This is an elec onic ep in o he o iginal a icle.
This ep in may di e om he o iginal in pagina ion and ypog aphic de ail.
Au ho (s):
Ti le:
Yea :
Ve sion:
Please ci e he o iginal e sion:
All ma e ial supplied ia JYX is p o ec ed by copy igh and o he in ellec ual p ope y igh s, and
duplica ion o sale o all o pa o any o he eposi o y collec ions is no pe mi ed, excep ha
ma e ial may be duplica ed by you o you esea ch use o educa ional pu poses in elec onic o
p in o m. You mus ob ain pe mission o any o he use. Elec onic o p in copies may no be
o e ed, whe he o sale o o he wise o anyone who is no an au ho ised use .
Sho elome e leng h is associa ed wi h impai ed cogni i e pe o mance in Eu opean
ances y coho s
Hägg, S.; Zhan, Y.; Ka lsson, R.; Ge i sen, L.; Plone , A.; an de Lee, S. J.; B oe , L.;
Deelen, J.; Ma ioni, R. E.; Wong, A.; Lundquis , A.; Zhu, G.; Hansell, N. K.; Sillanpää,
Elina; Fedko, I. O.; Amin, N. A.; Beekman, M.; C aen, A. J. M. de; Dege man, S.; Ha is,
S. E.; Kan, K.-J.; Ma in-Ruiz, C. M.; Mon gome y, G. W.; G oup, Neu oCHARGE
Cogni i e Wo king; Adol sson, A. N.; Reynolds, C. A.; Samani, N. J.; Suchiman, H. E. D.;
Viljanen, Anne; on Zglinicki, T.; W igh , M. J.; Ho enga, J.-J.; Boomsma, D. I.;
Ran anen, Taina; Kap io, J. A.; Nyhol , D. R.; Ma in, N. G.; Nybe g, L.; Adol sson, R.;
Kuh, D.; S a , J. M.; Dea y, I. J.; Slagboom, P. E.; an Duijn, C. M.; Codd, V.; Pede sen,
N. L.
Hägg, S., Zhan, Y., Ka lsson, R., Ge i sen, L., Plone , A., an de Lee, S. J., B oe , L.,
Deelen, J., Ma ioni, R. E., Wong, A., Lundquis , A., Zhu, G., Hansell, N. K., Sillanpää, E.,
Fedko, I. O., Amin, N. A., Beekman, M., C aen, A. J. M. D., Dege man, S., . . .
Pede sen, N. L. (2017). Sho elome e leng h is associa ed wi h impai ed cogni i e
pe o mance in Eu opean ances y coho s. T ansla ional Psychia y, 7(4), A icle
e1100. h ps://doi.o g/10.1038/ p.2017.73
2017
OPEN
ORIGINAL ARTICLE
Sho elome e leng h is associa ed wi h impai ed cogni i e
pe o mance in Eu opean ances y coho s
S Hägg
1
, Y Zhan
1
, R Ka lsson
1
, L Ge i sen
1
, A Plone
1
, SJ an de Lee
2
, L B oe
3
, J Deelen
4,5
, RE Ma ioni
6,7,8
, A Wong
9
, A Lundquis
10
,
G Zhu
11
, NK Hansell
8,11
, E Sillanpää
12
, IO Fedko
13
, NA Amin
2
, M Beekman
4
, AJM de C aen
14,✠
, S Dege man
15
, SE Ha is
6,7
, K-J Kan
13
,
CM Ma in-Ruiz
16
, GW Mon gome y
11
, Neu oCHARGE Cogni i e Wo king G oup, AN Adol sson
17
, CA Reynolds
18
, NJ Samani
19,20
,
HED Suchiman
4
, A Viljanen
12
, T on Zglinicki
21
, MJ W igh
8,22
, J-J Ho enga
13
, DI Boomsma
13
, T Ran anen
12
, JA Kap io
23,24,25
,
DR Nyhol
11,26
, NG Ma in
11
, L Nybe g
27,28,29
, R Adol sson
17
, D Kuh
9
, JM S a
7,30
, IJ Dea y
7,31
, PE Slagboom
4
, CM an Duijn
2
, V Codd
19,20
,
NL Pede sen
1
o he ENGAGE Conso ium
The associa ion be ween elome e leng h (TL) dynamics on cogni i e pe o mance o e he li e-cou se is no well unde s ood. This
s udy me a-analyses obse a ional and causal associa ions be ween TL and six cogni i e ai s, wi h s a ifica ions on APOE
geno ype, in a Mendelian Randomiza ion (MR) amewo k. Twel e Eu opean coho s (N= 17 052; mean age = 59.2 ± 8.8 yea s)
p o ided esul s o associa ions be ween qPCR-measu ed TL (T/S- a io scale) and gene al cogni i e unc ion, mini-men al s a e
exam (MMSE), p ocessing speed by digi symbol subs i u ion es (DSST), isuospa ial unc ioning, memo y and execu i e
unc ioning (STROOP). In addi ion, a gene ic isk sco e (GRS) o TL including se en known gene ic a ian s o TL was calcula ed,
and used in associa ions wi h cogni i e ai s as ou comes in all coho s. Obse a ional analyses showed ha longe elome es we e
associa ed wi h be e sco es on DSST (β= 0.051 pe s.d.-inc ease o TL; 95% confidence in e al (CI): 0.024, 0.077; P= 0.0002), and
MMSE (β= 0.025; 95% CI: 0.002, 0.047; P= 0.03), and as e STROOP (β=−0.053; 95% CI: −0.087, −0.018; P= 0.003). E ec s o DSST
we e s onge in APOE ε4 non-ca ie s (β= 0.081; 95% CI: 0.045, 0.117; P= 1.0 × 10
−5
), whe eas ca ie s pe o med be e in STROOP
(β=−0.074; 95% CI: −0.140, −0.009; P= 0.03). Causal associa ions we e ound o STROOP only (β=−0.598 pe s.d.-inc ease o TL;
95% CI: −1.125, −0.072; P= 0.026), wi h a la ge e ec in ε4-ca ie s (β=−0.699; 95% CI: −1.330, −0.069; P= 0.03). Two-sample
eplica ion analyses using CHARGE summa y s a is ics showed causal e ec s be ween TL and gene al cogni i e unc ion and
DSST, bu no wi h STROOP. In conclusion, we sugges causal e ec s om longe TL on be e cogni i e pe o mance, whe e
APOE ε4-ca ie s migh be a di e en ial isk.
