ARTICLE
Recei ed 3 Jun 2015 |Accep ed 24 Feb 2016 |Published 23 Ma 2016
Genome-wide s udy o ci cula ing me aboli es
iden ifies 62 loci and e eals no el sys emic e ec s
o LPA
Johannes Ke unen e al.#
Genome-wide associa ion s udies ha e iden ified nume ous loci linked wi h complex
diseases, o which he molecula mechanisms emain la gely unclea . Comp ehensi e
molecula p ofiling o ci cula ing me aboli es cap u es highly he i able ai s, which can help
o unco e me abolic pa hophysiology unde lying es ablished disease a ian s. We conduc
an ex ended genome-wide associa ion s udy o gene ic influences on 123 ci cula ing
me abolic ai s quan ified by nuclea magne ic esonance me abolomics om up o 24,925
indi iduals and iden i y eigh no el loci o amino acids, py u a e and a y acids. The LPA
locus link wi h ca dio ascula isk exemplifies how de ailed me abolic p ofiling may in o m
unde lying ae iology ia ex ensi e associa ions wi h e y-low-densi y lipop o ein and igly-
ce ide me abolism. Gene ic fine mapping and Mendelian andomiza ion unco e wide-sp ead
causal e ec s o lipop o ein(a) on o e all lipop o ein me abolism and we assess po en ial
pleio opic consequences o gene ically ele a ed lipop o ein(a) on di e se mo bidi ies ia
elec onic heal h-ca e eco ds. Ou findings s eng hen he a gumen o sa e LPA- a ge ed
in e en ion o educe ca dio ascula isk.
DOI: 10.1038/ncomms11122 OPEN
Co espondence and eques s o ma e ials should be add essed o J.K. (email: johannes.ke unen@oulu.fi).
#A ull lis o au ho s and hei a filia ions appea s a he end o he pape .
NATURE COMMUNICATIONS | 7:11122 | DOI: 10.1038/ncomms11122 | www.na u e.com/na u ecommunica ions 1
An unde s anding o he gene ic ac o s in ol ed in
sys emic me abolism and hei associa ions wi h ch onic
disease is a key objec i e, as la ge disease conso ia ha e
now unco e ed nume ous a ian s associa ed wi h me abolic
diseases1,2. Me abolic pheno ypes se e as good in e media e
ai s o a genome-wide associa ion s udy (GWAS) and
blood me aboli es can be po en ially used o disco e gene ic
de e minan s o ci cula ing me aboli es, and pa icula ly o
unde s and he me abolic con ex o disease-associa ed gene ic
a ian s. Ad ances in nuclea magne ic esonance (NMR)
spec oscopy and mass spec ome y ha e enabled analy ical
echniques ha can p o ide hund eds o quan i a i e me abolic
measu es om la ge biological sample collec ions3. GWAS me a-
analysis o me abolic measu es om hese me hodologies ha e
been pe o med, howe e , he sample sizes ha e only eached
se e al housand, which is s ill modes compa ed wi h disease
conso ia s udies4–6. The size o p e ious GWASs u ilizing
me abolic p ofiling echniques may pa ially explain he modes
new biological insigh added o known disease-associa ed
a ian s.
To o e come he challenge o small sample size, we pe o m an
expanded GWAS om ou p e ious s udy4by combining up o
24,925 indi iduals in a me a-analysis o 123 me abolic measu es.
We disco e eigh new loci o ci cula ing me aboli es. We ocus
on a new me aboli e associa ion wi h a ian s in LPA, a known
co ona y hea disease (CHD) isk locus. We ollow up he no el
associa ion by cons uc ing a s ong gene ic isk sco e o LPA
and use he isk sco e o he molecula cha ac e iza ion o he
me abolic e ec s o Lp(a) syn hesis and assessmen o causali y
o he me abolic associa ions. Finally, we pe o m e e se
gene ics using elec onic heal h eco ds oge he wi h he
gene ic isk sco e o es i LPA a ge ing ea men o
educing CHD isk would be associa ed wi h po en ial s ong
como bidi ies. To conclude, we demons a e how in e media e
pheno ypes can p o ide new biological in o ma ion o known
disease loci and how la ge mul i-omics biobank da a could be
used o in o m d ug disco e y al eady a an ea ly s age.
Resul s
Genome-wide associa ion s udy. Using he addi i e gene ic
model, we es ed o uni a ia e associa ions be ween genome-
wide single-nucleo ide polymo phism (SNP) panels impu ed o
39 million gene ic ma ke s and 123 human blood lipid and
me aboli e concen a ions quan ified by high- h oughpu NMR
spec oscopy me abolomics (Supplemen a y Table 1 o ai
in o ma ion, Me hods o analysis de ails) in 14 geno yped da a
se s de i ed om en Eu opean s udies (Fig. 1) o up o 24,925
indi iduals (Table 1 o s udy cha ac e is ics, Supplemen a y
Table 2 o s udy de ails and Supplemen a y No es 1 o s udy
desc ip ions). Coho s we e analysed indi idually and summa y
s a is ics we e combined in a me a-analysis (Me hods). Up o
12,133,295 SNPs, small inse ions and dele ions we e included in
he me a-analysis a e applying quali y con ol fil e s. All me a-
analysis esul s a e a ailable h ough URL: h p://www.compu-
a ionalmedicine.fi/da a/NMR_GWAS/. To co ec o mul iple
es ing, genome- and me abolome-wide s a is ical significance
was se o Po2.3 109, whe e he s anda d genome-wide
significance le el (5 108) is di ided by he numbe o p in-
cipal componen s (22) ha explain o e 95% o a ia ion in he
me abolomics da a. O e all, 62 loci we e significan ly associa ed
wi h a leas one me abolic measu e. Supplemen a y Fig. 1 p e-
sen s he associa ions in 2 Mb windows a ound he s onges
indi idual a ian o he 62 loci. The o es plo s o all 62 lead
a ian associa ions a e shown in Supplemen a y Fig. 2. We es ed
i he iden ified 62 loci ha bou ed addi ional independen
a ian s. In 9 ou o he 62 loci (PCSK9,LPL,PPM1K,HAL,
CETP,CILP, PLTP, APOB and LIPC), we ound a seconda y
s a is ically independen associa ion, in 2 o hese loci (APOB and
LIPC), we ound a hi d independen a ian and LIPC addi-
ionally ha bou ed a ou h independen a ian (Supplemen a y
Table 3 and Me hods). The o mal condi ional es ing was fi s
pe o med in a subse o coho s and a e condi ioning wi h
significan a ian s, he esul ing lead a ian was hen es ed
using he genomic co ela ion s uc u e in o ma ion and sum-
ma y s a is ics (Me hods). Ou co ela ion s uc u e was ob ained
om he Finnish popula ion ha has longe linkage dis-
equilib ium s uc u e han mo e ou b ed popula ions7and as a
esul ou disco e ed numbe o independen a ian s may be an
unde es ima e. The s eng h o ou app oach was o fi s op imize
he a iance explained by he nex bes a ian . Howe e , ou
app oach may esul in an unde es ima e o he numbe o
independen a ian s in a locus, as he a ian ha explains
la ges p opo ion o a iance in a ai may be agging wo o
mo e unc ional a ian s8. In con as , i he independen a ian
de ec ion elied only on co ela ion s uc u e and summa y
s a is ics, i may esul in a g oss o e es ima e on he numbe o
independen a ian s in a locus i da a a e impu ed wi h 1,000
GWAS o me abolic measu es
8 No el loci*
54 associa ions wi h known loci
Up o 24,925 indi iduals om
14 Eu opean coho s
Simila s ong associa ions o
ci cula ing Lp(a)*
E ec s Lp(a) on o e all lipop o ein
me abolism a e causal*
LPA SNP s me abolic measu es
Lp(a) s me abolic measu es
Mendelian andomiza ion
S ong associa ions wi h VLDL
and iglyce ide me abolism*
Elec onic heal h eco ds
Hypo hesis- ee disco e y
New biology o known loci
Can we make a good
gene ic ins umen o Lp(a)?
How does Lp(a) a ec
me abolism?
Is he gene ic isk sco e
associa ed wi h o he
mo bidi ies?
*New inding
Gene ic isk sco e GRSLp(a)
GRSLp(a) explains 45% o
a ia ion o ci cula ing Lp(a)*
GRSLp(a) is associa ed wi h
me abolic changes, pa icula ly
wi h VLDL*
GRSLp(a) associa es s ongly
only wi h CHD*
Sa e molecula in e en ion on
LPA mRNA o educe
ca dio ascula isk*
Figu e 1 | A genome-wide associa ion s udy o ci cula ing me aboli es.
