Resea ch pape
Associa ion o ma e nal p ena al smoking GFI1-locus and ca dio-
me abolic pheno ypes in 18,212 adul s
P iyanka Pa ma
a,b
,Es elleLow y
a,b
, Gio anni Cuglia i
c,d
,Ma hewSude man
e
, Ro y Wilson
,g
,
Ville Ka hunen
h
,TobyAnd ew
i
, Pe i Wiklund
a,h,j
, Ma hias Wielsche
h
, Simone a Gua e a
c,d
,
Alexande Teume
k,l
,BenjaminLehne
h
,LiliMilani
m,n
,NiekdeKlein
o
, Pashupa i P. Mish a
p,q
,
Phillip E. Mel on
,s
, Pooja R. Manda iya
, Sil a Kasela
m
,JanaNano
g,u
, Weihua Zhang
h,
, Yan Zhang
w
,
And e G. Ui e linden
,u
, Anne e Pe e s
,g,x
, Ben Schö ke
w,y
, Ch is ian Giege
,g,x
, Denise Ande son
z
,
Do e I. Boomsma
aa
,HansJ.G abe
ab,ac
, Sal a o e Panico
ad
, Jan H. Veldink
ae
, Joyce B.J. an Meu s
,
Leona d an den Be g
ae
, Law ence J. Beilin
a
, Lude F anke
o
,Ma ieLoh
h,ag,ah
, Ma leen M.J. an G ee enb oek
ai
,
Ma hias Nauck
l,aj
,MikaKähönen
ak,al
, Mikko A. Hu me
am
, Olli T. Rai aka i
an,ao
,Osca H.F anco
u
,
P.Eline Slagboom
ap
, Pim an de Ha s
o,aq,a
,SonjaKunze
,g
, S ephan B. Felix
l
, Tao Zhang
as,a
,WeiChen
as
,
T e o A. Mo i
a
, Amelie Bonne ond
i,au
, Bas iaan T. Heijmans
ap
, o he BIOS Conso ium, Taulan Muka
u
,
Jaspal S. Koone
,aw,h,ay
, K is a Fische
m
, Melanie Waldenbe ge
,g,x
, Philippe F oguel
i,au
, Rae-Chi Huang
z
,
Te ho Leh imäki
p,q
, Wol gang Ra hmann
ax
,Ca olineL.Rel on
e
,GiuseppeMa ullo
c,d
, He mann B enne
w,y
,
Niek Ve weij
aq
, Shengxu Li
ay
, John C. Chambe s
h, ,a ,az
, Ma jo-Rii a Jä elin
a,b,h,ba,
⁎⁎
,1
,
Syl ain Sebe
a,b,bb,
⁎
,1
, o he GLOBAL Me h QTL
a
Cen e o Li e Cou se Heal h Resea ch, Uni e si y o Oulu, Oulu, Finland
b
Biocen e Oulu, Uni e si y o Oulu, Oulu, Finland
c
Depa men o Medical Sciences, Uni e si y o Tu in, Tu in, I aly
d
I alian Ins i u e o Genomic Medicine, IIGM, Tu in, I aly
e
MRC In eg a i e Epidemiology Uni , Popula ion Heal h Sciences, B is ol Medical School, Uni e si y o B is ol, UK
Resea ch Uni o Molecula Epidemiology, Helmhol z Zen um München, Ge man Resea ch Cen e o En i onmen al Heal h, Neuhe be g, Ba a ia, Ge many
g
Helmhol z Zen um München, Ge man Resea ch Cen e o En i onmen al Heal h, Ins i u e o Epidemiology, Neuhe be g, Ba a ia, Ge many
h
Depa men o Epidemiology and Bios a is ics, MRC-PHE Cen e o En i onmen and Heal h, School o Public Heal h, Impe ial College London, London
i
Genomics o Common Disease, Depa men o Medicine, Impe ial College London, London, UK
j
Depa men o Heal h Sciences, Uni e si y o Jy äskylä, Jy äskylä, Finland
k
Depa men o In e nal Medicine B, Uni e si y Medicine G ei swald, G ei swald, Ge many
l
Pa ne Si e G ei swald, DZHK (Ge man Cen e o Ca dio ascula Resea ch), G ei swald, Ge many
m
Es onian Genome Cen e, Ins i u e o Genomics, Uni e si y o Ta u, Ta u, Es onia
n
Science o Li e Labo a o y, Depa men o Medical Sciences, Uppsala Uni e si y, Sweden
o
Depa men o Gene ics, Uni e si y Medical Cen e G oningen, Uni e si y o G oningen, G oningen, The Ne he lands
p
Depa men o Clinical Chemis y, Fimlab Labo a o ies, Tampe e, Finland
q
Depa men o Clinical Chemis y, Finnish Ca dio ascula Resea ch Cen e - Tampe e, Facul y o Medicine and Li e Sciences, Uni e si y o Tampe e, Tampe e, Finland
School o Pha macy and Biomedical Sciences, Cu in Uni e si y, Ben ley, Aus alia
s
Cu in UWA Cen e o Gene ic O igins o Heal h and Disease, School o Biomedical Sciences, The Uni e si y o Wes e n Aus alia, C awley, Aus alia
Depa men o In e nal Medicine, E asmus Uni e si y Medical Cen e, Ro e dam, The Ne he lands
u
Depa men o Epidemiology, E asmus Uni e si y Medical Cen e, Ro e dam, The Ne he lands
Depa men o Ca diology, Ealing Hospi al, No h Wes Heal hca e NHS T us , London, UK
w
Di ision o Clinical Epidemiology and Aging Resea ch, Ge man Cance Resea ch Cen e (DKFZ), Im Neuenheime Feld, Heidelbe g, Ge many
x
Ge man Cen e o Ca dio ascula Resea ch (DZHK), Pa ne Si e Munich Hea Alliance, Munich, Ge many
y
Ne wo k Aging Resea ch, Uni e si y o Heidelbe g, Be gheime S aße, Heidelbe g, Ge many
z
Tele hon Kids Ins i u e, Uni e si y o Wes e n Aus alia, Pe h, Aus alia
aa
Depa men o Biological Psychology, School o Public Heal h, V ije Uni e si ei Ams e dam, Ams e dam, The Ne he lands
ab
Depa men o Psychia y and Psycho he apy, Uni e si y Medicine G ei swald, G ei swald, Ge many
ac
Ge man Cen e o Neu odegene a i e Diseases DZNE, Si e Ros ock/G ei swald, G ei swald, Ge many
ad
Depa men o Clinical Medicine and Su ge y, Fede ico II Uni e si y, Naples, I aly
ae
Depa men o Neu ology, B ain Cen e Rudol Magnus, Uni e si y Medical Cen e U ech , U ech , The Ne he lands
EBioMedicine 38 (2018) 206–216
⁎Co espondence o: Syl ain Sebe , Cen e o Li e Cou se Heal h Resea ch, Facul y o Medicine, Uni e si y o Oulu, P.O. Box 5000, Oulu 90014, Finland.
⁎⁎ Co espondence o: Ma jo-Rii a Jä elin, Depa men o Epidemiology and Bios a is ics, Facul y o Medicine, Impe ial College London, S . Ma y's campus, UK.
E-mail add esses: m.ja elin@impe ial.ac.uk (M.-R. Jä elin), syl ain.sebe @oulu.fi(S. Sebe ).
1
Equal con ibu ions.
h ps://doi.o g/10.1016/j.ebiom.2018.10.066
2352-3964/© 2018 The Au ho s. Published by Else ie B.V. This is an open access a icle unde he CC BY-NC-ND license (h p://c ea i ecommons.o g/licenses/by-nc-nd/4.0/).