T ansla ional Psychia y (2017) 7, e1100; doi:10.1038/ p.2017.73; published online 18 Ap il 2017
INTRODUCTION
Telome es, sho DNA sequences a he end o ch omosomes, a e
conside ed ma ke s o biological age. Cell eplica ion and
oxida i e s esso s con ibu e o he loss o elome e nucleo ides
o e ime; below c i ical leng h, cellula senescence will ollow.
1
An inc easing numbe o s udies ha e shown he impo ance o
elome e leng h (TL) in ageing, specifically in he de elopmen o
demen ia and cogni i e impai men .
2–8
Using a ela i ely la ge
g oup o non-demen ed olde indi iduals, Ya e e al.
3
demon-
s a ed an associa ion be ween longe elome es and highe sco e
in he digi symbol subs i u ion es (DSST)—a measu e o
p ocessing speed—a baseline. A e se en yea s, he indi iduals
wi h longe elome es a baseline pe o med be e in he
modified mini-men al s a e exam (MMSE) bu no in DSST.
1
Depa men o Medical Epidemiology and Bios a is ics, Ka olinska Ins i u e , S ockholm, Sweden;
2
Depa men o Epidemiology, E asmus Uni e si y Medical Cen e , Ro e dam,
The Ne he lands;
3
Depa men o In e nal Medicine, E asmus Uni e si y Medical Cen e , Ro e dam, The Ne he lands;
4
Depa men o Molecula Epidemiology, Leiden Uni e si y
Medical Cen e , Leiden, The Ne he lands;
5
Max Planck Ins i u e o Biology o Ageing, Cologne, Ge many;
6
Cen e o Genomic and Expe imen al Medicine, Ins i u e o Gene ics
and Molecula Medicine, Uni e si y o Edinbu gh, Edinbu gh, UK;
7
Cen e o Cogni i e Ageing and Cogni i e Epidemiology, Uni e si y o Edinbu gh, Edinbu gh, UK;
8
Queensland
B ain Ins i u e, Uni e si y o Queensland, B isbane, QLD, Aus alia;
9
MRC Uni o Li elong Heal h and Ageing a UCL, London, UK;
10
Depa men o S a is ics, Umeå Uni e si y,
Umeå, Sweden;
11
QIMR Be gho e Medical Resea ch Ins i u e, B isbane, QLD, Aus alia;
12
Ge on ology Resea ch Cen e , Facul y o Spo and Heal h Sciences, Uni e si y o
Jy äskylä, Jy äskylä, Finland;
13
Depa men Biological Psychology, V ije Uni e si ei , Ams e dam, The Ne he lands;
14
Depa men o Ge on ology and Ge ia ics, Leiden Uni e si y
Medical Cen e , Leiden, The Ne he lands;
15
Depa men o Medical Biosciences, Umeå Uni e si y, Umeå, Sweden;
16
NIHR Newcas le Biomedical Resea ch Cen e & Uni , Ins i u e
o Neu osciences, Newcas le Uni e si y, Campus o Ageing and Vi ali y, Newcas le upon Tyne, UK;
17
Depa men o Clinical Sciences, Umeå Uni e si y, Umeå, Sweden;
18
Depa men o Psychology, Uni e si y Cali o nia Ri e side, Ri e side, CA, USA;
19
Depa men o Ca dio ascula Sciences, Uni e si y o Leices e , Leices e , UK;
20
Na ional Ins i u e
o Heal h Resea ch Leices e Ca dio ascula Biomedical Resea ch Uni , Glenfield Hospi al, Leices e , UK;
21
Newcas le Uni e si y Ins i u e o Ageing, Ins i u e o Cell & Molecula
Biosciences, Newcas le upon Tyne, UK;
22
Cen e o Ad anced Imaging, Uni e si y o Queensland, B isbane, QLD, Aus alia;
23
Depa men o Public Heal h, Clincum, Uni e si y o
Helsinki, Helsinki, Finland;
24
Ins i u e o Molecula Medicine Finland (FIMM), Uni e si y o Helsinki, Helsinki, Finland;
25
Na ional Ins i u e o Heal h and Wel a e (THL), Helsinki,
Finland;
26
Ins i u e o Heal h and Biomedical Inno a ion, Queensland Uni e si y o Technology, B isbane, QLD, Aus alia;
27
Depa men o Radia ion Sciences, Umeå Uni e si y,
Umeå, Sweden;
28
Depa men o In eg a i e Medical Biology (IMB), Umeå Uni e si y, Umeå, Sweden;
29
Umeå cen e o Func ional B ain Imaging, Umeå Uni e si y, Umeå,
Sweden;
30
Alzheime Sco land Demen ia Resea ch Cen e, Uni e si y o Edinbu gh, Edinbu gh, UK and
31
Depa men o Psychology, Uni e si y o Edinbu gh, Edinbu gh, UK.
Co espondence: D S Hägg, Depa men o Medical Epidemiology and Bios a is ics (MEB), Ka olinska Ins i u e , Box 281, S ockholm 171 77, Sweden.
E-mail: [email p o ec ed]
✠
Deceased.
Recei ed 13 Feb ua y 2017; accep ed 13 Feb ua y 2017
Ci a ion: T ansl Psychia y (2017) 7, e1100; doi:10.1038/ p.2017.73
www.na u e.com/ p
Cohen-Manheim e al.
8
in es iga ed TL in young adul s and ound
as e a i ion a es wi h poo e mid-li e gene al- and domain-
specific cogni i e pe o mance bu no associa ion wi h baseline
TL. In addi ion, a c oss-sec ional s udy o non-demen ed
indi iduals concluded ha APOE ε4-ca ie s had longe TL bu
as e a i ion a es indica ing abno mal cell u no e .
9
O he
s udies ha e also shown associa ions be ween TL and cogni ion
wi h conflic ing esul s o we e unde powe ed.
4,6,10,11
Hence, he
unde lying mechanisms by which elome es may be in ol ed in
cogni i e pe o mance a e complex, and la ge e o s a e needed
o elucida e his ela ionship. Mo eo e , i is s ill unclea whe he
sho elome es a e a cause, consequence o bo h o cogni i e
impai men .
One way o p edic a causal associa ion is o conduc a
Mendelian Randomiza ion (MR) s udy,
12
in which gene ic ma ke s
a e used as p oxies o an exposu e (TL), o in es iga e an un-
biased e ec on an ou come (cogni i e pe o mance). Because
gene ic a ian s a e andomly asso ed a meiosis, hey a e
gene ally ee om con en ional con ounding and hence he MR
s udy design is o en e e ed o as na u e’s own clinical ial.