S udy was conduc ed o elucida e he gene ic a ia ion o sys emic
me abolism and o disco e new me abolic associa ions in es ablished loci.
We also e ealed an in iguing no el ela ion be ween Lp(a) and sys emic
iglyce ide and VLDL me abolism. The eby, we highligh ed he LPA locus
and gene a ed he bes possible Lp(a) gene ic isk sco e (GRS
Lp(a)
) ha
enabled us o cla i y causal associa ions be ween Lp(a) and sys emic
iglyce ide and lipop o ein me abolism. Fu he , wi h he aid o ex ensi e
elec onic heal h-ca e eco ds, we we e able o use he GRS
Lp(a)
o show
ha Lp(a) is associa ed wi h ischaemic hea disease bu no s ongly wi h
o he mo bidi ies. Pu oge he , hese findings sugges sa e molecula
in e en ion on LPA o educe indi idual ca dio ascula isk.
ARTICLE NATURE COMMUNICATIONS | DOI: 10.1038/ncomms11122
2NATURE COMMUNICATIONS | 7:11122 | DOI: 10.1038/ncomms11122 | www.na u e.com/na u ecommunica ions
Genomes e e ence panel. This is because an algo i hm based on
-squa ed be ween ma ke s does no pe o m well wi h a e o
low- equency a ian s. O e all, his esul ed in a o al o 74
a ian s ha we e independen ly associa ed wi h one o mo e o
he 123 me abolic ai s. We es ima ed he p opo ion o
a iance explained by hese 74 a ian s on he me abolic ai s
(Supplemen a y Table 1 and Me hods). Fo all, bu six, me aboli e
ai s, we obse ed a leas one genome-wide significan
associa ion, wi h he p opo ion o a iance explained anging
om 0.2% o ace oace a e o 12.5% o glycine, wi h a median o
5%. The a e age inc ease in he p opo ion o a iance explained
was 1.1% when compa ing wi h ou p e ious s udy4. In 8 o he
62 loci, we ound he lead a ian o be a non-synonymous
subs i u ion. The SERPINA1 missense a ian ( s28929474) had
no been iden ified as he lead a ian o he associa ed
me aboli e in p io GWAS. The a ian had no been a ailable
om he HapMap-panel o h ough geno yping a ays bu has
become a ailable h ough 1,000 Genomes impu a ion. This may
explain why we see i as a new lead a ian . The PCSK9 locus also
ha bou ed a missense a ian as a seconda y signal ha was
independen o he lead a ian (Supplemen a y Table 3). We
used he Geno ype-Tissue Exp ession (GTEx) p ojec da abase o
u he e alua e i he independen SNPs would be associa ed
wi h he exp ession o nea by genes in a ious human issues9
and he exp ession quan i a i e loci (eQTLs) a e p esen ed in
Supplemen a y Table 4 (Me hods). Al hough GTEx is s ill in pilo
phase, we we e able o link 14 a ian s wi h gene exp ession o
nea by genes, and in six loci, ou manually cu a ed unc ional
candida e was confi med as an eQTL (Table 2 and Supplemen a y
Table 4). The eigh loci ha ha e no p e iously been associa ed
wi h he same o simila me abolic measu es in popula ion
samples a e lis ed in Table 2. Six o he eigh no el loci we e
associa ed wi h he blood concen a ion o amino acids, one wi h
py u a e and one wi h polyunsa u a ed a y acids. The glycine
deca boxylase (GLDC) on ch omosome 9q24.1 ( s140348140,
P¼3.7 1040) and glycine clea age sys em p o ein H (GCSH)
on ch omosome 16q23.2 ( s10083777, P¼3.0 1013) gene
egions showed associa ion wi h glycine concen a ions. In
addi ion, s10083777 was associa ed wi h he exp ession o
GCSH in he ibial ne e in he GTEx da a (Supplemen a y
Table 5). As a po en ial limi a ion, because o GTEx s ill being in
pilo phase, we canno assess i he a ian is also associa ed wi h
GCSH exp ession in o he issues. Mu a ions in hese wo genes
ha e been p e iously shown o cause Glycine Encephalopa hy
(OMIM: 605899), a a e ecessi e diso de o glycine me abolism
ha mani es s as se e e ea ly onse neu ological complica ions
and is diagnosed by abno mally high glycine concen a ion in he
blood. In his s udy, we ha e linked he neu onal exp ession o
GCSH and ci cula ing glycine le els wi h a common a ian on
he popula ion le el.
Table 1 | Sample demog aphics.
S udy Age BMI Female%
NMean s.d. Mean s.d.
EGCUT 3,287 46.3 19.5 26.4 5.4 58
ERF 2,118 48.2 14.7 26.7 4.7 58
FTC 664 23.9 2.1 23.1 3.7 50
FR97 3,661 45.3 12.8 26.3 4.5 55
COROGENE 828 53.2 13.2 26.6 4.1 54
GenMe s 572 55.8 7.3 27.2 4.5 57
HBCS 708 61.3 2.9 27.1 4.1 60
KORA 1,745 60.9 8.8 28.2 4.8 52
LLS 2,227 59.2 6.8 25.4 3.5 54
NTR 1,192 38.8 12.8 24.6 4.2 64
NFBC 1966 4,709 31.2 0.4 24.6 4.1 51
P edic CVD 374 47.5 14.6 26.6 4.4 37
PROTE 597 38.3 16 25.2 4.6 51
YFS 2,390 37.7 5.0 26 4.7 54
BMI, body mass index; COROGENE, Gene ic P edisposi ion o Co ona y Hea Disease in Pa ien s Ve ified wi h Co ona y Angiog am; EGCUT, Es onian Genome Cen e o Uni e si y o Ta u Coho ; ERF,
E asmus Rucphen Family S udy; FR97, a subsample o FINRISK 1997; FTC, Finnish Twin Coho ; GenMe s, Gene ics o METabolic Synd ome; HBCS, Helsinki Bi h Coho S udy; KORA, Coope a i e Heal h
Resea ch in he Region o Augsbu g; LLS, Leiden Longe i y S udy; N, numbe o indi iduals wi h bo h geno ype and me aboli e ai s analysed; NFBC 1966, No he n Finland Bi h Coho 1966; NTR,
Ne he lands Twin Regis e ; P edic CVD, FINRISK subsample o inciden ca dio ascula cases and con ols; PROTE, EGCUT sub-coho ; YFS, The Ca dio ascula Risk in Young Finns S udy.
Table 2 | No el significan loci iden ified in he GWAS.
T ai Va ian iden ifie Ch Posi ion ea/nea Ea Be a s.e. P- alue Q P- alue N
samples
Candida e
gene
eQTL Func ion
Glycine ch 3:125905336:D 3 125905336 A/ACCT
GACCCTGAC
0.40 0.07 0.01 1.1 1090.03 17,541 SLC41A3 — TFBS
Glycine s140348140 9 5877295 TA/T 0.05 0.33 0.03 3.7 1040 4.7 10617,535 GLDC ——
o PUFA s186183604 11 67128733 A/G 0.04 0.24 0.04 3.2 1011 0.71 13,545 CLCF1 — In on;LOC100130987
Alanine s4554975 12 47201814 G/A 0.64 0.07 0.01 6.1 10 13 0.76 24,792 SLC38A4 — In on
His idine s7954638 12 96314795 A/C 0.48 0.08 0.01 7.3 1015 0.53 19,240 HAL AMDH1 In on;CCDC38
His idine s1998848 14 21492229 A/G 0.05 0.15 0.02 4.9 10 10 0.06 19,239 NDRG2 — TFBS/5’UTR
Py u a e s74249229 16 69979271 T/C 0.05 0.15 0.02 2.1 10 11 0.17 23,561 PDPR ——
Glycine s10083777 16 81065282 T/C 0.17 0.11 0.01 3.0 10 13 0.92 18,732 GCSH GCSH, ATMIN,
LOC102724325
TFBS
Be a, e ec es ima e; ea, e ec allele; Ea , e ec allele equency; eQTL, exp ession quan i a i e ai locus om GTEx; GWAS, genome-wide associa ion s udy; nea, non-e ec allele; o PUFA,
polyunsa u a ed a y acids (o he han 18:2); Q, he e ogenei y s a is ics; TFBS, ansc ip ion ac o -binding si e.
I he SNP is loca ed in an in on o a di e en gene han he candida e, hen he gene is p esen ed in he Func ion column a e semicolon.
Be a e e s o one copy addi ion o he e ec allele in s.d. uni s.
