Cardiovascular Risk Factors From Childhood and Midlife Cognitive Performance : The Young Finns Study

1 Ea ly li e ca dio ascula isk ac o s and midli e cogni i e pe o mance: The Ca dio ascula Risk in Young Finns S udy Sho i le: Childhood ascula ac o s and midli e cogni ion Su i P. Ro io, PhD,* Ka ja Pahkala, PhD,*† Jaakko Ne alainen, PhD,‡ Ma kus Juonala, MD, PhD,§ Pia Salo, MD, PhD,* Mika Kähönen, MD, PhD,ǁ Nina Hu i-Kähönen, MD, PhD,¶ Te ho Leh imäki, MD, PhD,# Ee o Jokinen, MD, PhD,†† Tomi Lai inen, MD, PhD,‡‡ Leena Tai onen, MD, PhD,§§ǁǁ Päi i Tossa ainen, MD, PhD,ǁǁ Jo ma S.A. Viika i, MD, PhD,§ Juha O. Rinne, MD, PhD,¶¶ Olli T. Rai aka i, MD, PhD*## *Resea ch Cen e o Applied and P e en i e Ca dio ascula Medicine, Uni e si y o Tu ku, Tu ku, Finland; †Paa o Nu mi Cen e, Spo s & Exe cise Medicine Uni , Depa men o Physical Ac i i y and Heal h, Uni e si y o Tu ku, Tu ku, Finland; ‡School o Heal h Sciences, Uni e si y o Tampe e, Tampe e, Finland; §Depa men o Medicine, Uni e si y o Tu ku and Di ision o Medicine, Tu ku Uni e si y Hospi al, Tu ku, Finland; ǁDepa men o Clinical Physiology, Uni e si y o Tampe e and Tampe e Uni e si y Hospi al, Tampe e, Finland; ¶Depa men o Pedia ics, Uni e si y o Tampe e and Tampe e Uni e si y Hospi al, Tampe e, Finland; #Depa men o Clinical Chemis y, Fimlab Labo a o ies and School o Medicine, Uni e si y o Tampe e, Tampe e, Finland; ††Depa men o Paedia ic Ca diology, Hospi al o Child en and Adolescen s, Uni e si y o Helsinki, Helsinki, Finland; ‡‡Depa men o Clinical Physiology, Uni e si y o Eas e n Finland and Kuopio Uni e si y Hospi al, Kuopio, Finland; §§Vaasa Cen al Hospi al, Vaasa, Finland; ǁǁDepa men o Pedia ics, PEDEGO Resea ch Uni and Medical Resea ch Cen e Oulu, Uni e si y o Oulu, Oulu, Finland; ¶¶Tu ku PET Cen e and Depa men o Neu ology, Uni e si y o Tu ku and Tu ku Uni e si y Hospi al, Tu ku, Finland; ##Depa men o Clinical Physiology and Nuclea Medicine, Uni e si y o Tu ku, Tu ku, Finland. Add ess o co espondence: Su i Ro io 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 10 20520 Tu ku, Finland Telephone: +358 40 5665036 Fax: 23337477 Email: [email p o ec ed] Acknowledgemen s: Expe echnical assis ance in da a managemen and s a is ical analyses by I ina Lisinen and Johanna Ikonen is g a e ully acknowledged. Sou ces o unding: The Young Finns S udy has been inancially suppo ed by he Academy o Finland: g an s 286284 (T.L.), 134309 (Eye), 126925, 121584, 124282, 129378 (Sal e), 117797 (Gendi), and 41071 (Skidi), he Social Insu ance Ins i u ion o Finland, Kuopio, Tampe e and Tu ku Uni e si y Hospi al Medical Funds, Juho Vainio Founda ion, Sig id Juselius Founda ion, Y jö Jahnsson Founda ion, Paa o Nu mi Founda ion, Finnish Founda ion o Ca dio ascula Resea ch, Finnish Cul u al Founda ion, Tampe e Tube culosis Founda ion and Emil Aal onen Founda ion. The au ho s Published o iginal i le: Ca dio ascula Risk Fac o s F om Childhood and Midli e Cogni i e Pe o mance : The Young Finns S udy 2 epo no ela ionships wi h indus y. The unde s o he s udy had no ole in s udy design, da a collec ion, da a analysis, da a in e p e a ion, w i ing o he manusc ip , o he decision o submi he manusc ip o publica ion. 3 ABSTRACT Backg ound: In adul s, high blood p essu e (BP), ad e se se um lipids, and smoking associa e wi h cogni i e de ici s. The e ec s o hese isk ac o s om childhood on midli e cogni i e pe o mance a e unknown. Objec i es: The aim o his s udy was o in es iga e he associa ions be ween childhood/adolescence ca dio ascula isk ac o s and midli e cogni i e pe o mance. Me hods: F om 1980, a popula ion-based coho o 3596 child en (baseline age 3-18 yea s) ha e been ollowed-up o 31 yea s in 3-9 yea in e als. BP, se um lipids, body mass index and smoking we e assessed in all ollow-ups. Cumula i e exposu e as he a ea unde he cu e o each isk ac o was de e mined in childhood (6-12 yea s), adolescence (12-18 yea s) and young adul hood (18-24 yea s). In 2011, cogni i e es ing was pe o med in 2026 pa icipan s aged 34- 49 yea s. Resul s: High sys olic BP, ele a ed se um o al-choles e ol and smoking om childhood we e independen ly associa ed wi h wo se midli e cogni i e pe o mance, especially memo y and lea ning. The numbe o ea ly li e isk ac o s, including high le els (ex eme 75 h pe cen ile o cumula i e isk exposu e be ween ages 6-24) o sys olic BP, o al-choles e ol and smoking associa ed in e sely wi h midli e isual and episodic memo y and isuospa ial associa i e lea ning (-0.140 s anda d de ia ions pe isk ac o , p<0.0001), and emained signi ican a e adjus men o con empo aneous isk ac o s. Indi iduals wi h all isk ac o s wi hin ecommended le els be ween ages 6-24 pe o med 0.29 s anda d de ia ions be e (p=0.006) on his cogni i e domain han hose exceeding all isk ac o guidelines a leas wice. This di e ence co esponds o he e ec o six yea s aging on his cogni i e domain. Conclusions: Cumula i e bu den o ca dio ascula isk ac o s om childhood/adolescence associa e wi h wo se midli e cogni i e pe o mance independen o adul hood exposu e. Key wo ds: cogni i e pe o mance, blood p essu e, se um choles e ol, body mass index, smoking Abb e ia ions ApoE = Apolipop o ein E AUC = A ea unde he cu e BMI = Body mass index BP = Blood p essu e HDL = High densi y lipop o ein LDL = Low densi y lipop o ein PAL = Pai ed associa es lea ning es RTI = Reac ion ime es RVP = Rapid isual in o ma ion p ocessing es SE = S anda d e o SD = S anda d de ia ion SWM = Spa ial wo king memo y es YFS = The Ca dio ascula Risk in Young Finns S udy Condensed Abs ac : In adul s, high blood p essu e (BP), ad e se se um lipids, and smoking associa e wi h cogni i e de ici s, bu he e ec s om childhood a e unknown. F om 1980, a popula ion-based coho o 3596 child en (age 3-18 yea s) ha e been egula ly ollowed-up o 4 31 yea s. BP, se um lipids, body mass index and smoking we e assessed in all ollow-ups. In 2011, cogni i e es ing was pe o med in 2026 pa icipan s aged 34-49 yea s. High sys olic BP, ele a ed se um o al-choles e ol and smoking om childhood we e independen ly associa ed wi h wo se midli e cogni i e pe o mance, especially memo y and lea ning. Ou indings suppo ac i e con olling o ca dio ascula isk ac o s om childhood. 5 In oduc ion Epidemiological e idence indica es ha exposu e o midli e high blood p essu e (BP), ad e se se um lipids, and smoking associa e wi h cogni i e decline la e in li e (1-4). S udies in animal models ha e also obse ed associa ions be ween hese isk ac o s and cogni i e pe o mance. Spon aneously hype ensi e a s show cogni i e decline compa ed o no mo ensi e s ains (5). In oden s (6,7), he high- a die induced hype choles e olemia leads o memo y de ici s. In addi ion, adolescen a s exposed o nico ine show long-las ing cogni i e de ici s (8). Al hough he mechanisms unde lying hese associa ions a e la gely unclea , expe imen al s udies ha e sugges ed ha ca dio ascula isk ac o s may damage bo h neu onal and ascula issues o he b ain. S udies in oden s ha e shown ha hype ension may al e ce eb al ascula u e, and e en ually lead o es ained unc ion o he blood-b ain ba ie (9). Mo eo e , expe imen s on oden s ha e indica ed ha die induced hype choles e olemia may in luence he exp ession o genes in he b ain ele an o cellula mechanisms o lea ning, memo y and neu odegene a ion (10). Simul aneously, expe imen s on oden and abbi b ains ha e shown e idence ha hype choles e olemia may induce in lamma o y changes (6,7,11), and associa e wi h dis u bed be a-amyloid me abolism (6,7,11). Fu he mo e, expe imen al e idence on oden b ain ha e sugges ed ha smoking exposu e may induce oxida i e s ess (12-14) which may igge ce eb al in lamma o y changes (12), and lead o neu opa hological changes ela ed o cogni i e