen be ween a ian s asso-
cia ed wi h endome iosis (pa icula ly S age B) and WHRadjBMI
(all endome iosis e sus WHRadjBMI: P¼3.7 ×10
23
; S age
B endome iosis e sus WHRadjBMI: P¼4.5 ×10
24
), bu no
be ween endome iosis and BMI (all endome iosis e sus BMI:
P¼0.79; S age B endome iosis e sus BMI: P¼0.85) (Fig. 1;
Supplemen a y Ma e ial, Table S2). Resul s we e simila when
using emale-limi ed WHRadjBMI (N¼42 969 women) and
BMI (N¼73 137 women) GWAS summa y s a is ics (7); o op-
imize powe , in he emainde o he pape we he e o e ocus on
sex-combined WHRadjBMI and BMI da ase s (Supplemen a y
Ma e ial, Fig. S1). Empi ical es ing o s a is ical en ichmen
h ough pe mu a ion (see Supplemen a y Ma e ial, Me hods)
p o ided nea -iden ical esul s (Fig. 1; Supplemen a y Ma e ial,
Fig. S1).
The choice o signi icance h esholds in he disco e y and
lookup da ase s was based on a balance be ween applying a su -
icien ly s ingen signi icance h eshold in he disco e y da ase
ha would minimize he p opo ion o alse-posi i e associa ion
signals, while s ill ha ing su icien numbe s o loci in each o he
pheno ypic associa ion s a a o in es iga e s a is ical en ich-
men , and allow he pu sui o meaning ul biological pa hway
analyses subsequen ly. We conside ed he e ec o di e en sig-
ni icance h esholds, o bo h disco e y and lookup, which con-
i med esul s showing en ichmen o associa ion signals
be ween endome iosis and WHRadjBMI (Supplemen a y Ma-
e ial, Table S3), bu no en ichmen be ween endome iosis
and BMI (Supplemen a y Ma e ial, Table S4).
To in es iga e po en ial genome-wide sha ing o loci be ween
endome iosis and WHRadjBMI o BMI, we pe o med poly-
genic p edic ion analyses (11) e alua ing whe he he agg ega e
e ec o many a ian s o small e ec in he WHRadjBMI and
BMI GWAS could p edic endome iosis s a us in he IEC
GWAS (see Supplemen a y Ma e ial, Me hods). The e was no
signi ican associa ion be ween he WHRadjBMI- o BMI-
1186 Human Molecula Gene ics, 2015, Vol. 24, No. 4
a Tampe e Uni e si y Lib a y. Depa men o Heal h Sciences on Sep embe 27, 2016h p://hmg.ox o djou nals.o g/Downloaded om
Table 1. Associa ion esul s o published IEC genome-wide signi ican endome iosis loci (3) in he GIANT WHRadjBMI GWAS, and o WHRadjBMI loci (6,7) in endome iosis GWAS (lookup esul s a e
shown in bold)
GWAS SNP (p oxy;
2
) Ch Loca ion (B36) RAF (allele) S a us Endome iosis all cases Endome iosis S age B only O e all WHRadjBMI Female-limi ed WHRadjBMI Nea es gene
P- alue
c
OR (95% CI) P- alue
c
OR (95% CI) P- alue
d
E ec (SE) P- alue
e
E ec (SE)
Endome iosis s12700667 7 25 868 164 0.74 (A) G 5.1 ×10
27
1.21 (1.12–1.31) 3.3 ×10
28
1.36 (1.23–1.50) 4.4 310
25
20.023 (0.005) 3.3 310
28
20.023 (0.005) In e genic
Endome iosis s7521902 1 22 363 311 0.25 (A) G 8.9 ×10
25
1.16 (1.08–1.25) 7.5 ×10
25
1.26 (1.14–1.39) 1.3 310
23
20.020 (0.006) 6.1 310
23
20.023 (0.009) WNT4
WHRadjBMI s1055144
a
7 25 837 634 0.19 (T) G 3.7 310
25
0.84 (0.77–0.91) 3.1 ×10
24
0.78 (0.70–0.88) 1.5 ×10
28
0.034 (0.006) 2.3 ×10
26
0.039 (0.008) In e genic
WHRadjBMI s10195252 2 165 221 337 0.41 (C) G 9.8 310
23
0.92 (0.85–0.98) 0.56 0.92 (0.84–1.00) 3.2 ×10
210
20.031 (0.005) 6.3 ×10
215
20.053 (0.007) GRB14
Female WHRadjBMI s4684854 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
24
0.019 (0.005) 2.3 ×10
28
0.039 (0.007) PPARG
WHRadjBMI s718314 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
28
0.031 (0.005) 8.2 ×10
210
0.047 (0.008) ITPR2-SSPN
WHRadjBMI s6861681 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
26
0.026 (0.005) 2.1 ×10
24
0.027 (0.007) CPEB4
WHRadjBMI s6795735 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
27
20.025 (0.005) 7.8 ×10
27
20.033 (0.007) ADAMTS9
WHRadjBMI s2820446
( s4846567,
2
¼1)
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
212
0.037 (0.005) 8.5 ×10
218
0.064 (0.007) LYPLAL1
WHRadjBMI s498778
( s6784615,
2
¼1)
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
25
0.055 (0.010) 1.1 ×10
23
0.063 (0.019) NISCH-STAB1
WHRadjBMI s1294421 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
29
20.029 (0.005) 3.4 ×10
28
20.038 (0.007) LY86
WHRadjBMI s9491696 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
214
20.037 (0.005) 3.4 ×10
28
20.038 (0.007) RSPO3
WHRadjBMI s1443512 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
28
0.031 (0.005) 1.4 ×10
29
0.048 (0.008) HOXC13
WHRadjBMI s984222 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
214
20.037 (0.005) 1.2 ×10
27
20.036 (0.007) TBX15-WARS2
WHRadjBMI s4823006 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
210
0.030 (0.005) 6.9 ×10
28
0.037 (0.007) ZNRF3
Female WHRadjBMI s10478424 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
24
0.023 (0.006) 1.0 ×10
25
0.037 (0.009) HSD17B4
WHRadjBMI s1011731 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
210
0.031 (0.005) 2.1 ×10
25
0.028 (0.007) DNM3-PIGC
WHRadjBMI s6905288 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
210
0.033 (0.005) 7.7 ×10
213
0.052 (0.007) VEGFA
a
Logis ic eg ession analysis in he IEC GWAS shows ha s1055144 ma ks he same locus as s12700667 (condi ional P¼0.65;
2
¼0.8).