NATURE COMMUNICATIONS | DOI: 10.1038/ncomms11122 ARTICLE
NATURE COMMUNICATIONS | 7:11122 | DOI: 10.1038/ncomms11122 | www.na u e.com/na u ecommunica ions 3
Known loci and LPA associa ion. In addi ion o he new loci
disco e ed, we ound significan SNPs sp ead in 54 loci ha ha e
al eady been associa ed wi h he same o ela ed me abolic
measu es as p esen ed in he ca alogue o published GWASs1o
ecen ly disco e ed5(Supplemen a y Table 4). We hen wen
h ough he loci ha had been associa ed wi h simila me abolic
ai s compa ed wi h he p io published findings o pinpoin
po en ial no el biological unc ions o he al eady known loci.
He e, we no ed ha he Lp(a)- aising allele s10455872-G loca ed
in he in on o LPA was associa ed wi h a smalle diame e o
e y-low-densi y lipop o ein (VLDL) pa icles (P¼1.3 1012).
This allele was also associa ed wi h lowe concen a ions o ex a-
la ge, la ge and medium VLDL pa icles (Fig. 2). This me abolic
link ound be ween ci cula ing Lp(a) wi h VLDL me abolism is
no el. Lp(a) is hough o be comp ised o an low-densi y
lipop o ein (LDL) pa icle and a co alen ly bound p o ein
p oduc o he LPA gene, apo(a). Al hough he same a ian in
he LPA locus has been associa ed wi h LDL and o al choles e ol
in o e 100,000 indi iduals10, ou associa ion in his s udy had
nea ly wice he e ec es ima e o he VLDL associa ions using
he same a ian . Bo h s udies had s anda dized alues and
compa ed e ec es ima es we e in s anda d de ia ion uni s. The
LPA locus is known o i s associa ion wi h CHD isk11,12 and
he gene ic a ian s associa ed wi h highe CHD isk a e also
associa ed wi h highe Lp(a) concen a ions13 making i a
po en ially impo an d ug a ge o CHD. Fu he mo e, he
Lp(a) inc easing allele s10455872-G has also been shown o
educe s a in esponse, which implies ha LPA a ge ing
ea men could also po en ially imp o e s a in e ficacy14,15.
Gene ic isk sco e o LPA and me aboli e associa ions.We
ha e disco e ed new and s onge me abolic associa ions o a
known impo an CHD isk locus han iden ified p e iously10.
This in iguing finding di ec ed us o fine map he gene ic
a chi ec u e o Lp(a) in o de o gene a e he bes possible Lp(a)
gene ic isk sco e (GRS
Lp(a)
) ha would enable us o cla i y
associa ions wi h he in ica e aspec s o lipop o ein me abolism.
The gene sco e was gene a ed by pe o ming GWAS on
ci cula ing Lp(a) le els in FINRISK97 (N¼4,935) using
s epwise inc emen al condi ioning (Supplemen a y Me hods).
The esul ing gene sco e consis ed o 18 independen gene ic
a ian s loca ed nea he LPA gene and associa ed wi h Lp(a) a
genome-wide significance (Po5108). All 18 SNPs we e
u he eplica ed o ci cula ing Lp(a) in The Ca dio ascula Risk
in Young Finns S udy (YFS; N¼2,022, Supplemen a y Table 6).
XXL.VLDL.PL
XXL.VLDL.TG
XXL.VLDL.L
XXL.VLDL.P
XL.VLDL.PL
XL.VLDL.TG
XL.VLDL.L
XL.VLDL.P
L.VLDL.C
L.VLDL.FC
L.VLDL.PL
L.VLDL.TG
L.VLDL.CE
L.VLDL.L
L.VLDL.P
M.VLDL.C
M.VLDL.FC
M.VLDL.PL
M.VLDL.TG
M.VLDL.CE
M.VLDL.L
M.VLDL.P
S.VLDL.C
S.VLDL.FC
S.VLDL.PL
S.VLDL.TG
S.VLDL.L
S.VLDL.P
XS.VLDL.PL
XS.VLDL.TG
XS.VLDL.L
XS.VLDL.P
IDL.C
IDL.FC
IDL.PL
IDL.TG
IDL.L
IDL.P
L.LDL.C
L.LDL.FC
L.LDL.PL
L.LDL.CE
L.LDL.L
L.LDL.P
M.LDL.C
M.LDL.PL
M.LDL.CE
M.LDL.L
M.LDL.P
S.LDL.C
S.LDL.L
S.LDL.P
XL.HDL.C
XL.HDL.FC
XL.HDL.PL
XL.HDL.TG
XL.HDL.CE
XL.HDL.L
XL.HDL.P
L.HDL.C
L.HDL.FC
L.HDL.PL
L.HDL.CE
L.HDL.L
L.HDL.P
M.HDL.C
M.HDL.FC
M.HDL.PL
M.HDL.CE
M.HDL.L
M.HDL.P
S.HDL.TG
S.HDL.L
S.HDL.P
VLDL.D
LDL.D
HDL.D
Se um.C
Es .C
F ee.C
LDL.C
HDL.C
Se um.TG
To PG
PC
SM
ApoA1
ApoB
To .FA
FAw3
FAw6
FAw79S
LA
o PUFA
DHA
MUFA
CH2.in.FA
CH2.DB. a io
DB.in.FA
FALen
Bis.DB. a io
Bis.FA. a io
Glc
Lac
Py
Ci
Glol
Ala
Gln
Gly
His
Ile
Leu
Val
Phe
Ty
Ace
AcAce
bOHBu
C ea
Alb
U ea
Gp
0.5
Be a
0.05
0.005
0.0005
0.00005
5e-8
2.27e-9 0.05
0.10
0.15
0.20
Me abolic associa ions o G allele o s10455872 in Lp(a) locus in ch omosome 6q26
Posi i e Nega i e P- alue
Figu e 2 | The associa ion pa e n o he Lp(a) a ian s10455872 G-allele ac oss all ci cula ing me abolic ai s. Each ba ep esen s he associa ion
wi h espec i e me abolic ai , he size o he ba is he linea eg ession e ec es ima e, colou ing e e s o e ec di ec ion and significance is indica ed
wi h filled ci cles o Po2.27 109and unfilled ci cles o Po5108. Me aboli e abb e ia ions and sample sizes a e gi en in Supplemen a y Table 1,
he s onges associa ion was obse ed o he mean diame e o e y-low-densi y lipop o ein pa icles (VLDL.D).
ARTICLE NATURE COMMUNICATIONS | DOI: 10.1038/ncomms11122
4NATURE COMMUNICATIONS | 7:11122 | DOI: 10.1038/ncomms11122 | www.na u e.com/na u ecommunica ions
The e ec es ima e weigh ed gene sco e explained 54% o Lp(a)
a ia ion in he disco e y and 45% in he eplica ion coho .
No ably, he e ec es ima es o he 18 a ian s we e gene ally
la ge in he eplica ion coho , which migh be eflec i e o he
di e en assay me hods used (Supplemen a y Table 5). We hen
assessed whe he he me abolic associa ions we e s eng hened by
he be e ins umen o gene ically ele a ed Lp(a) by me a-
analysing isk sco e associa ions wi h he me abolic measu es in
FINRISK97 and YFS (Fig. 3). The s onges associa ion o he
GRS
Lp(a)
was again wi h he diame e o VLDL pa icles
(P¼8.6 1047,N¼7,365, Supplemen a y Da a 1 o all
associa ions in bo h indi iduals coho s). We used Mendelian
andomiza ion16 o e alua e causali y o Lp(a) o me abolic
dis u bances (Supplemen a y Me hods and Fig. 3); he de ailed
lipop o ein measu es and ci cula ing Lp(a) le els we e a ailable in
he FINRISK97 and YFS coho o 4,889 and 1,991 indi iduals,
espec i ely. The simila associa ion pa e n be ween obse -
a ional associa ions and causal e ec es ima es s ongly
suppo ha Lp(a) syn hesis is causally al e ing lipop o ein
concen a ions (obse a ional associa ions and ins umen al
a iable es ima es o he me aboli es a e p esen ed in Fig. 3,
and Supplemen a y Da a 1). These findings sugges , maybe
somewha su p isingly, ha Lp(a) syn hesis widely a ec s o e all
lipop o ein me abolism, and in pa icula , he syn hesis o la ge
VLDL pa icles in he li e and he eby he iglyce ide
me abolism in gene al. Based on hese esul s, we p opose ha
he apoB-con aining lipop o ein pa icle used o o m Lp(a) by
he co alen a achmen o apo(a), may ac ually also be a poo ly
lipida ed VLDL- ype o pa icle. This sugges s ha ci cula ing
Lp(a) pa icles a e likely o be a mo e he e ogeneous g oup han
simply an apo(a) componen added o LDL pa icles17.
