Association between exposure to air pollution and blood lipids in the general population of Spain

Eu J Clin In es . 2024;54:e14101. | 1 o 11 h ps://doi.o g/10.1111/eci.14101 wileyonlinelib a y.com/jou nal/eci Recei ed: 27 July 2023 | Accep ed: 23 Sep embe 2023 DOI: 10.1111/eci.14101 ORIGINAL ARTICLE Associa ion be ween exposu e o ai pollu ion and blood lipids in he gene al popula ion o Spain Se gioValdés1,2 | ViyeyDoula am- Gamga am1 | C is inaMaldonado- A aque1,2 | E aGa cía- Escoba 1,2 | Sa aGa cía- Se ano1,2 | WasimaOualla- Bachi i1,2 | Ma aGa cía- Vi anco3 | Juan LuisGa ido3 | Vic o iaGil3 | Fe nandoMa ín- Llo en e3 | Al onsoCalle- Pascual2,4 | LuisCas año2,5,6 | ElíasDelgado7 | Edelmi oMenéndez7 | JosepF anch- Nadal2,8 | SoniaGaz ambide2,6,9 | JoanGi bés10 | F. Ja ie Cha es2,11 | José L.Galán- Ga cía12 | Gab ielAguile a- Venegas12 | Joan Ca lesVall é2,13 | Nú iaAmigó2,14,15 | Mon seGua diola2,13 | JosepRibal a2,13 | GemmaRojo- Ma ínez1,2 1Depa men o Endoc inology and Nu i ion, Hospi al Regional Uni e si a io de Málaga/Uni e sidad de Málaga, Ins i u o de In es igación Biomedica de Málaga- IBIMA, Málaga, Spain 2Cen o de In es igación Biomédica en Red de Diabe es y En e medades Me abólicas Asociadas (CIBERDEM), Ins i u o de Salud Ca los III, Mad id, Spain 3Cen o de In es igaciones Ene gé icas, Medioambien ales y Tecnológicas (CIEMAT) – Di isión de Con aminación A mos é ica, Mad id, Spain 4Depa men o Endoc inology and Nu i ion and Ins i u o de In es igación Sani a ia Uni e si y Hospi al S. Ca los (IdISSC), Depa men Medicine II, Uni e sidad Complu ense (UCM), Mad id, Spain 5Hospi al Uni e si a io C uces, BioC uces, UPV/EHU, Ba akaldo, Spain 6Cen o de In es igación Biomédica en Red de En e medades Ra as (CIBERER), Ins i u o de Salud Ca los III, Mad id, Spain 7Depa men o Endoc inology and Nu i ion, Hospi al Uni e si a io Cen al de As u ias/Uni e si y o O iedo, Ins i u o de In es igación Sani a ia del P incipado de As u ias (ISPA), O iedo, Spain 8EAP Ra al Sud, Ins i u Ca alà de la Salu , Red GEDAPS, P ima y Ca e, Uni a de Supo a la Rece ca (IDIAP – Fundació Jo di Gol), Ba celona, Spain 9Depa men o Endoc inology and Nu i ion, Hospi al Uni e si a io C uces – BioC uces Bizkaia – UPV- EHU, Ba acaldo, Ba celona, Spain 10Diabe es Uni , Hospi al A nau de Vilano a, Valencia, Spain 11Genomic S udies and Gene ic Diagnosis Uni , Fundación de In es igación del Hospi al Clínico de Valencia – INCLIVA, Valencia, Spain 12Depa men o Applied Ma hema ics, Uni e si y o Málaga, Málaga, Spain 13Resea ch Uni on Lipids and A he oscle osis, San Joan Uni e si y Hospi al, Ro i a i Vi gili Uni e si y, IISPV, Reus, Spain 14Me abolomics Pla o m, Uni e si a Ro i a i Vi gili, IISRV, Reus, Spain 15Bios e Teslab, Reus, Spain This is an open access a icle unde he e ms o he C ea i e Commons A ibu ion-NonComme cial License, which pe mi s use, dis ibu ion and ep oduc ion in any medium, p o ided he o iginal wo k is p ope ly ci ed and is no used o comme cial pu poses. © 2023 The Au ho s. Eu opean Jou nal o Clinical In es iga ion published by John Wiley & Sons L d on behal o S ich ing Eu opean Socie y o Clinical In es iga ion Jou nal Founda ion. Valdés Se gio and Doula am- Gamga am Viyey ha e con ibu ed equally o his wo k and sha e i s au ho ship. Josep Ribal a and Gemma Rojo- Ma ínez ha e con ibu ed equally o his wo k and sha e las au ho ship. 2 o 11 | VALDÉS e al. 1 | BACKGROUND The Wo ld Heal h O ganiza ion (WHO) has iden i ied ai pollu ion as he la ges single en i onmen al heal h isk wo ldwide, wi h ou doo ai pollu ion accoun ing o mo e han 4.2 million dea hs e e y yea .1 Nea ly hal o ou doo ai pollu ion- ela ed p ema u e dea hs a e due o ischaemic hea disease and s oke.1 Ai pollu ion can impac he ca dio ascula sys em h ough a numbe o mechanisms, including endo helial dys unc ion, sys- emic and pulmona y oxida i e s ess and in lamma ion, au onomic ne ous sys em dys unc ion and epigene ic changes.2– 5 Al e a ions in lipid me abolism could be ano he po en ial pa hway in he associa ion be ween ai pollu ion and a e ioscle osis. In his ega d, se - e al p e ious epidemiological s udies ha e explo ed he ela ionships be ween he exposu e o se e al ambien ai pollu an s and blood lipid le els and he p esence o dyslipidaemia.6– 24 Howe e , mos o his e idence has elied on s anda d lipid measu es. S anda d lipid pan- els measu e he choles e ol o iglyce ide con en (in concen a ion pe decili