b
SNP was no geno yped in he endome iosis GWAS da ase ; esul shown is o p oxy SNP.
c
Resul s a e based on an upda ed GWAS pe o med using geno ype da a impu ed up o 1000 Genomes pilo e e ence panel (B36, June 2010).
d
Resul s a e om he GIANT WHRadjBMI disco e y GWAS da ase (N¼77 167); 3 o he 14 WHRadjBMI loci ha e P.5.0 ×10
28
, howe e , hey eached genome-wide signi icance combined wi h eplica ion analyses in up o a u he 113 636
indi iduals (6).
e
Resul s om he GIANT WHRadjBMI disco e y emale-limi ed GWAS da ase (N¼42 969); one o he wo emale-limi ed WHRadjBMI loci ha e P.5.0 ×10
28
, howe e , hey eached genome-wide signi icance combined wi h eplica ion
analyses in up o a u he 71 295 indi iduals (7).
Human Molecula Gene ics, 2015, Vol. 24, No. 4 1187
a Tampe e Uni e si y Lib a y. Depa men o Heal h Sciences on Sep embe 27, 2016h p://hmg.ox o djou nals.o g/Downloaded om
de i ed p o ile sco es (o e all o emale limi ed) and all/S age B
endome iosis (Supplemen a y Ma e ial, Tables S5–S8), sug-
ges ing no e idence o a di ec ionally consis en en masse,
genome-wide, sha ed common gene ic componen .
We nex in es iga ed he a ian s wi h mos signi ican e i-
dence o associa ion wi h bo h endome iosis (P,1×10
23
)
and WHRadjBMI (P,0.05) o p edominance in di ec ion o
pheno ypic e ec s (Supplemen a y Ma e ial, Tables S9 and
S10 and Fig. S2). No s a is ically signi ican di ec ional
consis ency was obse ed o hese a ian s (P.0.47), no
o he 17 loci (Table 1) ha we e genome-wide signi ican ly
associa ed wi h ei he ai (Fig. 2,P.0.44). In e genic
7p15.2 and WNT4 showed disco dan di ec ions o e ec ,
while he e ec o GRB14 was conco dan (Fig. 2). This could
sugges he p esence o mul iple biological pa hways h ough
which he a ian s in luence he wo pheno ypes. We nex se
ou o in es iga e he common biology sugges ed by gene ic a -
ian s associa ed wi h bo h endome iosis and WHRadjBMI.
Figu e 1. Gene ic en ichmen analyses be ween endome iosis, BMI and WHRadjBMI GWAS da ase s, using independen (
2
,0.2) SNPs. The panels show (i) The
p opo ion o SNPs nominally associa ed (P,0.05) wi h WHRadjBMI (A) o BMI (B) by signi icance o o e all and S age B endome iosis associa ion (P,1.0 ×
10
23
e sus P≥1×10
23
); (ii) The p opo iono SNPs nominally associa ed(P,0.05) wi h o e all and S age B endome iosis by signi icance o WHRadjBMI (C)
and BMI (D) associa ion (P,1.0 ×10
23
e sus P≥1×10
23
). P- alues o
x
2
es s assessing s a is ical di e ence be ween p opo ions a e shown abo e each se o
ba s, and 95% con idence in e als o he p opo ions a e gi en on each ba . Fo di e ences wi h P
chisq
,0.2, empi ical P- alues a e gi en in b acke s (see Supple-
men a y Ma e ial, Me hods).
1188 Human Molecula Gene ics, 2015, Vol. 24, No. 4
a Tampe e Uni e si y Lib a y. Depa men o Heal h Sciences on Sep embe 27, 2016h p://hmg.ox o djou nals.o g/Downloaded om
Biology o he loci sha ed be ween endome iosis
and a dis ibu ion
Ou analysis showing signi ican en ichmen be ween SNPs
associa ed wi h all o S age B endome iosis (P,1×10
23
)
and WHRadjBMI (P,0.05) shown in Figu e 1in ol ed 1284
independen (
2
.0.2) loci. We explo ed he biological unc ion
o he loci mos s ongly associa ed wi h WHRadjBMI, a
nominal P,0.005 (n¼16, Table 2; see Supplemen a y Ma e -
ial, Tables S11 and S12 o all a ian s associa ed a P,0.05).
Two no el loci, s560584 nea KIFAP3 (all endome iosis) and
s11619804 in CAB39L (S age B endome iosis), we e signi i-
can ly associa ed wi h WHRadjBMI a e Bon e oni co ec ion
allowing o 1284 independen es s (P,3.89 ×10
25
).