LPA gene ic isk sco e wi h elec onic heal h eco ds. Al hough
se e al pha maceu ical agen s a e known o cause a modes
dec ease o ci cula ing Lp(a), no d ugs exis ye o e ec i ely
lowe Lp(a)18. S a ins do no lowe he isk due o Lp(a) as s a in
use was no associa ed wi h a change in Lp(a) le els in a s udy by
Cobbae e al.19 and he JUPITER ial showed ha Lp(a) was a
significan p edic o o esidual isk in pa icipan s ea ed wi h
po en s a in he apy20. Howe e , an an isense oligonucleo ide
a ge ing LPA mRNA was shown o e ec i ely lowe ci cula ing
Lp(a) in a phase 1 ial21 and is now in phase 2 ial
(ClinicalT ials.go Iden ifie : NCT02160899). The pha ma-
cological use s ill depends on whe he he LPA exp ession
modi ying ea men would be associa ed wi h unin ended side
e ec s. Since we we e able o de i e an excep ionally s ong
gene ic ins umen o he me abolic associa ions, we used e e se
gene ics o assess whe he gene ically ele a ed Lp(a) would be
associa ed wi h any disease leading o hospi aliza ion o dea h
ac oss he na ion-wide elec onic heal h-ca e egis e s in Finland
in he FINRISK samples (N¼17,487; 429,357 pe son-yea ollow-
up; Supplemen a y Me hods). We ound ha he gene sco e o
Lp(a) was associa ed wi h ischaemic hea diseases (ICD10
I20-I25, P¼6.8 109,N
e en s
¼1,634, odds a io (OR) ¼1.25
pe uni inc emen in log(Lp(a))) bu no wi h any o he o he o
he 218 summa y diagnoses es ed (Supplemen a y Da a 2).
Consis en ly, he gene sco e was only associa ed wi h diagnoses
wi hin his ICD-block when es ing associa ions ac oss all he 615
diagnoses in he elec onic heal h-ca e eco ds a he accu acy o
h ee-digi ICD codes (Supplemen a y Da a 2). Ou s udy ex ends
he esul s om a loss-o - unc ion s udy by Lim e al.22 ha used
wo unca ing LPA splice a ian s se ing as a human knock-ou
model o LPA. Thei s udy ound no inc eased mo bidi y o he
indi iduals ca ying he LPA knockou alleles. The p esen s udy
had a conside ably s onge gene ic ins umen o e alua ing
co-mo bidi ies linked wi h Lp(a) in he gene al popula ion and
hus s eng hens he e idence ha no s ong common disease
co-mo bidi ies a e caused by Lp(a). Howe e , hese e e se
gene ic analyses p e en conclusions o a e disease e en s o
weak associa ion o common diseases. In addi ion, he Hospi al
Discha ge Regis e Diagnoses a e non- alida ed ou comes and
his may educe ou powe o de ec associa ions. Howe e , he
gene al alidi y o he Finnish Hospi al Discha ge Regis e
Diagnoses has been examined in nume ous s udies and ound
o be good23. Ne e heless, hese no el findings suppo he
no ion ha lowe ing ci cula ing Lp(a) le els would be a sui able
he apeu ic a ge o educe esidual CHD isk, and ha LPA
a ge ing he apy could be a beneficial addi ion o s a in
ea men .
Discussion
In his s udy, de ailed molecula p ofiles o ci cula ing me aboli es
we e analysed o almos 25,000 indi iduals o inc ease
−0.3 –0.2 –0.1 0 0.1 0.2
Se um.TG
LDL.D
VLDL.D
TG
P
L
CE
TG
PL
FC
C
P
L
CE
TG
PL
FC
C
P
L
TG
PL
P
L
TG
PL
XXL.VLDL
XL.VLDL
L.VLDL
M.VLDL
S.VLDL
O he s
SD−uni s di e ence in lipid (95% CI) pe
1−uni obse a ional log(Lp(a)), 1−uni GRS in log(Lp(a)
scale, and 1−uni causal log(Lp(a))
Obse a ional Gene sco e Causal es ima e
Figu e 3 | E alua ion o he causa i e ole o he Lp(a) on he ci cula ing
me abolic measu es ia Mendelian andomiza ion. Yellow linea
eg ession es ima es a e obse a ional associa ions, blue a e GRS
Lp(a)
es ima es and ed a e he causal e ec es ima es. Those me abolic ai s
a e lis ed o which he associa ions in he me a-analysis we e significan
wi h genome-wide h eshold (Po2.3 10 9). Me aboli e abb e ia ions
a e gi en in Supplemen a y Table 1.
NATURE COMMUNICATIONS | DOI: 10.1038/ncomms11122 ARTICLE
NATURE COMMUNICATIONS | 7:11122 | DOI: 10.1038/ncomms11122 | www.na u e.com/na u ecommunica ions 5

knowledge on gene ic egula ion o sys emic me abolism. Ou
main findings we e wo old. Fi s , a disco e y o eigh new gene ic
loci o ci cula ing me aboli es and a y acids. The new associa ed
loci con ained ei he anspo e s o enzymes closely in ol ed in
he me abolism o a ficking o he associa ed me aboli e as
shown in Table 2. These new da a a e now a ailable o be used o
s udy he po en ial causali y o a ple ho a o bioma ke s and o
be e unde s and he in ica e me abolic e ec s o known isk
ac o s. Second, in ou sea ch o new me abolic pa hways in
ela ion o known disease-associa ed a ian s, we ound ha a
known CHD-associa ed a ian nea LPA was linked wi h
ci cula ing iglyce ides and VLDL me abolism. Because o hese
new me abolic findings o his pa icula a ian , we ocused on
his egion and fine mapped he gene ic a chi ec u e o Lp(a). In
ac , we we e able o gene a e a gene sco e ha explained o e
45% o he a ia ion in Lp(a) in he eplica ion coho . The
me abolic associa ions we e s eng hened wi h he s onge
gene ic ins umen . Subsequen ly, we used he gene ic isk sco e
in Mendelian andomiza ion o show ha he disco e ed no el
e ec s o Lp(a) syn hesis on o e all lipop o ein and iglyce ide
me abolism a e causal. Fu he mo e, as we now had a s ong
gene ic isk sco e o Lp(a), we could use i o e e se gene ics in
combina ion wi h elec onic heal h eco ds. In iguingly, acco d-
ing o ex ensi e elec onic heal h eco d da a, he gene ic
a ia ion in LPA appea s o be associa ed wi h ischaemic hea
disease bu no wi h o he common ad e se disease e en s. Thus,
ou esul s p o ide he fi s e idence o he po en ial con-
sequences o lipop o ein me abolism when people a e ea ed
wi h eme ging d ugs (a phase 2 ial o LPA mRNA an isense
oligonucleo ides is cu en ly ac i e (ClinicalT ials.go Iden ifie :
NCT02160899)). Ou findings also p o ide suppo ha he
ea men may well be sui able o CHD isk educ ion and is
likely o be ee o o he s ong mo bidi ies. This s udy also se es
as a p oo o concep in e ms o how la ge mul iomics biobank
da a could be e ficien ly used o in o m d ug disco e y a an ea ly
s age.
Me hods
Me aboli e quan ifica ion.This wo k is an ex ension o ou p e ious
GWA-me abolomics s udy, in which he quan i a i e high- h oughpu NMR
me abolomics pla o m, used o quan i y human blood me aboli es, was applied4.
