Gene-Environment Interaction in Parkinson s Disease: Coffee, ADORA2A, and CYP1A2

Transcription

1 Originl Pper Received: July 22, 2016 Accepted: September 16, 2016 Published online: Jnury 31, 2016 Gene-Environment Interction in Prkinson s Disese: Coffee, ADORA2A, nd CYP1A2 Yu-Hsun Chung Christin M. Lill f Pei-Chen Lee h Johnni Hnsen i Christin F. Lssen i Lrs Bertrm g, j Nomi Greene Jnet S. Sinsheimer b, d Bete Ritz, c, e Deprtment of Epidemiology, b Deprtment of Biosttistics, c Deprtment of Environmentl Helth Sciences, Fielding School of Public Helth, University of Cliforni, Los Angeles (UCLA), d Deprtment of Humn Genetics nd Biomthemtics, nd e Deprtment of Neurology, Dvid Geffen School of Medicine t UCLA, Los Angeles, CA, USA; f Genetic nd Moleculr Epidemiology Group, Institute of Neurogenetics, nd g Interdisciplinry Pltform for Genome Anlytics (LIGA), Institutes of Neurogenetics nd Integrtive nd Experimentl Genomics, University of Lübeck, Lübeck, Germny; h Deprtment of Helth Cre Mngement, College of Helthcre Administrtion nd Mngement, Ntionl Tipei University of Nursing Helth Sciences, Tipei, Tiwn; i Dnish Cncer Society Reserch Center, Dnish Cncer Society, Copenhgen, Denmrk; j School of Public Helth, Fculty of Medicine, The Imperil College of Science, Technology, nd Medicine, London, UK Keywords Prkinson s disese Cffeine Adenosine A2A receptor ( ADORA2A ) Cytochrome P450 1A2 ( CYP1A2 ) Met-nlysis Abstrct Bckground nd Purpose: Drinking cffeinted coffee hs been reported to provide protection ginst Prkinson s disese (PD). Cffeine is n denosine A2A receptor (encoded by the gene ADORA2A ) ntgonist tht increses dopminergic neurotrnsmission nd Cytochrome P450 1A2 (gene: CYP1A2 ) metbolizes cffeine; thus, gene polymorphisms in ADORA2A nd CYP1A2 my influence the effect coffee consumption hs on PD risk. Methods: In popultion-bsed cse control study (PASIDA) in Denmrk (1,556 PD ptients nd 1,606 birth yer- nd gender-mtched controls), we ssessed interctions between lifetime coffee consumption nd 3 polymorphisms in ADORA2A nd CYP1A2 for ll subjects, nd incident nd prevlent PD cses seprtely using logistic regression models. We lso conducted met-nlysis combining our results with those from previous studies. Results: We estimted sttisticlly significnt interctions for ADORA2A rs nd hevy vs. light coffee consumption in incident (OR interction = 0.66 [95% CI ], p = 0.02) but not prevlent PD. We did not observe interctions for CYP1A2 rs nd rs in incident or prevlent PD. In met-nlyses, PD ssocitions with dily coffee consumption were strongest mong crriers of vrint lleles in both ADORA2A nd CYP1A2. Conclusion: We corroborted results from previous report tht described interctions between ADORA2A nd CYP1A2 polymorphisms nd coffee consumption. Our results lso suggest tht survivor bis my ffect results of studies tht enroll prevlent PD cses S. Krger AG, Bsel E-Mil krger@krger.com S. Krger AG, Bsel /17/ $39.50/0 Bete Ritz, MD, PhD Deprtment of Epidemiology, Fielding School of Public Helth University of Cliforni t Los Angeles, 650 Chrles E. Young Drive Los Angeles, CA (USA) E-Mil ucl.edu

2 Introduction More thn 90% of Prkinson s disese (PD) is considered to be idiopthic with genetic nd environmentl fctors incresing risk of disese. A protective effect of coffee on PD hs been postulted, since mny epidemiologic studies reported lower consumption of cffeinted coffee mong PD ptients [1, 2]. Ever vs. never drinkers hve 30% lower risk of PD nd 3 dditionl cups of coffee per dy lowered PD risk on verge by 25 32% [3]. Cffeinted coffee is very populr beverge in Northern Europen countries, especilly Denmrk, nd we recently reported 55% lower risk of PD mong moderte coffee drinkers in Denmrk [4]. A lndmrk erly prospective cohort study not only reported n inverse ssocition for coffee but lso for cffeine from non-coffee sources, suggesting it might be the protective gent [5]. This ssocition ws replicted in mny prospective cohort nd cse control studies nd further strengthened by observtions of exposure-response trends [3, 6, 7]. Animl studies lent dditionl support to the ide tht cffeine nd its metbolites re neuro-protective [8]. Yet, the biologicl mechnisms underlying neuroprotection derived from cffeine