decline such as accumula ion o be a-amyloid pep ide and phospho yla ion o au p o ein (13). I is he e o e plausible ha exposu e o ca dio ascula isk ac o s in ea ly li e may a ec la e cogni i e pe o mance. In suppo o his hypo hesis, a link be ween cumula i e bu den o young adul hood ca dio ascula isk ac o s and cogni i e pe o mance was shown in young and middle aged adul s in he CARDIA coho (15). Whe he a simila link exis s be ween 6 ea ly li e isk exposu e and adul hood cogni i e unc ion is unknown. We add essed his knowledge gap in he Ca dio ascula Risk in Young Finns S udy (YFS) ha has ollowed a popula ion-based sample o indi iduals om childhood o adul hood. As a pa o he 31-yea ollow-up s udy, cogni i e es ing was pe o med using a es ba e y ocusing on se e al cogni i e domains ha a e ela ed o b ain s uc u es ypically a ec ed in he ea ly s ages o cogni i e decline (16). We hypo hesized ha a g ea e bu den o ca dio ascula isk ac o s in childhood, adolescence and young adul hood, including epea edly assessed BP, se um lipids, and smoking, associa e wi h wo se cogni i e pe o mance in midli e. Me hods Popula ion This s udy is a pa o he na ional YFS, which is an ongoing longi udinal popula ion- based s udy on ca dio ascula isk ac o s om childhood o adul hood. The i s c oss-sec ional s udy including 3596 andomly selec ed child en and adolescen s (boys and gi ls; aged 3, 6, 9, 12, 15 and 18 yea s) was pe o med in 1980. The coho has been ollowed-up in egula in e als in 1983, 1986, 1989, 2001, 2007, and he la es ollow-up s udy was conduc ed in 2011. De ailed in o ma ion on he popula ion and p o ocol is epo ed elsewhe e (17). P ocedu es and measu emen s Ou come measu e cogni ion In 2011, a compu e ized cogni i e es ing ba e y (CANTAB®) was used o assess cogni i e pe o mance in 2026 pa icipan s. The YFS es ba e y included: 1) mo o sc eening es used as a aining/sc eening ool o indica e di icul ies in es execu ion, 2) pai ed associa es lea ning es (PAL) measu ed isual and episodic memo y and isuospa ial associa i e lea ning, 3) spa ial wo king memo y es (SWM) measu ed sho e m and spa ial 7 wo king memo y and p oblem sol ing, 4) eac ion ime es (RTI) measu ed eac ion and mo emen speed and a en ion, and 5) apid isual in o ma ion es (RVP) measu ed isual p ocessing, ecogni ion and sus ained a en ion. Each es p oduced se e al a iables. P incipal componen analysis was conduc ed o iden i y componen s accoun ing o he majo i y o he a ia ion wi hin he da ase . P incipal componen s we e c ea ed o each es o pe o mance in speci ic cogni i e domains. The mo o sc eening es componen was excluded om u he analyses due o ceiling e ec (i.e. all pa icipan s had he maximum sco e in his es ). O he componen s we e no malized using ank o de no maliza ion p ocedu e esul ing in ou a iables, each wi h mean 0 and s anda d de ia ion (SD) 1. All a ailable da a o each cogni i e es was used in he analyses, and he e o e, he numbe o pa icipan s a ies be ween he models (N=177 we e excluded due o echnical easons in some o he es domains and N=51 e used o pa icipa e in all o some o he es s). De ailed desc ip ion and alida ion o he cogni i e da a has been epo ed p e iously (18). Exposu e a iables blood p essu e, se um lipids, body mass index and smoking S anda d me hods we e used o measu ing BP, se um o al-choles e ol, high-densi y lipop o ein (HDL) choles e ol, and iglyce ides a baseline and all ollow-up s udies. Low densi y lipop o ein (LDL) choles e ol was calcula ed acco ding o F iedewald’s (19). De ails o hese me hods ha e been desc ibed p e iously (20). A all s udy phases, he pa icipan s’ weigh and heigh we e measu ed and BMI was calcula ed. To u ilize all a ailable epea edly measu ed exposu e da a o con inuous a iables, we es ima ed subjec -speci ic cu es o ca dio ascula isk ac o s by mixed model eg ession splines (21). The a ea unde he cu e (AUC) o con inuous isk a iables was e alua ed o indica e a long e m bu den o each measu ed a ibu e 8 (22). The AUC a iables we e de ined sepa a ely o childhood (6-12 yea s), adolescence (12-18 yea s), young adul hood (18-24 yea s) and ea ly li e (6-24 yea s). Fo in e p e abili y, he AUC a iables we e s anda dized esul ing in a iables wi h mean 0 and SD 1. Smoking exposu e was que ied h oughou he ollow-up ime. Smoking s a us was dicho omized in o daily smoke s and non-smoke s, de ined as cu en daily smoking (yes/no) a he baseline o a any o ollow-up s udies when he pa icipan s we e aged 12-24 yea s. In addi ion o s udying he associa ions using con inuous AUC a iables and he dicho omized smoking a iable, he combined e ec o se e al isk ac o s om childhood o young adul hood on midli e cogni i e pe o mance was analysed. Fo ha pu pose, a 3-le el a iable indica ing he numbe o ca dio ascula isk ac o s du ing ea ly li e (6-24 yea s) was c ea ed (1=no isk ac o s, 2=one isk ac o , 3= wo o h ee isk ac o s) including he a iables ha showed signi ican associa ion wi h cogni i e pe o mance in he mul i a ia e analyses (i.e. sys olic blood p essu e, se um o al-choles e ol and smoking). To de ine isk ac o s using he con inuous AUC a iables, he dis ibu ions o BP and se um o al-choles e ol we e dicho omized in o high (≥75 h pe cen ile) and low (<75 h pe cen ile) isk ac o le els (sensi i i y analyses we e addi ionally pe o med by using cu -poin s o 70 h, 80 h and 85 h pe cen iles). Such dicho omized a iables and he bina y smoking a iable we e summed o c ea e he a iable indica ing he numbe o isk ac o s ( ange 0-3) du ing ea ly li e. The age and sex speci ic mean alues o sys olic blood p essu e, o al- and LDL-choles e ol co esponding o he highes 75 h pe cen ile o he AUC a iables be ween ages 6-24 yea s a e p esen ed in Online Table 1. A simila a iable indica ing he numbe o midli e isk ac o s a he ime o cogni i e es ing was cons uc ed. Cu en midli e daily smoking was que ied a he ime o cogni i e es ing and ca ego ized as non-smoke s s. daily smoke s. 9 In addi ion o he a bi a ily selec ed isk ac o cu -poin s, we examined whe he he e ec o ea ly li e isk exposu e is a ibu able o le els o isk ac o s epea edly exceeding he ecommended guidelines. The age and sex speci ic cu -poin s we e conside ed o a iables showing signi ican e ec o cogni i e pe o mance in he mul i a ia e analyses i.e. sys olic BP (23), LDL-choles e ol (24), and smoking (25). The exac cu -poin s o hese isk ac o s a e p esen ed in Online Table 2. The pa icipan s we e classi ied in o: 1) no isk ac o le els exceeding guidelines o le els exceeding a mos once pe isk ac o , 2) isk ac o le els exceeding guidelines on one isk ac o a leas wice, 3) isk ac o le els exceeding guidelines a leas wice on wo isk ac o s, 4) isk ac o le els exceeding guidelines a leas wice on all isk ac o s. Co a ia es The ollowing p ima y co a ia es we e used in he analyses: age, sex, baseline household income, an ihype ensi e and dyslipidemia medica ion, diagnoses o ca dio ascula diseases and ype 1 and 2 diabe es melli us. Al oge he 1,901 indi iduals wi h cogni i e es da a had comple e da a on exposu e a iables and p ima y co a ia es. Age was de ined in ull yea s a he end o he yea 2011. Baseline household income, use o an ihype ensi e (N=192) o dyslipidemia (N=72) medica ion and ype 1 diabe es diagnoses (N=12) we e que ied. Pa icipan s we e classi ied as ha ing ype 2 diabe es based on sel - epo ed and egis e con i med diagnosis, sel - epo ed glucose lowe ing medica ion o abno mal plasma glucose o haemoglobin A1c alues (N=75). Diagnoses o ca dio ascula diseases (N=13) we e adjudica ed om he na ional hospi al discha ge egis e . Addi ionally, da a on ollowing co a ia es we e a ailable o es ic ed numbe s o pa icipan s, and we e he e o e used in addi ional analyses: childhood academic pe o mance (N=1684), adul hood educa ion (N=1894), apolipop o ein E 16 Online Table 2. The analyses conside ing isk ac o s exceeding ecommended guidelines indica ed ha he pe sons wi h ea ly li e isk ac o s wi hin he ecommended guidelines had 0.29 SD be e isual and episodic memo y and isuospa ial associa i e lea ning (PAL- es ) han hose exceeding he guidelines a leas wice on all isk ac o s (model adjus ed o age, sex, amily income, an ihype ension and dyslipidemia medica ions and diagnoses o ca dio ascula diseases and diabe es melli us: N=1568; β=-0.088, SE=0.032, p=0.007) (Figu e 2). This e ec emained iden ical a e u he adjus men s wi h childhood academic pe o mance, adul hood educa ion, apoEε4 geno ype, and physical ac i i y le el (N=1341; β=-0.077, SE=0.035, p=0.029). Discussion We ound ha inc easing cumula i e bu den o sys olic BP, se um o al- and LDL- choles e ol and smoking in childhood/adolescence associa ed wi h wo se isual and episodic memo y and isuospa ial associa i e lea ning a midli e (PAL- es ). Impo an ly, he associa ions we e independen o midli e exposu es o he same isk ac o s. These indings sugges ha he associa ions be ween ea ly li e ca dio ascula isk ac o s and midli e cogni i e pe o mance do no only e lec acking o isk ac o le els om childhood o adul hood, bu ha he isk ac o s po en ially s a o exe hei in luence on cogni i e pe o mance al eady in childhood. P e ious s udies ha e obse ed in e se associa ions be ween adul hood sys olic BP, se um choles e ol, BMI and smoking and midli e cogni i e pe o mance (15) o isk o la e-li e cogni i e de ici s (1-4). To he bes o ou knowledge, his is he i s s udy examining he cumula i e bu den o ca dio ascula isk ac o s om childhood in ela ion o cogni i e pe o mance in midli e. O he isk ac o s examined, we we e able o demons a e he independen e ec s o ea ly li e sys olic BP, se um o al- and LDL-choles e ol, and smoking. 17 Pa icipan s wi h hese ea ly li e isk ac o s, ac o ed as con inuous o bina y a iables, de ined ei he by using se e al a bi a y cu -poin s o by using cu en guidelines o a he oscle osis p e en ion, had wo se midli e cogni i e pe o mance han hose wi h low isk ac o le els. Al hough he obse a ional na u e o ou s udy p ecludes making clinical ecommenda ions, i p o ides e idence on cogni i e bene i s gained by main enance o ca dio ascula isk ac o s a low le els al eady om ea ly li e. I his hypo hesis is co ec , adop ing ac i e moni o ing and ea men s a egies agains ca dio ascula isk ac o s om childhood would be needed o u n he ocus o cogni i e de ici s p e en ion om seconda y and e ia y p e en ion o p imo dial p e en ion. Wi h he u u e YFS ollow-up da a on cogni i e pe o mance, we will be able o es ima e using obse a ional da a whe he changes in isk ac o le els om childhood o adul hood associa e wi h he de e io a ion o cogni i e unc ion be ween mid- o old adul hood. The ou cogni i e domains we e selec ed as ou comes in o de o ob ain a comp ehensi e ou look o cogni i e pe o mance. Based on p e ious expe imen al s udies in animals, we especially expec ed o see links be ween ca dio ascula isk ac o s and es s indica ing memo y and lea ning. Indeed, we ound ha he e ec s o ea ly li e isk exposu e was s ongly and obus ly associa ed wi h he isual and episodic memo y and isuospa ial associa i e lea ning (PAL- es ). Simila ly, he CARDIA s udy ound s ong associa ions be ween se um o al-choles e ol, blood p essu e and e bal memo y (15). Addi ionally, we ound weak links o smoking and BMI o he es measu ing sus ained a en ion and isual p ocessing (RVP- es ) ha we e o bo de line signi ican o no did su i e adjus men s. The CARDIA s udy used S oop es ha esembled he RVP- es used in ou s udy, and ound some associa ions wi h sys olic blood p essu e (15). We did no ind associa ions be ween isk exposu es and wo king memo y (SWM- es ) o eac ion ime (RTI- es ). E ec s on he eac ion ime we e no expec ed 18 based on p e ious animal da a. Addi ionally, he lack o e ec s o isk ac o exposu e on wo king memo y in midli e is pe haps also no su p ising because di icul ies in encoding and ecall migh become isible p io o di icul ies in o he ype o memo y unc ions (e.g. wo king memo y, ecogni ion) (26). Thus, ou esul s a e consis en wi h s udies in animal models ha ha e obse ed associa ions be ween ca dio ascula isk ac o s and memo y and lea ning. The neu al ne wo ks ela ed o he PAL- es localize mainly o medial empo al lobes, speci ically o hippocampus and pa ahippocampal gy us (16). These ana omical s uc u es a e esponsible o lea ning and memo y (27). Imaging s udies ha e epo ed ha exposu e o ca dio ascula isk ac o s is associa ed wi h inc eased amoun o ce eb al whi e ma e al e a ions and s uc u al b ain changes in he elde ly (28-31) and augmen he e ec o age on olume loss in hippocampus, en o hinal co ex and medial empo al lobes in heal hy adul s (32). These indings may o e insigh s in o he associa ions be ween ea ly ca dio ascula isk ac o s and medial empo al lobe ela ed b ain unc ions. Fu he mo e, in elde ly pa ien s wi h mild cogni i e impai men , de ici s in he PAL- es ha e been linked o p eclinical Alzheime ’s disease pa hology (33-36). The main neu opa hological mechanisms unde lying associa ions be ween ca dio ascula isk ac o s and old-age cogni i e de ici s a e sugges ed o be subclinical ischemia causing ce eb o ascula damage (27,37), s uc u al b ain changes and a ophy (28-31). Addi ionally, isk ac o s may in luence ce eb al β-amyloid p o ein me abolism (38,39). In young popula ions, howe e , he mechanisms emain unknown. Ca dio ascula isk ac o s cause sys emic a he oscle osis, loss o dis ensibili y in he ascula u e, essel ib osis, plasma p o ein leakage, and accumula ion o lipid-con aining mac ophages in he essels (40-42). In he b ain, hese changes may lead o ce eb al hypope usion and local in lamma ion which dis up he ulne able 19 su ounding neu onal milieu (37). Addi ionally, by inducing ce eb al hypope usion a he oscle o ic changes may ini ia e and/o accele a e neu odegene a i e changes (e.g. β-amyloid deposi ion and clea ance, synap ic and neu onal dys unc ion/loss) ha ul ima ely lead o al e a ions in cogni i e pe o mance (38,39). We ound ha he pa icipan s wi h se e al ea ly li e ca dio ascula isk ac o s, including ele a ed BP, high LDL-choles e ol and smoking, exceeding he ecommended guidelines had ~0.3 SD’s lowe pe o mance in he PAL- es han hose pa icipan s whose isk ac o s emained always wi hin he guidelines. When using a bi a y cu -poin s o BP and o al- choles e ol (exceeding 75 h pe cen iles), we simila ly ound ha indi iduals who had been exposed o all h ee isk ac o s in ea ly li e had ~0.4 SD’s lowe pe o mance in he PAL- es han indi iduals wi hou any ea ly li e isk ac o exposu es. In he YFS popula ion, we see a linea decline in he PAL- es pe o mance ha equals o 0.05 SD’s pe yea be ween ages 34 and 49 (18). The e o e, he p esen s udy sugges s ha he e ec o ea ly li e isk ac o clus e ing on he PAL- es pe o mance co esponds o 6-8 yea di e ence in cogni i e age. Fu he mo e, we ound ha he cumula i e e ec o he ea ly li e isk exposu e on he PAL- es pe o mance was s onge and independen o he e ec o cu en midli e isk exposu e. This obse a ion emphasizes he ole o ea ly li e isk exposu e on la e cogni i e pe o mance, and may be in line wi h he exis ence o di e en ial sensi i e pe iods and age windows o ulne abili y o en i onmen al ac o s and condi ions du ing b ain ma u a ion (43). Limi a ions o his s udy include ha cogni i e pe o mance was measu ed once. Cu en ly, his p e en s us om s udying he ole o ea ly li e ca dio ascula isk ac o s on changes in midli e cogni i e pe o mance. Fu he mo e, we we e unable o examine he ole o glucose le els as an ea ly li e isk