Where should the safe limits of alcohol consumption stand in light of liver enzyme abnormalities in alcohol consumers?

RESEARCH ARTICLE Whe e should he sa e limi s o alcohol consump ion s and in ligh o li e enzyme abno mali ies in alcohol consume s? Onni Niemela ¨ 1 *, Ma kus Niemela ¨ 2 , Ris o Bloigu 3 , Mau i Aal o 4 , Tiina Laa ikainen 5,6 1Depa men o Labo a o y Medicine and Medical Resea ch Uni , Seina ¨joki Cen al Hospi al and Uni e si y o Tampe e, Seina ¨joki, Finland, 2Depa men o Medicine, Uni e si y o Oulu, Oulu, Finland, 3Medical In o ma ics and S a is ics Resea ch G oup, Uni e si y o Oulu, Oulu, Finland, 4Depa men o Psychia y, Seina ¨joki Cen al Hospi al and Uni e si y o Tampe e, Tampe e, Finland, 5Na ional Ins i u e o Heal h and Wel a e (THL), Helsinki, Finland, 6The Ins i u e o Public Heal h and Clinical Nu i ion, Uni e si y o Eas e n Finland, Kuopio, Finland *onni.niemela@epshp. i Abs ac Objec i es To es ima e he p e alence and isk ac o s o abno mal li e enzymes in a la ge age- and gende s a i ied popula ion-based sample o appa en ly heal hy indi iduals wi h o wi hou alcohol consump ion and o he heal h- ela ed isk ac o s (adiposi y, physical inac i i y, smoking). Me hods Da a on alcohol use, smoking, die and physical ac i i y we e eco ded using s uc u ed ques ionnai es om 13,976 subjec s (6513 men, 7463 women, aged 25–74 yea s) in he na ional FINRISK s udies. Alcohol da a was used o ca ego ize he pa icipan s in o abs ain- e s, ligh d inke s, mode a e d inke s and hea y d inke s. Se um gamma-glu amyl ans e - ase (GGT) and alanine amino ans e ase (ALT) ac i i ies we e measu ed using s anda d kine ic me hods. Resul s Male ligh d inke s, mode a e d inke s and hea y d inke s showed signi ican ly highe ela- i e isks o abno mal GGT han abs aine s: 1.37 (95% con idence in e al 1.11 o 1.71, p < 0.01), 2.72 (2.08 o 3.56, p <0.0005), and 6.10 (4.55 o 7.17, p <0.0005), espec i ely. Co - esponding alues o women we e 1.22 (0.99 o 1.51, p = 0.065), 1.90 (1.44 o 2.51, p < 0.0005), and 5.91 (3.80 o 9.17, p <0.0005). Es ima ed h eshold doses o a signi ican GGT ele a ion was 14 s anda d weekly alcohol doses o men and 7 o women. Excess body weigh and age o e 40 yea s modula ed he h esholds owa ds smalle quan i ies o alcohol. The isk o abno mal GGT was also signi ican ly in luenced by physical inac i i y and smoking. The ela i e isks o abno mal ALT ac i i ies we e inc eased in male hea y d inke s, especially in hose p esen ing wi h adiposi y and seden a y li es yle. PLOS ONE | h ps://doi.o g/10.1371/jou nal.pone.0188574 Decembe 5, 2017 1 / 15 a1111111111 a1111111111 a1111111111 a1111111111 a1111111111 OPEN ACCESS Ci a ion: Niemela¨O, Niemela¨M, Bloigu R, Aal o M, Laa ikainen T (2017) Whe e should he sa e limi s o alcohol consump ion s and in ligh o li e enzyme abno mali ies in alcohol consume s? PLoS ONE 12(12): e0188574. h ps://doi.o g/10.1371/ jou nal.pone.0188574 Edi o : John G een, Uni e si y Hospi al Llandough, UNITED KINGDOM Recei ed: June 7, 2017 Accep ed: No embe 9, 2017 Published: Decembe 5, 2017 Copy igh : ©2017 Niemela¨e al. This is an open access a icle dis ibu ed unde he e ms o he C ea i e Commons A ibu ion License, which pe mi s un es ic ed use, dis ibu ion, and ep oduc ion in any medium, p o ided he o iginal au ho and sou ce a e c edi ed. Da a A ailabili y S a emen : All ele an da a a e a ailable o applica ion om FINRISK a h ps:// www. hl. i/en/web/ hl-biobank/ o - esea che s/ apply/. The da a con ain po en ially iden i ying in o