In his s udy, we ha e u ilized he same pla o m o quan i y 123 me aboli e
measu es ha ep esen a b oad molecula signa u e o sys emic me abolism. The
me aboli e se co e s mul iple me abolic pa hways, including lipop o ein lipids and
subclasses, a y acids as well as amino acids and glycolysis p ecu so s. Mos o he
NMR-based me abolomics analyses we e pe o med wi h he comp ehensi e
quan i a i e se um/plasma pla o m desc ibed o iginally by Soininen e al.24 and
e iewed ecen ly25. This same pla o m was used he e o analyse samples in
Es onian Genome Cen e o Uni e si y o Ta u Coho (EGCUT), Finnish Twin
Coho , a subsample o FINRISK 1997 (FR97), Gene ic P edisposi ion o Co ona y
Hea Disease in Pa ien s Ve ified wi h Co ona y Angiog am (COROGENE),
Gene ics o METabolic Synd ome, Helsinki Bi h Coho S udy (HBCS),
Coope a i e Heal h Resea ch in he Region o Augsbu g (KORA), No he n
Finland Bi h Coho 1966 (NFBC 1966), FINRISK subsample o inciden
ca dio ascula cases and con ols (P edic CVD), EGCUT sub-coho (PROTE) and
YFS. Me aboli e-specific un ans o med dis ibu ions and desc ip i e summa y
s a is ics om he la ges coho , NFBC 1966, a e p esen ed in Supplemen a y
Fig. 3. Chemical shi s and he coe ficien s o a ia ion o in e -assay a iabili y a e
p esen ed in Supplemen a y Da a 3 o each me aboli e. He e, he s udy was
ex ended wi h E asmus Rucphen Family S udy (ERF), Leiden Longe i y S udy
(LLS) and Ne he lands Twin Regis e (NTR) coho s o which he small-molecule
in o ma ion was a ailable om ano he NMR-based me hod (Supplemen a y
Table 2 o de ails)26. Me aboli e-specific un ans o med dis ibu ions and
desc ip i e summa y s a is ics o hese measu es om he ERF coho a e gi en in
Supplemen a y Fig. 4. Chemical shi s and he coe ficien s o a ia ion o in e -
assay a iabili y a e p esen ed in Supplemen a y Table 7. The sample ma e ial was
mos ly se um, excep o EGCUT, PROTE, NTR and LLS in which he sample
ma e ial was EDTA-plasma. The ERF coho had addi ional lipop o ein measu es
a ailable h ough he me hod de eloped by B uke L d. (h ps://www.b uke .com/
fileadmin/use _upload/8-PDF-Docs/Magne icResonance/NMR/b ochu es/
lipo-analysis_apps.pd ). The e minology o his me hod u ilized o lipop o ein
analyses in ERF was ma ched based on he lipop o ein pa icle size wi h he
comp ehensi e quan i a i e se um/plasma pla o m o enable me a-analyses. The
as majo i y o blood samples we e as ing, howe e , i a s udy did no ha e
o e nigh as ing samples, we co ec ed he as ing ime e ec by using R package
gam and fi ing a smoo hed spline o adjus o as ing. All me aboli es we e fi s
adjus ed o age, sex, ime om las meal, i applicable, and en fi s p incipal
componen s om genomic da a and he esul ing esiduals we e ans o med o
no mal dis ibu ion by in e se ank-based no mal ans o ma ion.
Genome-wide associa ion s udy.We pe o med a GWAS o me aboli es om
14 coho s om Eu ope, o aling up o 24,925 indi iduals (coho s a e desc ibed in
Table 1, Supplemen a y Table 2 and Supplemen a y No es 1) o include as many
samples wi h NMR me aboli e da a and genome-wide SNP a ay da a as possible.
W i en in o med consen was ob ained om all pa icipan s. S udies we e
app o ed by he ollowing e hical commi ees: E hical Commi ee o Oulu
Uni e si y Facul y o Medicine o NFBC 1966; E hics Commi ee o he Na ional
Public Heal h Ins i u e o Heal h2000 and HBCS; Helsinki Uni e si y Hospi al
Coo dina ing E hical Commi ee o FINRISK and Twins; The KORA s udies ha e
been app o ed by he e hics commi ee o he Ba a ian Medical Associa ion; NTR,
Cen al E hics Commi ee on Resea ch In ol ing Human Subjec s o he VU
Uni e si y Medical Cen e , Ams e dam; EGCUT, E hics Re iew Commi ee on
Human Resea ch o he Uni e si y o Ta u; ERF, medical e hics boa d o he
E asmus MC Ro e dam, he Ne he lands; LLS, Medical E hical Commi ee o he
Leiden Uni e si y Medical Cen e; and E hics Commi ee o he Hospi al Dis ic o
Sou hwes Finland o YFS. Indi iduals unde lipid-lowe ing medica ion o p eg-
nan we e excluded o m he analyses. FINRISK coho s included geno ype ba ches
P edic CVD, COROGENE, DILGOM and FINRISK97. Es onian biobank had wo
geno ype ba ches included in his s udy: EGCUT and PROTE. Geno ype ba ches
we e analysed sepa a ely. We used an addi i e model implemen ed in analysis
so wa e (Supplemen a y Table 2) o each coho . All s udies we e app o ed by
local e hical commi ees. SNPs we e impu ed up o 39 million ma ke s using a 1000
Genomes P ojec Ma ch 2012 e sion as desc ibed in Supplemen a y Table 2
( e . 27). The genomic posi ions used h oughou his s udy a e human genome
build 39. Each coho was analysed sepa a ely and SNPs wi h accu a e impu a ion
(p ope in o40.4) and mino allele coun 43 we e combined in fixed-e ec s
me a-analysis using double genomic con ol co ec ion, ha is, bo h indi idual
coho esul s and me a-analysis esul s we e co ec ed o he genomic infla ion
ac o as implemen ed in GWAMA28. Va ian s, a e fil e ing and me a-analysis,
p esen in mo e han se en s udies we e conside ed o he final esul s. A genome-
wide significance le el was se o 2.27 10 9co ec ing o 22 independen es s as
he me aboli e da a a e co ela ed (s anda d genome-wide significance h eshold o
5108/22, he numbe o p incipal componen s explaining o e 95% o he
a iance in he me abolomics da a). The numbe o independen es s was de i ed
om he numbe o p incipal componen s ha explain o e 95% o a ia ion in he
me aboli e da a. All ai s ga e genomic infla ion ac o s in he me a-analysis less
han 1.034 showing ha he e was li le e idence o sys ema ic bias in he es
s a is ics. Quan ile plo s o measu emen s lis ed in Supplemen a y Table 1 a e
p esen ed in Supplemen a y Fig. 5.
Condi ional analyses and p opo ion o a iance explained.We conduc ed an
ini ial o mal condi ional analysis o each o he 62 significan loci. We pe o med
an associa ion es o all SNP— ai pai s in a 2- o 4-Mb window a ound he lead
SNP. The 4-Mb window was used o se en loci whe e he associa ion peak was so
wide ha i spanned o e he 2-Mb window, as in he case o CPS1 locus. The
associa ions in each window we e fi s sc eened in he se en Finnish coho s only.
The lead SNP- ai pai was hen analysed using he me a-analysis summa y
s a is ics and co ela ion s uc u e om he FINRISK-coho o adjus o he
co ela ion be ween he lead SNP and possible seconda y a ian using he me hod
p oposed by Yang e al.29. Fu he associa ion was simila ly adjus ed o co ela ion
be ween p eceding a ian s. The p opo ion o a iance explained was calcula ed
based on he summa y s a is ics o each ai accoun ing o all independen SNPs
om he p ima y me a analysis and condi ional analyses ha we e significan a
he p e-specified h eshold (P¼2.27 109) o ha ai .
GTEx eQTL analyses.We in es iga ed whe he he lead SNPs o ou associa ed
loci we e also associa ed wi h he exp ession le els o nea by genes by que ying he
mul i- issue gene exp ession esou ce om The GTEx p ojec 9,30. The p ojec , da a
collec ion and analysis me hods we e ecen ly desc ibed in de ail9. B iefly, he pilo
da a se o he GTEx P ojec (dbGaP accession numbe phs000424. 3.p1) p o ides
exp ession da a o mul iple issues om up o 156 densely geno yped indi iduals
pe issue. The eQTL analysis was ocused on nine issues ha ing g ea e han 80
samples (Adipose—Subcu aneous, A e y—Tibial, Hea —Le Ven icle, Lung,
Muscle—Skele al, Ne e—Tibial, Skin—Sun Exposed Lowe leg, Thy oid and
Whole Blood) and genes exp essed a leas 0.1 eads pe kilobase pe million
mapped eads (RPKM) in wo o mo e indi iduals in a gi en issue. Fo his pape ,
cis-eQTLs we e calcula ed o hose 57 independen SNPs om he associa ion
analysis ha had mino allele equency (MAF)45% in he GTEx da a using a cis
window o 1 Mb up- and down-s eam om he ansc ip ion s a si e o a gene.
The analysis was conduc ed using he Ma ix-eQTL R package31 in linea
eg ession mode co ec ing o sex, he fi s 15 p obabilis ic es ima ion o
ARTICLE NATURE COMMUNICATIONS | DOI: 10.1038/ncomms11122
6NATURE COMMUNICATIONS | 7:11122 | DOI: 10.1038/ncomms11122 | www.na u e.com/na u ecommunica ions
exp ession esidual ac o s, and he fi s h ee p incipal componen s om he
geno ype da a. The alse disco e y a e was es ima ed ac oss he es ed SNP–gene
pai s (be ween 1,933 and 2,269 pe issue) using he Benjamini–Hochbe g
p ocedu e.