e) ca ied by each lipop o ein class, a he han he numbe s o hese pa icles. In con- as , NMR spec oscopy p o ides a di ec assessmen o he numbe and size o lipop o ein pa icles, p o iding addi ional in o ma ion abou ca dio ascula disease (CVD) isk.25– 29 Acco dingly, in he p esen s udy, we aimed o assess he associa ions be ween he exposu e o ai pollu an s [pa icles wi h an ae odynamic diame e o less han 10 mic ons (PM10), pa icles wi h an ae odynamic diame- e o less han 2.5 mic ons (PM2.5) and ni ogen dioxide (NO2)], wi h bo h he blood lipid le els o a s anda d lipid p o ile as well as he pa icle concen a ions o lipop o- eins, in a na ionwide sample ep esen a i e o he adul popula ion o Spain. Co espondence Se gio Valdés, Depa men o Endoc inology and Nu i ion, Hospi al Uni e si a io Regional de Málaga Plaza del Hospi al Ci il s/n, 29009 Málaga, Spain. Email: se gio. [email p o ec ed] Funding in o ma ion Conseje ía de Salud y Bienes a Social, Jun a de Andalucía, G an /Awa d Numbe : RC- 0006- 2016; Ins i u o de Salud Ca los III, G an /Awa d Numbe : CM21/00214, INT21/00037, PI17/02136 and PI20/01322 Abs ac Backg ound and Aims: We aimed o assess he associa ions o exposu e o ai pollu an s and s anda d and ad anced lipop o ein measu es, in a na ionwide sample ep esen a i e o he adul popula ion o Spain. Me hods: We included 4647 adul s (>18 yea s), pa icipan s in he na ional, c oss- sec ional, popula ion- based [email p o ec ed] s udy, conduc ed in 2008– 2010. S anda d lipid measu emen s we e analysed on an A chi ec C8000 Analyze (Abbo Labo a o ies SA). Lipop o ein analysis was made by an ad anced 1H- NMR lipop o ein es (Liposcale®). Pa icipan s we e assigned ai pollu ion con- cen a ions o pa icula e ma e <10 μm (PM10), <2.5 μm (PM2.5) and ni ogen dioxide (NO2), co esponding o he heal h examina ion yea , ob ained by mod- elling combined wi h measu emen s aken a ai quali y s a ions (CHIMERE chemis y- anspo model). Resul s: In mul i a ia e linea eg ession models, each IQR inc ease in PM10, PM2.5 and NO2 was associa ed wi h 3.3%, 3.3% and 3% lowe le els o HDL- c and 1.3%, 1.4% and 1.1% lowe HDL pa icle (HDL- p) concen a ions (p < .001 o all associa ions). In mul i a ia e logis ic eg ession, he e was a signi ican associa- ion be ween PM10, PM2.5 and NO2 concen a ions and he odds o p esen ing low HDL- c (<40 mg/dL), low HDL- p (<p25) and highe LDL pa icle (LDL- p) concen a ions (≥p75). In subg oup analyses he e we e s onge associa ions be- ween PM10 and NO2 and low HDL- p in men (p o in e ac ion .008 and .034), and be ween NO2 and low HDL- p in indi iduals wi h obesi y (p o in e ac ion .015). Conclusions: Ou s udy shows an associa ion be ween he exposu e o ai pol- lu an s and blood lipids in he gene al popula ion o Spain, sugges ing a link o a he oscle osis. KEYWORDS ai pollu ion, choles e ol, HDL, lipids, lipop o ein, iglyce ide 13652362, 2024, 2, Downloaded om h ps://onlinelib a y.wiley.com/doi/10.1111/eci.14101 by Readcube (Lab i a Inc.), Wiley Online Lib a y on [13/03/2024]. See he Te ms and Condi ions (h ps://onlinelib a y.wiley.com/ e ms-and-condi ions) on Wiley Online Lib a y o ules o use; OA a icles a e go e ned by he applicable C ea i e Commons License | 3 o 11 VALDÉS e al. 2 | METHODS 2.1 | S udy design, se ing and popula ion The [email p o ec ed] s udy is a na ional, c oss- sec ional, popula ion- based su ey conduc ed in 2008– 2010.30 A clus e sampling design was used o selec pa icipan s o o m a ep esen a i e andom sample o he Spanish pop- ula ion. One hund ed heal h cen es o hei equi alen om all a ound he coun y we e selec ed a andom, wi h a p obabili y o selec ion p opo ional o hei a ge pop- ula ion size, a e which 100 indi iduals aged ≥18 yea s we e andomly selec ed om each heal h cen e. O he eligible adul s, 55.8% came o examina ion, o which 9.9% we e excluded (ins i u ionalized, se e e disease, p eg- nancy o ecen deli e y), esul ing in a inal sample o 5072 indi iduals. The p esen s udy ocuses on 4647 indi iduals (92% o he s udy sample), in whom comple e da a on concen a- ions o ai pollu an s and blood lipids we e a ailable o analyses. This esea ch was ca ied ou in acco dance wi h he Decla a ion o Helsinki o he Wo ld Medical Associa- ion.31 W i en in o med consen was ob ained om all he pa icipan s. The s udy was app o ed by he E hics and Clinical In es iga ion Commi ee o he Hospi al Regional Uni e si a io de Málaga (Málaga, Spain) in ad- di ion o o he egional e hics and clinical in es iga ion commi ees all o e Spain. 