CETP (Cholesteryl Ester Transfer Protein) Concentration A Genome-Wide Association Study Followed by Mendelian Randomization on Coronary Artery Disease

r r r r r r

Choleste yl Este T ansfe P otein (CETP) Concent aton: a Genome-

d d d M d d

wi e Associaton Stu y followe by en elian Ran omizaton on r r r r

Co ona y A te y Disease

Lisanne L. Blauw, BSc^1,2, Ruifang Li-Gao, MSc², Raymon Noo am, PhDd rd ³, Renée e d Mutse t, PhDr ², r

Stella T ompet, PhD^3,4, Jimmy F.P. Be bée, PhDr ^1,5, Yanan Wang, PhD^1,5, Jan B. van Klinken, PhD ^5,6 ,Tim

r M

Ch isten, Sc², Diana van Heemst, PhD³, Dennis O. Mook-Kanamo i, PhDr ^2,7, F its R. Rosen aal , r d MD PhD ², J. Woute Jukema, r M D PhD⁴, Pat ick C.N. Rensen, PhDr ^1,5, Ko Willems van Dijk, PhD^1,5,6

1Dept. Inte nalr M de icine, Div. En oc inology,d r ²Dept. Clinical Epi emiology,d ³Dept. Inte nalr M de icine,

r d r r

Div. Ge ontology an Ge iat ics,⁴Dept. Ca iology,rd ⁵Einthoven Labo ato y fo Expe imental Vascular r r r r M de icine,⁶Dept. Human Genetcs,⁷Dept. Public Health an P ima y Ca e, Lei en Unive sityd r r r d r M de ical

r d r d

Cente , P.O. Box 9600, 2300 RC Lei en, The Nethe lan s

r r

Sho t unning ttle:Genome-wi e associaton stu y on CETP concent atond d r

rr d r

Co espon ing autho : Lisanne L. Blauw, Lei en Unive sity d r M de ical Cente , Dept.r M de icine, Div.

d r d r d

En oc inology, Post zone C7Q, P.O. Box 9600, 2300 RC Lei en, The Nethe lan s; Phone +31- 715265303; Email l.l.blauw@lumc.nl

rd

Wo count: 5,225 wo s an 2,552 wo s ford d rd r refe ences; 3 Tables, 2 Figu es, 3 Supplementa yr r r

r r

fgu es, 5 Supplementa y tables d

Subject co es: Ca iovascula Disease; Genetc, Associaton Stu ies; Lipi s an Choleste olrd r d d d r 1

2

3

4

5 6 7 8

9 10 11 12

13 14

15 16 17

18 19 20

r Abst act

r d

Backg oun —We aimed to di entfy in epen ent genetc d d dete minants of ci culatng choleste ylr r r r r r r

este t ansfe p otein (CETP)to assesscausal efects of va iaton in CETP concent aton on ci culatng r r r

d r d rd r d r

lipi concent atons an ca iovascula isease isk.

Metho s an Results— d d A genome-wi e associaton (GWA) iscove y and d r d replicaton stu y on se umd r

r r dd d r d d d d

CETP concent aton we e embe e in the Nethe lan s Epi emiology of Obesity (NEO) stu y. Base

d d d d r M d r d d d r d

on the in epen ent i entfe va iants, en elian an omizaton was con ucte on se um lipi s

d d r r r r d M r d r

(NEO stu y) an co ona y a te y isease (CAD) (CARDIoGRA plusC4D conso tum). In the iscove y

d d r d d r

analysis (N=4,248), we i entfe th ee in epen ent va iants (P<5×10^-8) that dete mine CETPr

r r d

concent aton. These SNPs we e mappe to CETP, and replicated in a sepa ate subpopulaton r

r r r r r

(N=1,458). Pe -allele inc ease (SE) in se um CETP was 0.32 (0.02) µg/mL fo s247616-C, 0.35 (0.02)

r r d r r d r r

µg/mL fo s12720922-A, an 0.12 (0.02) µg/mL fo s1968905-G. Combine , these th ee va iants

d r r r

explaine 16.4% of the total va iaton in CETP concent aton. One µg/mL inc ease in genetcally- dete mine CETP concent aton st ongly ec ease high- ensity lipop otein (HDL) choleste ol (-0.23r d r r d r d d r r

d r r d d r r

mmol/L; 95% CI -0.26, -0.20), mo e ately inc ease low ensity lipop otein (LDL) choleste ol (0.08

d d dd r r r

mmol/L; 0.00, 0.16), an was associate with an o s ato of 1.08 (0.94, 1.23) fo CAD isk.

Conclusions—This is the f st GWAS stu y i entfying in epen ent va iants that la gely ete mine r d d d d r r d r

r r r r

CETP concent aton. While HDL-C is not a causal isk facto fo CAD, it has been unequivocally demonst ate that LDL-C lowe ing is p opo tonally associate with a lowe CAD isk. The efo e, ther d r r r d r r r r results of ou stu y a e fully consistent with the noton that CETP concent aton is causally associate r d r r d

r

with CAD th ough LDL-C.

rd

Keywo s—CETP, Co ona y a te y isease, GWAS,r r r r d Men elian an omizatond r d 21

22 23 24

25 26 27 28 29 30 31 32 33 34 35 36

37 38 39 40 41

42

d rd r d r Non-stan a abb eviatons an ac onyms

r M d d r r r r r d

ApoB, apolipop otein B; B I, bo y mass in ex; C, choleste ol; CAD, co ona y a te y isease; CETP,

r r r r r rd r d d

choleste yl este t ansfe p otein; CVD, ca iovascula isease; ELISA, enzyme-linke immune r

so bent assay; eQTL, e rxp ession quanttatve t ait loci; r GCTA, genome-wi e complex t ait analysis;d r

d r r r r

GLGC, Global Lipi s Genetcs Conso tum; GRS, genetc isk sco e; GTEx, genotype-tssue exp ession;

d d d r r d

GWAS, genome-wi e associaton stu y; HDL, high- ensity lipop otein; NEO, Nethe lan s

d r d r

Epi emiology of Obesity; PC, p incipal component; SNP, single nucleot e polymo phism; (V)LDL

r d r

(ve y-) low- ensity lipop oteins 43

44 45 46 47 48 49 50

1. Int o uctonr d

r r r r r r r r

Choleste yl este t ansfe p otein (CETP) facilitates the net fux of choleste yl este s f om high- density lipop oteins (HDL) towa s (ve y-) low- ensity lipop oteins ((V)LDL), couple to a net fux ofr rd r d r d

r r d r

t iglyce i es f om (V)LDL to HDL.¹ As such, CETP cont ibutes to an athe ogenic lipop otein p ofle (i.e.r r r r

r r r d d d

high LDL-choleste ol/HDL-choleste ol ato), as has been extensively stu ie in both humans an in

r r

mice t ansgenic fo human CETP.^{2, 3}The efo e, inhibiton of CETP has long been ega e a p omisingr r r rd d r

r r d d d r d

the apeutc st ategy to atenuate yslipi aemia an ultmately p event the evelopment of

rd r d

ca iovascula isease (CVD).

r r d d r

Of the fou clinical t ials that have stu ie the efects of pha macological CETP inhibiton on CVD risk e ucton, only the fou th an most ecent REVEAL t ial with anacet apib i meet its p ima y r d r d r r r d d r r

d r d r r r

en point, i.e. a e ucton in majo co ona y events.⁴Cont a y to expectatons, the clinical t ials withr r r

r r r d r d r r r d r r d

the CETP the inhibito s to cet apib, alcet apib, an evacet apib, we e te minate : to cet apib ha

r d r r d d r rd r

of-ta get efects on bloo p essu e an cause an inc ease in ca iovascula events,⁵ an bothd dalcet apib anr d evacet apib lacker d efcacy in e ucing ca iovascula events on top of statn r d rd r

r

the apy.^{6, 7} All of these CETP inhibito s cause a la ge inc ease in HDL-C, an a low to mo e ate r d r r d d r d rec ease in LDL-C. ^5-8Although high HDL-C concent aton is associate with a ec easer d d r d risk of CVD in

d d

epi emiological stu ies,⁹Voight an colleaguesd ¹⁰showe in ad Men elian an omizaton stu y thatd r d d

d r d r r d d r r rd

genetcally- ete mine highe HDL-C concent atons o not ec ease the isk of myoca ial

r d d

infa cton, in icatng that the associaton between HDL-C an CVD is not causal. This may be one of

r r r d r

the explanatons fo the lack of efcacy of the th ee inital CETP inhibito s. Although the un e lying reason fo success of the fou th CETP inhibito is not yet eluci ate , anacet apib showe the la gest r r r d d r d r r de ucton in LDL-C concent aton compa e to the th ee inital CETP inhibito s, r r d r r ⁴which may possibly

r r d

explain its benefcial efects on CVD isk e ucton.

