Genome-wide association meta-analysis of fish and EPA+DHA consumption in 17 US and European cohorts

RESEARCH ARTICLE Genome-wide associa ion me a-analysis o ish and EPA+DHA consump ion in 17 US and Eu opean coho s Da iush Moza a ian 1 *, Hassan S Dash i 1,2 , Ma y K Wojczynski 3 , Aud ey Y Chu 4 , Jenni e A Ne le on 5 , Sa u Ma ¨nnis o ¨ 6 , Ka i K is iansson 6 , Ma ¨gi Reedik 7 , Ja i Lah i 8,9 , Denise K Hous on 10 , Ma ilyn C Co nelis 11 , F ank J. A an Rooij 12,13 , Ma ia Dimi iou 14 , S a oula Kanoni 15 , Ve a Mikkila ¨ 16,17 , Lyn M S e en 18 , Ma cia C de Oli ei a O o 5 , Lu Qi 13 , B uce Psa y 19 , Luc Djousse 20 , Je ome I Ro e 21 , Kenne Ha ald 6 , Ma kus Pe ola 6,22,23 , Ha i Rissanen 6 , An i Jula 6 , Fische K is a 7 , E elin Mihailo 7 , Ma y F Fei osa 24 , Julius S Ngwa 25,26 , Lu ing Xue 25 , Paul F Jacques 1,27 , Mia-Ma ia Pe a ¨la ¨ 28 , Aa no Palo ie 22,29 , Yongmei Liu 30 , Nike A Nalls 31 , Luigi Fe ucci 32 , Dena He nandez 31 , Ani Manichaikul 33 , Michael Y Tsai 34 , Jessica C Kie e-de Jong 35,36 , Albe Ho man 12,13,37 , And e ´G Ui e linden 12,13 , Loukianos Rallidis 38 , Paul M Ridke 6,39 , Lynda M Rose 4 , Julie E Bu ing 4,39 , Te ho Leh ima ¨ki 40 , Mika Ka ¨ho ¨nen 41 , Jo ma Viika i 42 , Rozenn Lemai e 43 , Veikko Salomaa 6 , Paul Knek 6 , And es Me spalu 7 , Ing id B Bo ecki 24 , L. Ad ienne Cupples 25,44 , Johan G E iksson 6,45,46 , S ephen B K i che sky 10 , S e ania Bandinelli 47 , Da id Sisco ick 48 , Osca H F anco 12,13 , Panos Deloukas 15,49 , Geo ge Dedoussis 14 , Daniel I Chasman 4,39 , Olli Rai aka i 17,50 , Toshiko Tanaka 32 1F iedman School o Nu i ion Science & Policy, Tu s Uni e si y, Bos on, MA, Uni ed S a es o Ame ica, 2Nu i ion and Genomics Labo a o y, Jean Maye US Depa men o Ag icul u e Human Nu i ion Resea ch Cen e on Aging, Tu s Uni e si y, Bos on, MA, Uni ed S a es o Ame ica, 3Di ision o S a is ical Genomics, Depa men o Gene ics, Washing on Uni e si y School o Medicine, S . Louis, MO, Uni ed S a es o Ame ica, 4Di ision o P e en i e Medicine, B igham and Women’s Hospi al, Bos on, MA, Uni ed S a es o Ame ica, 5Di ision o Epidemiology, Human Gene ics and En i onmen al Sciences, Uni e si y o Texas Heal h Science Cen e a Hous on, Hous on, TX, Uni ed S a es o Ame ica, 6Na ional Ins i u e o Heal h and Wel a e, Helsinki, Finland, 7Es onian Genome Cen e , Uni e si y o Ta u, Ta u, Es onia, 8Ins i u e o Beha iou al Sciences, Uni e si y o Helsinki, Helsinki, Finland, 9Folkha ¨lsan Resea ch Cen e, Helsinki, Finland, 10 S ich Cen e on Aging, Wake Fo es School o Medicine, Wins on Salem, NC, Uni ed S a es o Ame ica, 11 Depa men o Nu i ion, Ha a d School o Public Heal h, Bos on, MA, Uni ed S a es o Ame ica, 12 Depa men o Epidemiology, E asmus MC, Ro e dam, The Ne he lands, 13 Ne he lands Genomics Ini ia i e-sponso ed Ne he lands Conso ium o Heal hy Aging, Leiden, The Ne he lands, 14 Depa men o Die e ics and Nu i ion, Ha okopio Uni e si y, A hens, G eece, 15 William Ha ey Resea ch Ins i u e, Ba s and The London School o Medicine and Den is y, Queen Ma y Uni e si y o London, London, Uni ed Kingdom, 16 Depa men o Food and En i onmen al Sciences, Uni e si y o Helsinki, Finland, 17 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, 18 Di ision o Epidemiology and Communi y Heal h, Uni e si y o Minneso a School o Public Heal h, Minneapolis, MN, Uni ed S a es o Ame ica, 19 Depa men o Medicine, Epidemiology and Heal h Se ices, Uni e si y o Washing on, Sea le, WA, Uni ed S a es o Ame ica, 20 Depa men o Medicine, Ha a d Medical School, and Di ision o Aging B igham and Women’s Hospi al, Bos on, MA, Uni ed S a es o Ame ica, 21 Los Angeles Biomedical Resea ch Ins i u e and Depa men o Pedia ics, Ha bo -UCLA Medical Cen e , Los Angeles, CA, Uni ed S a es o Ame ica, 22 Ins i u e o Molecula Medicine Finland (FIMM), Uni e si y o Helsinki, Helsinki, Finland, 23 Uni e si y o Ta u, Es onian Genome Cen e , Ta u, Es onia, 24 Di ision o S a is ical Genomics, Depa men o Gene ics, Washing on Uni e si y School o Medicine, S . Louis, MO, Uni ed S a es o Ame ica, 25 Depa men o Bios a is ics, Bos on Uni e si y School o Public Heal h, Bos on, MA, Uni ed S a es o Ame ica, 26 Di ision o Ca dio ascula Medicine, Howa d Uni e si y College o Medicine, Washing on DC, Uni ed S a es o Ame ica, 27 Nu i ional Epidemiology P og am, USDA Human Nu i ion Resea ch Cen e on Aging, Tu s Uni e si y, Bos on, MA, Uni ed S a es o Ame ica, 28 Depa men o Ch onic Disease P e en ion, Na ional Ins i u e o Heal h and Wel a e, Helsinki, Finland, 29 Depa men o Medical Gene ics, Uni e si y o Helsinki and Uni e si y Cen al Hospi al, Helsinki, Finland, 30 Depa men o Bios a is ical Sciences, Wake Fo es School o Medicine, Wins on Salem, NC, Uni ed S a es o Ame ica, 31 Labo a o y o Neu ogene ics, Na ional Ins i u e o Aging, Be hesda, MD, Uni ed S a es o Ame ica, 32 Clinical Resea ch B anch, Na ional Ins i u e on Aging, Bal imo e, MD, Uni ed S a es o Ame ica, 33 Cen e o Public Heal h Genomics and Di ision o Bios a is ics and Epidemiology, Depa men o Public PLOS ONE | h ps://doi.o g/10.1371/jou nal.pone.0186456 Decembe 13, 2017 1 / 15 a1111111111 a1111111111 a1111111111 a1111111111 a1111111111 OPEN ACCESS Ci a ion: Moza a ian D, Dash i HS, Wojczynski MK, Chu AY, Ne le on JA, Ma¨nnis o¨S, e al. (2017) Genome-wide associa ion me a-analysis o ish and EPA+DHA consump ion in 17 US and Eu opean coho s. PLoS ONE 12(12): e0186456. h ps://doi.o g/10.1371/jou nal.pone.0186456 Edi o : Philipp D Koellinge , V ije Uni e si ei Ams e dam, NETHERLANDS Recei ed: July 9, 2015 Accep ed: Sep embe 14, 2017 Published: Decembe 13, 2017 