13
We hypo hesized ha TL is an indica o o cellula s abili y,
which as such a ec s unc ioning h oughou he body, including
pe o mance on all ypes o cogni i e ai s.
14
In addi ion,
indi iduals ca ying he APOE ε4 allele a e mo e suscep ible o
cogni i e impai men and a e he e o e o special in e es .
15
The
objec i e o ou s udy was o conduc a me a-analy ic MR s udy o
he associa ion be ween TL and six cogni i e ai s in 12 Eu opean
ances y coho s (N= 17 052). A seconda y aim was o s a i y on
APOE ε4 geno ype o in es iga e i ca ie s we e a di e en isks
gi en hei wo se cogni i e abili y. Mos coho s we e en olled
h ough he Eu opean Ne wo k o Genomic and Gene ic
Epidemiology (ENGAGE) Conso ium. Telome e measu emen s
we e pe o med by qPCR and a gene ic isk sco e (GRS) wi h
se en gene ic a ian s associa ed wi h TL
16
was calcula ed.
Obse a ional- as well as causal es ima es we e subsequen ly
ob ained using an MR design.
17
In a eplica ion e o , summa y
s a is ics om he Coho s o Hea and Aging Resea ch in
Genomic Epidemiology (CHARGE) Conso ium o he gene ic
associa ions wi h h ee cogni i e ai s
18,19
we e included in a wo-
sample MR app oach.
20
MATERIALS AND METHODS
S udy samples
Twel e coho s wi h a o al o 17 052 indi iduals (Table 1), all wi h
Eu opean ances y popula ions, pa icipa ed in he ENGAGE e o . The
sample-size weigh ed mean o age was 59.2 yea s wi h s.d. = 8.8. Mos
coho s con ibu ed da a measu ed a mid-li e o olde . The Leiden
Longe i y S udy 2 (LLS2) was he oldes coho (mean age = 93.3 yea s).
The Ne he lands Twin Regis e (NTR) included middle-aged adul s (mean
age = 40.3 yea s, s.d. = 16.4) and QIMR (Twin s udies a he Queensland
Ins i u e o Medical Resea ch) included adolescen s only (mean age = 14.1
yea s, s.d. = 2.4). All bu one s udy showed a ai ly e en p opo ion o sexes
( ange: 49–67% women); FITSA (The Finnish Twin S udy on Ageing)
included women only. Addi ional s udy-specific de ails a e ound in
Supplemen a y Table 1.
Cogni i e ai s
Six di e en cogni i e ai s we e es ed in a combined me a-analysis o
he ENGAGE coho s: (1) gene al cogni i e unc ion; (2) MMSE; (3)
p ocessing speed wi h DSST o he a ian symbol digi subs i u ion ask;
(4) isuospa ial unc ioning wi h block design es (BLOCK); (5) episodic
memo y by ei he e bal lea ning o pic u e lea ning es s (MEMORY); and
(6) execu i e unc ioning using S oop in e e ence sco e (STROOP).
De ailed desc ip ions o he di e en cogni i e ai s a e ound in he
supplemen (Supplemen a y Table 3). All coho s pa icipa ed wi h a leas
one cogni i e ai ; no single coho had all o hem (Table 1).
Table 1. Coho s pa icipa ing in he ENGAGE s udy
Coho Full name o coho NAge (y ) mean (s.
d.)
TL (T/S- a io) mean
(s.d.)
Women (%) GRS mean (s.
d.)
MMSE DSST BLOCK MEMORY STROOP Gene al
cogni ion
APOE
geno ype
BETULA1 The Be ula S udy
a
163 50.9 (7.8) 1.01 (0.17) 58.3 8.5 (1.57) × × × × × Yes
BETULA2 The Be ula S udy
a
396 62.4 (14.5) 0.93 (0.15) 54.8 8.87 (1.56) × × × × × Yes
BETULA3 The Be ula S udy
a
315 60 (15) 0.97 (0.17) 54.0 8.65 (1.51) × × × × × Yes
ERF E asmus Rucphen Family (EUROSPAN) 2502 51.6 (15.8) 1.76 (0.36) 55.6 8.65 (1.57) × × × Yes
FITSA The Finnish Twin S udy on Ageing 429 68.6 (3.4) 0.9 (0.19) 100.0 8.52 (1.46) × × Yes
Gende Sex di e ences in heal h and aging 466 74.5 (2.6) 0.68 (0.15) 49.0 8.43 (1.34) × × × × Yes
HRS Heal h and Re i emen S udy 4117 70.4 (9.4) 1.3 (0.3) 58.0 8.6 (1.52) × × ×
LBC1936 Lo hian Bi h Coho 999 69.6 (0.8) 1.3 (0.5) 49.0 8.36 (1.56) × × × × × Yes
LLS1 Leiden Longe i y S udy 1 2305 59.2 (6.8) 1.46 (0.26) 54.8 8.47 (1.52) × × × Yes
LLS2 Leiden Longe i y S udy 2 868 93.3 (2.6) 1.28 (0.22) 61.6 8.44 (1.57) × Yes
NSHD Na ional Su ey o Heal h and
De elopmen
2425 53 (0) 1.54 (0.91) 50.0 8.55 (1.4) × × Yes
NTR Ne he lands Twin Regis e 200 40.3 (16.4) 2.72 (0.56) 66.5 8.5 (1.56) ×
QIMR Twin s udies a he Queensland
Ins i u e o Medical Resea ch
1280 14.1 (2.4) 3.7 (0.6) 52.6 8.5 (1.5) ×
SATSA Swedish Adop ion/Twin S udy o
Aging
587 68.8 (9.6) 0.76 (0.27) 58.0 8.45 (1.45) × × × × Yes
Abb e ia ions: BLOCK, block design es ; CHARGE, Coho s o Hea and Aging Resea ch in Genomic Epidemiology Conso ium; DSST, digi symbol subs i u ion es ; ENGAGE, Eu opean Ne wo k o Gene ic and
Genomic Epidemiology; GRS, gene ic isk sco e; MEMORY, e bal memo y o pic u e lea ning es ; MMSE, mini-men al s a e exam; STROOP, s oop colo wo d ask in e e ence sco e; TL, elome e leng h; Y , yea .
a
BETULA1 and BETULA2-3 we e geno yped a wo di e en acili ies.