d r r d r d r r

Recent evi ence shows that se um CETP is la gely e ive f om hepatc mac ophages¹¹, but the

r r r r d d

genetc basis of se um the CETP concent aton in the gene al populaton emains to be eluci ate . A 51

52 53 54 55 56 57 58

59 60 61 62 63 64 65 66 67 68 69 70 71 72 73

74 75

r d d r r r d d la ge genome-wi e associaton stu y (GWAS) on ci culatng CETP has not been pe fo me to ate.

d d r r

With i entfcaton of the main genetc ete minants of ci culatng CETP, the causal efects of

r r r r d r d r

va iaton in se um CETP concent aton on ci culatng lipi concent atons an CVD isk can be

d M d r d r M d r d d

assesse using en elian an omizaton. In the past, seve al en elian an omizaton stu ies with

r r r d

a compa able aim have been pe fo me ,^12-14 inclu ing a ecent meta- en elian an omizatond r M d r d r

analysis by Fe ence et al. with ata f om ove 100,000 pa tcipants. d r r r ¹⁵Howeve , these stu ies user d d

d d r r d d r r

can i ate SNPs athe than GWAS-i entfe SNPs, which may be less powe ful genetc inst uments

r r r rd r d

fo assessing the causal ole of CETP concent aton in ca iovascula isease.^{16, 17}

r d d d d r d r

With the p esent stu y, we aim to i entfy in epen ent genetc va iants that ete mine

r r d r r d d r

ci culatng CETP concent aton, using a genome-wi e athe than a can i ate gene app oach. In

dd r r M d r d

a iton, we aim to use these va iants as genetc inst uments in en elian an omizaton to assess

r r d d r r r r d r

the causal efects of va iaton in CETP on se um lipi s an co ona y a te y isease (CAD) isk, which

d r d r d

may assist in un e stan ing the efectveness of pha maceutcal CETP inhibiton. To this en , we

r r d r r d r r d r

pe fo me a GWAS on se um CETP concent aton, using a iscove y coho t (n=4,248) an a sepa ate replicaton coho t (n=1,458) f om the Nethe lan s Epi emiology of Obesity (NEO) stu y.r r r d d d

d d d M d r d r d

Subsequently, we use the i entfe SNPs in en elian an omizaton analyses on se um lipi

r d d d r

concent atons in the NEO stu y populaton an the Global Lipi s Genetcs Conso tum (GLGC)¹⁸, and

d r M r

on CAD using publically-available ata f om the CARDGIoGRA plusC4D conso tum.¹⁹ 76

77 78 79 80 81 82 83

84 85 86 87 88 89 90 91 92 93

2. Metho sd

2.1 Stu y esign an populatond d d

d d r r d d d

The NEO stu y is a populaton-base p ospectve coho t stu y of men an women age between 45

d r r r r r d r d r r d

an 65 yea s. F om the g eate a ea of Lei en, The Nethe lan s, all inhabitants with a self- epo te d d M

bo y mass in ex (B I) of 27 kg/m² r o highe we e eligible to pa tcipate. In ar r r dditon, inhabitants

r r d rd r r d r r d

f om one nea by municipality (Lei e o p, The Nethe lan s) in the same age g oup we e invite to

r r rd r M r r r r M d r

pa tcipate ega less of thei B I, fo ming a efe ence populaton fo B I ist ibuton. In total,

r r d d r r r r d

6,671 pa tcipants we e inclu e f om Septembe 2008 untl Septembe 2012. Pa tcipants visite

d r r d r d r

the NEO stu y cente fo extensive physical examinaton. Venous bloo samples we e obtaine f om

r r r r r r rd d rr d

the antecubital vein afe a 10 hou ove night fast. Resea ch nu ses eco e cu ent me icaton use

d r r r d r d r

by means of a me icaton invento y. P io to the stu y visit, pa tcipants complete questonnai es at

r d r d r

home with espect to emog aphic, lifestyle, an clinical info maton.

d r d d d r M d

The NEO stu y was app ove by the me ical ethics commitee of the Lei en Unive sity e ical

r M d r r r r d

Cente (LU C), an all pa tcipants gave thei w iten info me consent. Detaile info maton aboutd r

d d d d d r d r

the stu y esign an ata collecton has been esc ibe elsewhe e.²⁰

Metho s used d fo genotyping anr d biochemical analyses a e r desc iber d in detail in the r

Supplemental mate ial.

d d d d r d

Due to consent issues, the in ivi ual ata of NEO stu y pa tcipants will not be ma e available to

r r r r r r r r d r r r r d r r

othe esea che s fo pu poses of ep o ucing the esults o eplicatng the p oce u e. Howeve , a

d r r d d

fle inclu ing GWAS summa y statstcs can be equeste via k.willems_van_ ijk@lumc.nl.

2.2 Genome-wi e associaton stu yd d

d d d rd d r r r

We con ucte the GWAS on the unstan a ize se um CETP concent aton fo all autosomal

r d d d d d r r

ch omosomes. We ivi e the total NEO stu y populaton base on the g aphical a ea of r rec uitment into a discove y coho t (n=4,248; g eate a ea of Lei en, The Nethe lan s) anr r r r r d r d d a 94

95

96 97 98 99 100 101 102 103 104 105

106 107 108

109 110

111 112 113

114

115 116 117

replicaton coho t (n=1,458; Lei e o p, The Nethe lan s). This was consi e e to be vali as wer d rd r r d d r d d recently showe , using ata of the NEO stu y populaton, that se um CETP concent aton was notd d d r r

d M r r r d

associate with B I no with othe measu es of bo y fat.²¹

dd r r r r r d d r r d r d

A itve (pe -allele) linea eg ession analyses we e con ucte sepa ately fo the iscove y an replicaton coho t in SNPTEST v2, a juste fo age, sex an the f st fou PCs. To i entfy va iants thatr d d r d r r d r

r d d d r r d d d

we e in epen ently associate with se um CETP concent aton, we use con itonal an joint

r r r d r d r

analyses to pe fo m a stepwise selecton p oce u e using the genome-wi e complex t ait analysis r

(GCTA) tool ve sion 1.24.4.²² A con itone P-value <5×10d d ^-8 was consi e e to be genome-wi ed r d d

d d d

signifcant, an a con itone P-value <1×10^-6was consi e e a suggestve signal. In epen ent singled r d d d

d r d d

nucleot e polymo phisms (SNPs) with a con itone P-value <1×10^-6 in the iscove y analysis we e d r r

d d r r r r

vali ate in the eplicaton sample. SNPs with a P-value <0.05 in the eplicaton coho t we e

d r d r d d d d r d r

consi e e to be eplicate . Upon i entfcaton of the lea SNPs, we ete mine whethe the d rist ibuton of the co ing alleles was simila in use s an non-use s of lipi -lowe ing d r r d r d r drugs. M ro e detaile info maton on this metho is esc ibe in the Supplemental mate ial.d r d d r d r

r

As Taq1B ( s708272)²³an -629C>A ( s1800775)d r ²⁴ r a e the most stu ie va iants in thed d r CETP gene

r r d r r r r

in lite atu e, we specifcally checke the GWAS esults fo thei associaton with CETP concent aton.