2.2 | Va iables and p ocedu es The pa icipan s we e in i ed o a end a single examina- ion isi a hei heal h cen e. In o ma ion was collec ed by means o an in e iewe - adminis e ed s uc u ed ques ionnai e, ollowed by a physical examina ion and blood sampling. In o ma ion on age, gende , educa ional le el (none/basic/high school/college), smoking habi (cu en , o me o ne e smoke s) and alcohol in ake (<30/30– 60/>60 se ings pe mon h), was ob ained by ques ionnai e. Food consump ion was de e mined by a ood equency ques ionnai e and adhe ence o he Medi- e anean die was es ima ed by an adap a ion o a 14- i em Medi e anean die sco e (MedSco e).32 The le el o daily physical ac i i y was es ima ed by he sho o m o he In e na ional Physical Ac i i y Ques ionnai e (IPAQ).33 Weigh and heigh we e measu ed and he body mass index (BMI) was calcula ed as weigh (kg)/heigh (m)2. Medical his o y and medica ions we e also eco ded. Blood samples, ob ained in as ing condi ions, we e immedia ely cen i uged and he se um was ozen un il analysis. Samples we e managed by he biochemis y lab- o a o y o he Hospi al Regional Uni e si a io de Málaga, he IBIMA Biobank and by he CIBERDEM Bio eposi o y (IDIBAPS Biobank). 2.3 | Lipid measu emen s S anda d lipid measu emen s Se um le els (mg/dL) o iglyce ides (TG), o al choles e ol (TC) and high- densi y lipop o ein choles e ol (HDL- c) we e measu ed on an A chi ec C8000 Analyze (Abbo Labo a o ies SA). Low- densi y lipop o ein choles e ol (LDL- c) was es ima ed by he F iedewald o mula. ‘High o al choles- e ol’ was de ined as TC ≥ 240 mg/dL; ‘high LDL- c’ was de ined as LDL- c ≥ 160 mg/dL; ‘low HDL- c’ was de ined as HDL- c < 40 mg/dL; and hype iglyce idaemia was de- ined as TG ≥ 200 mg/dL.34 Lipop o ein analysis Lipop o ein analysis was pe - o med by using Liposcale® Tes (CE, ISO 13.485 app o ed), an ad anced lipop o ein es based on wo- dimensional (2D) di usion- o de ed 1H- NMR (P o on nu- clea magne ic esonance) spec oscopy.35 Be o e 1H- NMR analysis, 200 μL o se um we e dilu ed wi h 50 μL deu e - a ed wa e and 300 μL o 50 mM phospha e bu e solu ion a pH 7.4. 1H- NMR spec a we e eco ded a 306K on a B uke A ance III 600 spec ome e ope a ing a a p o on equency o 600.20 MHz. The me hyl signal was decon o- lu ed by using Lo en zian unc ions o de e mine he lipid concen a ion o he main lipop o ein classes (VLDL, LDL and HDL), and hei size associa ed di usion coe icien s. Then, he lipid concen a ions we e combined wi h hei associa ed pa icle olume in o de o quan i y he num- be o pa icles equi ed o anspo he measu ed lipid concen a ion o each lipop o ein subclass calcula ing he pa icle concen a ions o VLDL (VLDL- p), LDL (LDL- p) and HDL (HDL- p). The a ia ion coe icien s o pa icle numbe we e be ween 2% and 4%, and o he pa icle sizes we e lowe han 0.3%. These analyses we e pe o med a he Bios e Teslab acili ies. We classi ied pa icipan s as ha ing no mal o abno mal le els o LDL- p, VLDL- p and HDL- p, using he equi alen s o he 75 h (LDL- p and VLDL- p) and 25 h (HDL- p) pe cen iles (p) o pa icle con- cen a ions wi hin ou s udy popula ion. 2.4 | Exposu e assessmen Modelled mean annual PM10, PM2.5 and NO2 concen a- ions in Spain o he pe iod 2008– 2010 we e calcula ed wi h he CHIMERE chemis y- anspo model.36 This model calcula es he concen a ion o gaseous species and bo h ino ganic and o ganic ae osols o p ima y and 13652362, 2024, 2, Downloaded om h ps://onlinelib a y.wiley.com/doi/10.1111/eci.14101 by Readcube (Lab i a Inc.), Wiley Online Lib a y on [13/03/2024]. See he Te ms and Condi ions (h ps://onlinelib a y.wiley.com/ e ms-and-condi ions) on Wiley Online Lib a y o ules o use; OA a icles a e go e ned by he applicable C ea i e Commons License 4 o 11 | VALDÉS e al. seconda y o igin, including p ima y pa icula e ma e , mine al dus , sulpha e, ni a e, ammonium, seconda y o ganic species and wa e . This model has been b oadly e alua ed in Spain by compa ison wi h measu ed ai pol- lu an s a a la ge se o moni o ing si es.37,38 The model was applied o a domain co e ing he Ibe ian Peninsula a a ho izon al esolu ion o 0.1 × 0.1°. The modelled concen- a ions we e co ec ed wi h obse ed alues, by conside - ing a me hodology desc ibed by Ma ín e al.39 in which (1) a bias is calcula ed wi h ega d o he obse a ions in he Spanish ai quali y ne wo k o moni o ing si