Telome e leng h and cogni i e pe o mance
S Hägg e al
2
T ansla ional Psychia y (2017), 1 –6
Telome e leng h measu emen s
Telome e leng h was measu ed in leukocy es in whole blood/bu y coa
excep o he HRS s udy, which used measu emen s om sali a. DNA om
sali a de i es o he mos pa (~74%) om leukocy es,
21
and TL
measu emen s om blood and sali a ha e been epo ed o ha e good
co ela ions (R= 0.72).
22
S anda d qPCR echniques o TL measu emen
we e applied as desc ibed by Caw hon
23
wi h mino modifica ions in he
Lo hian Bi h Coho (LBC)
11
and BETULA.
24
In b ie , elome e (T) and single
copy gene (S) quan i y we e measu ed and a T/S- a io was calcula ed. One
o se e al e e ence samples we e included in all uns and a ela i e
elome e leng h was calcula ed o each sample.
Geno yping
In o ma ion on geno yping pla o m, quali y con ol and single-nucleo ide
polymo phisms (SNPs) used in each coho is a ailable in he supplemen
(Supplemen a y Da a and Supplemen a y Table 2). An addi i e un-
weigh ed GRS was calcula ed o each indi idual by summa izing he
numbe o isk alleles om se en di e en loci (TERC, TERT, NAF1, OBFC1,
ZNF208, RTEL1 and ACYP2) whe e SNPs ( s10936599, s2736100, s7675998,
s9420907, s8105767, s755017 and s11125529) ha e been ound o
associa e wi h TL.
16
In es iga ions o possible pleio opic e ec s om he
di e en gene ic a ian s used in he GRS a e discussed in he supplemen
(Supplemen a y Da a). Fou coho s (ERF (E asmus Rucphen Family), LLS,
NTR and QIMR) om he cu en e o con ibu ed o he o iginal genome-
wide associa ion s udy (GWAS) o TL (see Supplemen a y Da a o u he
discussions on implica ions o ou s udy). No associa ions be ween any o
he se en TL genes and cogni i e ai s ha e been es ed hus a as
judged om he GWAS Ca alog.
25
Gene ic a ian s we e p io i ized as (1)
di ec ly geno yped, (2) impu ed wi h good quali y, (3) p oxies wi h
2
40.8
and (4) impu ed om summa y s a is ics. The weigh ed mean o he GRS
was 8.55 wi h s.d. = 1.51 (Table 1). APOE geno ype was assessed sepa a ely
and a ailable in mos coho s (Table 1).
Replica ion da a
Summa ized esul s om he CHARGE Conso ium’s me a-analyses o
genome-wide associa ion s udies be ween geno ypes and gene al
cogni i e unc ion (N= 53 949),
18
p ocessing speed by a me a-analysis o
ou es s o p ocessing speed including he DSST (N= 32 088)
19
and
execu i e unc ioning (STROOP; N= 7726)
19
we e used o assess causal
associa ions om he same se en TL associa ed gene ic a ian s.
16
The
CHARGE coho s we e all o Eu opean ances y and pa icipan s we e aged
45 yea s o olde . Some ENGAGE coho s con ibu ed o CHARGE analyses
o gene al cogni i e unc ion: BETULA1, ERF, HRS, and LBC1936; DSST and
STROOP: ERF; and DSST: LBC1936 (Table 1; see Supplemen a y Da a o
u he discussions on implica ions o ou s udy). The gene al cogni i e
unc ion pheno ype was c ea ed as a composi e sco e o mul iple cogni i e
es s
18
om p incipal componen analysis. The p ocessing speed a iable
was c ea ed om DSST and h ee simila es s, and execu i e unc ioning
was assessed by ei he T ail Making es s o S oop colo and wo d
in e e ence es s.
19
E ec sizes we e missing o DSST and STROOP; hence
esul s we e p esen ed in Z-sco es only.
S a is ical analyses
All a iables (TL and cogni i e ai s) we e Z- ans o med wi h sub ac ion
o he mean and di ision o s.d. o enable compa isons ac oss coho s. Age
g oups, ins ead o con inuous age, we e defined and used as co a ia es o
allow o non-linea age e ec s: (1) 0–29 yea s; (2) 30–59 yea s; (3) 60–79
yea s; and (4) 80+ yea s. Coho s wi h in o ma ion on APOE geno ype
pe o med addi ional analyses s a ified on ε4-ca ie s (ε4/ε4 and ε4/ε3)
and non-ca ie s (ε3/ε3, ε2/ε3 and ε2/ε2). Indi iduals wi h he geno ype ε2/
ε4 a e excluded om he analysis. All models a e desc ibed in de ail in he
Supplemen a y Da a; b iefly, all coho s con ibu ed wi h summa y da a
om h ee di e en models as depic ed in Figu e 1. Linea eg essions
we e fi ed o he associa ions o (1) TL on cogni i e ai (TL- ai ), (2) GRS
on TL (GRS-TL), and (3) GRS on cogni i e ai (GRS- ai ). All models we e
adjus ed o sex, age g oup and s udy-specific co a ia es. E ec es ima es
om all models we e pooled ac oss coho s ia fixed-e ec me a-analysis,
excep when e idence was ound o s a is ically significan he e ogenei y
(P- alueo0.05), in which case a andom-e ec s me a-analysis was
pe o med ins ead. Then, ins umen al a iable (IV) analysis was conduc ed
by calcula ing a Wald- ype causal es ima e o he e ec o TL on each
cogni i e ai (IV − ai = GRS − ai /GRS-TL). E ec di e ences be ween
obse ed (TL- ai ) and causal/p edic ed (IV- ai ) es ima es we e calcula ed
by sub ac ing he causal be a om he obse a ional be a in a Z- es
(Supplemen a y Da a). S a ified analyses on APOE geno ype we e done
simila ly. Fo summa y s a is ics da a om CHARGE, a causal es ima e was
calcula ed as desc ibed by Bu gess e al.
20
(Supplemen a y Da a). C ude
P- alues a e p esen ed o all associa ions, ha is, no mul iple es ing
co ec ions ha e been applied.