ac o because we did no ha e da a o cons uc a cumula i e 20 exposu e a iable simila ly o sys olic blood p essu e and se um o al-choles e ol. Howe e , he CARDIA s udy ound no associa ion be ween young adul hood/midli e cumula i e blood glucose bu den and midli e cogni i e pe o mance among no moglycemic popula ion (15). Ano he limi a ion is he possibili y o esidual con ounding due o unmeasu ed ac o s, which is always possible in obse a ional s udies like he YFS. Ou esul s emained unchanged a e con olling he analyses o a wide a ay o possible con ounding ac o s including he adul hood le els o ca dio ascula isk ac o s. I emains possible, howe e , ha some unmeasu ed ac o s con ibu e o he associa ion be ween ca dio ascula isk ac o s and cogni i e pe o mance. Fu he mo e, compu e ized cogni i e es s a e no ou inely used in clinical se ings o diagnose cogni i e pe o mance. In his s udy, he es ba e y was no used o clinical decision making, bu as a ool o e alua e cogni i e pe o mance among heal hy young and middle aged adul s on a popula ion le el. P e ious s udies ha e shown ha hese es s a e use ul in cap u ing a ia ion in cogni i e pe o mance in heal hy popula ions (44-46). The e o e, he es s used in he YFS may be conside ed adequa e in disc imina ing he cogni i e a ia ion among he pa icipan s. Ano he po en ial limi a ion is a possible selec ion in he ollow-up s udy - he pa icipan s we e mo e o en women and olde han non-pa icipan s, and hey o igina ed om amilies wi h highe income and had be e childhood academic pe o mance. Howe e , no di e ences we e obse ed in he le els o ea ly li e exposu e a iables. Fu he mo e, we conduc ed se e al s a is ical es s in ou s udy, which may inc ease he p obabili y o alse posi i e indings. Howe e , as he main analyses we e based on a p io i hypo heses, we did no apply mul iple es ing co ec ion. Mo eo e , wi h espec o he es ablishmen o causali y, obse a ional s udies a e p one o bias caused by e e se causa ion. Ne e heless, he use o exis ing popula ion coho s om childhood o adul hood is he only ealis ic app oach o es he hypo hesis ha ea ly li e isk exposu e is 21 causally linked wi h adul cogni i e pe o mance, as i is no possible o pe o m li e-long andomized con ol ials o es causal ela ions be ween childhood isk ac o s and adul ou comes. The mos impo an compe ing explana ion o hese associa ions is ha cogni i e unc ion in ea ly li e migh de e mine (o migh associa e wi h o he ac o s de e mining) he eme gence o ascula isk ac o s. I ha hypo heses was ue, midli e cogni i e pe o mance would simply be a ma ke o acked ea ly li e cogni i e unc ion. We we e able o es his hypo heses in a es ic ed numbe o he YFS pa icipan s by using a ailable da a on academic pe o mance as a p oxy o hei childhood cogni i e pe o mance. Academic pe o mance was de ined as g ade poin a e age, which indica es he mean o all school g ades du ing one school yea . As an o e all measu e o academic pe o mance i may be conside ed as an indica o o he pa icipan s’ cogni i e abili y a baseline. Indeed, when in oduced as a co a ia e, he e ec o smoking was dilu ed. This sugges s ha he associa ion be ween ea ly li e smoking and midli e cogni i e pe o mance migh be con ounded by baseline cogni i e pe o mance. Howe e , he e ec s o o he isk ac o s as well as he e ec o he ea ly li e isk ac o clus e ing emained essen ially simila a e aking accoun he childhood academic pe o mance. Ne e heless, as he possibili y o esidual con ounding emains, he esul s in YFS should be eplica ed in o he longi udinal coho s wi h ollow-up om childhood o adul hood. Conclusions In summa y, hese da a indica e ha he cumula i e bu den o BP, se um o al- and LDL- choles e ol, and smoking om childhood and adolescence associa e independen ly and combined wi h midli e cogni i e pe o mance. The indings gi e suppo o ac i e moni o ing/ ea men s a egies agains ca dio ascula isk ac o s om childhood in o de o u n he ocus o cogni i e de ici s p e en ion o p ima y p e en ion. 22 Pe spec i es Compe ency in medical knowledge Cumula i e bu den o sys olic BP, se um o al- and LDL-choles e ol and smoking in childhood/adolescence associa e wi h wo se cogni i e pe o mance, especially memo y and lea ning, a midli e. Impo an ly, he associa ions we e independen o midli e exposu es o he same isk ac o s. These esul s gi e suppo o ac i e moni o ing and ea men s a egies agains ca dio ascula isk ac o s al eady ea lie du ing he li espan in o de o p omo e adul hood cogni i e heal h. The indings om he cu en s udy elucida e he possibili ies o mo e he ocus o cogni i e decline p e en ion om seconda y and e ia y p e en ion o p imo dial p e en ion h ough a ec ing ca dio ascula isk ac o s al eady om childhood and adolescence. T ansla ional ou look The molecula mechanisms o childhood/adolescence ca dio ascula isk ac o s on cogni i e decline om young adul hood o middle and old age equi e u he in es iga ion. 23 Re e ences 1. Knopman D, Boland LL, Mosley T, e al. Ca dio ascula isk ac o s and cogni i e decline in middle-aged adul s. Neu ology. 2001;56:42-48. 2. Tolppanen A-M, Solomon A, Soininen H, Ki ipel o M. Midli e ascula isk ac o s and Alzheime ’s disease: e idence om epidemiological s udies. J Alzheime s Dis. 2012;32:531- 540. 3. Rusanen M, Ro io S, Ngandu T, e al. Midli e smoking, apolipop o ein E and isk o demen ia and Alzheime ’s disease: a popula ion-based ca dio ascula isk ac o s, aging and demen ia s udy. Demen Ge ia Cogn Diso d. 2010;30:277-284. 4. Go elick PB, Scu e i A, Black SE, e al. Vascula con ibu ions o cogni i e impai men and demen ia: a s a emen o heal hca e p o essionals om he ame ican hea associa ion/ame ican s oke associa ion. S oke. 2011;42:2672-2713. 5. G ünbla E, Ba l J, Iuhos D-I, e al. Cha ac e iza ion o cogni i e de ici s in spon aneously hype ensi e a s, accompanied by b ain insulin ecep o dys unc ion. J Mol psychia y. 2015;3:6. 6. Ull ich C, Pi chl M, Humpel C. Hype choles e olemia in a s impai s he choline gic sys em and leads o memo y de ici s. Mol Cell Neu osci. 2010;45:408-417. 7. Thi umangalakudi L, P akasam A, Zhang R, e al. High choles e ol-induced neu oin lamma ion and amyloid p ecu so p o ein p ocessing co ela e wi h loss o wo king memo y in mice. J Neu ochem. 2008;106:475-485. 8. Couno e DS, Spijke S, Van de Bu gwal LH, e al. Long-las ing cogni i e de ici s esul ing om adolescen nico ine exposu e in a s. Neu opsychopha macology. 2009;34:299- 306. 24 9. Ueno M, Sakamo o H, Tomimo o H, e al. Blood-b ain ba ie is impai ed in he hippocampus o young adul spon aneously hype ensi e a s. Ac a Neu opa hol. 2004;107:532- 538. 10. Ma eos L, Ak e in S, Gil-Bea F-J, e al. Ac i i y- egula ed cy oskele on-associa ed p o ein in oden b ain is down- egula ed by high a die in i o and by 27-hyd oxycholes e ol in i o. B ain Pa hol. 2009;19:69-80. 11. Spa ks DL, Kuo YM, Rohe A, Ma in T, Lukas RJ. Al e a ions o Alzheime ’s disease in he choles e ol- ed abbi , including ascula in lamma ion. P elimina y obse a ions. Ann N Y Acad Sci. 2000;903:335-344. 12. Khanna A, Guo M, Meh a M, Royal W. In lamma ion and oxida i e s ess induced by ciga e e smoke in Lewis a b ains. J Neu oimmunol. 2013;254:69-75. 13. Ho Y-S, Yang X, Yeung S-C, e al. Ciga e e smoking accele a ed b ain aging and induced p e-Alzheime -like neu opa hology in a s. PLoS One. 2012;7:e36752. 14. Rue -Ba oso CR, T ajano ETL, Al es JN, e al. O gan- ela ed ciga e e smoke-induced oxida i e s ess is s ain-dependen . Med Sci Moni . 2010;16:BR218-26. 15. Ya e K, Vi ingho E, Ple che MJ, e al. Ea ly adul o midli e ca dio ascula isk ac o s and cogni i e unc ion. Ci cula ion. 2014;129:1560-1567. 16. de Ro e M, Pi on i VA, McCabe JA, e al. Hippocampal dys unc ion in pa ien s wi h mild cogni i e impai men : a unc ional neu oimaging s udy o a isuospa ial pai ed associa es lea ning ask. Neu opsychologia. 