re yet to be estblished. Importntly, epidemiologic dt even from prospective studies do not preclude reverse cuslity since those who lter develop PD my stop drinking cffeinted coffee due to sleep disorders, nxiety, gstro-intestinl problems or simply loss of smell tht could mke coffee drinking less enjoyble in the very long pre-motor stges of PD. Evidence for interctions between cffeinted coffee nd genes tht metbolize cffeine or encode brin receptors trgeted by cffeine, could help strengthen rguments tht cffeine indeed plys biologicl role in reducing PD risk. Recently, lrge consortium (PEGASUS) combined dt from 1,325 PD cses nd 1,735 controls nd reported tht PD risk ws influenced by interctions between the single-nucleotide polymorphisms (SNP) rs nd rs in ADORA2A, which encodes the denosine A2A receptor in dopmine neurons, nd cffeinted coffee consumption [9] ; however, two much smller studies did not find evidence for such interction [10, 11]. In ddition, the PEGASUS study lso observed stronger coffee-pd ssocitions mong crriers of the CC genotype of rs in CYP1A2 compred with CA or AA crriers [9], gene tht encodes the cytochrome P450, fmily 1, subfmily A, polypeptide 2, the min cffeine-metbolizing enzyme. Relying on dt from lrge popultion-bsed cse control study of PD (Prkinson s Disese in Denmrk [PASIDA]), here we re-exmine interctions of coffee consumption with ADORA2A nd CYP1A2 polymorphisms nd lso ssess whether relince on prevlent versus incident PD cses influences results, distinction tht my hve cused previous study results to disgree [12]. Method The PASIDA study ws pproved by the Institutionl Review Bords of the Dnish Dt Protection Agency nd the Ethics Committee of Copenhgen. Informed consent ws obtined from ll study prticipnts. Study Popultion The PASIDA study enrolled idiopthic PD ptients (ICD nd ICD-10 G20) treted t 10 neurologicl tretment centers nd identified from the Dnish Ntionl Hospitl Register between 1996 nd mid-2009 with subsequent vlidtion of their dignoses by medicl record review. Popultion controls, free of PD when mtched cses were dignosed, were selected from the Dnish Centrl Popultion Registry (individully mtched on yer of birth nd gender). Detiled recruitment informtion ws published previously [4]. Of 3,700 recruited subjects, 1,575 (87%) PD cses nd 1,607 (85%) controls provided DNA smples (sliv) for genotyping. We further excluded subjects who were dignosed with dementi prior to interview, leving 1,556 PD cses nd 1,606 controls for nlyses. Exposure Assessment nd Vrible Definition Stndrdized telephone interviews were conducted between 2008 nd 2010 to obtin prticipnts lifetime cffeinted coffee consumption history (drip- nd instnt-coffee) nd informtion on other lifestyle fctors. Due to the high prevlence (>90%) of cffeinted coffee drinking in Denmrk, but little te nd cffeinted sod consumption during the study period, we omitted the ltter cffeine sources. We collected lifetime mount nd durtion of cffeinted coffee-drinking, sking prticipnts to report strt nd stop ges nd the verge number of cups they consumed per dy. We consider n ever coffee drinker someone who consumed t lest one cup (6 oz) of coffee per week for yer. To obtin the mount of cffeine intke, we converted coffee cups per dy into dily cffeine consumption (mg) using the US Deprtment of Agriculture criteri [13]. Only consumption before the index dte contributed to our exposure mesures, tht is, the dte of first motor symptom recorded on the medicl record, or the dte of PD dignosis for both cses nd their mtched controls. Genotyping DNA ws extrcted from sliv using stndrd protocols. Smples were genotyped on the QuntStudio TM 12K Flex Rel-Time PCR System using multiplex Tqmn llelic discrimintion ssys (Applied Biosystems) ccording to the mnufcturer s protocol. Ech 384-well plte included 5% HpMp CEU smples genotyped in duplictes cross pltes to ssess genotyping ccurcy. To control for genotyping qulity, we excluded smples with genotyping efficiency less thn 80% nd SNPs with low genotyping efficiency (<95%) nd ccurcy (<99.5%); ll 3 SNPs (rs , rs762551, nd rs ) in this study met these criteri. Coffee, ADORA2A nd CYP1A2 Interction in PD 193

