genetic evidence that lower circulating
TRANSCRIPT
-
7/24/2019 Genetic Evidence That Lower Circulating
1/9
GENETIC evidence that lower circulating
FSH levels lengthen menstrual cycle,
increase age at menopause and impact
female reproductive health
atherine S! "uth,"o#in N! $eaumont, %essicaTyrrell, Samuel E! %ones,
&arcus '! Tu(e, Hanieh )aghoot(ar, 'ndrew "!*ood,"achel &! Freathy, &ichael N!*eedon, Timothy &! Frayling, and
'nna &urray+
Genetics of Comple Traits, -niversity of Eeter &edical School, "I./ .evel 0, "oyal
/evon and Eeter Hospital,
$arrac( "oad, Eeter E12 3/*, -
+Correspondence address! E4mail5 a!murray6eeter!ac!u(
Su#mitted on Septem#er 78, 2973: resu#mitted on ;cto#er 2
summaryanswer5 The T allele of the FSH$ promoter polymorphism ?rs79@03A0@: c!4277G!TBresults in longer menstrual cycles and
later menopause and, while having detrimental effects on fertility, is protective against
endometriosis!
what is (nown already5 The FSH$ promoter polymorphism ?rs79@03A0@: c!4277G!TB affects
levels of FSH$ transcription and, as a
result, circulating levels of FSH! FSH is re=uired for normal fertility and genetic variants at
the FSH$ locus are associated with age at menopause and
polycystic ovary syndrome ?C;SB!
study design, siDe, duration5 *e used cross4sectional data from the - $io#an( to loo( at
associations #etween the FSH$
promoter polymorphism and reproductive traits, and performed a genome4wide association
study ?G*'SB for length of menstrual cycle!
participantsmaterials, setting, methods5 *e included white $ritish individuals aged 9A8
years in 299A2979, in the
&ay 2973 release of genetic data from - $io#an(! *e tested the FSH4lowering T allele of
the FSH$ promoter polymorphism ?rs79@03A0@:
c!4277G!TB for associations with 28, mainly female, reproductive phenotypes in up to A0 039
women and 3A A9@ men! *e conducted a
G*'S in 830 individuals to identify genetic variants associated with length of menstrual
cycle!
main results and the role of chance5 The FSH4lowering T allele of the FSH$ promoterpolymorphism ?rs79@03A0@: &'F
9!7AB was associated with longer menstrual cycles 9!7A S/ ?c! 7 dayB per minor allele: 83
confidence interval ?CIB 9!729!29:
J A K 7927AL, later age at menopause ?9!70 years per minor allele: 83 CI 9!99!22: J
3!< K 7920B, greater female nulliparity odds
ratio ?;"B J 7!9A: 83 CI 7!927!77: J !@ K 7920L and lower ris( of endometriosis ?;"
J 9!
-
7/24/2019 Genetic Evidence That Lower Circulating
2/9
limitations, reasons for caution5 The data included might #e affected #y recall #ias! Cycle
length was not availa#le for 23 of
women still cycling ?7 did not answer, A did not (now and for 7@ cycle length was
recorded as MirregularB! *omen with a cycle length
recorded were aged over 9 and were approaching menopause: however, we did not find
evidence that this affected the results! &any of thegroups with illnesses had relatively small sample siDes and so the study may have #een under4
powered to detect an effect!
wider implications of the findings5*e found a strong novel association #etween a genetic
variant that lowers FSH levels and
longer menstrual cycles, at a locus previously ro#ustly associated with age at menopause! The
variant was also associated with nulliparity and
endometriosis ris(! These findings should now #e verified in a second independent group of
patients!*e conclude that lifetime differences in
circulating levels of FSH #etween individuals can influence menstrual cycle length and arange of reproductive outcomes, including menopause
timing, infertility, endometriosis and C;S!
study fundingcompeting interest?sB5None!
trial registration num#er5Not applica#le!
