The effects of height and BMI on prostate cancer incidence and mortality: a Mendelian randomization study in 20,848 cases and 20,214 controls from the PRACTICAL consortium

ORIGINAL PAPER The e ec s o heigh and BMI on p os a e cance incidence and mo ali y: a Mendelian andomiza ion s udy in 20,848 cases and 20,214 con ols om he PRACTICAL conso ium Neil M. Da ies 1,2 •Tom R. Gaun 1,2 •Sa ah J. Lewis 1,2 •Je Holly 3 •Jenny L. Dono an 1 • F eddie C. Hamdy 4 •John P. Kemp 2,5 •Rosalind Eeles 6,7 •Doug Eas on 8 •Zso ia Ko e-Ja ai 6 • Ali Amin Al Olama 8 •Sa a Benlloch 8 •Kenne h Mui 9 •G aham G. Giles 10,11 •F ed ik Wiklund 12 • Hen ik G onbe g 12 •Ch is ophe A. Haiman 13 •Johanna Schleu ke 14,15 •Bø ge G. No des gaa d 16 • Ru h C. T a is 17 •Da id Neal 18,19 •No a Pashayan 20,41 •Kay-Tee Khaw 21 •Jane L. S an o d 22,23 • William J. Blo 24 •S ephen Thibodeau 25 •Ch is iane Maie 26,27 •Adam S. Kibel 28,29 •Ceza y Cybulski 30 • Lisa Cannon-Alb igh 31 •He mann B enne 32,33,34 •Jong Pa k 35 •Radka Kane a 36 •Jyo sna Ba a 37 • Manuel R. Teixei a 38,39 •Ha de Pandha 40 •PRACTICAL conso ium •Ma k La h op 42,43 • Geo ge Da ey Smi h 1,2 •Richa d M. Ma in 1,2,44 Recei ed: 9 Ap il 2015 / Accep ed: 12 Augus 2015 / Published online: 19 Sep embe 2015 ÓThe Au ho (s) 2015. This a icle is published wi h open access a Sp inge link.com Abs ac Backg ound Epidemiological s udies sugges a po en ial ole o obesi y and de e minan s o adul s a u e in p os a e cance isk and mo ali y, bu he ela ionships desc ibed in he li e a u e a e complex. To add ess unce ain y o e he causal na u e o p e ious obse a ional indings, we in es iga ed associa ions o heigh - and adiposi y- ela ed gene ic a ian s wi h p os a e cance isk and mo ali y. Me hods We conduc ed a case–con ol s udy based on 20,848 p os a e cance s and 20,214 con ols o Eu opean ances y om 22 s udies in he PRACTICAL conso ium. We cons uc ed gene ic isk sco es ha summed each man’s numbe o heigh and BMI inc easing alleles ac oss mul iple single nucleo ide polymo phisms obus ly asso- cia ed wi h each pheno ype om published genome-wide associa ion s udies. Resul s The gene ic isk sco es explained 6.31 and 1.46 % o he a iabili y in heigh and BMI, espec i ely. The e was only weak e idence ha gene ic a ian s p e iously associ- a ed wi h inc eased BMI we e associa ed wi h a lowe p os a e cance isk (odds a io pe s anda d de ia ion PRACTICAL conso ium is p o ided in appendix sec ion. Elec onic supplemen a y ma e ial The online e sion o his a icle (doi:10.1007/s10552-015-0654-9) con ains supplemen a y ma e ial, which is a ailable o au ho ized use s. &Richa d M. Ma in [email p o ec ed] Neil M. Da ies [email p o ec ed] 1 School o Social and Communi y Medicine, Uni e si y o B is ol, B is ol, UK 2 MRC In eg a i e Epidemiology Uni , Uni e si y o B is ol, B is ol, UK 3 School o Clinical Sciences, Uni e si y o B is ol, B is ol BS10 5NB, UK 4 Nu ield Depa men o Su ge y, Uni e si y o Ox o d, Ox o d, UK 5 Uni e si y o Queensland Diaman ina Ins i u e, T ansla ional Resea ch Ins i u e, B isbane, QLD, Aus alia 6 The Ins i u e o Cance Resea ch, London SM2 5NG, UK 7 The Royal Ma sden NHS Founda ion T us , London SW3 6JJ, UK 8 S angeways Labo a o y, Cen e o Cance Gene ic Epidemiology, Depa men o Public Heal h and P ima y Ca e, Uni e si y o Camb idge, Wo s Causeway, Camb idge, UK 9 Ins i u e o Popula ion Heal h, Uni e si y o Manches e , Manches e , UK 10 Cance Epidemiology Cen e, The Cance Council Vic o ia, 615 S Kilda Road, Melbou ne, VIC, Aus alia 11 Cen e o Epidemiology and Bios a is ics, Melbou ne School o Popula ion and Global Heal h, The Uni e si y o Melbou ne, Melbou ne, VIC, Aus alia 123 Cance Causes Con ol (2015) 26:1603–1616 DOI 10.1007/s10552-015-0654-9 inc ease in BMI gene ic sco e 0.98; 95 % CI 0.96, 1.00; p=0.07). Gene ic a ian s associa ed wi h inc eased heigh we e no associa ed wi h p os a e cance incidence (OR 0.99; 95 % CI 0.97, 1.01; p=0.23), bu we e associa ed wi h an inc ease (OR 1.13; 95 % CI 1.08, 1.20) in p os a e cance mo ali y among low-g ade disease (phe e ogenei y, low s. high g ade 0.001). Gene ic a ian s associa ed wi h inc eased BMI we e associa ed wi h an inc ease (OR 1.08; 95 % CI 1.03, 1.14) in all-cause mo ali y among men wi h low-g ade disease (phe e ogenei y =0.03). Conclusions We ound li le e idence o a subs an ial e ec o gene ically ele a ed heigh o BMI on p os a e cance isk, sugges ing ha p e iously epo ed obse a- ional associa ions may e lec common en i onmen al de e minan s o heigh