Con en s lis s a ailable a ScienceDi ec
EBioMedicine
jou nal homepage: www.ebiomedicine.com
a
Medical School, Uni e si y o Wes e n Aus alia, Pe h, Aus alia
ag
T ansla ional Labo a o y in Gene ic Medicine (TLGM), Agency o Science, Technology and Resea ch (A*STAR), 8A Biomedical G o e, Immunos, Le el 5, Singapo e, Singapo e
ah
Ins i u e o Heal h Sciences, Uni e si y o Oulu, Finland
ai
Depa men o In e nal Medicine and School o Ca dio ascula Diseases (CARIM), Maas ich Uni e si y Medical Cen e, Maas ich , The Ne he lands
aj
Ins i u e o Clinical Chemis y and Labo a o y Medicine, Uni e si y Medicine G ei swald, G ei swald, Ge many
ak
Depa men o Clinical Physiology, Tampe e Uni e si y Hospi al, Tampe e, Finland
al
Depa men o Clinical Physiology, Finnish Ca dio ascula Resea ch Cen e - Tampe e, Facul y o Medicine and Li e Sciences, Uni e si y o Tampe e,Tampe e,Finland
am
Depa men o Mic obiology and Immunology, Facul y o Medicine and Li e Sciences, Uni e si y o Tampe e, Tampe e, Finland
an
Depa men o Clinical Physiology and Nuclea Medicine, Tu ku Uni e si y Hospi al, Tu ku, Finland
ao
Resea ch Cen e o Applied and P e en i e Ca dio ascula Medicine, Uni e si y o Tu ku, Tu ku, Finland
ap
Molecula Epidemiology, Depa men o Biomedical Da a Sciences, Leiden Uni e si y Medical Cen e, Leiden, The Ne he lands
aq
Depa men o Ca diology, Uni e si y Medical Cen e G oningen, Uni e si y o G oningen, G oningen, The Ne he lands
a
Du e Cen e o Ca diogene ic Resea ch, ICIN - Ne he lands Hea Ins i u e, U ech , The Ne he lands
as
Depa men o Epidemiology, School o Public Heal h and T opical Medicine, Tulane Uni e si y, New O leans, USA
a
Depa men o Bios a is ics, School o Public Heal h, Shandong Uni e si y, Jinan, China
au
Eu opean Genomic Ins i u e o Diabe es (EGID), Ins i u Pas eu de Lille, Uni e si y o Lille, CNRS UMR 8199, Lille, F ance
a
Impe ial College Heal hca e NHS T us , London, UK
aw
Impe ial College London, Na ional Hea and Lung Ins i u e, London, UK
ax
Ins i u e o Biome ics and Epidemiology, Ge man Diabe es Cen e, Leibniz Cen e o Diabe es Resea ch a Hein ich, Heine Uni e si y, Düsseldo , Ge many
ay
Child en's Hospi als and Clinics o Minneso a, Child en's Minneso a Resea ch Ins i u e, Minneapolis, MN 55404, USA
az
Lee Kong Chian School o Medicine, Nanyang Technological Uni e si y, Singapo e, Singapo e
ba
Depa men o Li e Sciences, College o Heal h and Li e Sciences, B unel Uni e si y London, Uxb idge, UK
bb
Medical Resea ch Cen e (MRC) Oulu, Oulu Uni e si y Hospi al, Uni e si y o Oulu, Oulu, Finland
abs ac a icle in o
A icle his o y:
Recei ed 22 Augus 2018
Recei ed in e ised o m 26 Oc obe 2018
Accep ed 26 Oc obe 2018
A ailable online 13 No embe 2018
Backg ound: DNA me hyla ion a he GFI1-locus has been epea edly associa ed wi h exposu e o smoking om
he oe al pe iod onwa ds. We explo ed whe he DNA me hyla ion may be a mechanism ha links exposu e o
ma e nal p ena al smoking wi h o sp ing's adul ca dio-me abolic heal h.
Me hods: We me a-analysed he associa ion be ween DNA me hyla ion a GFI1-locus wi h ma e nal p ena al
smoking, adul own smoking, and ca dio-me abolic pheno ypes in 22 popula ion-based s udies om Eu ope,
Aus alia, and USA (n= 18,212). DNA me hyla ion a he GFI1-locus was measu ed in whole-blood. Mul i a i-
able eg ession models we e fi ed o examine i s associa ion wi h exposu e o p ena al and own adul smoking.
DNA me hyla ion le els we e analysed in ela ion o body mass index (BMI), wais ci cum e ence (WC), as ing
glucose (FG), high-densi y lipop o ein choles e ol (HDL—C), iglyce ides (TG), dias olic, and sys olic blood p es-
su e (BP).
Findings: Lowe DNA me hyla ion a h ee ou o eigh GFI1-CpGs was associa ed wi h exposu e o ma e nal p e-
na al smoking, whe eas, all eigh CpGs we e associa ed wi h adul own smoking. Lowe DNA me hyla ion a
cg14179389, he s onges ma e nal p ena al smoking locus, was associa ed wi h inc eased WC and BP when ad-
jus ed o sex, age, and adul smoking wi h Bon e oni-co ec ed Pb0·012. In con as , lowe DNA me hyla ion
a cg09935388, he s onges adul own smoking locus, was associa ed wi h dec eased BMI, WC, and BP (adjus ed
1×10
−7
bPb0.01). Simila ly, lowe DNA me hyla ion a cg12876356, cg18316974, cg09662411, and
cg18146737 was associa ed wi h dec eased BMI and WC (5 × 10
−8
bPb0.001). Lowe DNA me hyla ion a all
he CpGs was consis en ly associa ed wi h highe TG le els.
In e p e a ion: Epigene ic changes a he GFI1 we e linked o smoking exposu e in-u e o/in-adul hood and o-
bus ly associa ed wi h ca dio-me abolic isk ac o s.
Fund: Eu opean Union's Ho izon 2020 esea ch and inno a ion p og amme unde g an ag eemen no. 633595
DynaHEALTH.
© 2018 The Au ho s. Published by Else ie B.V. This is an open access a icle unde he CC BY-NC-ND license
(h p://c ea i ecommons.o g/licenses/by-nc-nd/4.0/).
1. In oduc ion
Ciga e e smoking, including second-hand exposu e, is es ima ed o
accoun o nea ly 6 million dea hs annually [1]. Fi s and second-hand
exposu es a e widely ecognized as independen isk ac o s o ca dio-
ascula diseases (CVD), la gely de e mined by dose and du a ion [1,2].
P oposed di ec mechanisms linking ciga e e smoking and CVD include
inc eased hea a e and myoca dial con ac ili y, inflamma ion, insulin
esis ance, and oxida i e s ess [3,4]. Mo eo e , he isk may emain
e en a e success ul long- e m smoking cessa ion [5]. Simila ly, ma e -
nal p ena al smoking has implica ions o bi h ou comes, including low
bi h weigh and isk o p e e m bi h [6], as well as inc eased isk o he
o sp ing's la e ca dio-me abolic heal h [7,8]. A ecen global e iew e-
po ed he highes es ima ed p e alence o ma e nal p ena al smoking
in Eu ope, despi e he widely known isks [9].
Eme ging esea ch sugges s ha pa o he downs eam impac o
smoking likely pe sis s h ough al e ed epigene ic pa e ns, many o
which ha e been associa ed wi h al e a ions in he gene exp ession
[10]. O pa icula impo ance, al e ed DNA me hyla ion a AHHR,GFI1,
and MYO1G genes is consis en ly obse ed among bo h adul smoke s
and new-bo ns exposed o ma e nal p ena al smoking [10–13]. E i-
dence on he s abili y o smoking- ela ed-loci DNA me hyla ion o e
he li e ime is inconsis en . Many CpGs in o me smoke s show a e e -
sal o dis up ed DNA me hyla ion equi alen o non-smoke s wi hin
fi e yea s o cessa ion, whe eas o he s show no e e sibili y e en 20–
30 yea s a e cessa ion [10,14]. Simila ly, Richmond e al. sugges ed
he e we e bo h e e sible and pe manen changes a smoking- ela ed
DNA me hyla ion loci in o sp ing exposed o ma e nal smoking du ing
p egnancy [15].
Fu he mo e, eigh GFI1-linked-CpGs wi h abe an DNA me hyla-
ion we e epo ed o pa ially media e he associa ion o ma e nal p e-
na al smoking wi h bi hweigh .