ma ion and a e es ic ed by he Pe sonal Da a Ac (Finlex 523/1999). The au ho s did no ecei e any special access p i ileges when applying o he da a. In e es ed, quali ied esea che s may apply o his da a in he same manne as he he au ho s did. Conclusions Alcohol use ma kedly inc eases he isk o abno mal li e enzyme ac i i ies in hose p e- sen ing wi h age o e 40 yea s, obesi y, smoking o seden a y li es yle. The da a should be conside ed in public heal h ecommenda ions and in he de ini ions o sa e limi s o alcohol use. In oduc ion Alcohol consump ion accoun s o a signi ican p opo ion o li e yea s los o disabili y [1–3]. Based on scien i ic e idence abou he ha m ul consequences o alcohol d inking, he de ini- ion o hea y o a - isk d inking has been apidly e ol ing o e he pas wo decades. The da a ga he ed so a has also p omp ed e isions o go e nmen al guidelines in se e al coun ies [2, 4–6]. In cu en socie ies app oxima ely e e y six h adul indi idual ha e been es ima ed o d ink alcohol in amoun s, which exceed he limi s ha a e known o inc ease he isk o mul i- ple medical p oblems (24 s anda d d inks o alcohol pe week o men and 16 d inks o women) [1,4,5]. Howe e , he heal h isks esul ing om consump ion o mo e modes amoun s o alcohol and he issue o sa e limi s o alcohol use ha e emained as ma e s o long- s anding con o e sy. Recen ly, e idence has accumula ed o indica e ha e en ligh o mode a e alcohol d inking may inc ease he isk o ce ain cance s [6–8] and ad e se b ain ou comes in long- e m ol- low-ups [9]. S udies ha e u he indica ed ha he ac i i ies o common li e enzymes, ALT and GGT, may inc ease om baseline as a esul o ela i ely low alcohol d inking le els espe- cially in indi iduals wi h adiposi y [10–13]. Wi h epidemic le els o bo h obesi y and excessi e alcohol in ake doc o s a e cu en ly encoun e ing inc easing numbe s o abno mal li e unc- ion es s which may indica e silen li e disease [2,10,14]. The ea ly changes in he ac i i ies o li e enzymes in such pa ien s may also p edic bo h hepa ic and ex a-hepa ic heal h isks, including me abolic synd ome, and ca dio- o ce eb o ascula e en s [15–17]. Changes in GGT ac i i ies appea o be mechanis ically linked wi h he ac i a ion o oxida i e s ess [18– 20], whe eas se um ALT a he ma ks dis u bed li e cell in eg i y o cellula adap a ion o main ain ene gy homeos asis [21,22]. Recen ad ances in indi idualized medicine ha e also emphasized a mo e sys ema ic use o bioma ke da a o he assessmen o li es yle and die a y in e en ions aimed a educing dis- o de s caused by haza dous d inking o excess body weigh [1,10,23]. A mo e widesp ead use o labo a o y es s has, howe e , been hampe ed by he lack o uni o m de ini ions o bio- ma ke no mal anges o e en he mos commonly used bioma ke s, such as li e enzymes, and lack o knowledge on he sa e limi s o alcohol in ake [24,25]. The e is also a pauci y o s udies obse ing he ela ionships be ween mild o mode a e le els o alcohol d inking and he ea ly-phase changes in li e enzymes in la ge popula ions s a i ied by key co a ia es [10, 25]. In he p esen s udy we collec ed da a om a la ge na ional FINRISK popula ion heal h su - ey o de e mine he ela i e isks o abno mal li e enzyme indings in appa en ly heal hy indi iduals wi h de ailed eco ds on alcohol consump ion, die and o he heal h- ela ed beha - iou , including physical ac i i y. The ecen ly es ablished ULNs o li e enzymes based on abs aine s wi h no mal body weigh we e used as e e ence [11]. The pu pose o he s udy