2011;49:2060-2070. 17. Rai aka i OT, Juonala M, Rönnemaa T, e al. Coho p o ile: he ca dio ascula isk in Young Finns S udy. In J Epidemiol. 2008;37:1220-1226. 25 18. Ro io SP, Pahkala K, Ne alainen J, e al. Cogni i e Pe o mance in Young Adul hood and Midli e: Rela ions Wi h Age, Sex, and Educa ion-The Ca dio ascula Risk in Young Finns S udy. Neu opsychology. 2016:30;532-42. 19. F iedewald WT, Le y RI, F ed ickson DS. Es ima ion o he concen a ion o low- densi y lipop o ein choles e ol in plasma, wi hou use o he p epa a i e ul acen i uge. Clin Chem. 1972;18:499-502. 20. Po kka K V, Rai aka i OT, Leino A, e al. T ends in se um lipid le els du ing 1980-1992 in child en and young adul s. The Ca dio ascula Risk in Young Finns S udy. Am J Epidemiol. 1997;146:64-77. 21. Welham S. Longi udinal Da a Analysis. In: Fi zmau ice G, Da idian M, Ve beke G, Molenbe ghs G, eds. Longi udinal Da a Analysis. Chapman & Hall /CRC; 2009:253-289. 22. Lai C-C, Sun D, Cen R, e al. Impac o long- e m bu den o excessi e adiposi y and ele a ed blood p essu e om childhood on adul hood le en icula emodeling pa e ns: he Bogalusa Hea S udy. J Am Coll Ca diol. 2014;64:1580-1587. 23. Kaelbe DC, Picke F. Simple able o iden i y child en and adolescen s needing u he e alua ion o blood p essu e. Pedia ics. 2009;123:e972-4. 24. Jolli e CJ, Janssen I. Dis ibu ion o lipop o eins by age and gende in adolescen s. Ci cula ion. 2006;114:1056-1062. 25. Ka ey R-EW, Daniels SR, Laue RM, A kins DL, Hayman LL, Taube K. Ame ican Hea Associa ion guidelines o p ima y p e en ion o a he oscle o ic ca dio ascula disease beginning in childhood. Ci cula ion. 2003;107:1562-1566. 26. D ag LL Bieliauskas LA. Con empo a y Re iew 2009: Cogni i e Aging. J Ge ia Psychia y Neu ol. 2010;23:75-93. 32 showing signi ican associa ion o cogni i e pe o mance and wi h guideline ecommenda ions o child en/adolescence (i.e. LDL-choles e ol, sys olic blood p essu e and smoking) we e included in he a iable o he ca dio ascula isk ac o clus e ing. The ba s indica e he mean alues o he p incipal componen o isual and episodic memo y and isuospa ial associa i e lea ning and he whiske s a e s anda d e o s. The indi iduals we e classi ied in o: 0) no isk ac o le els exceeding guidelines o le els exceeding a mos once pe isk ac o , 1) isk ac o le els exceeding guidelines wice o mo e on one isk ac o , 2) isk ac o le els exceeding guidelines wice o mo e on wo isk ac o s, 3) isk ac o le els exceeding guidelines wice o mo e on all isk ac o s. The in e se dose- esponse ela ion wi h cogni i e pe o mance was signi ican (β=-0.088, p=0.007), adjus ed o age, sex, amily income, an ihype ension and dyslipidemia medica ions and diagnoses o ca dio ascula diseases and diabe es melli us. The e e ence line is se on he popula ion mean. 33 Table 1. Pe o mance in he Pai ed Associa es Lea ning- es , es ima ed di e ence in cogni i e age and mean alues o isk ac o a iables ac oss he ea ly li e cumula i e ca dio ascula isk ac o bu den. Ea ly li e isk ac o bu den (be ween ages 6 and 24 yea s). PAL- es Di e ence in cogni i e age* Mean (SD) sys olic blood p essu e a age 6-9 yea s Mean (SD) sys olic blood p essu e a age 12-15 yea s Mean (SD) sys olic blood p essu e a age 18-24 yea s The a ea unde he cu e a iable o sys olic blood p essu e 1s qua ile (n=462) 2nd qua ile (n=461) 3 d qua ile (n=462) 4 h qua ile (n=463) 0.21 (1.00) 0.09 (0.99) -0.09 (0.98) -0.21 (0.98) e . +2.4 yea s +6.0 yea s +8.4 yea s 101.8 (6.1) 108.3 (6.0) 111.1 (6.3) 117.5 (6.8) 102.9 (6.3) 110.0 (5.4) 115.3 (5.5) 124.3 (8.1) 107.1 (7.4) 114.9 (6.1) 122.6 (5.9) 132.6 (8.4) Mean (SD) se um o al-choles e ol a age 6-9 yea s Mean (SD) se um o al-choles e ol a age 12-15 yea s Mean (SD) se um o al-choles e ol a age 18-24 yea s The a ea unde he cu e a iable o se um o al-choles e ol 1s qua ile (n=462) 2nd qua ile (n=461) 3 d qua ile (n=462) 4 h qua ile (n=463) 0.16 (0.96) -0.02 (0.98) 0.03 (0.98) -0.17 (1.04) e . +3.6 yea s +2.6 yea s +6.6 yea s 4.7 (0.5) 5.3 (0.4) 5.8 (0.4) 6.6 (0.6) 4.1 (0.5) 4.8 (0.4) 5.2 (0.4) 6.1 (0.7) 3.9 (0.5) 4.6 (0.4) 5.1 (0.5) 6.0 (0.7) Daily smoking age 6-12 yea s Daily smoking age 12-18 yea s Daily smoking age 18-24 yea s Daily smoking (be ween ages 12-24 yea s) No (n=1306) Yes (n=491) 0.04 (1.00) -0.13 (0.97) e . +3.4 yea s All pa icipan s we e non-smoke s 0 % 71.3 % 0 % 94.5 % Values a e means (s anda d de ia ions) o he pe o mance in he pai ed associa es lea ning (PAL)- es indica ing lea ning and memo y and o he ca dio ascula isk ac o alues in he qua iles o he a ea unde he cu e (AUC)- a iables o sys olic blood p essu e and se um o al-choles e ol and o dicho omized ea ly li e (age 12-24 yea s) smoking. Fo he PAL componen highe alues indica e be e pe o mance. 34 *The di e ence in cogni i e age has been calcula ed compa ing he di e ence in he PAL- es pe o mance be ween he isk ac o qua iles o ou p e ious inding on he e ec o age (-0.0.5 SD pe yea ) on he PAL- es .18 The lowes qua ile has been used as he e e ence ca ego y o all compa isons o cogni i e age. 35 Table 2. Associa ions be ween cumula i e bu den o ea ly li e ascula isk ac o s (age 6- 24 yea s) and midli e isual and episodic memo y and isuospa ial associa i e lea ning (PAL- es ) in 1733 YFS pa icipan s. ROW MODEL β coe icien (SE) p- alue A Unadjus ed bi a ia e models Sys olic blood p essu e -0.152 (0.023) <0.0001 Se um o al-choles e ol -0.122 (0.024) <0.0001 Body mass index 0.018 (0.026) 0.490 Smoking -0.182 (0.054) 0.001 B Age and sex adjus ed models Sys olic blood p essu e -0.067 (0.026) 0.010 Se um o al-choles e ol -0.064 (0.025) 0.010 Body mass index -0.004 (0.025) 0.870 Smoking -0.123 (0.053) 0.001 C Mul i a ia e model 1 Sys olic blood p essu e -0.076 (0.028) 0.006 Se um o al-choles e ol -0.059 (0.025) 0.018 Body mass index 0.026 (0.026) 0.315 Smoking -0.140 (0.053) 0.008 D Mul i a ia e model 2 Sys olic blood p essu e -0.064 (0.028) 0.023 Se um o al-choles e ol -0.053 (0.025) 0.037 Body mass index 0.029 (0.026) 0.281 36 Smoking -0.140 (0.053) 0.008 Values a e β coe icien s (s anda d e o s) and p- alues om linea models. Unadjus ed models a e conduc ed sepa a ely o a iables indica ing cumula i e bu den o ea ly li e (6-24 yea s) ca dio ascula isk ac o s (i.e. sys olic blood p essu e, se um o al-choles e ol and body mass index, adolescence and young adul hood smoking a he age o 12 o 24 yea s) wi hou co a ia es. Age and sex adjus ed models a e conduc ed sepa a ely o each ea ly li e ca dio ascula isk ac o . In he mul i a ia e model 1, all a iables indica ing cumula i e bu den o ea ly li e (6-24 yea s) ca dio ascula isk ac o s a e en e ed simul aneously in an age and sex adjus ed model. The mul i a ia e model 2 is adjus ed addi ionally o childhood amily income, adul hood an ihype ension and dyslipidemia medica ions, and diagnoses o ca dio ascula diseases and diabe es melli us. Va iables o cogni i e pe o mance (PAL- es ) and ca dio ascula isk ac o s a e s anda dized (mean 0, s anda d de ia ion 1), hus he be a coe icien s indica e he amoun o change in he PAL- es pe o mance in s anda d de ia ions when he cumula i e bu den (6-24 yea s) o an ea ly li e isk ac o inc eases one s anda d de ia ion. One s anda d de ia ion uni is equi alen o ~6 mmHg o sys olic blood p essu e, ~0.7 mmol/l (~27 mg/dL) o o al-choles e ol and ~2.4 kg/m2 o body mass index. Fo smoking, he be a coe icien es ima es he e ec o daily smoking be ween ages 12-24 yea s. Fo he PAL- es , lowe alues indica e lowe cogni i e pe o mance. In all models, he pa icipan s wi h missing da a on any o he a iables in he ully adjus ed model we e excluded om he analyses. Online Appendix – Ro io e al. 1 ONLINE APPENDIX SUPPLEMENTAL