The endome iosis isk allele T o s560584 (OR ¼1.14
(1.07–1.22), P¼1.42 ×10
24
) was associa ed wi h lowe
WHRadjBMI (
b
¼20.021, P¼1.47 ×10
25
), and loca ed in
an in e genic egion 46 kb downs eam o KIFAP3
(Kinesin-associa ed p o ein 3). Toge he wi h KIF3A and
KIF3B,KIFAP3 o ms a kinesin mo o complex, KIF3, ha
media es cellula anspo o N-cadhe in and b-ca enins (12),
which a e in ol ed in cell adhesion, he Wn canonical
pa hway and cell cycle p og ession (13). The Wn /
b
-ca enin sig-
nalling pa hway ac s as a molecula swi ch o adipogenesis (14)
and has mul iple sugges ed oles in endome iosis h ough sex
ho mone homeos asis egula ion (15), i s ole in de elopmen
o emale ep oduc i e o gans (16), molecula mechanisms o
in e ili y (17) and media ion o ib ogenesis (18).
The S age B endome iosis isk allele C o s11619804 (OR ¼
1.17 (1.07–1.28); P¼4.88 ×10
24
), loca ed in CAB39L
(Calcium-Binding P o ein 39-Like), was associa ed wi h inc eased
WHRadjBMI (
b
¼0.022, P¼1.06 ×10
25
;Table2). The unc-
ion o his gene is no well cha ac e ized bu he encoded p o ein
in e ac s wi h a se ine h eonine kinase (STK11) ha unc ions as
a umou supp esso (19).
Rs12700667 in he in e genic egion 7p15.2 emained among
he s onges associa ed sha ed signals, wi h he endome iosis
isk allele A associa ed wi h educed WHRadjBMI (
b
¼20.023,
P¼4.4 ×10
25
). The associa ion maps o an in e genic high
LD egion o 48 kb (
2
.0.8) o unknown unc ionali y.
Fu he in e es ing nea by loci include he miRNA hsa-mi -
148a, wi h a pu po ed ole in Wn /
b
-ca enin signalling (14);
NFE2L3 (nuclea ac o (e y h oid-de i ed 2)-like 3), a an-
sc ip ion ac o sugges ed o be in ol ed in cell di e en ia ion,
in lamma ion and ca cinogenesis (20). The WNT signalling
pa hway was u he highligh ed by he nominal associa ion o
wo independen (
2
¼0.06) endome iosis isk a ian s nea
WNT4 (wingless- ype MMTV in eg a ion si e amily), s3820282
(genic) and s2807357 (22.4 kb downs eam), wi h educed
WHRadjBMI (
b
¼20.019, P¼5.0 ×10
23
;
b
¼20.015, P¼
3.7 ×10
23
;Table2). O no e is ha all sha ed a ian s implica ed
in WNT signalling (in/nea in e genic 7p15.2, WNT4,KIFAP3)
showed consis en —disco dan —pheno ypic di ec ions o e ec .
Risk a ian s10195252, 34.6 kb downs eam o GRB14
(g ow h ac o ecep o -bound p o ein 14) was he hi d locus
wi h signi ican e idence o associa ion wi h bo h o e all (no
S age B) endome iosis and WHRadjBMI (Table 1). GRB14
has an inhibi o y e ec on insulin ecep o signalling (21),
may ha e a ole in signalling pa hways ha egula e g ow h
and me abolism and has been shown o in e ac wi h ib oblas
g ow h ac o ecep o s (22). This sha ed a ian is also in high
LD (
2
¼0.93 and ¼0.87, espec i ely) wi h a ype 2 diabe es
isk a ian s13389219 (23) and as ing insulin isk a ian
s6717858 (24).
O he loci o in e es include s2921188 in PPARG and
s6556301 nea FGFR4 (Table 2).The endome iosis isk
allele A o s2921188 (OR ¼1.13, 95% CI: 1.05–1.21), P¼
5.9 ×10
24
)inPPARG (pe oxisome p oli e a o -ac i a ed
ecep o gamma) is associa ed wi h inc eased WHRadjBMI
(
b
¼0.017; P¼1.1 ×10
23
). PPARG is a nuclea ho mone
ecep o ha egula es a y acid s o age and glucose me abol-
ism. Syn he ic ligands, such as insulin sensi izing d ugs, a ge
PPARG in ea men o diabe es o lowe se um glucose le els
(25) and a e also documen ed o ha e an i-in lamma o y, an i-
angiogenic and an i-p oli e a i e e ec s on endome ium, wi h
baboon models sugges ing a ole in a ge ing endome io ic
disease (26). S age B endome iosis isk allele G o s6556301
nea FGFR4 ( ib oblas g ow h ac o ecep o ,OR¼1.17
[1.07–1.28], P¼7.4 ×10
24
) is associa ed wi h educed
WHRadjBMI (
b
¼20.021, P¼1.9 ×10
24
). FGFR4 in e ac s
Figu e 2. Di ec ions o e ec o 17 independen SNPs genome-wide signi ican ly associa ed wi h all (A) o S age B (B) endome iosis, o WHRadjBMI. In e genic
7p15.2, WNT4, and GRB14 a e shown in ed. Linea eg ession R
2
and P- alues used o es o signi ican di ec ionali y o e ec s (35) a e shown.