ey words5 FSH # su#unit menstrual cycle menopause endometriosis fertility
Introduction
FSH is a (ey pituitary hormone, which stimulates maturation of oocytes
and is a #iomar(er of ovarian reserve! FSH is a heterodimer comprised a
hormone4specific #4chain ?FSH4#B associated with an a4chain shared #y
other mem#ers of the glycoprotein hormone family ?NagirnaOa et al!,
2979B! The anterior pituitary produces FSH, with transcription of FSH$
#eing the rate4limiting step for FSH production! FSH stimulates target
cells #y #inding to the FSH receptor ?FSH"B, a G4protein4coupled receptor
?Fan and Hendric(son, 2993B, promoting follicle maturation and estrogen
production in women, and Sertoli cell proliferation and
spermatogenesis in men ?NagirnaOa et al!, 2979B!
"are mutations in the FSH$ gene cause truncation of the FSH4#
protein and result in hypogonadism and primary amenorrhoea in
females ?.ayman et al!, 788
-
7/24/2019 Genetic Evidence That Lower Circulating
3/9
homeodomain transcription factor #inding ?$enson et al!, 2970B! The
T allele of rs79@03A0@ ?c!4277G!TB is associated with lower FSH
levels in men and women, and with higher .H and lower testicular
volume, sperm count, FSH.H ratio, inhi#in $ and testosterone in
men, and has #een found at a higher prevalence in infertile men ?Grigorova
et al!, 299@, 2979, 2977: Tuttelmann et al!, 2972: .a &arca et al!,2970: Schuring et al!, 2970: Simoni and Casarini, 297: "uth et al!,
2973B! Genetic association studies have identified signals at the FSH$
locus associated with age at menopause ?Stol( et al!, 2972: /ay et al!,
2973B, polycystic ovary syndrome ?C;SB ?Hayes et al!, 2973B and
levels of .H ?Hayes et al!, 2973: "uth et al!, 2973B!
-sing the uni=ue resource of the - $io#an( ?'llen et al!, 297B,
we show that a common genetic variant (nown to alter FSH levels
impacts a wide range of traits important to female reproductive
health, including fertility, endometriosis and menstrual cycle length! In
the first genome4wide association study ?G*'SB for menstrual cycle
length, we identified the FSH$ locus as the only signal associatedwith this trait!
&aterials and ðods
Source of data
The - $io#an( includes data for 390 023 people aged 9A8 years recruited
in 299A2979 from across the - ?'llen et al!, 297B!*e analysed data from
the&ay 2973 interim release of imputed genetic data from- $io#an(,which
contains
-
7/24/2019 Genetic Evidence That Lower Circulating
4/9
loss ?!7 caseB!
*e defined two infertility4related #inary phenotypes: never pregnant
?femalesB and never fathered a child ?malesB! *e analysed female medical
conditions as #inary outcomes, comparing people reporting a condition
?caseB with those who did not ?controlB! &edical conditions included
dysmenorrhoea, endometriosis, fi#roids, irregular menstrual cycles, menopausalsymptoms, menorrhagia, ovarian cysts, C;S, uterine polyps,
vaginaluterine prolapse and #reast, endometrial and ovarian cancer! 's
more general indicators of gynaecological health, we included the medical
interventions #ilateral oophorectomy or hysterectomy in our analysis!
articipants
In our analysis, we included individuals who #oth self4identified as white
$ritish and were confirmed as ancestrally Caucasian #y - $io#an( from
genetic information ?n J 72@ 2AAB! *e calculated principal components
?CsB for inclusion as covariates in our analyses using FlashC' ?'#raham
and Inouye, 297B! Cs were calculated in 729 2@A unrelated participants
?as identified #y - $io#an(B #ased on 83 303 independent, directly genotyped
SNs ?pairwise r2 , 9!7B! These SNs had a minor allele fre=uency
?&'FB 2!3 and missing4ness ,7!3 across all participants in the &ay
2973 interim release of genetic data, and had a Hardy*ein#erg e=uili#rium
?H*EB ! 7 K 792A within the white $ritish participants!
