scieee Open visual document viewer

Genome-wide association study of bronchopulmonary dysplasia: a potential role for variants near the CRP gene

Mahlman, Mari,Karjalainen, Minna,Huusko, Johanna,Andersson, Sture,Kari, M Anneli,Tammela, Outi,Sankilampi, Ulla,Lehtonen, Liisa,Marttila, Riitta,Basler, Dirk,Poets, Christian,Lacaze-Masmonteil, Thierry,Danan, Claude,Delacourt, Christophe,Palotie, Aarne,M

Full text

1 Scien i ic RepoR s | 7: 9271 | DOI:10.1038/s41598-017-08977-w www.na u e.com/scien i ic epo s Genome-wide associa ion s udy o b onchopulmona y dysplasia: a po en ial ole o a ian s nea he CRP gene Ma i Mahlman1,2, Minna K. Ka jalainen1,2, Johanna M. Huusko1,2,21, S u e Ande sson3, M. Anneli Ka i3, Ou i K. T. Tammela4, Ulla Sankilampi5, Liisa Leh onen6, Rii a H. Ma ila1,2, Di k Bassle 7, Ch is ian F. Poe s8, Thie y Lacaze-Masmon eil9, Claude Danan10,11,12, Ch is ophe Delacou 10,13,14, Aa no Palo ie15,16,17,18,19,20, Louis J. Muglia 21, Pascal M. La oie22, Alice Hadchouel10,13,14, Mika Räme 1,2,23 & Mikko Hallman1,2 B onchopulmona y dysplasia (BPD), he main consequence o p ema u i y, has a signi ican he i abili y, bu li le is known abou p edisposing genes. The aim o his s udy was o iden i y gene loci p edisposing in an s o BPD. The ini ial genome-wide associa ion s udy (GWAS) included 174 Finnish p e e m in an s o ges a ional age 24–30 weeks. The ea e , he mos p omising single-nucleo ide polymo phisms (SNPs) associa ed wi h BPD we e geno yped in bo h Finnish (n = 555) and non-Finnish (n = 388) eplica ion coho s. Finally, plasma CRP le els om he i s week o li e and he isk o BPD we e assessed. SNP s11265269, lanking he CRP gene, showed he s onges signal in GWAS (odds a io [OR] 3.2, p = 3.4 × 10−6). This associa ion was nominally eplica ed in Finnish and F ench A ican popula ions. A numbe o o he SNPs in he CRP egion, including s3093059, had nominal associa ions wi h BPD. Du ing he i s week o li e he ele a ed plasma le els o CRP p edic ed he isk o BPD (OR 3.4, p = 2.9 × 10–4) and he SNP s3093059 associa ed nominally wi h plasma CRP le els. Finally, SNP s11265269 was iden i ied as a isk ac o o BPD (OR 1.8, p = 5.3 × 10−5), independen ly o he obus an ena al isk ac o s. As such, in BPD, a po en ial ole o a ian s nea CRP gene is p oposed. 1PEDEGO Resea ch Uni , Medical Resea ch Cen e Oulu, Uni e si y o Oulu, Oulu, Finland. 2Depa men o Child en and Adolescen s, Oulu Uni e si y Hospi al, Oulu, Finland. 3Child en’s Hospi al, Uni e si y o Helsinki, and Helsinki Uni e si y Hospi al, Helsinki, Finland. 4Tampe e Uni e si y Hospi al, Tampe e Uni e si y, and Cen e o Pedia ic Child Heal h, Tampe e, Finland. 5Depa men o Pedia ics, Kuopio Uni e si y Hospi al, Kuopio, Finland. 6Tu ku Uni e si y Hospi al, and he Uni e si y o Tu ku, Tu ku, Finland. 7Depa men o Neona ology, Uni e si y Hospi al Zu ich, and Uni e si y o Zu ich, Zu ich, Swi ze land. 8Depa men o Neona ology, Tuebingen Uni e si y Hospi al, Tuebingen, Ge many. 9Depa men o Paedia ics, Cumming School o Medicine, Uni e si y o Calga y, Albe a, Canada. 10Inse m, U955, C é eil, F ance. 11CRB, CHI-C e eil, F ance. 12Depa men o neona ology, CHI-C e eil, C e eil, F ance. 13AP-HP, Hôpi al Necke -En an s Malades, Se ice de Pneumologie Pédia ique, Pa is, F ance. 14Uni e si é Pa is-Desca es, Pa is, F ance. 15Analy ic and T ansla ional Gene ics Uni , Depa men o Medicine, Massachuse s Gene al Hospi al, Bos on, MA, USA. 16P og am in Medical and Popula ion Gene ics, The B oad Ins i u e o MIT and Ha a d, Camb idge, MA, USA. 17The S anley Cen e o Psychia ic Resea ch, The B oad Ins i u e o MIT and Ha a d, Camb idge, MA, USA. 18Ins i u e o Molecula Medicine Finland, Uni e si y o Helsinki, Helsinki, Finland. 19Psychia ic & Neu ode elopmen al Gene ics Uni , Depa men o Psychia y, Massachuse s Gene al Hospi al, Bos on, MA, USA. 20Depa men o Neu ology, Massachuse s Gene al Hospi al, Bos on, MA, USA. 21Pe ina al Ins i u e, Cincinna i Child en’s Hospi al Medical Cen e and Depa men o Pedia ics, Uni e si y o Cincinna i College o Medicine, Cincinna i, OH, USA. 22BC Child en’s Hospi al Resea ch Ins i u e, Vancou e Canada, Vancou e , Canada. 23BioMediTech Ins i u e and Facul y o Medical and Li e Sciences, Uni e si y o Tampe e, Tampe e, Finland. Ma i Mahlman and Minna K. Ka jalainen con ibu ed equally o his wo k. Mika Räme and Mikko Hallman join ly supe ised his wo k. Co espondence and eques s o ma e ials should be add essed o M.M. (email: [email p o ec ed]) Recei ed: 8 Feb ua y 2017 Accep ed: 20 July 2017 Published: xx xx xxxx OPEN www.na u e.com/scien i ic epo s/ 2 Scien i ic RepoR s | 7: 9271 | DOI:10.1038/s41598-017-08977-w B onchopulmona y dysplasia (BPD) is a majo complica ion o p ema u i y. In he USA alone, abou 10,000– 15,000 new cases a e diagnosed each yea 1. The po en ial ad e se consequences o BPD include as hma and ch onic obs uc i e pulmona y disease (COPD)2–4. E ec i e p e en ion o BPD does no exis , and e en wi h mode n s a e-o - he-a he apies he incidence has no allen1. In BPD, poo al eola iza ion and dis up ed pulmona y ascula iza ion lead o impai ed gas exchange, clini- cally seen as p olonged need o espi a o y suppo 5. In lamma ion plays a key ole in he pa hogenesis o BPD, bu he molecula mechanisms emain unknown6, 7. Acco ding o s udies in wins, gene ic ac o s accoun o 53–82% o he a iance in suscep ibili y o BPD8, 9. Howe e , iden i ica ion o p edisposing genes has been chal- lenging. Al hough a sugges i e associa ion be ween SPOCK2 gene and BPD was disco e ed in he i s pub- lished genome-wide associa ion s udy (GWAS) on BPD10, wo subsequen GWASs did no e eal any signi ican associa ions a he genome-wide le el11, 12. Mo eo e , candida e gene s udies ha e been la gely unsuccess ul a p oducing eplicable esul s13, 14. One p obable eason o his “missing he i abili y” is he he e ogenei y o s udy popula ions. In Hadchouel e al.’s s udy, he in an s we e o ei he Caucasian o A ican