dd r r d d r d d r r

In a iton, we epo te the linkage isequilib ium (NEO stu y) of the lea SNPs f om the p esent GWAS with the eightCETPSNPs that we e use in a ecent meta- en elian an omizaton stu y onr d r M d r d d

r r r d d r d r

co ona y hea t isease an se um lipi s by Fe ence et al.¹⁵(i.e. s3764261, s1800775, s1864163, r r r rs9929488, s9989419, s12708967, s289714 an r r r d rs5880). This allows compa ison between our r

r d d d r r r d

genetc inst ument compose of SNPs i entfe fo m a GWAS on se um CETP concent aton an

r r d d d

thei genetc inst ument compose of can i ate SNPs.

d r r r r d d r d

The explaine va iance in se um CETP concent aton fo the in epen ent va iants was estmate

r r r d d d r d r

in the eplicaton coho t. Fo each in ivi ual SNP the explaine va iance was estmate as the pa tal R² rf om the linear r reg ession mo el with the SNP as in epen ent va iable and d d r d se um CETPr 118

119 120

121 122 123 124 125 126 127 128 129 130 131

132 133 134 135 136 137 138 139

140 141 142

r d d r r d d d concent aton as epen ent va iable. To estmate the total va iance explaine by all in epen ent

d d r r d r d d

lea SNPs, a weighte genetc isk sco e (GRS) was calculate pe in ivi ual. The GRS was

r d r r d d r

const ucte as the sum of the numbe of isk alleles on the lea SNPs weighte by thei efect size on

r d r r d d r r

CETP concent aton in the iscove y coho t. The combine explaine va iance was the pa tal R² rf om

r r r d d d d r d r

the linea eg ession mo el, with the weighte GRS as in epen ent va iable an se um CETP

r d d r

concent aton as epen ent va iable.

d d d r r

To quantfy the genome-wi e cumulatve efects of in epen ent va iants infuencing va ious

rr r r r d r

phenotypes, genetc co elatons of se um CETP concent aton with se um lipi concent atons (i.e.

r r d d r d M r d r

HDL-C, LDL-C, t iglyce i es an total choleste ol), an B I we e calculate (Supplemental mate ial).

2.3 E rxp ession quanttatve t ait loci (eQTLs) analysis r

r d d d d r r

To investgate whethe the i entfe lea SNPs coul explain se um CETP concent aton via

r r r d r r

t ansc iptonal gene egulaton, we checke if these SNPs we e eQTLs fo CETPusing ata f om the d r r

genotype-tssue exp ession (GTEx) p oject po tal (V7) r r ²⁵an the Bloo eQTL b owse .d d r r²⁶

2.4 M den elian an omizaton r d

d d d d d d r d r r r r d

Base on the i entfe in epen ent an eplicate SNPs fo se um CETP concent aton in ou stu y

d d M d r d r d r

populaton, we con ucte en elian an omizaton analyses on se um lipi concent atons in the

d d r d r d

NEO stu y populaton, an on the isk of CAD base on publically-available summa y statstcs ata

r M d

f om the CARDGIoGRA plusC4D 1000 Genomes stu y.¹⁹ dA etailed desc ipton of ther Men eliand ran omizaton analyses on CAD isk using ata f om the CARDIoGRA plusC4D conso tum can bed r d r M r

d r r r r r r d dd r

foun in the Supplemental mate ial. Efect estmates fo CAD isk we e epo te as o s ato with

rr d d

co espon ing 95% CI. We use a publically available tool²⁷ to con uct a powe analysis fo thed r r Men elian an omizaton analysis on CAD, which was base on the fn ings f om the GWAS ond r d d d r

r r d r d d

se um CETP concent aton, the explaine va iance of the SNPs use to compose the GRS, an the

M d r

sample size of the CARDIoGRA plusC4D 1000 Genomes stu y (60,801 cases; 123,504 cont ols).

143 144 145 146 147 148

149 150 151

152

153 154 155

156

157 158 159 160 161 162 163 164 165 166

d d d d d d d d In the NEO stu y populaton, we calculate the in ivi ual weighte GRS base on the i entfe

d d r d r d r d r

SNPs an ete mine the efect of 1 µg/mL inc ease in genetcally- ete mine CETP concent aton

r r r r d r r

on the concent atons of total choleste ol, HDL-C, t iglyce i es, LDL-C, total choleste ol/HDL-C ato,

r d r r r r d d r d

LDL-C/HDL-C ato an apolipop otein B (ApoB), using linea eg ession analysis a juste fo age an

d r r r d r r d r r r d

sex. Beta coefcients an 95% CIs we e epo te . We epo te total choleste ol, t iglyce i es, HDL-C

d r r r r r r d d

an LDL-C concent atons in mmol/L. We epo t total choleste ol, t iglyce i es, HDL-C an LDL-C

r d r d r d d

concent atons in mmol/L. To yiel concent atons in mg/ L, choleste ol values shoul be multplie

d r r d

by 38.67 an t iglyce i e values by 88.57.²⁸

dd r r r r d d d d r

In a iton, fo eplicaton pu poses, we ext acte the in epen ent lea s SNPs f om publically

d

available atasets of the GLGC.¹⁸We also ext acte the Taq1B ( s708272) an -629C>A ( s1800775)r d r d r

r r M d d

polymo phisms f om the CARDGIoGRA plusC4D 1000 Genomes an GLGC atasets.¹⁸ 167

168 169 170 171 172 173 174

175 176 177

178

3 Results

3.1 Populaton cha acte istcsr r

r r d r d r r r r d r d

Cha acte istcs of the iscove y an eplicaton coho ts a e summa ize in Table 1. Compa e with

r r r r r d r r r

the eplicaton coho t, the e we e fewe women in the iscove y coho t (50.6% ve sus 56.1%). Also,

r d r r d r M

pa tcipants in the iscove y coho t ha a highe B I (30.3 kg/m² rve sus 25.6 kg/m²) an mo e ofend r

d d r dr r r r r r

use lipi -lowe ing ugs (17.5% ve sus 10.4%) than pa tcipants in the eplicaton coho t. Se um

d d r r r r

CETP an lipi concent atons we e compa able between both coho ts.

3.2 Genome-wi e associaton analysisd

r r r d

The -Log(P-value) plot fo the GWAS is shown in Figu e 1. The accompanying list of SNPs that eache d

genome-wi e signifcance (P<5×10^-8) is p esente in Supplementa y table 1. Afe con itoning on r d r r d

d r d d r r d d d d

the lea SNPs, th ee in epen ent va iants eache genome-wi e signifcance (con itone P<5×10^-8;

r r d r d d d d

Supplementa y Figu e 2) an seven suggestve signals we e i entfe (con itone P<1×10^-6) in the discove y coho t (Table 2). The th ee genome-wi e signifcant va iants we e all mappe to ther r r d r r d CETP

d d r r r

gene. Notably, these in epen ent va iants we e s247616 (P=1.86×10^-64), s12720922 (P=6.68×10 r ^-13) d r

an s1968905 (P=1.66×10^-12), which ha a pe -allele inc ease (SE) in se um CETP of 0.32 (0.02)d r r r

r r d r

µg/mL ( s247616-C), 0.35 (0.02) µg/mL ( s12720922-A) an 0.12 (0.02) µg/mL ( s1968905-G). These

r r r r d r d r

th ee va iants we e all eplicate in the eplicaton analysis (P<0.05). In the NEO stu y, these va iants

r d r r d r

togethe explaine 16.4% of the se um CETP concent aton. The ist ibutons of the efect alleles of

r d r r r d d d r dr d d

the th ee lea SNPs we e simila fo in ivi uals taking lipi -lowe ing ugs an not taking lipi -

r dr r r r d r

lowe ing ugs (Supplementa y table 2). A numbe of SNPs we e suggestvely associate with se um

r d d

CETP concent aton, inclu ing SNPs mappe toADA TS3M ,PPARGand LPL.