RESULTS
Obse a ional analyses o elome e leng h and cogni i e ai s
Significan associa ions, suppo ing he ela ionship be ween
longe elome es and be e cogni i e abili y, we e seen be ween
TL and MMSE, DSST and STROOP using fixed-e ec s me a-analysis
(Table 2, Figu e 2). Posi i e associa ions we e obse ed o MMSE
(0.025 pe s.d.-inc ease in TL; 95% confidence in e al (CI) 0.002,
0.047) and DSST (0.051; 95% CI 0.024, 0.077). Fo STROOP, a
nega i e be a (−0.053 pe s.d.-inc ease in TL; 95% CI −0.087,
−0.018) was seen, which was in acco dance wi h he hypo hesis
ha longe elome es a e associa ed wi h sho e ime o
comple ion o he S oop in e e ence es . Howe e , a e
co ec ions o mul iple compa isons, he associa ion wi h MMSE
was no significan .
Gene ic isk sco e o elome e leng h
The combined e ec o he GRS on TL was −0.048 s.d. −change o
TL pe allele (95% CI: −0.064, −0.032, P- alue = 4.0*10
−9
)
calcula ed using andom-e ec s me a-analysis (Supplemen a y
Da a: Supplemen a y Figu e S1). The co esponding F-s a is ic was
36, indica ing ha he GRS-TL es ima e p o ided a su ficien ly
s ong ins umen o u he use in IV analyses.
26
Al hough
he e ogenei y was de ec ed, all coho s showed nega i e e ec
sizes anging om −0.01 o −0.13 (Supplemen a y Table S1).
Addi ional es s in es iga ing possible pleio opic e ec s o SNP-
ai associa ions we e done and ound no e idence o such
(Supplemen a y Da a).
Ins umen al a iable analyses o elome e leng h and cogni i e
ai s
Ins umen al a iable analyses o causal associa ions o TL on
cogni i e pe o mance we e conduc ed o all cogni i e ai s.
Only he associa ion be ween TL and STROOP was ound o be
causal (Table 2), and o each s.d.-dec ease in TL an e ec change
o −0.60 in S oop sco e was de ec ed (95% CI: −1.12, −0.07,
P- alue = 0.026). Howe e , he associa ion would no be significan
Figu e 1. G aph desc ibing he design o he s udy. A gene ic isk
sco e (GRS) o elome e leng h (TL) is used in he ins umen al
a iable (IV) analysis o de e mine he p edic ed e ec o TL on
di e en cogni i e ai s (IV- ai ). The Mendelian Randomiza ion
design allows o calcula ion o an es ima e independen o
con ounde s (C) in compa ison o he obse ed e ec es ima ed
be ween TL and cogni i e ai s (TL- ai ).
Telome e leng h and cogni i e pe o mance
S Hägg e al
3
T ansla ional Psychia y (2017), 1 –6
a e mul iple es ing adjus men s. The di e ence in e ec sizes
be ween obse a ional and causal be as o STROOP was
s a is ically significan (Table 2).
S a ified analyses
S a ified me a-analyses we e pe o med o APOE ε4-ca ie s
(n⩽2380) and ε4 non-ca ie s (n⩽5669) sepa a ely (Supplemen a y
Da a). Obse a ional associa ions we e seen be ween TL and DSST
in non-ca ie s (β= 0.081, 95% CI: 0.045, 0.117, P- alue = 1 × 10
−5
)
and wi h be e pe o mance o STROOP in ca ie s (β=−0.074,
95% CI: −0.140, −0.009, P- alue = 0.027) (Supplemen a y Table
S4). A causal associa ion be ween long elome es and be e
pe o mance o STROOP was de ec ed amongs APOE ε4-ca ie s
(β=−0.70, 95% CI: −1.33, −0.07, P- alue = 0.030), al hough he
finding would no hold a e mul iple es ing adjus men s. No
o he causal e ec s o TL on cogni i e ai s we e seen in ei he
ca ie s o non-ca ie s (Supplemen a y Table S6).
Replica ion analyses
In eplica ion e o s, wo-sample MR analyses we e ca ied ou using
summa y s a is ics om CHARGE GWAS on gene al cogni i e
unc ion, DSST and STROOP. Da a om he TL GWAS we e used o
he gene ic ins umen (Supplemen a y Da a). Resul s p o ided
e idence o a causal associa ion be ween longe leukocy e TL and
be e gene al cogni i e unc ion (β= 0.086 pe s.d.-inc ease o TL,
95% CI: 0.016–0.156, P- alue = 0.016, Table 2) and be e DSST
Table 2. Associa ions be ween p edic ed and obse ed elome e leng h and di e en cogni i e ai s
Cogni i e ai ENGAGE
a
CHARGE
P edic ed e ec (IV- ai ) P edic ed e ec (IV- ai )
NBe a (95% CI) P- alue NBe a (95% CI) P- alue
MMSE 7066 0.291 (−0.05, 0.631) 0.095
DSST 4419 −0.016 (−0.437, 0.405) 0.941 32 088 2.021
b
0.043
BLOCK 5001 −0.192 (−0.594, 0.21) 0.349
MEMORY 13 060 −0.022 (−0.264, 0.22) 0.860
STROOP 2940 −0.598 (−1.125, −0.072) 0.026 7726 −0.780
b
0.435
Gene al 12 283 0.039 (−0.229, 0.306) 0.778 53 949 0.086 (0.016, 0.156) 0.016
Cogni i e ai Obse ed e ec (TL- ai )
Di -P- alue Be a (95% CI) P- alue
MMSE 0.13 0.025 (0.002, 0.047) 0.030
DSST 0.76 0.051 (0.024, 0.077) 0.0002
BLOCK 0.34 0.004 (−0.024, 0.032) 0.781
MEMORY 0.79 0.011 (−0.005, 0.028) 0.187
STROOP 0.04 −0.053 (−0.087, −0.018) 0.003
Gene al 0.89 0.020 (−0.008, 0.047) 0.156
Abb e ia ions: BLOCK, Block-design es ; CI, confidence in e al; Di -P- alue, es s o di e ence in es ima o s be ween obse ed and p edic ed e ec s; DSST,
Digi -symbol subs i u ion es ; Gene al, Gene al cogni i e pe o mance; IV, ins umen al a iable; MEMORY, Ve bal memo y o Pic u e lea ning es ; MMSE,
Mini-men al s a e exam; STROOP, S oop colo wo d ask in e e ence sco e; TL, elome e leng h.
a
All models a e adjus ed o age g oup and sex. No
adjus men s o mul iple es ing ha e been done on he epo ed P- alues.
b
Z-sco es, e ec sizes a e missing.
