Germline variants in IL4, MGMT and AKT1 are associated with prostate cancer-specific mortality : an analysis of 12,082 prostate cancer cases

P os a e Cance and P os a ic Diseases (2018) 21:228–237 h ps://doi.o g/10.1038/s41391-017-0029-2 ARTICLE Ge mline a ian s in IL4,MGMT and AKT1 a e associa ed wi h p os a e cance -specific mo ali y: An analysis o 12,082 p os a e cance cases L. M. Fi zGe ald 1,2 ●S. Zhao3●A. Leona dson4●M. S. Geybels4,5 ●S. Kolb4●D. W. Lin4,6 ●J. L. W igh 4,6 ● R. Eeles 7,8 ●Z. Ko e-Ja ai7●K. Go indasami7●G. G. Giles2,9 ●M. C. Sou hey10 ●J. Schleu ke 11,12 ●T. L. Tammela13,14 ● C. Sipeky 11 ●K. L. Penney15,16 ●M. J. S amp e 15,16,17 ●H. G onbe g18 ●F. Wiklund18 ●P. S a in19,20 ●J. Hugosson21 ● D. M. Ka yadi22 ●E. A. Os ande 22 ●Z. Feng23 ●J. L. S an o d4,24 Recei ed: 9 Sep embe 2017 / Re ised: 9 No embe 2017 / Accep ed: 20 No embe 2017 / Published online: 3 Janua y 2018 © The Au ho (s) 2018. This a icle is published wi h open access Abs ac Backg ound P os a e cance (PCa) is a leading cause o mo ali y and gene ic ac o s can influence umou agg essi eness. Se e al ge mline a ian s ha e been associa ed wi h PCa-specific mo ali y (PCSM), bu u he eplica ion e idence is needed. Me hods Twen y- wo p e iously iden ified PCSM-associa ed gene ic a ian s we e geno yped in se en PCa coho s (12,082 pa ien s; 1544 PCa dea hs). Fo each coho , Cox p opo ional haza ds models we e used o calcula e haza d a ios and 95% confidence in e als o isk o PCSM associa ed wi h each a ian . Da a we e hen combined using a me a-analysis app oach. Resul s Fi een SNPs we e associa ed wi h PCSM in a leas one o he se en coho s. In he me a-analysis, a e adjus men o clinicopa hological ac o s, a ian s in he MGMT ( s2308327; HR 0.90; p- alue =3.5 ×10−2)andIL4 ( s2070874; HR 1.22; p- alue =1.1 ×10−3) genes we e confi med o be associa ed wi h isk o PCSM. In analyses limi ed o men diagnosed wi h local o egional s age disease, a a ian in AKT1, s2494750, was also confi med o be associa ed wi h PCSM isk (HR 0.81; p- alue =3.6 ×10−2). Conclusions This me a-analysis confi ms he associa ion o h ee gene ic a ian s wi h isk o PCSM, p o iding u he e idence ha gene ic backg ound plays a ole in PCa-specific su i al. While hese a ian s alone a e no su ficien as p ognos ic bioma ke s, hese esul s may p o ide insigh s in o he biological pa hways modula ing umou agg essi eness. In oduc ion Fo men in many de eloped coun ies, p os a e cance (PCa) is he second leading cause o cance - ela ed dea hs. While PCa mo ali y a es ha e been declining, he numbe o PCa dea hs in Wes e n coun ies is p ojec ed o be sus- ained o decades based on aging o he popula ions [1]. T adi ionally, clinicians ha e assessed a man’s likelihood o ha ing a biologically agg essi e p os a e umou using clinical and pa hological ea u es assessed a diagnosis, including Gleason sco e, umou s age, and se um p os a e- specific an igen (PSA) le el [2]. Resea che s ha e pos u- la ed ha genomic bioma ke s may be able o dis inguish indolen om agg essi e PCa umou s, and se e al s udies ha e iden ified issue-based bioma ke s [3,4]. Tes ing p i- ma y umou issue o p ognos ic bioma ke s may help iden i y cases a highe isk o PCa-specific mo ali y (PCSM) [5] and who would benefi mos om being ea ed agg essi ely ea ly in he disease cou se. We pos ula e ha i is also impo an o conside he hos ’s gene ic backg ound and i s po en ial influence on PCa ou comes. To add ess his issue, ou g oup p e iously assessed ge mline gene ic a ian s in genes om specific biological pa hways hypo hesised o a ec me as a ic p og ession, o de e mine i geno ype was ela ed o PCSM [6]. Twen y- wo PCSM-associa ed a ian s we e iden ified in a *J. L. S an o d [email p o ec ed] Ex ended au ho in o ma ion a ailable on he las page o he a icle Supplemen a y In o ma ion The online e sion o his a icle (h ps://doi.o g/10.1038/s41391-017-0029-2) con ains supplemen a y ma e ial, which is a ailable o au ho ised use s. 1234567890 Sea le-based disco e y coho , and alida ion in a Swedish coho confi med ha he a ian s in fi e genes, LEPR, CRY1,RNASEL,IL4 and ARVCF, we e significan ly asso- cia ed wi h PCSM. Subsequen s udies p o ided addi ional e idence o eplica ion o he ARVCF a ian in he Phy- sicians’Heal h S udy (PHS) pa icipan s wi h PCa [7] and a ian s in RNASEL,XRCC1 and AKT1 in PCa cases pa - icipa ing in he amily-based P os a e Cance Gene ic Resea ch S udy (PROGRESS)[8]. As hese p io s udies had some sho comings, including a limi ed numbe o dea hs due o PCa, we sough o u he e alua e his panel o 22 SNPs in ela ion o PCSM in addi ional independen pa ien coho s, and in a me a-analysis yielding g ea e s a is ical powe om combining hese new da ase s wi h p e iously s udied coho s. Pa ien s and me hods S udy popula ions—new PCa coho s Melbou ne PCa coho s The Melbou ne coho s we e om he P os a e Cance Resea ch P og amme o he Cance Council Vic o ia. The Melbou ne Collabo a i e Coho S udy (MCCS) is a p o- spec i e coho s udy o 41,514 pa icipan s, which has been desc ibed elsewhe e [9]. The MCCS is ma ched o cance egis ies in all Aus alian s a es and na ional dea h indices o asce ain cance diagnoses and dea hs. Fo his s udy, DNA samples we e a ailable o 1100 PCa cases, including 147 who died o PCa. The Ea ly-Onse P os a e Cance Family S udy (EOPCFS) is a popula ion-based amily se ies o 1428 men diagnosed wi h PCa and has been desc ibed elsewhe e [10]. Cases we e asce ained using he popula ion-based Vic o ian Cance Regis y (VCR) and 1531 un ela ed cases wi h a DNA sample we e a ailable o his s udy, including 91 confi med PCa dea hs. Clinical da a we e ob ained om he VCR and