Human Molecula Gene ics, 2015, Vol. 24, No. 4 1189
a Tampe e Uni e si y Lib a y. Depa men o Heal h Sciences on Sep embe 27, 2016h p://hmg.ox o djou nals.o g/Downloaded om

Table 2. Resul s o he op all/S age B endome iosis loci (P,1×10
23
) associa ed wi h WHRadjBMI a P,0.005
SNP Ch Posi ion (B36) RAF (allele) Endome iosis O e all WHRadjBMI Female-limi ed WHRadjBMI Nea es loci
P- alue OR (95% CI) P- alue E ec SE P- alue E ec SE (dis ance)
All cases
s560584 1 168 357 136 0.41 (T) 1.4 ×10
24
1.14 (1.07–1.22) 1.4 ×10
25
20.021 0.005 1.1 ×10
23
20.022 0.677 KIFAP3 (46 632)
s12700667 7 25 868 164 0.74 (A) 5.1 ×10
27
1.22 (1.13–1.32) 4.4 ×10
25
20.023 0.005 3.4 ×10
24
20.028 0.284 NFE2L3 (2 90 221)
s2921188 3 12 387 115 0.64 (A) 5.9 ×10
24
1.13 (1.05–1.21) 1.1 ×10
23
0.017 0.005 1.8 ×10
24
0.026 0.054 PPARG (0)
s1250248 2 215 995 338 0.27 (A) 1.6 ×10
25
1.17 (1.09–1.26) 1.0 ×10
23
0.018 0.005 9.9 ×10
24
0.025 0.242 FN1 (0)
s2630787 3 21 847 339 0.52 (C) 9.2 ×10
24
1.12 (1.05–1.19) 1.9 ×10
23
20.015 0.004 0.38 20.006 0.030 ZNF659 (79 518)
s1430788 2 67 721 916 0.31 (C) 9.3 ×10
25
1.15 (1.07–1.23) 2.7 ×10
23
0.016 0.005 3.1 ×10
23
0.022 0.330 ETAA1 (230 878)
s906721 3 184 687 691 0.41 (A) 6.1 ×10
25
1.16 (1.08–1.24) 4.2 ×10
23
0.015 0.005 1.7 ×10
23
0.023 0.140 KLHL6 (322)
s1868894 4 187 606 728 0.80 (C) 2.3 ×10
24
1.16 (1.07–1.26) 4.9 ×10
23
20.018 0.006 0.13 20.013 0.524 MTNR1A (85 075)
s3820282 1 22 340 802 0.16 (T) 3.3 ×10
27
1.26 (1.15–1.37) 5.0 ×10
23
20.019 0.007 0.09 20.016 0.749 WNT4 (0)
S age B cases
s11619804 13 49 888 131 0.53 (C) 4.8 ×10
24
1.17 (1.07–1.28) 1.1 ×10
25
0.022 0.005 2.2 ×10
22
0.016 0.022 CAB39L (0)
s12700667 7 25 868 164 0.74 (A) 3.3 ×10
29
1.36 (1.23–1.50) 4.4 ×10
25
20.023 0.005 3.4 ×10
24
20.028 0.284 NFE2L3 (290 221)
s2782659 6 45 794 768 0.33 (G) 4.2 ×10
24
1.18 (1.08–1.30) 9.2 ×10
25
0.020 0.005 1.7 ×10
24
0.027 0.108 RUNX2 (167 970)
s6556301 5 176 460 183 0.63 (G) 7.4 ×10
24
1.17 (1.07–1.28) 1.9 ×10
24
20.021 0.005 7.8 ×10
23
20.021 0.845 FGFR4 (2450)
s1250248 2 215 995 338 0.27 (A) 2.9 ×10
28
1.32 (1.19–1.45) 1.2 ×10
23
0.018 0.005 9.9 ×10
24
0.025 0.242 FN1 (0)
s4131816 1 161 662 648 0.85 (T) 5.4 ×10
24
1.24 (1.10–1.41) 1.5 ×10
23
0.022 0.007 0.25 0.011 0.072 NUF2 (70 470)
s9912335 17 77 552 948 0.69 (T) 3.1 ×10
24
1.19 (1.08–1.31) 3.5 ×10
23
20.021 0.007 0.10 20.016 0.454 ASPSCR1 (0)
s10878362 12 64 703 760 0.69 (C) 4.9 ×10
24
1.19 (1.08–1.31) 3.6 ×10
23
0.015 0.005 3.1 ×10
23
0.022 0.204 HMGA2 (57 421)
s2807357 1 22 364 571 0.64 (A) 9.7 ×10
24
1.16 (1.06–1.27) 3.7 ×10
23
20.015 0.005 1.0 ×10
23
20.024 0.081 WNT4 (22 373)
s906721 3 184 687 691 0.41 (A) 1.4 ×10
24
1.20 (1.09–1.32) 4.2 ×10
23
0.015 0.005 1.7 ×10
23
0.023 0.140 KLHL6 (322)
s12267660 10 4 419 530 0.85 (G) 7.9 ×10
24
1.24 (1.09–1.40) 4.6 ×10
23
0.02 0.007 8.0 ×10
23
0.030 0.133 CR749391 (191 913)
s11685481 2 67 590 253 0.15 (C) 8.4 ×10
24
1.23 (1.09–1.38) 4.8 ×10
23
0.018 0.006 1.1 ×10
22
0.022 0.451 ETAA1 (99 215)
1190 Human Molecula Gene ics, 2015, Vol. 24, No. 4
a Tampe e Uni e si y Lib a y. Depa men o Heal h Sciences on Sep embe 27, 2016h p://hmg.ox o djou nals.o g/Downloaded om
wi h ib oblas g ow h ac o s, which ha e oles in angiogenesis,
wound healing and cell mig a ion (27).