Testing for associations of theFSH$promoter
polymorphism with reproductive phenotypes
*e tested the FSH4lowering T allele of the FSH$ promoter polymorphism
?rs79@03A0@: c!4277G!TB for associations with reproductive phenotypes
?up to A0 039 women and 3A A9@ menB! SN rs79@03A0@ was wellimputed in the data ?imputation =uality 9!883: H*E J 9!7A: missing
rate J 9!0B! 'll analyses were carried out in males or females as appropriate
?#ased on self4defined seB using Stata ?v70B ?StataCorp ., College
Station, T1, -S'B!
For continuous phenotypes, we transformed the phenotype #y adOusting
for recruitment centre, age at recruitment and the first five Cs prior
to inverse4normaliDation! *e performed linear regression of transformed
phenotype on imputed minor4allele dosages at SN rs79@03A0@ with genotyping
chip as a covariate!*e carried out a sensitivity analysis of the effect of
different transformations, e!g! inverse normaliDing the trait prior to calculating
the residuals:however, this did not materially affect our results! Since the dataon length of menstrual cycle included a wide range of values ?Supplementary
data, Figs S7 and S2B,wecarried out analyses on cycles from27 to 03 days and
in women aged ,3 and 3 years at recruitment!*e validated our results
for length of menstrual cycle #y carrying out analyses in two randomly chosen,
e=ually siDed groups! For age at menopause and age at menarche,we also ran
analysis using the phenotype definition from the "eproGen Consortium
G*'S ?www!reprogen!orgB ?untransformed age at menopause #etween
9 and A9 years not adOusted for age, untransformed age at menarcheB to
allow comparisons with pu#lished data ?Stol( et al!, 2972: erry et al!,
297a,#: /ay et al!, 2973B!
For #inary outcomes, we performed logistic regression of the phenotypeon minor4allele dosages at SN rs79@03A0@ including the first five Cs, recruitment
-
7/24/2019 Genetic Evidence That Lower Circulating
5/9
centre, age at recruitment and genotyping chip as covariates!
G*'S of length of menstrual cycle
*e conducted aG*'S to identify genetic variants associated with length of
menstrual cycle ?n J 830B using the $;.T4.&& algorithm ?descri#ed in
.oh et al!, 2973B from the freely availa#le $;.T4.&& software pac(age
version 2!2, https5data!#roadinstitute!orgal(esgroup$;.T4.&&?7< /ecem#er 2973, date last accessedBL to account for relatedness and
population structure! This allowed us to include related individuals who
were ecluded from the association analysis of the FSH$ promoter polymorphism
?Supplementary data, Ta#le SIB!*e transformed length of menstrual
cycle #y adOusting for recruitment centre and age at recruitment
prior to inverse4normaliDation, and performed association testing while
adOusting for genotype chip! *e filtered results on imputation =uality !9!,
H*E ! 7 K 7923, and &'F !9!7, resulting in P7A!@ million variants
that were tested! 's the - $io#an( G*'S included more variants than a
standard G*'S and we did not have a replication sample availa#le, we
chose a threshold of , 3 K 7928, #ased on a $onferroni correction forthe num#er of variants tested, rather than the conventional , 3 K 792@!
"esults
' common allele in the FSH$ gene, (nown to
lower FSH levels, is associated with longer
length of menstrual cycle
The FSH4lowering T allele of the FSH$ promoter polymorphism
?rs79@03A0@: &'F 9!7AB was associated with longer menstrual cycles
9!7A S/ ?P7 dayB per minor allele: 83 confidence interval ?CIB
9!729!29: J A K 7927AL! ;f the reproductive traits tested ?Ta#les I
and IIB, length of menstrual cycle was the most strongly associated with
rs79@03A0@ ?Fig! 7 and Ta#le IIIB! The SN was also associated with
cycle length when we dichotomiDed data into women reporting a cycle
length of R29 days compared with those reporting an average length
of 2@ days odds ratio ?;"B J 9!
-
7/24/2019 Genetic Evidence That Lower Circulating
6/9
per minor allele: 83 CI 9!939!77: J 7!8 K 792@B!