16
Conside ing he consis en
associa ion obse ed be ween low bi h weigh and ad e se adul ca -
dio-me abolic heal h [8], we aimed o pu sue a li e-cou se app oach o
207P. Pa ma e al. / EBioMedicine 38 (2018) 206–216
e alua e he possibili y ha exposu e o ma e nal smoking in p eg-
nancy influences he heal h o o sp ing ia epigene ic mechanisms.
We hypo hesized ha DNA me hyla ion changes a GFI1-CpGs, a
po en ial smoking bioma ke , pe sis h oughou he li e-cou se and as-
socia e wi h ca dio-me abolic pheno ypes in adul s. We es ed his hy-
po hesis in a la ge me a-analysis in ol ing 22 popula ion-based s udies.
2. Ma e ial and me hods
2.1. Pa icipa ing s udies
We included 22 s udies consis ing o 18,212 pa icipan s, including
fi e p egnancy-bi h coho s, 17 o he popula ion-based da ase s and
hei sub-s udies: he A on Longi udinal S udy o Pa en s and Child en
(ALSPAC) (specifically subse wi h DNA me hyla ion p ofiles in he Ac-
cessible Resou ce o In eg a ed Epigenomic S udies), wo s udies om
he Bogalusa Hea S udy (BHS – he Eu opean-Ame ican and A ican-
Ame ican coho s), he BIOS conso ium, he Es onian Genome Cen e
Uni e si y o Ta u (EGCUT), he Eu opean P ospec i e In es iga ion
in o Cance and Nu i ion (EPIC), he I alian Ca dio ascula sec ion
(EPICOR), wo independen subse s o he ESTHER s udy, he Coope a-
i e Heal h Resea ch in he Augsbu g Region F4 (KORAF4), he Li elines
Deep (LLD), he London Li e Science Popula ion s udy (LOLIPOP), wo
ollow-up da ase s om he No he n Finland Bi h coho 1966
(NFBC1966 - 31 yea s and NFBC1966 - 46 yea s) and No he n Finland
Bi h coho 1986 (NFBC1986), he Wes e n Aus alian P egnancy Co-
ho (RAINE) s udy, wo independen s udies om he Ro e dam
S udy (RS) –RSIII-1 and RSII-3_III-2, he S udy o Heal h In Pome ania
–T end (SHIP-T end), and he Young Finns S udy 2011 (YFS). The
BIOS conso ium ep esen s ou s udies wi h coo dina ed DNA me hyl-
a ion measu emen s: he Coho On Diabe es And A he oscle osis
Maas ich (CODAM), he Leiden Longe i y S udy (LLS), he Ne he -
lands Twin Regis e S udy (NTR) and he p ospec i e Amyo ophic La -
e al Scle osis (ALS) s udy, he Ne he lands (PAN). Among hese, fi e
s udies (ALSPAC, NFBC1966–31 y , NFBC1966–46 y , NFBC1986, and
RAINE) pa icipa ed in he me a-analysis o associa ions be ween he
eigh GFI1-CpGs and ma e nal p ena al smoking (n= 4230). De ailed
da a collec ion and e hical app o al o each s udy a e desc ibed in sup-
plemen a y me hods in he Appendix A. Subjec s wi h missing in o ma-
ion on DNA me hyla ion and mul iple bi hs we e excluded.
2.2. Smoking
Ma e nal p ena al smoking and o sp ing's own adul smoking we e
sel - epo ed. Ques ions we e ha monized o de i e a dicho omous a -
iable o ma e nal smoking as ‘no ma e nal smoking’and ‘any ma e nal
smoking’du ing p egnancy. Adul own smoking was ca ego ized as cu -
en non-smoke s and smoke s (adul own smoking ≥one ciga e e/
day).
2.3. DNA me hyla ion measu emen and quali y con ol
We used eigh GFI1-linked-CpGs: cg04535902, cg09662411,
cg09935388, cg10399789, cg12876356, cg18146737, cg14179389,
and cg18316974. Each s udy conduc ed DNA me hyla ion measu e-
men s and quali y con ol. DNA me hyla ion was measu ed in
pe iphe al whole blood by s anda d p ocedu es o Illumina
HumanMe hyla ion450 o EPIC a ay. DNA Me hyla ion is desc ibed as
β- alue anging be ween 0 (no cy osine me hyla ion) and 1 (comple e
cy osine me hyla ion). Each s udy excluded ailed samples based on de-
ec ion P- alues, CpG-specific pe cen age, low DNA concen a ion,
bisulphi e con e sion e ficiency, and o he s udy-specific con ol me -
ics (Appendix A) [17].
2.4. Co a ia es
Co a ia es we e age, sex, and echnical co a ia es o CpGs (ba che -
ec s, con ol p obe adjus men s, and cell ype p opo ions).
Resea ch in con ex
E idence be o e his s udy
Ma e nal p ena al smoking is associa ed wi h un a ou able bi h
ou comes and has implica ions on o sp ing ca dio-me abolic
heal h in la e li e. Despi e he widely known isk, a ecen global
sys ema ic e iew epo ed ha 52.9% o women, who smoke
daily, con inue smoking du ing p egnancy, wi h he highes p e -
alence in he Eu opean egion and mos s able o e he yea s. As
o ye , he mechanism unde lying smoking associa ed ad e se
ca dio-me abolic heal h ou comes emains poo ly unde s ood
and is sugges i e o he associa ed DNA me hyla ion changes.
We sea ched PubMed o a icles on ma e nal smoking associa ed
DNA me hyla ion changes in o sp ing using he sea ch e ms ‘ma-
e nal smoking’,‘p egnancy’,‘DNA me hyla ion’,‘epigene ic
ma ke s’,and‘o sp ing’ o wo k published un il Decembe
2017. We no ed ha p e ious s udies ha e iden i ied many epige-
ne ic ma ke s, especially DNA me hyla ion changes in he o -
sp ing exposed o ma e nal p ena al smoking, bu no s udy has
in es iga ed he po en ial unde lying ole o hese epigene ic
ma ke s on long- e m heal h. One s udy iden i ied he media ing
ole o ma e nal smoking ela ed DNA me hyla ion changes a
he GFI1 in he associa ion be ween ma e nal p ena al smoking
and low bi h weigh , and hus ou esea ch aimed o go beyond
he di ec e ec o ma e nal smoking on bi h weigh .
Added alue o his s udy
To he bes o ou knowledge, ou s udy is one o he la ges me a-
analysis conduc ed and includes 22 s udies om Eu ope, US and
Aus alia o subs an ia e he deba e o he epigene ic pa hways
o li e-long heal h. Ou esea ch is key o unde s anding he
causal, molecula pa hways associa ed wi h such consis en ly ob-
se ed DNA me hyla ion pa e ns, and how hey may media e he
associa ion be ween smoking and clinical isks a ibu ed o
smoking. In he p esen esea ch, we b ing e idence o asce ain-
ing he clinical ele ance o hese indings in he eme ging ield o
epigenomics. Epigenomics helps o unde s and why he isk o
diseases, in ou case ch onic ca dio-me abolic diseases, may
exis e en in he absence o a di ec exposu e. We uniquely iden-
i ied lowe DNA me hyla ion a cg14179389, a s ong ma e nal
smoking locus as a isk ac o o adul adiposi y, highe iglyce -
ides le els, and blood p essu e. The s udy deli e s s ong e i-
dence o suppo he concep o he ea ly li e epigene ic
in luence on adul heal h.
Implica ions o all he a ailable e idence
We epo no el indings on he ma e nal smoking- ela ed epige-
ne ic ac o s a he GFI1-locus linking i o ca dio-me abolic heal h
in he adul . The indings ma ched known ca dio-me abolic dis-
eases isk a ibu ed o ma e nal smoking exposu e o adul
smoking, suppo ing an unde lying epigene ic componen ha
can help bio-ma king exposu e o pas isk. These indings p o ide
a s ong ounda ion o u he wo k o un a el eme ging smoking
epigene ic ma ke s wi h downs eam de imen al heal h ou -
comes and u he d aws a en ion o inc ease awa eness on
smoking cessa ion and be e p e en ion s a egies.