was o o e new pe spec i es on he ela ionships be ween alcohol use, adiposi y, seden a y li e s yle and li e enzyme abno mali ies, which would ha e impo an implica ions o public heal h. Sa e limi s o alcohol consump ion based on li e enzyme abno mali ies PLOS ONE | h ps://doi.o g/10.1371/jou nal.pone.0188574 Decembe 5, 2017 2 / 15 Funding: The au ho s ecei ed no speci ic unding o his wo k. Compe ing in e es s: The au ho s ha e decla ed ha no compe ing in e es s exis . Ma e ials and me hods Pa icipan s Da a was collec ed om a c oss-sec ional popula ion heal h su ey (The Na ional FINRISK s udy) ca ied ou in six di e en geog aphic a eas in Finland in yea s 1997, 2002, and 2007. This su ey was o iginally se ou o examine isk ac o s o ch onic, non-communicable dis- eases and modi iable beha iou al isk ac o s. Fo his pu pose, an age- and gende s a i ied andom sample was d awn om he popula ion egis e acco ding o an in e na ional WHO MONICA (Moni o ing ends and de e minan s in ca dio ascula disease) p o ocol [26]. The clinical examina ions included physical measu emen s, labo a o y es s and de ailed ques ion- nai es encompassing alcohol in ake, cu en heal h s a us, die , smoking, cu en physical ac i - i y, medical his o y and socioeconomic ac o s [26–28]. Body weigh and heigh we e measu ed o he nea es 0.1 kg and 0.1 cm, espec i ely. Body mass index (BMI, kg/m 2 ) was calcula ed as a measu e o ela i e body weigh . Wais ci cum e ence was measu ed o he nea es 0.5 cm be ween he lowes ib and he iliac c es while he subjec was a minimal espi a ion. The subjec s included we e de oid o any clinical signs o li e disease, ischaemic hea o b ain disease, diabe es o abno mal glucose ole ance es , hype ension o ac i e in ec ion a he ime o he s udy. The inal s udy popula ion consis ed o 13,976 appa en ly heal hy indi- iduals: 6513 men and 7463 women (mean age 45 ±13 yea s, ange 25–74 yea s) who com- ple ed he ques ionnai es and a ended he medical examina ions. The esponse a es in he 1997, 2002 and 2007 su eys we e 73.4%, 71.4% and 66.9%, espec i ely. Alcohol consump ion was assessed wi h s uc u ed ques ionnai es including in o ma ion on he ype o be e age consumed and he quan i y and equency o consump ion using a e ospec i e ecall me hod egis e ing consump ion om he pas weeks and one yea p io o blood sampling [28,29]. The amoun o e hanol in di e en be e ages was quan i a ed based on de ined po ion sizes as ollows: egula bee 12 g ams (1/3 L), s ong bee 15.5 g ams (1/3 L), long d ink 15.5 g ams (1/3 L), spi i 12 g ams (4 cL), wine 12 g ams (12 cL) and cide 12 g ams (1/3 L). A dose o 12 g ams o pu e e hanol was conside ed as one s anda d d ink. The da a on alcohol consump ion was subsequen ly used o ca ego ize he popula ion by gende and d inking habi s as ollows: 1. pe sons who epo ed no cu en alcohol consump- ion we e e e ed o as non-d inke s (abs aine s), 2. ligh d inke s consumed be ween 1–13 d inks (men) o 1–6 d inks (women), 3. mode a e d inke s consumed 14–23 d inks (men) o 7–15 d inks (women) and 4. hea y d inke s consumed mo e han 23 d inks (men) o mo e han 15 d inks (women) pe week. In he assessmen o ela i e isks o abno mal li e enzyme ac i i ies in mo e de ail he ca ego y o ligh d inke s was u he sepa a ed in o subg oups o e y ligh d inke s: men <7 (n = 2166), women <3.5 (n = 2092) d inks pe week and ligh d inke s men 7 and <14 (n = 1436), women 3.5 and <7 (n = 1200) d inks pe week. Smoking and co ee consump ion we e assessed wi h a se o s anda dized ques ions