o igin. Wang e al.’s pop- ula ion consis ed o ou iden i ied e hnic g oups, and Ambala anan e al.’s s udy included bo h Caucasians and A ican-Ame icans. Di e en e hnic g oups di e in hei allele equencies, hus complica ing gene ic analyses. The popula ion o Finland is gene ically ela i ely homogeneous and has he e o e been used ex ensi ely in gene ic s udies15, 16. Fu he mo e, neona al p ac ices a y li le in Finland due o s anda dized ea men p o- ocols. This is o impo ance when s udying a ai such as BPD, he incidence o which is a ec ed by neona al ea men p ac ices. In he p esen in es iga ion, we conduc ed a GWAS on BPD in a Finnish popula ion. To make he esul s mo e gene alizable, we s udied wo eplica ion popula ions om Finland and wo popula ions om Canada and F ance. We iden i ied a single nucleo ide polymo phism (SNP), s11265269, as an independen isk ac o o BPD. Me hods S udy design and s udy popula ions. A low cha o he s udy is shown in Fig.1. The o al numbe s o BPD cases and con ols in he en i e s udy we e 319 and 798, espec i ely. Inclusion c i e ia o all in an s included in he s udy we e (1) ges a ional age (GA) <31 wk and (2) no majo congeni al mal o ma ions. The disco e y popula ion used o he GWAS (s ep 1 o he s udy, n = 174) and he wo in e nal eplica ion popula ions ( i s in e nal eplica ion, s ep 2 o he s udy, n = 326 and second in e nal eplica- ion, s ep 4 o he s udy, n = 229) consis ed o Finnish in an s collec ed p ospec i ely a he i e uni e si y hospi- als in Finland. The ex e nal eplica ion popula ions (s ep 3 o he s udy, n = 388) comp ised in an s om Canada and F ance. To augmen he powe o he GWAS s udy, wo con ol in an s we e chosen o each BPD case. Only one in an om each sibling pai was selec ed. Fo de ails o pa ien selec ion, see Supplemen a y In o ma ion. Clinical cha ac e is ics o he popula ion in he GWAS a e shown in Table1, and hose o he eplica ion popu- la ions a e shown in Supplemen a y TablesS1 and S2. The s udy was app o ed by he E hics Commi ee o Oulu Uni e si y Hospi al, he Uni e si y o B i ish Columbia Clinical Resea ch E hics Boa d, he Uni e si y o Albe a E hics Boa d, and he Comi és de P o ec ion des Pe sonnes (CPP) Ile de F ance VI and IX. W i en in o med consen was ob ained om he in an s’ pa en s. All expe imen s we e pe o med in acco dance wi h ele an guidelines and egula ions. Diagnos ic c i e ia o BPD. BPD was de ined as a con inuing need o supplemen al oxygen o posi i e p essu e suppo (including high- low 5–6 li e pe minu e ia nasal p obes) a 36 wks. pos mens ual age (PMA); i.e., mode a e- o-se e e BPD acco ding o Na ional Ins i u e o Child Heal h and Human De elopmen (NICHD) c i e ia17. This c i e ion was based on win s udies, in which mode a e- o-se e e BPD is signi ican ly a ec ed by gene ic ac o s whe eas mild BPD is no 9. The con ols we e in an s wi hou BPD o wi h mild BPD. Acco ding o NICHD c i e ia, mild BPD is de ined as a need o supplemen a y oxygen o posi i e p essu e suppo a he age o 28 days bu no longe a he age o 36 wks. PMA. No BPD is de ined as no need o supplemen a y oxygen o posi i e p essu e suppo a he age o 28 days. Figu e 1. Flow cha o he s udy. www.na u e.com/scien i ic epo s/ 3 Scien i ic RepoR s | 7: 9271 | DOI:10.1038/s41598-017-08977-w S a is ical analyses and ou line o o he me hods. Speci ic de ails o s a is ical analyses, geno yping, da a p ocessing, impu a ion and selec ion o SNPs o eplica ion analyses, and analyses o CRP alues, a e shown in Supplemen a y In o ma ion. In o ma ion abou DNA sample p epa a ion is shown in Supplemen a y TableS3. Genome-wide SNP geno yping was pe o med wi h he In inium HumanCo eExome BeadChip (Illumina, San Diego, CA, USA). GWAS was used o iden i y SNPs associa ed wi h mode a e- o-se e e BPD; da a p ocessing and s a is ical analyses we e pe o med wi h PLINK, . 1.07 o 1.0918. A e quali y con ol, 276,306 SNPs wi h mino allele equency >0.01 emained o GWAS. We used a il e ing s a egy o na ow down he SNPs ana- lyzed o associa ion wi h BPD in each s ep (Fig.1): he mos p omising GWAS SNPs we e analyzed in he in e - nal Finnish eplica e, and he mos consis en SNPs we e u he analyzed in wo ex e nal popula ions (Caucasian and F ench A ican). A he inal s ep, ou SNPs we e analyzed in he second in e nal Finnish eplica e. Logis ic eg ession unde he addi i e model wi h GA as a co a ia e was used o assess associa ions wi h BPD in hese popula ions. The SNP selec ion c i e ia a e lis ed in he Supplemen a y In o ma ion. A he inal s age, we also included small- o -ges a ional age (SGA, Z-sco e ≤−2 SD) as a co a ia e in a logis ic eg ession model o assess how he signi ican an ena al ac o s oge he (GA and SGA) wi h he SNP wi h mos consis en signal a ec he isk o BPD. Because he SNP wi h mos consis en signal was loca ed nea he CRP gene, we analyzed plasma CRP le els o associa ion wi h BPD and SNP geno ypes. Daily CRP le els we e a ailable because o he cu en p ac ice in sc eening o in ec ion du ing he i s week a e e y p ema u e bi h. The in an s included in he plasma CRP s udies we e a subse o he popula ions analyzed in he GWAS and he i s in e nal eplica e; in an s o whom CRP le els could be eco ded we e included (n = 275; 112 in an s om GWAS, 163 in an s om i s in e nal eplica e). Logis ic eg ession was used o e alua e whe he SNP geno ypes we e associa ed wi h maximum and mean CRP le el (CRP alues abo e/below median) du ing he i s week o li e wi h PLINK, . 