Exp ession quan i a i e ai loci analysis o he sha ed
endome iosis and a dis ibu ion loci
We in es iga ed he po en ial impac o he desc ibed 16 genes
(Table 2) sha ed be ween endome iosis and WHRadjBMI on
ansc ip ional unc ion using h ee public exp ession da a
esou ces: (i) he Mammalian Gene Exp ession U e us da abase
(MGEx-Udb) (28) con aining published in o ma ion on an-
sc ip ional ac i i y o speci ic genes in human endome ial
issue om indi iduals wi h and wi hou endome iosis; (ii)
he MuTHER s udy which collec ed exp ession and eQTL da a
om 776 abdominal a issues (29); and (iii) he MOLOBB
da ase o di e en ial exp ession le els be ween abdominal
and glu eal a om 49 indi iduals (30). Based on he limi ed
a ailable e idence in he MGEx-Udb da abase, wo genes a e
ansc ibed in endome ial issue o women wi h endome iosis
bu do man in hose wi hou endome iosis: PPARG and
FGFR4 (Supplemen a y Ma e ial, Table S13). O he 16
genes, 15 had p obes p esen wi hin 1 Mb ei he side o he
SNP in he MuTHER da abase; howe e , none showed signi i-
can associa ion wi h nea by ansc ip s in abdominal a issue
(Supplemen a y Ma e ial, Table S14). The MOLOBB s udy
da a showed cis-eQTL e idence o di e en ial exp ession o
wo genes; KIFAP3 ( s560584; old change ¼0.14, adjus ed
P¼0.04) (Supplemen a y Ma e ial, Table S15). Addi ional
ansc ip ional e idence ele an o he in e genic 7p15.2 locus
includes he p esence o an exp ession QTL associa ed wi h a
ansc ip o unknown unc ion, AA553656, in subcu aneous
abdominal a issue (6), and he di e en ial exp ession o
nea by hsa-miR-148a be ween glu eal and abdominal a issue
samples (31).
Pa hway analysis
To iden i y po en ial common biological pa hways in ol ed
in he ae iology o endome iosis and he a iabili y o a
dis ibu ion, we conduc ed pa hway analyses using genes wi h
e idence o en ichmen be ween he ai s using (i) he
PANTHER da abase (32) and (ii) GRAIL (33). Fo he
PANTHER analysis, we selec ed he 91 and 108 genes loca ed
in a 1 Mb in e al su ounding each independen SNP associa ed
wi h all endome iosis (P,1.0 ×10
23
) and WHRadjBMI
(P,0.05), and S age B endome iosis (P,1.0 ×10
23
) and
WHRadjBMI (P,0.05), espec i ely (see Supplemen a y Ma-
e ial, Me hods). This excluded in e genic loci wi hou a gene
wi hin 1 Mb, such as ou op sha ed locus a 7p15.2. We es ed
whe he he wo se s o genes showed signi ican o e ep esen-
a ion o a pa icula pa hway, o each o 176 cu a ed pa hways
and 241 biological p ocesses. The op en iched pa hways we e
‘de elopmen al p ocesses’ (all endome iosis: P¼1.2 ×
10
25
; S age B: P¼1.25 ×10
24
), ‘WNT signalling’ (all endo-
me iosis: P¼1.07 ×10
24
), ‘gonado opin- eleasing
ho mone ecep o ’ (all endome iosis: P¼1.48 ×10
23
), ‘cad-
he in signalling’ (S age B: P¼6.42 ×10
24
), ‘FGF signalling’
(S ageB: P¼2.96 ×10
23
)and‘TGF-be a signalling’(S age B:
P¼1.48 ×10
23
) pa hways (Supplemen a y Ma e ial, Tables
S16 and S17). Bon e oni co ec ion o he numbe o pa hways
es ed (see Supplemen a y Ma e ial, Me hods) ende ed ‘WNT
signalling’, ‘de elopmen al p ocesses’, ‘cellula p ocesses’
and ‘cell communica ion’ signi ican ly en iched; howe e , his
adjus men is conse a i e, as exempli ied by ‘cadhe in signal-
ling’ genes being a subse o hose in he ‘WNT signalling’
pa hway. Sensi i i y analyses explo ing he e ec o di e en
endome iosis associa ion h esholds on pa hway analyses
showed e y consis en esul s o h eshold P,1.0 ×10
22
,
wi h he same op h ee en iched pa hways—WNT signalling,
Cadhe in signalling and Gonado opin- eleasing ho mone e-
cep o pa hway. No meaning ul pa hway analyses could be con-
duc ed on he limi ed numbe o genes passing associa ion
h eshold P,1×10
24
(Supplemen a y Ma e ial, Table S18).
We used GRAIL (33) o sea ch o connec i i y be ween he
91 and 108 genes all/S age B endome iosis and WHRadjBMI-
associa ed genes and speci ic keywo ds om he published
li e a u e ha desc ibe po en ial unc ional connec ions. We iden-
i ied 17 genes wi h nominal signi icance (P,0.05) o po en ial
unc ional connec i i y o ‘all’ endome iosis and WHRadjBMI
and six genes o S age B endome iosis and WHRadjBMI
(Supplemen a y Ma e ial, Fig. S3 and Tables S19 and S20).
The keywo ds associa ed wi h hese connec ions included ‘cad-
he in’, ‘di e en ia ion’, ‘de elopmen ’ and ‘insulin’ o ‘all’
endo, and ‘de elopmen ’ and ‘emb yos’ o S age B endome i-
osis, ma king again de elopmen alp ocesses and cadhe in signal-
ling as biological pa hways sha ed in he o igins o endome iosis
and a dis ibu ion.