and exp essed as he amoun o ciga e es pe day and as he in ake o s anda d se ings o co ee (cups) pe day, espec i ely. Physical ac i i y and he numbe and o al ime used o physical exe cises we e egis e ed using s uc u ed ques ionnai es [28,30]. The da a was used o classi y he popula ion in o he subg oups o 1. mode a e o igo ous ac i i y (o e 4 hou s o ac i i y pe week including b isk unning, walking, c oss-coun y skiing, swimming o o he s enu- ous exe cises) 2. ligh (0.5–4 hou s pe week including walking, cycling, ga dening o o he mode a e ac i i ies), and 3. seden a y ac i i y (less han 0.5 hou s pe week). All su eys we e conduc ed in acco dance wi h he Decla a ion o Helsinki acco ding o he e hical ules o he Na ional Public Heal h Ins i u e. W i en in o med consen was ob ained om all pa icipan s. The app o al o his s udy was ecei ed om he Coo dina ing E hics Commi ee o he Helsinki Hospi al Dis ic . Sa e limi s o alcohol consump ion based on li e enzyme abno mali ies PLOS ONE | h ps://doi.o g/10.1371/jou nal.pone.0188574 Decembe 5, 2017 3 / 15 Labo a o y analyses Se um ALT and GGT we e measu ed by s anda d kine ic me hods ollowing ecommenda- ions o he Eu opean Commi ee o Clinical Labo a o y S anda ds (ECCLS) on an Abbo A chi ec clinical chemis y analyse (Abbo Labo a o ies, Abbo Pa k, IL, USA). S a is ical me hods Values a e exp essed as mean ±SD o mean ±95% con idence in e al (CI). Compa isons in he dis ibu ion o he da a we e ca ied ou wi h a chi squa e es o end and p - es on he equali y o p opo ions using S a a s a is ical da a analysis so wa e (S a aCo p LP, TX, USA). Fo compa isons be ween g oups, S uden ’s - es o ANOVA wi h Tukey’s HSD as a pos hoc es o mul iple ac o s we e used. ANOVA wi h Dunne pos hoc es was used o de e mine he h eshold le els o alcohol doses o ini ia ing a signi ican ele a ion in li e enzymes based on he da a wi h a sepa a e ca ego y o each weekly s anda d d ink. Logis ic eg ession was used o assess whe he d inking s a us and he co a ia es associa ed wi h he abno mal li e enzyme le els. Bo h uni a iable and mul i a iable analyses we e pe o med and he esul s o mul i a iable analyses a e epo ed. In he analyses he pa icipan s wi h missing da a we e lis - wise excluded. The analyses we e ca ied ou wi h IBM SPSS S a is ics 22.0 (A monk, NY: IBM Co p.). A p- alue <0.05 was conside ed s a is ically signi ican . Resul s The main demog aphic cha ac e is ics and li es yle ac o s o he pa icipan s classi ied o sub- g oups acco ding o alcohol consump ion and gende a e summa ized in Table 1. Among men, 26.0% o he popula ion we e abs aine s, 55.3% we e ligh d inke s, 12.0% we e mode a e d inke s and 6.7% we e hea y d inke s. In women he co esponding pe cen ages we e 41.2%, 44.1%, 12.7%, and 2.0%. Table 1. Main cha ac e is ics o he s udy popula ion, as classi ied acco ding o d inking s a us. Men Abs aine s Ligh d inke s Mode a e d inke s Hea y d inke s Alcohol consump ion 0 d inks/week 1–13 d inks/week 14–23 d inks/week 24 d inks/week N (%) 1696 (26.0) 3602 (55.3) 780 (12.0) 435 (6.7) Age, yea s, mean ±SD 46.1 ±13.8 (n = 1696) 44.8 ±12.9 (n = 3602) 44.0 ±11.9 (n = 780) 44.8 ±11.3 (n = 435) BMI 26.3 ±3.8 (n = 1554) 26.2 ±3.5 (n = 3279) 26.6 ±3.9 (n = 699) 26.9 ±4.4 (n = 382) Wais ci cum e ence, cm 92.8 ±10.9 (n = 1554) 92.6 ±10.2 (n = 3267) 94.2 ±11.2 (n = 696) 96.3 ±12.4 (n = 380) Smoking, ciga e es/day 4.3 ±8.8 (n = 1685) 4.4 ±8.1 (n = 3562) 8.0 ±10.2 (n = 773) 11.4 ±12.6 (n = 426) Co ee, cups/day 4.8 ±3.6 (n = 1670) 4.7 ±3.1 (n = 