es, (2) hese biases a e spa ially in e pola ed using a k iging me hodol- ogy o ob ain a g idded bias and (3) his g idded bias is applied o he modelled concen a ion g id. This me hod- ology conside s a di e en bias g id o u al and u ban si es ha a e hen combined and weigh ed by popula ion densi y. We assigned he a e age annual exposu e o ai pollu an s co esponding o he heal h examina ion yea o each pa icipan by in e pola ing he es ima ed concen- a ions o he cen oid o hei esiden ial pos al codes. Da a on mean annual empe a u e (°C) om each mu- nicipali y o esidence we e ob ained om he Spanish Na ional Me eo ological Agency.40 2.5 | S a is ical analysis We applied linea eg ession models o assess associa- ions be ween ai pollu an and lipid measu emen s, which we e log ans o med o no malize dis ibu ions and also o limi he in luence o ex eme alues. Asso- cia ion es ima es we e p esen ed as pe cen changes wi h co esponding 95% con idence in e als (calcula ed by 100 × [exp(b) − 1]), pe each in e qua ile ange (IQR) in- c ease in ai pollu an concen a ions which equa ed o 7.7 μg/m3 PM10, 4.8 μg/m3 PM2.5 and 12.1 μg/m3 NO2. We also used logis ic eg ession models o in es iga e he associa ions o ambien ai pollu an s wi h high o al choles e ol (TC ≥ 240 mg/dL); high LDL- c (LDL- c ≥ 160 mg/ dL); low HDL- c (HDL- c < 40 mg/dL); and hype iglyce- idaemia (TG ≥ 200 mg/dL) as de ined by he s anda d lipid measu emen s, and wi h high LDL- p and VLDL- p concen a ions (≥p75), and low HDL- p concen a ions (<p25), measu ed by he Liposcale® es . These esul s a e p esen ed as odds a ios (ORs) wi h co esponding 95% CIs again pe each IQR inc ease in ai pollu an s. All hese models we e con olled o possible con ound- e s such as age, sex, BMI, educa ion le el, smoking s a us, alcohol in ake, municipali y popula ion, MedSco e, IPAQ, mean ambien empe a u e o he municipali y and lipid lowe ing medica ion. In addi ion, we in es iga ed po en ial e ec modi- ica ion o he associa ions by sex (male/ emale), age (<40/40– 60 o ≥60 yea s) and BMI (<30 o ≥30 kg/m2). Each po en ial modi ie was examined in a sepa a e model by adding an in e ac ion e m. All he s a is ical analyses we e pe o med wi h IBM SPSS s a is ics 23.0. Repo ed p alues we e based on wo- sided es s wi h s a is ical signi icance se a .05. Bon e - oni co ec ion o mul iple compa isons was applied a a le el o alpha =.05/7 = .007. 3 | RESULTS A o al o 4647 indi iduals we e included in he analysis (Table1). The sample was composed o 1976 men (42.5%) and 2671 women (57.5%). Mean age o he popula ion was 50.5 ± 17.0 yea s ( ange: 18– 93 yea s). The cha ac e is ics o he s udy popula ion ollowed a dis ibu ion as ex- pec ed in he Spanish gene al popula ion. Dis ibu ions o s anda d lipids and NMR pa icle concen a ions a e also displayed in he able. Table2 summa izes esiden ial es ima es o ou doo ai pollu ion concen a ions assigned o he s udy pa ici- pan s in he yea o examina ion. The median (25 h– 75 h pe cen ile) PM10, PM2.5 and NO2 exposu e le els we e 23.7 (19.6– 27.3), 12.2 (10.5– 15.3) and 16.6 (12.5– 24.6) μg/m3, espec i ely. Mos alues we e wi hin he cu en Eu o- pean Ambien Ai Quali y Di ec i e a ge alues (Di ec- i e 2008/50/EC).41 Table3 shows he esul s o he linea co ela ions be ween ai pollu an concen a ions and s anda d lipid bioma ke s (TC, LDL- c, HDL- c and TG) and NMR pa - icle concen a ions (LDL- p, VLDL- p and HDL- p), in c ude, and mul i a ia e adjus ed linea eg ession mod- els. G ea e ai pollu an exposu es we e associa ed wi h lowe TC and LDL- c concen a ions in c ude models, bu hese associa ions we e a enua ed a e mul i a ia e ad- jus men . This end was no obse ed by NMR analyses. In ac he associa ion be ween ai pollu an s and LDL- p al hough no signi ican ended o be posi i e. The h ee pollu an s es ed showed a s ong and highly signi ican nega i e associa ion wi h HDL- c concen a ions in bo h c ude and mul i a ia e models. In he ully adjus ed model, an IQR inc ease in PM10, PM2.5 and NO2 was as- socia ed wi h 3.3%, 3.3% and 3% lowe le els o HDL- c, espec i ely. The same end was obse ed ega ding he associa ion be ween con aminan s and he concen a ion o HDL pa icles measu ed by NMR. In he mul i a ia e analysis, each IQR inc ease in PM10, PM2.5 and NO2 was associa ed wi h 1.3%, 1.4% and 1.1% lowe HDL- p. PM2.5 concen a ions also showed a posi i e associa ion wi h TG le els in mul i a ia e analyses. Table4 shows