Cluster Analysis on Longitudinal Data of Patients With Adult-Onset Asthma

O iginal A icle Clus e Analysis on Longi udinal Da a o Pa ien s wi h Adul -Onse As hma Pinja Ilma inen, PhD a , Leena E. Tuomis o, MD, PhD a , Onni Niemelä, MD, PhD b,c , Minna Tommola, MD a , Jussi Haanpää, MSc d , and Hannu Kankaan an a, MD, PhD a,e Seinäjoki and Tampe e, Finland Wha is al eady known abou his opic? Many pheno ypes o as hma ha e been iden ified in p e ious clus e analyses, mos ly on he basis o c oss-sec ional da a wi h limi ed inclusion o pa ien s. Some s udies ha e p o ided sho - e m 1- o 3-yea p ognosis o he pheno ypes. Wha does his a icle add o ou knowledge? This is he fi s s udy ha epo s long- e m 12-yea p ognosis o clus e s o adul -onse as hma s a ing om diagnosis. We epo di e en disease p ognoses o smoking, obesi y- ela ed, emale, a opic, and non hini ic as hma. How does his s udy impac cu en managemen guidelines? In o ma ion on long- e m ou come o as hma can be used o in o m and mo i a e pa ien s. We show he poo es ou come and he mos unme needs in he he apy o smoking and obesi y- ela ed as hma, sugges ing need o special guidance. BACKGROUND: P e ious clus e analyses on as hma a e based on c oss-sec ional da a. OBJECTIVE: To iden i y pheno ypes o adul -onse as hma by using da a om baseline (diagnos ic) and 12-yea ollow-up isi s. METHODS: The Seinäjoki Adul As hma S udy is a 12-yea ollow-up s udy o pa ien s wi h new-onse adul as hma. K-means clus e analysis was pe o med by using a iables om baseline and ollow-up isi s on 171 pa ien s o iden i y pheno ypes. RESULTS: Fi e clus e s we e iden ified. Pa ien s in clus e 1(n[38) we e p edominan ly nona opic males wi h mode a e smoking his o y a baseline. A ollow-up, 40% o hese pa ien s had de eloped pe sis en obs uc ion bu he numbe o pa ien s wi h uncon olled as hma (5%) and hini is (10%) was he lowes . Clus e 2 (n [19) was cha ac e ized by olde men wi h hea y smoking his o y, poo lung unc ion, and pe sis en obs uc ion a baseline. A ollow-up, hese pa ien s we e mos ly uncon olled (84%) despi e daily use o inhaled co icos e oid (ICS) wi h add-on he apy. Clus e 3 (n [50) consis ed mos ly o nonsmoking emales wi h good lung unc ion a diagnosis/ ollow-up and well-con olled/pa ially con olled as hma a ollow-up. Clus e 4 (n [25) had obese and symp oma ic pa ien s a baseline/ ollow-up. A ollow-up, hese pa ien s had se e al como bidi ies (40% psychia ic disease) and we e ea ed daily wi h ICS and add-on he apy. Pa ien s in clus e 5(n[39) we e mos ly a opic and had he ea lies onse o as hma, he highes blood eosinophils, and FEV 1 e e sibili y a diagnosis. A ollow-up, hese pa ien s used he lowes ICS dose bu 56% we e well con olled. CONCLUSIONS: Resul s can be used o p edic ou comes o pa ien s wi h adul -onse as hma and o aid in de elopmen o a Depa men o Respi a o y Medicine, Seinäjoki Cen al Hospi al, Seinäjoki, Finland b Depa men o Labo a o y Medicine, Seinäjoki Cen al Hospi al, Seinäjoki, Finland c Uni e si y o Tampe e, Tampe e, Finland d Depa men o Clinical Physiology, Seinäjoki Cen al Hospi al, Seinäjoki, Finland e Depa men o Respi a o y Medicine, Uni e si y o Tampe e, Tampe e, Finland This s udy was suppo ed by he Finnish An i-Tube culosis Associa ion Founda- ion (Helsinki, Finland), he Tampe e Tube culosis Founda ion (Tampe e, Finland), he Jalma i and Rauha Ahokas Founda ion (Helsinki, Finland), he Resea ch Founda ion o he Pulmona y Diseases (Helsinki, Finland), he Compe i i e S a e Resea ch Financing o he Expe Responsibili y A ea o Tampe e Uni e si y Hospi al (Tampe e, Finland), and he Medical Resea ch Fund o Seinäjoki Cen al Hospi al (Seinäjoki, Finland). None o he sponso s had any in ol emen in he planning