208 P. Pa ma e al. / EBioMedicine 38 (2018) 206–216
Adjus men s o echnical a ia ion and cell ype p opo ion in each
s udy a e desc ibed in he Appendix A.
2.5. Ca dio-me abolic pheno ypes
We used se en ca dio-me abolic pheno ypes de i ed om clinical
examina ions: body mass index (BMI, weigh (kg)/heigh (m)
2
), wais
ci cum e ence (WC), high-densi y lipop o ein choles e ol (HDL-C), i-
glyce ides (TG), as ing glucose (FG), dias olic blood p essu e (DBP),
and sys olic blood p essu e (SBP). All ca dio-me abolic pheno ypes
we e used as con inuous a iables and s anda dized (mean = 0, s an-
da d de ia ion = 1). Co ec ion cons an s we e applied o HDL-C, TG,
and BP alues, i pa icipan epo ed lipid o blood p essu e medica ion
use (Appendix A). Acco ding o a ailabili y in he pa icipa ing s udies,
BMI was a ailable in 18,212, WC in 14,665, HDL-C in 18,212, TG in
18,212, FG in 16,529, DBP in 16,529, and SBP in 16,529 indi iduals o
he o al.
2.6. S udy-specific s a is ical analyses
Each s udy conduc ed s a is ical analyses acco ding o he analysis
plan. F equencies and means we e compu ed o desc ip i e pu poses.
We used mul i a ia e eg ession o e alua e h ee se s o associa ions:
GFI1-CpGs wi h (i) ma e nal p ena al smoking (n= 4230), (ii) adul
own smoking (n= 13,551), and (iii) ca dio-me abolic pheno ypes (n
= 18,212) (Appendix B Fig. S1). Fi s ly, analyses in fi e p egnancy-
bi h coho s udies we e pe o med using: baseline model, which
used any ma e nal smoking du ing p egnancy as an exposu e plus ech-
nical co a ia es eg essed on DNA me hyla ion as an ou come (be a-
alues), and adjus ed model wi h sex, age (whe e applicable) and
adul smoking as co a ia es. To assess he impac o adul own smoking
on DNA me hyla ion le el, we included: baseline model, which used
adul own smoking as an exposu e plus echnical co a ia es, and DNA
me hyla ion as an ou come, and adjus ed model including sex and age
as co a ia es. These wo analyses we e assessed in 20 pa icipa ing
s udies. Bo h ma e nal and adul smoking showed lowe DNA me hyla-
ion a GFI1-CpGs, and hus we assessed ca dio-me abolic pheno ypes
wi h espec o isk o lowe DNA me hyla ion. In he final analyses, co-
a ia e-adjus ed models we e pe o med in all pa icipa ing s udies
wi h: baseline model, using DNA me hyla ion as an exposu e plus ech-
nical co a ia es, and each ca dio-me abolic pheno ype as an ou come,
and adjus ed model, including sex, age and adul smoking as addi ional
co a ia es.
2.7. Me a-analysis
We used METAL so wa e o conduc in e se a iance-weigh ed
fixed e ec s me a-analysis. We assessed he e ogenei y using he I
2
s a-
is ic (low-he e ogenei y = I
2
b50%). S a is ical significance was de-
fined by Bon e oni co ec ion o mul iple es ing as 0·05/4 (P≤
0·012), accoun ing o ou clus e s o ca dio-me abolic pheno ypes.
2.8. Supplemen a y analyses
In he NFBC1966 and 1986, we also examined he co ela ion be-
ween eigh GFI1-CpGs. In a condi ional analysis, we assessed associa-
ion be ween adul own smoking and GFI1-CpGs addi ionally adjus ed
o all o he GFI1-CpGs. Fu he mo e, as he ull sample is mul i-e hnic,
he sensi i i y analysis was pe o med o in es iga e he associa ion be-
ween lowe DNA me hyla ion a eigh GFI1-CpGs and ca dio-me abolic
pheno ypes in a subse o Eu opean ances y. Addi ionally, we also
assessed he associa ion o he eigh GFI1-CpGs wi h o me and cu en
adul own smoking in NFBC1966 (Appendix B Tables S2, S5, S7 and S8).
3. Resul s
3.1. Pa icipan cha ac e is ics
Pa icipan s we e aged 16–81 yea s a he ime o ca dio-me abolic
pheno ype measu emen s, wi h he majo i y be ween 40 and 60
yea s. Among hese, 17% we e cu en smoke s (Table 1). 18% o he
pa icipan s we e exposed o ma e nal p ena al smoking in he fi e
s udies (Appendix B Table S1). All eigh GFI1-CpGs had lowe mean
DNA me hyla ion le els in he g oup exposed o ma e nal p ena al
smoking compa ed wi h unexposed g oup.
3.2. Co ela ion s uc u e o he GFI1-CpGs
Fig. 1 displays he co ela ion ma ix be ween heeigh s udied GFI1-
CpGs in ela ion o hei genomic loca ion. The analysis pe o med in he
NFBC1986 and NFBC1966 desc ibed a s ong co ela ion be ween se en
CpGs (cg04535902, cg09662411, cg09935388, cg10399789,
cg12876356, cg18146737, and cg18316974). In con as cg14179389
was weakly co ela ed wi h cg09935388 (0·35; Pb0·0001) only (Fig.
1and Appendix B Table S2).
3.3. GFI1-CpGs DNA me hyla ion and p ena al ma e nal smoking and
o sp ing's own smoking exposu es
Following me a-analysis om fi e s udies, he p ena al ma e nal
smoking exposu e s a us was associa ed wi h lowe DNA me hyla ion
a cg14179389 (P=6×10
−30
), cg09935388 (P=9×10
−11
), and
cg12876356 (P= 0·008) (Fig. 2, Appendix B Table S3). Cg14179389
was ound o be he s onges ma e nal smoking locus and he associa-
ion was no a enua ed when adjus ed o age, sex, and adul own
smoking (β=−0·03, P=2·0×10
−27
,I
2
= 19·3). Simila ly, adul
own smoking s a us was associa ed wi h lowe DNA me hyla ion a all
he s udied CpGs. Howe e , cg09935388 was ound o be he s onges
adul smoking locus (β=−0·07, P=4·4×10
−67
); he associa ion
being independen o o he CpGs in he condi ional analysis (Fig. 3,Ap-
pendix B Table S4 and S5). In con as , Cg14179389, he s onges
abo e-men ioned p ena al ma e nal smoking signal did no show asso-
cia ion wi h adul smoking s a us when condi ioned by he DNA me h-
yla ion a he o he se en GFI1-CpGs. In ac , o he eigh CpGs s udied,
only h ee o hem emained associa ed wi h adul smoking ollowing
condi ional analysis including cg09935388, cg18316974, and
cg18146737 (Pb0·001) (Appendix B Table S5).
Since smoking exposu es we e consis en ly nega i ely associa ed
wi h DNA me hyla ion a he GFI1-locus, we assessed he associa ions
o ca dio-me abolic pheno ypes agains lowe DNA me hyla ion, o be
consis en wi h he en i onmen al isk i sel i.e. inc ease in smoking.