3 Sttisticl Anlysis We tested for devition from the Hrdy Weinberg equilibrium mong controls using Person s chi-squre test (ll p 0.05). We broke the mtched pirs nd conducted unconditionl logistic regression nlyses djusting for gender, birth yer, nd onset/index ge to estimte mrginl ssocitions between cffeinted coffee consumption nd PD sttus s well s between the three ADORA2A or CYP1A2 polymorphisms nd PD sttus (dditive genetic model), respectively. We broke the mtched sets to void loss of entire pirs with only one subject when conducting strtified nlyses nd to increse efficiency since mny pirs shred the sme mtching vrible vlues [4]. However, we compred the overll results from conditionl with the results from unconditionl logistic regression djusted for ll of the mtching vribles nd found them to be identicl. Mtching vribles (i.e., yer of birth, gender, nd onset/index ge), potentil confounders (i.e., ny kind of tobcco smoking) nd strong predictors of PD (i.e., fmily history) were included in ll models. We treted coffee intke s binry vrible with light vs. hevy consumption (defined s 0 to medin vs. >medin cup-yers [14] ) nd lso s continuous vrible (number of cups per dy). We further creted ctegories of cffeine intke in mg per dy nd yers of coffee consumed using ctegory definitions from our previous pper [15]. The Wld test for trend ws pplied to ctegorized coffee vribles testing for liner reltionship with PD. Informtion bout ethnic diversity ws not vilble, but bsed on demogrphics of the Dnish popultion provided by Denmrk, we re confident tht the lrge mjority were non-hispnic Whites [16]. We used multiplictive terms in logistic regression djusted for confounders to ssess whether the ADORA2A or CYP1A2 polymorphisms modify cffeine-pd ssocitions, nd the likelihood rtio chi-squre tests ws used to evlute sttisticl significnce. We lso restricted ll nlyses to incident PD ptients nd their mtched controls, tht is, those dignosed close to their dte of interview during , to ssess whether survivl or recll bis my hve influenced results with prevlent ptients. Anlyses were conducted using SAS version 9.4 (SAS Institute, Inc., Cry, NC, USA). Lstly, we conducted met-nlyses to ggregte results from PASIDA (incident cses only) nd non-hispnic Whites from PEGASUS [9], bsed on the type of PD cse (i.e., incident), control selection (i.e., popultion-bsed controls), nd ethnicity (i.e., non- Hispnic Whites) using the metgen pckge in the R environment, which llows to fit fixed-effects nd rndom-effects models [17]. In the met-nlysis, results of ADORA2A rs in PASIDA were equted with rs in PEGASUS becuse they re in high linkge disequilibrium (LD) [18] ; lso, we combined our coffee ctegory of hevy use with ever consumption in PEGASUS s well s nd light consumption in PASIDA with never in PEGASUS, since less thn 10% of PASIDA prticipnts reported hving never consumed coffee. Results Our initil nlysis included 3,162 Dnish prticipnts in the PASIDA study with high-qulity genotyping dt. The verge ge of PD onset or index ge ws 61 yers for ll prticipnts ( Tble 1 ) nd 64 yers for incident PD ptients nd their mtched controls only. Sixty percent of prticipnts were mle nd, compred with popultion controls, PD cses were more likely to hve positive fmily history of PD nd smoke less. Ninety-four percent of PD cses nd 97% of controls were ever coffee consumers. Hevy coffee drinking in PASIDA is ssocited with 25% lower risk of PD (OR 0.75 [95% CI ]), nd ech dditionl cup of coffee consumed per dy on verge is ssocited with 4% lower PD risk (OR 0.96 [95% CI ]); online suppl. Tble S1; for ll online suppl. mteril, see inverse coffee-pd ssocitions re estimted for both prevlent nd incident PD. Of note, the per-cup mesure of dily coffee consumption ws not ssocited with PD mong incident cses. OR estimtes djusted solely for birth yer, gender, nd onset/index ge did not substntilly differ from estimtes further djusted for ever smoking nd PD fmily history. Finlly, mrginl ssocitions of ADORA2A rs s well s of CYP1A2 rs nd rs (in LD with rs762551: r 2 = 0.87, D = 0.99) with PD sttus (incident nd prevlent) were null (online suppl. Tble S2). Interction nlyses bsed on ll subjects did not show sttisticlly nd significntly vrying effects of cffeine cross genotypes of ADORA2A or CYP1A2 