he c ude and mul i a ia e- adjus ed ORs o p esen ing lipid abno mali ies in he s anda d lipid 13652362, 2024, 2, Downloaded om h ps://onlinelib a y.wiley.com/doi/10.1111/eci.14101 by Readcube (Lab i a Inc.), Wiley Online Lib a y on [13/03/2024]. See he Te ms and Condi ions (h ps://onlinelib a y.wiley.com/ e ms-and-condi ions) on Wiley Online Lib a y o ules o use; OA a icles a e go e ned by he applicable C ea i e Commons License | 5 o 11 VALDÉS e al. p o ile and NMR, pe each IQR inc ease in ai pollu an concen a ions. Replica ing wha was seen in he linea model, he e was a s ong nega i e associa ion be ween he h ee ai pollu an s es ed, and HDL- c. In he ully ad- jus ed model, he odds o p esen ing low HDL- c (<40 mg/ dL), we e 1.36 (95% CI 1.21– 1.54) p < .001, 1.41 (95% CI 1.25– 1.59) p < .001 and 1.31 (95% CI 1.18– 1.44) p < .001 pe each IQR inc ease in PM10, PM2.5 and NO2. Again, his same end was obse ed o he associa ion be ween ai pollu an s and HDL- p, wi h mul i a ia e ORs o p e- sen ing low HDL- p (<p25) o 1.23 (95% CI 1.09– 1.37) p < .001, 1.23 (95% CI 1.10– 1.37) p < .001 and 1.18 (95% CI 1.07– 1.29) p = .001 pe each IQR inc ease in PM10, PM2.5 and NO2, espec i ely. In he logis ic eg ession models, he e was also a posi i e associa ion be ween PM10 and PM2.5 concen a ions and he odds o p esen ing highe LDL- p, wi h mul i a ia e ORs o p esen ing LDL- p con- cen a ions ≥p75 o 1.24 (95% CI 1.11– 1.39) p < .001 and 1.16 (95% CI 1.04– 1.29) p = .007, pe each IQR inc ease, espec i ely. In he subg oup analysis, we ound a signi ican in- e ac ion be ween sex and he associa ion be ween PM10 and NO2 and low HDL- p (p o in e ac ion .008 and .034, espec i ely), wi h s onge associa ions in men. We also ound a s onge associa ion be ween NO2 and low HDL- p in indi iduals wi h obesi y (p o in e ac ion .015) (Figu e1). 4 | DISCUSSION In his na ionwide sample ep esen a i e o he adul popula ion o Spain we ound signi ican associa ions be ween exposu e o a ious ai pollu an s and se e al s anda d and no el blood lipop o ein measu es, poin - ing o a p o- a he ogenic lipid p o ile in subjec s exposed o a highe deg ee o pollu ion. In pa icula we ound s ong nega i e associa ions be ween PM10, PM2.5 and NO2 exposu es, and bo h HDL- c and HDL- p concen- a ions bo h in linea and in logis ic mul i a ia e e- g ession models. Addi ionally, we ound a signi ican associa ion be ween exposu e o PM10 and PM2.5, and TABLE 1 Clinical cha ac e is ics and lipid alues o he s udy sample (n = 4647). % Mean ± SD Range Age (yea s) 50.5 ± 17.0 18– 93 Women 57.5 Smoking Cu en 23.8 Fo me 26.0 Ne e 50.2 Alcohol in ake (se ings- mon h) <30 73.9 30– 60 14.9 >60 11.3 Municipali y popula ion <10,000 18.4 10,000– 50,000 27.8 >50,000 53.8 Educa ion le el No s udies 12.8 Basic 47.6 High school- college 39.6 BMI (kg/m2) 28.0 ± 5.2 12.2– 61.3 Med die sco e 7.8 ± 1.8 1– 13 Physical ac i i y (IPAQ) Low 42.0 Medium 34.5 High 23.5 Lipid lowe ing medica ion 13.3 S anda d lipid bioma ke s (mg/dL) TC 196 ± 40 58– 395 LDL- C 105 ± 30 21– 254 HDL- C 52 ± 13 7– 14 TG 121 ± 88 17– 2095 Pa icle concen a ion (1H- NMR) (nmol/L) LDL- p 1392 ± 278 358– 2702 VLDL- p 53 ± 40 12– 619 HDL- p 28 ± 5 6– 56 Abb e ia ions: HDL- C, high- densi y lipop o ein choles e ol; HDL- p, HDL pa icles; LDL- C, low- densi y lipop o ein choles e ol; LDL- p, LDL pa icles; TC, o al choles e ol; TG: iglyce ides; VLDL- p, VLDL pa icles; 1H- NMR, p o on nuclea magne ic esonance. TABLE 2 Desc ip i e s a is ics o ai pollu an s concen a ions (μg/m3) in he s udy sample. Pollu an Pe cen ile IQR Mean Minimum Maximum5 h 25 h 50 h 75 h 95 h PM10 14.4 19.6 23.7 27.3 33.1 7.7 23.2 4.0 42.3 PM2.5 8.1 10.5 12.2 15.3 20.0 4.8 12.7 3.4 22.3 NO26.6 12.5 16.6 24.6 50.3 12.1 20.2 3.6 51.4 Abb e ia ions: IQR, in e qua ile ange; NO2, ni ogen dioxide; PM10, pa icles wi h an ae odynamic diame e o less han 10 mic ons; PM2.5, pa icles wi h an ae odynamic diame e o less han 2.5 mic ons. 13652362, 2024, 2, Downloaded om h ps://onlinelib a y.wiley.com/doi/10.1111/eci.14101 by Readcube (Lab i a Inc.), Wiley Online Lib a y on [13/03/2024]. See he Te ms and Condi ions (h ps://onlinelib a y.wiley.com/ e ms-and-condi ions) on Wiley Online Lib a y o ules o use; OA a icles a e go e ned by he applicable C ea i e Commons License 6 o 11 | VALDÉS e al. TABLE 3 Associa ions be ween ai pollu an s (pe IQR concen a ions inc ease) and blood lipid le els. S anda d lipid bioma