Gene sco e o ele a ed Lp(a).Lp(a) was measu ed in YFS om se um s o ed a
–70 °C by he immuno u bidime ic me hod (Lp(a)-HA eagen , Wako Chemicals
GmbH). Lp(a) was measu ed in FINRISK97 om se um s o ed a –70 °C using a
comme cially a ailable la ex immunoassay on an A chi ec c8000 sys em
(Quan ia Lp(a), Abbo Diagnos ics). The impu ed geno ype ba ches P edic CVD,
COROGENE and FR97 we e combined o gene a e as comple e a geno yped
sample as possible o he gene ic analyses (FINRISK97). We hen pe o med a
GWAS o na u al loga i hm ans o med Lp(a) in FINRISK97 using sex, age
and en gene ic p incipal componen s as co a ia es in linea models. Va ian s
associa ed wi h Lp(a) a genome-wide significance we e i e a i ely added o he
associa ion model o iden ifica ion o independen a ian s. SNPs wi h in o 40.7
and mino allele equency 40.5% we e conside ed. All 18 independen a ian s
iden ified in FINRISK97 we e eplica ed in he independen YFS coho
(Supplemen a y Da a 1).
We used weigh ed e ec es ima es om FINRISK97 o gene a e a gene sco e o
Lp(a) and es ed he p opo ion o a iance explained in he FINRISK97 disco e y
and YFS eplica ion coho s. We also es ed he associa ion be ween he Lp(a) gene
sco e and me aboli es using linea eg ession adjus ed o he same co a ia es as o
he GWAS.
Causali y es ima es o Lp(a) on lipop o ein me abolism.We used na u al
loga i hm o ans o m he Lp(a) dis ibu ion and pe o med linea eg ession o
es o associa ion be ween Lp(a) and me aboli es using linea eg ession adjus ed
o he same co a ia es as o he GWAS. As he e ec es ima es in FINRISK97 o
ci cula ing Lp(a) we e la ge han in YFS, we es ed i di e ences in as ing ime
could accoun o he de ia ions—pa icipan s in FINRISK97 we e only ins uc ed
o as 4 h be o e he blood samples in con as o o e nigh as ing in YFS. We
obse ed no di e ences in he e ec es ima es be ween as ing o e 8 h (N¼4,269)
o as ing less han 8 h (N¼620) subg oups (b¼0.40 o bo h g oups) in
FINRISK97.
Causal es ima es o Lp(a) on me aboli e measu es we e assessed by wo-s age
leas -squa es eg ession wi h he Lp(a) gene sco e as ins umen . To enable
compa ison be ween he obse a ional and causal e ec es ima es om hese
Mendelian andomiza ion analyses, Lp(a) and me aboli es had been co ec ed and
ans o med as in he GWAS. Obse a ional associa ions, gene ic isk sco e
associa ions and ins umen al a iable es ima es om FINRISK97 and YFS we e
combined wi h in e se a iance weigh ed me a-analysis.
Re e se gene ics o LPA wi h na ionwide elec on heal h eco ds.The gene
sco e o Lp(a) was es ed in he FINRISK field s udies conduc ed in 1992, 1997,
2002 and 2007 o associa ion wi h any disease e en leading o hospi aliza ion o
dea h in Finland du ing Janua y 1987 o Decembe 2010. P edic CVD, COR-
OGENE, FR97 and addi ional Illumina co e-exome geno yped sample o 9,906
FINRISK indi iduals we e combined a e impu a ion o o m as comple e
and an un ela ed da a se as possible om FINRISK-samples. Gene ic p incipal
componen s we e gene a ed om he combined geno yped SNPs o accoun o
popula ion s a ifica ion and also o exclude ela ed indi iduals. Maximum
ela edness be ween indi iduals was se o 0.1 be ween geno yping ba ches o
emo e ela ed indi iduals om he combined sample esul ing in 17,496 un ela ed
indi iduals. Disease acking was enabled by uni o m diagnosis da a ob ained om
he Finnish Na ional Hospi al Discha ge Regis e and he Na ional Causes-o -
Dea h Regis e . These egis e s co e all e en s ha ha e led o ei he o e nigh
hospi aliza ion o dea h in Finland. The disease e en s a e linked o s udy
pa icipan s using hei social secu i y numbe , which is assigned o e e y
pe manen esiden o Finland. Bo h di ec and side causes o he diagnoses and
cause o dea h we e analysed. Hospi aliza ion and a al e en s we e combined.
The disease diagnoses we e encoded acco ding o he In e na ional
Classifica ion o Diseases 10 h Re ision (ICD-10) om 1997 onwa ds. Disease
e en s occu ing om 1987 o 1996 we e encoded in ICD-9 o ma ac oss Finland;
hese diagnoses we e con e ed o ICD-10 o ma by he scheme p o ided by he
Uni ed S a es Cen e o Disease Con ol Diagnosis Code Se Gene al Equi alence
Mappings ( p:// p.cdc.go /pub/Heal h_S a is ics/NCHS/Publica ions/ICD10CM/
2011/), including combina ion codes. All diagnosis con e sions we e u he
e ified acco ding o he mapping scheme p o ided by he New Zealand Minis y
o Heal h, Na ional Da a Policy G oup (h p://www.heal h.go .nz/sys em/files/
documen s/pages/mas e 4.xls). Manual cu a ion o he con e sion was
conduc ed o diagnoses wi h misma ch in he con e sion o he deg ee o
h ee digi s.
Tes ing o he Lp(a) gene sco e agains disease e en s was conduc ed by logis ic
eg ession, using he gene sco e as a p edic o and e e -occu ence o a disease
om 1987 onwa ds as a bina y ou come, wi h adjus men o sex, age a end-o -
ollow-up, s udy-collec ion-yea and he fi s ou p incipal componen s
accoun ing o popula ion s uc u e. The coho s we e analysed he e join ly.
Simila esul s we e ob ained when he coho s we e also analysed sepa a ely and
me a-analysed (da a no shown). Diagnoses wi h mo e han en e en s we e
analysed. The disease e en s es ed we e 18 ICD-10 chap e s ( o example, Diseases
o he ci cula o y sys em: I00-I99), 189 ICD-10 blocks (A00-09 o T90-98, o
example, Ischaemic hea diseases: I20-I25) and 615 ICD-10 codes ( om A00 o
T98, o example, Angina pec o is: I20). O e all, 822 ou comes we e es ed.
We he e o e co ec ed o mul iple es ing o 822 es s. The o al ollow-up ime
amoun ed o 429,357 pe son-yea s. Disease-specific ollow-up o ime o e en
models a e gi en in he Supplemen a y Da a 2.
Re e ences
1. Wel e , D. e al. The NHGRI GWAS Ca alog, a cu a ed esou ce o SNP- ai
associa ions. Nucleic Acids Res. 42, D1001–D1006 (2014).
2. Hindo , L. A. e al. Po en ial e iologic and unc ional implica ions o genome-
wide associa ion loci o human diseases and ai s. P oc. Na l Acad. Sci. USA
106, 9362–9367 (2009).
3. Suh e, K. & Giege , C. Gene ic a ia ion in me abolic pheno ypes: s udy designs
and applica ions. Na . Re . Gene . 13, 759–769 (2012).
4. Ke unen, J. e al. Genome-wide associa ion s udy iden ifies mul iple
loci influencing human se um me aboli e le els. Na u e Gene . 44, 269–276
(2012).
5. Shin, S. Y. e al. An a las o gene ic influences on human blood me aboli es.
Na u e Gene . 46, 543–550 (2014).
6. Suh e, K. e al. Human me abolic indi iduali y in biomedical and
pha maceu ical esea ch. Na u e 477, 54–60 (2011).
7. Se ice, S. e al. Magni ude and dis ibu ion o linkage disequilib ium in
popula ion isola es and implica ions o genome-wide associa ion s udies.
Na u e Gene . 38, 556–560 (2006).
8. Su akka, I. e al. The impac o low- equency and a e a ian s on lipid le els.
Na u e Gene . 47, 589–597 (2015).
9. Conso ium, G. T. Human genomics. The Geno ype-Tissue Exp ession (GTEx)
pilo analysis: mul i issue gene egula ion in humans. Science 348, 648–660
(2015).
10. Teslo ich, T. M. e al. Biological, clinical and popula ion ele ance o 95 loci o
blood lipids. Na u e 466, 707–713 (2010).