polymorphisms, respectively ( Tble 2 ). However, there ppered to be trend in coffee-pd effect estimtes cross ADORA2A rs genotypes: the OR for PD mong hevy coffee drinkers, reltive to light coffee drinkers, ws 0.81 (95% CI ) for GG crriers compred with 0.68 (95% CI ) for GT nd 0.54 (95% CI ) for TT crriers (OR interction = 0.85 [95% CI ], p for interction = 0.14). Further djustment for smoking nd PD fmily history did not chnge results (dt not shown). When we restricted our nlyses to incident PD only, we observed sttisticlly significnt interction for the ADORA2A rs nd hevy coffee drinking (OR interction = 0.66 [95% CI ], p for interction = 0.02): the OR for drinking coffee in GG crriers ws 1.10 (95% CI ), 0.63 (95% CI ) for GT, nd 0.58 (95% CI ) for TT crriers. When the durtion of cffeine intke ws removed from the cffeine mesure, the interction of ADORA2A polymorphism nd coffee ws not sttisticlly significnt ( p for interction = 0.28 in cup/dy for ll cses, nd p = 0.55 for incident cses, respectively). There ws no evidence for ssocition mesure modifiction for CYP1A2 rs nd rs ( p for interction = 0.45 nd 194 Chung/Lill/Lee/Hnsen/Lssen/Bertrm/ Greene/Sinsheimer/Ritz

4 Tble 1. Chrcteristics of study prticipnts (n = 3,162) All Incident cses only Prevlent cses only cses (n = 1,556) controls (n = 1,606) cses (n = 554) controls (n = 566) cses (n = 1,002) controls (n = 1,040) Onset/index ge, men ± SD 61.1± ± ± ± ± ±9.9 Mle, % 932 (59.9) 975 (60.7) 335 (60.5) 357 (63.1) 597 (59.6) 618 (59.4) Ever smoking, % 775 (49.9) 1,026 (64.2) 271 (49.2) 349 (62.2) 504 (50.4) 677 (65.3) Pckyers of cigrette smoking (SD) 7.0 (13.3) 11.3 (16.3) 8.0 (14.7) 11.0 (16.8) 6.5 (12.3) 11.5 (16.0) Fmily history of PD, % 211 (13.6) 89 (5.5) 74 (13.36) 28 (5.0) 137 (13.7) 61 (5.9) Cffeinted coffee consumption Cup per dy b, % Never 85 (6.4) 43 (3.0) 25 (5.2) 18 (3.5) 60 (7.1) 25 (2.7) (8.4) 79 (5.5) 36 (7.5) 23 (4.5) 75 (8.9) 56 (6.1) (18.4) 243 (17.0) 77 (16.0) 71 (13.8) 167 (19.8) 172 (18.8) (66.8) 1,064 (74.5) 342 (71.3) 403 (78.3) 542 (64.2) 661 (72.3) Men ± SD 3.9± ± ± ± ± ±2.9 Cup-yers b, % Light (0, medin) 806 (60.9) 746 (52.2) 253 (52.7) 236 (45.8) 553 (65.5) 510 (55.8) Hevy (>medin) 518 (39.1) 683 (47.8) 227 (47.3) 279 (54.2) 291 (34.5) 404 (44.2) Cffeine, mg/dy, medin (qurtile) b, % (32.4) 357 (24.9) 136 (28.3) 108 (21.0) 293 (34.7) 249 (27.2) > (32.5) 445 (31.1) 154 (32.1) 171 (33.2) 276 (32.7) 274 (30.0) > (15.4) 269 (18.8) 73 (15.2) 95 (18.4) 131 (15.5) 174 (19.1) > (19.7) 358 (25.1) 117 (24.4) 141 (27.4) 144 (17.1) 217 (23.7) Yers of coffee drinking b, % (38.9) 467 (32.5) 155 (32.2) 123 (23.8) 364 (42.8) 344 (37.4) > (23.4) 352 (24.5) 111 (23.1) 131 (25.4) 200 (23.5) 221 (24.0) > (22.6) 354 (24.6) 124 (25.8) 142 (27.5) 177 (20.8) 212 (23.0) > (15.1) 264 (18.4) 91 (18.9) 120 (23.3) 110 (12.9) 144 (15.6) Polymorphisms ADORA2A rs c, % GG 534 (34.5) 533 (33.3) 201 (36.4) 193 (34.2) 333 (33.4) 340 (32.9) GT 735 (47.5) 781 (48.9) 255 (46.2) 270 (47.8) 480 (48.2) 511 (49.5) TT 279 (18.0) 284 (17.8) 96 (17.4) 102 (18.0) 183 (18.4) 182 (17.6) CYP1A2 rs c, % AA 874 (56.3) 873 (54.7) 312 (56.4) 322 (57.4) 562 (56.2) 551 (53.3) CA 563 (36.3) 607 (38.1) 205 (37.1) 202 (36.0) 358 (35.8) 405 (39.2) CC 116 (7.4) 115 (7.2) 36 (6.5) 37 (6.6) 80 (8.0) 78 (7.5) CYP1A2 rs c, % AA 750 (48.4) 750 (46.8) 260 (47.1) 283 (50.3) 490 (49.1) 466 (44.9) GA 630 (40.7) 681 (42.5) 234 (42.4) 229 (40.7) 396 (39.7) 452 (43.5) GG 169 (10.9) 171 (10.7) 58 (10.5) 51 (9.0) 111 (11.2) 120 (11.6) Age rnge t interview: cses yers, controls yers. Age t PD onset: yers; ge t dignosis: yers. Age t the dte of first motor symptoms recorded on medicl record. b Missing informtion: smoking (n = 13), pckyer (n = 416), coffee cup/dy (n = 409), cup-yers (n = 409), cffeine mg/dy (n = 409), yers of coffee drinking (n = 393); medin vlue ws determined bsed on controls (excluding non-drinkers [14]). c Genotyping filures: rs (n = 16), rs (n = 14), rs (n = 12). Coffee, ADORA2A nd CYP1A2 Interction in PD 195