DISCUSSION
In his s udy, we ha e in es iga ed he o e lap in gene ic asso-
cia ion signals om he la ges GWA s udies o da e o
endome iosis, o e all adiposi y (BMI) and a dis ibu ion
(WHRadjBMI). Ou esul s demons a ed ha he e is a sha ed
gene ic basis be ween endome iosis and a dis ibu ion ha
ex ends o e and abo e he single genome-wide signi ican
locus ha has been epo ed in GWAS o he sepa a e ai s.
Ou analyses highligh no el loci in/nea KIFAP3 and
CAB39L, which oge he wi h in e genic 7p15.2, WNT4 and
GRB14, showed signi ican e idence o ai associa ion
sha ing. The s eng h o e idence o en ichmen was simila
o o e all e sus emale-limi ed WHRadjBMI loci, which
may be unexpec ed, gi en ha endome iosis is a emale con-
di ion. Howe e , he lack o a s onge en ichmen be ween
emale-speci ic WHRadjBMI GWAS esul s and endome i-
osis, compa ed wi h all WHRadjBMI esul s should be consid-
e ed agains he e ec s o a educed sample size used o
emale-speci ic WHRadjBMI analyses on powe o associa ion
de ec ion.
The en ichmen o associa ed a ian s was gene ally s onge
when he endome iosis cases we e es ic ed o mode a e–
se e e (S age B) disease, despi e he smalle sample size.
Indeed, he associa ion o he op in e genic GWAS locus on
7p15.2, also genome-wide signi ican ly associa ed wi h
WHRadjBMI, is limi ed o S age B endome iosis. S age B—
o ASRM S ages III/IV disease (34)—is ypically cha ac e ized
by o a ian (endome ioma) o deep in il a ing ( ec o aginal)
lesions, which we e shown o ha e a subs an ially g ea e unde -
lying gene ic con ibu ion han milde , pe i oneal disease
(ASRM S age I/II) (3). The pa icula en ichmen be ween
Human Molecula Gene ics, 2015, Vol. 24, No. 4 1191
a Tampe e Uni e si y Lib a y. Depa men o Heal h Sciences on Sep embe 27, 2016h p://hmg.ox o djou nals.o g/Downloaded om
WHRadjBMI and S ages III/IV endome iosis is in iguing, and
ano he eason o u he unc ional wo k o concen a e on his
endome iosis sub- ype. The e a e, howe e , speci ic loci ha
show en ichmen o associa ion wi h WHRadjBMI and o e all
endome iosis, he analysis o which he e o e emains o in e -
es . An example is GRB14, which did no show signi ican asso-
cia ion wi h S age B disease, displayed a conco dan di ec ion o
e ec be ween endome iosis and WHRadjBMI, and he bio-
logical unc ion o which also seems o sugges an en i ely di e -
en con ibu ion o he o igins o bo h pheno ypes han he 7p15.2
and WNT4 loci.
The limi ed a ailable eQTL da a showed signi ican e idence
o di e en ial exp ession o KIFAP3 be ween di e en a
depo s. The a ian s wi h mos e idence o en ichmen
be ween he ai s, in/nea in e genic 7p15.2, KIFAP3 and
WNT4, we e all implica ed in WNT signalling and had consis -
en —disco dan —di ec ions o e ec , wi h endome iosis isk
alleles associa ed wi h a dec eased WHRadjBMI. Indeed, bio-
logical pa hway analyses showed signi ican e idence o he in-
ol emen o de elopmen al p ocesses and WNT signalling in
endome iosis ae iology and egula ion o a dis ibu ion, a po-
en ial pleio opic connec ion ha has no been epo ed o da e.
The ela i ely limi ed epidemiological e idence o pheno yp-
ic co ela ion be ween endome iosis and WHRadjBMI (8,9)is
consis en wi h he absence o s ong di ec ional consis ency
o pheno ypic e ec s o gene ic a ian s unde lying bo h ai s
a a genome-wide le el. Mos s udies o gene ic pleio opy
be ween ai s o da e ha e ocused on genome-wide di ec ional
consis ency be ween epidemiologically o clinically (pos u-
la ed) co ela ed ai s, such as di e en me abolic ai s (6,35)
o psychia ic condi ions (36). Howe e , genome-wide consis -
ency in di ec ionali y o pheno ypic e ec s would mos likely
apply o ai s ha sha e a la ge p opo ion o causali y, and
ha epidemiologically lie on he same causal pa hway(s) and
a e hus mo e likely o be examples o media ed (gene ic a ian s
in luencing one pheno ype indi ec ly h ough associa ion wi h
a second pheno ype) a he han biological (gene ic a ian s
exe ing a di ec biological in luence on mo e han one pheno-
ype) pleio opy (37). Thus, ou esul s o gene ic en ichmen
be ween endome iosis and WHRadjBMI demons a e an
example o he biological complexi y o ae iological associa ions
be ween complex ai s, and sugges ha he unde lying sha ed
loci a e po en ially biologically pleio opic, gi en he absence
o pheno ypic co ela ion be ween endome iosis and
WHRadjBMI and absence o en masse di ec ional consis ency
o sha ed gene ic a ian s on he pheno ypes (37,38). I also
demons a es mo e gene ally how po en ial pe u ba ion o a
causal pa hway h ough, o example, d ug ea men a ge ing
one ai could ha e unexpec ed e ec s on ano he , e en when
he e is no clea e idence ha hese ai s a e associa ed clinically
o epidemiologically—a p oblem o en encoun e ed in d ug de-
elopmen . Sys ema ic explo a ion o biological pleio opy o
gene ic a ian s ma king po en ial d ug a ge s may help in high-
ligh ing he po en ial o such unwan ed o unexpec ed e ec s.