3.4. Me a-analysis: eigh GFI1-CpGs wi h lowe DNA me hyla ion and ca -
dio-me abolic pheno ypes
The associa ions be ween GFI1-CpGs and ca dio-me abolic pheno-
ypes om he me a-analysis a e p esen ed in Fig. 4 and Appendix B
Table S6. Lowe DNA me hyla ion a cg14179389 was associa ed wi h
inc eased WC, TG, and BP a e a Bon e oni-co ec ion se a P ≤
0.012, wi h associa ions being enhanced wi h WC and BP when ad-
jus ed o sex, age, and adul own smoking (WC β=0·04;BPβ=
0·04; 0.0002 ≤P≤0.001). Cg14179389 consis en ly showed he lowes
he e ogenei y o he eigh CpGs (I
2
≤25·4). In con as , lowe DNA
me hyla ion a cg09935388 was associa ed wi h dec eased BMI, WC,
and BP, al hough simila ly o cg14179389 showed associa ion wi h in-
c eased TG. A e adjus men s, he associa ions emained showing mod-
e a e a enua ion wi h TG (BMI β=−0·06, WC β=−0·05; BP β=
−0·03, TG β=0·01;1×10
−7
≤P≤0.01). Lowe DNA me hyla ion a
cg12876356, cg18316974, and cg09662411 was associa ed wi h de-
c eased BMI, WC, BP and inc eased TG and a e adjus men s,
209P. Pa ma e al. / EBioMedicine 38 (2018) 206–216
associa ions wi h dec eased BMI and WC su i ed Bon e oni-co ec-
ion (5 × 10
−8
≤Pb9×10
−5
). Simila ly, lowe DNA me hyla ion a
cg18146737 was associa ed wi h dec eased BMI and WC and a
cg04535902, wi h dec eased BMI when adjus ed (1 × 10
−7
≤P≤
0.001). Lowe DNA me hyla ion a cg10399789 showed no associa ions
ollowing adjus men s (P ≥0.04). Lowe he e ogenei y was obse ed in
only Eu opean ances y indi iduals, a he han he ull sample, o he
associa ion be ween GFI1-CpGs wi h BMI and WC (0 ≤I
2
b40)
Table 1
Cha ac e is ics o he pa icipan s o s udies in he me a-analysis.
S udy
Ac onym
a
Sample
size
b
Males,
N (%)
Age, mean
(SD), yea s
Cu en
smoke s
c
, N (%)
BMI, mean
(SD), kg/m
2
WC, mean
(SD), cm
TG, mean
(SD), mmol/l
HDL-C, mean
(SD), mmol/l
FG, mean
(SD), mmol/l
DBP, mean
(SD), mmHg
SBP, mean (SD),
mmHg
ALSPAC 1530 554 (36) 49·1 (5·8) 172 (11) 26·8 (4·8) 89·2 (13·2) 1·2 (0·7) 1·4 (0·4) 5·4 (1·1) 74 (10) 123 (14)
BHS_EA 680 308 (45) 43·2 (4·5) 168 (25) 30·0 (6·9) 98·6 (16·4) 1·7 (1·2) 1·2 (0·3) 4·7 (1·2) 81 (9) 117 (14)
BHS_AA 288 113 (39) 43·2 (4·5) 96 (33) 32·5 (8·6) 100·9 (17·7) 1·3 (1·0) 1·3 (0·4) 5·0 (2·0) 89 (14) 131 (22)
CODAM 160 86 (54) 65·5 (6·8) 25 (16) 28·9 (4·3) NA 1·5 (0·7) 1·4 (0·3) NA NA NA
EGCUT 312 156 (50) 50·2 (17) 56 (18) 27·4 (5·6) 91·6 (14·9) 1·3 (0·8) 1·6 (0·5) 5·1 (0·7) 80 (10) 128 (18)
EPICOR 584 376 (64) 53 194 (33) 26·6 (3·8) 90·6 (11·7) 1·5 (0·9) 1·7 (0·5) 6·1 (1·6) 87 (11) 140 (21)
ESTHERa 1000 500 (50) 62·1 (6·5) 186 (19) 27·8 (4·3) NA 1·3 (0·9) 1·3 (0·4) 5·6 (1·2) 87 (12) 146 (22)
ESTHERb 864 390 (45) 62·1 (6·5) 174 (21) 27·7 (4·8) NA 1·5 (0·9) 1·3 (0·4) 5·7 (2·0) 89 (12) 148 (22)
KORAF4 1701 831 (49) 60·9 (8·9) 243 (14) 28·1 (4·6) 95·4 (13·9) 1·6 (1·1) 1·5 (0·4) 5·0 (0·9) 79 (11) 130 (21)
LLD 1057 446 (42) 45·2 (13·5) 439 (42) 25·3 (4·1) 88·2 (12·5) 1·4 (0·2) 1·6 (0·4) 5·2 (0·1) 70 (9) 119 (13)
LLS 631 300 (48) 58·9 (6·6) 85 (13) 25·4 (3·5) NA 1·9 (1·2) 1·4 (0·4) NA NA NA
LOLIPOP 3842 2386
(62)
52 (10·3) 304 (8) 27·5 (4·4) 96·9 (11·2) 1·7 (1·1) 1·3 (0·3) 5·4 (1·1) 81 (11) 131 (19)
NFBC1966–31 740 325 (44) 31 194 (26) 24·5 (4·0) 82·7 (11·4) 1·1 (0·7) 1·6 (0·4) 5·0 (0·8) 76 (11) 124 (13)
NFBC1966–46 716 315 (44) 46 113 (16) 26·8 (4·8) 91·4 (13·2) 1·3 (0·9) 1·6 (0·4) 6·1 (0·7) 86 (11) 129 (17)
NFBC1986 512 232 (45) 16 101 (20) 21·4 (3·5) 74·4 (9·2) 0·9 (0·4) 1·4 (0·3) 5·2 (0·5) 68 (7) 115 (12)
NTR 729 256 (35) 40·3 (15·1) 137 (19) 24·6 (4·1) NA 1·3 (0·7) 1·5 (0·4) NA NA NA
PAN 163 100 (61) 62·6 (9·5) 45 (28) 26·1 (3·7) NA 1·9 (1·1) 1·4 (0·3) NA NA NA
RAINE 819 418 (51) 17 NA 23·2 (4·5) 79·7 (11·6) 1·1 (0·5) 1·3 (0·3) 4·7 (0·6) 58 (6) 113 (11)
RSIII-1 731 336 (46) 59·9 (8·2) 197 (27) 27·5 (4·8) 93·5 (12·8) 1·5 (0·8) 1·4 (0·4) 5·6 (1·2) 81 (11) 132 (20)
RSII-3_III-2 719 305 (42) 67·6 (5·9) 77 (11) 27·7 (4·1) 94·4 (12·0) 1·5 (0·9) 1·5 (0·4) 5·7 (1·2) 84 (11) 144 (21)
SHIP 248 118 (48) 51·6(13·8) 53 (21) 27·3 (4·0) 89·0 (12·5) 1·5 (0·8) 1·4 (0·4) 5·4 (0·6) 76 (9) 124 (17)
YFS 186 72 (39) 44·2 21 (11) 26·2 (4·7) 88·2 (13·7) 1·2 (0·8) 1·4 (0·3) 5·4 (1·1) 73 (9) 119 (13)
Da a shown as N (%) o mean (SD). Acco ding o a ailabili y in he pa icipa ing s udies, BMI was a ailable in 18212, WC in 14665, HDL-C in 18212, TG in 18212, FG in 16529, DBP in 16529,
SBP in 16529 indi iduals o he o al.
Abb e ia ions: BMI –Body Mass Index; WC –Wais Ci cum e ence; TG –T iglyce ides; HDL-C –High Densi y Lipop o ein Choles e ol; FG –Fas ing Glucose; DBP –Dias olic Blood P essu e;
SBP –Sys olic Blood P essu e, NA - no a ailable.
a
S udy names: The A on Longi udinal S udy o Pa en s and Child en (ALSPAC) (specifically subse wi h DNA me hyla ion p ofiles in he Accessible Resou ce o In eg a ed Epigenomic
S udies, ARIES), he wo s udies om Bogalusa Hea S udy (BHS –Eu opean Ame ican(EA) and A ican Ame ican (AA)), he Coho On Diabe es And A he oscle osisMaas ich (CODAM),
he Es onian Genome Cen e, Uni e si y o Ta u (EGCUT), he I alian ca dio ascula sec ion o EPIC (EPICOR), he Coope a i e Gesundhei s o schung in de Region Augsbu g (Coope a i e
Heal h Resea ch in he Augsbu g Region) F4 (KORAF4), he wo independen subse s o he Epidemiologische S udie zu Chancen de Ve hü ung, F ühe kennung und op imie en The apie
ch onische E k ankungen in de äl e en Be ölke ung (ESTHERa and ESTHERb), he Li elines Deep (LLD), he Leiden Longe i y S udy (LLS), he London Li e Science Popula ion s udy
(LOLIPOP), he wo ollow-up da ase s om No he n Finland Bi h coho 1966 (NFBC1966–31 yea s and NFBC1966–46 yea s), No he n Finland Bi h coho 1986 (NFBC1986), he
Ne he lands Twin Regis e s udy (NTR), he P ospec i e Amyo ophic La e al Scle osis s udy Ne he lands (PAN), The Wes e n Aus alian P egnancy Coho s udy (RAINE), he wo inde-
penden s udies om Ro e dam S udy (RS) –RSIII-1 and RSII-3_III-2, he S udy o Heal h in Pome ania –T end (SHIP-T end), and he Young Finns S udy 2011 (YFS). The CODAM, LLS, NTR,
and PAN belong o he BIOS conso ium wi h coo dina ed DNA me hyla ion measu emen s.
b
Sample size o he s udies wi h DNA me hyla ion da a.
c
Cu en smoking was defined as smoking ≥1 ciga e e pe day.