o BMI and p os a e cance isk. Gene ically ele a ed heigh and BMI we e associa ed wi h inc eased mo ali y (p os a e cance -speci ic and all-cause, espec i ely) in men wi h low-g ade disease, a po en ially in o ma i e bu no el inding ha equi es eplica ion. Keywo ds Heigh Body mass index P os a e cance  Mendelian andomiza ion Single nucleo ide polymo phisms Ins umen al a iables analysis In oduc ion P os a e cance is he mos common male cance in Eu ope and No h Ame ica, bu he obus iden i ica ion o po en- ially modi iable isk ac o s has p o en elusi e [1]. Epi- demiological s udies sugges a po en ial ole o obesi y [2–5] and de e minan s o adul s a u e [6], bu he ela ionships desc ibed in he li e a u e a e complex [7–9]. In e se associa ions ha e gene ally been obse ed be ween adiposi y and localized p os a e cance , bu associa ions a e la gely posi i e wi h ad anced o high-g ade [2,10] and a al [3] cance and may a y in di ec ion depending on whe he obesi y was obse ed in ea ly o middle o la e adul hood [4]. Adul s a u e is gene ally posi i ely associa ed wi h p os a e cance , al hough associa ions may be s onge o a al [11] o high- compa ed wi h low-g ade disease [6]. The explana ion o hese associa ions is unclea . Obse - a ions ega ding obesi y could be due o con ounding by common causes o bo h obesi y and p os a e cance (e.g., calo ie and die a y a in ake) [12]; he mi ogenic ho mones insulin and insulin-like g ow h ac o -I [13,14]; delayed de ec ion in obese men [8,9]; o a eal biological e ec [15]. Obse ed heigh associa ions could e lec ea ly-li e en i- onmen al (e.g., e al, die a y, social, ho mones, and psy- chological ci cums ances) o sha ed gene ic con ibu ions o s a u e and p os a e cance isk [16–18]. Gene ic epidemiological s udies a e less suscep ible o con ounding han obse a ional epidemiology. This is because condi ional on popula ion s uc u e, gene ic a ian s a e mo e likely o be independen o la e en i onmen and li es yle ac o s [19]; hey a e also unlikely o be a ec ed by e e se causa ion. Thus, he exis ence o gene ic a ia ion in obesi y and heigh can p o ide obus e idence abou how associa ions o pheno ypes, in his case obesi y and heigh , wi h diseases a ise [15]. We p e iously epo ed ha a single nucleo ide polymo phism (SNP) associa ed wi h obesi y (FTO s9939609-A) was in e sely associa ed wi h low- g ade p os a e cance (odds a io, OR 0.90 pe A allele; 95 % CI 0.81, 0.99; p=0.03), bu posi i ely associa ed wi h high- 12 Depa men o Medical Epidemiology and Bios a is ics, Ka olinska Ins i u e, S ockholm, Sweden 13 Depa men o P e en i e Medicine, Keck School o Medicine, Uni e si y o Sou he n Cali o nia/No is Comp ehensi e Cance Cen e , Los Angeles, CA, USA 14 Depa men o Medical Biochemis y and Gene ics, Uni e si y o Tu ku, Tu ku, Finland 15 Ins i u e o Biomedical Technology/BioMediTech, Uni e si y o Tampe e and FimLab Labo a o ies, Tampe e, Finland 16 Depa men o Clinical Biochemis y, He le Hospi al, Copenhagen Uni e si y Hospi al, He le Ring ej 75, 2730 He le , Denma k 17 Cance Epidemiology Uni , Nu ield Depa men o Clinical Medicine, Uni e si y o Ox o d, Ox o d, UK 18 Su gical Oncology (U o-Oncology: S4), Uni e si y o Camb idge, Addenb ooke’s Hospi al, Hills Road, Box 279, Camb idge, UK 19 Li Ka Shing Cen e, Cance Resea ch UK Camb idge Resea ch Ins i u e, Camb idge, UK 20 S angeways Labo a o y, Cen e o Cance Gene ic Epidemiology, Depa men o Oncology, Uni e si y o Camb idge, Wo s Causeway, Camb idge, UK 21 Camb idge Ins i u e o Public Heal h, Uni e si y o Camb idge, Fo ie Si e, Robinson Way, Camb idge CB2 0SR, UK 22 Di ision o Public Heal h Sciences, F ed Hu chinson Cance Resea ch Cen e , Sea le, WA, USA 23 Depa men o Epidemiology, School o Public Heal h, Uni e si y o Washing on, Sea le, WA, USA 24 In e na ional Epidemiology Ins i u e, 1455 Resea ch Bl d., Sui e 550, Rock ille, MD 20850, USA 25 Mayo Clinic, Roches e , MN, USA 26 Depa men o U ology, Uni e si y Hospi al Ulm, Ulm, Ge many 27 Ins i u e o Human Gene ics, Uni e si y Hospi al Ulm, Ulm, Ge many 1604 Cance Causes Con ol (2015) 26:1603–1616 123 g ade cance (OR 1.16; 0.99, 1.37; p=0.07) [15]. These da a sugges ha he compa able obse a ional associa ions be ween adiposi y pheno ypes and p os a e cance ou comes a e no con ounded. Howe e , he e idence o hese e ec s was weak, o igina ing om a single s udy o mode a e size (1,550 cases) using only a single a ian , and he e is no e idence we a e awa e o linking gene ic a ia ion in heigh wi h p os a e cance . The esul