1.0918. The num- be o su ac an doses was included as a co a ia e, because his a iable associa ed wi h CRP le els (p < 0.001). Associa ion o he BPD incidence o SNP geno ypes wi h CRP le els was assessed by he χ2 o K uskal–Wallis es wi h SPSS S a is ics 20.0, IBM Co po a ion. A he inal phase, logis ic eg ession wi h SPSS was used o e alua e whe he CRP le els a e signi ican p edic o s o BPD using GA and SGA as co a ia es. Resul s Associa ions in GWAS. The GWAS was pe o med in a disco e y coho o 60 cases wi h mode a e- o-se- e e BPD and 114 con ols. GA, gende , p opo ion o single ons, exposu e o an ena al o pos na al s e oids o o clinical cho ioamnioni is did no di e be ween he cases and con ols. As expec ed19, 20, bo h bi h weigh and bi h-weigh Z-sco e we e lowe in BPD cases compa ed wi h con ols, sugges ing a lowe oe al g ow h a e among BPD in an s. P e-eclampsia was mo e common in he BPD g oup (Table1). In he GWAS, none o he SNPs eached ou s ingen genome-wide signi icance le el (p < 1.8 × 10−7), bu he e we e many sugges i e associa ions (Supplemen a y Figu eS1). Resul s o he SNPs wi h he s onges sugges i e signals (p < 1 × 10−4) a e shown in Table2. The mos p omising SNP was s11265269 (odds a io [OR] 3.22, p = 3.43 × 10−6), which is loca ed be ween he CRP and DUSP23 genes (app oxima ely 44 and 23 kb ups eam o each gene, espec i ely). Popula ion s a i ica ion was minimal (Supplemen a y Figu eS2). The impu ed SNPs did no show any associa ion signals ha we e clea ly mo e signi ican han hose o he geno yped SNPs (Supplemen a y Figu eS3). The SNPs wi h p < 5 × 10−4 in he GWAS we e selec ed o u he geno yping, and a e lis ed in Supplemen a y TableS4. Associa ion o s11265269 wi h BPD in eplica ion popula ions. Fi s , we chose 155 SNPs o analysis in he 1s in e nal eplica ion popula ion (105 cases and 221 con ols) (Supplemen a y TableS1). These SNPs we e selec ed on he basis o he s onges signals in GWAS; in addi ion, p e iously BPD-associa ed SNPs Cha ac e is ics BPD cases Con ols p To al n60 114 Mode a e/se e e BPD, n (%) 33/27 (55.0/45.0) No BPD/mild BPD, n (%) 64/50 (56.1/43.9) GA, weeks*,† 27.1 ± 1.8 (24.1–30.9) 27.5 ± 1.7 (23.3–30.9) 0.114 GA < 28 wk, n (%) 42 (70.0) 66 (57.9) 0.118 Bi h weigh , g ams*865 ± 244 (440–1470) 1017 ± 264 (520–1695) 2.85 × 10–4 Bi h weigh Z-sco e*‡−1.5 ± 1.34 (−4.0–1.1) −0.9 ± 1.20 (−4.1–1.6) 3.00 × 10−3 SGA, n (%) 23 (38.3) 20 (17.5) 3.00 × 10−3 Male gende , n (%) 32 (53.3) 57 (50.0) 0.676 Single ons, n (%) 57 (95.0) 105 (92.1) 0.474 An ena al s e oids, n (%) 57 (95.0) 108 (94.7) 0.941 P eeclampsia, n (%) 23 (38.3) 24 (21.1) 0.046 Cho ioamnioni is§15 (26.8) 38 (34.5) 0.311 Table 1. Clinical cha ac e is ics o he s udy popula ion in he genome-wide associa ion s udy o b onchopulmona y dysplasia. De ini ion o abb e ia ions: BPD, b onchopulmona y dysplasia; GA, ges a ional age; SD, s anda d de ia ion; SGA, small o ges a ional age (Z-sco e ≤−2 SD). *Mean ± s anda d de ia ion ( ange). †GA de ined on he basis o oe al ul asound be o e 15 weeks o p egnancy. ‡Bi hweigh Z-sco e desc ibes dis ibu ion o bi hweigh a gi en leng h o ges a ion in SD. §Da a no a ailable o eigh in an s. www.na u e.com/scien i ic epo s/ 4 Scien i ic RepoR s | 7: 9271 | DOI:10.1038/s41598-017-08977-w wi h consis en signals we e included. The selec ion c i e ia a e gi en in de ail in Supplemen a y In o ma ion. The ea e , he op 41 SNPs, based on GWAS esul s and 1s in e nal eplica ion, we e u he analysed in he ex e nal eplica ion popula ions (Caucasian, n = 312, and F ench A ican, n = 76) (Supplemen a y TableS2 and Supplemen a y In o ma ion). Finally, o inc ease powe and o alida e he mos p omising associa ions, SNPs s11265269 ( he op SNP iden i ied by GWAS, nea CRP/DUSP23), s1889268 (COL15A1) and s5999125 (LARGE) we e geno yped in a second in e nal eplica ion popula ion (Finnish, n = 229). O all he SNPs ana- lysed in he ex e nal eplica ion popula ions, hese h ee SNPs showed he mos consis en esul s and we e hus selec ed o his inal geno yping s ep. Because GA was signi ican ly di e en be ween cases and con ols in he eplica ion popula ions, he analyses we e pe o med using logis ic eg ession (LR) analysis wi h GA as a co a ia e. In he i s in e nal eplica e (Supplemen a y TableS4), only one SNP, s6690148, showed nominal associa- ion wi h BPD (p < 0.05). The op SNP in GWAS, s11265269 showed an e ec simila o ha de ec ed in GWAS (OR 1.42, p = 0.097). O he SNPs analysed u he , wo showed nominal associa ion in he ex e nal eplica ion popula ions: s11265269 in he F ench A ican popula ion and COL15A1 SNP ( s1889268) in he Caucasian pop- ula ion (Supplemen a y TableS5). O e all, SNP s11265269 ( he SNP wi h he s onges signal in GWAS) was he mos consis en SNP. Table3 summa izes he esul s o his SNP in all he eplica ion popula ions. In addi ion o he nominal eplica ion in he F ench A ican popula ion, his SNP associa ed wi h BPD when he in e nal eplica ion popula ions we e combined (p = 0.029). Wi h all o he Finnish popula ions combined, he OR o he mino allele o SNP s11265269 was 1.84 (p = 2.39 × 10−5) (Table3); i.e., he e ec was smalle han in he ini ial GWAS, which may be due o o e es ima ed e ec size in GWAS. The case–con ol allele equency di e ences o he COL15A1 SNP ( s1889268) we e also consis en in all Caucasian popula ions, al hough he di e ence was no signi ican in he in e nal eplica ion popula ions (Supplemen a y TablesS5 and S6). The esul s o he LARGE SNP in o ma ion GWAS SNP*Ch Posi ion†Gene‡Case/Con ol mino allele equency OR p s11265269 1 159,728,127 CRP, DUSP23 0.392/0.167 3.22 3.43 × 10−6 s1481294 11 38,604,075 LOC103312105, LOC105376635 0.325/0.575 0.36 9.56 × 10−6 s2351857 7 137,467,829 DGKI 0.617/0.373 2.71 1.42 × 10−5 s11691168 2 74,999,114 LOC102724482 0.475/0.250 2.71 2.13 × 10−5 s2149564 9 98,607,989 LINC00476 0.642/0.404 2.65 2.39 × 10−5 s6562965 13 77,351,486 LOC105370265 0.442/0.224 2.75 2.42 × 10−5 s11745686 5 74,198,611 LOC105379039, FAM169A 0.400/0.193 2.79 3.15 × 10−5 s1403617 3 165,255,278 LINC01322, BCHE 0.475/0.254 2.65 3.20 × 10−5 s12788032 11 38,632,770 LOC103312105, LOC105376635 0.275/0.504 0.37 3.89 × 10−5 s1822471 15 79,327,227 RASGRF1 0.092/0.281 0.26 4.57 × 10−5 s9552800 13 23,599,673 LOC105370111, SGCG 0.242/0.083 3.51 4.67 × 10−5 s17537018 5 155,458,803 SGCD, LOC105377674 0.367/0.171 2.81 4.70 × 10−5 s2527506 7 2,968,361 CARD11 0.325/0.140 2.95 4.85 × 10−5 s4704970 5 155,500,992 