we e limi ed o diagnosis age and Gleason sco e. Finnish PCa coho The Finnish coho consis s o PCa cases om he edi a y PCa amilies and om a case-con ol s udy popula ion, desc ibed elsewhe e in de ail [11,12]. All 2629 cases we e o Finnish he i age. PCa diagnoses we e confi med using medical eco ds and su i al da a we e ob ained h ough annual upda es om he Finnish Cance Regis y. Fo PCa dea hs (n=281) iden ified ia annual linkage o he Cance Regis y, unde lying cause was confi med using medical eco ds. Uni ed Kingdom (UK) PCa coho The UK coho comp ises men diagnosed wi h PCa and ec ui ed o he UK Gene ic P os a e Cance S udy (UKGPCS), which has been desc ibed elsewhe e [13]. O he 1560 cases a ailable o his s udy, diagnoses and clinical-pa hological da a we e confi med using medical eco ds. Vi al s a us and cause o dea h we e ob ained om he Na ional Heal h Se ice In o ma ion Cen e and Cen al Regis e , wi h a o al o 221 PCa-specific dea hs. S udy popula ions—p e iously analysed PCa coho s Sea le amily-based PCa coho The PROGRESS [14] includes cases om high- isk he - edi a y PCa amilies. Asce ainmen , eligibili y c i e ia and da a collec ion o his s udy ha e been desc ibed p e- iously [14,15]. Medical eco ds we e ob ained o 961 PCa cases and we e used o ex ac clinical da a on Gleason sco e, s age o disease, and se um PSA le el a diagnosis. Dea h ce ifica es confi med unde lying cause (PCSM o o he ), da e and age a dea h. Fo his coho , 957 cases o Eu opean ances y had DNA a ailable o geno yping, including 98 men who died o PCa [8]. Swedish PCa coho The Swedish popula ion-based PCa coho comp ises cases en oled in Cance o he P os a e in Sweden, which has been desc ibed elsewhe e [6,16]. Fo he cu en s udy, 2875 cases o Eu opean descen had DNA a ailable o geno yping and 501 had PCa confi med as he unde lying cause o dea h [17]. Clinical da a we e ob ained om he Swedish cance egis y. PHS PCa coho The PHS began as a andomised, double-blind placebo- con olled ial o aspi in, and β-ca o ene o he p e en ion o ca dio ascula disease and cance and has been desc ibed in de ail elsewhe e [18]. The 1430 PCa cases in his s udy we e p e iously chosen o a nes ed case-con ol s udy [19] and a e es ic ed o sel - epo ed Caucasians. Fo hese analyses, 194 PCa dea hs and 11 men wi h bone me as ases we e included [7]. Clinical da a we e abs ac ed om medical eco ds. All s udies we e app o ed by hei local Ins i u ional Re iew Boa d o Human Resea ch E hics Commi ee. W i en in o med consen was ob ained om all s udy pa icipan s. Inhe i ed p edisposi ion o agg essi e p os a e cance 229 Geno yping Twen y- wo candida e SNPs [6] we e geno yped o his alida ion s udy. The MassARRAY iPLEX sys em (Sequenom, Inc.) was used o geno ype he Swedish and Finnish samples, and 20 o he 22 SNPs in he PROGRESS samples. The emaining wo SNPs (PROGRESS) and all 22 SNPs we e geno yped in he Aus alian coho s using TaqMan assays (Applied Biosys ems). The PHS samples we e geno yped using BioT o e OpenA ay Technology (Applied Biosys ems) and he UK samples we e geno yped on he Infinium OncoA ay 500K BeadChip (Illumina, Inc.). Two SNPs ailed geno yping in he Swedish coho , s228697 and s1029153. In he UK coho , nine o he 22 SNPs we e eplaced wi h a su oga e SNP ha was in s ong linkage disequilib ium (LD; 2≥0.85) wi h he o iginal SNP (Supplemen a y Table 1). Blind duplica e samples we e dis ibu ed e enly ac oss all geno yping ba ches om each s udy coho . Con- co dance o he 22 SNP geno ypes was 100% o he 53 Finnish duplica es, 99% o he 24 EOPCFS duplica es, 97% o he 21 MCCS duplica es and >93% o he 16 UK duplica es. Samples wi h ≥5 ailed SNPs we e emo ed om u he analyses (n=49 Finnish, n=110 EOPCFS, and n=384 MCCS cases). One Finnish case was emo ed due o missing ollow-up da a. Quali y con ol (QC) esul s o he Swedish, PROGRESS and PHS s udies ha e been epo ed p e iously [6–8]. A e QC measu es, 12,082 PCa cases, including 1544 confi med PCa dea hs, we e a ailable o analysis. The mino allele equencies (MAF) o he 22 SNPs in men who did no die o PCa om each pa ien coho a e shown in Supplemen a y Table 2. Fo mos SNPs, he MAF is ai ly simila ac oss he coho s wi h he excep ion o he Finnish coho . Se e al SNPs in he Finnish coho ha e a MAF a leas 10% highe han wha was ound in he o he coho s, e.g., s1137100, s627839, s4583514, and s2070874. The dis ibu ion o MAF o he h ee SNPs associa ed wi h PCSM o each g oup o pa ien s (ali e, o he cause o dea h, and PCa-specific dea h) o each coho is shown in Supplemen a y Table 3, excluding he PHS (only summa y geno yping da a we e a ailable) and Swedish (missing o he cause dea h in o ma ion) coho s. S a is ical analyses Haza d a ios (HR), 95% confidence in e als (95% CI) and p- alues o each SNP in ela ion o PCSM we e calcula ed using Cox p opo ional haza ds eg ession models o each o he se en independen coho s. Men we e ollowed om da e o diagnosis o da e o : (1) PCa-specific dea h; (2) dea h om ano he cause; o (3) las ollow-up. Those who died o o he causes and su i o s we e ea ed as censo ed obse a ions. The mino allele o each SNP in he Sea le- based PCa disco e y coho was conside ed he “a isk” allele. Fo each SNP, wo Cox models we e es ed. In he fi s model, bo h he gene ic model (addi i e, dominan , o ecessi e) and clinicopa hological co a ia es (age a diag- nosis, Gleason sco e, s age, diagnos ic PSA, and p ima y ea men ) ha we e ound o be significan in he o iginal Sea le coho we e fixed [6]. In he second model, he gene ic model emained fixed based on he o iginal Sea le coho , bu he clinicopa hological co a ia es we e allowed o a y acco ding o he bes -fi ing model o each coho . Missing