d d r r d

The uncon itone pe -allele efect size of the well-known Taq1B ( s708272) an -629C>A

r r r r r d

( s1800775) va iants was 0.27 µg/mL fo both SNPs (Supplementa y table 1). Of the th ee lea SNPs, 179

180

181 182 183 184 185

186

187 188 189 190 191 192 193 194 195 196 197 198 199

200 201

d r r

Taq1B was in high linkage isequilib ium with s247616 (LD=0.55), as was -629C>A (LD=0.51). Taq1B

d r r

an -629C>A we e also in high linkage with each othe (LD=0.83).

d r r d d r d

The linkage isequilib ium between the th ee lea SNPs an eight SNPs that we e use as

r r

genetc inst uments fo aCETPGRS in a ecent meta- en elian an omizaton analysis on co ona y r M d r d r r

r d d r d

hea t isease an se um lipi s¹⁵ r a e shown in Supplementa y fgu e 3. The two st ongest lea SNPs r r r d

r r r d r r d r

f om the p esent GWAS, i.e. s247616 an s12720922, we e in high linkage isequilib ium with the

d d d d r

eight can i ate SNPs of the GRS that was use in the stu y of Fe ence et al.,¹⁵with the highest

d r r d r d r d

linkage isequilib ium between s247616 an s3764261 (0.996), an between s12720922 an rs1864163 (0.646).

rr r r r d r d M

The genetc co elaton of se um CETP concent aton with se um lipi concent atons an B I is repo te in Supplementa y table 3. The genetc co elaton of se um CETP concent aton was highestr d r rr r r

r r d r r r d r d

with se um HDL-C concent aton (0.17) an se um t iglyce i e concent aton (-0.29), an lowest with

r r r r r d M

se um total choleste ol concent aton (-0.020), se um LDL-C concent aton (0.074) an B I (0.032).

3.3 eQTL analysis of the lea SNPs d

r r r r d r

Table 3 shows the eQTLs fo the genetc va iants s247616, s12720922 an s1968905. The SNP that

r d r r r

was most st ongly associate with se um CETP concent aton in the GWAS, i.e. s247616, was

d d r

i entfe as an eQTL fo the CETP gene in seve al tssues (P-value ange 6.3×10 r r ^-10to 4.1×10^-5).

r

Rs12720922 was an eQTL fo NLRC5, but not fo r CETP. Howeve , we founr d rs1864163, which is in

d r d r r r

mo e ate linkage isequilib ium with s12720922 (LD=0.65), to be an eQTL fo CETP in whole blood (P=8.5×10^-4, efect size A-allele 3.33). The third lea SNP, i.e. s1968905, was not i entfe as and r d d

r

eQTL fo CETP in the stu ie tssues, neithe we e any va iants in st ong linkage isequilib ium withd d r r r r d r this SNP.

3.4 M den elian an omizaton r d 202

203

204 205 206 207 208 209 210

211 212 213 214

215

216 217 218 219 220 221 222 223

224

r r r M d r d r d r Figu e 2 shows the esults f om the en elian an omizaton analyses on CAD isk an se um

d d r d r d dd r

lipi an ApoB concent atons. We ha a powe of 0.90 to etect an o s ato of 1.04 with

M d r d r d r

conventonal en elian an omizaton analyses (which makes use of a fo mal weighte genetc isk

r d r

sco e), when taking into account an alpha of 0.05, the explaine va iance of the SNPs that compose

d M d r

the GRS (i.e. 16.4%), an the sample size of the CARDIoGRA plusC4D 1000Genomes stu y. Pe 1

r d r d r r dd r r r

µg/mL inc ease in genetcally- ete mine se um CETP concent aton the o s ato fo CAD isk was

r d r dd r r r

1.08 (95% CI: 0.94, 1.23). Fo the lea SNPs sepa ately, o s atos we e 1.11 (95% CI: 1.03, 1.17) fo rs247616, 1.08 (95% CI: 1.01, 1.15) fo s12720922, an 0.89 (95% CI: 0.72, 1.09) fo s1968905. Fo r r d r r r

d dd r r r r d

Taq1B an -629C>A, o s atos fo CAD isk we e 1.02 (95% CI 1.01, 1.04) an 1.03 (95% CI 1.01,

r r

1.05), espectvely (Supplementa y table 4).

r d r d r r d

A 1 µg/mL inc ease in genetcally- ete mine se um CETP concent aton was associate with d rec eased total choleste ol concent aton, i.e. -0.14 (95% CI: -0.22, -0.05) mmol/L, anr r d HDL-C

r d r d r

concent aton, i.e. -0.23 (95% CI: -0.26, -0.20) mmol/L, while it was associate with inc ease se um

r d r

LDL-C concent aton, i.e. 0.08 (95% CI: 0.00, 0.16) mmol/L, an ApoB concent aton, i.e. 0.02 (95% CI:

d r d r r d r

0.00, 0.03) g/L. Genetcally- ete mine se um CETP concent aton was not associate with se um

r r d r r

t iglyce i es concent aton, i.e. 0.02 (95% CI -0.05, 0.09) mmol/L. Supplementa y table 5 shows the results f om the r Men elian an omizaton analysis withd r d data f om GLGC. The esults fo total r r r

r r r d d r r r r r

choleste ol, t iglyce i es an LDL-C concent atons f om GLGC we e compa able with the esults

r d r r r r d

f om the NEO stu y. Efect sizes fo HDL-C we e la ge in GLGC than in the NEO stu y.

225 226 227 228 229 230 231 232 233 234

235 236 237 238 239 240 241 242 243

244

4 Discussion

r r r r d d d r d r

With this f st la ge GWAS on se um CETP concent aton, we i entfe an eplicate th ee

d d

in epen ent SNPs, all mapping to the CETP region. These th ee va iants, notably s12720922, r r r rs247616 and rs1968905, explaine 16.4% of the total va iaton in se um CETP concent aton. Efectd r r r

d r r r r r

sizes of all lea SNPs we e la ge, with s12720922-A having the la gest efect on se um CETP: +0.35

r dd r d d r d r r

µg/mL pe a itonal isk allele. Also, we showe that genetcally- ete mine va iaton in ci culatng

d r r d r r

CETP associates with a stepwise substantal ec ease in HDL-C concent aton, a mo e ate inc ease in

d r d rd r r

LDL-C an ApoB concent aton, an a conco ant 8% inc ease in CAD isk.

d r d d

We foun th ee in epen ent SNPs in theCETP region that la gely explaine CETP concent aton. r d r

r r r d d r r

The associaton of the s12720922 va iant with ci culatng CETP, bloo lipi s o isk of CAD has, to the

r d r d r d r

best of ou knowle ge, neve been esc ibe befo e. In the eQTL stu ies that we consi e e ,d d r d^{25, 26} rs12720922 was not epo te as an eQTL fo r r d r CETP. Howeve , s1864163, which is in mo e ater r d r

d r r d r

linkage isequilib ium with s12720922 (LD=0.65), was foun to be an eQTL fo CETP in whole d

bloo ,²⁶although not in the live . r²⁵Despite the absence of a i ect associaton between s12720922 d r r

d r d r r r

an CETP mRNA levels, it is not ule out that s12720922 afects CETP levels via t ansc iptonal regulaton. Possibly, we we e not able to i entfy eQTLs of s12720922 for d r r CETP due to the low

d r d d r d d d r

sample size of eQTL stu ies in live tssue, an the iluton that is int o uce by consi e ing whole r r

live exp ession, sinceCETP is specifcally exp esse by hepatc mac ophages (i.e. Kupfe cells).r d r r ²⁹The

d d d d r d r r r

secon in epen ent lea SNP, s247616, is locate in the p omoto egion of the CETP gene.³⁰ This

d r r r r

SNP has also not been associate with se um CETP concent aton befo e, but it has p eviously been

r r r d d r d

shown that the mino allele of this va iant ( s247616-T) is associate with ec ease CETP mRNA

r r d r d r

exp ession in human live an inc ease HDL-C concent atons^30-33, which is in line with ou fn ings. r d

rd d d r r r r d d

The thi i entfe va iant, s1968905, was epo te by one stu y to associate with HDL-C

r r d r r

concent aton specifcally in Af icans, but has not been linke to se um CETP concent aton r

p eviously.³⁴ Of note, s1968905 is in st ong linkage isequilib ium (LD=0.89) with s1801706 (i.e. r r d r r 245