3571) 5.1 ±3.4 (n = 780) 4.9 ±4.0 (n = 430) Physical ac i i y, numbe o exe cises/week 2.5 ±2.2 (n = 600) 2.3 ±1.9 (n = 1366) 2.0 ±2.1 (n = 328) 2.0 ±2.3 (n = 166) Women Abs aine s Ligh d inke s Mode a e d inke s Hea y d inke s Alcohol consump ion 0 d inks/week 1–6 d inks/week 7–15 d inks/week 16 d inks/week N (%) 3072 (41.2) 3292 (44.1) 947 (12.7) 152 (2.0) Age, yea s, mean ±SD 44.3 ±13.5 (n = 3072) 43.8 ±12.2 (n = 3292) 42.4 ±11.4 (n = 947) 45.4 ±10.8 (n = 152) BMI 25.8 ±4.8 (n = 2818) 25.1 ±4.4 (n = 3022) 24.9 ±4.0 (n = 851) 25.6 ±4.1 (n = 125) Wais ci cum e ence, cm 82.0 ±12.1 (n = 2722) 80.5 ±11.3 (n = 3010) 80.6 ±10.9 (n = 849) 83.6 ±10.8 (n = 125) Smoking, ciga e es/day 2.0 ±5.3 (n = 3059) 2.1 ±5.2 (n = 3274) 3.9 ±6.4 (n = 937) 8.7 ±9.0 (n = 149) Co ee, cups/day 3.7 ±2.6 (n = 3038) 3.7 ±2.5 (n = 3281) 3.8 ±2.5 (n = 940) 3.8 ±2.4 (n = 152) Physical ac i i y, numbe o exe cises/week 2.5 ±2.1 (n = 1146) 2.6 ±2.1 (n = 1300) 2.3 ±1.9 (n = 394) 2.7 ±2.4 (n = 62) BMI, body mass index h ps://doi.o g/10.1371/jou nal.pone.0188574. 001 Sa e limi s o alcohol consump ion based on li e enzyme abno mali ies PLOS ONE | h ps://doi.o g/10.1371/jou nal.pone.0188574 Decembe 5, 2017 4 / 15 The uppe no mal limi s used o ALT and GGT we e based on p e iously desc ibed alues de ined by calcula ing 97.5 h pe cen iles o he da a based on no mal weigh non-d inke s [11]. The amoun o alcohol (mean ±SD) consumed in hose exceeding he ULNs in GGT analyses in he p esen popula ion was 170 ±206 g ams pe week o men and 55 ±78 g ams pe week o women (Table 2). Fo ALT, he co esponding le els we e 144 ±194 g ams pe week o men and 42 ±63 g ams pe week o women (Table 2). Fig 1 shows he equencies o abno mal li e enzymes om he di e en subg oups classi- ied acco ding o d inking s a us. In men, abno mal GGT indings we e signi ican ly mo e e- quen in ligh d inke s (10.9%) (p <0.01), mode a e d inke s (21.8%) (p <0.001) and hea y d inke s (40.6%) (p <0.001) han in abs aine s (8.7%). In women, he likelihood o abno mal GGT indings was inc eased in mode a e d inke s (10.6%) (p <0.001) and in hea y d inke s (31.2%) (p <0.001) when compa ed o abs aine s (6.7%) o ligh d inke s (7.2%) (p <0.001 o each compa ison). Fo ALT, he g oup o hea y d inke s showed abno mal alues mo e equen ly han he g oup o abs aine s (p <0.001 o men, p <0.01 o women) (Fig 1). In he analyses o he incidences o abno mal li e enzyme alues based on alcohol con- sump ion epo ed om he pas 12 mon hs, he es ima es o weekly alcohol consump ion we e ound o be in good ag eemen wi h he es ima es ob ained om he da a on mo e ecen d inking ( = 0.664, p <0.0001). The incidences o ele a ed li e enzymes we e, howe e , highe in he subg oups classi ied as hea y d inke s based on he la e app oach (Fig 1) han in he co esponding subg oups o med based on 12-mo da a (GGT: 31.4% men, 26.5% women; ALT 21.1% men, 12.8% women). No signi ican di e ences occu ed in he co e- sponding compa isons o abs aine s, ligh d inke s o mode a e d inke s. The indi iduals who could be classi ied as o me d inke s (n = 314, 159 men, 155 women) epo ing p e ious alco- hol consump ion bu no alcohol consump ion om he pas 12 mon hs showed low a es o ele a ed li e enzyme alues (GGT: 7.1% men, 2.4% women; ALT 6.0% men, 1.4% women). The es ima ed h eshold alcohol doses o ini ia ing a signi ican ele a ion in GGT ac i i ies we e 14 s anda d d inks o weekly alcohol consump ion o men and 7 d inks o women (Fig 2). Excess body weigh and inc easing age we e ound o modula e he