Copy igh : This is an open access a icle, ee o all copy igh , and may be eely ep oduced, dis ibu ed, ansmi ed, modi ied, buil upon, o o he wise used by anyone o any law ul pu pose. The wo k is made a ailable unde he C ea i e Commons CC0 public domain dedica ion. Da a A ailabili y S a emen : The me a-analysis esul s om his s udy a e a ailable a dbGAP (accession numbe phs000930). Funding: The A he oscle osis Risk in Communi ies (ARIC) s udy is ca ied ou as a collabo a i e s udy suppo ed by Na ional Hea , Lung, and Blood Ins i u e con ac s N01-HC-55015, N01-HC-55016, N01-HC-55018, N01-HC-55019, N01-HC-55020, N01-HC-55021, N01-HC-55022, R01HL087641, R01HL59367 and R01HL086694; Na ional Human Genome Resea ch Ins i u e con ac Heal h Sciences, Uni e si y o Vi ginia, Cha lo es ille, VA, Uni ed S a es o Ame ica, 34 Depa men o Labo a o y Medicine and Pa hology, Uni e si y o Minneso a, Minneapolis, MN, Uni ed S a es o Ame ica, 35 Depa men o Epidemiology, E asmus MC, Ro e dam, The Ne he lands, 36 Leiden Uni e si y College, The Hague, The Ne he lands, 37 Depa men o In e nal Medicine, E asmus MC, Ro e dam, The Ne he lands, 38 Second Depa men o Ca diology, Uni e si y Gene al Hospi al A ikon, A hens, G eece, 39 Ha a d Medical School, Bos on MA, Uni ed S a es o Ame ica, 40 Depa men o Clinical Chemis y, Fimlab Labo a o ies, Uni e si y o Tampe e School o Medicine, Tampe e, Finland, 41 Depa men o Clinical Physiology, Uni e si y o Tampe e and Tampe e Uni e si y Hospi al, Tampe e, Finland, 42 Depa men o Medicine, Uni e si y o Tu ku and Tu ku Uni e si y Hospi al, Tu ku, Finland, 43 Depa men o Medicine, Uni e si y o Washing on, Sea le, WA, Uni ed S a es o Ame ica, 44 NHLBI F amingham Hea S udy, F amingham, MA, Uni ed S a es o Ame ica, 45 Depa men o Gene al P ac ice and P ima y heal h Ca e, Uni e si y o Helsinki, Helsinki, Finland, 46 Helsinki Uni e si y Cen al Hospi al, Uni o Gene al P ac ice, Helsinki, Finland, 47 Ge ia ic Rehabili a ion Uni , Azienda Sani a ia Fi enze, Flo ence, I aly, 48 New Yo k Academy o Medicine, New Yo k, NY, Uni ed S a es o Ame ica, 49 P incess Al-Jawha a Al-B ahim Cen e o Excellence in Resea ch o He edi a y Diso de , King Abdulaziz Uni e si y, Jeddah, Saudi A abia, 50 Depa men o Clinical Physiology and Nuclea Medicine, Tu ku Uni e si y Hospi al, Tu ku, Finland *da iush.moza a ian@ u s.edu Abs ac Backg ound Regula ish and omega-3 consump ion may ha e se e al heal h bene i s and a e ecom- mended by majo die a y guidelines. Ye , hei in akes emain ema kably a iable bo h wi hin and ac oss popula ions, which could pa ly owe o gene ic in luences. Objec i e To iden i y common gene ic a ian s ha in luence ish and die a y eicosapen aenoic acid plus docosahexaenoic acid (EPA+DHA) consump ion. Design We conduc ed genome-wide associa ion (GWA) me a-analysis o ish (n= 86,467) and EPA +DHA (n= 62,265) consump ion in 17 coho s o Eu opean descen om he CHARGE (Coho s o Hea and Aging Resea ch in Genomic Epidemiology) Conso ium Nu i ion Wo king G oup. Resul s om coho -speci ic GWA analyses (addi i e model) o ish and EPA+DHA consump ion we e adjus ed o age, sex, ene gy in ake, and popula ion s a i ica- ion, and me a-analyzed sepa a ely using ixed-e ec me a-analysis wi h in e se a iance weigh s (METAL so wa e). Addi ionally, he i abili y was es ima ed in 2 coho s. Resul s He i abili y es ima es o ish and EPA+DHA consump ion anged om 0.13–0.24 and 0.12– 0.22, espec i ely. A signi ican GWA o ish in ake was obse ed o s9502823 on ch omo- some 6: each copy o he mino allele (F eq A = 0.015) was associa ed wi h 0.029 se ings/ day (~1 se ing/mon h) lowe ish consump ion (P = 1.96x10 -8 ). No signi ican associa ion was obse ed o EPA+DHA, al hough s7206790 in he obesi y-associa ed FTO gene was among op hi s (P= 8.18x10 -7 ). Pos -hoc calcula ions demons a ed 95% s a is ical powe o de ec a gene ic a ian associa ed wi h e ec size o 0.05% o ish and 0.08% o EPA +DHA. Genome-wide associa ion o die a y ish and EPA+DHA consump ion PLOS ONE | h ps://doi.o g/10.1371/jou nal.pone.0186456 Decembe 13, 2017 2 / 15 U01HG004402; and Na ional Ins i u es o Heal h con ac HHSN268200625226C. The au ho s hank he s a and pa icipan s o he ARIC s udy o hei impo an con ibu ions. In as uc u e was pa ly suppo ed by G an Numbe UL1RR025005, a componen o he Na ional Ins i u es o Heal h and NIH Roadmap o Medical Resea ch. D . Ne le on was suppo ed by a K01 om he Na ional Ins i u es o Heal h, Na ional Ins i u e o Diabe es and Diges i e and Kidney Diseases (5K01DK082729-02). The Ca dio ascula Heal h S udy (CHS) esea ch epo ed in his a icle was suppo ed by con ac s HHSN268201200036C, HHSN268200800007C, N01HC55222, N01HC85079, N01HC85080, N01HC85081, N01HC85082, N01HC85083, N01HC85086, and g an U01HL080295 om he Na ional Hea , Lung, and Blood Ins i u e, wi h addi ional con ibu ion om he Na ional Ins i u e o Neu ological Diso de s and S oke. Addi ional suppo was p o ided by R01AG023629 om he Na ional Ins i u e on Aging. A ull lis o p incipal CHS in es iga o s and ins i u ions can be ound a CHS-NHLBI.o g. DNA handling and geno yping was suppo ed in pa by Na ional Cen e o Resea ch Resou ces g an M01RR00069 o he Ceda s-Sinai Gene al Clinical Resea ch Cen e Geno yping co e and Na ional Ins i u e o Diabe es and Diges i e and Kidney Diseases g an DK063491 o he Sou he n Cali o nia Diabe es Endoc inology Resea ch Cen e . D . Moza a ian was suppo ed by R01 HL085710 om he Na ional Hea , Lung, and Blood Ins i u e. The DILGOM s udy has been unded by he Academy o Finland (g an numbe s 139635, 129494, 118065, 129322, 250207, 136895, 141005), he O ion- Fa mos