s, he e o e, equi e con i - ma ion and ex ension in la ge da ase s, using heigh - and addi ional adiposi y- ela ed gene ic a ian s. Ou aim was o use gene ic a ia ion in heigh and body mass index (BMI) as uncon ounded exposu es o in es i- ga e he causal associa ions o obesi y and s a u e wi h p os a e cance isk and ou comes (Mendelian andomiza- ion [20]). Ins ead o he single- a ian , single-sample app oach used p e iously, we employ a mo e powe ul wo-sample, mul iple- a ian app oach [21,22] ha com- bines se e al polymo phisms (based on con i med gene ic a ian -in e media e pheno ype associa ions [23,24]) in o gene ic isk sco es in o de o explain mo e o he a iance in BMI and heigh exposu es and hus inc ease powe and a oid weak ins umen bias [21]. Me hods Pa icipan s in his s udy we e men o Eu opean geno ypic ances y om 22 independen s udies con ibu ing o he in e na ional PRACTICAL Conso ium (PRos a e cance AssoCia ion g oup To In es iga e Cance -Associa ed aL e a ions in he genome, h p://www.p ac ical.ccge. medschl.cam.ac.uk)[25,26]. The indi idual s udies a e desc ibed a h p://www.na u e.com/ng/jou nal/ 45/n4/ ex e /ng.2560-S1.pd , wi h summa y da a in Table 1.O he s udies wi hin he PRACTICAL Conso ium a he ime o da a ex ac ion, we excluded he EPIC-No olk, CAPS, and SEARCH s udies (in ol ing 3,005 cases and 2,825 con ols), because hey we e included in he genome-wide s udies ha o iginally de ec ed he heigh and BMI gene ic a ian s [23, 24]. Cance s we e ca ego ized as low g ade (Gleason sco e B6) o high g ade (Gleason sco e C7) and localized (T1 o T2 on TNM s aging, o i no a ailable, ‘‘localized’’ on SEER s aging) o ad anced (T3 o T4 on TNM s aging, o i no a ailable, ‘‘ egional’’ o ‘‘dis an ’’ on SEER s aging). All s udies me he app op ia e e hical c i e ia o each coun y in acco dance wi h he p inciples embodied in he Decla a ion o Helsinki. Geno yping Geno yping was ca ied ou using an Illumina Cus om In inium geno yping a ay (iCOGS), designed o he Collabo a i e Oncological Gene-en i onmen S udy (COGS), and consis ed o 211,155 SNPs (de ails a h p:// ec.eu opa.eu/ esea ch/heal h/medical- esea ch/cance / p7- p ojec s/cogs_en.h ml)[25,26]. This a ay was de ised o e alua e gene ic a ian s o associa ions wi h b eas , o a - ian, and p os a e cance ; 68,638 we e speci ically chosen o hei po en ial ele ance o p os a e cance . The emaining 125,877 SNPs measu ed by he a ay we e chosen o 28 B igham and Women’s Hospi al/Dana-Fa be Cance Ins i u e, 45 F ancis S ee -ASB II-3, Bos on, MA 02115, USA 29 Washing on Uni e si y, S . Louis, Missou i 30 In e na ional He edi a y Cance Cen e , Depa men o Gene ics and Pa hology, Pome anian Medical Uni e si y, Szczecin, Poland 31 Di ision o Gene ic Epidemiology, Depa men o Medicine, Uni e si y o U ah School o Medicine, Sal Lake Ci y, UT, USA 32 Di ision o Clinical Epidemiology and Aging Resea ch, Ge man Cance Resea ch Cen e (DKFZ), Heidelbe g, Ge many 33 Di ision o P e en i e Oncology, Ge man Cance Resea ch Cen e (DKFZ), Heidelbe g, Ge many 34 Ge man Cance Conso ium (DKTK), Ge man Cance Resea ch Cen e (DKFZ), Heidelbe g, Ge many 35 Di ision o Cance P e en ion and Con ol, H. Lee Mo i Cance Cen e , 12902 Magnolia D ., Tampa, FL, USA 36 Molecula Medicine Cen e and Depa men o Medical Chemis y and Biochemis y, Medical Uni e si y So ia, 2 Zd a e S , 1431 So ia, Bulga ia 37 Aus alian P os a e Cance Resea ch Cen e-Qld, Ins i u e o Heal h and Biomedical Inno a ion and School o Biomedical Sciences, Queensland Uni e si y o Technology, B isbane, QLD, Aus alia 38 Depa men o Gene ics, Po uguese Oncology Ins i u e, Po o, Po ugal 39 Biomedical Sciences Ins i u e (ICBAS), Po o Uni e si y, Po o, Po ugal 40 The Uni e si y o Su ey, Guild o d, Su ey GU2 7XH, UK 41 Depa men o Applied Heal h Resea ch, Uni e si y College London, 1-19 To ing on Place, London WC1E 7HB, UK 42 Commissa ia a `l’Ene gie A omique, Cen e Na ional de Ge ´no ypage, E y, F ance 43 McGill Uni e si y-Ge ´nome Que ´bec Inno a ion Cen e, Mon eal, Canada 44 B is ol Nu i ion Biomedical Resea ch Uni , Na ional Ins i u e o Heal h Resea ch, B is ol, UK Cance Causes Con ol (2015) 26:1603–1616 1605 123 ele ance o o he cance s and common SNPs which had been p e iously associa ed wi h any ai . Pa icipan s wi h low call a es ( 95 %) and high o low he e ozygosi y (p 1910 -5 ) we e excluded; 201,598 SNPs passed quali y con ol o he Eu opean ances y samples. We used hese geno ypic da a o impu e 2.6 million SNPs based on he HapMap 2 CEU e e ence panel and using