and execu ion o his s udy o in he w i ing o his a icle. Conflic s o in e es : P. Ilma inen has ecei ed lec u e ees om MundiPha ma. L. E. Tuomis o has ecei ed lec u e ees om Mundipha ma and has ecei ed a el suppo om Takeda, Chiesi, and O ion. M. Tommola has ecei ed lec u e ees om As aZeneca and GlaxoSmi hKline. H. Kankaan an a has ecei ed lec u e and consul ancy ees and a el suppo om Almi all, As aZeneca, and Boeh inge Ingelheim; has ecei ed consul ancy ees om Chiesi Pha ma AB; has ecei ed lec u e and consul ancy ees om GlaxoSmi hKline, Lei as-Takesa, and No a is; has ecei ed lec u e ees om MSD, Mundipha ma, Medi h, Resmed Finland, and O ion Pha ma; has ecei ed a el suppo om In e mune; and has ecei ed consul ancy ees om Roche. The es o he au ho s decla e ha hey ha e no ele an conflic s o in e es . Recei ed o publica ion Oc obe 25, 2016; e ised Janua y 3, 2017; accep ed o publica ion Janua y 31, 2017. A ailable online Ap il 25, 2017. Co esponding au ho : Pinja Ilma inen, PhD, Depa men o Respi a o y Medicine, Seinäjoki Cen al Hospi al, Hanneksen inne 7, Seinäjoki 60220, Finland. E-mail: pinja.ilma inen@epshp.fi. 2213-2198 Ó2017 The Au ho s. Published by Else ie Inc. on behal o he Ame ican Academy o Alle gy, As hma & Immunology. This is an open access a icle unde he CC BY-NC-ND license (h p://c ea i ecommons.o g/licenses/by-nc-nd/4.0/). h p://dx.doi.o g/10.1016/j.jaip.2017.01.027 967 Abb e ia ions used ACOS- As hma-COPD o e lap synd ome ACT- As hma con ol es AQ20- Ai ways Ques ionnai e 20 BD- B onchodila o COPD- Ch onic obs uc i e pulmona y disease FVC- Fo ced i al capaci y ICS- Inhaled co icos e oid Max 0-2.5 - Maximum lung unc ion du ing he fi s 2.5 yea s a e diagnosis (and s a o an i-inflamma o y he apy) SAAS- Seinäjoki Adul As hma S udy pe sonalized he apy (NCT02733016 a ClinicalT ials. go ). Ó2017 The Au ho s. Published by Else ie Inc. on behal o he Ame ican Academy o Alle gy, As hma & Immunology. This is an open access a icle unde he CC BY- NC-ND license (h p://c ea i ecommons.o g/licenses/by-nc- nd/4.0/). (J Alle gy Clin Immunol P ac 2017;5:967-78) Key wo ds: As hma; Adul -onse ; La e-onse ; Clus e analysis; Pheno ypes; Follow-up; Longi udinal; Smoking; Obesi y; Ea ly-onse Adul - o la e-onse as hma has been sugges ed o be a dis inc i e pheno ype o as hma. 1,2 Pa ien s wi h adul -onse as hma ha e lesse alle gic p ocesses, lowe lung unc ion despi e sho e du a ion o disease, and mo e o en a p onounced eosinophilic inflamma ion wi hou e idence o T H 2eassocia ed inflamma ion when compa ed wi h pa ien s wi h childhood- onse as hma. 1 These findings sugges ha adul -onse as hma is mo e he e ogeneous when compa ed wi h childhood-onse as hma. In p e ious s udies, subpheno ypes o adul -onse as hma such as eosinophil-p edominan , mild o mode a e well-con olled, obesi y- ela ed, smoking, and se e e obs uc i e as hma ha e been p oposed. 3,4 To iden i y pheno ypes o as hma, unsupe ised hie a chical clus e analyses ha e been ca ied ou . Howe e , he clus e an- alyses ha e mos ly been based on c oss-sec ional da a on pa ien s wi h mixed du a ion o as hma. 2,4-6 As hma is known as a disease wi h a high deg ee o a iabili y, making one ime poin a agile basis o clus e analysis. Fu he mo e, no in o ma ion on he diagnos ic phase has been included in he p e ious analyses. In addi ion, many p e ious analyses ha e clus e ed pa ien s wi h se e e as hma, 7,8 lea ing milde o ms wi h less a en ion. Some s udies ha e in ol ed sho ollow-ups (1-3 yea s). 6,7,9,10 In a p e ious p ospec i e longi udinal analysis o se e e as hma, he clus e s did no show clus e -specific disease cou ses ega ding ou come o as hma, sugges ing a po en ial limi a ion in he way o pe o ming cu en clus e analyses. 