Resea ch Founda ion, he Finnish Founda ion o Ca dio ascula Resea ch, and he Sig id Juse ´lius Founda ion. We hank he many colleagues who con ibu ed o collec ion and pheno ypic cha ac e iza ion o he clinical samples, and DNA ex ac ion and geno yping o he da a, especially Eija Ha¨ma¨la¨inen, Min u Jussila, Ou i To¨ nwall, Pa¨i i Laiho, and he s a om he Geno yping Facili ies a he Wellcome T us Sange Ins i u e. We would also like o acknowledge hose who ag eed o pa icipa e in he DILGOM s udy. EGCUT ecei ed inancing by FP7 g an s (278913, 306031, 313010), Cen e o Excellence in Genomics (EXCEGEN) and Uni e si y o Ta u (SP1GVARENG). We acknowledge EGCUT echnical pe sonnel, especially M V. Soo and S. Smi . Da a analyzes we e ca ied ou in pa in he High Pe o mance Compu ing Cen e o Uni e si y o Ta u. The Family Hea S udy (FamHS) wo k was suppo ed in pa by NIH g an s 5R01 HL08770003, 5R01 HL08821502 (Michael A. Conclusions These no el indings sugges ha non-gene ic pe sonal and en i onmen al ac o s a e p in- cipal de e minan s o he ema kable a ia ion in ish consump ion, ep esen ing modi iable a ge s o inc easing in akes among all indi iduals. Genes unde lying he signal a s72838923 and mechanisms o he associa ion wa an u he in es iga ion. In oduc ion Consump ion o ish (including in ish and shell ish) and long-chain omega-3 a y acids is linked o lowe isk o se e al ch onic diseases, in pa icula a al co ona y hea disease [1]. These bene icial associa ions in obse a ional s udies a e suppo ed by andomized con olled ials demons a ing a o able e ec s o ish o ish oil on nume ous ch onic disease isk ac- o s and on ca diac mo ali y [1,2]. As a esul , egula ish consump ion is ecommended by all majo na ional and in e na ional die a y guidelines [1]. In con as o hese guidelines and in compa ison o many o he oods, ema kable a ia- ion exis s in he amoun o ish consump ion wi hin and ac oss popula ions. In many Wes e n na ions, app oxima ely one- hi d o indi iduals consume no ish a all, app oxima ely one- hi d consume ish bu ela i ely a ely (up o once pe week), and app oxima ely one- hi d consume ish mo e equen ly [3]. While some o his wide a ia ion in ish consump ion is undoub edly due o pe sonal and en i onmen al ac o s (e.g., cul u e, geog aphic esidence, amily habi s, socioeconomic s a us), he po en ial con ibu ion o in insic biologic ac o s, such as gene ic a ia ion, is no well es ablished. In one analysis among Danish wins, he es i- ma ed he i abili y o ish consump ion was 17% in men and 61% in women, based on addi i e gene ic e ec s [4]. Fo example, po en ial he i abili y could ela e o di e ences in genes ela ed o as e, diges ion, a y acid me abolism, o o he unknown p ocesses ela ed o ood p e e ences. Ye , he po en ial gene ic a ian s unde lying his es ima ed he i abili y a e unknown; and such he i abili y es ima es also equi e u he eplica ion. A basic concep unde lying “pe sonalized nu i ion” is ha a pe son’s genes can in luence hei beha io s and esponses o he en i onmen . Die a y habi s, including he consump ion o ish, a e among he mos ele an ac o s ha in luence he de elopmen o ch onic diseases. Elucida ing whe he , and in wha manne , speci ic genes al e ish and long-chain omega-3 a y acid consump ion would ha e implica ions o unde s anding in luences on a ia ion in ish in ake wi hin popula ions and he biology o pa iali y o oods. Fu he mo e, iden i ica- ion o such a ian s could also in o m he de elopmen o pe sonalized nu i ion—die a y ecommenda ions based on gene ic p e e ences o consump ion. As has been seen wi h o he cha ac e is ics such as physiologic isk ac o s, genome-wide associa ion (GWA) s udies may lead o disco e y o no el genes and biologic pa hways ha in luence he indi idual cha ac e is ic o in e es . Al hough such s udies ha e been pe o med o majo mac onu ien s (e.g., a , ca bohyd a e, p o ein) [5,6], ew analyses ha e been done o speci ic oods[7], whose in akes may be in luenced by complex cha ac e is ics o as es, ex- u es, a omas, and nu ien con en s. The abili y o unde ake ood-speci ic gene ic analyses has been limi ed by he modes sample sizes o indi idual coho s ha ing bo h die a y and gene ic in o ma ion and he po en ial lack o ep oducibili y o gene ic indings disco e ed in any single coho . We he e o e pe o med a collabo a i e in es iga ion o es ima e he i abili y o and assess how common gene ic a ia ion ela es o die a y consump ion o bo h ish and long-chain omega-3 a y acids (eicosapen aenoic acid (EPA) and docosahexaenoic acid (DHA)) as pa o Genome-wide associa ion o die a y ish and EPA+DHA consump ion PLOS ONE | h ps://doi.o g/10.1371/jou nal.pone.0186456 Decembe 13, 2017 3 / 15 P o ince) om NHLBI, and 5R01 DK07568102, 5R01 DK06833603 om NIDDK (Ing id B. Bo ecki), and by he Na ional Hea , Lung, and Blood Ins i u e coope a i e ag eemen g an s U01 HL 67893, U01 HL67894, U01 HL67895, U01 HL67896, U01 HL67897, U01 HL67898, U01 HL67899, U01 HL67900, U01 HL67901, U01 HL67902, U01 HL56563, U01 HL56564, U01 HL56565, U01 HL56566, U01 HL56567, U01 HL56568, and U01 HL56569. The in es iga o s hank he s a and pa icipan s o he FHS o hei impo an con ibu ions. The F amingham O sp ing S udy and F amingham Thi d Gene a ion S udy (FHS) we e conduc ed in pa using da a and esou ces om he F amingham Hea S udy o he Na ional Hea Lung and Blood Ins i u e o he Na ional Ins i u es o Heal h and Bos on Uni e si y School o Medicine. The analyses e lec in ellec ual inpu and esou ce de elopmen om he F amingham Hea S udy in es iga o s pa icipa ing in he SNP Heal h Associa ion Resou ce (SHARe) p ojec . This wo k