IMPUTE2 so wa e [27]. We excluded poo ly impu ed SNPs (R 2 0.3). Cons uc ing gene ic isk sco es o BMI and heigh We cons uc ed gene ic isk sco es [21] o heigh and BMI using 179 and 32 a ian s, espec i ely, p e iously epo ed in genome-wide associa ion s udies (GWAS) o be asso- cia ed wi h heigh [23] and BMI [24]. We used allele dosages om he impu a ion o cons uc he gene ic isk sco e. The dosages code each SNP con inuously om 0 o 2, and he dosages ac oss all SNPs a e summed o es ima e Table 1 Clinical cha ac e is ics o he men in each o he s udies con ibu ing o he PRACTICAL conso ium (n=41,062) S udy Coun y nMean % Con ols Cases Age a diagnosis (yea s) PSA a diagnosis (ng/ml) Sc een de ec ed b (%) Family his o y p os a e cance Gleason sco e 8–10 Ad anced s age (T3 o T4) Dis an sp ead (SEER) CPCS1 Denma k 2,771 848 69.5 48.0 0.0 8.2 35.0 – – CPCS2 Denma k 1,009 265 64.9 36.0 0.0 14.7 10.6 – – EPIC Eu ope a 1,079 722 64.9 19.7 0.0 – 3.6 3.8 0.9 ESTHER Ge many 318 313 65.5 58.7 61.9 8.9 9.1 26.4 3.4 FHCRC USA 730 761 59.7 16.1 – 21.7 10.4 – 2.6 IPO-Po o Po ugal 66 183 59.3 8.3 82.8 20.0 15.8 64.5 0.0 MAYO USA 488 767 65.2 15.5 73.7 29.1 33.0 44.4 0.5 MCCS Aus alia 1,169 1,698 58.5 136.6 – 23.4 11.0 14.0 0.8 MEC USA 829 819 69.5 – – 13.0 36.0 – 2.8 MOFFITT USA 100 412 64.9 7.3 0.0 22.9 11.2 3.5 0.5 PCMUS Bulga ia 140 151 69.3 32.5 21.2 5.3 29.8 42.4 18.5 PPF-UNIS UK 176 244 68.9 32.0 – 25.3 10.9 25.7 9.0 Poland Poland 359 438 67.7 40.2 0.0 10.6 14.0 36.8 2.8 P oMPT UK 1 166 66.3 33.0 0.0 34.6 18.9 32.7 7.8 P o ecT UK 1,474 1,542 62.8 9.6 100.0 7.9 5.7 11.3 0.4 QLD/ P osCan Aus alia 87 186 61.3 6.7 – 36.2 4.0 0.0 0.0 STHMI Sweden 2,224 2,002 66.2 – – 20.2 10.2 14.2 1.6 TAMPERE Finland 2,413 2,753 68.2 69.1 46.8 – 15.4 21.0 7.3 UKGPCS UK 4,182 4,549 63.8 83.9 28.9 23.4 17.2 32.9 10.7 ULM Ge many 354 601 63.8 19.1 – 44.9 15.5 39.9 1.1 UTAH USA 245 440 62.6 – – 51.4 16.1 – 4.7 WUGS USA 0 988 60.8 6.2 – 42.3 7.9 24.2 0.1 S udies: Copenhagen P os a e Cance S udy 1 (CPCS1); Copenhagen P os a e Cance S udy 2 (CPCS2); Eu opean P ospec i e In es iga ion In o Cance and Nu i ion (EPIC); Epidemiological in es iga ions o he chances o p e en ing, ecognizing ea ly and op imally ea ing ch onic diseases in an elde ly popula ion (ESTHER); F ed Hu chinson Cance Resea ch Cen e (FHCRC); Po uguese Oncology Ins i u e, Po o (IPO- Po o); Mayo Clinic (MAYO); Melbou ne Collabo a i e Coho S udy (MCCS); Mul ie hnic Coho S udy (MEC); The Mo i G oup (MOFFITT); P os a e Cance s udy Medical Uni e si y So ia (PCMUS); P os a e P ojec Founda ion-Pos g adua e Medical School, Su ey (PPF- UNIS); The Poland G oup (Poland); P os a e cance : Mechanisms o p og ession and T ea men (P oMPT); P os a e es ing o cance and T ea men (P o ecT); Re ospec i e Queensland S udy (QLD) and he P os a e Cance Suppo i e Ca e and Pa ien Ou comes P ojec (P osCan); S ockholm 1 (STHMI); Finnish Gene ic P edisposi ion o P os a e Cance S udy (TAMPERE); U.K. Gene ic P os a e Cance S udy and The P os a e Cance Resea ch Founda ion S udy (UKGPCS); Familial P os a e Cance S udy Ulm (ULM); UTAH S udy (UTAH); Washing on Uni e si y Gene ics S udy (WUGS) a Ge many, G eece, I aly, Ne he lands, Spain, Sweden, Ox o d b S udies wi h 0 % sc een de ec ed a e en i ely based on clinically de ec ed cases, and s udies wi h no in o ma ion abou me hod o de ec ion ha e a missing alue; 12,231 indi iduals ha e in o ma ion o me hod o de ec ion 1606 Cance Causes Con ol (2015) 26:1603–1616 123 he numbe o heigh o BMI inc easing isk alleles pe man. Each gene ic a ian was gi en a weigh equal o he e ec o he a ian on heigh o BMI epo ed by he p e ious GWASs [23,24]. The gene ic isk sco e is he e o e a weigh ed sum o he es ima ed numbe o isk alleles ac oss se e al geno ypes, which can imp o e he p ecision o he esul s compa ed o an unweigh ed sco e [21]. Supplemen a y Tables 1 and 2 p o ide de ails o he a ian s used and weigh s assigned. S a is ical analysis We es ima ed associa ions o he gene ic isk sco es wi h measu ed heigh and BMI using linea eg ession based on 1,270 men wi hou p os a e cance [i.e., p os a e-speci ic an igen (PSA) le el 3.0 ng/ml o men wi h a aised PSA bu who we e biopsy nega i e] om he P o ecT popula- ion-based s udy [15,28], one o he PRACTICAL s udies wi h he ele an pheno ypic da a in a well-de ined con ol g oup. We compu ed Fs a is ics and R 2 alues ( he p o- po ion o a ia ion in heigh and BMI explained by he gene ic isk sco e) om he linea eg ession o e alua e he s eng h o he gene ic isk sco e ins umen