246 247 248 249 250 251 252

253 254 255 256 257 258 259 260 261 262 263 264 265 266

267 268 269

r r d r r r d

G84A), which has been epo te as a isk facto fo CAD in South In ians.³⁵Inte estngly, s1801706 isr r

d r d r

locate in the 3' unt anslate egion (3' UTR) of the CETP gene, suggestng involvement in

r r r

post ansc iptonal egulaton.³⁶ In a dditon to these lea SNPs, we foun a suggestve signal (i.e.d d rs117427818) that coul be a potental fou th hit in thed r CETPgene. Its statstcal signifcance was,

r r r d d d d d d r

howeve , la gely e uce upon con itonal analyses. In ee , this SNP is in linkage isequilib ium

d d d d d r

with (one of) the lea SNPs an not completely in epen ently associate with se um CETP

r d r r

concent aton (highest linkage isequilib ium with s12720922; LD=0.20).

r d r

To obtain insight in the ole of LDL-C an HDL-C in the causal associaton between se um CETP

d r r r d M d r d r d r

an CAD isk, we pe fo me en elian an omizaton analyses on se um lipi concent atons. A

r d r d r r d d r

highe CETP GRS was associate with a la ge ec ease in HDL-C concent aton an a mo e ate

r r r d r d r

inc ease in LDL-C concent aton. The st ong associaton of genetcally- ete mine se um CETP

r r r r d r d

concent aton with HDL-C concent aton is p obably pa tally explaine by a sha e genetc

r d rr r r

backg oun of these two phenotypes, as the genetc co elaton between se um CETP concent aton

d r r d r d d

an HDL-C concent aton was elatvely high, which in icates pleiot opy. It shoul be note ,

r r d r

howeve , that HDL-C has been obse vatonally, but not causally associate with CVD isk.¹⁰ Thus,

d r d r r d d r

although a genetcally- ete mine inc ease in se um CETP is causally associate with a ec ease in

r d r d r

HDL-C concent aton, this likely oes not explain the associaton between se um CETP an CAD isk.

r d r r r r d

On the othe han , a causal, p opo tonal, log-linea associaton between LDL-C concent aton an

r r d

CAD isk has been f mly establishe .³⁷ rIn a p eviously pe fo mer r d Men elian an omizaton stu yd r d d

dd r r r r

using an LDL-C GRS, it was shown that the o s ato fo CAD isk was 1.68 (1.51-1.87) pe 1 SD

r

inc ease in LDL-C (i.e.0.98 mmol/L).³⁸To compa e, we showe that pe 1 µg/mL inc ease in se umr d r r r

r r d d dd r r r

CETP, LDL-C concent aton inc ease with 0.08 mmol/L an the o s ato fo CAD isk was 1.08.

r d r r r r d r

Thus, exp esse pe 0.98 mmol/L inc ease in LDL-C concent aton, we obse ve 1.98 tmes inc ease

r r d

in CAD isk using the CETP GRS, which is compa able with the efect estmate foun with the LDL-C GRS.³⁸Taken togethe , ou stu y suggests that the causal associaton between CETP concent atonr r d r

d r d r r r r

an CAD isk may be explaine by efects on LDL-C concent aton. Inte estngly, a ecent la ge meta- 270

271 272 273 274 275 276

277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295

Men elian an omizaton analysis by Fe enced r d r et al.,¹⁵ din icate that ApoB concent aton is an evend r

r r d r r

mo e impo tant causal link between CETP an CAD isk than LDL-C concent aton. This implies that

r r d r r d

an inc ease in the absolute numbe of VLDL, IDL an LDL (i.e. non-HDL) pa tcles, as efecte by

r d r d r r

ApoB concent aton, ue to inc ease ci culatng CETP may explain the associaton with CAD isk, rathe than the amount of choleste ol in LDL pa tcles.r r r

r d r r M d r d d r

Ou fn ings a e in line with this ecent meta- en elian an omizaton analysis, inclu ing ove a

r d r r r d

100,000 pa tcipants that showe compa able efects fo a CETP GRS on CAD isk, LDL-C an ApoB r

concent aton.¹⁵ In that meta-analysis, a CETP GRS was compose of eight can i ate SNPs selecte d d d d r

f om the CETP gene with a fo wa r rd con itonal eg ession analysis on HDL-C concent aton. In thed r r r

r d d d r d r

p esent stu y, we i entfe th ee ife ent CETP SNPs that in epen ently ete mine ci culatng d d d r r

r r r d d

CETP concent aton by using a hypothesis-f ee app oach (i.e. GWAS). These GWAS-i entfe SNPs

r r r d r r d M d

a e the efo e i ect genetc inst uments to stu y the causal efects of CETP in en elian ran omizaton. Although none of the lea SNPs that we i entfe with GWAS we e inclu e in thed d d d r d d

d r

GRS compose by Fe ence et al.¹⁵, we obse ve that the two st ongest lea SNPs f om ou GWASr d r d r r

r r d r r d r d d

we e in high ( s247616) to mo e ate ( s12720922) linkage isequilib ium with the eight can i ate

d d d d

SNPs of that GRS. This in icates that the GRS compose of can i ateCETPSNPs is a eliable genetc r

r d d r r r r d r

inst ument to stu y the causal efects of CETP, an ou esults the efo e exten this ecent meta- Men elian an omizaton analysis.d r d ¹⁵

d r d r r

Although we showe a causal associaton between CETP concent aton an CAD isk, the th ee

r d d r d r rd r dd

inital CETP inhibito s i not e uce the isk of ca iovascula events when given in a iton to statn

r r r r r d d r

t eatment. In fact, clinical t ials with those CETP inhibito s we e even te minate ue to of-ta get

r r r d r d r

efects (to cet apib) o a lack of efcacy ( alcet apib an evacet apib).^5-7 Dalcet apib ha minimalr d

r

efects in LDL-C concent aton possibly explaining its futlity.⁶Evacet apib i signifcantly e uce LDL-r d d r d

r d d rd d r

C concent aton, but i not evoke a conco ant ec ease in ApoB,³⁹ din icatng unfavou able LDLr

r r d r r r r r

pa tcle emo elling athe than emoval f om the ci culaton.^{37, 40}This explanaton fo the failu e of r r 296

297 298 299 300

301 302 303 304 305 306 307 308 309 310 311 312 313

314 315 316 317 318 319 320

r r r M d r d d r

the evacet apib t ial is in line with a ecent meta- en elian an omizaton stu y by Fe ence et al.¹⁵

r d d r r r d

Data f om that stu y in icate that the success of CETP inhibito s when p esc ibe on top of statn

r d d r r d r r

t eatment is epen ent on thei capability to e uce the absolute numbe non-HDL pa tcles as refecte by a e ucton in ApoB concent aton. A e ucton in the LDL-C concent aton th ough CETPd r d r r d r r

rd r d r

inhibiton may thus only be benefcial when a conco ant e ucton in ApoB concent aton is d

achieve .¹⁵ In ee , anacet apibd d r d d i show a conco ant rd r de ucton in non-HDL and ApoB

r d r d d r r r r r r

concent aton of -18%, which was accompanie by a e uce ate ato fo majo co ona y events of

0.91 (95% CI 0.85, 0.97).⁴

r d d r d dd r

Ou stu y may have ha insufcient statstcal powe to i entfy a itonal va iants with small

r r r r d d r r

efects on se um CETP concent aton o with low allele f equencies. As we i not eplicate ou

d dd r d r

GWAS fn ings in a itonal hete ogeneous populatons, cauton shoul be taken when ext apolatng

r r d r d d

the esults to othe populatons. Also, espite the simila associatons between the lea SNPs an

r d d r r r r

LDL-C concent atons in GLGC an the NEO stu y, efect sizes fo HDL-C concent atons we e highe

d r d

in GLGC. A possible explanaton might involve ife ences in the compositon of the stu y

d r r

populatons, as GLGC is a meta-analysis of a wi e va iety of coho ts.