was pa ially suppo ed by he Na ional Hea , Lung and Blood Ins i u e’s F amingham Hea S udy (Con ac No. N01-HC-25195) and i s con ac wi h A yme ix, Inc. o geno yping se ices (Con ac No. N02-HL- 6-4278). A po ion o his esea ch u ilized he Linux Clus e o Gene ic Analysis (LinGA-II) unded by he Robe Dawson E ans Endowmen o he Depa men o Medicine a Bos on Uni e si y School o Medicine and Bos on Medical Cen e . Also suppo ed by Na ional Ins i u e o Diabe es and Diges i e and Kidney Diseases (NIDDK) R01 DK078616 o D s. Meigs, Dupuis and Flo ez, NIDDK K24 DK080140 o D . Meigs, and a Massachuse s Gene al Hospi al Physician Scien is De elopmen Awa d and a Do is Duke Cha i able Founda ion Clinical Scien is De elopmen Awa d o D . Flo ez. D . Hi e was suppo ed by he Cen e de Reche che Medicale de l’Uni e si e de She b ooke (CRMUS) and a Canadian Ins i u e o Heal h Resea ch (CHIR) Fellowships Heal h P o essional Awa d. D . Nicola McKeown is suppo ed by he USDA ag eemen No. 58-1950-7-707. We hank all s udy pa icipan s as well as e e ybody in ol ed in he Helsinki Bi h Coho S udy. Helsinki Bi h Coho S udy has been suppo ed by g an s om he Academy o Finland, he Finnish Diabe es Resea ch Socie y, Folkha¨lsan Resea ch Founda ion, No o No disk Founda ion, Finska La¨ka esa¨llskape , Signe and Ane Gyllenbe g Founda ion, Uni e si y o Helsinki, Eu opean Science Founda ion (EUROSTRESS), Minis y o Educa ion, Ahokas Founda ion, Emil Aal onen Founda ion. The Heal h, Aging and Body Composi ion (Heal h ABC) s udy was suppo ed in pa by he In amu al Resea ch P og am o he he o he CHARGE (Coho s o Hea and Aging Resea ch in Genomic Epidemiology) Con- so ium Nu i ion Wo king G oup, b inging oge he in es iga o s and da a om 17 US and Eu opean popula ion-based coho s udies o aling 86,467 pa icipan s o Eu opean descen . Subjec s and me hods Coho s The p esen wo k was a collabo a ion among 17 US and Eu opean popula ion-based coho s udies pa icipa ing in he Nu i ion Wo king G oup o he CHARGE Conso ium (S1 Table). These included he A he oscle osis Risk in Communi ies S udy (ARIC); Ca dio ascu- la Heal h S udy (CHS); Die a y, Li es yle, and Gene ic De e minan s o Obesi y and Me a- bolic Synd ome (DILGOM); Es onian S udy; Family Hea S udy (FamHS); F amingham Hea S udy (FHS); Helsinki Bi h Coho S udy (HBCS); Heal h 2000 su ey (H2000); Heal h, Aging, and Body Composi ion (Heal hABC) S udy; Heal h P o essionals Follow-up S udy (HPFS); In ecchia e [Aging] in Chian i A ea (InCHIANTI); Mul i-E hnic S udy o A he o- scle osis (MESA); Nu ses’ Heal h S udy (NHS); Ro e dam S udy; The Hellenic S udy o In e - ac ions be ween SNPs and Ea ing in A he oscle osis Suscep ibili y (THESIAS); Women’s Genome and Heal h s udy (WGHS); and Young Finns S udy (YFS). Addi ional de ails on hese coho s ha e been published p e iously [5,6] and a e p o ided in S1 Table. All pe sons s udied we e o Eu opean descen , consen ed o gene ic esea ch, and p o ided w i en in o med consen . Fo each s udy, examina ion p o ocols we e app o ed by local ins i u ional e iew boa ds a Johns Hopkins Uni e si y (ARIC, MESA), Uni e si y o Washing on (CHS), Epidemiology and Public Heal h o he Hospi al Dis ic o Helsinki and Uusimaa (DILGOM), Washing on Uni e si y (FamHS), Bos on Uni e si y (FHS), Uni e si y o Pi sbu gh (Heal hABC), Ha a d Uni e si y (HPFS, NHS), Na ional Public Heal h Ins i u e o Finland (HBCS), Epidemiology and Public Heal h o he Hospi al Dis ic o Helsinki and Uusimaa (H2000), I alian Na ional Ins i u e o Resea ch and Ca e o Aging (InCHIANTI), E asmus Medical Cen e and The Ne he lands Minis y o Heal h, Wel a e and Spo s (Ro e dam), Ha okopio Uni e si y (THESIAS), B igham and Women’s Hospi al (WGHS), Uni e si y o Helsinki (YFS), and p ocedu es we e in acco dance wi h he e hical s anda ds o he esponsi- ble ins i u ional o egional commi ee on human subjec esea ch. Assessmen o ish and omega-3 a y acid consump ion Usual die a y in ake was assessed in each coho using de ailed ood equency ques ionnai es designed o cap u e he die a y habi s o he popula ion unde s udy (S2 Table). Typically, pa - icipan s we e asked o indica e how o en, on a e age, hey had consumed a ious oods and be e ages o e he pas yea acco ding o mul iple equency ca ego ies (e.g., 9 ca ego ies ang- ing om <1/mon h o 6+/day), wi h usual po ion sizes speci ied on he ques ionnai e o by he pa icipan . Fish in ake was gene ally assessed using mul iple ques ions, such as on con- sump ion o una ish; da k mea ish such as salmon o sa dines; o he whi e ish; shell ish; and ied ish o ish sandwiches. Fo each ques ion, he midpoin o each equency ca ego y was used o es ima e usual in ake which was hen mul iplied by he speci ied po ion size; hese in akes we e summed ac oss all ques ions on ish. Fo his analysis, we s anda dized ish consump ion in each coho o 100g se ings/day. In 12 coho s, o al die a y consump ion o eicosapen aenoic acid (EPA) and docosahexaenoic acid (DHA) was es ima ed by linking he die a y assessmen ool o a ood composi ion able speci ic o he coho (e.g., he USDA ood composi ion da abase in he US). Fo each o he ypes o oods consumed, he equency and a e age po ion size we e mul ipled by he con en o EPA/DHA in he ood. The o al was cal- cula ed by summing ac oss all oods in he ques ionnai e. Fo coho s ha included nu ien s Genome-wide associa ion o die a y ish and EPA+DHA consump ion PLOS ONE | h ps://doi.o g/10.1371/jou nal.pone.0186456 Decembe 13, 2017 4 / 15 NIH, Na ional Ins i u e on Aging con ac s N01AG62101, N01AG62103, and N01AG62106. The genome-wide associa ion s udy was unded by NIA g an R01 