indica o a iables we e included i clin- icopa hological co a ia es had some (bu no all) missing da a. Fo bo h Aus alian coho s, only wo co a ia es (age a diagnosis and Gleason sco e) we e conside ed in hese models due o missing da a. We hen pe o med me a-analyses o agg ega e e idence ac oss hese s udies using he R package, Me a o [20]. Da a om he o iginal Sea le-based disco e y coho we e no included in he me a-analyses. We fi ed an in e cep -only linea model o each SNP, wi h log HRs es ima ed om he se en coho s as he ou comes, and weigh ed by he in e se o hei co esponding s anda d e o squa es. The fi s me a-analysis was un based on he coe ficien s es ima ed wi h he combina ion o co a ia es ha we e significan in he o iginal Sea le coho (fi s model) and he second was based on he bes fi ing co a ia es o each coho (second model). As we we e es ing an a p io i defined hypo hesis o each SNP, an associa ion was conside ed s a is ically significan i he nominal p- alue was <0.05 (one-sided es ). A one- ailed es was used because o alida ion we equi ed ha he e ec o he isk allele on PCSM be in he same di ec ion as in he o iginal Sea le da ase [6]. Due o he di e en MAFs in he Finnish coho and missing clinicopa hological co a ia es in he Aus alian da ase s, sensi i i y analyses we e pe o med whe e he Finnish o bo h he Finnish and Aus alian da ase s we e excluded. In o he sensi i i y analyses, men diagnosed wi h dis an o unknown s age PCa we e excluded due o unce ain y in defining he p ocess o me as a ic p og ession o le hali y in such pa ien s, and o e alua e SNP associa- ions in men diagnosed wi h less ad anced disease. Resul s The cha ac e is ics o he se en geno yped PCa coho s a e p esen ed in Table 1. O e all, he e we e 12,082 cases wi h geno yping da a om ac oss he s udies, o which 1544 (12.8%) had died o PCa. As di e en coho s may ha e di e en unde lying gene ic suscep ibili ies and dis ibu ions o clin- icopa hological ea u es, each coho was fi s e alua ed 230 L. M. Fi zGe ald e al. independen ly o associa ions be ween he 22 SNP geno- ypes and isk o PCSM. Fi een SNPs we e significan ly associa ed wi h PCSM in a leas one o he se en coho s, and he isk alleles o ou SNPs, s1137100 (LEPR), s2070874 (IL4), s2494750 (AKT1), and s5993891 (ARVCF), we e associa ed wi h PCSM in wo o he coho s (Supplemen a y Table 4). Me a-analysis o he se en coho s confi med ha wo SNPs we e associa ed wi h PCSM (Table 2). The In e - leukin 4 (IL4) SNP, s2070874, was associa ed wi h PCSM unde he same gene ic model adjus ed o he same co - a ia es as in he o iginal Sea le disco e y coho (dominan ; adjus ed o age a diagnosis; p=1.1 ×10−2), and also when clinicopa hological co a ia es we e included in he Table 1 Cha ac e is ics o he se en independen p os a e cance coho s Swedish PROGRESS PHS Aus alia (EOPCFS) Aus alia (MCCS) Finnish UK Sample size 2875 957 1430 1531 1100 2629 1560 Age a diagnosis (yea s) Mean 65.8 64.5 70.5 52.9 67.9 68.6 61.7 Range 44.6–80.4 40.0–87.0 45.5–100.9 38.0–87.0 47.0–86.0 37.0–95.0 36.9–88.9 Follow-up ime (yea s) Mean 6.0 12.73 11.0 8.1 8.5 9.1 3.8 Range 0.3–8.6 0.3–32.6 0.01–27.9 0.9–18.2 0.01–24.5 0.1–31.9 0.01–11.1 Age a dea h (yea s) Mean 71.2 80.5 84.2 61.0 76.4 77.7 67.7 Range 48.5–85.7 54.0–99.0 60.9–104.3 42.9–89.9 53.1–91.7 44.2–105.6 41.7–94.2 PCa-specific mo ali y Noa2374 (82.6) 782 (81.7) 1225 (85.7) 1439 (94.2) 953 (86.6) 2196 (83.5) 1335 (85.6) Yes 501 (17.4) 98 (10.2) 205 (14.3) 91 (5.8) 147 (13.4) 281 (10.7) 221 (14.2) Unknownb0 (0) 77 (8.0) 0 (0) 1 (0.1) 0 (0.0) 152 (5.8) 4 (0.3) S age Local 1885 (65.6) 623 (65.1) 1293 (90.4) NA NA 1910 (72.7) 967 (62.0) Regional 651 (22.6) 240 (25.1) 54 (3.8) NA NA 467 (17.8) 324 (20.8) Dis an 266 (9.3) 27 (2.8) 54 (3.8) NA NA 203 (7.7) 105 (6.7) Missing 73 (2.5) 67 (7.0) 29 (2.0) NA NA 49 (1.9) 164 (10.5) Gleason sco e ≤6 1375 (47.8) 561 (58.6) 662 (46.3) 744 (48.6) 527 (47.9) 1187 (45.2) 627 (40.2) 7 782 (27.2) 204 (21.3) 424 (29.7) 617 (40.3) 285 (25.9) 753 (28.6) 523 (33.5) 8–10 467 (16.2) 80 (8.4) 192 (13.4) 116 (7.6) 146 (13.3) 410 (15.6) 289 (18.5) Missing 251 (8.7) 112 (11.7) 152 (10.6) 54 (3.5) 142 (12.9) 279 (10.6) 121 (7.8) Diagnos ic PSA le el (ng/mL) <4 148 (5.1) 77 (8.0) 116 (8.1) NA NA 217 (8.3) 272 (17.4) 4–9.9 993 (34.5) 360 (37.6) 558 (39.0) NA NA 1114 (42.4) 641 (41.1) 10–19.9 651 (22.6) 157 (16.4) 206 (14.4) NA NA 654 (24.9) 285 (18.3) ≥20 1003 (34.9) 130 (13.6) 136 (9.5) NA NA 574 (21.8) 293 (18.8) Missing 80 (2.8) 233 (24.3) 414 (28.9) NA NA 70 (2.7) 69 (4.4) P ima y he apy Radical p os a ec omy 713 (24.8) 501 (52.4) 579 (40.5) NA NA 823 (31.3) 415 (26.6) Radia ion he apy 682 (23.7) 256 (26.8) 383 (26.8) NA NA 624 (23.7) 707 (45.3) And ogen dep i a ion 927 (32.2) 21 (2.2) 119 (8.3) NA NA 149 (5.7) 208 (13.3) Ac i e su eillance 488 (17.0) 51 (5.3) 91 (6.4) NA NA 970 (36.9) 183 (11.7) O he 22 (0.8) 42 (4.4) 21 (1.5) NA NA 15 (0.6) 47 (3.0) Missing 43 (1.5) 86 (9.0) 237 (16.6) NA NA 48 (1.8) 0 NA no a ailable aDied o o he causes bDied bu unknown cause o dea h Inhe i ed p edisposi ion o agg essi e p os a e cance 231 model and bes -fi ed o each coho (p=1.1 ×10−3). The O6-me hylguanine-DNA me hyl ans e ase (MGMT) SNP, s2308327, was also associa ed wi h PCSM when analysed using he same gene ic model as de e mined using he o i- ginal Sea le-based coho , bu only when clin- icopa hological co a ia es we e included in he model (addi i e; p=3.5 ×10−2). Th ee o he SNPs, s228697 (PER3), s12467911 (SRD5A2), and s4645959 (c-MYC) we e associa ed wi h PCSM in he me a-analysis, bu he di ec ion o associa ion was opposi e o ha obse ed in he o iginal Sea le disco e y coho , so hese a ian s we e