d d d r d r d d

In conclusion, with a GWAS, we i entfe an eplicate th ee in epen ent SNPs mapping to the CETP gene that togethe explaine 16.4% of the total va iaton in se um CETP concent aton,r d r r r

r r r d r d M d

which shows that se um CETP concent aton is st ongly genetcally ete mine . Using en elian ran omizaton, we showe that 1 µg/mL inc ease in se um CETP causally associates with a la ged d r r r d rec ease in HDL-C choleste ol of -0.23 mmol/L, mo e ate inc eases in LDL-C concent aton of 0.08 r d r r r

d r d dd r r r

mmol/L an ApoB concent aton of 0.02 g/L, an an o s ato of 1.08 fo CAD isk. While HDL-C is

r r r d r d r

not a causal isk facto fo CAD, it has been unequivocally emonst ate that LDL-C lowe ing is

r r d r r r r r r d r

p opo tonally associate with a lowe CAD isk. The efo e, the esults of ou stu y a e fully

r d r

consistent with the noton that CETP concent aton is causally associate with CAD th ough LDL-C.

321 322 323 324 325 326 327 328

329 330 331 332 333 334 335

336 337 338 339 340 341 342 343 344

d Acknowle gments

r r r d d d r d r r

We exp ess ou g attu e to all in ivi uals who pa tcipate in the NEO stu y. We a e g ateful to all

r r r r r r r r r

pa tcipatng gene al p acttone s fo invitng eligible pa tcipants. We fu the mo e thank Pat van

d r r r r d r rd d r r r r

Beelen an all esea ch nu ses fo collectng the ata, Pet a Noo ijk an he team fo labo ato y

d d r d r d d

management an DNA isolaton, an Ingebo g e Jonge fo all ata management of the NEO stu y.

r r d r r r r r r r

We since ely thank Ch is van e Bent fo pe fo ming the se um CETP concent aton measu ements.

d r d r d r

The genotyping in the NEO stu y was suppo te by the Cent e Natonal e Génotypage (Pa is,

r d d r r r r r d r d

F ance), hea e by Jean-F ancois Deleuze. Data on co ona y a te y isease have been cont ibute by

M r d d d d r

CARDIoGRA plusC4D investgato s an have been ownloa e f om

M

www.CARDIOGRA PLUSC4D.ORG.

d r

Fun ing sou ces

d r d r r d rd

The NEO stu y was suppo te by the pa tcipatng Depa tments, the Division an the Boa of

r r d r M d r d d r r r

Di ecto s of the Lei en Unive sity e ical Cente , an by the Lei en Unive sity, Resea ch P ofle

r r d r M d r d r r rd

A ea ‘Vascula an Regene atve e icine’. L.L. Blauw was suppo te by a g ant f om the Boa of

r r d r M d r r d

Di ecto s of the Lei en Unive sity e ical Cente . Y. Wang is suppo te by the Dutch Science

r r d r r r d M

O ganizaton [The Nethe lan s O ganisaton fo Health Resea ch an Development (Zon W) VENI

r r d r d d r

G ant 91617027]. D. van Heemst was suppo te by the Eu opean Commission fun e p oject

M M r r d

HU AN [Health-2013-INNOVATION-1-602757]. D. ook-Kanamo i is suppo te by the Dutch Science

r M r d r

O ganizaton [Zon W VENI G ant 91614023]. P.C.N. Rensen is an Establishe Investgato of the

r d

Dutch Hea t Foun aton [2009T038].

r Disclosu es None 345

346 347 348 349 350 351 352 353 354 355

356 357 358 359 360 361 362 363 364

365 366

r Refe ences

1. Tall AR. Plasma choleste yl este t ansfe p otein.r r r r r Jou nal of lipir d resea chr .1993;34:1255- 1274

2. Chapman M J, Le Gof W, Gue inr M , Kontush A. Choleste yl este t ansfe p otein: At the r r r r r

r d d r r d r

hea t of the acton of lipi -mo ulatng the apy with statns, fb ates, niacin, an choleste yl

r r r r r

este t ansfe p otein inhibito s.Eu opean hea t jou nalr r r .2010;31:149-164

3. Weste te pr r M , van e Hoogt CC, e Haan W, Ofe man EH, Dallinga-Thie G , Jukema JW, d r d r M M

Havekes L , Rensen PC.Choleste yl este t ansfe p otein ec eases high- ensity lipop oteinr r r r r d r d r

d r r r r d

an seve ely agg avates athe oscle osis in apoe*3-lei en mice.A te ioscle osis, th ombosis,r r r r

d r

an vascula biology .2006;26:2552-2559

4. Bowman L, Hopewell JC, Chen F, Wallen szus K, Stevens W, Collins R, et al. Efects of d

r r r r d

anacet apib in patents with athe oscle otc vascula isease.The New Englan jou nal ofd r d

me icine .2017;377:1217-1227

5. Ba te PJ, Caulfelr r d M r, E iksson M r, G un y S , Kastelein JJ, Komaj ad M d M , et al. Efects of

r r r r r r

to cet apib in patents at high isk fo co ona y events.The New Englan jou nal of me icined r d . 2007;357:2109-2122

6. Schwa tz GG, Olsson AG, Abtr M , Ballantyne C , Ba te PJ, B umm J, et al. Efects of M r r r

dalcet apib in patents with a ecent acute co ona y syn ome.r r r r dr The New Englan jou nal ofd r d

me icine .2012;367:2089-2099

7. Lincof A , Nicholls SJ, Riesmeye JS, Ba te PJ, B ewe HB, Fox KAA, et al. Evacet apib an M r r r r r r d

rd r r r d

ca iovascula outcomes in high- isk vascula isease.The New Englan jou nal of me icined r d . 2017;376:1933-1942

8. Eyvazian VA, F ishman WH. Evacet apib: Anothe cetp inhibito fo r r r r r dyslipi emia with nod clinical beneft.Ca iology in eviewrd r .2017;25:43-52

9. Go on DJ, P obstiel JL, Ga ison RJ, Neaton JD, Castelli WP, Knoke JD, et al. High- ensityrd r d rr d

r r d rd r d r r r d

lipop otein choleste ol an ca iovascula isease. Fou p ospectve ame ican stu ies.

r

Ci culaton .1989;79:8-15

10. Voight BF, Peloso G , O ho- elan e M r M d r M r, F ikke-Schmi t R, Ba balicd r M , Jensen M K, et al.

d r d r rd r d r d d

Plasma h l choleste ol an isk of myoca ial infa cton: A men elian an omisaton stu y.

d d

Lancet (Lon on, Englan ) .2012;380:572-580

11. Wang Y, van e Tuin S, Tjee ema N, van Dam AD, Rensen SS, Hen ikx T, et al. d r rd dr Plasma

r r r r r r d d r d r r

choleste yl este t ansfe p otein is p e ominantly e ive f om kupfe cells.Hepatology r Md

(Baltmo e, .) .2015;62:1710-1722

12. Thompson A, Di Angelantonio E, Sa wa N, E qou S, Saleheen D, Dullaa t RP, et al. Associaton r r r r

r r r r r d d d

of choleste yl este t ansfe p otein genotypes with cetp mass an actvity, lipi levels, an r r r

co ona y isk.Jama .2008;299:2777-2788

13. Niu W, Qi Y. Ci culatng choleste yl este t ansfe p otein an co ona y hea t isease: r r r r r r d r r r d Men elian an omizaton meta-analysis.d r d Ci culaton. Ca iovascula genetcsr rd r .2015;8:114- 121

14. Johannsen TH, F ikke-Schmi t R, Schou J, No estgaa r d rd rd BG, Tybjae g-Hansen A. Genetcr

r d d r d d r

inhibiton of cetp, ischemic vascula isease an mo tality, an possible a ve se efects.