While he obse ed gene ic en ichmen be ween endome i-
osis and WHRadjBMI p esen s new a enues o explo ing
common biology, he o al absence o any gene ic en ichmen
be ween endome iosis and BMI (wi hin he limi s o powe
p esen ed by hese la ge da ase s) is in iguing gi en he consis -
en , p ospec i e, obse a ional epidemiological e idence o
pheno ypic associa ion be ween educed BMI and endome iosis
isk (8). Ou analyses ep esen an adap a ion o Mendelian an-
domiza ion analyses (39,40), in which gene ic a ian s obus ly
associa ed wi h BMI in he la ges GWAS analyses o da e (10)
a e in es iga ed o associa ion wi h endome iosis. The o al
lack o gene ic en ichmen sugges s ha educed BMI is no
causally ela ed o endome iosis isk. Ra he , i sugges s ha
he obse ed pheno ypic associa ion (8) is ei he d i en by
sha ed en i onmen al ac o s, o is due o con ounding ac o s
ela ed o BMI a ec ing, o example, diagnos ic oppo uni y
o endome iosis.
These no el indings p esen an en i ely new oppo uni y o
unc ional a ge ed ollow-up o pleio opic loci be ween endo-
me iosis and WHRadjBMI in ele an disease issues such as
endome ium and a issue, cellula sys ems, animal models
and u he c oss- ai compa isons, o unco e hei biological
unc ions and o assess how s udies in he a dis ibu ion esea ch
ield can in o m esea ch in o endome iosis pa hogenesis, bio-
ma ke iden i ica ion and d ug a ge disco e y and alida ion.
MATERIALS AND METHODS
Genome-wide associa ion s udies
IEC endome iosis GWAS
This GWAS included 3194 su gically con i med endome iosis
cases and 7060 con ols om Aus alia and he UK. Disease se-
e i y o he endome iosis cases was assessed e ospec i ely
om su gical eco ds using he AFS classi ica ion sys em and
g ouped in o wo pheno ypes: S age A (S age I o II disease o
some o a ian disease wi h a ew adhesions; n¼1686)o S age
B (S age III o IV disease; n¼1364). We p e iously showed an
inc eased gene ic loading among 1364 cases wi h S age B endo-
me iosis compa ed wi h 1666 wi h S age A disease (3), which
led o wo GWA analyses, including (i) 3194 ‘all’ endome iosis
case and (ii) 1364 S age B cases (Table 3). The geno yped da a
we e impu ed up o 1000 Genomes pilo e e ence panel (B36,
June 2010) and he GWAS was pe o med again, using a
missing da a likelihood in a logis ic eg ession model including
Table 3. Summa y desc ip ion o he GWAS used in he gene ic en ichmen
analysis
GWAS Conso ium Sample
size
No. o
SNPs
(million)
Re e ences
Endome iosis—
all cases
IEC 3194 cases,
7060
con ols
12.5 Pain e e al.(3)
Endome iosis—
S age B cases
IEC 1363 cases,
7060
con ols
12.5 Pain e e al.(3)
WHRadjBMI GIANT 77 167 2.85 Heid e al.(6)
Female-limi ed
WHRadjBMI
GIANT 42 969 2.85 Randall e al.(7)
BMI GIANT 123 865 2.85 Spelio es e al.(10)
Female-limi ed
BMI
GIANT 73 137 2.85 Randall e al.(7)
IEC, In e na ional Endogene Conso ium; GIANT, Gene ic In es iga ion o
An h opome ic T ai s Conso ium; BMI, body mass index adjus ed o age;
WHRadjBMI, wais o hip a io adjus ed o BMI and age.
1192 Human Molecula Gene ics, 2015, Vol. 24, No. 4
a Tampe e Uni e si y Lib a y. Depa men o Heal h Sciences on Sep embe 27, 2016h p://hmg.ox o djou nals.o g/Downloaded om
a co a ia e ep esen ing he Aus alian and he UKs a a, wi h he
impu ed da a (N¼12.5 million SNPs). The en ichmen analysis
we p esen is om his se o esul s.
GIANT Conso ium
WHR GWAS. A o al o 77 167 subjec s o Eu opean ances y in-
o ma i e o body a dis ibu ion measu emen WHR om 32
GWAS we e included (6). The geno ype da a we e impu ed up
o HapMap 2 CEU e e ence panel. The associa ions o 2.85
million SNPs wi h WHR we e examined in a ixed-e ec s
me a-analysis, a e in e se no mal ans o ma ion o WHR
and adjus ing o BMI and age wi hin each s udy in an addi i e
gene ic model; analyses we e conduc ed o males and emales
combined (6) and limi ed o emales only (7) (Table 3).
BMI GWAS. A o al o 123 865 subjec s wi h o e all adiposi y
measu emen BMI om 46 GWAS we e included (10). The
geno ype da a we e impu ed up o HapMap wo CEU e e ence
panels. The associa ions o 2.85 million SNPs wi h BMI we e
es ed in an in e se- a iance me a-analysis, a e in e se no -
mally ans o ma ion o BMI and adjus ing o age and o he ap-
p op ia e co a ia es in an addi i e gene ic model wi hin each
s udy; analyses we e conduc ed o males and emales combined
(10) and limi ed o emales only (7) (Table 3).