no conside ed alida ed (Table 2). A sensi i i y analysis was pe o med o e alua e he SNP–PCSM associa ions when pa ien s p esen ing wi h dis an o unknown s age we e excluded. The esul s o he IL4 and MGMT SNPs we e obus o his sensi i i y ana- lysis. In addi ion, when limi ing he analysis o men diag- nosed wi h local o egional s age he e was confi ma o y e idence ha he SNP ( s2494750) in AKT1 was associa ed wi h PCSM unde he same gene ic model adjus ed o he same co a ia es as in he o iginal Sea le disco e y coho (addi i e; adjus ed o clinicopa hological co a ia es; HR = 0.81, 95% CI 0.67–0.98, p=3.6 ×10–2), and also when clinicopa hological co a ia es we e included in he model and bes -fi ed o each coho (HR =0.83, 95% CI 0.70–0.98, p=3.1 ×10−2). O he sensi i i y analyses excluded he Finnish and/o Aus alian da ase s. When he Finnish coho was excluded he associa ion be ween PCSM and s2070874 (IL4) geno- ype emained significan whe eas he associa ion wi h Table 2 Me a-analysis esul s o 22 SNPs geno yped in se en p os a e cance coho s SNP Gene Risk allele equencya Disco e y coho gene ic modelb& adjus men co a ia esc Haza d a iod 95% CIep- alue Haza d a iog 95% CIep- alue s1137100 LEPR G: 0.27 Dom–ACP 1.01 0.91–1.12 NR 0.98 0.88–1.08 0.35 s228697hPER3 G: 0.11 Dom–ACP 1.33 1.14–1.55 NR 1.24 1.07–1.44 NR s635261 RNASEL C: 0.36 Rec–ACP 0.96 0.82–1.11 0.31 0.92 0.80–1.06 0.18 s627839 RNASEL T: 0.47 Dom–ACP 1.05 0.94–1.18 0.24 1.07 0.96–1.19 0.17 s4583514 MSH2 A: 0.38 Dom–ACP 1.03 0.92–1.14 0.35 1.04 0.94–1.15 0.28 s4608577 MSH2 G: 0.17 Add–A 1.02 0.94–1.11 0.34 1.03 0.95–1.13 0.27 s523349 SRD5A2 G: 0.29 Dom–A 1.08 0.99–1.19 NR 1.07 0.97–1.19 NR s12467911 SRD5A2 T: 0.28 Dom–A 1.10 1.00–1.20 NR 1.09 0.99–1.21 NR s11710277 SEMA3F G: 0.09 Dom–ACP 1.03 0.89–1.20 0.35 0.95 0.83–1.09 NR s11205 HSD17B4 G: 0.39 Rec–ACP 1.06 0.92–1.21 NR 1.06 0.93–1.21 NR s2070874 IL4 T: 0.16 Dom–A 1.14 1.04–1.26 0.01 1.22 1.10–1.35 1.1 ×10–3 s1799964 TNF C: 0.21 Dom–A 1.07 0.98–1.18 NR 1.06 0.96–1.17 NR s4645959 C-MYC G: 0.04 Add–ACP 1.11 0.92–1.35 NR 1.24 1.03–1.50 NR s1029153hCXCL12 C: 0.31 Add–A 1.04 0.95–1.14 NR 1.04 0.95–1.14 NR s2839685 CXCL12 T: 0.15 Rec–ACP 0.71 0.45–1.14 NR 0.66 0.41–1.05 NR s2308327 MGMT G: 0.13 Add–A0.93 0.84–1.02 0.11 0.90 0.81–0.99 0.03 s10778534 CRY1 C: 0.36 Dom–A 1.03 0.94–1.13 0.28 1.03 0.93–1.13 0.32 s2494750 AKT1 G: 0.07 Add–ACP 0.92 0.79–1.08 0.21 0.92 0.79–1.06 0.16 s1799814 CYP1A1 A: 0.05 Add–ACP 1.09 0.90–1.32 NR 1.05 0.88–1.27 NR s25487 XRCC1 A: 0.36 Add–A 0.95 0.89–1.02 0.11 0.94 0.88–1.01 0.09 s915927 XRCC1 G: 0.43 Dom–A 1.03 0.94–1.14 0.30 0.98 0.89–1.10 NR s5993891 ARVCF T: 0.05 Dom–ACP 0.92 0.76–1.12 0.25 0.89 0.75–1.06 0.14 aBased on he Sea le p os a e cance disco e y coho bDom dominan model; Rec ecessi e model; Add addi i e model cAdjus ed o age a diagnosis (A) o age +clinicopa hological (ACP) ac o s (Gleason sco e, s age, PSA, p ima y ea men ) dBo h he gene ic model and adjus men co a ia es we e fixed based on he Sea le disco e y coho eOne-sided 95% confidence in e als NR no eplica ed (HR is in he opposi e di ec ion o he HR in he disco e y coho ) gThe gene ic model was fixed based on he Sea le disco e y coho , bu he adjus men co a ia es a y ac oss he se en coho s hThese SNPs we e no geno yped in he Swedish coho SNPs wi h s a is ically significan e idence o alida ion in he o e all me a-analysis a e shown in bold ace. 232 L. M. Fi zGe ald e al. s2308327 (MGMT) was a enua ed (Supplemen a y Table 5). When bo h Aus alian coho s we e excluded om he analyses, esul s o he IL4 and MGMT SNPs we e simila o hose shown in Table 2. The ATK1 SNP was also associa ed wi h PCSM (HR =0.83; 95% CI 0.70–0.98; p= 0.04) unde an addi i e gene ic model, allowing clin- icopa hological co a ia es o a y by coho o ob ain he bes -fi ing model. Las ly, esul s o IL4 and MGMT a - ian s (Table 2) we e simila a e excluding he Finnish and bo h Aus alian coho s. Discussion Twen y- wo PCSM-associa ed a ian s we e p e iously iden ified in a Sea le-based disco e y coho , ye sub- sequen indi idual eplica ion s udies ha e only confi med subse s o hese a ian s. In his la ge me a-analysis o 12,082 PCa pa ien s om se en coho s, we confi m asso- cia ions be ween wo SNPs, s2070874 (IL4) and s2308327 (MGMT), and isk o PCSM. In addi ion, he me a-analysis highligh ed an associa ion wi h an AKT1 SNP in he subse o men diagnosed wi h less ad anced PCa (i.e., local o egional s age disease) o when bo h Aus alian da ase s missing s age da a we e excluded. Findings om sensi i i y analyses we e obus o he IL4 and MGMT SNPs, and p o ide suppo i e e idence ha a ian s in h ee genes (IL4,MGMT, and AKT1) may play a ole in media ing PCa agg essi eness. P e ious s udies ha e shown ha MGMT and AKT1 a ian s a e no associa ed wi h o e all PCa isk [21–24] and while a nominal associa ion has been obse ed be ween isk and he IL4 a ian , s2243228 [22], his a ian is no linked o s2070874 ( 2=0.0127). Howe e , ano he IL4 a ian , s2243250, which is in comple e link- age disequilib ium wi h s2070874, was ecen ly associa ed wi h Gleason sco e 7–10 PCa in men andomised o he finas e ide a m o he P os a e Cance P e en ion T ial [25]. A s udy in 2011 ound nominal e idence o sugges ha AKT1 gene ic a ia ion had a possible ole in ela ion o isk o mo e agg essi e PCa [22], bu he esul s we e no confi med in la ge s udies (i.e., OncoA ay da a). Collec- i ely, hese esul s ha e a numbe o impo an implica- ions in ela ion o PCa ou comes. Fi s , hey suppo he hypo hesis ha unde lying gene ic backg ound can influ- ence an indi idual’s isk o PCSM. Second, a deepe unde s anding o his gene ic p edisposi ion could e en- ually lead