r r rd

Jou nal of the Ame ican College of Ca iology .2012;60:2041-2048

15. Fe ence BA, Kastelein JJP, Ginsbe g HN, Chapmanr r M J, Nicholls SJ, Ray KK, et al. Associaton of

r r d r d r d

genetc va iants elate to cetp inhibito s an statns with lipop otein levels an

rd r r

ca iovascula isk.Jama .2017;318:947-956

16. Thompson JF, Li a r M E, Du ham LK, Cla k RW, Bambe ger r r r M MJ, ilos P . Polymo phisms in the M r

d d d

cetp gene an associaton with cetp mass an h l levels.Athe oscle osisr r .2003;167:195-204 17. Thompson JF, Woo LS, Picke ing EH, Dechai o B, Hy e CL. High- ensity genotyping an d r r d d d

functonal snp localizaton in the cetp gene.Jou nal of lipir d resea chr .2007;48:434-443 367

368 369 370 371 372 373 374 375 376 377 378 379 380 381 382 383 384 385 386 387 388 389 390 391 392 393 394 395 396 397 398 399 400 401 402 403 404 405 406 407 408 409 410 411 412 413 414 415

18. Wille CJ, Schmi t E , Sengupta S, Peloso G , Gustafsson S, Kanoni S, et al. Discove y anr d M M r d refnement of loci associate with lipi levels. d d Natu e genetcsr .2013;45:1274-1283

19. Nikpay M , Goel A, Won HH, Hall L , Willenbo g C, Kanoni S, et al. M r A comp ehensive 1,000r

d d r r r r d

genomes-base genome-wi e associaton meta-analysis of co ona y a te y isease.Natu er genetcs .2015;47:1121-1130

20. d Me utse t R, en Heijer d r M , Rabelink TJ, Smit JW, Romijn JA, Jukema JW, et al. The

r d d d d d d d

nethe lan s epi emiology of obesity (neo) stu y: Stu y esign an ata collecton.Eu opeanr

r d

jou nal of epi emiology .2013;28:513-523

21. Blauw LL, e d Mutse t R, Lamb HJ, e Roos A, Rosen aal FR, Jukema JW, et al. Se um cetpr d d r

r d r d d

concent aton is not associate with measu es of bo y fat: The neo stu y.Athe oscle osisr r . 2016;246:267-273

22. Yang J, Lee SH, Go dd rd M a E, Vissche P . Gcta: A tool fo genome-wi e complex t aitr M r d r analysis.Ame ican jou nal of human genetcsr r .2011;88:76-82

23. D ayna D, Lawn R.r Multple fps at the human choleste yl este t ansfe p otein (cetp) locus. r r r r r r d r r

Nucleic aci s esea ch .1987;15:4698

24. Dachet C, Poi ie O, Cambien F, Chapman J, Rouis r r M . New functonal p omote r r

r r r r r r r d

polymo phism, cetp/−629, in choleste yl este t ansfe p otein (cetp) gene elate to cetp

d d r r

mass an high ensity lipop otein choleste ol levels.Role of Sp1/Sp3 in T ansc iptonal r r Regulaton .2000;20:507-515

25. The GC. The genotype-tssue exp ession (gtex) p oject.r r Natu e genetcsr .2013;45:580-585 26. West a HJ, Pete sr r M J, Esko T, Yaghootka H, Schu mann C, Ketunen J, et al. Systematc r r

d r dr r d

i entfcaton of t ans eqtls as putatve ive s of known isease associatons.Natu er genetcs .2013;45:1238-1243

27. B ionr M J, Shakhbazov K, Vissche P . Calculatng statstcal powe in men elian r M r d ran omizaton stu ies.d d Inte natonal jou nal of epi emiologyr r d .2013;42:1497-1501

28. Rugge B, Balshem H, Sehgal R, al. e. Sc eening an t eatment of subclinical hypothy oi ism r d r r d

r r r d r d r r r r d

o hype thy oi ism [inte net]. Rockville (m ): Agency fo healthca e esea ch an quality

r r r

(us); (compa atve efectveness eviews, no. 24.) available f om:

Htps://www.Ncbi.Nlm.Nih.Gov/books/nbk83496/. 2011 Oct

29. Wang Y, Tuin Sv , Tjee ema N, Dam ADv, Rensen SS, Hen ikx T, Be bée JFP, Atanasovska B, d rd dr r

r M r r r r M r rd

Fu J, Hoekst a , Bekke ing S, Riksen NP, Buu man WA, G eve JW, Hofe H, Shi i Sve lov‐

M r M r r

R, eije OC, Smit JWA, Havekes L , Dijk KWv, Rensen PCN. Plasma choleste yl este

r r r r d d r d r r

t ansfe p otein is p e ominantly e ive f om kupfe cells.Hepatology .2015;62:1710-1722 30. Papp AC, Pinsonneault JK, Wang D, Newman LC, Gong Y, Johnson JA, et al. Choleste yl este r r

r r r r r d d d d

t ansfe p otein (cetp) polymo phisms afect m na splicing, h l levels, an sex- epen ent

rd r r

ca iovascula isk.PloS one .2012;7:e31930

31. Suhy A, Ha tmann K, Newman L, Papp A, Tonef T, Hook V, et al. Genetc va iants afectng r r

r r r r r r

alte natve splicing of human choleste yl este t ansfe p otein.Biochemical an biophysical d resea ch communicatonsr .2014;443:1270-1274

32. Smith EN, Chen W, Kahonen M , Ketunen J, Lehtmaki T, Peltonen L, et al. Longitu inal d

d rd r d r r r d

genome-wi e associaton of ca iovascula isease isk facto s in the bogalusa hea t stu y.

PLoS genetcs .2010;6:e1001094

33. Suhy A, Ha tmann K, Papp AC, Wang D, Sa ee W. Regulaton of choleste yl este t ansfe r d r r r r

r r r r d d r d

p otein exp ession by upst eam polymo phisms: Re uce exp ession associate with rs247616.Pha macogenetcs an genomicsr d .2015;25:394-401

34. Pi im D, Wang X, Niemsi i V, Ra wan ZH, Bunke CH, Hokanson JE, et al. Resequencing of ther r d r

r d r r r d r

cetp gene in ame ican whites an af ican blacks: Associaton of a e an common va iants

d r

with h l-choleste ol levels.Metabolism: clinical an expe imental d r .2016;65:36-47

35. Ganesan M , Nizamudd in S, Katkam SK, Kuma aswami K, Hosa UK, Lobo LL, et al. C.*84g>ar d

d r r r r d d

mutaton in cetp is associate with co ona y a te y isease in south in ians.PloS one . 2016;11:e0164151

416 417 418 419 420 421 422 423 424 425 426 427 428 429 430 431 432 433 434 435 436 437 438 439 440 441 442 443 444 445 446 447 448 449 450 451 452 453 454 455 456 457 458 459 460 461 462 463 464 465 466

36. Matoulkova E, Michalova E, Vojtesek B, H stka R. The ole of the 3' unt anslate r r r d region in

r r r r r

post-t ansc iptonal egulaton of p otein exp ession in mammalian cells.RNA biology . 2012;9:563-576

37. Fe ence BA, Ginsbe g HN, G aham I, Ray KK, Packar r r rd CJ, B ucke t E, et al. Low- ensity r r d

r r r rd r d d r

lipop oteins cause athe oscle otc ca iovascula isease. 1. Evi ence f om genetc,

d d d r r

epi emiologic, an clinical stu ies. A consensus statement f om the eu opean

r r

athe oscle osis society consensus panel.Eu opean hea t jou nalr r r .2017;38:2459-2472 38. White J, Swe low DI, P eiss D, Fai hu st-Hunte Z, Keatng BJ, Asselbe gs FW, et al. rd r r r r r

d r r r r r r r d d d

Associaton of lipi f actons with isks fo co ona y a te y isease an iabetes.JA AM rd

ca iology .2016;1:692-699

39. Eli lilly an company. Lilly to iscontnue evelopment of evacet apib fo high- isk d d d r r r

r r rd r d

athe oscle otc ca iovascula isease.htps://investo .lilly.com/ elease etail.cfm?r r d ReleaseID=936130, Accesse 15 Octobe 2015d r .2015