The FSH$ allele associated with longer
cycle length is associated with later
menopause
The FSH4lowering Tallele of rs79@03A0@ was associated with later age at
menopause for those in the - $io#an( 9!70 years per minor allele?"eproGen definitionB: 83 CI 9!99!22: J 3!< K 7920L! There was
no association #etween rs79@03A0@ and menopause age when we
dichotomiDed the phenotype into early menopause compared with
later menopause ?Ta#le IIIB! The FSH$ locus is (nown to #e associated
with timing of menopause5 in aG*'S conducted #y the "eproGen consortium,
the signal at this locus ?rs722879B increases age at menopause
#y 9!20 years ?83 CI 9!7A9!28: J 7!3 K 79277B ?Stol(
et al!, 2972B! .ater menopause has #een shown to #e associated with
later age at last #irth ?'yatollahi et al!, 2990: /orOgochoo et al!, 299@B
and rs79@03A0@ was also associated with later age at last #irth 9!92
S/ ?P9!7 yearsB per T allele: 83 CI 9!999!9: J !2 K 7922L!.onger cycle length is not a general feature of
alleles associated with later age at menopause
*e net tested the role of all 3A genetic variants associated with age
at menopause! In addition to the age at menopause signal at the FSH$
locus ?rs722879B, only one of the other 33 pu#lished age at menopause
signals was nominally associated with cycle length ? ! 9!93B5
rs79
-
7/24/2019 Genetic Evidence That Lower Circulating
7/9
having the medical interventions #ilateral oophorectomy ?;" J 7!72:
83 CI 7!9A7!78: J 7! K 792B and hysterectomy ?;" J 7!70:
83 CI 7!9A7!29: J 7!9 K 792B, which are used as treatments for
a range of gynaecological conditions including endometriosis!
The common FSH$ variant, associated with
FSH levels, is not associated with reproductivetraits more generally
There was no consistent evidence that the FSH$ variant ?rs79@03A0@B
was associated with age at menarche! There was a 9!904year increase
in age at menarche ?"eproGen definitionB per T allele of rs79@03A0@
?83 CI 9!979!93: J 7! K 7922B and the #inary phenotype of
early menarchewas associated at ! 9!93 ?Ta#le IIIB! None of 722 pu#lished
G*'S signals for menarche ?erry et al!, 297a,#B were associated
with length of menstrual cycle at , 9!99@!
The FSH$ promoter polymorphism ?rs79@03A0@B was not associated
with other reproductive illnesses or conditions at , 9!93 ?Ta#le IIIB,
ecept for menopausal symptoms ?;" J 9!A2: 83 CI 9!79!80: J 9!92B ?Fig! 7B! No associations were found with dysmenorrhoea,
fi#roids, irregular menstrual cycles, menorrhagia, multiple pregnancy
loss, ovarian cysts, C;S, uterine polyps or vaginaluterine prolapse,
or with female #reast, ovarian or endometrial cancer!
/iscussion
In the first G*'S of menstrual cycle length, we found a strong association
#etween an FSH lowering, li(ely functional, variant in the FSH$ promoter
and longer cycles ?Hoogendoorn et al!, 2990: Grigorova et al!,
299@, 2979: Tuttelmann et al!, 2972: $enson et al!, 2970: .a &arca
et al!, 2970: Simoni and Casarini, 297: "uth et al!, 2973B! This locus
has #een previously ro#ustly associated with age at menopause in the
"eproGen consortium G*'S of menopause timing ?Stol( et al!, 2972:
/ay et al!, 2973B and the allele associated with longer cycle length is
also associated with later age at menopause!*e did not o#serve associations
for the maOority of age at menopauseG*'S signals with length
of menstrual cycle, including the four signals with effects of over one4third
of a year per allele on menopause timing, implying that the association is
specific to FSH$5 either FSH4# has independent effects on #oth cycle
length and menopause or changes in cycle length are causally influencing
menopause timing!;ur results are consistent with the o#served epidemiological relationship
#etween longer menstrual cycles and later age at menopause
?*helan et al!, 7889: acDmare(, 299
-
7/24/2019 Genetic Evidence That Lower Circulating
8/9
pill influences menopause timing ?van Noord et al!, 788
-
7/24/2019 Genetic Evidence That Lower Circulating
9/9