s in a popula ion o men a inc eased isk o cance . We had 82 and 78 % powe o de ec an odds a io o 1.12 and 1.25 o he e ec s o heigh and BMI on p os a e cance isk, assuming a sample size o 41,062 and ha he gene ic isk sco es explained 6.31 and 1.46 % o he a ia ion in heigh and BMI, espec i ely [29]. We in es iga ed associa ions o he pheno ypes (heigh and BMI) and he gene ic isk sco es ( o heigh and BMI) wi h measu ed co a iables in he P o ecT cases o assess whe he he sco es we e likely o be independen o po en ial en i onmen al con ounding ac o s and o assess he po en ial o pleio opy (gene ic con ounding). We included he ollowing po en ial con ounde s: diabe es; occupa ion (manage ial s. nonmanage ial); exe cise (s enuous; mode a e o s enuous, s. ligh ); alcohol in ake ( h ee o mo e d inks a week s. wo o less); smoking (passi e, cu en , o ex-smoke s. ne e ); diagnos ic PSA le el; and age a ec ui men . We in es iga ed whe he he sco es p edic ed ci cula ing insulin-like g ow h ac o (IGF-I) le els (a po en ial mechanism linking size wi h p os a e cance [13,14]) and benign p os a ic hype plasia (a po en ial cause o de ec ion bias [30]). We assessed he ela ionship o he heigh and BMI gene ic isk sco es wi h p os a e cance isk, s age, and g ade ac oss all 22 eligible s udies con ibu ing o PRAC- TICAL using logis ic eg ession o compu e ORs, wi h obus s anda d e o s o accoun o wi hin-s udy clus e - ing. The gene ic isk sco e was s anda dized o mean ze o and s anda d de ia ion one, and he ORs we e pa ame e - ized as he change in ou come pe s anda d de ia ion inc ease in gene ic isk sco e. In a seconda y analysis, we also compu ed ORs compa ing he highes e sus he lowes quin ile o each gene ic isk sco e o illus a e he di e ences in ou comes be ween he ex emes o he BMI o heigh allele sco e dis ibu ions. This educed o m, he associa ion o he ins umen ( he gene ic isk sco e) wi h he ou come, is a alid es o he di ec ion o he e ec o a pheno ype on an ou come [31,32]. We in es iga ed be ween-s udy he e ogenei y by es ima ing he logis ic eg essions indi idually o each s udy and using he S a a me an command o es ima e he I 2 s a is ic assuming a ixed-e ec model. As we ound li le e idence o he e o- genei y, we epo he ORs om he logis ic eg ession analyses conduc ed ac oss he 22 included s udies. We calcula ed ORs o all p os a e cance s and hen sepa a ely o localized e sus ad anced and low-g ade (Gleason sco e B6) e sus high-g ade (Gleason sco e C7) cance s. Among men wi h p os a e cance (case-only analysis), we es ima ed associa ions o he s anda dized heigh and BMI weigh ed gene ic isk sco es wi h all-cause and p os a e cance -speci ic mo ali y using Cox p opo ional haza ds eg ession, wi h age a diagnosis as he s a da e and age a dea h o inal ollow-up ime- poin as he exi da e, wi h s anda d e o s clus e ed by s udy ( he e was no e idence ha he p opo ional haza ds assump ion was iola ed). We es ed o he e ogenei y in associa ion o he gene ic isk sco es wi h localized e sus ad anced and low- e sus high-g ade p os a e cance isk using a mul i a ia e logis ic eg ession. We es ed o he e ogenei y in he associa ion o he gene ic isk sco es and su i al o pa ien s wi h localized e sus ad anced and low e sus high g ade using he es p oposed by Al man and Bland [33]. Sensi i i y analyses We assessed he po en ial o pleio opy, since i is possible ha a ian s iden i ied in he genome-wide scans a e no speci ic o heigh o BMI and ha e e ec s on he p os a e cance ou comes independen o hei e ec s on he exposu es (heigh o BMI) [34]. I he no-e ec modi ica- ion assump ion holds, simila ins umen al a iable es i- ma es acqui ed using independen ins umen s would p o ide sugges i e e idence agains an in luence o pleio- opic e ec s, as i is unlikely ha hey ha e sha ed pleio opy [21,35]. The e o e, as a sensi i i y analysis we es ed o e idence o he e ogenei y ac oss di e en SNPs o each o ou baseline esul s which di e ed om he null. We gene a ed wo independen gene ic ins umen s o BMI using (1) s1558902 in FTO, he indi idual SNP wi h he la ges e ec size in he me a-analysis o GWASs o BMI [24] and (2) a weigh ed allelic sco e cons uc ed om he emaining BMI-associa ed SNPs. We andomly Cance Causes Con ol (2015) 26:1603–1616 1607 123 spli he heigh allele sco e in o wo independen weigh ed sco es con aining 89 and 90 SNPs ( o de ails o he SNPs in each sco e see Supplemen a y Table 3). The heigh SNPs we e in linkage equilib ium, and hence, hese sco es we e s a is ically independen . We es ima ed he associa- ion o each ins umen wi h p os a e cance and es ed o he e ogenei y [33]. The op eigh p incipal componen s ha e lec he popula ion’s gene ic s uc u e we e es ima ed and included as co a ia es in adjus ed eg ession models o accoun o con ounding by popula ion s a i ica ion. We also epo he associa ions o he gene ic isk sco es wi h su i al addi ionally adjus ed o PSA le el, g ade, and s age. We an all s a is ical analyses in S a a e sion 13.1 (S a aCo p LP, 2014, College S a ion, TX). Resul s Ou sample consis ed o 20,848 cases and 20,214 con ols o Eu opean gene ic descen , wi h geno ypic da a om he iCOGs a ay ha had passed quali y con ol and was no included in he GIANT conso ium used o gene a e he gene ic isk sco es (EPIC-No olk, CAPS, and SEARCH s udies) (Table 1). The pe cen age o high-g ade cance s epo ed a ied be ween s udies (3.6–35.0 %), as did he p opo ion o ad anced s age cance s (3.5–64.5 %). The case-only su i al analysis was based on 15,491 men, because 5,357 o he 20,848 men wi h p os a e cance did no ha e age a en y o exi in he da ase . Associa ions o gene ic isk sco es wi h measu ed heigh and BMI in P o ecT Associa ions o he weigh ed gene ic isk sco es wi h heigh and BMI in he P o ecT sub-sample a e shown in Table 2. The esul s wi h he unweigh ed sco e we e sim- ila , bu less p ecise ( esul s no shown). The gene ic isk sco es explained 6.31 and 1.46 % o he a iabili y in heigh and BMI, espec i ely, consis en wi h p e ious s udies [23,24], which sugges ha he gene ic isk sco es a e s ong ins umen s o he pheno ypes. Associa ions o gene ic isk sco es wi h po en ial con ounde s in P o ecT Talle men we e mo e likely o ha e manage ial jobs, ha e lowe PSA le els, and ha e joined he P o ecT s udy a a younge age (Table 3), bu he e was li le e idence ha he heigh gene ic isk sco e was associa ed wi h any o he con ounde s excep benign hype ophy o he p os a e (all p alues [0.05). Hea ie men we e mo e likely o ha e diabe es; be inac i e; d ink ewe han 3 d inks a week; be a nonsmoke ; and ha e lowe IGF-I le els (Table 3), bu we ound li le e idence ha he BMI gene ic isk sco e was associa ed wi h any o he po en ial con ounde s (all p alues [0.05). Associa ion o he gene ic isk sco es and p os a e cance isk and mo ali y Associa ions o he gene ic isk sco es o heigh and BMI wi h p os a e cance isk a e shown in Table 4, wi h he s udy-speci ic es ima es in Supplemen a y Figu es 1–10. The e was li le consis en e idence ha he gene ic isk sco e o heigh was associa ed wi h p os a e cance , al hough he e was weak e idence o an in e se associa ion wi h ad anced p os a e cance [OR, pe s anda d de ia ion inc ease in heigh gene ic sco e 0.96; 95 % CI 0.93, 0.99, p=0.01; phe e ogenei y, ad anced s. localized 0.05]. The e was weak e idence ha he gene ic isk sco e o BMI was associa ed wi h a educed p os a e cance isk (OR pe s anda d de ia ion inc ease in BMI gene ic sco e 0.98; 95 % CI 0.96, 1.00; p=0.07), bu li le e idence o a ia ion by s age o g ade (phe e ogenei y 0.64 and 0.13, espec i ely). The heigh gene ic isk sco e was associa ed wi h an inc ease in p os a e cance -speci ic mo ali y among men wi h low-g ade disease (OR pe s anda d de ia ion inc ease in he heigh sco e 1.13; 95 % CI 1.08, 1.20, phe e o- genei y, low s. high g ade 0.001), bu he e was li le e idence o associa ions wi h all-cause mo ali y (Table 5). The BMI gene ic isk sco e was associa ed wi h highe all- cause mo ali y among low-g ade disease (OR pe s anda d de ia ion inc ease in he BMI sco e 1.08; 95 % CI 1.03, 1.14, phe e ogenei y low s. high g ade =0.03), bu he e was li le e idence o associa ions wi h p os a e cance - speci ic mo ali y. Table 2 Associa ion o weigh ed heigh and BMI gene ic isk sco es wi h measu ed heigh and weigh in 907 con ols in P o ecT [28] nMean di e ence 95 % CI 2 (%) F-s a is ic Lowe limi Uppe limi Heigh 907 0.26 0.20 0.33 6.31 67.6 BMI 901 0.12 0.06 0.19 1.46 13.6 To allow di ec compa ison o e ec sizes, BMI and heigh pheno ypic measu emen s and he gene ic isk sco es we e no malized o mean ze o and s anda d de ia ion one 1608 Cance Causes Con ol (2015) 26:1603–1616 123 Table 3 Odds a io o change in con inuous a iable co a ia es pe s anda d de ia ion change in ei he heigh and BMI (pheno ypes) o gene ic isk sco es o heigh and BMI (ins umen s) in he P o ecT s udy cases [28] nObse ed pheno ype a Gene ic isk sco es a E ec es ima e Con idence in e al b p alue E ec es ima e Con idence in e al b p alue Lowe Uppe Lowe Uppe S anda dized