40. Sni e man AD, Islam S, Yusuf S,d r McQueen M J. Disco ance analysis of apolipop otein b anrd r d

d r r r r rd r r r r

non-high ensity lipop otein choleste ol as ma ke s of ca iovascula isk in the inte hea t d

stu y.Athe oscle osisr r .2012;225:444-449 467

468 469 470 471 472 473 474 475 476 477 478 479 480 481 482 483

Table 1 Cha acte istcs of the iscove y anr r d r d replicaton coho t f om the Nethe lan s Epi emiology ofr r r d d d

Obesity (NEO) stu y.

r r

Cha acte istcs Discove y coho tr r Replicaton coho tr r r

Numbe of pa tcipants 4,248* 1,458^†

r

Age (yea ) 56 (51, 61) 57 (51, 61)

Women 2,148 (50.6%) 818 (56.1%)

d d

Bo y mass in ex (kg/m²) 30.3 (28.4, 33.0) 25.6 (23.2, 28.2)

d r dr r

Lipi -lowe ing ug use s 745 (17.5%) 151 (10.4%)

r r

Fastng se um concent atons

CETP (µg/mL) 2.50 (0.67) 2.43 (0.64)

r

Total choleste ol (mmol/L) 5.66 (1.08) 5.69 (1.07)

r

LDL-choleste ol (mmol/L) 3.58 (0.99) 3.56 (0.98)

r

HDL-choleste ol (mmol/L) 1.38 (0.38) 1.58 (0.46)

r r d

T iglyce i es (mmol/L) 1.34 (0.95, 1.87) 1.00 (0.71, 1.45)

r r d d r r r r r d r d d

Results a e p esente as me ian (inte qua tle ange) fo not no mally ist ibute ata, mean (SD)

r r r

o numbe (pe centage).

M d r r r r r r

* issing ata: n=13 fo total choleste ol concent aton, n=14 fo HDL-choleste ol concent aton,

r r d d r r

t iglyce i es an LDL-choleste ol concent aton

d

† No missing ata 484

485

486 487

488 489

490

Table 2Summa y statstcs of the associatons of the th ee in epen ent lea SNPs that eache genome-wi e signifcance an seven suggestve signals withr r d d d r d d d

r r

se um CETP concent aton.

r r

Discove y coho t (n=4,248) Replicaton coho t (n=1,458)r r

Ch SNP Positon Locaton Closest gene Co ing/d

Non- d co ing allele

d Co ing allele

r f equency*

Imputaton quality

Efect size r pe allele^† (µg/mL)

SE P-value Con itoned d P-value^‡

Efect r size pe allele^† (µg/mL)

SE P-value

d Lea SNPs

16 rs247616 56989590 Inte genicr CETP C/T 0.67 1 0.32 0.015 3.98×10^-100 1.86×10^-64 0.31 0.024 1.24×10^-37

16 rs12720922 57000885 Int onr CETP A/G 0.17 0.98 0.35 0.019 3.48×10^-74 6.68×10^-13 0.36 0.030 3.27×10^-33

16 rs1968905 57010948 Int onr CETP G/T 0.82 0.85 0.12 0.02 4.12×10^-9 1.66×10^-12 0.098 0.031 1.80×10^-3

Suggestve SNPs

2 rs185550357 50249349 Int onr NRXN1 C/T 0.007 0.70 0.63 0.12 3.94×10^-7 4.57×10^-7 0.064 0.15 0.67

3 rs6442310 12358230 Int onr PPARG A/T 0.54 0.92 0.08 0.015 5.52×10^-8 6.13×10^-8 0.0044 0.024 0.85

8 ch 8:19811023:Ir 19811023 Int onr LPL ATG/A 0.12 0.87 0.12 0.023 1.45×10^-7 1.59×10^-7 -0.018 0.038 0.63

9 rs3094377 136312119 Int onr ADA TS13M T/C 0.03 0.41 0.35 0.064 7.65×10^-8 9.58×10^-8 -0.18 0.11 0.098

10 rs12253367 62003462 Int onr ANK3 G/A 0.17 0.94 0.10 0.019 1.08×10^-7 1.18×10^-7 0.0051 0.031 0.87

16 rs117427818 57010486 Int onr CETP T/C 0.050 0.75 0.46 0.038 3.26×10^-33 1.02×10^-7 0.54 0.061 1.93×10^-18

20 rs150904289 18845428 Inte genicr G/A 0.11 0.79 0.13 0.026 6.31×10^-7 6.87×10^-7 -0.0084 0.042 0.84

r d r d

Th eshol fo genome-wi e signifcance is 5×10^-8, base on the con itone P-valued d d ^‡. Th eshol fo the suggestve signals is 1×10r d r ^-6. Th eshol fo eplicatonr d r r is 0.05.

492 493

494 495

d r r r d d r r r

* In the iscove y coho t; † Beta coefcient exp esse as the ife ence in se um CETP concent aton; ‡P-value con itone on the top lea SNPs using step-d d d d

wise con itonal analysis

r r

Ch , ch omosome 496

497

498

Table 3Exp ession quanttatve t ait loci (eQTL) fo the th ee GWAS-i entfe lea SNPs.r r r r d d d

SNP Assessed

Allele

Gene Chr P-value Efect size

Tissue Database

rs247616 C NLRC5 16 9.5×10^-14 0.34 T ansfo me fb oblastsr r d r GTEx

rs247616 C CETP 16 6.3×10^-10 0.30 Lung GTEx

rs247616 C CETP 16 1.1×10^-7 0.41 T ansve se colonr r GTEx

rs247616 C CETP 16 1.4×10^-6 0.45 Te minal Ileumr GTEx

rs247616 C CETP 16 3.6×10^-6 0.32 Liver GTEx

rs247616 C CETP 16 7.9×10^-6 0.30 Esophagus (mucosa) GTEx

rs247616 C CETP 16 9.3×10^-6 0.41 Panc easr GTEx

rs247616 C BBS2 16 1.7×10^-5 0.37 Ce ebella Hemisphe er r r GTEx

rs247616 C CETP 16 4.1×10^-5 4.10 Whole blood Bloo eQTLd

r r

b owse rs12720922 A NLRC5 16 1.9×10^-6 0.24 T ansfo me fb oblastsr r d r GTEx

rs1968905 G - 16 - - -

499

500

501

r

Figu e 1 –Log(P-value) plot fo the genome-wi e associaton stu y in the iscove y coho t (n=4,248). r d d d r r

TheCETPgene is locate on ch omosome 16. The e line ep esents the th eshol fo genome-wi e d r r d r r r d r d signifcance (P<5×10^-8). The blue line ep esents the th eshol fo suggestve signals (P<1×10 r r r d r ^-6).

502

503

504 505

506

r

Figu e 2 Results f om the r Men elian an omizaton stu y on co ona y a te y isease in thed r d d r r r r d

M r r d r

CARDIoGRA plusC4D 1000 Genomes Conso tum (60,801 cases; 123,504 cont ols), an on se um

d d r r r d d d

lipi an lipop otein B concent atons in the total Nethe lan s Epi emiology of Obesity (NEO) stu y populaton (n=5,706*).

M d r r r r r r

* issing ata: n=13 fo total choleste ol concent aton, n=14 fo HDL-choleste ol concent aton,

r r d d r r r r r

t iglyce i es an LDL-choleste ol concent aton, n=41 fo apolipop otein B concent aton. Results

r d d r d

we e a juste fo age an sex.

r r r r d r r r r r r r

CAD, co ona y a te y isease; CETP, choleste yl este t ansfe p otein; IVW, inve se-va iance d

weighte . 507

508 509 510

511

512 513 514

515 516