METHODS Cogni ion Du ing he ollow-up examina ion in 2011, a cogni i e es ing ba e y de eloped by he Camb idge Cogni ion (CANTAB®) was used o assess cogni i e unc ion among he YFS pa icipan s. The CANTAB® es is a compu e ized, p edominan ly non-linguis ic and cul u ally neu al es ocusing on a wide ange o cogni i e domains. The es is pe o med using a alida ed ouch-sc een compu e sys em. The ull es ba e y includes 25 indi idual es s om which, a sui able es ba e y o each pa icula s udy may be selec ed. In YFS, he es ba e y was selec ed so ha i could be accomplished in 20-30 minu es, and included es s ha a e sensi i e o aging.1,2 The es s included in he es ba e y measu ed se e al cogni i e domains: 1) sho e m memo y, 2) spa ial wo king memo y, 3) p oblem sol ing, 4) eac ion ime, 5) a en ion, 6) apid isual p ocessing, 7) isual memo y, 8) episodic memo y, and 9) isuospa ial lea ning. Cogni i e es ing was pe o med du ing clinical examina ion. Due o he blood sampling included in he s udy p o ocol, he subjec s came o he examina ions a e as ing a leas 12 hou s. They we e ins uc ed o a oid smoking, hea y physical ac i i y as well as d inking alcohol and co ee du ing he p e ious e ening and he mo ning be o e he examina ions. Be o e he cogni i e es ing he subjec s we e p o ided wi h a ligh snack including a whole meal oa -based snack biscui , a small po ion o ui /be y oa meal and weak ui /be y squash. Du ing cogni i e es ing he pa icipan s i s conduc ed a mo o sc eening es (MOT es ) measu ing psychomo o speed and accu acy. In his s udy, he mo o sc eening es was conside ed as a aining p ocedu e whe e he pa icipan s we e in oduced o he equipmen Online Appendix – Ro io e al. 2 used in he es ing, and a sc eening ool o poin ou any di icul ies in ision, mo emen , comp ehension o abili y o ollow simple ins uc ions. Pai ed associa es lea ning es (PAL- es ) was used o assess isual and episodic memo y as well as isuospa ial associa i e lea ning con aining aspec s o bo h delayed esponse p ocedu e and condi ional lea ning. Spa ial wo king memo y es (SWM- es ) was used o measu e abili y o e ain spa ial in o ma ion and o manipula e i ems s o ed in he wo king memo y, p oblem sol ing as well as he abili y o conduc a sel -o ganized sea ch s a egy. Reac ion ime es (RTI- es ) assessed speed o esponse and mo emen on asks whe e he s imulus was ei he p edic able (simple loca ion ask) o unp edic able ( i e-choice loca ion ask). Rapid isual in o ma ion es (RVP- es ) was used o assess, isual p ocessing, ecogni ion and sus ained a en ion. Each o he CANTAB® es s p oduced se e al a iables. Fo his s udy, p incipal componen analysis was conduc ed as a mul i a ia e echnique o da a educ ion and o iden i y componen s accoun ing o he majo i y o he a ia ion wi hin he cogni ion da ase . P incipal componen analysis was selec ed since i allows analyzing he majo sou ces o a ia ion in a mul i-dimensional da a wi hou in oducing inhe en bias. P incipal componen analyses we e pe o med sepa a ely o all indi idual es s. The i s componen s esul ing om hese analyses we e conside ed o ep esen cogni i e pe o mance ela ed o he pa icula cogni i e domain. A e c ea ing he es wise p incipal componen s hei dis ibu ions we e analyzed. The componen o mo o sc eening es was excluded om u he analyses because i did no disc imina e he subjec s indica ing a ceiling e ec . All o he componen s we e no malized based on he ank o de no maliza ion p ocedu e esul ing in i e sepa a e a iables, each wi h mean alue o 0 and s anda d de ia ion (SD) o 1. Online Appendix – Ro io e al. 3 Co a ia es Age was de ined in ull yea s a he end o he yea 2011. A he baseline, he sum o household income was assessed wi h an eigh -ca ego y ques ion and used as an indica o o he socio- economic s a us o he amily (he ea e amily income). Fo his s udy, he o iginal income ca ego ies we e con e ed o co espond o he alue o money in 2011, and a e ha combined in o ou ca ego ies: 1) <17 000 eu os/yea , 2) 17 000–27 000, 3) 27 000–37 000 eu os/yea , and 4) >37 000 eu os/yea . Da a on an ihype ensi e (N=192) and dyslipidemia (N=72) medica ions we e ob ained om he ques ionnai es in he la es ollow-up s udy (2011). Diagnoses o ca dio ascula diseases (N=13) we e adjudica ed om he na ional hospi al discha ge egis e . Diagnoses o ype 1 diabe es we e sel - epo ed in he las ollow-up s udy (yea 2011). Pa icipan s we e classi ied as ha ing ype 2 diabe es i , a any o he ollow-up isi s (2001, 2007 o 2011-2012), hei as ing plasma glucose alue was equal o g ea e han 7 mmol/l, o i hey epo ed ha ing he diagnosis made by a physician. In addi ion, indi iduals whose hemoglobin A1c was equal o g ea e han 6.5% (48 mmol/l) a 2011 ollow-up o who epo ed aking glucose lowe ing medica ion a 2007 o 2011 ollow-up we e classi ied as ha ing ype 2 diabe es. Finally, ype 2 diabe es diagnoses we e ob ained om he Na ional Social Insu ance Ins i u ion's D ug Reimbu semen Regis y (N=75). Fu he mo e, childhood academic pe o mance o he s udy pa icipan was exp essed by g ade poin a e age ha was calcula ed as a mean o g ades in all indi idual school subjec s epo ed in a ques ionnai e. In Finland, he school g ades a y on a scale be ween 4 indica ing ailu e (US g ade: F) and 10 indica ing excellen knowledge and skills (US g ade: A). In his s udy, he g ade poin a e age alues we e used as a p oxy o childhood cogni i e abili y. Adul hood educa ion was assessed wi h ques ionnai es a he ollow-up s udies in 2001, 2007 and 2011. Fo his s udy, he maximum yea s o educa ion was de e mined as a con inuous a iable om sel - epo ed da a conce ning o al yea s o educa ion. Apolipop o ein E (APOE) geno ypes we e analyzed wi h Online Appendix – Ro io e al. 4 2 single nucleo ide polymo phisms ( s429358 and s7412), and he APOE p omo e polymo phisms −219 and + 113 ( s405509 and s440446, espec i ely). In his s udy, subjec s we e di ided in o 1) apoEε4 ca ie s (a leas one ε4 allele) and 2) non-ca ie s (no ε4 alleles). Physical ac i i y was measu ed wi h a sel -adminis e ed ques ionnai e a baseline and in all ollow-ups. The ques ionnai e consis ed o i ems on he equency and in ensi y o physical ac i i y, equency o igo ous physical ac i i y, hou s spen on igo ous physical ac i i y, a e age du a ion o a physical ac i i y session, and pa icipa ion in o ganized physical ac i i y. A comp ehensi e physical ac i i y index was calcula ed o baseline and each ollow-up s udy simila ly o p e ious s udies.3 Mean o he adul hood (ages 24-49 yea s) physical ac i i y indexes was conside ed as an indica o o physical ac i i y le el in his s udy. S a is ical analyses Subjec -speci ic cu es o sys olic BP, se um o al-, HDL- and LDL-choles e ols, iglyce ides and BMI we e es ima ed by mixed model eg ession splines.4 The co a iance s uc u e o he longi udinal se ing was modelled by allowing o subjec speci ic eg ession spline coe icien s, which we e inco po a ed as andom e ec s o he model. To a oid o e i ing we educed he numbe o kno s ( wo kno s on he calenda ime om 1980 o 2011) o he subjec -speci ic pa om ha o he ixed e ec s pa ( ou kno s on age om 3 o 34 yea s). The mean p o ile was allowed o a y ac oss bi h coho s and sex in e ms o possibly di e en ixed e ec s pa s. Simila o he app oach o Lai e al (2014), we hen e alua ed he a ea unde he cu e (AUC) as a measu e o a long e m bu den o each o he measu ed a ibu es.5 Fo his s udy, he AUC a iables o BP, se um o al-, HDL- and LDL-choles e ols, iglyce ides and BMI we e de ined sepa a ely o childhood (age 6-12 yea s), adolescence (12-18 yea s), young adul hood (18-24 yea s) and ea ly li e (6-24 yea s). Fo in e p e abili y, he AUC a iables we e s anda dized esul ing in no mally dis ibu ed a iables wi h mean 0 and SD 1. Online Appendix – Ro io e al. 