Fig. 1. Map and co ela ion clus e ing o DNA me hyla ion a eigh GFI1 CpGs on human ch omosome 1 (HapMap build 37).
210 P. Pa ma e al. / EBioMedicine 38 (2018) 206–216

(Appendix B Table S7). The independen esul s o all he associa ions
om each o he 22 s udies a e p esen in he Appendix B Tables S9,
S10, S11 and S12.
4. Discussion
The p esen me a-analysis has co obo a ed he associa ion o lowe
DNA me hyla ion a he eigh GFI1-CpGs wi h ma e nal p ena al and
adul own smoking exposu e, as well as uniquely iden i ying lowe
DNA me hyla ion a cg14179389, a p ena al ma e nal smoking- ela ed
locus, as a isk ac o o adul adiposi y and blood p essu e le els. Im-
po an ly, lowe DNA me hyla ion a all he CpGs indica es isk o
highe iglyce ide le els.
Recen ly, s udies ha e shown GFI1-CpGs o media e low bi h
weigh due o p ena al ma e nal smoking exposu e [16], and o associ-
a e wi h sudden in an dea h synd ome (SIDS) [18]. One s iking finding
Fig. 2. Fo es plo showing me a-analysis e ec sizes o DNA me hyla ion a eigh GFI1-CpGs by ma e nal p ena al smoking ac oss fi e s udies (n= 4230). Model 1: CpG = ma e nal
p ena al smoking + echnical co a ia es; Model 2: CpG = ma e nal p ena al smoking + echnical co a ia es + sex + age + adul own smoking. 95% CI, 95% Confidence In e al.
Bon e oni co ec ed P[HYPHEN] alue h eshold o Pb0·012. Open and closed symbol indica e pN0·012 and pb0·012, espec i ely. Ma e nal p ena al smoking was defined as any
ma e nal smoking du ing p egnancy (0 No, 1 Yes). S anda dized alues wi h mean = 0 and s anda d de ia ion = 1 we e used o CpG me hyla ion ac oss all he s udies. DNA
me hyla ion be a alues can be in e p e ed as SD change in me hyla ion o ma e nal p ena al smoking s a us om 0 o 1.
Fig. 3. Fo es plo showing me a-analysis e ec sizes o DNA me hyla ion a eigh GFI1-CpGs by adul smoking ac oss 20 pa icipa ing s udies (n= 13,551). Model 1: CpG = adul own
smoking + echnical co a ia es; Model 2: CpG = adul own smoking + echnical co a ia es + sex + age. CI: Confidence In e al. Bon e oni co ec ed P[HYPHEN] alue h eshold o
Pb0.012. Open and closed symbol indica e pN0·012 and pb0·012, espec i ely. Adul own smoking was defined as 1 o mo e ciga e e pe day (0 No, 1 Yes). S anda dized alues wi h
mean = 0 and s anda d de ia ion = 1 we e used o CpG me hyla ion ac oss all he s udies.
211P. Pa ma e al. / EBioMedicine 38 (2018) 206–216
om ou s udy was he long las ing associa ion be ween exposu e o
ma e nal p ena al smoking and lowe DNA me hyla ion a
cg14179389 un il adul hood. Mo eo e , lowe DNA me hyla ion a
cg14179389 was also associa ed wi h inc eased adul WC, SBP, and
DBP, sugges ing a isk o adiposi y and hype ension. The me a-analysis
e ealed a consis en e ec size and di ec ion o associa ion in all s ud-
ies, highligh ing he ep oducibili y o findings. Fu he mo e, he associ-
a ions pe sis ed and we e ein o ced a e adjus ing o adul own
smoking, suppo ing obus ness and pos na al s abili y o ma e nal
smoking- ela ed DNA me hyla ion locus. P e ious s udies ha e ob-
se ed cg14179389 as he mos consis en and s onges signal
associa ed wi h ma e nal p ena al smoking among GFI1-CpGs [13,16].
These findings also include an app eciable o e lap wi h p e iously iden-
ified e idence o influence o ma e nal smoking on he o sp ing's isk
o obesi y, hype ension, hype lipidaemia and ca dio ascula disease
[19,20]. As hypo hesized, simila i y in influences o ma e nal smoking
and cg14179389 on ca dio-me abolic heal h iden ifies consequences
o childhood de elopmen , and sugges s he e may be an unde lying
egula o y ole o epigene ic changes in ela ion o de imen al ca -
dio-me abolic heal h ou comes. We specula e ha unc ionally impo -
an DNA me hyla ion changes a cg14179389 in adul s a e p esen
om bi h due o smoking exposu e in-u e o.
Fig. 4. Fo es plo showing me a-analysis e ec sizes o ca dio-me abolic pheno ypes in SD change by one SD lowe DNA me hyla ion a eigh GFI1-CpGs ac oss all he pa icipa ing s udies
(n= 18,212). Model 1: Ca dio-me abolic Pheno ype = CpG + echnical co a ia es; Model 2: Ca dio-me abolic Pheno ype = CpG + echnical co a ia es + sex + adul smoking + age.
Bon e oni co ec ed P- alue h eshold o P b0·012 has been used o his analysis. Open and closed symbol indica e pN0·012 and pb0·012, espec i ely. S anda dized alues wi h mean
= 0 and s anda d de ia ion = 1 we e used o ca dio-me abolic pheno ypes and CpG me hyla ion ac oss all he s udies. βcan be in e p e ed as SD change in ca dio-me abolic pheno ype
pe 1-SD dec ease in me hyla ion. Acco ding o a ailabili y in he pa icipa ing s udies, we had BMI o 18212, WC o 14665, HDL-C o 18212, TG o 18212, FG o 16529, DBP o 16529,
and SBP o 16529 o he o al 18212 indi iduals. Abb e ia ions: BMI –Body Mass Index, BP –Blood P essu e, CI - Confidence In e al, HDL-C –High Densi y Lipop o ein Choles e ol.
Measu emen uni s o each ca dio-me abolic pheno ype a e gi en in Table 1.
212 P. Pa ma e al. / EBioMedicine 38 (2018) 206–216
In con as , lowe DNA me hyla ion o he o he six GFI1-CpGs
(cg09935388, cg12876356, cg18316974, cg09662411, cg18146737,
cg04535902) was associa ed wi h dec eased BMI, WC, and BP. O
hese all BMI and WC and he mos o BP associa ions su i ed
Bon e oni co ec ion. The associa ions we e o simila magni ude, al-
hough di ec ionally opposi e o cg14179389. Fu he mo e, adul own
smoking showed a con ounding e ec in a enua ing he associa ions.