o ea ly isk s a ifica ion and he disco e y o he apeu ic a ge s o ea ing high- isk cases. In ac , IL4, MGMT, and AKT1 ha e well documen ed oles in ca ci- nogenesis and hey, o hei ecep o s, ha e been sugges ed as he apeu ic a ge s o PCa. In he immune sys em, IL4, a T helpe ype 2 (TH2) cy okine, egula es he su i al, g ow h, and di e en ia ion o B and T lymphocy es [26], mas cells [27], and endo- helial cells [28] h ough ac i a ion o he Type I IL4 ecep o (IL4R). In umo igenesis, s udies o he e ec s o IL4 a e conflic ing; ea ly wo k sugges ed he cy okine had an i- umou e ec s [29,30], bu mo e ecen s udies ha e demons a ed umo igenic e ec s, including he p omo ion o cance cell su i al and p oli e a ion [31], g ea e mig a ion and in asion [32], enhanced me abolism o umou g ow h [33] and highe me as a ic umou bu den [32]. In PCa, s udies ha e shown ha IL4 le els a e ele- a ed in ho mone e ac o y disease [34], ha IL4 can ac i a e he and ogen ecep o when and ogen is abla ed o p esen a e y low le els [35], and ha o e exp ession o IL4 enhances he g ow h o and ogen-sensi i e LNCaP cells in and ogen-dep i ed condi ions [36]. In epi helial cance cells, IL4 exe s i s e ec s h ough he Type II IL4R ( e iewed in [37]), which was ound o be o e exp essed in PCa cell lines, p ima y cul u es es ablished om esh p os a e umou s and p os a e umou specimens [38]. No ably, se e al he apies ha e been designed o a ge he IL4/IL4R signalling axis h ough i s ole in as hma and alle gy ( e iewed in [37]). While he apies specific o he Type II IL4R a e s ill in he disco e y phase, a Pseudo- monas endo oxin-based IL4 chime ic p o ein, IL4-CTx, which a ge s bo h IL4 ecep o s, has been shown o cause emission o xenog a umou s de eloped om wo PCa cell lines, DU145 and LNCaP [38]. This is pa icula ly ele an o ou finding ha he s2070874 a ian o IL4 is associa ed wi h a g ea e isk o PCSM, and i is possible ha cases ca ying his a ian could benefi om adju an ea men wi h eme ging Type II IL4R he apies. The MGMT p o ein is esponsible o epai o DNA adduc s gene a ed by alkyla ing agen s. Alkyla ion o DNA in ol es he addi ion o an alkyl g oup o he O6-posi ion o guanine, which induces mu a ion and malignan ans o - ma ion due o me hylguanine: hymine mispai ing du ing DNA eplica ion [39]. MGMT epai occu s h ough he co alen ans e o he alkyl g oup o i s ac i e si e, which esul s in a con o ma ional change, ubiqui ina ion and a apid deg ada ion o he p o ein [40]. While MGMT has an impo an ole in p e en ing ca cinogenesis h ough i s ole in DNA epai , MGMT ac i i y in umou s ea ed wi h chemo he apeu ic O6-alkyla ing agen s is ac ually de i- men al, educing he sensi i i y o he cance cells o che- mo he apy. MGMT p o ein le els ha e been shown o a y widely bo h wi hin and be ween indi iduals [41], and he e is e idence o sugges his is due o inhe i ed gene ic a - ia ion [42], which also al e s MGMT ac i i y [43]. Ma gi- son and colleagues [42] ha e shown ha he a ian alleles o wo SNPs in pe ec linkage disequilib ium, s2308321 (I143V) and s2308327 (K178R), a e associa ed wi h a highe le el o MGMT ac i i y and a e mo e esis an o inac i a ing pseudosubs a es. This may be due o mo e Inhe i ed p edisposi ion o agg essi e p os a e cance 233 e ficien epai o bulky adduc s as a esul o he s2308321 amino acid change, which is wi hin he MGMT-binding pocke and in close p oximi y o he ac i e si e C145 [43]. He e, we obse ed ha he s2308327 a ian was asso- cia ed wi h a educed isk o PCSM, sugges ing ha inhe i ance o he mo e ac i e p o ein o m may p o ec cases om de eloping a high equency o mu a ions in genes c i ical o umo igenesis and ha push he umou owa d an agg essi e pheno ype. Howe e , cases ca ying he s2308327 a ian may also be mo e esis an o che- mo he apeu ic O6-alkyla ing agen s and may benefi om concu en ea men wi h an MGMT inac i a o , such as lomegua ib [44]. AKT1 is a membe o he AKT amily o se ine/ h eonine kinases, and wi hin he PI3K/AKT pa hway, plays a key ole in cellula me abolism, g ow h, p oli e a ion, di e - en ia ion, and su i al [45,46]. The PI3K/AKT pa hway also has a cen al unc ion in epi helial o mesenchymal ansi ion (EMT), a key p ocess in umou p og ession and me as asis [47]. Fu he mo e, al e a ions in he PI3K/AKT pa hway ha e been epo ed in bo h p ima y and me as a ic p os a e umou s [48], including cons i u i e ac i a ion o AKT1 ia loss o he inhibi o y phospha ase, PTEN [49–51], and he de elopmen o doce axel esis ance has been linked o his pa hway in PCa pa ien s [52]. The in ol emen o AKT1 in cance de elopmen and p og ession has made i a a ge o he apeu ic in e en ion [53–55] and se e al Phase I and II ials, p edominan ly in b eas cance pa ien s, a e cu en ly unde way es ing AKT1 o PI3K/AKT pa h- way inhibi o s. Ou s udy also illus a es how MAFs ha a y sub- s an ially ac oss popula ions can impac es ima es o isk. This is pa icula ly s iking in he Finnish popula ion whe e he MAF o se e al gene a ian s (MSH2,HSD17B4,IL4, and CXCL12) is qui e di e en o ha o he o he s udy popula ions. As he IL4 s2070874 a ian is mo e common in he Finnish PCa coho , i may also be mo e equen in he o e all Finnish popula ion, hus explaining why his a ian is mo e s ongly associa ed wi h PCSM when his coho is emo ed om he me a-analysis, especially as he associa ion may be d i en by Swedish and Aus alian coho s (Supple- men a y Tables 4 and 5). Whe eas he MAF o he MGMT a ian is simila ac oss he popula ions and i s associa ion wi h PCSM is a enua ed when he Finnish coho is emo ed; his may be due o a loss o powe as he asso- cia ion be ween MGMT and PCSM appea s o be d i en by all and no indi idual coho s. These findings