SGCD, LOC105377674 0.358/0.167 2.79 5.78 × 10−5 s1358603 7 52,757,480 LOC105375280, LOC101928257 0.617/0.390 2.51 5.78 × 10−5 s4506388 23 130,210,648 ARHGAP36 0.534/0.281 2.94 6.13 × 10−5 s9979500 21 48,029,698 S100B, PRMT2 0.450/0.241 2.57 6.57 × 10−5 s4640066 13 77,354,747 LOC105370265 0.475/0.263 2.53 7.00 × 10−5 s2352931 16 86,169,068 LOC105376778, LOC101928582 0.200/0.412 0.36 7.01 × 10−5 s12603672 17 51,115,243 LOC105371831, LOC101927337 0.117/0.018 7.40 7.23 × 10−5 s2279073 19 44,739,303 ZNF227 0.325/0.548 0.40 7.32 × 10−5 s1044189 12 7,053,149 RNU7-1 0.383/0.189 2.67 7.57 × 10−5 s7934284 11 38,577,786 LOC103312105, LOC105376635 0.608/0.386 2.47 7.67 × 10−5 s200642524 8 10,470,709 RP1L1 0.067/0.000 — 8.00 × 10−5 s11200206 10 123,635,883 ATE1 0.442/0.237 2.55 8.29 × 10−5 s2543361 14 76,593,690 GPATCHL2L, IFT43 0.575/0.355 2.46 8.34 × 10−5 s314277 6 105,407,662 LIN28B 0.275/0.110 3.08 8.35 × 10−5 s4583363 18 65,162,006 DSEL, LOC643542 0.408/0.211 2.59 9.16 × 10−5 s11178156 12 70,635,747 LINC01481, CNOT2 0.258/0.474 0.39 9.67 × 10−5 Table 2. Single-nucleo ide polymo phisms showing sugges i e associa ion signals in he genome-wide associa ion s udy o b onchopulmona y dysplasia. De ini ion o abb e ia ions: BPD, b onchopulmona y dysplasia; Ch , ch omosome; GWAS, genome-wide associa ion s udy; OR, odds a io; SNP, single-nucleo ide polymo phism. *SNPs wi h p < 1 × 10−4 in GWAS a e shown. †Ch omosomal posi ions e e o human genome build 37 (GRCh37/hg19). ‡Respec i e locus shown o SNPs wi hin genes; wo nea es loci shown o in e genic SNPs. www.na u e.com/scien i ic epo s/ 5 Scien i ic RepoR s | 7: 9271 | DOI:10.1038/s41598-017-08977-w SNP ( s5999125) we e mo e inconsis en (Supplemen a y TablesS5 and S6). The esul s o he o he SNPs ana- lysed in he eplica ion popula ions a e lis ed in Supplemen a y TablesS4 and S5. We eanalysed he h ee SNPs wi h he s onges signals ( s11265269, s1889268 and s5999125) in he Finnish popula ions by excluding he mild BPD in an s om he con ols. This exclusion did no no ably al e he esul s (Supplemen a y TableS7). Fo s11265269 he OR was 3.28 in GWAS (p = 5.94 × 10−5) and 1.79 in he join anal- ysis o all Finnish popula ions (p = 1.41 × 10−3). Al e ed ea men p ac ices o changes in he diagnos ics o BPD (Supplemen a y In o ma ion) could also in luence he esul s. The e o e, he gene ic da a was eanalysed sepa- a ely o in an s bo n du ing 1998–2010 and a e 2010. Du ing bo h ime pe iods he associa ion o s11265269 wi h he isk o BPD was main ained in GWAS (OR 4.78 and 2.37, espec i ely) and in join analysis o all Finnish popula ions (OR 1.87 and 1.71, espec i ely). SNPs p e iously associa ed wi h BPD: no clea associa ion signals. Nex , we analysed, i s in ou GWAS da a, and he ea e he mos p omising SNPs in he in e nal eplica ion popula ion, he SNPs in p oximi y o genes ha we e p e iously ound o be associa ed wi h BPD10, 11. Some o hese SNPs showed a nominal associa ion in he GWAS, bu none we e associa ed wi h BPD in he i s in e nal eplica ion popu- la ion (Supplemen a y TablesS8 and S9). Consis en ly, none o he SNPs in p e iously-associa ed genes ha we e analysed u he associa ed wi h BPD in he Caucasian and F ench A ican ex e nal eplica ion popula- ions (Supplemen a y TableS10). O e all, SNP s2536512 in SOD3 was he mos p omising. This SNP associa ed nominally wi h BPD in GWAS (OR = 0.53, p = 0.017), and showed a simila e ec in he i s in e nal epli- ca e (OR = 0.75, p = 0.173) and ex e nal Caucasian popula ion (OR = 0.70, p = 0.072). SNP s2536512 eached p < 0.05 when he Caucasian popula ion was combined wi h he in e nal eplica ion popula ion. Howe e , he e was no associa ion in he F ench A ican eplica e o in he second in e nal eplica e (Supplemen a y TableS10). Wi h all he Finnish popula ions combined, his SNP had an OR o 0.76 (p = 0.060). E alua ion o BPD isk and SNPs in he CRP egion and in o he egions p e iously associa ed wi h plasma CRP le els. Because SNP s11265269 showed he mos p omising associa ion wi h BPD and is loca ed be ween DUSP23 and CRP genes, we in es iga ed his genomic egion in mo e de ail. We de e mined ha s11265269 displays linkage disequilib ium (LD) wi h some o he SNPs wi hin o lanking he CRP gene a he han hose lanking he DUSP23 gene (Supplemen a y Figu eS4, Fig.2). This is consis en wi h 1000 Genomes da a (h p://www.1000genomes.o g). Fu he mo e, some o he o he SNPs in he icini y o only CRP gene also showed associa ion signals in GWAS. Because many SNPs in his egion a e known o be associa ed wi h se um CRP le els, we u he analysed 13 SNPs in he egion ex ending om 69 kb 5′ o 3 kb 3′ o CRP o cap u e all common a ia ions wi hin his egion in he i s in e nal eplica ion popula ion (Supplemen a y TableS11). In addi ion o s11265269, wo SNPs ( s3093059 and s12091403) in his egion showed p < 0.05 o associa- ions wi h BPD when he i s in e nal eplica ion popula ion was combined wi h he popula ion analysed in he GWAS. Haplo ype analysis in his egion e ealed ha h ee haplo ypes we e associa ed wi h BPD (p < 0.05); hese could be disc imina ed by he h ee SNPs wi h p < 0.05 in he single SNP analyses (Table4, Fig.2). SNP s11265269 was he only one o hese h ee SNPs nominally associa ed wi h BPD in he Finnish and F ench A ican eplica ion popula ions (Supplemen a y TableS12). Using ou GWAS da a, we also analysed 108 SNPs (23 SNPs nea CRP and 85 SNPs in o he genes) ha we e p e iously epo ed o be associa ed wi h CRP le els21–23. Some o he SNPs had nominal associa ions wi h BPD (p < 0.05). Howe e , in addi ion o hose in he CRP gene egion (Supplemen a y TableS11), none o hem ul- illed ou c i e ia o be included o geno yping in he eplica ion se s. SNPs in he CRP gene egion and plasma CRP le els. Nex , we analysed whe he SNPs nea he CRP gene a e associa ed wi h plasma CRP le els du ing he i s week o li e. Al oge he , 275 Finnish in an s we e included. CRP alues we e ob ained om clinical labo a