These findings a e in ag eemen wi h he obse a ional s udies ha
show highly complex and non-linea associa ions be ween smoking
and ca dio-me abolic heal h [2,5,21]. Sne e e al. obse ed a U-shaped
ela ionship be ween he numbe o ciga e es/day and BMI, wi h low-
es BMI in hose smoking 6–10 ciga e es/day; smoking cessa ion was
associa ed wi h an ini ial inc ease in weigh compa ed o hose who
con inued smoking [22]. Inc eased isk o obesi y among smoke s is ob-
se ed in a dose dependen manne , whe e o me hea y smoke s a e
mo e likely o be obese han o me ligh smoke s and ha e g ea e
isk o CVD e en s [5,21]. Highe BMI in hea y smoke s likely eflec s
clus e ing o isky beha iou s ha is conduci e o weigh gain. Pa adox-
ically, whils smoking acu ely inc eases BP, smoke s a e obse ed o
ha e sligh ly lowe BP le els han non-smoke s, especially in young
adul hood, in la ge epidemiological s udies [23]. The compa able ob-
se a ions be ween six GFI1-CpGs and smoking wi h ca dio-me abolic
pheno ypes aises he in iguing possibili y ha ciga e e smoking in-
duces epigene ic modifica ions a hese CpGs, which, a leas in pa ,
may eflec he de imen al impac o smoking on ca dio-me abolic
heal h. Significan associa ions in ou s udy be ween adul own
smoking and lowe DNA me hyla ion a GFI1-CpGs ac oss he pa icipa -
ing s udies suppo his hypo hesis (Fig. 3, Appendix B Table S4). The
obse ed epigene ic al e a ions may also pa ly indica e po en ial pa h-
ways o complex associa ions be ween smoking and BP.
In e es ingly, lowe DNA me hyla ion a all eigh CpGs showed asso-
cia ion wi h highe TG in echnically co ec ed models, bu adjus men
o adul smoking a enua ed he associa ions, indica ing a s ong con-
ounding o media ion e ec . Consis ency in di ec ion o e ec ac oss
all CpGs implies a conco dan influence o bo h ma e nal and adul
smoking induced epigene ic al e a ions a he GFI1-CpGs on TG. P e i-
ous e idence shows ha ma e nal p ena al smoking exposu e is associ-
a ed wi h hype lipidaemia in o sp ing [19]. Simila ly, adul smoke s
ha e hype lipidaemia and he influence o smoking cessa ion on lipid
p ofiles seems o be qui e modes and highe iglyce ide le els pose
significan isk o CVDs [24,25].
Lowe he e ogenei y was obse ed only in he Eu opean ances y in-
di iduals, a he han he ull sample, o he associa ion be ween GFI1-
CpGs wi h BMI and WC, indica ing popula ion-specificinfluence
pe aining o adiposi y (Appendix B Table S7). The s udies excluded
he e we e o he A ican Ame ican and Sou h Asian ances y. The e is
s ong e idence ha a any gi en BMI, he heal h isks a e ma kedly
highe in some e hnic g oups han o he s. Asians ha e highe weigh -
ela ed disease isks a a lowe BMI and Sou h Asians, in pa icula ,
ha e especially high le els o body a and a e mo e p one o de eloping
abdominal obesi y han Caucasians [26].
We obse ed con as ing associa ions o he di e en GFI1-CpGs
(cg14179389 s six o he CpGs) wi h ca dio-me abolic pheno ypes. In
he NFBC1966 and NFBC1986, we obse ed ha cg14179389 di e ed
and did no co ela e wi h he o he CpGs, while he o he se en CpGs
we e highly co ela ed wi h each o he (Fig. 1, Appendix B Table S2).
This is suppo ed by a ecen popula ion-le el s udy ha iden ified di -
e en ial DNA me hyla ion quan i a i e ai loci (meQTL) a hese eigh
GFI1-CpGs, whe e all bu one o he CpG si es (cg14179389) we e highly
co ela ed wi h he o he s, and o med con iguous clus e s unde he
con ol o one meQTL [27]. Fu he mo e, cg14179389 associa ion wi h
adul smoking disappea ed when adjus ed o o he CpGs whils
cg09935388, cg18316974, and cg18146737 showed independen asso-
cia ions (Appendix B Table S5). This explains he di e ences in hei in-
dependen biological unc ions. Al hough some CpGs we e associa ed
wi h bo h ma e nal and adul smoking, pe haps due o he e e sible
na u e o DNA me hyla ion, hei associa ion wi h ca dio-me abolic
pheno ypes was simila o unc ional consequence o own smoking in
la e li e. Longi udinal analysis has p o ided e idence o apid e e s-
ibili y o DNA me hyla ion in gene al du ing ea ly de elopmen , pa ic-
ula ly du ing he immedia e pos na al yea s, wi h s abiliza ion beyond
age 7, sugges ing a ‘ca ch up’mechanism in ea ly li e [15,28]. In con as ,
he unpe u bed e ec o cg14179389 due o ma e nal smoking in ou
s udy indica es pe sis en dis up ion o DNA me hyla ion due o in-
u e o smoking exposu e a his pa icula si e.
Collec i ely, hese obse a ions sugges e idence o wo concep s
(Fig. 5). Fi s , ma e nal p ena al smoking induces a oe al esponse mod-
ula ed h ough pe sis en epigene ic dis up ion. Second, adul own
smoking, ha may be po en ially influenced by ma e nal smoking, in-
duces simila epigene ic changes, which may play a ole in he unde ly-
ing pa hways owa ds ad e se consequences o smoking on
ca dio ascula isk. I is impo an o conside ha many o he en i on-
men al ac o s con ibu e o he ca dio ascula isk (e.g. physical ac i -
i y, s ess, seden a y li es yle, die , alcohol consump ion), and
associa ed DNA me hyla ion dis up ion ollowing exposu es o he ma-
e nal p ena al and own smoking could only explain a pa ial media ing
ole.
This s udy ep esen s a majo e o o pe o m a la ge-scale me a-
analysis o ma e nal p ena al smoking, DNA me hyla ion a GFI1-locus
and ca dio-me abolic pheno ypes. A wide ange o pheno ypic da a
was a ailable, acili a ing assessmen o he unc ional consequences
o DNA me hyla ion changes o e a a ied age ange. The s udy epli-
ca es and confi ms p e iously epo ed associa ions o lowe DNA
me hyla ion a he GFI1-CpGs wi h exposu e o own and ma e nal p e-
na al smoking.
We acknowledge ha la ge collabo a ions u ilizing summa y le el
da a, al hough use ul in enhancing powe o de ec associa ions, may
limi he abili y o unde ake mul iple sensi i i y analyses. We we e un-
able o ully analyse DNA me hyla ion changes o e he li e-cou se and
disen angle he in e ac ion o age wi h DNA me hyla ion, o suppo
eme ging e idence ha shows e e sibili y o DNA me hyla ion pa -
e ns [29]. Ano he limi a ion was he use o leucocy es, which we e
he sou ce o DNA used. They a e composed o se e al cell ypes each
wi h cell- ype specific DNA me hyla ion pa e ns and hus di e ences
in hese cell ypes could po en ially con ound he obse ed associa ions.
Adjus men o de i ed cell ype p opo ions was included in he analy-
sis o o e come his e en uali y. Ou s udy included eigh GFI1-si es as-
socia ed wi h ma e nal p ena al smoking. We ecognize ha u he
wo k explo ing he associa ions be ween DNA me hyla ion o o he
adul and ma e nal p ena al smoking ela ed loci and ca dio-me abolic
pheno ypes could yield addi ional insigh s in o he ole o epigene ic
ma ke s ha may join ly a ec ca dio-me abolic heal h. In addi ion,
lack o gene exp ession da a ac oss s udies limi ed insigh in o he mo-
lecula mechanisms. Addi ional e idence is needed o suppo GFI1 as
he causal gene esponsible o he obse ed findings. Howe e , in a e-
cen animal s udy, GFI1 did a ec he sys emic inflamma ion h ough
he NE-dependen -C/EBPa-GFI1 pa hway ha p edisposes o me abolic
dys unc ion and obesi y [30]. T ansla ing hese findings o human da a
would be clinically ele an in ligh o ou findings.