ke s TC LDL- C HDL- C TG % change (95% CI) p % change (95% CI) p % change (95% CI) p % change (95% CI) p PM10 C ude −1.7 (−2.5, −1.0) <.001 −1.6 (−2.7, −0.5) .004 −3.2 (−4,1, −2.3) <.001 0.4 (−1.4, 2.0) .667 Mul i a ia e −1.1 (−1.9, −0.2) .013 −0.3 (−1.6, 0.9) .590 −3.3 (−4.3, −2.3) <.001 2.0 (0.1, 3.9) .037 PM2.5 C ude −1.1 (−1.9, −0.3) .01 −0.8 (−2.0, 0.4) .190 −3.0 (−4.0, −2.1) <.001 1.0 (−1.0, 2.9) .323 Mul i a ia e −0.6 (−1.4, 0.3) .215 −0.3 (−1.0, 1.5) .628 −3.3 (−4.3, −2.3) <.001 2.8 (0.9, 4.7) .003 NO2 C ude −1.8 (−2.5, −1.2) <.001 −1.8 (−2.7, −0.9) <.001 −2.7 (−3.4, −1.9) <.001 −1.5 (−2.9, 0.0) .053 Mul i a ia e −1.4 (−2.2, −0.7) <.001 −0.9 (−1.9, 0.2) .118 −3.0 (−3.8, −2.2) <.001 −0.5 (−1.2, 2.1) .570 Pa icle concen a ion (1H- NMR) LDL- p VLDL- p HDL- p % change (95% CI) p% change (95% CI) p% change (95% CI) p PM10 C ude 0.2 (−0.6, 0.9) .703 1.1 (−1.0, 3.2) .324 −0.7 (−1.3, 0.0) .049 Mul i a ia e 0.6 (−0.3, 1.4) .205 1.9 (−0.3, 4.1) .091 −1.3 (−2.0, −0.6) <.001 PM2.5 C ude 0.4 (−0.4, 1.2) .364 1.0 (−1.3, 3.2) .391 −0.8 (−1.5, −0.1) .018 Mul i a ia e 0.6 (−0.2, 1.5) .159 1.9 (−0.3, 4.1) .089 −1.4 (−2.1, −0.7) <.001 NO2 C ude 0.6 (0.0, 1.2) .068 0.3 (−1.3, 2.0) .703 −0.5 (−1.1, 0.0) .040 Mul i a ia e 0.5 (−0.2, 1.3) .155 1.8 (−0.1, 3.6) .059 −1.1 (−1.6, −0.5) <.001 No e: % changes and p alues calcula ed by linea eg ession pe in e qua ile ange (IQR) inc ease in ai pollu an s concen a ions (PM10: 7.7 μg/m3, PM2.5: 4.8 μg/m3, NO2: 12.1 μg/m3). Mul i a ia e model: adjus ed o age, sex, BMI, smoking s a us, alcohol in ake, educa ion le el, MedSco e, IPAQ, municipali y popula ion, ambien empe a u e and lipid lowe ing medica ion. In bold: % changes wi h p alues < .007 (alpha .05 co ec ed by Bon e oni = .05/7). Abb e ia ions: HDL- C, high- densi y lipop o ein choles e ol; HDL- p, HDL pa icles; LDL- C, low- densi y lipop o ein choles e ol; LDL- p, LDL pa icles; NO2, ni ogen dioxide; PM10, pa icles wi h an ae odynamic diame e o less han 10 mic ons; PM2.5, pa icles wi h an ae odynamic diame e o less han 2.5 mic ons; TC, o al choles e ol; TG, iglyce ides; VLDL- p, VLDL pa icles; 1H- NMR, p o on nuclea magne ic esonance. 13652362, 2024, 2, Downloaded om h ps://onlinelib a y.wiley.com/doi/10.1111/eci.14101 by Readcube (Lab i a Inc.), Wiley Online Lib a y on [13/03/2024]. See he Te ms and Condi ions (h ps://onlinelib a y.wiley.com/ e ms-and-condi ions) on Wiley Online Lib a y o ules o use; OA a icles a e go e ned by he applicable C ea i e Commons License | 7 o 11 VALDÉS e al. highe LDL- p concen a ions (≥p75), al hough he asso- cia ions be ween ai pollu an s and he s anda d LDL- c measu es we e null. These indings a e in line wi h p e ious s udies sug- ges ing ha ai pollu ion migh nega i ely impac blood lipids in he gene al popula ion,6– 24 and may be a po en- ial con ibu o o ai pollu ion- ela ed CVD isk. In ac , low HDL- c is a well- es ablished a he oscle o ic CVD isk ac o ,34 whe eas lowe HDL and highe LDL pa icle numbe s ha e also been associa ed wi h inc easing ca - dio ascula e en s.42– 44 In e es ingly, al hough he ela ionship o he ex- posu e o di e en ai pollu an s wi h s anda d lipid bioma ke s has been widely s udied, o he bes ou knowledge, only wo p e ious s udies ha e assessed as- socia ions wi h pa icle concen a ion measu es. Bell e al.9 examined he ela ionship be ween ai pollu ion and bo h HDL- c and HDL- p, in 6654 men and women ee o p e alen clinical CVD, pa icipan s in he Mul i- E hnic S udy o A he oscle osis Ai Pollu ion s udy (MESA Ai ). A 5 μg/m3 highe PM2.5 was associa ed wi h lowe HDL- p (−0.64 μmol/L [95% CI −1.01, −0.26]), bu no HDL- c (−0.05 mg/dL [95% CI −0.82, 0.71]). Mc- Guinn e al.13 s udied linea associa ions o es ima e change in lipop o ein le els wi h each μg/m3 inc ease in annual a e age PM2.5 in 6587 pa ien s who had a ca diac ca he e iza ion in Duke Uni e si y be ween 2001 and 2010 (CATHGEN s udy). The pe cen change om he mean ou come le el was 2.00% (95% CI 1.38%, 2.64%) o o al LDL- p. Howe e , he associa ions be ween ai pollu ion and HDL pa icle concen a ions in his s udy we e inconsis en . This is in con as o he indings TABLE 4 Odd a ios (OR) o p esen ing lipid abno mali ies pe in e qua ile ange (IQR) inc ease in ai pollu an s concen a ions. S anda d lipid bioma ke s TC ≥ 240 mg/dL LDL- C ≥ 160 mg/dL HDL- C < 40 mg/dL T iglyce ides ≥ 200 mg/dL OR 95% CI pOR 95% CI pOR 95% CI pOR 95% CI p PM10 C ude 0.94 