demons a e he impo ance o conside ing unde lying a ian equencies when combining da a om di e en popula ions. A limi a ion o ou s udy was he le el o missing clin- icopa hological da a o some PCa pa ien coho s. Fo example, we we e unable o s a i y Gleason sco e 7 pa ien s in o Gleason pa e n 3 +4 e sus 4 +3, es ic ing ou abili y o e alua e associa ions o hese wo dis inc umou g ades ha ha e di e en su i al ou comes. We we e able o exclude men wi h missing da a on s age and men wi h dis an s age disease, which demons a ed obus findings o IL4,MGMT, and AKT1 a ian s in men diag- nosed wi h localised o egional s age disease. I should also be no ed ha he e was no cen al e iew o pa hology slides o assign Gleason sco e, he e o e he e may ha e been some umou g ade misclassifica ion ac oss cases in hese coho s; bu i is unlikely ha such misclassifica ion would di e subs an ially be ween coho s o ha i would be influenced by geno ype. In addi ion, while all bu wo o he PCa coho s had in o ma ion a ailable on p ima y ea men , we did no ha e in o ma ion on seconda y he apies ha may ha e been used o ea PCa p og ession and could ha e a ied be ween popula ions. The e is cu - en ly no e idence ha he IL4,MGMT,o AKT1 a ian alleles al e esponse o he apy, and i seems unlikely ha use o seconda y ea men (s) by pa ien s in hese PCa coho s would a y subs an ially by geno ype. Ano he limi a ion o ou s udy is ha all o he coho s we e com- p ised o pa ien s o Eu opean ances y. Men o A ican ances y ha e a highe PCa mo ali y a e compa ed o men o Eu opean ances y [56,57], and u u e s udies o hese gene ic a ian s in ela ion o PCSM a e impe a i e in ha high- isk popula ion. Unde s anding which gene ic pa hways a e in ol ed in media ing PCa p og ession o a a al endpoin may lead o he disco e y o no el p ognos ic bioma ke s and he - apeu ic a ge s. While he IL4,MGMT, and AKT1 isk alleles confi med in his s udy a e insu ficien as p ognos ic bioma ke s on hei own, he iden ifica ion o u he bio- ma ke s in he same o simila biological pa hways may lead o he de elopmen o a bioma ke panel ha could imp o e s a ifica ion o cases a diagnosis in o low- isk and high- isk ca ego ies [58]. Such in o ma ion could be use ul o decide on ini ial and adju an ea men s, clinical ial en olmen , ea ly sal age he apy and mo e in ensi e su - eillance in men a highe isk o PCSM. Fu he mo e, as ou s udy sugges s a common gene ic suscep ibili y ac oss se e al in e na ional PCa coho s, such a panel may be ele an a a global le el. Acknowledgemen s The au ho s g a e ully acknowledge ha hese s udies we e made possible by he con ibu ion o many people, including he o iginal in es iga o s and he diligen eams who ec ui ed he pa icipan s and con inue o wo k on ollow-up. We also exp ess ou g a i ude o he men who ook pa in hese s udies and p o ided blood samples. This wo k was suppo ed by awa ds om he Mo embe Founda ion ( o R.E. and J.L.S.) and g an s om he Na ional Cance Ins i u e (P50 CA097186 and K05 CA175147 o J.L. S.), wi h addi ional suppo om he F ed Hu chinson Cance Resea ch Cen e . The PROGRESS s udy was suppo ed by g an R01 CA080122 om he Na ional Cance Ins i u e and an awa d om he P os a e Cance Founda ion. The Finnish s udy was suppo ed by 234 L. M. Fi zGe ald e al. g an s om he Finnish Cance O ganisa ions, Sig id Juselius Foun- da ion, and he Academy o Finland (251074). The MCCS and EOPCFS we e suppo ed by Cance Council Vic o ia and he MCCS u he by he Aus alian Na ional Heal h and Medical Resea ch Council p ojec g an s, 940394, 126402, 209057, and 1043616. E.A. O. and D.M.K. acknowledge he In amu al P og amme o he Na ional Human Genome Resea ch Ins i u e. The PHS was suppo ed by Depa men o De ence g an s PC050569 and PC073618, Na ional Cance Ins i u e g an s CA-42182, CA-34944, CA-40360, and CA- 097193, and Na ional Hea , Lung, and Blood Ins i u e g an s HL- 26490 and HL-34595. S.Z. was suppo ed in pa by he In amu al Resea ch P og amme o he NIH, Na ional Ins i u e o En i onmen al Heal h Sciences. K.L.P. was suppo ed by Na ional Resea ch Se ice Awa ds T32 CA009001-32 and R25 CA098566. R.E. acknowledges he suppo o he Na ional Ins i u e o Heal h Resea ch o he Bio- medical Resea ch Cen e a The Ins i u e o Cance Resea ch and Royal Ma sden Na ional Heal h Se ice Founda ion T us . The au ho s hank he UK Gene ic P os a e Cance S udy Collabo a o s (Supple- men a y No e; www.ic .ac.uk/ukgpcs). OncoA ay geno yping was unded by he US Na ional Ins i u es o Heal h (U19 CA 148537: ELLIPSE; ×01HG007492: Cen e o Inhe i ed Disease Resea ch, con ac numbe HHSN268201200008I]. The PCa componen o he OncoA ay (PRACTICAL conso ium; h p://p ac ical.ccge.medschl. cam.ac.uk/) was suppo ed by Eu opean Commission’s Se en h F a- mewo k P og amme g an ag eemen no. 223175 (HEALTH-F2-2009- 223175), Cance Resea ch UK G an s C5047/A7357, C1287/A10118, C1287/A16563, C5047/A3354, C5047/A10692, C16913/A6135, and he Na ional Ins i u e o Heal h Cance Pos -Cance GWAS ini ia i e g an no. 1 U19 CA 148537-01 ( he GAME-ON ini ia i e). Funding This wo k was suppo ed by awa ds om he Mo embe Founda ion ( o R.E. and J.L.S.) and g an s om he Na ional Cance Ins i u e (P50 CA097186 and K05 CA175147 o J.L.S.), wi h addi- ional suppo om he F ed Hu chinson Cance Resea ch Cen e . Compliance wi h e hical s anda ds Conflic o in e es The au ho s decla e ha hey ha e no conflic o in e es . Open Access This a icle is licensed unde a C ea i e Commons A ibu ion-NonComme cial-Sha eAlike 4.0 In e na ional License, which pe mi s any non-comme cial use, sha ing, adap a ion, dis ibu ion and ep oduc ion in any medium o o ma , as long as you gi e app op ia e c edi o he o iginal