Figu e 2. Obse ed e ec s be ween elome e leng h and cogni i e ai s. Fixed-e ec s me a-analyses we e pe o med ac oss coho s and
domains o all cogni i e ai s. Significan e ec s (s.d.-change in cogni i e sco e o an s.d.-change in elome e leng h (TL)) we e ound o (a).
digi symbol subs i u ion es apping p ocessing speed (DSST), (b). Mini-men al s a e exam (MMSE), and (c). S oop in e e ence sco e apping
execu i e unc ioning (STROOP). All models we e adjus ed o age g oup, sex and s udy-specific co a ia es.
Telome e leng h and cogni i e pe o mance
S Hägg e al
4
T ansla ional Psychia y (2017), 1 –6

sco ing (Z-sco e = 2.02, P- alue = 0.043, Table 2). No e idence o a
causal e ec by TL on STROOP was ound (Table 2).
DISCUSSION
In he p esen s udy, we p o ide e idence o obse a ional and
causal associa ions be ween longe elome es and be e cogni-
i e pe o mance. By conduc ing a la ge me a-analysis o 12
coho s om Eu opean ances y popula ions wi h measu ed
elome es and assessmen s o cogni i e unc ion, we we e able
o obse e associa ions be ween TL and be e sco ing on MMSE,
DSST and STROOP. Mo eo e , APOE ε4-ca ie s seemed o ha e
di e en e ec s o he obse ed associa ion wi h wo se
pe o mance in DSST bu be e in STROOP. The associa ion
be ween longe elome es and as e comple ion o he S oop
in e e ence es was also ound o be significan in causal analysis
o all indi iduals and in APOE ε4-ca ie s only. Howe e , none o
he significan causal associa ions de ec ed o STROOP passed
mul iple es ing co ec ions. Hence, in line wi h his, using
summa y da a om CHARGE, we ound suppo o a causal
associa ion om TL on gene al cogni i e unc ion and DSST, bu
no on STROOP.
In he biology o aging, elome e leng h has long been
conside ed as a bioma ke eflec ing he unde lying cellula s a e.
Recen ly, howe e , se e al esea ch pape s ha e p esen ed
e idence o elome es being in ol ed in he p ocess o cellula
senescence causing inc eased isk o disease.
5,16,27
Mo eo e ,
o he s udies sugges elome es migh e en elonga e in soma ic
cells o main ain cellula s abili y,
21,28,29
al hough his phenom-
enon could be pa ly explained by leukocy e u no e o imp ecise
measu emen s. Ne e heless, elome e biology has implica ions
o he aging p ocesses and s udies a e wa an ed o elucida e he
ull complexi y.
Wi h his e o we demons a e se e al obse a ional associa-
ions be ween TL and cogni i e ai s, bo h confi ming ea lie
s udies and p esen ing new links. Gene al cogni i e unc ion is
usually ope a ionalized as a composi e sco e ac oss a numbe o
di e se cogni i e domains cap u ing mos o he cogni i e
a ia ion.
18
As an o e all measu e o cogni ion i also p edic s
mo ali y;
30
likewise, he leng h o elome es can be used o
p edic mo ali y.
31,32
Thus, i bo h gene al cogni i e unc ion and
TL se e as alid bioma ke s o aging, associa ions be ween hese
ma ke s a e expec ed, al hough causali y needs o be u he
in es iga ed. In he ENGAGE da a (N= 12 283), we we e no able o
de ec any associa ion be ween gene al cogni i e unc ion and TL,
bu using he CHARGE summa y da a (N= 53 949) we p o ided
e idence o a causal associa ion. I is likely ha he ENGAGE
analysis was low in powe ; e ec sizes had o e lapping CI’s. A
possible biological mechanism o a causal associa ion could be
explained by o e all body ail y; he leng hs o elome es a e
impo an o main aining cellula s abili y a old age and hence
also impo an o biological aging p ocesses such as decline in
cogni i e pe o mance.
1
Causal links ha e also been demons a ed
using animal models. A mouse wi h elome ase deficiency,
exp essing accele a ed aging wi h mal unc ioning issue epai
and impai ed neu ological unc ion, had es o ed unc ions again
upon elome ase eac i a ion.
33
The DSST es assesses p ocessing speed equi ed o ansla e a
code o symbols and digi s as as as possible in a gi en ime
ame. P ocessing speed has been demons a ed o ha e a s eady,
almos linea decline wi h ad ancing age, and i s decline leads
o he o ms o cogni i e decline.
34,35
Hence, in ligh o his i is no
su p ising ha we, and o he s,
3
de ec a ai ly s able obse a ional
associa ion o longe elome es and be e DSST sco ing. The MR
analysis did no indica e a causal associa ion in ou samples
(N= 4419); on he o he hand, when inc easing powe using
CHARGE da a (N= 32 088) we we e able o find suppo o a
posi i e causal ela ionship om longe TL on DSST sco ing.
Mo eo e , APOE ε4 non-ca ie s sco ed be e on he es wi h a
la ge e ec size seen in obse a ional analysis om TL on DSST.
Thus, as APOE geno ype is impo an o elucida ing di e en isk
g oups o many age- ela ed pheno ypes, i is possible ha i
applies o TL dynamics and cogni i e pe o mance as well.
Ya e e al.
3
showed ha longe TL a baseline ga e less
longi udinal decline in MMSE, and we p esen ed c oss-sec ional
e idence om obse a ional associa ions in line wi h hese
findings; longe TL is consis en wi h be e MMSE sco ing.
Howe e , while causal es ima es suppo hese associa ions, he
CI´s we e wide and esul s did no each s a is ical significance.
Un o una ely, CHARGE da a on MMSE we e no a ailable o
eplica ion analysis.
The STROOP a iable aps he execu i e unc ioning by a
combined colo and wo d es o be comple ed as quickly as
possible. To he bes o ou knowledge, he e has been only one
ea lie small s udy in es iga ing baseline and a i ion TL
associa ions wi h execu i e unc ioning, wi h inconclusi e
esul s.
8
Ou ENGAGE analysis included 2940 indi iduals whe e
we ound bo h an obse a ional and causal associa ion be ween
longe elome es and as e comple ion o he S oop es . The
la ge e ec size di e ence was howe e dis u bing and no
explained by addi ional adjus men s o smoking and alcohol
(Supplemen a y Da a). Fu he , causal associa ions did no hold
a e mul iple es ing co ec ions and when using he wo-sample
app oach including he la ge CHARGE da a (N= 7726) we could
no eplica e he associa ion, al hough he e ec s we e in he
same di ec ion. Hence, i is possible ha he STROOP finding
obse ed in he ENGAGE da a is a alse disco e y. In addi ion, he
s a ified analyses by APOE ε4 geno ype ound ε4-ca ie s o
pe o m be e , which is con adic o y o wha would be expec ed.