o y epo s. In an s ca ying he BPD-p edisposing S udy popula ion nCase/con ol mino allele equency*OR (95% CI) p In e nal eplica ion popula ion 1 (Finnish) 326 0.278/0.230 1.42 (0.94–2.13)†0.097 In e nal eplica ion popula ion 2 (Finnish) 229 0.274/0.204 1.50 (0.75–2.86)†0.216 In e nal eplica ion popula ions combined 555 0.277/0.218 1.47 (1.04–2.06)†0.029 All Finnish popula ions combined 729 0.313/0.207 1.84 (1.39–2.45)†2.4 × 10–5 Ex e nal eplica ion popula ion 1 (Caucasian) 312 0.263/0.259 0.90 (0.59–1.37)†0.629 Ex e nal eplica ion popula ion 2 (F ench A ican) 76 0.440/0.235 2.48 (1.17–5.24)†0.017 Table 3. Associa ion o single-nucleo ide polymo phism s11265269 wi h mode a e- o-se e e b onchopulmona y dysplasia. De ini ion o abb e ia ions: CI, con idence in e al; OR, odds a io. * Mino allele equencies o he con ols a e simila o hose o he popula ions o he 1000genomes p ojec popula ions (0.239 and 0.223 o he Eu opean and A ican popula ions, espec i ely; h p://www.1000genomes.o g). †Odds a io o mino allele unde addi i e model in logis ic eg ession analysis wi h ges a ional age as a co a ia e. www.na u e.com/scien i ic epo s/ 6 Scien i ic RepoR s | 7: 9271 | DOI:10.1038/s41598-017-08977-w alleles o SNPs s11265269 and s3093059 ended o ha e highe CRP le els han non-ca ie s. In LR analysis wi h he numbe o su ac an doses as he es ablished signi ican co a ia e, SNP s3093059 was associa ed wi h mean plasma CRP concen a ions (p = 0.046) bu no wi h maximum plasma CRP concen a ions (p = 0.074) (Supplemen a y TableS13) du ing he i s week o li e. SNP s11265269 did no show associa ion. Plasma CRP le els du ing he i s week o li e associa ed wi h BPD. Because he SNP wi h he s onges associa ion signal in GWAS was loca ed ups eam o he CRP gene and o he SNPs nea CRP (including s3093059) associa ed wi h plasma CRP le els in adul popula ions22, 23, and also showed nominal associa ions wi h BPD, we analysed whe he maximum o mean CRP le els du ing he i s week o li e could p edic de elop- men o BPD. A ailable labo a o y da a om 275 Finnish p e e m in an s we e s udied. Figu e 2. Linkage disequilib ium plo o single-nucleo ide polymo phisms (SNPs) wi hin he egion lanking s11265269, he op SNP in genome-wide associa ion s udy (GWAS) o b onchopulmona y dysplasia. Pai wise D′ alues o each SNP pai in he combined Finnish popula ion (disco e y GWAS and i s in e nal eplica e) a e shown. Da ke colo s indica e s onge linkage disequilib ium. O he SNPs analysed, s2794520 and s3093059 (D′ 0.86 wi h s11265269) a e loca ed down- and ups eam o he CRP gene, espec i ely; hese SNPs a e known o be associa ed wi h plasma le els o CRP. SNP s1129923 is loca ed wi hin he DUSP23 (dual speci ici y phospha ase 23) gene; s7519478 and s4233356 a e up- and downs eam o DUSP23. Haplo ype block*Haplo ype Case/con ol haplo ype equency†p 1AGACG 0.335/0.355 0.522 AACCA 0.223/0.202 0.444 AAACG 0.146/0.186 0.122 AAACA 0.110/0.139 0.193 AAAAG 0.091/0.064 0.129 GAACA 0.092/0.049 8.2 × 10–3 2 GAA 0.385/0.473 8.8 × 10−3 AGA 0.281/0.308 0.383 GGG 0.324/0.210 9.1 × 10−5 3 AAGG 0.372/0.426 0.100 GGGA 0.252/0.232 0.499 GAAG 0.134/0.129 0.809 GGGG 0.137/0.123 0.529 GAGG 0.090/0.082 0.660 Table 4. Haplo ype analysis wi hin he egion lanking s11265269, he op single-nucleo ide polymo phism (SNP) in genome-wide associa ion s udy (GWAS) o b onchopulmona y dysplasia (BPD). *Haplo ype blocks a e illus a ed in Fig.2. Block 1 consis s o SNPs s3093059, s3122012, s2808635, s11265263, and s4285692 adjacen o he CRP gene. Block 2 consis s o SNPs s4656849, s12094103, and s11265269. Block 3 consis s o SNPs s4604689, s7519487, s1129923, and s4233356 encompassing he DUSP23 gene. BPD-p edisposing alleles o SNPs s3093059 and s11265269 and p o ec i e allele o s12091403 a e unde lined; hey disc imina e he haplo ypes showing associa ion (p < 0.05). †F equency o haplo ype in BPD cases and con ols in he combined Finnish popula ion (disco e y GWAS and i s in e nal eplica e). www.na u e.com/scien i ic epo s/ 7 Scien i ic RepoR s | 7: 9271 | DOI:10.1038/s41598-017-08977-w Plasma le els o CRP be o e 12 hou s’ age we e mos ly low; only 8.4% o he in an s had a CRP concen a ion o mo e han 1 mg/L; he maximum was 53.4 mg/L. These plasma le els we e no associa ed wi h he isk o BPD (p = 0.66). We nex s udied whe he maximum and mean plasma CRP concen a ions du ing he 1s week we e associa ed wi h he isk o BPD. The median alues o he maximum and mean CRP le els du ing he i s week we e 4.0 mg/L and 1.6 mg/L, espec i ely. The maximum plasma CRP concen a ion was abo e he median in 72.6% o he BPD cases and in 40.6% o he con ols (p = 3.0 × 10−6). Mean CRP le els du ing he 1s week o li e exceeded he median in 74.0% o he BPD cases and 41.6% o he con ols (p = 2.0 × 10−6) (Table5). The numbe o su ac an doses gi en was associa ed wi h hese CRP le els (p = 8.86 × 10−8 o he maximum and p = 3.25 × 10−8 o he mean le el). In con as , he leng h o ges a ion, p olonged up u e o memb anes, cho- ioamnioni is, co d blood pH o blood cul u e-p o en sepsis (ei he ea ly- o la e-onse du ing he i s week) showed no de ec able associa ion. In LR analysis wi h GA and SGA as co a ia es, bo h he maximum CRP le el du ing he i s week and he mean CRP le el du ing he i s week we e signi ican p edic o s o BPD, wi h ORs o 3.4 and 3.6, espec i ely (Table5). SNP s11265269 in he CRP gene egion is an independen isk ac o o BPD. To consolida e he e idence on he ole o s11265269 as a isk ac o o BPD, his SNP was analysed oge he wi h he signi ican an ena al isk ac o s, GA and SGA. Table6 shows he esul s o LR analysis ha included s11265269 in he model. This SNP p edic ed he isk o BPD in each popula ion, independen ly o he deg ee o p ema u i y and SGA s a us. Discussion BPD is a mul i ac o ial disease wi h high he i abili y. We conduc ed a GWAS on a ela i ely gene ically homog- enous Finnish popula ion and obse ed a sugges i e associa ion be ween mode a e- o-se e