h esholds owa ds smalle quan i ies (Fig 3). Table 3 summa izes he mul i a iable ela i e isks o abno mal GGT and ALT indings acco ding o d inking habi s, age, wais ci cum e ence, physical ac i i y and smoking. When compa ed wi h abs aine s, male ligh d inke s, mode a e d inke s and hea y d inke s had ela- i e isks o abno mal GGT ac i i ies o 1.37 (95% con idence in e al 1.11 o 1.71, p <0.01), 2.72 (2.08 o 3.56, p <0.0005), and 6.10 (4.55 o 7.17, p <0.0005), espec i ely. Co esponding alues o women we e 1.22 (0.99 o 1.51, p = 0.065), 1.90 (1.44 o 2.51, p <0.0005), and 5.91 (3.80 o 9.17, p <0.0005). When he ca ego y o ligh d inke s was u he sepa a ed in o sub- g oups o e y ligh d inke s (men <7, women <3.5 d inks pe week) and ligh d inke s Table 2. Mean alcohol consump ion (g ams pe week) in g oups acco ding o labo a o y es ULN. <ULN ULN mean ±SD mean ±SD p GGT Men 82.1 ±110.1 170.3 ±206.3 <0.0005 Women 33.7 ±50.8 54.6 ±78.4 <0.0005 ALT Men 91.1 ±122.4 144.0 ±193.9 <0.0005 Women 35.1 ±50.4 41.8 ±62.9 0.160 GGT, gamma-glu amyl ans e ase; ALT, alanine amino ans e ase; ULN, uppe limi o no mal h ps://doi.o g/10.1371/jou nal.pone.0188574. 002 Sa e limi s o alcohol consump ion based on li e enzyme abno mali ies PLOS ONE | h ps://doi.o g/10.1371/jou nal.pone.0188574 Decembe 5, 2017 5 / 15 (men 7 and <14, women 3.5 and <7 d inks pe week) he ela i e isks o abno mal ac i i ies we e no ound o be inc eased in any o he compa isons among e y ligh d inke s (da a no shown). Fig 1. Incidence o abno mal GGT and ALT ac i i ies in g oups classi ied acco ding o d inking s a us. h ps://doi.o g/10.1371/jou nal.pone.0188574.g001 Fig 2. Es ima ed h eshold doses o alcohol consump ion (s anda d d inks /week) o ini ia ing a signi ican GGT ac i a ion. The le els leading o GGT inc eases especially in hose abo e 40 yea s o age a e ma kedly lowe han he cu en limi s o hea y d inking in many Wes e n coun ies (men: 24 d inks, women: 16 d inks) (dashed line). h ps://doi.o g/10.1371/jou nal.pone.0188574.g002 Sa e limi s o alcohol consump ion based on li e enzyme abno mali ies PLOS ONE | h ps://doi.o g/10.1371/jou nal.pone.0188574 Decembe 5, 2017 6 / 15 The isk o abno mal GGT was ound o be signi ican ly in luenced by age, adiposi y, physi- cal ac i i y and smoking (Table 3). Fo ALT, he ela i e isks o abno mal ac i i ies we e inc eased only in male hea y d inke s. Adiposi y and seden a y li es yle also inc eased he like- lihood o abno mal ALT le els (Table 3). The indi iduals wi h low o seden a y physical ac i - i y showed ma kedly highe ela i e isks o ele a ed li e enzymes han hose engaged in physically demanding ac i i ies o a leas 3 hou s pe week. Fig 3. Se um GGT and ALT in subg oups classi ied acco ding o d inking s a us, age and BMI. Da a a e shown as median and in e qua ile anges. h ps://doi.o g/10.1371/jou nal.pone.0188574.g003 Sa e limi s o alcohol consump ion based on li e enzyme abno mali ies PLOS ONE | h ps://doi.o g/10.1371/jou nal.pone.0188574 Decembe 5, 2017 7 / 15 The analysis o he associa ions be ween co ee consump ion and he isk o ele a ed li e enzyme alues showed ha GGT ac i i ies in hose classi ied as hea y d inke s (whe eas no in o he d inking ca ego ies) we e also signi ican ly in luenced by co ee d inking. Hea y d ink- e s consuming 4 cups o co ee showed signi ican ly lowe GGT le els (men 65 ±74 U/L, women 41 ±76 U/L) han he co esponding g oups consuming 1–3 cups (men 91 ±90 U/L, women 62 ±120 U/L) o no co ee (men 120 ±171 U/L, women 76 ±95 U/L) (p <0.001 o