0.85– 1.05 .282 0.99 0.82– 1.20 .952 1.29 1.17– 1.43 .000 0.96 0.85– 1.08 .489 Mul i a ia e 1.02 0.90– 1.16 .718 1.10 0.87– 1.37 .428 1.36 1.21– 1.54 .000 0.97 0.83– 1.12 .655 PM2.5 C ude 1.00 0.89– 1.13 .958 1.07 0.87– 1.31 .521 1.33 1.20– 1.47 .000 0.94 0.82– 1.07 .353 Mul i a ia e 1.07 0.94– 1.22 .314 1.16 0.92– 1.46 .198 1.41 1.25– 1.59 .000 0.95 0.82– 1.11 .542 NO2 C ude 0.93 0.85– 1.02 .139 0.97 0.83– 1.14 .975 1.23 1.14– 1.32 .000 0.91 0.82– 1.01 .073 Mul i a ia e 0.99 0.88– 1.11 .853 1.11 0.91– 1.36 .285 1.31 1.18– 1.44 .000 0.96 0.84– 1.10 .558 Pa icle concen a ion (1H- NMR) LDL- p ≥ p75 VLDL- p ≥ p75 HDL- p < p25 OR 95% CI pOR 95% CI pOR 95% CI p PM10 C ude 1.15 1.06– 1.26 .002 1.03 0.94– 1.12 .578 1.12 1.03– 1.23 .011 Mul i a ia e 1.24 1.11– 1.39 .000 1.08 0.96– 1.21 .188 1.23 1.09– 1.37 .000 PM2.5 C ude 1.12 1.02– 1.23 .016 1.02 0.93– 1.12 .631 1.13 1.03– 1.24 .007 Mul i a ia e 1.16 1.04– 1.29 .007 1.09 0.97– 1.21 .144 1.23 1.10– 1.37 .000 NO2 C ude 1.08 1.01– 1.16 .023 0.99 0.92– 1.06 .741 1.10 1.03– 1.17 .007 Mul i a ia e 1.11 1.02– 1.22 .018 1.05 0.95– 1.15 .334 1.18 1.07– 1.29 .001 No e: ORs. 95% CI and p alues we e calcula ed by logis ic eg ession pe in e qua ile ange (IQR) inc ease in ai pollu an s concen a ions (PM10: 7.7 μg/m3, PM2.5: 4.8 μg/m3, NO2: 12.1 μg/m3). In bold: % changes wi h p alues < .007 (alpha .05 co ec ed by Bon e oni = .05/7). Mul i a ia e model: adjus ed o age, sex, BMI, smoking s a us, alcohol in ake, educa ion le el, MedSco e, IPAQ, municipali y popula ion, ambien empe a u e and lipid lowe ing medica ion. Abb e ia ions: HDL- C, high- densi y lipop o ein choles e ol; HDL- p, HDL pa icles; LDL- C, low- densi y lipop o ein choles e ol; LDL- p, LDL pa icles; NO2, ni ogen dioxide; PM10, pa icles wi h an ae odynamic diame e o less han 10 mic ons; PM2.5, pa icles wi h an ae odynamic diame e o less han 2.5 mic ons; TC, o al choles e ol; TG, iglyce ides; VLDL- p, VLDL pa icles; 1H- NMR, p o on nuclea magne ic esonance. 13652362, 2024, 2, Downloaded om h ps://onlinelib a y.wiley.com/doi/10.1111/eci.14101 by Readcube (Lab i a Inc.), Wiley Online Lib a y on [13/03/2024]. See he Te ms and Condi ions (h ps://onlinelib a y.wiley.com/ e ms-and-condi ions) on Wiley Online Lib a y o ules o use; OA a icles a e go e ned by he applicable C ea i e Commons License 8 o 11 | VALDÉS e al. epo ed by Bell e al. in he MESA Ai s udy9 and o ou esul s in [email p o ec ed], which indica e s ong nega i e associa ions be ween ai pollu an s and HDL ma ke s. O no e, he CATHGEN s udy sample consis ed o pa- ien s who unde wen ca diac ca he e iza ion, who may ep esen a highly selec i e popula ion, while he MESA Ai and he [email p o ec ed] s udy popula ions may be mo e gene alizable o he gene al backg ound popula ion. Al hough he obse ed di e ences in blood lipids ound in ou s udy may seem small, i is o no e ha , as FIGURE 1 Logis ic eg ession analyses be ween PM2.5, PM10 and NO2 exposu es and HDL- c <40 mg/dL, HDL- p <p25 and LDL- p ≥p75 in di e en popula ion subg oups. 13652362, 2024, 2, Downloaded om h ps://onlinelib a y.wiley.com/doi/10.1111/eci.14101 by Readcube (Lab i a Inc.), Wiley Online Lib a y on [13/03/2024]. See he Te ms and Condi ions (h ps://onlinelib a y.wiley.com/ e ms-and-condi ions) on Wiley Online Lib a y o ules o use; OA a icles a e go e ned by he applicable C ea i e Commons License | 9 o 11 VALDÉS e al. poin ed ou by Bell e al.,9 he magni ude o he dec ease o HDL- c and HDL- p we ha e obse ed (be ween −3.0% and −3.3% dec ease in HDL- c and − 1.1% o −1.4% de- c ease in HDL- p pe IQR inc ease in ai pollu an s con- cen a ion), can be compa ed o he e ec o smoking on hese measu emen s as obse ed in smoking cessa ion s udies (2.4 mg/dL and 1.0 μmol/L change on HDL- c and HDL- p, espec i ely).45 On he o he hand, exposu e o ai pollu ion has also been shown o induce he de elopmen o dys unc ional HDL esul ing in al e a ion o i s a he op o ec i e capaci- ies,46,47 which may con ibu e o a he oscle osis p og es- sion beyond he o al lipop o ein concen a ions measu ed in ou s udy. The appa en ly coun e in ui i e nega i e associa ion be ween ai pollu an s and TC obse ed is o small magni- ude and could be explained by he con ibu ion o HDL- c o he TC le els, which was signi ican ly educed when as- socia ed wi h highe exposu es. In line wi h his, no associ- a ions wi h LDL- c le els in he mul i a ia e analyses we e ound. This appa en ly