5 To analyze he e ec o isk ac o clus e ing om childhood o young adul hood on cogni i e pe o mance a midli e, a a iable indica ing he numbe o isk ac o s du ing ea ly li e (6-24 yea s) was c ea ed. Fi s , he AUC a iables o BP and se um o al-choles e ol we e dicho omized in o 1) high isk ac o le el (≥ 75 h pe cen ile; coded as 1) and 2) low isk ac o le el (<75 h pe cen ile; coded as 0). Smoking s a us was dicho omized simila ly in o 1) smoke s (coded as 1) and 2) non-smoke s (coded as 0). Finally, he dicho omic a iables o BP, se um o al-choles e ol and smoking we e summed o c ea e he a iable indica ing he numbe o isk ac o s ( ange 0-3) du ing ea ly li e. The numbe o pe sons wi h h ee ea ly li e ca dio ascula isk ac o s was subs an ially low (n=62) and he e o e, he pe sons wi h 2 o 3 isk ac o s we e conside ed as he highes g oup. A simila a iable indica ing he numbe o ca dio ascula isk ac o s a he ime o cogni i e es ing was o med based on he BP and se um o al-choles e ol alues (dicho omized om he 75 h pe cen ile simila ly o he ea ly li e AUC a iables) and smoking s a us du ing he la es ollow-up s udy ( ange o he sum a iable 0-3). The numbe o pa icipan s wi h h ee ca dio ascula isk ac o s in he la es ollow-up s udy was low (n=28), and he e o e, he pe sons wi h 2 o 3 isk ac o s we e again add essed o he same ca ego y. Sensi i i y analyses we e addi ionally pe o med using cu -poin s o 70 h, 80 h and 85 h pe cen iles o con inuous a iables (Online able 5). To in es iga e whe he he e ec o ca dio ascula isk ac o clus e ing is a ibu able o isk ac o exposu e le els exceeding ecommended guidelines, he age and sex speci ic guideline alues we e conside ed o sys olic BP6, LDL-choles e ol7 and smoking.8 Each subjec was classi ied as ha ing isk ac o le els below/abo e he ecommended guidelines a each age poin . Subsequen ly, he subjec s we e classi ied acco ding o he numbe o isk ac o s (i.e. sys olic BP, LDL-choles e ol and smoking) and imes when hey had isk ac o le els exceeding he guidelines be ween ages 6-24 in o: 1) no isk ac o le els exceeding guidelines o le els exceeding guidelines a mos once on each isk ac o , 2) isk ac o le els exceeding Online Appendix - Ro io e al 12 Online Table 3. Cha ac e is ics o he s udy popula ion. All (N=2026) Men (N=922) Women (N=1104) Backg ound cha ac e is ics Age, yea s (N=2026) A baseline 10.8 (5.0) 10.7 (5.1) 10.9 (5.0) A cogni i e es ing 41.8 (5.0) 41.7 (5.1) 41.9 (5.0) Family income a baseline, N (%), (N=1956) <17 000 eu os/yea 512 (26.2) 229 (25.7) 283 (26.6) 17 000–27 000 eu os/yea 575 (29.4) 270 (30.3) 305 (28.6) 27 000–37 000 eu os/yea 425 (21.7) 191 (21.5) 234 (22.0) >37 000 eu os/yea 444 (22.7) 200 (22.5) 244 (22.9) Childhood academic pe o mance (N=1777) 7.77 (0.7) 7.57 (0.7) 7.94 (0.7) ApoE ε4 ca ie s (≥one ε4 allele) (N=1909) 680 (35.6) 314 (36.6) 366 (34.8) Ea ly li e smoking, N (%), yes (N=1968) 544 (27.6) 291 (32.6) 253 (23.5) Physical ac i i y index ( ange 5-15), (N=2005) 9.21 (1.72) 9.14 (1.84) 9.26 (1.62) Ca dio ascula isk ac o s a baseline Sys olic blood p essu e, mmHg (N=2009) 112.8 (11.9) 113.8 (13.0) 112.0 (10.9) Dias olic blood p essu e, mmHg (N=1732) 68.6 (9.4) 68.9 (9.6) 68.3 (9.2) To al-choles e ol, mmol/l (N=2003) 5.3 (0.9) 5.2 (0.9) 5.4 (0.9) HDL-choles e ol, mmol/l (N=2002) 1.6 (0.3) 1.6 (0.3) 1.6 (0.3) LDL-choles e ol, mmol/l (N=2002) 3.4 (0.8) 3.4 (0.8) 3.5 (0.8) T iglyce ides, mmol/l (N=2003) 0.7 (0.3) 0.7 (0.3) 0.7 (0.3) Body mass index, kg/m2 (N=2011) 18.0 (3.1) 18.0 (3.2) 17.9 (3.1) Ca dio ascula isk ac o s a cogni i e es ing Sys olic blood p essu e, mmHg (N=2019) 118.9 (14.1) 122.9 (13.3) 115.6 (13.8) Dias olic blood p essu e, mmHg (N=2019) 74.9 (10.5) 77.8 (10.8) 72.4 (9.5) To al-choles e ol, mmol/l (N=2008) 5.2 (1.0) 5.3 (1.0) 5.1 (0.9) HDL-choles e ol, mmol/l (N=2006) 1.3 (0.3) 1.2 (0.3) 1.4 (0.3) LDL-choles e ol, mmol/l (N=1961) 3.3 (0.8) 3.4 (0.9) 3.1 (0.8) Online Appendix - Ro io e al 13 T iglyce ides, mmol/l (N=2008) 1.3 (1.2) 1.6 (1.2) 1.1 (1.2) Body mass index, kg/m2 (N=2020) 26.5 (5.1) 27.0 (4.4) 26.1 (5.5) A ea unde he cu e (AUC) a iables o ca dio ascula isk ac o s Childhood (6-12 yea s), mean (SD) Sys olic blood p essu e, mmHg*y s (N=2026) 647.5 (31.9) 646.0 (32.4) 648.8 (31.5) To al-choles e ol, mmol/l*y s (N=2026) 33.1 (4.0) 32.8 (4.1) 33.3 (3.9) HDL-choles e ol, mmol/l*y s (N=2026) 9.1 (1.4) 9.2 (1.4) 8.9 (1.4) LDL-choles e ol, mmol/l*y s (N=2026) 22.1 (4.3) 21.6 (4.3) 22.5 (4.2) T iglyce ides, mmol/l*y s (N=2026) 3.8 (1.1) 3.6 (1.1) 3.9 (1.1) Body mass index, kg/m2*y s (N=2026) 99.7 (9.7) 100.0 (9.8) 99.4 (9.6) Adolescence (12-18 yea s), mean (SD) Sys olic blood p essu e, mmhg*y s (N=2026) 685.8 (40.6) 699.3 (40.0) 674.5 (37.6) To al-choles e ol, mmol/l*y s (N=2026) 29.4 (4.1) 28.4 (4.0) 30.2 (4.0) HDL-choles e ol, mmol/l*y s (N=2026) 8.7 (1.4) 8.3 (1.3) 9.0 (1.4) LDL-choles e ol, mmol/l*y s (N=2026) 18.4 (3.9) 17.9 (3.9) 18.9 (3.8) T iglyce ides, mmol/l*y s (N=2026) 4.9 (1.4) 4.8 (1.4) 5.0 (1.3) Body mass index, kg/m2*y s (N=2026) 120.6 (14.3) 120.5 (14.4) 120.6 (14.2) Young adul hood (18-24 yea s), mean (SD) Sys olic blood p essu e, mmHg*y s (N=2026) 713.2 (53.1) 746.2 (45.6) 685.7 (42.1) To al-choles e ol, mmol/l*y s (N=2026) 29.4 (4.2) 28.3 (4.1) 30.4 (4.1) HDL-choles e ol, mmol/l*y s (N=2026) 8.1 (1.6) 7.2 (1.3) 8.8 (1.5) LDL-choles e ol, mmol/l*y s (N=2026) 17.7 (3.7) 17.2 (3.7) 18.1 (3.7) T iglyce ides, mmol/l*y s (N=2026) 6.2 (1.6) 6.2 (1.7) 6.2 (1.5) Body mass index, kg/m2*y s (N=2026) 135.2 (17.3) 138.1 (16.9) 132.8 (17.2) Cogni i e componen s, mean (SD) PAL- es (N=1848) Cogni i e componen s we e s anda dized; mean 0, SD 1 -0.1 (1.0) 0.1 (1.0) SWM- es (N=2011) 0.2 (1.0) -0.2 (1.0) RTI- es (N=1822) 0.2 (1.0) -0.2 (0.9) RVP- es (N=1975) 0.1 (1.0) -0.1 (1.0) Online Appendix - Ro io e al 14 Values a e means (s anda d de ia ions) o con inuous a iables and numbe s (pe cen ages) o ca ego ical a iables. S uden ’s - es and χ2- es we e used. Ea ly li e smoking was dicho omized in o hose smoking a any ollow-up phase be ween ages 12-24. Family income was measu ed as sum o pa icipan s’ pa en s’ income and ans o med o co espond wi h he cu en alue o money. Physical ac i i y index was calcula ed as a mean o adul hood (ages 24-29 yea s) physical ac i i y indexes ha we e c ea ed based on da a que ied a each ollow-up s udy on equency and in ensi y o physical ac i i y, equency o igo ous physical ac i i y, hou s spen on igo ous physical ac i i y, a e age du a ion o a physical ac i i y session, and pa icipa ion in o ganized physical ac i i y ( ange 5-15). P incipal componen analyses was used o calcula e componen s indica ing isual and episodic memo y and isuospa ial associa i e lea ning (PAL- es ), wo king memo y and p oblem sol ing (SWM- es ), eac ion ime (RTI- es ) and ecogni ion, isual p ocessing and sus ained a en ion (RVP- es ) based on CANTAB® cogni i e es ba e y. All compa isons be ween men and women we e s a is ically signi ican (p<0.05) excep o age a baseline and a he ollow-up, amily income, apoE ε4 allele, adul hood physical ac i i y index, baseline dias olic blood p essu e, baseline HDL-choles e ol, baseline body mass index, childhood and adolescence AUC’s o body mass index, and young adul hood iglyce ides o which he compa isons we e non-signi ican . The alues o se um o al-, HDL- and LDL-choles e ol a e con e ed in o mg/dl by mul iplying he alues in mmol/l by 39. The alues o iglyce ides a e con e ed in o mg/dl by mul iplying he alues in mmol/l by 89. Online Appendix - Ro io e al 15 Online Table 4. Baseline and ea ly li e cha ac e is ics o pa icipan s and non-pa icipan s. Pa icipan s (N=2026) Non-pa icipan s (N=1570) P- alue Sex, male 922 (45.51) 842 (53.63) <0.0001 Age a baseline, yea s 10.84 (5.01) 9.92 (4.92) <0.0001 