5 Tble 2. Adjusted ORs nd 95% CIs for cffeinted coffee consumption nd PD in PASIDA, by ADORA2A nd CYP1A2 genotypes Cffeinted coffee Homozygous mjor Heterozygous Homozygous minor OR interction (95% CI) cses controls cor 95% CI cses controls cor 95% CI cses controls cor 95% CI p vlue Incident nd prevlent cses: 1,324 cses, 1,429 controls ADORA2A rs Cups/dy, men ± SD b 4.1± ± ± ± ± ± ( ) 0.28 Light (0, medin cup-yers) Hevy (>medin cup-yers) ( ) 0.14 CYP1A2 rs Cups/dy, men ± SD b 4.3± ± ± ± ± ± ( ) 0.62 Light (0, medin cup-yers) Hevy (>medin cup-yers) ( ) 0.45 CYP1A2 rs Cups/dy, men ± SD b 4.3± ± ± ± ± ± ( ) 0.85 Light (0, medin cup-yers) Hevy (>medin cup-yers) (0.80 1/27) 0.93 Incident cses only: 480 cses, 515 controls ADORA2A rs Cups/dy, men ± SD b 4.6± ± ± ± ± ± ( ) 0.55 Light (0, medin cup-yers) Hevy (>medin cup-yers) ( ) 0.02 CYP1A2 rs Cups/dy, men ± SD b 4.6± ± ± ± ± ± ( ) 0.20 Light (0, medin cup-yers) Hevy (>medin cup-yers) ( ) 0.89 CYP1A2 rs Cups/dy, men ± SD b 4.6± ± ± ± ± ± ( ) 0.21 Light (0, medin cup-yers) Hevy (>medin cup-yers) ( ) 0.54 Prevlent cses only: 844 cses, 914 controls ADORA2A rs Cups/dy, men ± SD b 3.8± ± ± ± ± ± ( ) 0.05 Light (0, medin cup-yers) Hevy (>medin cup-yers) ( ) 0.94 CYP1A2 rs Cups/dy, men ± SD b 4.1± ± ± ± ± ± ( ) 0.73 Light (0, medin cup-yers) Hevy (>medin cup-yers) ( ) 0.31 CYP1A2 rs Cups/dy, men ± SD b 4.1± ± ± ± ± ± ( ) 0.48 Light (0, medin cup-yers) Hevy (>medin cup-yers) ( ) 0.66 cor, crude OR, tht is, djusted only for the covrites yer of birth, gender, nd onset/index ge (continuous); p for interction (coffee*genotype) bsed on chi-squre test with df = 1 (dditive genetic model). Subjects with missing coffee informtion or SNP dt were excluded. b Coffee non-drinkers were excluded [9]. 196 Chung/Lill/Lee/Hnsen/Lssen/Bertrm/ Greene/Sinsheimer/Ritz

6 Tble 3. Adjusted ORs nd 95% CIs for the ssocition between cffeinted coffee consumption nd PD in the PEGASUS nd PASIDA studies, by ADORA2A nd CYP1A2 genotypes: met-nlytic results using rndom-effect models Cffeinted coffee Homozygous mjor Heterozygous Homozygous minor OR 95 % CI OR 95% CI OR 95% CI ADORA2A rs in PEGASUS (LD with rs in PASIDA) Cups/dy, men ± SD Ever vs. never* CYP1A2 rs Cups/dy, men ± SD Ever vs. never* CYP1A2 rs Cups/dy, men ± SD Ever vs. never* * Hevy vs. light coffee consumption in the PASIDA study. Adjusted for ge, gender nd site in PEGASUS; djusted for the covrites yer of birth, gender, nd onset/index ge (continuous) in PASIDA. 0.93, respectively). Similrly, restricting to incident cses only did not revel sttisticlly significnt interctions for CYP1A2 SNPs. No interctions were found in prevlent cse nlyses. Results of interction nlyses using other coffee mesures, dily intke of cffeine nd yers of coffee drinking re presented in online supplementl Tble S3. We did not observe evidence for effect-mesure modifiction with these mesures nd the SNPs we investigted. Our met-nlytic results for coffee-pd ssocitions did not differ much when we used rndom-effects versus fixed-effects models ( Tble 3 ; online suppl. Tble S4). Bsed on rndom effects models, the ADORA2A gene polymorphisms nd dily coffee consumption ssocition for PD ws strongest mong rs TT crriers, OR 0.77 (95% CI ), compred with GT nd GG crriers (OR 0.96 [95% CI ] nd 0.97 [95% CI ], respectively). We sw similr ptterns for CYP1A2 polymorphisms in both rs nd rs , tht is, the coffee-pd ssocitions were strongest mong homozygotes for the vrint lleles (OR 0.86 [95% CI ] for rs CC crriers nd 0.84 [95% CI ] for rs GG crriers; Fig. 1 ). Discussion We conducted nlyses of gene-environment interctions in Dnish cse control study of PD. Our study follows up on the results from previous consortium study (PEGASUS) with n eqully lrge smple size. When we include both prevlent nd incident PD cses in our nlysis, we found no evidence for interctions with ADORA2A/CYP1A2 polymorphisms. However, when we restricted nlyses to incident cses only, we observed interctions between the ADORA2A polymorphism nd coffee drinking. This difference in results suggests tht survivl or recll bis my ffect studies tht rely on or include prevlent PD cses. Moreover, when we combined our results for incident cses with the PEGASUS incident cses of Europen ncestry in met-nlyticl pproch, both ADORA2A nd CYP1A2 polymorphisms modified coffee-pd ssocitions, lthough the CYP1A2 interction ws solely due to the influence of the PEGASUS study. Our PASIDA findings re mostly consistent with those published by the