benign e ec o ai pollu an s in he s anda d LDL- c measu es was howe e accompanied by a signi ican associa ion be ween highe exposu es o PM10 and PM2.5, and highe LDL- p concen a ions (≥p75). LDL- p has been ound o be a be e p edic o o CVD han LDL- c, especially in indi iduals wi h LDL- p/LDL- c disco dance.28,29,48 So again, hese associa ions could be ega ded as p o- a he ogenic. In subg oup analyses, we ound s onge associa ions be ween PM10 and NO2 and low HDL- p in men, and be ween NO2 and low HDL- p in indi iduals wi h obesi y, sugges ing ha hese subg oups may be mo e suscep ible. A s onge suscep ibili y in o e weigh – obese indi iduals is consis en wi h Sø ensen e al.,6 Yang e al.,10 Kim e al.,12 Mao e al.,16 Zhang e al.22 and Kim e al.19 among o he s. Bo h ai pollu ion expo- su e and o e weigh /obesi y a e associa ed wi h highe sys emic in lamma ion2,49 which may explain an in e play o hese ac o s. The modi ica ion e ec s o sex in p e ious s udies ha e been mixed and ha de o in e p e . The unde lying mechanisms explaining he e ec o ai pollu an s on lipids ha e no been ully elucida ed. The main hypo hesis is ha he oxida i e s ess and sys- emic in lamma ion caused by inhaled ai pollu ion could induce ad e se lipid me abolism and lipid oxida ion.50,51 Ai pollu an s ha e also been ound o cause DNA me h- yla ion o genes ela ed o lipid me abolism.52 Fu he s udies a e wa an ed o cla i y he ull spec um o hese mechanisms. Ou s udy has se e al s eng hs, including he la ge popula ion- based design and he inclusion o ad anced li- pop o ein measu es, as well as o he ex ensi e indi idual- le el da a o clinical, demog aphic and li es yle a iables, which allowed us o pe o m a comple e mul i a ia e adjus men o he da a. Ou na ionwide pe spec i e, al- lows us o ex apola e ou esul s mo e widely han local o egional s udies, inc easing he public heal h implica- ions o he indings. The limi a ions o ou s udy include i s obse a ional c oss- sec ional na u e; so ha we canno es ablish causal associa ions o exclude esidual con ounding in he e- la ion be ween ai pollu an s and lipids. Also, we used ambien ou doo measu emen s modelled a he esiden- ial add esses o he pa icipan s as a p oxy o exposu e o ai pollu ion, whe eas no o he ele an in o ma ion such as ime– ac i i y pa e ns, p oximi y o main oads, occupa ional exposu es o pe sonal moni o ing da a was a ailable. Exposu e measu emen e o is possible when using modelled pollu an le els and his could, in ac , ha e a enua ed ou e ec es ima es. Finally, ou exposu e models we e de eloped based on yea ly exposu es o ai pollu an s, whe eas mo e e ined measu es o look a di - e en lags we e no a ailable. 5 | CONCLUSIONS In summa y, ou s udy, in keeping wi h p e ious da a, sugges s a dele e ious e ec o he exposu e o ai pollu - an s on blood lipids, in he gene al popula ion o Spain. Ou esul s ein o ce he need o imp o ing ai quali y as much as possible o dec ease he isk o a he oscle osis in ou popula ion, as he lipid changes obse ed in ou s udy may be a po en ial con ibu o o ai pollu ion- ela ed CVD isk. AUTHOR CONTRIBUTIONS Concep ion and design: G.R.M and J.R. Acquisi ion o epidemiological da a: C.M.A., E.G.E., S.G.S., A.C.P., L.C., E.D., E.M., J.F.N., S.G., J.G., F.J.C., S.V. and G.R.M. anda d biochemical samples managemen : W.O.B. and G.R.M. Liposcale® samples managemen and analyses: J.R., N.A. and M.G. Ai pollu ion modelling: M.G.V., J.L.G., M.T., V.G. and F.M.L. C ea ion o new so wa e used in he wo k: J.L.G.G and G.A.V. Analysis and in e - p e a ion o da a: S.V. J.C.V. and G.R.M. D a ing he a - icle: S.V., V.K.D.G. and G.R.M. All au ho s e ised and app o ed he inal manusc ip . ACKNO WLE DGE MENTS The [email p o ec ed] p ojec is a collabo a i e s udy wi h a ious phases and sub- p ojec s in which a la ge numbe o esea ch- e s and echnicians ha e collabo a ed, o whom we a e in- deb ed. Ou p o ound app ecia ion goes o he p ima y ca e manage s and pe sonnel o he pa icipa ing heal h cen es, o all he ieldwo ke s, nu ses and echnicians and o he s udy pa icipan s o hei al uis ic pa icipa ion. 13652362, 2024, 2, Downloaded om h ps://onlinelib a y.wiley.com/doi/10.1111/eci.14101 by Readcube (Lab i a Inc.), Wiley Online Lib a y on [13/03/2024]. See he Te ms and Condi ions (h ps://onlinelib a y.wiley.com/ e ms-and-condi ions) on Wiley Online Lib a y o ules o use; OA a icles a e go e ned by he applicable C ea i e Commons License