e BPD and SNP s11265269 nea he CRP gene. This associa ion was eplica ed in an independen Finnish popula ion and in popula ion o in an s o A ican descen om F ance. Fu he mo e, plasma le els o CRP du ing he i s week o li e p edic ed he de elopmen o BPD. Addi ionally, he known an ena al isk ac o s and s11265269 addi i ely p edisposed o BPD. Finally, ano he SNP om he CRP-gene egion, SNP s3093059, associa ed wi h plasma CRP le els. No signi ican associa ion was no ed wi h he SNP s11265269 and plasma CRP, howe e . CRP is a p ominen clinical ma ke o in ec ions and in lamma o y disease. Pe sis ing in lamma ion is a majo inding in in an s de eloping BPD. Mechanical en ila ion and hype oxia a e known pos na al isk ac o s o BPD and ha e been shown o induce in lamma ion24, 25. CRP is mainly p oduced in he li e . One o he main ligands o CRP is phosphocholine, a cons i uen o many bac e ial and ungal polysaccha ides. Phosphocholine is also p esen in dead o dying cells26. IL-6 and IL-1 ha e a majo syne gis ic e ec on he induc ion o CRP exp es- sion. High le el o CRP ac i a es phagocy ic cells, s imula es p oduc ion o in lamma o y cy okines, egula es he classical complemen pa hway and may cause ad e se ascula e en s when p esen in excess27. The plasma le els o CRP a e gene ally low a bi h28, and a mechanism o de icien CRP syn hesis in e al li e has been p oposed29. In p esen s udy, plasma le els o CRP be o e age 12 hou s we e mos ly low and p edic ed nei he BPD no sepsis. By con as , i was he inc ease in plasma CRP a e bi h and he main enance o a iably ele a ed CRP le els du ing he i s week o li e ha disc imina ed be ween in an s who will de elop BPD and hose who will no . This is in line wi h wha ea lie has been epo ed abou CRP and BPD, namely ha CRP on day 28 and BPD associa ed30, and ha he e was a quali a i e associa ion be ween CRP le els om days 0 o 21 and BPD31. In p esen s udy, he associa ion be ween CRP alues and BPD was demons a ed al eady du ing he i s week o li e, which is o clinical impo ance. Ou no el inding, ha ele a ed plasma CRP alues om he i s week o li e can be used as a bioma ke o an inc eased isk o BPD, needs o be con i med in independen s udies. The clinical easibili y o CRP makes i supe io o he o he ma ke s iden i ied wi h an inc eased isk o BPD32. No ewo hy he CRP alues did no co ela e wi h ei he ea ly- o la e-onse sepsis. This is in consonance wi h wha is known abou CRP in he p e e m i.e. ha bo h he baseline le el o CRP and he esponse o in ec ion a e lowe in p ema u e in an s compa ed o he e m new-bo ns. Also he sensi i i y and speci ici y o CRP in diagnosing sepsis a e lowe a p e e m neona es compa ed o hei e m pee s33. As all o he labo a o y alues, he in e p e a ion o CRP equi es he con ex o o he indings and clinical signs o he pa ien . Besides in lamma ion, in ec ion and o he en i onmen al ac o s, gene ic polymo phisms in luence he exp ession o CRP. The he i abili y o plasma CRP is es ima ed o be 25–40%, and le els o CRP a y among di - e en e hnic g oups34, 35. Mo eo e , he SNPs epo ed o a ec CRP le els a y among di e en popula ions23, 36. Va iable*BPD cases/ con ols, nBPD cases/con ols wi h CRP abo e median, n (%)†OR (95% CI)‡p‡ Maximum CRP 73/202 53 (72.6)/82 (40.6) 3.40 (1.76–6.58) 3.0 × 10−4 Mean CRP 73/202 54 (74.0)/84 (41.6) 3.57 (1.88–6.77) 9.7 × 10−5 Table 5. Associa ion o plasma C- eac i e p o ein (CRP) le el wi h b onchopulmona y dysplasia (BPD). De ini ion o abb e ia ions: BPD, b onchopulmona y dysplasia; CI, con idence in e al; CRP, C- eac i e p o ein; OR, odds a io. *Maximum o mean CRP le el (mg/L) du ing he i s week o li e. In Finland, a CRP le el o <3 mg/L is conside ed o be wi hin he no mal e e ence ange, bu no no ma i e alues o p e e m in an s a e a ailable. †χ2 es p alues we e 3.0 × 10−6 and 2.0 × 10−6 o maximum and mean CRP, espec i ely. ‡Logis ic eg ession wi h ges a ional age and small- o ges a ional (SGA, de ined as bi hweigh Z-sco e o ≤−2 SD) as co a ia es. Odds a io gi en o maximum and mean plasma CRP (abo e/below median) du ing i s week o li e. www.na u e.com/scien i ic epo s/ 8 Scien i ic RepoR s | 7: 9271 | DOI:10.1038/s41598-017-08977-w As such, he lack o an associa ion be ween s11265269 and BPD in one o he ex e nal eplica ion popula ions is no su p ising. CRP esponse a ies also be ween species and o ins ance in mice C p has gene ally low exp es- sion le els37, hus complica ing u he con i ma ion o he indings o his s udy. A s eng h o his s udy is he ela i e homogenei y o he popula ion in he GWAS and in he in e nal eplica ion se s. Because o he Finnish popula ion his o y wi h a es ic ed numbe o ounde s, he allelic di e si y in Finland is lowe han in mos o he popula ions38. This compensa es o he main limi a ion o he s udy, namely he small sam- ple size in he GWAS. Howe e , i is possible ha he small sample size limi s he o e all gene aliza ion o ou GWAS indings o o he s udy popula ions. Ano he limi a ion is he lack o associa ion be ween s11265269 and plasma CRP le els. This SNP, howe e , is loca ed in he same LD egion wi h s3093059 and o he CRP-associa ed SNPs, and also haplo ypes wi hin he egion associa ed wi h BPD. We p opose ha s11265269 may show milde associa ion wi h CRP plasma le els ha emains unde ec ed in ou se o s udy subjec s, o i may associa e wi h he CRP syn- hesis in lung cells39. Un o una ely, po en ially aluable da a on exp ession o CRP in lung compa men s was no a ailable. Fu he c edence o associa ion be ween s11265269 and mode a e- o-se e e BPD is gi en by he inding ha he associa ion emained simila when in an s wi h mild BPD we e excluded om he con ols. Fu he mo e, as al e ed ea men p ac ices and changes in diagnos ics may ha e in luenced he esul s, we epea ed he analyses in subg oups o in an s bo n in wo ime pe iods (be o e and a e 2010) and he associa ion emained simila . I is also possible ha he po en ial p edisposing ole o s11265269 in BPD is