PEGASUS consortium, which previously reported ORs for PD risk of ech dditionl cup of coffee consumed per dy mong coffee drinkers s 0.93 (95% CI ) nd 0.92 (95% CI ) for CC or CT crriers of ADORA2A rs , respectively, nd 0.61 (95% CI ) for TT crriers ( p for interction = 0.01) in non-hispnic Whites [9]. Yet, two smller studies did not find sttisticlly significnt interctions for ADORA2A polymorphisms nd coffee in PD [10, 11]. One study ws conducted in mixed-rce popultion tht did not find the expected inverse min effect for coffee consumption on PD, possibly becuse sibling controls were used [10]. Sibling controls re likely too similr to cses in terms of coffee consumption, mking it hrd to Coffee, ADORA2A nd CYP1A2 Interction in PD 197

7 Odds rtio rs CC (rs GG) ADORA2A rs , cup/dy rs CT (rs GT) 0.77 rs TT (rs TT) Odds rtio b CYP1A2 rs762551, cup/dy rs AA rs AC rs CC Color version vilble online Odds rtio c CYP1A2 rs , cup/dy rs AA rs AG rs GG PEGASUS (popt et l.) PASIDA Met-nlysis (rndom effect) Fig. 1. c ORs for cups of coffee consumed per dy nd PD cross ADORA2A/CYP1A2 gene polymorphisms. estimte effects of coffee consumption on PD risk. The second null result ws reported for n Asin popultion with low verge coffee consumption (2.9 in cses vs. 4.7 in controls [11] cup-yers compred with vs cup-yers in PASIDA), such tht the exposure levels nd contrsts were likely insufficient. Animl studies hve shown tht dministrtion of cffeine or other denosine A2A receptor ntgonists before dosing the niml with the neurotoxin 1-methyl-4-phenyl-1,2,3,6-tetrhydropyridine reduces loss of dopmine nd dopminergic neurons, suggesting tht cffeine reduces PD risk by dectivting the A2A receptors [19, 20]. Also, the ADORA2A rs vrint (in LD with rs ) hs been shown to reduce protein expression [21], nd thus my result in reduced A2A receptor function tht together with further inhibition through coffee consumption my exert protective effects [9]. We would thus expect the inverse coffee-pd ssocition to be the strongest in those with TT genotype in ADORA2A rs (in LD with rs ). Adenosine A2A receptors hve lso become the ltest trget for non-dopminergic therpies in PD bsed on their interction with dopmine D 2 receptors in stritopllidl neurons [22, 23]. Cytochrome P450 1A2 is the min cffeine-metbolizing enzyme tht converts over 90% of cffeine in the liver to prxnthine, nd its ctivity depends on ge, gender, smoking, nd CYP1A2 polymorphisms [24 26]. Thus, we would expect neuroprotection due to cffeine to be stronger in slow metbolizers who crry the vrint lleles we investigted. The PEGASUS consortium pooling incident cse control studies ( n = 3,060), found evidence for coffee- CYP1A2 interctions with inverse PD ssocitions for coffee drinking being strongest in CC genotype crriers t rs nd GG genotype crriers t rs [9]. The NeuroGenetic Reserch Consortium ( n = 2,389) did not find interctions with CYP1A2 but included prevlent cses nd some studies used spousl controls [12]. Previously, concerns were rised tht confounding by popultion structure in PEGASUS produced spurious results [12, 27] since llele frequencies for CYP1A2 SNPs vry strongly cross ethnicities nd in PEGASUS the coffee- CYP1A2 interctions did not rech sttisticl significnce in non-hispnic Whites lone [9]. However, combining PEGASUS nd PASIDA non-hispnic Whites, our met-nlysis produced decresing trend cross CYP1A2 rs genotypes (green/middle line; Fig. 1 b) nd suggested n interction with cups per dy of consumption. Interestingly, ever (vs. never ) coffee consumption produced n inverted-u shpe for the CYP1A2 rs polymorphisms, but ever (vs. never ) coffee consumption is poor mesure of verge cffeine intke ( Tble 3 ). 198 Chung/Lill/Lee/Hnsen/Lssen/Bertrm/ Greene/Sinsheimer/Ritz