The s eng h o his s udy is he e o o combining mul iple
Eu opean coho s wi h TL, cogni i e and gene ic da a a ailable as
well as APOE geno ype. By doing so, we we e able o de ec
pa e ns o associa ions o di e en cogni i e ai s ha would
no be possible o find in single s udy analyses. Mo eo e , we
included la ge-scale CHARGE GWAS da a se s o pe o m wo-
sample MR analyses as eplica ion. The weaknesses o he s udy
include gene alizabili y, as he analyses we e pe o med solely in
Eu opean ances y popula ions, and some o he coho s we e
included in bo h ENGAGE and CHARGE analyses as desc ibed in
he supplemen . Mo eo e , he e ogenei y due o di e en issues
used (blood and sali a) and lab-specific echnical a iances (TL
es ima es om all 12 coho s we e done in fi e di e en labs) may
ha e d i en he esul s owa d null. Ano he limi a ion ela es o
he h ee assump ions o conduc ing MR s udies, which ha e
been conside ed as ollows: (1) a s ong gene ic ins umen should
be demons a ed be ween he GRS and TL (GRS-TL) which we
ha e (F-s a is ic = 36); (2) he gene ic ins umen should no be
con ounded by e.g., age and sex (unlikely conside ing he
andomiza ion o alleles a concep ion); and (3) pleio opic e ec s
(when o he pa hways exis om he TL SNPs o he ou come
(cogni i e ai ) wi hou going h ough he in e media e pheno-
ype (TL)) om he SNPs included in he GRS should be uled ou
as much as possible. We did no find e idence o pleio opic
e ec s (Supplemen a y Da a). Finally, also wo h men ioning a e
he ela i ely weak P- alues o some o he associa ions. The
obse a ional associa ion o MMSE would no hold a e
Bon e oni co ec ion o he P- alue o he six cogni i e ai s
es ed, likewise o he causal associa ions ound o STROOP.
To conclude, his s udy demons a es an o e all pic u e o he
impo ance o biological aging p ocesses such as TL dynamics o
main aining cogni i e unc ion h oughou li e. Mo e specifically,
we we e able o show obse a ional as well as causal associa ions
be ween TL and di e en cogni i e ai s ha ha e ne e been
elucida ed be o e. Hence, he cu en e o p esen s new
impo an pieces o e idence o he con inued sea ch o a
Telome e leng h and cogni i e pe o mance
S Hägg e al
5
T ansla ional Psychia y (2017), 1 –6
be e unde s anding o he biology behind aging and he ac o s
explaining heal hy aging.
CONFLICT OF INTEREST
The au ho s decla e no conflic o in e es .
ACKNOWLEDGMENTS
Eu opean Ne wo k o Gene ic and Genomic Epidemiology (ENGAGE) wi h FP7-
HEALTH-F4-2007 g an ag eemen n . 201413. The Coho s o Hea and Aging
Resea ch in Genomic Epidemiology Conso ium (CHARGE) which is suppo ed in pa
by he Na ional Hea , Lung and Blood Ins i u e g an HL105756 and he
neu oCHARGE cogni i e wo king g oup h ough he Na ional Ins i u e on Aging
g an AG033193. All coho -specific acknowledgmen s a e ound in he
Supplemen a y Da a.
REFERENCES
1 Blackbu n EH, Epel ES, Lin J. Human elome e biology: a con ibu o y and in e -
ac i e ac o in aging, disease isks, and p o ec ion. Science 2015; 350: 1193–1198.
2 Wikg en M, Ka lsson T, Lind J, Nilb ink T, Hul din J, Sleege s K e al. Longe
leukocy e elome e leng h is associa ed wi h smalle hippocampal olume among
non-demen ed APOE epsilon3/epsilon3 subjec s. PLoS ONE 2012; 7: e34292.
3 Ya e K, Lindquis K, Kluse M, Caw hon R, Ha is T, Hsueh WC e al. Telome e
leng h and cogni i e unc ion in communi y-dwelling elde s: findings om he
Heal h ABC S udy. Neu obiol Aging 2011; 32: 2055–2060.
4 Zek y D, He mann FR, I minge -Finge I, G a C, Gene C, Vi ale AM e al. Telome e
leng h and ApoE polymo phism in mild cogni i e impai men , degene a i e and
ascula demen ia. J Neu ol Sci 2010; 299: 108–111.
5 Zhan Y, Song C, Ka lsson R, Tillande A, Reynolds CA, Pede sen NL e al. Telome e
leng h sho ening and Alzheime disease—a Mendelian andomiza ion s udy.
JAMA Neu ology 2015; 72: 1202–1203.
6 De o e EE, P esco J, De Vi o I, G ods ein F. Rela i e elome e leng h and cog-
ni i e decline in he Nu ses' Heal h S udy. Neu osci Le 2011; 492:15–18.
7 Honig LS, Kang MS, Schup N, Lee JH, Mayeux R. Associa ion o sho e leukocy e
elome e epea leng h wi h demen ia and mo ali y. A ch Neu ol 2012; 69:
1332–1339.
8 Cohen-Manheim I, Donige GM, Sinn eich R, Simon ES, Pinchas R, A i A e al.
Inc eased a i ion o leukocy e elome e leng h in young adul s is associa ed wi h
poo e cogni i e unc ion in midli e. Eu J Epidemiol 2016; 31: 147–157.
9 Wikg en M, Ka lsson T, Nilb ink T, No d jall K, Hul din J, Sleege s K e al. APOE
epsilon4 is associa ed wi h longe elome es, and longe elome es among
epsilon4 ca ie s p edic s wo se episodic memo y. Neu obiol Aging 2012; 33:
335–344.
10 Zek y D, He mann FR, I minge -Finge I, O olan L, Gene C, Vi ale AM e al.
Telome e leng h is no p edic i e o demen ia o MCI con e sion in he oldes old.
Neu obiol Aging 2010; 31:719–720.
11 Ha is SE, Ma ioni RE, Ma in-Ruiz C, Pa ie A, Gow AJ, Cox SR e al. Longi udinal
elome e leng h sho ening and cogni i e and physical decline in la e li e: The
Lo hian Bi h Coho s 1936 and 1921. Mech Ageing De 2016; 154:43–48.