Family income a baseline, N (%) <17 000 eu os/yea 512 (26.18) 438 (29.26) 17 000–27 000 eu os/yea 575 (29.40) 479 (32.00) 27 000–37 000 eu os/yea 425 (21.73) 309 (20.64) 0.003 >37 000 eu os/yea 444 (22.70) 271 (18.10) Ea ly li e smoking, N (%) 544 (27.64) 397 (28.14) 0.752 Childhood academic pe o mance 7.77 (0.73) 7.65 (0.74) <0.0001 ApoE ε4 ca ie s (≥one ε4 allele) 680 (35.62) 273 (37.19) 0.451 Sys olic blood p essu e, mmHg 112.80 (11.92) 112.21 (12.53) 0.165 Dias olic blood p essu e, mmHg 68.58 (9.39) 69.02 (9.84) 0.221 To al-choles e ol, mmol/l 5.29 (0.90) 5.31 (0.93) 0.551 HDL-choles e ol, mmol/l 1.56 (0.31) 1.56 (0.31) 0.939 LDL-choles e ol, mmol/l 3.42 (0.82) 3.45 (0.86) 0.427 T iglyce ides, mmol/l 0.67 (0.31) 0.66 (0.32) 0.434 Body mass index, kg/m2 17.97 (3.12) 17.69 (3.10) 0.009 Values a e means (s anda d de ia ions) o con inuous a iables and numbe o subjec s (pe cen ages) o ca ego ical a iables. Age and sex adjus ed p- alues we e calcula ed using linea models. The pa icipan s a e subjec s who pa icipa ed in he cogni i e es ing du ing he la es ollow-up s udy o he Ca dio ascula Risk in Young Finns S udy. The non-pa icipan s a e subjec s who did no pa icipa e in cogni i e es ing. Blood p essu e, se um lipids and body mass index a e measu ed a he baseline. Ea ly li e smoking was dicho omized in o hose smoking a any ollow-up phase be ween ages 12-24. Family income was measu ed as sum o Online Appendix - Ro io e al 16 pa icipan s’ pa en s’ income, ans o med o co espond wi h he cu en alue o money. G ade poin a e age ( ange om 4.0 o 10.0) is calcula ed as he mean o g ades on all school subjec s a baseline o ei he o he wo ollowing ollow-ups ( o hose pa icipan s who we e no o school age a he baseline). G ade poin a e age is conside ed as a measu e o childhood academic pe o mance. The alues o se um o al-, HDL- and LDL-choles e ol a e con e ed in o mg/dl by mul iplying he alues in mmol/l by 39. The alues o iglyce ides a e con e ed in o mg/dl by mul iplying he alues in mmol/l by 89. Online Appendix - Ro io e al 17 Online Table 5. Associa ions be ween childhood, adolescence and young adul hood ca dio ascula isk ac o s and midli e cogni i e pe o mance. Childhood (6-12 yea s) Adolescence (12-18 yea s) Young adul hood (18-24 yea s) β coe icien (SE) p- alue β coe icien (SE) p- alue β coe icien (SE) p- alue Sys olic blood p essu e PAL- es -0.058 (0.023) 0.013 -0.067 (0.026) 0.011 -0.097 (0.030) 0.001 SWM- es 0.006(0.022) 0.770 -0.001 (0.025) 0.956 -0.012 (0.029) 0.681 RTI- es 0.039 (0.024) 0.103 0.036 (0.027) 0.180 0.028 (0.030) 0.360 RVP- es -0.020 (0.023) 0.393 -0.034 (0.026) 0.196 -0.046 (0.030) 0.121 Dias olic blood p essu e PAL- es -0.024 (0.027) 0.382 -0.035 (0.027) 0.185 -0.053 (0.025) 0.035 SWM- es 0.032 (0.026) 0.216 0.029 (0.026) 0.264 0.020 (0.024) 0.406 RTI- es 0.027 (0.028) 0.328 -0.008 (0.027) 0.761 -0.009 (0.025) 0.728 RVP- es -0.030 (0.027) 0.266 -0.025 (0.027) 0.348 -0.028 (0.025) 0.266 Se um o al-choles e ol PAL- es -0.063(0.025) 0.010 -0.066 (0.025) 0.009 -0.066 (0.024) 0.007 SWM- es -0.031 (0.023) 0.180 -0.027(0.024) 0.263 -0.019 (0.023) 0.427 RTI- es 0.003 (0.025) 0.917 -0.001 (0.026) 0.976 -0.005 (0.025) 0.836 RVP- es -0.046 (0.025) 0.063 -0.047 (0.025) 0.062 -0.050 (0.024) 0.041 Se um LDL-choles e ol PAL- es -0.059 (0.028) 0.031 -0.056 (0.025) 0.025 -0.052 (0.023) 0.024 SWM- es -0.044 (0.026) 0.091 -0.036 (0.024) 0.133 -0.021 (0.022) 0.333 RTI- es -0.013 (0.028) 0.635 -0.016 (0.025) 0.515 -0.012 (0.024) 0.613 RVP- es -0.047 (0.027) 0.084 -0.039(0.025) 0.119 -0.041 (0.023) 0.078 Online Appendix - Ro io e al 18 Online Table 5. Associa ions be ween childhood, adolescence and young adul hood ascula isk ac o s and midli e cogni i e pe o mance 1 (con inued). 2 Childhood (6-12 yea s) Adolescence (12-18 yea s) Young adul hood (18-24 yea s) β coe icien (SE) p- alue β coe icien (SE) p- alue β coe icien (SE) p- alue Se um HDL-choles e ol PAL- es -0.064 (0.037) 0.087 -0.059 (0.035) 0.098 -0.048 (0.031) 0.117 SWM- es -0.020 (0.036) 0.576 -0.016 (0.034) 0.640 -0.019 (0.029) 0.519 RTI- es 0.020 (0.038) 0.594 0.014 (0.036) 0.708 -0.003 (0.031) 0.919 RVP- es -0.014 (0.038) 0.706 -0.012 (0.036) 0.746 -0.001 (0.031) 0.980 Se um iglyce ides PAL- es -0.045 (0.027) 0.093 -0.053 (0.027) 0.049 -0.042 (0.022) 0.055 SWM- es -0.034 (0.026) 0.182 -0.012 (0.025) 0.647 -0.001 (0.021) 0.972 RTI- es -0.030 (0.028) 0.283 0.001 (0.027) 0.976 -0.010 (0.022) 0.646 RVP- es -0.031 (0.027) 0.242 -0.053 (0.027) 0.045 -0.040 (0.022) 0.070 Body mass index PAL- es -0.020 (0.024) 0.415 -0.014 (0.025) 0.570 -0.009 (0.024) 0.709 SWM- es 0.027 (0.023) 0.240 0.026 (0.023) 0.272 0.034 (0.023 0.149 RTI- es 0.024 (0.025) 0.324 0.020 (0.025) 0.414 0.005 (0.025) 0.850 RVP- es -0.036 (0.024) 0.136 -0.052 (0.024) 0.034 -0.076 (0.024) 0.024 Adolescence and young adul hood (12-24 yea s) Smoking β coe icien (SE) p- alue PAL- es All 6-12-yea -old pa icipan s we e non-smoke s -0.123 (0.053) 0.020 SWM- es 0.020 (0.050) 0.692 RTI- es -0.037 (0.053) 0.489 Online Appendix - Ro io e al 19 RVP- es -0.116 (0.052) 0.027 1 Online Appendix - Ro io e al 20 The alues a e β coe icien s (s anda d e o s) and p- alues om linea models. Va iables o cogni i e pe o mance and ca dio ascula isk ac o s a e s anda dized (mean 0, s anda d de ia ion 1) and hus he be a coe icien s indica e he amoun o change in s anda d de ia ions when isk ac o s inc ease one s anda d de ia ion. All models we e adjus ed o age and sex. Sys olic and dias olic blood p essu e, se um lipids and body mass index we e ea ed as con inuous a iables. P incipal componen analyses was used o calcula e componen s indica ing isual and episodic memo y and isuospa ial associa i e lea ning (PAL- es ), wo king memo y and p oblem sol ing (SWM- es ), eac ion ime (RTI- es ) and ecogni ion, isual p ocessing and sus ained a en ion (RVP- es ) based on CANTAB cogni i e es ba e y. In he a iables o cogni i e pe o mance, lowe alues indica e lowe cogni i e pe o mance. The analyses we e conduc ed using all a ailable da a o each cogni i e es ; in he analyses o sys olic and dias olic blood p essu e, se um o al- and LDL-choles e ol, se um iglyce ides, and body mass index N=1781 in he PAL- es , N=1941 in he SWM- es , N=1756 in he RTI- es , and N=1906 in he RVP- es . In he analyses o HDL-choles e ol N=1780 in he PAL- es , N=1940 in he SWM- es , N=1755 in he RTI- es , and N=1905 in he RVP- es and in he analyses o smoking N=1733 in he PAL- es , N=1886 in he SWM- es , N=1708 in he RTI- es , and N=1851 in he RVP- es . Online Appendix - Ro io e al 21 Online Table 6. Sensi i i y analyses o he numbe o ea ly and midli e ca dio ascula isk ac o s. Ex eme 70 h pe cen ile Ex eme 75 h pe cen ile Ex eme 80 h pe cen ile Ex eme 85 h pe cen ile β es ima e (SE) p- alue β es ima e (SE) p- alue β es ima e (SE) p- alue β es ima e (SE) p- alue Uni a ia e models Numbe o ea ly li e isk ac o s Numbe o midli e isk ac o s -0.136 (0.033) -0.090 (0.032) <0.001 0.006 -0.135 (0.034) -0.078 (0.034) <0.001 0.022 -0.120 (0.035) -0.063 (0.035) <0.001 0.071 -0.111 (0.037) -0.061 (0.037) 0.002 0.103 Mul i a ia e model Numbe o ea ly li e isk ac o s Numbe o midli e isk ac o s -0.115 (0.035) -0.056 (0.035) 0.001 0.110 -0.119 (0.036) -0.045 (0.037) 0.001 0.220 -0.107 (0.038) -0.035 (0.038) 0.004 0.351 -0.099 (0.039) -0.038 (0.040) 0.011 0.345 A 3-le el a iable indica ing he numbe o isk ac o s du ing ea ly (age 6-24 yea s) and midli e (a he la es ollow-up in 2011; age 34-49 yea s) was c ea ed om he a ea unde he cu e (AUC) a iables o each ca dio ascula isk ac o (1=no isk ac o s, 2=one isk ac o , 3= wo o h ee isk ac o s). The AUC a iables o BP and se um o al-choles e ol we e dicho omized in o high isk ac o le el (≥ 70 h/75 h/80 h/85 h pe cen ile)