heigh Odds a io c Odds a io c Bina y a iables Diabe es 726 0.91 0.64 1.30 0.62 0.94 0.70 1.25 0.66 Manage ial occupa ion 818 1.21 1.06 1.40 0.006 0.91 0.79 1.04 0.17 S enuous exe cise 621 1.13 0.96 1.33 0.13 1.03 0.87 1.21 0.75 Mode a e o s enuous exe cise 621 1.15 0.96 1.37 0.12 1.01 0.85 1.20 0.90 C3 d inks in he las week 820 1.13 0.98 1.30 0.09 1.05 0.91 1.22 0.47 Passi e smoke 752 1.03 0.89 1.19 0.72 1.00 0.86 1.16 0.99 E e smoke 780 1.10 0.95 1.27 0.21 1.08 0.93 1.25 0.33 Cu en smoke 552 1.09 0.89 1.35 0.40 1.21 0.97 1.51 0.08 Benign hype ophy o he p os a e 704 0.77 0.58 1.02 0.07 1.38 1.01 1.88 0.05 Reg ession coe icien c Reg ession coe icien c Con inuous a iables PSA (ng/ml) 828 -1.15 -2.01 -0.29 0.009 -0.31 -1.32 0.70 0.55 IGF-I (ng/ml) 718 1.80 -2.23 5.83 0.38 -2.53 -6.31 1.25 0.19 Age (yea s) 1,109 -0.53 -0.83 -0.24 0.001 0.07 -0.23 0.38 0.64 S anda dized BMI Odds a io c Odds a io c Bina y a iables Diabe es 724 1.90 1.45 2.48 0.001 1.16 0.86 1.57 0.33 Manage ial occupa ion 813 0.96 0.83 1.10 0.54 0.91 0.79 1.05 0.20 S enuous exe cise 617 0.91 0.77 1.08 0.28 1.05 0.89 1.23 0.57 Mode a e o s enuous exe cise 617 0.80 0.66 0.96 0.02 1.01 0.85 1.20 0.94 C3 d inks in he las week 814 0.87 0.75 1.01 0.07 0.90 0.78 1.04 0.17 Passi e smoke 748 1.12 0.97 1.30 0.13 0.97 0.84 1.12 0.65 E e smoke 776 1.09 0.93 1.27 0.29 1.09 0.94 1.27 0.25 Cu en smoke 548 0.71 0.54 0.94 0.02 1.18 0.95 1.48 0.13 Benign hype ophy o he p os a e 700 0.92 0.69 1.23 0.56 0.94 0.71 1.25 0.66 Reg ession coe icien c Reg ession coe icien c Con inuous a iables PSA (ng/ml) 822 -0.25 -1.55 1.05 0.70 -0.21 -0.96 0.54 0.58 IGF-I (ng/ml) 714 -5.38 -9.12 -1.64 0.005 1.77 -1.97 5.51 0.35 Age (yea s) 1,101 -0.28 -0.60 0.04 0.08 -0.04 -0.33 0.25 0.79 a Obse ed pheno ypes and gene ic isk sco es no malized o mean ze o and s anda d de ia ion one b Robus s anda d e o s c Odds a io o change in con inuous a iable pe s anda d de ia ion change in heigh and BMI (pheno ype o gene ic isk sco e) Cance Causes Con ol (2015) 26:1603–1616 1609 123 Sensi i i y analysis P os a e cance isk The e was li le e idence ha men wi h heigh a ian s wi h la ge e ec s on he heigh pheno ype we e mo e o less likely o be diagnosed wi h p os a e cance ( 2 =0.0071) (Fig. 1; see Supplemen a y Table 4 o associa ions o each o he heigh a ian s wi h p os a e cance isk). The e was some e idence ha BMI a ian s wi h he la ges e ec on BMI we e mos s ongly in e sely associa ed wi h p os a e cance ( 2 =0.0231) (Fig. 2; Supplemen a y Table 5 o associa ions o each o he BMI a ian s wi h p os a e cance isk). We ound li le e i- dence o he e ogenei y in he e ec o BMI p oxied by independen ins umen s based on independen gene ic sco es made up o di e en se s o SNPs. Indi iduals wi h mo e BMI inc easing FTO alleles we e less likely o be diagnosed wi h p os a e cance (OR pe BMI inc easing allele s1558902-A 0.97; 95 % CI 0.94, 1.01, p=0.10). In line wi h his, he allele sco e based on he emaining 31 BMI SNPs was also in e sely associa ed wi h p os a e cance (OR pe s anda d de ia ion inc ease in BMI gene ic sco e excluding FTO 0.99; 95 % CI 0.97, 1.01, p=0.33; p alue o he e ogenei y be ween he wo independen ins umen s =0.38). All-cause mo ali y The e was li le e idence ha he wo se s o independen heigh o BMI allele sco es we e associa ed wi h an inc eased isk o all-cause mo ali y in men diagnosed wi h p os a e cance (see Supplemen a y Table 10 o associa- ions o all 179 heigh SNPs and all-cause mo ali y and Supplemen a y Table 7 o associa ions o each o he 32 BMI SNPs wi h all-cause mo ali y and p os a e cance - speci ic mo ali y). P os a e cance -speci ic mo ali y Bo h se s o independen heigh allele sco es we e associ- a ed wi h an inc eased isk o p os a e cance -speci ic mo ali y in men diagnosed wi h low-g ade p os a e cance (haza d a io pe one s anda d de ia ion inc ease in he i s heigh allele sco e 1.10; 95 % CI 1.03, 1.19, p=0.008; and in he second heigh allele sco e 1.09; 95 % CI 1.05, 1.13, p 0.001; p alue o he e ogenei y =0.86; see Table 4 Odds a io o p os a e cance pe one s anda d de ia ion change in heigh o BMI gene ic sco e nUnadjus ed Adjus ed a Odds a io c Con idence in e al b p alue Odds a io c Con idence in e al b p alue phe e ogenei y d Lowe Uppe Lowe Uppe Heigh Con ols 20,214 1.00 – – – 1.00 – – – – All p os a e cance s 20,848 0.96 0.91 1.01 0.12 0.99 0.97 1.01 0.23 Localized p os a e cance 12,975 0.96 0.88 1.03 0.27 1.00 0.98 1.02 0.72 0.05 Ad anced p os a e cance 4,325 0.90 0.83 0.98 0.02 0.96 