5. Conclusion
Ou findings suppo e idence ha epigene ic ac o s a he GFI1-
locus, ha a e associa ed wi h exposu es o smoking in-u e o o adul -
hood a e also linked o ca dio-me abolic isk ac o s, specifically sug-
ges ing a ole in hype iglyce idemia. The findings suppo an
unde lying epigene ic componen o he epidemiologically obse ed
ca dio-me abolic isk by ma e nal p ena al and adul smoking. The
ac ha hese epigene ic ac o s associa e wi h ca dio-me abolic isk
in la e li e e en among non-smoke s exposed o in-u e o smoking
may ha e impo an clinical implica ions. Such epigene ic loci migh
se e as objec i e bioma ke s o pas en i onmen al exposu es ha
213P. Pa ma e al. / EBioMedicine 38 (2018) 206–216
could be used o p e en i e heal h measu es. Ou findings p o ide a
s ong ounda ion o u he wo k o un a el eme ging epigene ic
ma ke s wi h downs eam de imen al heal h ou comes, and deli e
s ong e idence o suppo he ea ly o igin o adul heal h. I d aws a -
en ion o inc ease awa eness on smoking cessa ion and be e p e en-
ion s a egies.
Funding s a emen s
The UK Medical Resea ch Council and Wellcome (g an numbe
102215/2/13/2) and he Uni e si y o B is ol p o ide co e suppo o
ALSPAC. This publica ion is he wo k o he au ho s and Ma hew
Sude man will se e as gua an o o he ALSPAC- ela ed con en s o
his pape . Analysis o he ALSPAC da a was unded by UK Economic &
Social Resea ch Council g an (g an numbe ES/N000498/1). ARIES
was unded by he BBSRC (BBI025751/1 and BB/I025263/ 1). Supple-
men a y unding o gene a e DNA me hyla ion da a which a e (o will
be) included in ARIES has been ob ained om he MRC, ESRC, NIH and
o he sou ces. ARIES is main ained unde he auspices o he MRC In e-
g a i e Epidemiology Uni a he Uni e si y o B is ol (g an numbe s
MC_UU_12013/2, MC_UU_12013/8 and MC_UU_12013/9). Ma hew
Sude man was suppo ed by he BBSRC and ESRC (g an numbe ES/
N000498/1). BHS The s udy is suppo ed by g an 2R01AG016592
om he Na ional Ins i u e on Aging. Shengxu Li was pa ly suppo ly
by g an 13SDG14650068 om he Ame ican Hea Associa ion. NTR,
LLS, CODAM and PAN we e suppo ed om he Ne he lands Ca dio as-
cula Resea ch Ini ia i e ( he Du ch Hea Founda ion, Du ch Fede a ion
o Uni e si y Medical Cen es, he Ne he lands O ganiza ion o Heal h
Resea ch and De elopmen , and he Royal Ne he lands Academy o Sci-
ences) o he GENIUS p ojec “Gene a ing he bes e idence-based
pha maceu ical a ge s o a he oscle osis”(CVON2011–19). This
wo k was pe o med wi hin he amewo k o he Biobank-Based In e-
g a i e Omics S udies (BIOS) Conso ium unded by BBMRI-NL, a e-
sea ch in as uc u e financed by he Du ch go e nmen (NWO
184.021.007). EGCUT was suppo ed by he Es onian Resea ch Council
G an s PRG184, PUT1665, IUT20-60, EU H2020 g an ePe Med (G an
No. 692145), and Eu opean Union h ough he Eu opean Regional De-
elopmen Fund (P ojec No. 2014–2020.4.01.15–0012). Lili Milani
was suppo ed by Uppsala Uni e si y S a egic Resea ch G an as pa
o he Science o Li e Labo a o y ellowship p og am. EPICOR was sup-
po ed by he Compagnia di San Paolo o he EPIC-I aly and EPICOR p o-
jec s, he I alian Ins i u e o Genomic Medicine (IIGM, o me ly Human
Gene ics Founda ion-To ino, HuGeF, Tu in, I aly) and he MIUR ex60%
g an . EPIC-I aly is u he suppo ed by a g an om he “Associazione
I aliana pe la Rice ca sul Canc o”(AIRC, Milan). The EPICOR s udy was
also suppo ed by a MIUR g an o he Depa men o Medical Sciences,
p ojec “Dipa imen i di Eccellenza 2018 –2022”. The ESTHER s udy
was suppo ed by he Baden-Wü embe g S a e Minis y o Science,
Resea ch and A s (S u ga , Ge many), he Fede al Minis y o Educa-
ion and Resea ch (Be lin, Ge many), and he Fede al Minis y o Family
A ai s, Senio Ci izens, Women and You h (Be lin, Ge many). The KORA
s udy was ini ia ed and financed by he Helmhol z Zen um München –
Ge man Resea ch Cen e o En i onmen al Heal h, which is unded by
he Ge man Fede al Minis y o Educa ion and Resea ch (BMBF) and by
he S a e o Ba a ia. Fu he mo e, KORA esea ch has been suppo ed
wi hin he Munich Cen e o Heal h Sciences (MC-Heal h), Ludwig-
Maximilians-Uni e si ä , as pa o LMUinno a i . The esea ch leading
o hese esul s has ecei ed unding om he Eu opean Union Se en h
F amewo k P og am unde g an ag eemen [n°602736] (Mul i-dimen-
sional omics app oach o s a ifica ion o pa ien s wi h low back pain -
PAIN-OMICS; h p://www.painomics.eu/) and unde g an ag eemen
[n°603288] (Sys ems Biology o Iden i y Molecula Ta ge s o Vascula
Disease T ea men –SysVasc; h p://www.sys asc.eu/). This wo k was
u he suppo ed by a g an (WA 4081/1–1) om he Ge man Resea ch
Founda ion. N. Ve weij was suppo ed by NWO VENI (016.186.125). L.
F anke is suppo ed by ZonMW-VIDI 917.14.374 and ERC S a ing
G an , g an ag eemen 637640 (ImmRisk). The LOLIPOP s udy is sup-
po ed by he Na ional Ins i u e o Heal h Resea ch (NIHR) Comp e-
hensi e Biomedical Resea ch Cen e Impe ial College Heal hca e NHS
T us , he B i ish Hea Founda ion (SP/04/002), he Medical Resea ch
Council (G0601966, G0700931), he Wellcome T us (084723/Z/08/Z,
090532 & 098381) he NIHR (RP-PG-0407-10371), he NIHR O ficial De-
elopmen Assis ance (ODA, awa d 16/136/68), he Eu opean Union
FP7 (EpiMig an , 279143) and H2020 p og ams (iHeal h-T2D,
643774). John Chambe s is suppo ed by Singapo e Minis y o Heal h's
Na ional Medical Resea chCouncil unde i s Singapo e T ansla ional Re-
sea ch In es iga o (STaR) Awa d (NMRC/STaR/0028/2017). NFBC1966
ecei ed financial suppo om Uni e si y o Oulu G an no. 65354,
Oulu Uni e si y Hospi al G an no. 2/97, 8/97, Minis y o Heal h and
Social A ai s G an no. 23/251/97, 160/97, 190/97, Na ional Ins i u e
o Heal h and Wel a e, Helsinki G an no. 54121, Regional Ins i u e o
Occupa ional Heal h, Oulu, Finland G an no. 50621, 54231. NFBC1986
ecei ed financial suppo om EU QLG1-CT-2000-01643 (EUROBLCS)
G an no. E51560, No FA G an no. 731, 20056, 30167, USA / NIHH
2000 G DF682 G an no. 50945. MRC no: MR/M013138/1, ERDF Eu o-
pean Regional De elopmen Fund G an no. 539/2010 A31592. Ma hias
Adul own smoking
TG
BMI
Li es yle (alcohol, die , physical ac i i y),
Socio-economic ac o s,
Men al heal h (dep ession, anxie y)
Ma e nal p ena al smoking
Fe al hypoxia,
Fe al g ow h es ic ion,
Low bi h weigh
Pos na al ca ch-up g ow h
BP
TG
BMI
BP
TG
BMI
BP
Epigene ic changes
Pe sis en epigene ic changes
Epigene ic changes
Fig. 5. Model ep esen ing he po en ial mechanis ic pa hways in he s udy.
214 P. Pa ma e al. / EBioMedicine 38 (2018) 206–216