8 Smoking is positively ssocited with coffee drinking nd negtively with PD risk, which we hve previously interpreted s consequence of pre-motor prodroml PD [3]. Moreover, study reported tht the metbolic ctivity did not differ between AC or CC nd AA crriers t rs in non-smokers suggesting tht CYP1A2 genotypes my influence enzyme ctivity only in smokers [26]. In our study, djustment for smoking did not chnge the interction estimtes for either of the CYP1A2 SNPs. Complicting the mtter further, both cffeine nd its CYP1A2 metbolite prxnthine my non-selectively bind to denosine receptor nd ct s neuroprotector diminishing somewht the potentil importnce of CYP1A2 enzyme ctivity [8]. The verge Dnish study prticipnts drnk s much s 4 cups per dy over 40 yers implying tht levels of cffeine nd its metbolites might be chroniclly higher thn in other popultions consuming less coffee possibly rendering the contributions of the metbolizing enzyme less importnt. Our study hs severl strengths. We hve lrge smple size with homogenous ncestry; we selected popultion controls from Dnish registers, ssessed confounding (including smoking) extensively, nd were ble to distinguish between incident nd prevlent cses. High coffee consumption in Denmrk llowed us to ssess dose-response reltionships for coffee nd PD with gret sttisticl power nd since we collected detiled informtion on lifetime coffee consumption, we were ble to define exposures in vrious wys. Limittions to this study re tht very few (<10%) prticipnts reported not drinking coffee such tht CYP1A2 enzyme ctivity my not ffect cffeine levels in the blood more thn minimlly mking it hrd to ssess the influence of CYP1A2 polymorphisms. PD-prevlent cses tend to hve more memory loss, nd therefore lifetime coffee consumption could be misreported or reflect chnges in drinking hbits fter dignosis such s due to sleep problems common in PD ptients. Also, recll might lso be impired in ll cses nd controls s the popultion ws on verge 68 yers of ge t the time of interview, which could cuse non-differentil exposure misclssifiction. In conclusion, our study corrobortes previous findings tht interctions between ADORA2A rs , CYP1A2 rs nd rs vrints nd coffee consumption ffect PD risk. However, since our study only found interction between ADORA2A rs nd coffee for mesure of totl cup-yers of coffee consumed but not verge number of coffee cups per dy, which is mesure used in the previous study, we cnnot exclude the possibility tht reverse custion contributed to these results. The lck of cup-per-dy ssocition my, however, lso be explined by the generlly very high coffee consumption levels mong Dnes, tht is, tht few Dnes consumed so little coffee tht ech dditionl cup would mke difference [9]. Therefore, dditionl dt nd studies re still needed in support of the hypothesis tht biologicl effect of cffeine protects ginst PD. Acknowledgement Y.C. ws funded by the Burroughs Wellcome Fund Inter- School Trining Progrm in Chronic Diseses in the pst 12 months. The funding source hd no role in the design, conduct, or nlysis of the study. Reserch reported in this journl ws supported by NIEHS of the Ntionl Institutes of Helth under wrd number R01ES Authors Roles Y.C. nd C.L. nlyzed the dt. Y.C. wrote the first drft. All co-uthors contributed to study concept, design, nd review nd critique the mnuscript. Finncil Disclosures None. Disclosure Sttement The uthors hve no conflicts of interest to declre. Funding This work ws supported by Ntionl Institutes of Helth grnt R01ES References 1 Nefzger MD, Qudfsel FA, Krl VC: A retrospective study of smoking in Prkinson s disese. Am J Epidemiol 1968; 88: Hellenbrnd W, Seidler A, Boeing H, Robr BP, Vieregge P, Nischn P, Joerg J, Oertel WH, Schneider E, Ulm G: Diet nd Prkinson s disese. I: possible role for the pst intke of specific foods nd food groups. Results from self-dministered food-frequency questionnire in cse-control study. Neurology 1996; 47: Coffee, ADORA2A nd CYP1A2 Interction in PD 199

9 3 Hernán MA, Tkkouche B, Cmño-Isorn F, Gestl-Otero JJ: A met-nlysis of coffee drinking, cigrette smoking, nd the risk of Prkinson s disese. Ann Neurol 2002; 52: Kenborg L, Lssen CF, Ritz B, Andersen KK, Christensen J, Schernhmmer ES, Hnsen J, Wermuth L, Rod NH, Olsen JH: Lifestyle, fmily history, nd risk of idiopthic Prkinson disese: lrge Dnish cse-control study. Am J Epidemiol 2015; 181: Ross GW, Abbott RD, Petrovitch H, Morens DM, Grndinetti A, Tung KH, Tnner CM, Mski KH, Blnchette PL, Curb JD, Popper JS, White LR: Assocition of coffee nd cffeine intke with the risk of Prkinson disese. JAMA 2000; 283: Sksjrvi K, Knekt P, Rissnen H, Lksonen MA, Reunnen A, Mnnisto S: Prospective study of coffee consumption nd