11. T egoue , D. A. e al. Genome-wide haplo ype associa ion s udy iden ifies he
SLC22A3-LPAL2-LPA gene clus e as a isk locus o co ona y a e y disease.
Na u e Gene . 41, 283–285 (2009).
12. Schunke , H. e al. La ge-scale associa ion analysis iden ifies 13 new
suscep ibili y loci o co ona y a e y disease. Na u e Gene . 43, 333–338
(2011).
13. Kams up, P. R., Tybjae g-Hansen, A., S e ensen, R. & No des gaa d, B. G.
Gene ically ele a ed lipop o ein(a) and inc eased isk o myoca dial in a c ion.
JAMA 301, 2331–2339 (2009).
14. Deshmukh, H. A. e al. Genome-wide associa ion s udy o gene ic de e minan s
o LDL-c esponse o a o as a in he apy: impo ance o Lp(a). J. Lipid Res. 53,
1000–1011 (2012).
15. Chasman, D. I. e al. Gene ic de e minan s o s a in-induced low-densi y
lipop o ein choles e ol educ ion: he Jus ifica ion o he Use o S a ins in
P e en ion: an In e en ion T ial E alua ing Rosu as a in (JUPITER) ial.
Ci c. Ca dio asc. Gene . 5, 257–264 (2012).
16. Da ey Smi h, G. & Hemani, G. Mendelian andomiza ion: gene ic ancho s o
causal in e ence in epidemiological s udies. Hum. Mol. Gene . 23, R89–R98
(2014).
17. Hobbs, H. H. & Whi e, A. L. Lipop o ein(a): in igues and insigh s. Cu . Opin.
Lipidol. 10, 225–236 (1999).
18. Dube, J. B., Bo a, M. B., Hegele, R. A. & Koschinsky, M. L. Lipop o ein(a):
mo e in e es ing han e e a e 50 yea s. Cu . Opin. Lipidol. 23, 133–140
(2012).
19. Cobbae , C. e al. Modula ion o lipop o ein(a) a he ogenici y by high densi y
lipop o ein choles e ol le els in middle-aged men wi h symp oma ic co ona y
a e y disease and no mal o mode a ely ele a ed se um choles e ol. Reg ession
G ow h E alua ion S a in S udy (REGRESS) S udy G oup. J. Am. Coll. Ca diol.
30, 1491–1499 (1997).
20. Khe a, A. V. e al. Lipop o ein(a) concen a ions, osu as a in he apy, and
esidual ascula isk: an analysis om he JUPITER T ial (Jus ifica ion o he
Use o S a ins in P e en ion: an In e en ion T ial E alua ing Rosu as a in).
Ci cula ion 129, 635–642 (2014).
21. Tsimikas, S. e al. An isense he apy a ge ing apolipop o ein(a): a andomised,
double-blind, placebo-con olled phase 1 s udy. Lance 386, 1472–1483
ð2015Þ:
22. Lim, E. T. e al. Dis ibu ion and medical impac o loss-o - unc ion a ian s in
he Finnish ounde popula ion. PLoS Gene . 10, e1004494 (2014).
23. Sund, R. Quali y o he Finnish Hospi al Discha ge Regis e : a sys ema ic
e iew. Scand. J. Public Heal h 40, 505–515 (2012).
24. Soininen, P. e al. High- h oughpu se um NMR me abonomics o cos -
e ec i e holis ic s udies on sys emic me abolism. Analys 134, 1781–1785
(2009).
NATURE COMMUNICATIONS | DOI: 10.1038/ncomms11122 ARTICLE
NATURE COMMUNICATIONS | 7:11122 | DOI: 10.1038/ncomms11122 | www.na u e.com/na u ecommunica ions 7
25. Soininen, P., Kangas, A. J., Wu z, P., Suna, T. & Ala-Ko pela, M. Quan i a i e
se um nuclea magne ic esonance me abolomics in ca dio ascula
epidemiology and gene ics. Ci c. Ca dio asc Gene . 8, 192–206 (2015).
26. Demi kan, A. e al. Insigh in genome-wide associa ion o me aboli e
quan i a i e ai s by exome sequence analyses. PLoS Gene . 11, e1004835
(2015).
27. Genomes P ojec , C. e al. A map o human genome a ia ion om popula ion-
scale sequencing. Na u e 467, 1061–1073 (2010).
28. Magi, R. & Mo is, A. P. GWAMA: so wa e o genome-wide associa ion
me a-analysis. BMC Bioin o ma ics 11, 288 (2010).
29. Yang, J. e al. Condi ional and join mul iple-SNP analysis o GWAS summa y
s a is ics iden ifies addi ional a ian s influencing complex ai s. Na u e Gene .
44(S1–S3): 369–375 (2012).
30. Conso ium, G.T. The Geno ype-Tissue Exp ession (GTEx) p ojec . Na u e
Gene . 45, 580–585 (2013).
31. Shabalin, A. A. Ma ix eQTL: ul a as eQTL analysis ia la ge ma ix
ope a ions. Bioin o ma ics 28, 1353–1358 (2012).
Acknowledgemen s
All acknowledgemen s a e p o ided in he Supplemen a y No es 2.
Au ho con ibu ions
J.K., A.D., P.W., S.R. and M.A.-K. w o e he manusc ip . J.K., A.D., P.W., H.H.M.D.,
T.H., R.R., Ta.T., A.P.-S. and A.V. pe o med he s a is ical analyses. A.J.K,, L.P.L., M.Pi.,
R.P., P.S., Q.W., M.T., Tu.T., N.A., T.Z., M.B., J.D., K.W. an D., T.E., J.-J.H., E.M. an L.,
T.L., E.M., R.J.R., A.J.M. de C., A.Ve., J.V., G.W., D.I.B., C.M. an D., J.E., A.J., M.-R.J.,
J.Ka., A.M., O.R., V.S., P.E.S., M.W., C.G., M.K., M.Pe., S.B. and M.S. ga e c i ical
commen s ega ding he manusc ip . J.K., S.R. and M.A.K. supe ised he esea ch.
A.J.K., L.P.L., M.Pi., R.P., P.S., Q.W., M.T., Tu.T., N.A., T.Z., M.B., J.D., K.W. an D.,
T.E., J.-J.H., E.M. an L., T.L., E.M., R.J.R., A.J.M. de C., A.Ve., J.V., G.W., D.I.B., C.M.
an D., J.E., A.J., M.-R.J., J.Ka., A.M., O.R., V.S., P.E.S., M.W., S.R., M.A.-K., C.G., M.K.,
M.Pe., S.B. and M.S. o ganized he da a collec ions.
Addi ional in o ma ion
Supplemen a y In o ma ion accompanies his pape a h p://www.na u e.com/
na u ecommunica ions
Compe ing financial in e es s: S.B. has ecei ed hono a ia om Abbo Diagnos ics,
SIEMENS, The mo Fishe and Roche Diagnos ics and is a consul an o The mo Fishe .
The sponso played no ole in he design o conduc o his s udy; in he managemen ,
analysis o in e p e a ion o he da a; o in he p epa a ion, e iew o app o al o he
manusc ip o in he decision o submi he manusc ip o publica ion. P.W., A.J.K., P.S.
and M.A.-K. a e sha eholde s o B ainshake L d. (www.b ainshake.fi), a company
o e ing NMR-based me aboli e p ofiling. Jo.K., P.W., A.J.K., P.S., Q.W., M.T. and Tu.T.
epo employmen and consul ing o B ainshake L d. The emaining au ho s decla e no
compe ing financial in e es s.
Rep in s and pe mission in o ma ion is a ailable online a h p://npg.na u e.com/
ep in sandpe missions/
How o ci e his a icle: Ke unen, J. e al. Genome-wide s udy o ci cula ing me a-
boli es iden ifies 62 loci and e eals no el sys emic e ec s o LPA.Na . Commun. 7:11122
doi: 10.1038/ncomms11122 (2016).
This wo k is licensed unde a C ea i e Commons A ibu ion 4.0
In e na ional License. The images o o he hi d pa y ma e ial in his
a icle a e included in he a icle’s C ea i e Commons license, unless indica ed o he wise
in he c edi line; i he ma e ial is no included unde he C ea i e Commons license,
use s will need o ob ain pe mission om he license holde o ep oduce he ma e ial.