12 Palme TM, S e ne JA, Ha bo d RM, Lawlo DA, Sheehan NA, Meng S e al.
Ins umen al a iable es ima ion o causal isk a ios and causal odds a ios in
Mendelian andomiza ion analyses. Am J Epidemiol 2011; 173: 1392–1403.
13 Bu gess S, Bu e wo h A, Mala s ig A, Thompson SG. Use o Mendelian ando-
misa ion o assess po en ial benefi o clinical in e en ion. BMJ 2012; 345: e7325.
14 Clous on SA, B ews e P, Kuh D, Richa ds M, Coope R, Ha dy R e al. The dynamic
ela ionship be ween physical unc ion and cogni ion in longi udinal aging
coho s. Epidemiol Re 2013; 35:33–50.
15 Salmon DP, Fe is SH, Thomas RG, Sano M, Cummings JL, Spe ling RA e al. Age
and apolipop o ein E geno ype influence a e o cogni i e decline in non-
demen ed elde ly. Neu opsychology 2013; 27:391–401.
16 Codd V, Nelson CP, Alb ech E, Mangino M, Deelen J, Bux on JL e al. Iden ifica ion
o se en loci a ec ing mean elome e leng h and hei associa ion wi h disease.
Na Gene 2013; 45: 422–427.
17 Hagg S, Fall T, Plone A, Magi R, Fische K, D aisma HH e al. Adiposi y as a cause
o ca dio ascula disease: a Mendelian andomiza ion s udy. In J Epidemiol 2015;
44:578–586.
18 Da ies G, A ms ong N, Bis JC, B essle J, Chou aki V, Giddalu u S e al. Gene ic
con ibu ions o a ia ion in gene al cogni i e unc ion: a me a-analysis o
genome-wide associa ion s udies in he CHARGE conso ium (N = 53949). Mol
Psychia y 2015; 20: 183–192.
19 Ib ahim-Ve baas CA, B essle J, Debe e S, Schuu M, Smi h AV, Bis JC e al. GWAS
o execu i e unc ion and p ocessing speed sugges s in ol emen o he
CADM2 gene. Mol Psychia y 2016; 21:189–197.
20 Bu gess S, Sco RA, Timpson NJ, Da ey Smi h G, Thompson SG. Conso ium E-I
Using published da a in Mendelian andomiza ion: a bluep in o e ficien
iden ifica ion o causal isk ac o s. Eu J Epidemiol 2015; 30:543–552.
21 Cai N, Chang S, Li Y, Li Q, Hu J, Liang J e al. Molecula signa u es o majo
dep ession. Cu Biol 2015; 25: 1146–1156.
22 Mi chell C, Hobc a J, McLanahan SS, Siegel SR, Be g A, B ooks-Gunn J e al. Social
disad an age, gene ic sensi i i y, and child en's elome e leng h. P oc Na l Acad
Sci USA 2014; 111: 5944–5949.
23 Caw hon RM. Telome e measu emen by quan i a i e PCR. Nucleic Acids Res 2002;
30:e47.
24 Dege man S, Domello M, Land o s M, Linde J, Lundin M, Ha aldsson S e al. Long
leukocy e elome e leng h a diagnosis is a isk ac o o demen ia p og ession in
idiopa hic pa kinsonism. PLoS ONE 2014; 9: e113387.
25 Wel e D, MacA hu J, Mo ales J, Bu de T, Hall P, Junkins H e al. The NHGRI
GWAS Ca alog, a cu a ed esou ce o SNP- ai associa ions. Nucleic Acids Res 2014;
42(Da abase issue): D1001–D1006.
26 Bu gess S, Thompson SG. Use o allele sco es as ins umen al a iables o
Mendelian andomiza ion. In J Epidemiol 2013; 42: 1134–1144.
27 Zhang C, Dohe y JA, Bu gess S, Hung RJ, Linds om S, K a P e al. Gene ic
de e minan s o elome e leng h and isk o common cance s: a Mendelian
andomiza ion s udy. Hum Mol Gene 2015; 24: 5356–5366.
28 Ka o S, Shiels PG, McGuinness D, Lindholm B, S en inkel P, No d o s L e al.
Telome e a i ion and elonga ion a e ch onic dialysis ini ia ion in pa ien s wi h
end-s age enal disease. Blood Pu i 2016; 41:25–33.
29 A i A, Chen W, Ga dne JP, Kimu a M, B imacombe M, Cao X e al. Leukocy e
elome e dynamics: longi udinal findings among young adul s in he Bogalusa
Hea S udy. Am J Epidemiol 2009; 169:323–329.
30 Cal in CM, Dea y IJ, Fen on C, Robe s BA, De G, Leckenby N e al. In elligence in
you h and all-cause-mo ali y: sys ema ic e iew wi h me a-analysis. In J Epide-
miol 2011; 40:626–644.
31 Deelen J, Beekman M, Codd V, T ompe S, B oe L, Hagg S e al. Leukocy e
elome e leng h associa es wi h p ospec i e mo ali y independen o immune-
ela ed pa ame e s and known gene ic ma ke s. In J Epidemiol 2014; 43: 878–886.
32 Bakaysa SL, Mucci LA, Slagboom PE, Boomsma DI, McClea n GE, Johansson B e al.
Telome e leng h p edic s su i al independen o gene ic influences. Aging Cell
2007; 6: 769–774.
33 Jaskelio M, Mulle FL, Paik JH, Thomas E, Jiang S, Adams AC e al. Telome ase
eac i a ion e e ses issue degene a ion in aged elome ase-deficien mice.
Na u e 2011; 469:102–106.
34 Finkel D, Reynolds CA, McA dle JJ, Pede sen NL. Age changes in p ocessing speed
as a leading indica o o cogni i e aging. Psychol Aging 2007; 22:558–568.
35 Hagenaa s SP, Ha is SE, Da ies G, Hill WD, Liewald DC, Ri chie SJ e al. Sha ed
gene ic ae iology be ween cogni i e unc ions and physical and men al heal h in
UK Biobank (N = 112 151) and 24 GWAS conso ia. Mol Psychia y 2016; 21:
1624–1632.
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/
© The Au ho (s) 2017
Supplemen a y In o ma ion accompanies he pape on he T ansla ional Psychia y websi e (h p://www.na u e.com/ p)
Telome e leng h and cogni i e pe o mance
S Hägg e al
6
T ansla ional Psychia y (2017), 1 –6