compa isons in bo h gende s). Discussion In his la ge c oss-sec ional popula ion-based sample o appa en ly heal hy indi iduals, we showed ha he ela i e isk o abno mal li e enzyme ac i i ies is inc eased in indi iduals consuming alcohol in amoun s conside ed as ligh o mode a e d inking le els. In ligh o Table 3. Rela i e isks o abno mal li e enzyme le els in s udy subg oups om a c oss-sec ional FINRISK popula ion su ey. GGT ALT Men Women Men Women Mul i a iable ela i e isk p Mul i a iable ela i e isk p Mul i a iable ela i e isk p Mul i a iable ela i e isk p D inking s a us Abs aine s Ligh d inke s 1.37 (1.11 o 1.71) 0.004 1.22 (0.99 o 1.51) 0.065 1.09 (0.78 o 1.53) 0.614 1.08 (0.76 o 1.52) 0.666 Mode a e d inke s 2.72 (2.08 o 3.56) <0.0005 1.90 (1.44 o 2.51) <0.0005 1.19 (0.74 o 1.90) 0.481 1.27 (0.78 o 2.07) 0.344 Hea y d inke s 6.10 (4.55 o 8.17) <0.0005 5.91 (3.80 o 9.17) <0.0005 3.10 (1.86 o 5.15) <0.0005 2.23 (0.94 o 5.33) 0.070 Age g oups <40 y s 40–49 y s 1.27 (1.03 o 1.57) 0.026 1.75 (1.34 o 2.29) <0.0005 0.91 (0.66 o 1.26) 0.563 0.96 (0.63 o 1.45) 0.838 50–64 y s 1.23 (1.00 o 1.51) 0.051 2.70 (2.10 o 3.47) <0.0005 0.35 (0.24 o 0.51) <0.0005 1.36 (0.93 o 1.97) 0.109 >64 y s 0.65 (0.46 o 0.93) 0.019 2.83 (1.99 o 4.02) <0.0005 0.17 (0.07 o 0.44) <0.0005 0.39 (0.14 o 1.10) 0.074 Wais ci cum e ence Low isk * High isk ** 2.67 (2.19 o 3.27) <0.0005 1.54 (1.20 o 1.97) 0.001 3.08 (2.19 o 4.34) <0.0005 1.69 (1.12 o 2.54) 0.012 Ve y high isk *** 5.07 (4.12 o 6.24) <0.0005 3.07 (2.45 o 3.84) <0.0005 6.00 (4.26 o 8.46) <0.0005 3.28 (2.25 o 4.78) <0.0005 Physical ac i i y Mode a e/ igo ous ( e e ence) Ligh 1.37 (1.11 o 1.69) 0.004 1.46 (1.12 o 1.89) 0.005 1.49 (1.03 o 2.15) 0.033 0.85 (0.58 o 1.25) 0.406 Seden a y 1.51 (1.18 o 1.93) 0.001 1.55 (1.15 o 2.10) 0.004 1.96 (1.31 o 2.93) 0.001 0.83 (0.51 o 1.33) 0.437 Smoking No smoking 1–9 ciga e es/day 1.03 (0.74 o 1.44) 0.869 1.26 (0.90 o 1.77) 0.179 1.26 (0.78 o 2.03) 0.350 1.02 (0.59 o 1.77) 0.954 10–19 ciga e es/ day 1.24 (0.97 o 1.58) 0.093 1.22 (0.91 o 1.65) 0.189 1.10 (0.73 o 1.65) 0.651 0.69 (0.40 o 1.19) 0.183 20–29 ciga e es/ day 1.61 (1.26 o 2.05) <0.0005 1.12 (0.70 o 1.79) 0.635 0.79 (0.51 o 1.24) 0.303 0.40 (0.12 o 1.30) 0.128 30 ciga e es/day 1.56 (1.06 o 2.30) 0.024 2.05 (0.87 o 4.86) 0.102 0.81 (0.40 o 1.63) 0.549 4.14 (1.01 o 16.92) 0.048 GGT, gamma-glu amyl ans e ase; ALT, alanine amino ans e ase *, men <94 cm, women <80 cm **, men 94–102 cm, women 80–88 cm ***, men >102 cm, women >88 cm h ps://doi.o g/10.1371/jou nal.pone.0188574. 003 Sa e limi s o alcohol consump ion based on li e enzyme abno mali ies PLOS ONE | h ps://doi.o g/10.1371/jou nal.pone.0188574 Decembe 5, 2017 8 / 15 ecen indings indica ing ha he ea ly changes in he ac i i ies o GGT and ALT should also be ega ded as impo an p ognos ic ma ke s o bo h hepa ic and ex a-hepa ic heal h isks [17,19,22], i may be assumed ha he isk ca ego ies o alcohol d inking and associa ed indi- idual heal h- ela ed beha iou need u he a en ion. The la ge numbe o well-cha ac e ized s udy subjec s o e ed us wi h he possibili y o pe - o m a comp ehensi e analysis o he ela i e isks o abno mal bioma ke indings ac oss all le els o alcohol consump ion and o he ac o s o li es yle. Ou s udy elied la gely on he da a ob ained om common li e enzymes, ALT and GGT, bo h o which a e eadily induced by excessi e alcohol consump ion and a e associa ed wi h ele an clinical ou comes. These li e enzymes may also be conside ed bioma ke s o choice since hey a e