linked o a gene o he han CRP, such as DUSP23, he adjacen gene, encoding dual speci ici y phospha ase 23. DUSP23 is a phospha ase in ol ed in egula- ion o cell-cell adhesion40. I s po en ial link o BPD emains o be in es iga ed. Al hough none o he ini ial GWAS signals we e consis en ly eplica ed in ou eplica ion popula ions, some o hese may ep esen eal associa ion signals. As an example, o he SNPs sugges i ely associa ed wi h BPD in GWAS, hose wi hin RASGFR1 (encoding Ras p o ein speci ic guanine nucleo ide eleasing ac o 1) may ep- esen eal associa ions, since o he SNPs loca ed nea his gene showed associa ion signals in a p e ious GWAS o BPD11. Fu he mo e, some o he o he genes wi h sugges i ely associa ed SNPs could be biologically ele an o he BPD pheno ype. As an example, a ian s nea he SGCD gene (encoding sa coglycan del a) a e known o be associa ed wi h ai way esponsi eness, a p ocess ha is linked o decline in lung unc ion in subjec s wi h COPD41. La ge s udies a e equi ed o assess he po en ial ole o hese genes in BPD. Al hough he he i abili y o BPD is es ima ed o be high, consis en and s a is ically signi ican associa ions ha e no been de ec ed in genomic s udies. Cu en e idence sugges s a mul iplici y o genes and pa hways in ol ed in he pa hogenesis o BPD. Besides GWAS, hese pa hways ha e been sugges ed in o he la ge-scale s udies, includ- ing exome sequencing and ansc ip omic s udies. I is possible ha a e p edisposing a ian s de ec ed by exome sequencing can explain pa o he missing he i abili y o BPD; hese a e a ian s would be missed wi h he GWAS s a egy. In line wi h ou p esen inding, CRP was one o he op i e genes cha ac e izing se e e BPD compa ed o non-BPD in an s in a ecen exome sequencing s udy42. In ano he exome sequencing s udy, Li e al. en a i ely iden i ied 258 genes wi h a e nonsynonymous a ian s in pa ien s wi h mode a e- o-se e e BPD. These genes ep- esen ed pulmona y s uc u e and unc ion, mo phogenesis o emb yonic epi helium, and egula ion o he Wn signalling pa hway43. On he o he hand, Ambala anan and co-wo ke s combined a GWAS wi h pa hway analysis and ound e idence sugges ing in ol emen o bo h known (phospho us oxygen lyase ac i i y) and new ( a ge s o miR-219) pa hways in he pa hophysiology o BPD12. Bha acha ya e al. pe o med a genome-wide gene-exp ession s udy and epo ed e idence on accumula ion o connec i e issue in lamma o y mas cells in he lungs o in an s who died o BPD44. In addi ion o u he exome and whole-genome sequencing s udies o de ec a e p edisposing a ian s, addi ional app oaches, such as me a-analyses, pha macogenomics and expe imen al s udies a e equi ed. Iden i ica ion o BPD-p edisposing genes would b ing in o ma ion abou he pa hogenesis o BPD, ul ima ely ena- bling disco e y o new indi idualized s a egies agains his complex synd ome. We disco e ed a sugges i e associa ion be ween s11265269 polymo phism nea CRP gene and he isk o BPD. Plasma CRP sho ly a e bi h p edic s he isk o BPD. Toge he wi h he known an ena al isk ac o s s11265269 in luences suscep ibili y o BPD. As such, we p opose ha gene ic and en i onmen al ac o s in lu- ence he exp ession o CRP ha may s imula e pa hways p edisposing o BPD. Re e ences 1. S oll, B. J. e al. T ends in Ca e P ac ices, Mo bidi y, and Mo ali y o Ex emely P e e m Neona es, 1993–2012. JAMA 314, 1039–1051 (2015). 2. Ba aldi, E. & Filippone, M. Ch onic lung disease a e p ema u e bi h. N. Engl. J. Med. 357, 1946–1955 (2007). 3. Islam, J. Y., Kelle , R. L., Aschne , J. L., Ha e , T. V. & Moo e, P. E. Unde s anding he Sho - and Long-Te m Respi a o y Ou comes o P ema u i y and B onchopulmona y Dysplasia. Am. J. Respi . C i . Ca e Med. 192, 134–156 (2015). Va iable GWAS In e nal eplica ion popula ions join All Finnish popula ions join OR (95% CI) pOR (95% CI) pOR (95% CI) p SNP s11265269 3.37 (1.89–6.00)*3.76 × 10–5 1.45 (1.02–2.05)*0.038 1.82 1.36–2.43)*5.32 × 10−5 Ges a ional age†0.83 (0.68–1.02) 0.076 0.67 (0.60–0.75) 5.31 × 10−12 0.70 (0.63–0.77) 2.67 × 10−13 SGA‡3.50 (1.58–7.76) 1.98 × 10–3 2.65 (1.64−4.26) 6.45 × 10−52.86 (1.91–4.28) 3.05 × 10−7 Table 6. Logis ic eg ession model desc ibing p edic o s o b onchopulmona y dysplasia. De ini ion o abb e ia ions: GWAS, genome-wide associa ion s udy; CI, con idence in e al; OR, odds a io; SNP, single- nucleo ide polymo phism; SGA, small o ges a ional age. *Odds a io o mino allele unde addi i e model. †As con inuous a iable in he model. ‡De ined as bi hweigh Z-sco e o ≤−2 SD. Bi hweigh Z-sco e desc ibes dis ibu ion o bi hweigh a gi en leng h o ges a ion in SD. www.na u e.com/scien i ic epo s/ 9 Scien i ic RepoR s | 7: 9271 | DOI:10.1038/s41598-017-08977-w 4. Gough, A., Spence, D., Linden, M., Halliday, H. L. & McGa ey, L. P. Gene al and espi a o y heal h ou comes in adul su i o s o b onchopulmona y dysplasia: a sys ema ic e iew. Ches 141, 1554–1567 (2012). 5. Abman, S. H. B onchopulmona y dysplasia: “a ascula hypo hesis”. Am. J. Respi . C i . Ca e Med. 164, 1755–1756 (2001). 6. Balany, J. & Bhanda i, V. Unde s anding he Impac o In ec ion, In lamma ion, and Thei Pe sis ence in he Pa hogenesis o B onchopulmona y Dysplasia. F on . Med. (Lausanne) 2, 90 (2015). 7. Jobe, A. H. Mechanisms o Lung Inju y and B onchopulmona y Dysplasia. Am. J. Pe ina ol. 33, 1076–1078 (2016). 8. Bhanda i, V. e al. Familial and gene ic suscep ibili y o majo neona al mo bidi ies in p e e m wins. Pedia ics 117, 1901–1906 (2006). 9. La oie, P. M., Pham, C. & Jang, K. L. He i abili y o b onchopulmona y dysplasia, de ined acco ding o he consensus s a emen o he na ional ins i u es o heal h. Pedia ics 122, 479–485 (2008). 10. Hadchouel, A. e al. Iden i ica ion o SPOCK2 as a suscep ibili y gene o b onchopulmona y dysplasia. Am. J. Respi . C i . Ca e Med. 184, 1164–1170 (2011). 11. Wang, H. e al. A genome-wide associa ion s udy (GWAS) o b onchopulmona y dysplasia. Pedia ics 132, 290–297 (2013). 