AG032098 o Wake Fo es Uni e si y Heal h Sciences and geno yping se ices we e p o ided by he Cen e o Inhe i ed Disease Resea ch (CIDR). CIDR is ully unded h ough a ede al con ac om he Na ional Ins i u es o Heal h o The Johns Hopkins Uni e si y, con ac numbe HHSN268200782096C. The use o Heal h 2000 da a in his s udy has been inancially suppo ed by he Academy o Finland (g an 250207) and O ion- Fa mos Resea ch Founda ion. The au ho s would like o hank he many colleagues who con ibu ed o collec ion and pheno ypic cha ac e iza ion o he clinical samples, and DNA ex ac ion and geno yping o he da a, especially Eija Ha¨ma¨la¨inen, Min u Jussila, Ou i To¨ nwall, Pa¨i i Laiho, and he s a om he Geno yping Facili ies a he Wellcome T us Sange Ins i u e. They would also like o acknowledge hose who ag eed o pa icipa e in he H2000 s udy. In ecchia e in Chian i (aging in he Chian i a ea, InCHIANTI) s udy in es iga o s hank he In amu al Resea ch P og am o he NIH, Na ional Ins i u e on Aging who a e esponsible o he InCHIANTI samples. In es iga o s also hank he InCHIANTI pa icipan s. The InCHIANTI s udy baseline (1998- 2000) was suppo ed as a “ a ge ed p ojec ” (ICS110.1/RF97.71) by he I alian Minis y o Heal h and in pa by he U.S. Na ional Ins i u e on Aging (Con ac s: 263 MD 9164 and 263 MD 821336). The Mul i-E hnic S udy o A he oscle osis (MESA) and MESA SHARe p ojec a e conduc ed and suppo ed by con ac s N01-HC-95159 h ough N01-HC-95169 and RR-024156 om he Na ional Hea , Lung, and Blood Ins i u e (NHLBI). Funding o MESA SHARe geno yping was p o ided by NHLBI Con ac N02-HL-6-4278. The au ho s hank he pa icipan s o he MESA s udy, he Coo dina ing Cen e , MESA in es iga o s, and s udy s a o hei aluable con ibu ions. A ull lis o pa icipa ing MESA in es iga o s and ins i u ions can be ound a h p://www.mesa-nhlbi.o g. The NHS and HPFS a e suppo ed by he Na ional Cance Ins i u e (NHS: UM1 CA186107, HPFS:UM1 CA167552) wi h addi ional suppo o geno yping. The NHS B eas Cance GW scan was pe o med as pa o he Cance Gene ic Ma ke s o Suscep ibili y ini ia i e o he NCI (R01CA40356, U01-CA98233). The NHS/HPFS ype 2 diabe es GWAS (U01HG004399) is a componen o a collabo a i e p ojec ha includes 13 o he GWAS unded as pa o he Gene En i onmen - Associa ion S udies (GENEVA) unde he NIH Genes, En i onmen and Heal h Ini ia i e (GEI) om supplemen s, he po ion o EPA+DHA om supplemen s was excluded om ou analysis. He i abili y es ima es To e alua e po en ial he i abili y o ish and EPA+DHA consump ion, we es ima ed he i abil- i y using amily-based me hods in wo amily-based coho s (FamHS and FHS) using he a i- ance componen s me hod in Sequen ial Oligogenic Linkage Analysis Rou ines (SOLAR; Texas Biomedical Resea ch Ins i u e; San An onio, TX), and adjus ing o age and sex. B ie ly, he i a- bili y is calcula ed using a maximum likelihood me hod using he a io o he gene ic a iance o o al pheno ypic a iance.[8] Geno yping and analysis Genome-wide geno yping was conduc ed in each coho using A yme ix o Illumina pla - o ms. Each s udy pe o med quali y con ol o geno yped single nucleo ide polymo phisms (SNPs) based on mino allele equency (MAF), call a e, and depa u e om Ha dy-Weinbe g equilib ium (S3 Table). Phased haplo ypes om HapMap CEU we e used o impu e ~2.5 mil- lion au osomal SNPs using a Hidden Ma ko Model algo i hm implemen ed in MACH, IMPUTE, o BimBam. S udy-speci ic GWA analyses we e conduc ed wi hin each coho using geno yped and impu ed SNP dosages assuming an addi i e gene ic model. Fish and EPA +DHA consump ion we e sepa a ely e alua ed as he dependen a iable using linea eg es- sion wi h obus s anda d e o , adjus ed o age, sex, ene gy in ake (kcal/d), s udy-speci ic cen e s whe e applicable, and popula ion s a i ica ion p incipal componen s when he coho lambda was >1. SNPs wi h low MAF (<1%), low impu a ion quali y (MACH: R 2 <0.3; o IMPUTE: p ope in o <0.4), we e excluded. Quali y con ol o coho -le el GWAS esul s was pe o med o ensu e co ec speci ica ion o he mino allele and ag eemen in equencies wi h he e e ence popula ion (HapMap CEU), consis en dis ibu ion o e ec sizes and s an- da d e o , and examina ion o QQ plo s o assess any la ge in la ion o es s a is ics. Resul s ac oss s udies we e combined using ixed-e ec me a-analysis wi h in e se a iance weigh s (METAL so wa e)[9]. The associa ion esul s om indi idual s udies as well as me a-analyses we e adjus ed o genomic con ol. To explo e po en ial he e ogenei y by demog aphic egion, a me a-analysis wi hin coho s om Eu ope and USA was pe o med. Genome-wide signi i- cance was conside ed a he Bon e oni-co ec ed h eshold o P<5x10 -8 . S a is ical powe o de ec a ue associa ion a a ious e ec sizes (he i abili y) was calcula ed using GWAPowe so wa e o he analysis o 1 million independen SNPs o bo h ish and EPA+DHA (Feng S, 2011 BMC Gene ics), assuming linkage disequilib ium ( 2 ) o 0.5 be ween a SNP and pu a i e causal a ian and 10% a iance explained by h ee co a ia es. Explo a o y analysis o plasma phospholipid EPA and DHA Resul om genome-wide associa ion analyses o ci cula ing EPA and DHA a e publically a ailable (h p:// acul y.washing on.edu/ ozenl/ iles/) [10]. These da abases we e mined o es whe he he op SNPs om he ish and EPA+DHA in ake GWAS a e associa ed wi h ci cula - ing le els o EPA and DHA. Resul s The 17 coho s we e om he US, Es onia, Finland, G eece, I aly, and he Ne he lands and included 86,467 pa icipan s wi h in o ma ion on ish consump ion and 62,265 wi h in o ma ion on EPA+DHA consump ion. Ac oss pa icipa ing coho s, mean ish consump ion anged om Genome-wide associa