au ho (s) and he sou ce, p o ide a link o he C ea i e Commons license, and indica e i changes we e made. I you emix, ans o m, o build upon his a icle o a pa he eo , you mus dis ibu e you con ibu ions unde he same license as he o iginal. The images o o he hi d pa y ma e ial in his a icle a e included in he a icle’s C ea i e Commons license, unless indica ed o he wise in a c edi line o he ma e ial. I ma e ial is no included in he a icle’s C ea i e Commons license and you in ended use is no pe mi ed by s a u o y egula ion o exceeds he pe mi ed use, you will need o ob ain pe mission di ec ly om he copy igh holde . To iew a copy o his license, isi h p://c ea i ecommons. o g/licenses/by-nc-sa/4.0/. Re e ences 1. B ay F, Soe joma a am I. The changing global bu den o cance : ansi ions in human de elopmen and implica ions o cance p e en ion and con ol. In: Gelband H, Jha P, Sanka ana ayanan R, Ho on S (eds) Cance : disease con ol p io i ies. The In e na ional Bank o Recons uc ion and De elopmen /The Wo ld Bank. Washing on, DC; 2015. 2. Eifle JB, Feng Z, Lin BM, Pa in MT, Humph eys EB, Han M, e al. An upda ed p os a e cance s aging nomog am (Pa in ables) based on cases om 2006 o 2011. BJU In . 2013;111:22–9. 3. Ma in NE. New de elopmen s in p os a e cance bioma ke s. Cu Opin Oncol. 2016;28:248–52. 4. Moschini M, Spahn M, Ma ei A, Che ille J, Ka nes RJ. Inco - po a ion o issue-based genomic bioma ke s in o localized p os- a e cance clinics. BMC Med. 2016;14:67. 5. Ross AE, D’Amico AV, F eedland SJ. Which, when and why? Ra ional use o issue-based molecula es ing in localized p os a e cance . P os a e Cance P os a ic Dis. 2016;19:1–6. 6. Lin DW, Fi zGe ald LM, Fu R, Kwon EM, Zheng SL, Kolb S, e al. Gene ic a ian s in he LEPR, CRY1, RNASEL, IL4, and ARVCF genes a e p ognos ic ma ke s o p os a e cance -specific mo ali y. Cance Epidemiol Bioma k P e . 2011;20:1928–36. 7. Penney KL, Shui IM, Feng Z, Sesso HD, S amp e MJ, S an o d JL. Replica ion o a gene ic a ian o p os a e cance -specific mo ali y. P os a e Cance P os a ic Dis. 2015;18:260–3. 8. Ka yadi DM, Zhao S, He Q, McIn osh L, W igh JL, Os ande EA, e al. Confi ma ion o gene ic a ian s associa ed wi h le hal p os a e cance in a coho o men om he edi a y p os a e cance amilies. In J Cance . 2015;136:2166–71. 9. Giles G, English D. The Melbou ne Collabo a i e Coho S udy. IARC Sci Publ. 2002;156:69–70. 10. Eeles RA, Olama AA, Benlloch S, Saunde s EJ, Leongamo nle DA, Tym akiewicz M, e al. Iden ifica ion o 23 new p os a e cance suscep ibili y loci using he iCOGS cus om geno yping a ay. Na Gene . 2013;45:385–91. 391e381–2 11. Lai inen VH, Wahl o s T, Saa is o L, Ran ape o T, Pel a i LM, Kilpi aa a O, e al. HOXB13 G84E mu a ion in Finland: popula ion-based analysis o p os a e, b eas , and colo ec al cance isk. Cance Epidemiol Bioma k P e . 2013;22:452–60. 12. Schleu ke J, Ma ikainen M, Smi h J, Koi is o P, Ba oe-Bonnie A, Kainu T, e al. A gene ic epidemiological s udy o he edi a y p os a e cance (HPC) in Finland: equen HPCX linkage in amilies wi h la e-onse disease. Clin Cance Res. 2000;6:4810–5. 13. Eeles RA. Gene ic p edisposi ion o p os a e cance . P os a e Cance P os a ic Dis. 1999;2:9–15. 14. S an o d JL, Fi zGe ald LM, McDonnell SK, Ca lson EE, McIn- osh LM, Deu sch K, e al. Dense genome-wide SNP linkage scan in 301 he edi a y p os a e cance amilies iden ifies mul iple egions wi h sugges i e e idence o linkage. Hum Mol Gene . 2009;18:1839–48. 15. Jane M, F ied ichsen DM, S an o d JL, Badzioch MD, Kolb S, Deu sch K, e al. Genomic scan o 254 he edi a y p os a e cance amilies. P os a e. 2003;57:309–19. 16. Van Hemel ijck M, Wige z A, Sandin F, Ga mo H, Hells om K, F ansson P, e al. Coho p ofile: he Na ional P os a e Cance Regis e o Sweden and P os a e Cance da a Base Sweden 2.0. In J Epidemiol. 2013;42:956–67. 17. Wo ld Heal h O ganisa ion. Manual o he in e na ional s a is ical classifica ion o diseases, inju ies, and causes o dea h. Gene a: Wo ld Heal h O ganisa ion; 1977. 18. S ee ing Commi ee o he Physicians’Heal h S udy Resea ch G oup. Final epo on he aspi in componen o he ongoing Physicians’Heal h S udy. N Engl J Med. 1989;321:129–35. 19. Penney KL, Schumache FR, Li H, K a P, Mo is JS, Ku h T, e al. A la ge p ospec i e s udy o SEP15 gene ic a ia ion, in e ac ion wi h plasma selenium le els, and p os a e cance isk and su i al. Cance P e Res. 2010;3:604–10. 20. Bauja B, Mahe C, Pignon JP, Hill C. A g aphical me hod o explo ing he e ogenei y in me a-analyses: applica ion o a me a- analysis o 65 ials. S a Med. 2002;21:2641–52. Inhe i ed p edisposi ion o agg essi e p os a e cance 235 21. Agalliu I, Kwon EM, Salinas CA, Koopmeine s JS, Os ande EA, S an o d JL. Gene ic a ia ion in DNA epai genes and p os a e cance isk: esul s om a popula ion-based s udy. Cance Causes Con ol. 2010;21:289–300. 22. Kwon EM, Salinas CA, Kolb S, Fu R, Feng Z, S an o d JL, e al. Gene ic polymo phisms in inflamma ion pa hway genes and p os a e cance isk. Cance Epidemiol Bioma k P e . 2011;20:923–33. 23. Loh YH, Mi ou PN, Bowman R, Wood A, Je e y H, Luben RN, e al. MGMT Ile143Val polymo phism, die a y ac o s and he isk o b eas , colo ec al and p os a e cance in he Eu opean P ospec i e In es iga ion in o Cance and Nu i ion (EPIC)-No - olk s udy. DNA Repai . 2010;9:421–8. 24. Ri chey JD, Huang WY, Chokkalingam AP, Gao YT, Deng J, Le ine P, e al. Gene ic a ian s o DNA epai genes and p os a e cance : a popula ion-based s udy. Cance Epidemiol Bioma k P e . 2005;14:1703–9. 25. Winches e DA, Till C, Goodman PJ, Tangen CM, San ella RM, Johnson-Pais TL, e al. Associa ion be ween a ian s in genes in ol ed in he immune esponse and p os a e cance isk in men andomized o he finas e ide a m in he P os a e Cance P e en- ion T ial. P os a e. 2017;77:908–19. 