0.93 0.99 0.01 Low-g ade p os a e cance 8,784 0.96 0.90 1.02 0.20 0.99 0.96 1.01 0.30 0.55 High-g ade p os a e cance 8,230 0.97 0.92 1.02 0.26 1.00 0.98 1.02 0.85 BMI Con ols 20,214 1.00 – – – 1.00 – – – – All p os a e cance s 20,848 0.98 0.96 1.01 0.15 0.98 0.96 1.00 0.07 Localized p os a e cance 12,975 0.98 0.96 1.00 0.10 0.98 0.96 1.00 0.05 0.64 Ad anced p os a e cance 4,325 1.01 0.97 1.05 0.69 1.01 0.97 1.05 0.62 Low-g ade p os a e cance 8,784 0.98 0.94 1.02 0.25 0.97 0.94 1.00 0.09 0.13 High-g ade p os a e cance 8,230 1.00 0.97 1.02 0.69 1.00 0.98 1.01 0.65 a Adjus ed o he eigh p incipal componen s o popula ion s a i ica ion b Based in obus s anda d e o s o accoun o wi hin-s udy clus e ing c Change in odds a io pe s anda d de ia ion change in heigh and BMI gene ic isk sco e (s anda dized o mean ze o s anda d de ia ion one) d Localized e sus ad anced, o high- e sus low-g ade using mul i a ia e logis ic eg ession 1610 Cance Causes Con ol (2015) 26:1603–1616 123 Supplemen a y Table 8 o he associa ion o p os a e cance -speci ic mo ali y and each o he 179 heigh SNPs). None o he BMI independen ins umen s o indi idual SNPs we e associa ed wi h p os a e cance -speci ic mo - ali y (Supplemen a y Table 9). Fu he adjus ing he associa ions o he gene ic isk sco es and su i al o PSA le el, g ade, and s age made no subs an ial di e ences o he esul s (Supplemen a y Table 10). Discussion We ound weak e idence ha gene ically ele a ed BMI was associa ed wi h a educed isk o p os a e cance , bu ha gene ically ele a ed heigh was no associa ed wi h p os a e cance isk. The heigh and BMI allele sco es we e posi i ely associa ed wi h p os a e cance -speci ic and all- cause mo ali y, espec i ely, bu only among men wi h low-g ade disease (phe e ogenei y, low- s. high-g ade p os a e cance 0.05). Al hough e idence o hese associa ions was ela i ely weak, he in e se ela ionship o BMI wi h p os a e cance isk is in line wi h bo h obse a ional da a [8] and ou p e ious gene ic s udy [15]. The la e epo showed an in e se ela ionship o a single obesi y- ela ed SNP (FTO s9939609) wi h o e all- and low-g ade p os a e cance in P o ecT, a much smalle popula ion-based sample o 1,550 sc een-de ec ed p os a e cance s and 1,815 con ols [15]. We ound in e se associa ions o a ela ed SNP in FTO ( s1558902, which is in linkage disequilib ium wi h Table 5 Haza d a io o all-cause and p os a e cance -speci ic mo ali y among men wi h p os a e cance pe one s anda d change in heigh o BMI gene ic sco e Numbe o pa icipan s Numbe o ailu es Yea s a isk (1000s) Unadjus ed Adjus ed a Haza d a io c Con idence in e al b p alue Haza d a io c Con idence in e al b p alue phe e ogenei y d Lowe Uppe Lowe Uppe All-cause mo ali y Heigh All cases 14,649 3,591 105 1.02 0.97 1.08 0.47 1.00 0.96 1.04 0.88 Localized 8,553 1,447 65 1.01 0.93 1.09 0.81 1.00 0.93 1.07 0.97 0.20 Ad anced 3,435 1,332 25 1.08 0.98 1.18 0.11 1.07 0.99 1.14 0.07 Low g ade 5,684 905 43 1.04 0.97 1.11 0.32 1.02 0.95 1.09 0.57 0.80 High g ade 5,892 1,365 36 1.02 0.97 1.08 0.36 1.01 0.96 1.06 0.71 BMI All cases 14,649 3,591 105 1.02 0.99 1.05 0.18 1.02 0.99 1.05 0.23 Localized 8,553 1,447 65 1.04 0.99 1.10 0.09 1.04 0.99 1.10 0.09 0.28 Ad anced 3,435 1,332 25 1.01 0.98 1.04 0.50 1.01 0.98 1.05 0.59 Low g ade 5,684 905 43 1.09 1.04 1.15 0.001 1.08 1.03 1.14 0.002 0.03 High g ade 5,892 1,365 36 1.00 0.96 1.05 0.89 1.00 0.95 1.05 0.98 P os a e cance -speci ic mo ali y Heigh All cases 14,649 1,483 105 1.02 0.98 1.06 0.44 1.00 0.97 1.04 0.87 Localized 8,553 363 65 0.98 0.91 1.07 0.72 0.99 0.91 1.08 0.79 0.29 Ad anced 3,435 745 25 1.05 1.00 1.10 0.06 1.04 1.00 1.09 0.07 Low g ade 5,684 188 43 1.13 1.06 1.21 0.001 1.13 1.08 1.20 0.001 0.001 High g ade 5,892 678 36 0.97 0.93 1.02 0.20 0.97 0.93 1.01 0.19 BMI All cases 14,649 1,483 105 0.99 0.96 1.03 0.76 1.00 0.96 1.04 0.94 Localized 8,553 363 65 0.95 0.88 1.03 0.22 0.95 0.87 1.05 0.31 0.09 Ad anced 3,435 745 25 1.04 0.98 1.10 0.18 1.05 0.99 1.10 0.11 Low g ade 5,684 188 43 0.95 0.89 1.01 0.08 0.95 0.88 1.01 0.12 0.03 High g ade 5,892 678 36 1.05 0.99 1.11 0.12 1.05 0.98 1.13 0.14 a Adjus ed o he i s eigh p incipal componen s o popula ion s a i ica ion b Based in obus s anda d e o s o accoun o wi hin-s udy clus e ing c Change in haza d a io pe s anda d de ia ion change in heigh and BMI gene ic isk sco e (s anda dized o mean ze o s anda d de ia ion one) d Localized e sus ad anced, o high- e sus low-g ade using Bland–Al man es s Cance Causes Con ol (2015) 26:1603–1616 1611 123