To iew a copy o his license, isi h p://c ea i ecommons.o g/licenses/by/4.0/
Johannes Ke unen1,2,3,4,Ays¸e Demi kan5,6,w, Pe e Wu¨ z1, Ha men H.M. D aisma7,8,9, Toomas Halle 10,
Rajesh Rawal11,12, Anika Vaa ho s 13, An i J. Kangas1, Leo-Pekka Lyy ika
¨inen14, Ma i Pi inen15,Rene
´Pool7,8,
An i-Pekka Sa in2,15, Pasi Soininen1,3, Ta u Tukiainen16,17,18, Qin Wang1,3, Mika Tiainen1,3, Tuulia Tynkkynen1,3,
Naja Amin6, Tanja Zelle 19,20, Ma ian Beekman13, Jo is Deelen13, Ko Willems an Dijk5,21,w,To
˜nu Esko10,
Jouke-Jan Ho enga7,8, Elisabe h M. an Leeuwen6, Te ho Leh ima
¨ki14, E elin Mihailo 10, Richa d J. Rose22,23,
An on J.M. de C aen24, Ch is ian Giege 11,12, Mika Ka
¨ho
¨nen25, Ma kus Pe ola2,10,15, S e an Blankenbe g19,20,
Ma kku J. Sa olainen4,26, Aswin Ve hoe en27, Jo ma Viika i28, Gonneke Willemsen7,8, Do e I. Boomsma7,8,
Co nelia M. an Duijn6, Johan E iksson2,29,30, An i Jula2, Ma jo-Rii a Ja
¨ elin4,31,32,33, Jaakko Kap io15,22,34,
And es Me spalu10, Olli Rai aka i35,36, Veikko Salomaa2, P. Eline Slagboom13, Melanie Waldenbe ge 11,12,
Samuli Ripa i2,15,22,37,* & Mika Ala-Ko pela1,3,4,38,39,40,*
1Compu a ional Medicine, Facul y o Medicine, Uni e si y o Oulu, PO Box 5000, 90014 Oulu, Finland. 2Na ional Ins i u e o Heal h and Wel a e, PO Box
30, FI-00271 Helsinki, Finland. 3NMR Me abolomics Labo a o y, School o Pha macy, Uni e si y o Eas e n Finland, Yliopis on an a 1C, Kuopio 70210,
Finland. 4Biocen e Oulu, Uni e si y o Oulu, PO Box 5000, FI-90014 Oulu, Finland. 5Depa men o Human Gene ics, Leiden Uni e si y Medical Cen e ,
PO Box 9600, 2300 RC Leiden, The Ne he lands. 6Depa men o Epidemiology, E asmus Medical Cen e , PO Box 2040, 3000 CA Ro e dam, The
Ne he lands. 7Depa men o Biological Psychology, VU Uni e si y Ams e dam, Van de Boecho s s aa 1, Room 2B-29, 1081 BT Ams e dam, The
Ne he lands. 8EMGO Ins i u e o Heal h and Ca e Resea ch, Van de Boecho s s aa 7, 1081BT Ams e dam, The Ne he lands. 9Neu oscience Campus
Ams e dam, De Boelelaan 1085, 1081HV Ams e dam, The Ne he lands. 10 Es onian Genome Cen e , Uni e si y o Ta u, Riia 23b, 51010 Ta u, Es onia.
11 Resea ch Uni o Molecula Epidemiology, Helmhol z Zen um Mu
¨nchen, Ingols a
¨d e Lands ae 1, 85764 Neuhe be g, Ge many. 12 Ins i u e o
Epidemiology II, Helmhol z Zen um Mu
¨nchen, Ingols a
¨d e Lands ae 1, 85764 Neuhe be g, Ge many. 13 Depa men o Molecula Epidemiology, Leiden
Uni e si y Medical Cen e , PO Box 9600, 2300 RC Leiden, The Ne he lands. 14 Depa men o Clinical Chemis y, Fimlab Labo a o ies, Uni e si y o Tampe e
School o Medicine, Tampe e Uni e si y, Kale an ie 4, Tampe e 33014, Finland. 15 Ins i u e o Molecula Medicine (FIMM), Uni e si y o Helsinki,
Biomedicum 2, Tukholmanka u 8, Helsinki 00290, Finland. 16 Analy ic and T ansla ional Gene ics Uni , Depa men o Medicine, Massachuse s Gene al
Hospi al, 55 F ui S ee , Bos on, Massachuse s 02114, USA. 17 P og am in Medical and Popula ion Gene ics, B oad Ins i u e, 415 Main S ee Camb idge,
Massachuse s 02142, USA. 18 Depa men o Gene ics, Ha a d Medical School, 77 A enue Louis Pas eu , NRB 0330, Bos on, Massachuse s 02115, USA.
19 Ge man Cen e o Ca dio ascula Resea ch (DZHK e.V.), Pa ne Si e Hambu g/Lu
¨beck/Kiel, Ma inis ae 52, 20246 Hambu g, Ge many. 20 Uni e si y
Hea Cen e Hambu g, Clinic o gene al and in e en ional Ca diology, Ma inis ae 52, 20246 Hambu g, Ge many. 21 Depa men o Endoc inology, Leiden
Uni e si y Medical Cen e , PO Box 9600, 2300 RC Leiden, The Ne he lands. 22 Depa men o Public Heal h, Hjel Ins i u e, Uni e si y o Helsinki, PO Box 41
Manne heimin ie 172, Helsinki 00014, Finland. 23 Depa men o Psychological and B ain Sciences, Indiana Uni e si y, 1101 E 10 h S ee , Blooming on,
Indiana 47405, USA. 24 Depa men o Ge ia ics and Ge on ology, Leiden Uni e si y Medical Cen e , Pos zone C7-Q, PO Box 9600, 2300RC Leiden,
The Ne he lands. 25 Depa men o Clinical Physiology, Uni e si y o Tampe e and Tampe e, Uni e si y Hospi al, PO Box 2000, FIN-33521 Tampe e, Finland.
26 Medical Resea ch Cen e , In e nal Medicine, Oulu Uni e si y Hospi al, Uni e si y o Oulu, Aapis ie 5A, Oulu FI-90220, Finland. 27 Cen e o P o eomics
and Me abolomics, Leiden Uni e si y Medical Cen e , Albinusd ee 2, 2333 ZA Leiden, The Ne he lands. 28 Depa men o Medicine, Uni e si y o Tu ku and
ARTICLE NATURE COMMUNICATIONS | DOI: 10.1038/ncomms11122
8NATURE COMMUNICATIONS | 7:11122 | DOI: 10.1038/ncomms11122 | www.na u e.com/na u ecommunica ions
Tu ku Uni e si y Hospi al, PB 52, 20521 Tu ku, Finland. 29 Depa men o Gene al P ac ice and P ima y Heal h Ca e, Uni e si y o Helsinki, PL 20,
Tukholmanka u 8B, Helsinki 00029, Finland. 30 Folkha
¨lsan Resea ch Cen e, Helsing o s Uni e si e , PB 63, Helsinki 00014, Finland. 31 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 SW7 2AZ, UK.
32 Cen e o Li e Cou se and Sys ems Epidemiology, Facul y o Medicine, Uni e si y o Oulu, PL 5000, 90014 Oulu, Finland. 33 Uni o P ima y Ca e, Oulu
Uni e si y Hospi al, P.O. Box 20, OYS, Oulu 90029, Finland. 34 Depa men o Men al Heal h and Subs ance Abuse Se ices, Na ional Ins i u e o Heal h and
Wel a e, PO Box 30 (Manne heimin ie 166), Helsinki 00300, Finland. 35 Resea ch Cen e o Applied and P e en i e Ca dio ascula Medicine, Uni e si y o
Tu ku, Kiinamyllynka u 4-8, Tu ku 20521, Finland. 36 Depa men o Clinical Physiology, Tu ku Uni e si y Hospi al, Kiinamyllynka u 4-8, Tu ku 20521, Finland.
37 Human Gene ics, Wellcome T us Sange Ins i u e, Wellcome T us Genome Campus, Hinx on CB10 1SA, UK. 38 Oulu Uni e si y Hospi al, Kajaanin ie 50,
Oulu 90220, Finland. 39 Compu a ional Medicine, School o Social and Communi y Medicine, Uni e si y o B is ol, Sena e House, Tyndall A enue,
B is ol, B is ol BS8 1TH, UK. 40 Medical Resea ch Council In eg a i e Epidemiology Uni , Uni e si y o B is ol, B is ol, B is ol BS8 1TH, UK. * These au ho s
con ibu ed equally o his wo k. wP esen add ess: LUMC, Building 2, Ein ho enweg 20, 2333 ZC Leiden (A.D.); (K.W. .D).
NATURE COMMUNICATIONS | DOI: 10.1038/ncomms11122 ARTICLE
NATURE COMMUNICATIONS | 7:11122 | DOI: 10.1038/ncomms11122 | www.na u e.com/na u ecommunica ions 9