easily measu able, cos - e ec i e and easy o in e p e by clinicians. Al hough cu en p og ess in clinical labo a o ies and quali y con ol egimes has led o mo e consis en measu emen p ocedu es, se e al p ac- ical p oblems ha e emained un esol ed including de ini ions o uni e sal no mal anges. The ULNs o e en he mos commonly used bioma ke s o li e s a us, ALT and GGT, cu - en ly show signi ican a ia ion be ween indi idual labo a o ies and geog aphic a eas [11,25, 31,32]. Thus, he e is a con inuing need o addi ional bioma ke alida ion o ansla e he nume ical alues in o clinically meaning ul in o ma ion, which would also allow mo e e icien in e na ional compa isons o he da a. O e he pas decades educ ion o ha m ul alcohol d inking has been an impo an a ge o public heal h policies. Al hough hea y alcohol use is known o be causally linked o o e 60 dis inc diseases, he associa ion be ween ligh o mode a e d inking and he o e all isk o heal h p oblems has emained unclea [1,5]. The concep s o bene icial, sa e o ha m ul le els o e hanol in ake ha e also emained as opics o g ea in e es and li ely deba e. While o he s udies ha e sugges ed ha ligh o mode a e d inking is associa ed wi h bene icial heal h e ec s especially in hose ollowing a Medi e anean die [33], o he g oups o in es iga o s ha e no eached simila conclusions [34–36]. The ques ions on he dose-e ec ela ionships be ween e hanol in ake, bioma ke changes and possible issue inju y ha e also emained un esol ed. The p esen da a suppo s he iew ha biochemical esponses in he sequence o e en s leading o heal h p oblems may be expec ed o occu e en in associa ion wi h ligh o mode a e d inking le els especially in hose wi h p ecipi a ing isk ac o s. Recen indings by o he g oups o in es iga o s ha e also emphasized he iew ha ligh o mode a e d inking could be associa ed wi h an ele a ed isk o cance [6–8], a ial ib illa ion [37], le en icula dias olic dys unc ion [38], ad e se b ain ou comes [9] and an inc ease in all-cause mo ali y [39]. Al hough educ ions in alcohol-a ibu able disease bu den may ob iously be expec ed o be achie ed h ough a mo e e icien iden i ica ion o high- isk indi iduals, alcohol-consum- ing pa ien s seem a ely ecei e speci ic in e en ion in cu en clinical p ac ice unless he si u- a ion is complica ed by como bidi y. The p esen da a indica es ha he in luence o e hanol on li e enzymes is signi ican ly d i en by ac o s such as gende , age, excess body weigh , physical inac i i y o smoking, which should all be aken in o accoun when discussing indi idually on he mos app op ia e le els o d inking. Al hough he p ima y mechanisms unde lying such obse a ions emain unknown a his ime i is possible ha all hese condi ions s imula e oxida i e s ess in an addi i e, gende - and age-dependen manne . GGT enzyme plays a pi o al ole in he me abo- lism o glu a hione (GSH), and ele a ed ac i i ies could sign a need o main ain in acellula GSH le els du ing oxida i e s ess [11,15,40–44]. While alcohol use and ela ed ha m a e mo e p e alen in men, women seem o show ele a ed li e enzyme ac i i ies ollowing con- sump ion o smalle amoun s o alcohol. In acco dance wi h p e ious indings, alcohol con- sump ion in hose abo e 40 yea s o age appea s o pose a highe isk owa ds ele a ed GGT le els [23], which could indica e a g ea e sensi i i y owa ds oxida i e s ess upon aging o Sa e limi s o alcohol consump ion based on li e enzyme abno mali ies PLOS ONE | h ps://doi.o g/10.1371/jou nal.pone.0188574 Decembe 5, 2017 9 / 15