ion o die a y ish and EPA+DHA consump ion PLOS ONE | h ps://doi.o g/10.1371/jou nal.pone.0186456 Decembe 13, 2017 5 / 15 (U01HG004738, U01HG004422, U01HG004402, U01HG004729, U01HG004726, 01HG004735, U01HG004415, U01HG004436, U01HG004423, U01HG004728, AHG006033) wi h addi ional suppo om indi idual NIH Ins i u es (NIDCR: U01DE018993, U01DE018903; NIAAA: U10AA008401, NIDA: P01CA089392, 01DA013423; NCI: CA63464, CA54281, CA136792, Z01CP010200). Assis ance wi h pheno ype ha moniza ion and geno ype cleaning, as well as wi h gene al s udy coo dina ion, was p o ided by he GENEVA Coo dina ing Cen e (U01HG004446).Geno yping was pe o med a he B oad Ins i u e o MIT and Ha a d, wi h unding suppo om he NIH GEI (U01HG04424), and Johns Hopkins Uni e si y Cen e o Inhe i ed Disease Resea ch, wi h suppo om he NIH GEI (U01HG004438) and he NIH con ac "High h oughpu geno yping o s udying he gene ic con ibu ions o human disease”(HHSN268200782096C). The NHS/HPFS CHD GWAS was suppo ed by Me ck/Rose a Resea ch Labo a o ies, No h Wales, PA. The NHS/ HPFS Kidney GWAS was suppo ed by NIDDK: 5P01DK070756. The gene a ion and managemen o GWAS geno ype da a o he Ro e dam S udy is suppo ed by he Ne he lands O ganiza ion o Scien i ic Resea ch NWO In es men s (n . 175.010.2005.011, 911-03-012), he Resea ch Ins i u e o Diseases in he Elde ly (014-93-015; RIDE2),EUROSPAN (Eu opean Special Popula ions Resea ch Ne wo k;LSHG-CT-2006-01947), he Ne he lands O ganiza ion o Scien i ic Resea ch (Pionie , 047.016.009, 047.017.043;050-060-810), E asmus Medical Cen e and he Cen e o Medical Sys ems Biology (CMSB I and II and G and; Na ional Genomics Ini ia i e) o he Ne he lands Genomics Ini ia i e (NGI); The Ro e dam S udy is u he unded by E asmus Medical Cen e and E asmus Uni e si y, Ro e dam, Ne he lands O ganiza ion o he Heal h Resea ch and De elopmen (ZonMw), he Resea ch Ins i u e o Diseases in he Elde ly (RIDE), he Minis y o Educa ion, Cul u e and Science, he Minis y o Heal h, Wel a e and Spo s, he Eu opean Commission (DG XII), and he Municipali y o Ro e dam. We hank Pascal A p, Mila Jhamai, D Michael Moo house, Ma ijn Ve ke k, and Sande Be oe s o hei help in c ea ing he GWAS da abase. The au ho s a e g a e ul o he s udy pa icipan s, he s a om he Ro e dam S udy and he pa icipa ing gene al p ac i ione s and pha macis s. The Hellenic s udy o In e ac ions be ween SNPs and Ea ing in A he oscle osis Suscep ibili y (THISEAS) s udy hanks he Geno yping Facili y a he Wellcome T us Sange Ins i u e o yping he THISEAS samples and in 0.19 se ings/day (FHS) o 0.75 se ings/day (THISEAS) (Table 1). Mean in ake o EPA+DHA consump ion anged om 89 (Ro e dam) o 563 (HBCS) mg/d and was gene ally consis en wi h indings on ish in ake, excep in THISEAS (G eece) which had ela i ely highe in akes o ish han EPA+DHA, sugges ing p edominan consump ion o whi e (non-oily) ish. In gene al, pa icipan s in Eu opean coho s had highe ish consump ion han hose in US coho s. The he i abili y es ima es o ish in ake we e 0.13±0.03 (FamHS) and 0.24±0.02 (FHS); and o EPA+DHA in ake, 0.12±0.03 (FamHS) and 0.22±0.02 (FHS). In GWA me a-analyses o ish (17 coho s) and EPA+DHA (11 coho s) consump ion, he genomic con ol lambda alues we e 1.07 and 0.99 espec i ely (S1 and S2 Figs). A genome-wide signi ican associa ion was obse ed o ish in ake on ch omosome 6 o s9502823 (Table 2). The mino allele (F eq A = 0.015) was associa ed wi h 0.029 se ings/day lowe ish consump ion (P = 1.96x10 -8 ). This SNP was mapped o LOC285768 gene o unknown unc ion (Fig 1, op panel); and was no iden i ied in NHGRI-EBI GWAS ca alogue (h p://www.ebi.ac.uk/gwas/, sea ch on Feb 23, 2017). The sec- ond op hi was s17396472 on ch omosome 3, no achie ing genome-wide s a is ical signi i- cance (P = 5.62x10 -8 ). No genome-wide signi ican associa ion was obse ed o EPA+DHA consump ion (S1 and S2 Figs). The op associa ion o EPA+DHA consump ion was obse ed o s11877506 (P= 1.18x10 -7 ) (Table 2). Addi ionally, s7206790 in he obesi y-associa ed FTO gene was among he op SNPs o EPA+DHA in ake: he body mass index- aising G allele was associ- a ed wi h 7mg/day g ea e EPA+DHA in ake (Fig 1,bo om panel; P = 7.44x10 -7 ). To ob ain mo e in o ma ion on he locus associa ed wi h ish consump ion on ch omo- some 6, we in es iga ed da a om he ENCODE p ojec . Using CEU 1000genomes da a, we calcula ed he LD wi hin he egion 250kb ups eam and downs eam om s9502823. Map- ping he SNPs ound in he s9502823 LD block o ENCODE egula o y egions, we iden i ied s72838923 (in comple e LD wi h s9502823) as a unc ional candida e. s72838923 alls wi hin a expe imen ally de e mined H3k27Ac egion, iden i ied in se e al cell ypes in ENCODE. H3k27Ac egions a e hough o be ma ke s o ac i e enhance ac i i y. In addi ion, s72838923 alls wi hin DNAse Hype sensi i i y Peak which we e iden i ied expe imen ally ac oss 65 cell ypes om he ENCODE p ojec . The e is u he e idence o ansc ip ion ac- o binding si es o FOXA1, among o he s in his egion, om ENCODE CHIP-Seq expe i- men s. In addi ion, mapping s72838923 on he UCSC genome b owse sugges ed ha his SNP is ound wi hin a egion o conse a ion ac oss mammals. Due o he a ying anges o a e age ish consump ion in US e sus Eu opean s udies, we pe o med explo a o y subg oup GWA me a-analysis s a i ied by geog aphic loca ion. No sig- ni ican associa ions we e iden i ied in USA no Eu opean s udies (S3 Fig) o ish o EPA +DHA consump ion (S4 Fig). A p io conso ium analysis including se e al o hese same coho s epo ed on genome- wide associa ion o SNP a ian s wi h plasma phospholipid EPA and DHA, he concen a ions o which a e de e mined by bo h die a y in ake and endogenous me abolism egula ion.