risk of Prkinson s disese. Eur J Clin Nutr 2008; 62: Qi H, Li S: Dose-response met-nlysis on coffee, te nd cffeine consumption with risk of Prkinson s disese. Geritr Gerontol Int 2014; 14: Xu K, Xu YH, Chen JF, Schwrzschild MA: Neuroprotection by cffeine: time course nd role of its metbolites in the MPTP model of Prkinson s disese. Neuroscience 2010; 167: Popt RA, Vn Den Eeden SK, Tnner CM, Kmel F, Umbch DM, Mrder K, Myeux R, Ritz B, Ross GW, Petrovitch H, Topol B, McGuire V, Costello S, Mnthriprgd AD, Southwick A, Myers RM, Nelson LM: Coffee, ADORA2A, nd CYP1A2: the cffeine connection in Prkinson s disese. Eur J Neurol 2011; 18: Fcheris MF, Schneider NK, Lesnick TG, de Andrde M, Cunninghm JM, Rocc WA, Mrgnore DM: Coffee, cffeine-relted genes, nd Prkinson s disese: cse-control study. Mov Disord 2008; 23: Tn EK, Lu ZY, Fook-Chong SM, Tn E, Shen H, Chu E, Yih Y, Teo YY, Zho Y: Exploring n interction of denosine A2A receptor vribility with coffee nd te intke in Prkinson s disese. Am J Med Genet B Neuropsychitr Genet 2006; 141B: Hill-Burns EM, Hmz TH, Zbetin CP, Fctor SA, Pymi H: An ttempt to replicte interction between coffee nd CYP1A2 gene in connection to Prkinson s disese. Eur J Neurol 2011; 18:e107 e108; uthor reply e Gebhrdt SE, Thoms RG: Nitritive Vlue of Foods. Beltsville, US Deprtment of Agriculture, Agriculturl Reserch Service, Nutrient Dt Lbortory, 2000, p Hmz TH, Chen H, Hill-Burns EM, Rhodes SL, Montimurro J, Ky DM, Tenes A, Kusel VI, Sheehn P, Eswrkhnth M, Yerout D, Smii A, Roberts JW, Agrwl P, Bordelon Y, Prk Y, Wng L, Go J, Vnce JM, Kendler KS, Bcnu SA, Scott WK, Ritz B, Nutt J, Fctor SA, Zbetin CP, Pymi H: Genomewide gene-environment study identifies glutmte receptor gene GRIN2A s Prkinson s disese modifier gene vi interction with coffee. PLoS Genet 2011; 7:e Ahmed I, Lee PC, Lill CM, Serles Nielsen S, Artud F, Gllgher LG, Loriot MA, Mulot C, Ncfer M, Liu T, Biernck JM, Armsu S, Anderson K, Frin FM, Lssen CF, Hnsen J, Olsen JH, Bertrm L, Mrgnore DM, Checkowy H, Ritz B, Elbz A: Lck of repliction of the GRIN2A-by-coffee interction in Prkinson disese. PLoS Genet 2014; 10:e Dnmrks Sttistik: Indvndrere i Dnmrk Publiktioner/VisPub?cid= Friedrich T: Generlised Inference in the Rndom Effects Met Regression Model pckges/metgen/metgen.pdf. 18 Byrne EM, Johnson J, McRe AF, Nyholt DR, Medlnd SE, Gehrmn PR, Heth AC, Mdden PA, Montgomery GW, Chenevix-Trench G, Mrtin NG: A genome-wide ssocition study of cffeine-relted sleep disturbnce: confirmtion of role for common vrint in the denosine receptor. Sleep 2012; 35: Chen JF, Xu K, Petzer JP, Stl R, Xu YH, Beilstein M, Sonsll PK, Cstgnoli K, Cstgnoli N Jr, Schwrzschild MA: Neuroprotection by cffeine nd A(2A) denosine receptor inctivtion in model of Prkinson s disese. J Neurosci 2001; 21:RC Pierri M, Vudno E, Sger T, Englund U: KW-6002 protects from MPTP induced dopminergic toxicity in the mouse. Neurophrmcology 2005; 48: Alsene K, Deckert J, Snd P, de Wit H: Assocition between A2A receptor gene polymorphisms nd cffeine-induced nxiety. Neuropsychophrmcology 2003; 28: Ferre S, Quiroz C, Woods AS, Cunh R, Popoli P, Ciruel F, Lluis C, Frnco R, Azdd K, Schiffmnn SN: An updte on denosine A2Adopmine D2 receptor interctions: implictions for the function of G protein-coupled receptors. Curr Phrm Des 2008; 14: Frnco R, Ferre S, Agnti L, Torvinen M, Gines S, Hillion J, Csdo V, Lledo P, Zoli M, Lluis C, Fuxe K: Evidence for denosine/dopmine receptor interctions: indictions for heteromeriztion. Neuropsychophrmcology 2000; 23(4 suppl):s50 S Gunes A, Dhl ML: Vrition in CYP1A2 ctivity nd its clinicl implictions: influence of environmentl fctors nd genetic polymorphisms. Phrmcogenomics 2008; 9: Bckmn JT, Schroder MT, Neuvonen PJ: Effects of gender nd moderte smoking on the phrmcokinetics nd effects of the CYP1A2 substrte tiznidine. Eur J Clin Phrmcol 2008; 64: Schse C, Brockmoller J, Buer S, Roots I: Functionl significnce of C >A polymorphism in intron 1 of the cytochrome P450 CYP1A2 gene tested with cffeine. Br J Clin Phrmcol 1999; 47: Mellick GD, Ross OA: Cffeine nd Prkinson s disese: re we getting our fix on riskmodifying gene-environment interctions? Eur J Neurol 2011; 18: Chung/Lill/Lee/Hnsen/Lssen/Bertrm/ Greene/Sinsheimer/Ritz