12. Ambala anan, N. e al. In eg a ed Genomic Analyses in B onchopulmona y Dysplasia. J. Pedia . (2014). 13. Shaw, G. M. & O’B odo ich, H. M. P og ess in unde s anding he gene ics o b onchopulmona y dysplasia. Semin. Pe ina ol. 37, 85–93 (2013). 14. La oie, P. In B onchopulmona y Dysplasia (ed Bhanda i, V.) 109–27 (Sp inge In e na ional, 2016). 15. Pel omaki, P. e al. Gene ic mapping o a locus p edisposing o human colo ec al cance . Science 260, 810–812 (1993). 16. Saba i, C. e al. Genome-wide associa ion analysis o me abolic ai s in a bi h coho om a ounde popula ion. Na . Gene . 41, 35–46 (2009). 17. Jobe, A. H. & Bancala i, E. B onchopulmona y dysplasia. Am. J. Respi . C i . Ca e Med. 163, 1723–1729 (2001). 18. Pu cell, S. e al. PLINK: a ool se o whole-genome associa ion and popula ion-based linkage analyses. Am. J. Hum. Gene . 81, 559–575 (2007). 19. K linge , G. e al. Pe ina al isk ac o s o b onchopulmona y dysplasia in a na ional coho o e y-low-bi hweigh in an s. Am. J. Obs e . Gynecol. 208, 115.e1–115.e9 (2013). 20. Bos e, C. e al. Fe al g ow h es ic ion and ch onic lung disease among in an s bo n be o e he 28 h week o ges a ion. Pedia ics 124, e450–8 (2009). 21. Schick, U. M. e al. Associa ion o exome sequences wi h plasma C- eac i e p o ein le els in >9000 pa icipan s. Hum. Mol. Gene . 24, 559–571 (2015). 22. Ridke , P. M. e al. Loci ela ed o me abolic-synd ome pa hways including LEPR,HNF1A, IL6R, and GCKR associa e wi h plasma C- eac i e p o ein: he Women’s Genome Heal h S udy. Am. J. Hum. Gene . 82, 1185–1192 (2008). 23. Okada, Y. e al. Genome-wide associa ion s udy o C- eac i e p o ein le els iden i ied pleio opic associa ions in he IL6 locus. Hum. Mol. Gene . 20, 1224–1231 (2011). 24. Bhanda i, V. & Elias, J. A. Cy okines in ole ance o hype oxia-induced inju y in he de eloping and adul lung. F ee Radic. Biol. Med. 41, 4–18 (2006). 25. Allison, B. J. e al. Ven ila ion o he e y imma u e lung in u e o induces inju y and BPD-like changes in lung s uc u e in e al sheep. Pedia . Res. 64, 387–392 (2008). 26. Weismann, D. & Binde , C. J. The inna e immune esponse o p oduc s o phospholipid pe oxida ion. Biochim. Biophys. Ac a 1818, 2465–2475 (2012). 27. Volanakis, J. E. Human C- eac i e p o ein: exp ession, s uc u e, and unc ion. Mol. Immunol. 38, 189–197 (2001). 28. Ishibashi, M., Takemu a, Y., Ishida, H., Wa anabe, K. & Kawai, T. C- eac i e p o ein kine ics in newbo ns: applica ion o a high- sensi i i y analy ic me hod in i s de e mina ion. Clin. Chem. 48, 1103–1106 (2002). 29. Nishikawa, T. e al. T ansc ip ional complex o ma ion o c-Fos, STAT3, and hepa ocy e NF-1 alpha is essen ial o cy okine-d i en C- eac i e p o ein gene exp ession. J. Immunol. 180, 3492–3501 (2008). 30. Ambala anan, N., Ross, A. C. & Ca lo, W. A. Re inol-binding p o ein, ans hy e in, and C- eac i e p o ein in ex emely low bi h weigh (ELBW) in an s. J. Pe ina ol. 25, 714–719 (2005). 31. D’Angio, C. T. e al. Blood Cy okine P o iles Associa ed wi h Dis inc Pa e ns o B onchopulmona y Dysplasia among Ex emely Low Bi h Weigh In an s. J. Pedia . (2016). 32. Lal, C. V. & Ambala anan, N. Cellula and humo al bioma ke s o B onchopulmona y Dysplasia. Ea ly Hum. De . (2016). 33. Ho e , N., Zacha ias, E., Mulle , W. & Resch, B. An upda e on he use o C- eac i e p o ein in ea ly-onse neona al sepsis: cu en insigh s and new asks. Neona ology 102, 25–36 (2012). 34. Pankow, J. S. e al. Familial and gene ic de e minan s o sys emic ma ke s o in lamma ion: he NHLBI amily hea s udy. A he oscle osis 154, 681–689 (2001). 35. Anand, S. S. e al. C- eac i e p o ein as a sc eening es o ca dio ascula isk in a mul ie hnic popula ion. A e ioscle . Th omb. Vasc. Biol. 24, 1509–1515 (2004). 36. Reine , A. P. e al. Genome-wide associa ion and popula ion gene ic analysis o C- eac i e p o ein in A ican Ame ican and Hispanic Ame ican women. Am. J. Hum. Gene . 91, 502–512 (2012). 37. Pepys, M. B. & Hi sch ield, G. M. C- eac i e p o ein: a c i ical upda e. J. Clin. In es . 111, 1805–1812 (2003). 38. Pel onen, L., Jalanko, A. & Va ilo, T. Molecula gene ics o he Finnish disease he i age. Hum. Mol. Gene . 8, 1913–1923 (1999). 39. Dong, Q. & W igh , J. R. Exp ession o C- eac i e p o ein by al eola mac ophages. J. Immunol. 156, 4815–4820 (1996). 40. Gallegos, L. L. e al. A p o ein in e ac ion map o cell-cell adhesion egula o s iden i ies DUSP23 as a no el phospha ase o be a- ca enin. Sci. Rep. 6, 27114 (2016). 41. Hansel, N. N. e al. Genome-Wide Associa ion S udy Iden i ica ion o No el Loci Associa ed wi h Ai way Responsi eness in Ch onic Obs uc i e Pulmona y Disease. Am. J. Respi . Cell Mol. Biol. 53, 226–234 (2015). 42. Ca e a, P. e al. Exome sequencing and pa hway analysis o iden i ica ion o gene ic a iabili y ele an o b onchopulmona y dysplasia (BPD) in p e e m newbo ns: A pilo s udy. Clin. Chim. Ac a 451, 39–45 (2015). 43. L i, J. e al. Exome Sequencing o Neona al Blood Spo s and he Iden i ica ion o Genes Implica ed in B onchopulmona y Dysplasia. Am. J. Respi . C i . Ca e Med. 192, 589–596 (2015). 44. Bha acha ya, S. e al. Genome-wide ansc ip ional p o iling e eals connec i e issue mas cell accumula ion in b onchopulmona y dysplasia. Am. J. Respi . C i . Ca e Med. 186, 349–358 (2012). Acknowledgemen s The au ho s wish o hank Maa i Haa ala o labo a o y assis ance; Rii a Vike äinen, Ma i a Suni, Kaisa Jaakkonen, Leena Ki inen, Jaana Vuolle -Puu unen, and Sa u Ekblad o sample and da a collec ion; Leena Haa aja and Helena Lapinleimu o subjec en ollmen and sample collec ion; Mihoko Ladd and Ba b Kams a o coo dina ing sample collec ion and en ollmen in Vancou e and Edmon on (Canada), espec i ely; Claude Danan, Pie e-Hen i Ja eau, Richa d Lenclen and he EPIPAGEADO s udy g oup o coo dina ing sample collec ion and subjec en olmen in F ance, and he Cen e de Ressou ces Biologiques (CRB, CHI C é eil) o F ench DNA s o ing; and Ki a O’Day Helle o language edi ing o he manusc ip . Geno yping o SNP ma ke s