[11] In explo a o y analysis, we e alua ed whe he he op 5 hi s o ish consump ion and he op 4 hi s o es ima ed die a y EPA+DHA consump ion iden i ied in he p esen analysis we e asso- cia ed wi h plasma phospholipid concen a ions o EPA o DHA in ha p io analysis [10], adjus ing o mul iple compa isons (9 SNPs x 2 a y acids = Bon e oni-co ec ed alpha o 0.05/18 = 0.0028). No signi ican associa ions we e iden i ied (S4 Table). Discussion In his la ge GWA me a-analysis o 17 US and Eu opean coho s o aling 86,467 pa icipan s, we ound e idence ha common gene ic a ia ion may be associa ed wi h consump ion o Genome-wide associa ion o die a y ish and EPA+DHA consump ion PLOS ONE | h ps://doi.o g/10.1371/jou nal.pone.0186456 Decembe 13, 2017 6 / 15 pa icula Sa ah Edkins and Co delia Lang o d. PD is suppo ed by he Wellcome T us . The WGHS is suppo ed by HL043851 and HL080467 om he Na ional Hea , Lung, and Blood Ins i u e and CA047988 om he Na ional Cance Ins i u e, he Donald W. Reynolds Founda ion and he Fonda ion Leducq, wi h collabo a i e scien i ic suppo and unding o geno yping p o ided by Amgen. The Young Finns S udy has been inancially suppo ed by he Academy o Finland: g an s 126925, 121584, 124282, 129378 (Sal e), 117787 (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 (g an 9M048 o TeLeh ), Juho Vainio Founda ion, Paa o Nu mi Founda ion, Finnish Founda ion o Ca dio ascula Resea ch and Finnish Cul u al Founda ion, Tampe e Tube culosis Founda ion and Emil Aal onen Founda ion (T.L). The expe echnical assis ance in he s a is ical analyses by I ina Lisinen, Ville Aal o and Mika Helminen a e g a e ully acknowledged. The con en is solely he esponsibili y o he au ho s and does no necessa ily ep esen he o icial iews o he Na ional Ins i u es o Heal h o he o he unde s. Compe ing in e es s: Luc Djousse epo s ecei ing in es iga o -ini ia ed g an s (omega-3 a y acid s udies) om NIH and Ama in Pha ma, Inc. Cu en ly se ing as ad hoc consul an o Ama in Pha ma, Inc. B uce Psa y epo s se ing on he DSMB o a clinical ial o a de ice unded by he manu ac u e (Zoll Li eCo ) and on he S ee ing Commi ee o he Yale Open Da a Access P ojec unded by Johnson & Johnson." Paul Ridke has ecei ed esea ch g an unds om As aZeneca, a manu ac u e o a p esc ip ion ish oil p oduc . Osca F anco wo ks in E asmusAGE, a cen e o aging esea ch ac oss he li e cou se unded by Nes le ´Nu i ion (Nes ec L d.); Me agenics Inc.; and AXA. Nes le ´Nu i ion (Nes ec L d.); Me agenics Inc.; and AXA had no ole in design and conduc o he s udy; collec ion, managemen , analysis, and in e p e a ion o he da a; and p epa a ion, e iew o app o al o he manusc ip . D . Moza a ian epo s epo s ad hoc hono a ia o consul ing om Bunge, Haas A ocado Boa d, Nu i ion Impac , Ama in, As a Zeneca, Bos on Hea Diagnos ics, GOED, and Li e Sciences Resea ch O ganiza ion; and scien i ic ad iso y boa ds, Unile e No h Ame ica and Elysium Heal h.Ha a d Uni e si y holds a pa en , lis ing D . Moza a ian as one o h ee co-in en o s, o use o ans-palmi oleic acid o p e en and ea insulin esis ance, ype 2 diabe es, and ela ed condi ions. All o he au ho s epo no con lic s o in e es . This does no al e ou adhe ence o PLOS ONE policies on sha ing da a and ma e ials. ish. We ound no genome-wide signi ican associa ion o common a ian s wi h EPA+DHA in ake. While he sample size o he analysis o EPA+DHA was smalle han he ish in ake analysis, wi h 62,265 indi iduals he analysis had 95% powe o de ec an e ec size (he i abil- i y) o 0.08%. We iden i ied one locus on ch omosome 6 in associa ion wi h ish consump ion. The SNP was mapped o LOC285768wi h unknown unc ion. The nex closes gene is o khead box Q1 (FOXQ1) which is a membe o he cance -associa ed o khead-box (FOX) gene amily [12]. Ou e alua ion o da a om ENCODE, aken oge he , iden i ied a unc ional candida e, s72838923, ha appea s o lie wi hin a ansc ip ionally ac i e egion o he genome. While he associa ion was s a is ically signi ican , he magni ude o e ec was small, wi h he mino allele being associa ed wi h a di e ence o 0.03 se ings/day o app oxima ely 1 se ing/mon h o ish. This inding is mo e likely o be ele an o unde s anding he biology o ood p e e - ences han o in luencing clinical ou comes, al hough e en small di e ences in ish consump- ion, o e a li e ime, could in luence heal h. The nonsigni ican op associa ions iden i ied in ch omosomes 1, 3, and 12 each ep esen in agenic egions o genes highly exp essed in he b ain, bu hese associa ions did no achie e genome-wide signi icance. In he i abili y analyses, we ound e idence o modes he i abili y o ish (0.13 o 0.24) and EPA+DHA (0.12 o 0.22). Ou GWA esul s iden i ied one locus in associa ion wi h ish in ake ha canno ully accoun o his obse ed he i abili y, sugges ing ha obse ed he i abili y migh be due o ema kably small e ec s ac oss a la ge numbe o SNPs, o he ypes o gene ic Table 1. Cha ac e is ics o he coho s included in his analysis o he gene ics o ish consump ion. S udy Maximum N Age % Female To al Fish In ake (se /day) Median (5-95 h%) Die a y EPA+DHA (mg/d) Median (5-95 h%) ARIC 9557 54.3 ±5.7 53 0.21 (0.0–0.9) 180 (10–730) CHS 3190 72.3 ±5.4 61 0.29 (0.1–0.8) 191 (27–569) DILGOM_METABO 3467 51.5 ±13.4 55 0.42 (0.1–1.3) 431 (119–1277) DILGOM_GWA 604 52.4 ±13.5 52 0.44 (0.1–1.2) 444 (107–1233) ESTONIAN S udy 9920 48.8 ±20.1 53 0.21 (0.0–0.6) - - FamHS 3640 52.2 ±13.7 53 0.14 (0.0–0.7) 170 (1–680) FHS 7044 47.3 ±11.8 54 0.13 (0.0–0.6) 200 (40–640) Heal h ABC 1494 74.8 ±2.9 48 0.20 (0.2–0.3) - - Heal h 2000 