Gene ic en ichmen analysis
Wi h one es o associa ion conduc ed o each SNP, he GWAS
analyses p oduced a genome-wide dis ibu ion o P- alues o in-
di idual SNP associa ions. P io o es ing en ichmen : (i) he
o e lap o SNPs p esen in endome iosis GWAS e sus
WHRadjBMI and BMI GWAS was aken, (ii) all SNPs wi h
MAF ≤0.01 we e emo ed, (iii) all SNPs wi h A/T and C/G
base pai s we e emo ed, (i ) co ela ed SNPs (
2
.0.2) we e
emo ed as p e iously epo ed (41) by aking he mos signi i-
can ly associa ed SNP and elimina ing all SNPs ha ha e a
HapMap CEU pai wise co ela ion coe icien (
2
).0.2 wi h
ha SNP, hen p ocessing o he nex s ongly associa ed SNP
emaining. This esul ed in 173 157 independen SNPs in endo-
me iosis e sus WHRadjBMI and 173 223 in endome iosis
e sus BMI en ichmen analyses.
The independen SNPs in he ails (P,1×10
23
) o he asso-
cia ion esul s dis ibu ion o he wo endome iosis GWAS (all
endome iosis and ‘S age B’ cases) we e in es iga ed o en ich-
men o WHRadjBMI o BMI low P- alue (P,0.05) associ-
a ion signals; in e e sal, SNPs in he ails o WHRadjBMI
and BMI GWAS (P,1×10
23
) we e in es iga ed o e idence
o nominal associa ion (P,0.05) in he wo endome iosis
GWAS. The h eshold o P,1×10
23
co esponded o he
poin a which endome iosis GWAS esul s s a ed o de ia e
om he null dis ibu ion (e idence o associa ion) in he
o e all and S age B endome iosis Q–Q plo s (Supplemen a y
Ma e ial, Fig. S4). En ichmen was assessed in R by means o
Pea son’s
x
2
es s wi h Ya es’ con inui y co ec ion, es ing o
he di e ence in p opo ion o SNPs wi h associa ion P,0.05
in he lookup da ase acco ding o associa ion in he disco e y
da ase (P,1×10
23
e sus P≥1×10
23
). To es o con-
sis ency in di ec ionali y o pheno ypic e ec s o he SNPs
wi h e idence o en ichmen , linea eg ession analysis was pe -
o med on he e ec (
b
) o each SNP o WHRadjBMI as
p edic o a iable and he e ec (
b
) o endome iosis isk as
he ou come a iable (35). In addi ion, a wo-sided binomial
es was pe o med wi h null hypo hesis P¼0.50.
Pe mu a ion-based en ichmen analysis
Fo hose esul s ha showed nominally signi ican (P,0.10)
e idence o en ichmen in
x
2
es s o con ingency ables, we
pe o med pe mu a ion-based analyses o ob ain empi ical es i-
ma es o signi icance o en ichmen . We (i) andomly picked he
same numbe o independen SNPs ‘associa ed’ wi h he disco -
e y ai a P,1×10
23
(e.g. he numbe o SNPs associa ed
wi h all endome iosis a P,1×10
23
was n¼717) om he
WHRadjBMI da ase ; (ii) coun ed how many o he andomly
selec ed SNPs had P- alues o associa ion wi h WHRadjBMI
,0.05; (iii) epea ed S eps (i) and (ii) 10 000 imes; (i ) de e -
mined he numbe o ins ances among he 10 000 d aws in
which he numbe o SNPs associa ed a P,0.05 wi h
WHRadjBMI was g ea e o equal o he numbe we obse ed
in ou o iginal analysis (e.g. ≥52/717). Fo example, o
o e all endome iosis and o e all WHRadjBMI, we obse ed
his in 26/10 000 ins ances, co esponding o a P- alue o
2.6 ×10
23
, which was e y simila o he P- alue ob ained
om he
x
2
es (P¼3.7 ×10
23
).
Polygenic p edic ion analysis
The independen SNPs in bo h WHRadjBMI and endome iosis
da ase s we e used o conduc a polygenic p edic ion analysis
(11). The aim o his analysis was o e alua e he agg ega e
e ec s o many SNPs o small e ec and assess whe he
subse s o SNPs selec ed in his manne om one disease/ ai
GWAS p edic disease/ ai s a us in ano he , hus p o iding a
measu e o a common polygenic componen wi h conco dan
di ec ions o e ec unde lying he ai s. B ie ly, subse s o
SNPs we e selec ed om he WHRadjBMI GWAS da a based
on hei associa ion wi h WHRadjBMI using inc easingly
libe al h esholds, ha is, P,0.01, P,0.05, P,0.1, P,
0.2, P,0.3, P,0.4, P,0.5 and P,0.75. Using hese
h esholds, we de ined se s o allele-speci ic sco es in he
WHRadjBMI da ase o gene a e isk p o ile sco es o indi i-
duals in he endome iosis da ase . Fo each indi idual in he
endome iosis da ase , we calcula ed he numbe o sco e
alleles hey possessed, each weigh ed by hei e ec size
(
b
- alue) o associa ion in he WHRadjBMI da ase . To assess
whe he he agg ega e sco es we e associa ed wi h endome i-
osis isk, we es ed o a highe mean sco e in cases compa ed
wi h con ols. Logis ic eg ession was used o assess he ela ion-
ship be ween endome iosis disease s a us and agg ega e isk
sco e.
Exp ession analyses
MGEx-Udb
The mammalian gene exp ession u e us da abase (MGEx-Udb)
is a manually cu a ed u e us-speci ic da abase c ea ed using a
me a-analysis app oach om published pape s (28) ha p o-
ides lis s o ansc ibed and do man genes o a ious
no mal, pa hological (e.g. endome iosis, ce ical cance and
endome ial cance ) and expe imen al (e.g. ea men and
Human Molecula Gene ics, 2015, Vol. 24, No. 4 1193
a Tampe e Uni e si y Lib a y. Depa men o Heal h Sciences on Sep embe 27, 2016h p://hmg.ox o djou nals.o g/Downloaded om