26. De ance T, Vanbe lie B, Aub y JP, Takebe Y, A ai N, Miya- jima A, e al. B cell g ow h-p omo ing ac i i y o ecombinan human in e leukin 4. J Immunol. 1987;139:1135–41. 27. Mosmann TR, Bond MW, Co man RL, Oha a J, Paul WE. T-cell and mas cell lines espond o B-cell s imula o y ac o 1. P oc Na l Acad Sci USA. 1986;83:5654–8. 28. Toi M, Ha is AL, Bicknell R. In e leukin-4 is a po en mi ogen o capilla y endo helium. Biochem Biophys Res Commun. 1991;174:1287–93. 29. Lahm H, Schnyde B, Wynige J, Bo benyi Z, Yilmaz A, Ca BD, e al. G ow h inhibi ion o human colo ec al-ca cinoma cells by in e leukin-4 and exp ession o unc ional in e leukin-4 ecep o s. In J Cance . 1994;59:440–7. 30. Volpe OV, Fong T, Koch AE, Pe e son JD, Wal enbaugh C, Teppe RI, e al. Inhibi ion o angiogenesis by in e leukin 4. J Exp Med. 1998;188:1039–46. 31. Halle MA, Venma KT, Fingle on B. Cy okine s imula ion o epi helial cance cells: he simila and di e gen unc ions o IL-4 and IL-13. Cance Res. 2012;72:6338–43. 32. DeNa do DG, Ba e o JB, And eu P, Vasquez L, TawfikD, Kolha ka N, e al. CD4(+) T cells egula e pulmona y me as asis o mamma y ca cinomas by enhancing p o umo p ope ies o mac ophages. Cance Cell. 2009;16:91–102. 33. Venma KT, Kimmel DW, Cli el DE, Fingle on B. IL4 ecep o alpha media es enhanced glucose and glu amine me abolism o suppo b eas cance g ow h. Biochim Biophys Ac a. 2015;1853:1219–28. 34. Wise GJ, Ma ella VK, Tallu i G, Shi azian D. Cy okine a ia ions in pa ien s wi h ho mone ea ed p os a e cance . J U ol. 2000;164:722–5. 35. Lee SO, Lou W, Hou M, Ona e SA, Gao AC. In e leukin-4 enhances p os a e-specific an igen exp ession by ac i a ion o he and ogen ecep o and Ak pa hway. Oncogene. 2003;22:7981–8. 36. Lee SO, Pinde E, Chun JY, Lou W, Sun M, Gao AC. In e leukin- 4 s imula es and ogen-independen g ow h in LNCaP human p os a e cance cells. P os a e. 2008;68:85–91. 37. Bankai is KV, Fingle on B. Ta ge ing IL4/IL4R o he ea men o epi helial cance me as asis. Clin Exp Me as asis. 2015;32:847–56. 38. Husain SR, Kawakami K, Kawakami M, Pu i RK. In e leukin-4 ecep o - a ge ed cy o oxin he apy o and ogen-dependen and -independen p os a e ca cinoma in xenog a models. Mol Cance The . 2003;2:245–54. 39. Ma gison GP, San ibanez Ko e MF, Po ey AC. Mechanisms o ca cinogenici y/chemo he apy by O6-me hylguanine. Mu agen- esis. 2002;17:483–7. 40. Xu-Welli e M, Pegg AE. Deg ada ion o he alkyla ed o m o he DNA epai p o ein, O(6)-alkylguanine-DNA alkyl ans e ase. Ca cinogenesis. 2002;23:823–30. 41. Janssen K, Eichho n-G ombache U, Schlink K, Ni zsche S, Oesch F, Kaina B. Long- ime exp ession o DNA epai enzymes MGMT and APE in human pe iphe al blood mononuclea cells. A ch Toxicol. 2001;75:306–12. 42. Ma gison GP, Heighway J, Pea son S, McGown G, Tho nc o MR, Wa son AJ, e al. Quan i a i e ai locus analysis e eals wo in agenic si es ha influence O6-alkylguanine-DNA alkyl- ans e ase ac i i y in pe iphe al blood mononuclea cells. Ca ci- nogenesis. 2005;26:1473–80. 43. Mijal RS, Kanugula S, Vu CC, Fang Q, Pegg AE, Pe e son LA. DNA sequence con ex a ec s epai o he obacco-specific adduc O(6)-[4-Oxo-4-(3-py idyl)bu yl]guanine by human O(6)- alkylguanine-DNA alkyl ans e ases. Cance Res. 2006;66: 4968–74. 44. Wa son AJ, Sabha wal A, Tho nc o M, McGown G, Ke R, Bojanic S, e al. Tumo O(6)-me hylguanine-DNA me hyl- ans e ase inac i a ion by o al lomegua ib. Clin Cance Res. 2010;16:743–9. 45. Cheung M, Tes a JR. Di e se mechanisms o AKT pa hway ac i a ion in human malignancy. Cu Cance D ug Ta ge s. 2013;13:234–44. 46. Pal I, Mandal M. PI3K and Ak as molecula a ge s o cance he apy: cu en clinical ou comes. Ac a Pha macol Sin. 2012;33:1441–58. 47. La ue L, Bellacosa A. Epi helial-mesenchymal ansi ion in de elopmen and cance : ole o phospha idylinosi ol 3’kinase/ AKT pa hways. Oncogene. 2005;24:7443–54. 48. Taylo BS, Schul z N, Hie onymus H, Gopalan A, Xiao Y, Ca e BS, e al. In eg a i e genomic p ofiling o human p os a e cance . Cance Cell. 2010;18:11–22. 49. Da ies MA, Koul D, Dhesi H, Be man R, McDonnell TJ, McConkey D, e al. Regula ion o Ak /PKB ac i i y, cellula g ow h, and apop osis in p os a e ca cinoma cells by MMAC/ PTEN. Cance Res. 1999;59:2551–6. 50. S ambolic V, Suzuki A, de la Pompa JL, B o he s GM, Mi sos C, Sasaki T, e al. Nega i e egula ion o PKB/Ak -dependen cell su i al by he umo supp esso PTEN. Cell. 1998;95: 29–39. 51. Sun H, Lesche R, Li DM, Lilien al J, Zhang H, Gao J, e al. PTEN modula es cell cycle p og ession and cell su i al by egula ing phospha idylinosi ol 3,4,5,- isphospha e and Ak /p o ein kinase B signaling pa hway. P oc Na l Acad Sci USA. 1999;96:6199–204. 52. McCub ey JA, S eelman LS, Kemp CR, Chappell WH, Ab ams SL, S i ala F, e al. The apeu ic esis ance esul ing om mu a- ions in Ra /MEK/ERK and PI3K/PTEN/Ak /mTOR signaling pa hways. J Cell Physiol. 2011;226:2762–81. 53. Bi ing RL, A ms ong AJ. Ta ge ing he PI3K/Ak /mTOR pa h- way in cas a ion- esis an p os a e cance . Endoc Rela Cance . 2013;20:R83–99. 54. Cou ney KD, Co co an RB, Engelman JA. The PI3K pa hway as d ug a ge in human cance . J Clin Oncol. 2010;28:1075–83. 55. Ni ulescu GM, Ma gina D, Juzenas P, Peng Q, Ola u OT, Sal- ous os E, e al. Ak inhibi o s in cance ea men : The long jou ney om d ug disco e y o clinical use (Re iew). In J Oncol. 2016;48:869–85. 56. DeSan is CE, Siegel RL, Saue AG, Mille KD, Fedewa SA, Alca az KI, e al. Cance s a is ics o A ican Ame icans, 2016: p og ess and oppo uni ies in educing acial dispa i ies. CA Cance J Clin. 2016;66:290–308. 236 L. M. Fi zGe ald e al.