1935 50.5 ±10.9 51 0.39 (0.1–1.1) 505 (102–1477) HBCS 1701 61.5 ±2.9 57 0.44 (0.1–1.3) 563 (145–1895) HPFS 4133 58.6 ±8.7 0 0.29 (0.1–0.9) - - InCHIANTI 1194 68.3 ±15.4 55 0.19 (0.0–0.5) - - MESA 2305 62.7 ±10.2 52 0.17 (0.0–0.7) 100 (20–300) NHS 6776 54.4 ±6.7 100 0.28 (0.1–0.8) - - Ro e dam 4606 67.6 ±7.7 59 0.07 (0.0–0.5) 89 (8–443) THISEAS 395 59.4 ±13.0 41 0.59 (0.0–2.0) 137 (4–479) WGHS 22691 54.7 ±7.1 100 0.20 (0.0–0.7) 150 (30–470) YFS 1815 37.8 ±5.0 56 0.34 (0.1–0.9) 357 (92–902) Abb e ia ions: A he oscle osis Risk in Communi ies S udy (ARIC); Ca dio ascula Heal h S udy (CHS); Die a y, Li es yle, and Gene ic De e minan s o Obesi y and Me abolic Synd ome (DILGOM); Es onian S udy; Family Hea S udy (FamHS); F amingham Hea S udy (FHS); Helsinki Bi h Coho S udy (HBCS); Heal h 2000 su ey (H2000); Heal h, Aging, and Body Composi ion (Heal hABC) S udy; Heal h P o essionals Follow-up S udy (HPFS); In ecchia e [Aging] in Chian i A ea (InCHIANTI); Mul i-E hnic S udy o A he oscle osis (MESA); Nu ses’ Heal h S udy (NHS); Ro e dam S udy; The Hellenic S udy o In e ac ions be ween SNPs and Ea ing in A he oscle osis Suscep ibili y (THESIAS); Women’s Genome and Heal h s udy (WGHS); and Young Finns S udy (YFS). h ps://doi.o g/10.1371/jou nal.pone.0186456. 001 Genome-wide associa ion o die a y ish and EPA+DHA consump ion PLOS ONE | h ps://doi.o g/10.1371/jou nal.pone.0186456 Decembe 13, 2017 7 / 15 Abb e ia ions: EPA, eicosapen aenoic acid; DHA, docosahexaenoic acid; LD, linkage disequilib ium; GWAS, genome-wide associa ion s udy. a ia ion such as copy numbe a ian s, epigene ic modi ica ions, o mul iple unobse ed gene ic in e ac ions wi h unknown en i onmen al ac o s. This challenge o “missing” o unaccoun ed o he i abili y is a equen inding in GWA analyses o common diseases and ai s [13]. He i abili y analyses may o e es ima e he i abili y due o unmeasu ed sha ed en i- onmen al in luences, o example om in u e o/placen al in luences h ough childhood and adul li e. In his ligh , ou he i abili y indings a e lowe han hose p e iously epo ed [4] and ep esen an addi ional impo an new con ibu ion. Ou indings suppo he need o u u e in es iga ions o he possible explana ions o he modes bu as ye missing he i abili y o ish and EPA+DHA consump ion. This in es iga ion had se e al s eng hs. Ou pooling o mul iple la ge, well-es ablished coho s p o ided a e y la ge sample o pa icipan s o in es iga ing ou esea ch ques ions. Ou pos -hoc powe calcula ions demons a e 95% s a is ical powe o de ec a gene ic a ian associa ed wi h an e ec size o 0.05% o ish consump ion and 0.08% o EPA+DHA con- sump ion. We adjus ed o o al epo ed ene gy in ake, which helps o add ess any sys ema ic o e - o unde - epo ing by indi iduals and also eal di e ences in o al ood consumed (i.e., due o di e ences in age, sex, body size, o physical ac i i y), acili a ing e alua ion o die a y composi ion. All he s udies in he me a-analysis used compa able die a y assessmen ools ha we e app op ia e o he popula ion unde s udy, p o iding he highes quali y da a ha can be easonably collec ed ac oss mul iple la ge epidemiological s udies. Limi a ions should be conside ed. While die a y in akes assessed by ood equency ques- ionnai e ep esen a easonably alid me hod o collec da a on usual die a y habi s in la ge popula ions [14], such da a also include measu emen e o , which could limi he abili y o de ec ue associa ions. Howe e , many alida ion s udies ha e demons a ed ha ish and EPA+DHA consump ion a e measu ed easonably well by ood equency ques ionnai es, whe he compa ed wi h mul iple die eco ds o wi h objec i e ci cula ing o issue bioma ke s [15,16,17,18]. Indeed, because bioma ke le els also ep esen impe ec measu es o “ ue” habi ual consump ion wi h unco ela ed e o s compa ed o ques ionnai e es ima es, he ac ual co ela ions o es ima ed ish o EPA+DHA consump ion wi h “ ue” consump ion a e likely much highe , in he ange o 0.8 o mo e. Compa ed wi h a candida e gene app oach, GWA has lowe s a is ical powe o de ec small gene ic e ec s. Ye , candida e gene app oaches o e alua ing ish consump ion would be s ongly limi ed by impe ec knowledge o which genes a ec known sys ems and biologic p ocesses ela ed o ood p e e ences and, e en mo e so, which genes may a ec cu en ly unknown sys ems and biologic in luences on ood p e e ences. In summa y, his la ge pooling p ojec ac oss 17 es ablished coho s iden i ied modes he i- abili y o ish and omega-3 a y acid consump ion and one gene ic locus associa ed wi h ish Table 2. The mos signi ican associa ions om genome-wide associa ion me a-analysis o ish and EPA+DHA consump ion. T ai SNPID Ch Posi ion N E ec / Non-e ec F eq (E ec ) E ec S dE P- alue Fish In ake s9502823 6 1050674 71910 A/G 0.02 -0.029 0.005 1.96E-08 (se ing/day) s17396472 3 68469758 63102 A/T 0.98 0.031 0.005 5.62E-08 s1860343 12 4583970 78153 T/C 0.52 -0.006 0.001 4.30E-07 s1562806 15 35038800 61909 T/C 0.94 0.022 0.004 1.41E-06 s16834168 1 150754770 77898 A/G 0.02 -0.020 0.004 1.58E-06 EPA+DHA s11877506 18 49176846 52909 A/G 0.96 16.313 3.091 1.18E-07 (mg/day) s2456163 19 1427432 36937 T/C 0.04 -15.560 3.087 4.20E-07 s7476409 10 24809042 52909 T/C 0.05 -17.907 3.589 5.50E-07 s7206790 16 52355409 56099 C/G 0.55 -7.000 1.420 7.44E-07 h ps://doi.o g/10.1371/jou nal.pone.0186456. 002 Genome-wide associa ion o die a y ish and EPA+DHA consump ion PLOS ONE | h ps://doi.o g/10.1371/jou nal.pone.0186456 Decembe 13, 2017 8 / 15 Genome-wide associa ion o die a y ish and EPA+DHA consump ion PLOS ONE | h ps://doi.o g/10.1371/jou nal.pone.0186456 Decembe 13, 2017 9 / 15