9 In addi ion o he na u al disease a iabili y, many ac o s such as he apy, li es yle, and como bidi ies may modi y he disease cou se. Reliabili y o he esul s o a clus e analysis would be inc eased by including clinical da a om se e al ime poin s o he disease ollow-up in o he analysis. He e, we used a long- e m ollow-up app oach o cons uc pheno ypes o adul -onse as hma by ca ying ou a clus e analysis wi h inclusion o a iables om diagnosis o a 12-yea ollow-up isi . This app oach p o ides no el insigh s in o he pheno ypes o as hma wi h p ognos ic significance. METHODS Pa ien s and s udy design The p esen s udy was pa o he Seinäjoki Adul As hma S udy (SAAS), which is a p ospec i e, single-cen e (Seinäjoki Cen al Hospi al, Seinäjoki, Finland), 12-yea ollow-up s udy o a coho o consecu i e whi e pa ien s ha ing new-onse as hma diagnosed a adul age (15 yea s). SAAS has been egis e ed on ClinicalT ials.go wi h ID NCT02733016. Ins i u ional pe missions (TU1114 and LET) we e ob ained and he pa icipan s ga e w i en in o med consen o he s udy p o ocol app o ed by he E hics Commi ee o Tampe e Uni e si y Hospi al, Tampe e, Finland (R12122). The p o ocol, inclusion and exclusion c i e ia, and he backg ound da a o SAAS ha e been published elsewhe e. 11 B iefly, as hma was diagnosed by a specialized espi a o y physician du ing he pe iod 1999 o 2002 on he basis o ypical clinical symp oms and confi med by objec i e lung unc ion measu emen s. 11,12 The main diagnos ic ea u es o as hma in each clus e a e p esen ed in Table E1 in his a icle’s Online Reposi o y a www.jaci-inp ac ice.o g. Smoke s and pa ien s wi h como bidi ies we e no excluded. A e diagnosis, he pa ien s we e ea ed and moni o ed in specialized o p ima y ca e as equi ed. The o al coho consis ed o 257 pa ien s and 203 pa ien s comple ed he ollow-up isi (mean ollow-up ime, 12.2 yea s; ange, 10.8-13.9 yea s). A 12-yea ollow-up isi , as hma s a us and disease con ol, como bidi ies, and medica ion we e e alua ed using s uc u ed ques ionnai es and lung unc ion was measu ed. Da a on as hma- ela ed isi s o heal h ca e and hospi aliza ions we e also collec ed om p ima y ca e, occupa ional heal h ca e, p i a e clinics, and hospi als. A e excluding hose wi h missing da a, 171 pa ien s wi h adul -onse as hma emained in he coho o clus e analysis (Figu e 1). Lung unc ion, como bidi ies, in lamma o y pa ame e s, and o he clinical measu emen s Lung unc ion was measu ed wi h a spi ome e acco ding o in- e na ional ecommenda ions. 13 The ollowing we e he lung unc- ion measu emen poin s: (1) baseline ( ime o as hma diagnosis), (2) he maximum p eb onchodila o FEV 1 (P e-BD FEV 1 ) du ing he fi s 2.5 yea s a e diagnosis (Max 0-2.5 ) (and a e s a o an i-inflamma o y he apy), and (3) 12-yea ollow-up. 14 De ailed in o ma ion on de e mina ion o lung unc ion, inflamma o y pa- ame e s, and como bidi ies can be ound in his a icle’s Online Reposi o y a www.jaci-inp ac ice.o g. As hma con ol was assessed acco ding o he Global Ini ia i e o As hma 2010 epo . 15 Pa ien s filled ou he Ai ways Ques ionnai e 20 (AQ20) a baseline isi and AQ20 and as hma con ol es (ACT) ques ionnai es a he ollow- up isi . The AQ20 is a sho and simple well- alida ed ques ion- nai e o measu e and quan i y dis u bances in he ai way-specific quali y o li e. 16 ACT is a widely used pa ien sel -adminis e ed ool o iden i ying hose wi h poo ly con olled as hma. 17 Va iable selec ion Inpu a iables o he clus e analysis we e selec ed on he basis o ac o analysis (see Table E2 in his a icle’s Online Reposi o y a www.jaci-inp ac ice.o g). Basic and clinical a iables included in ac o analysis we e chosen o co e as wide a ange as possible om diagnosis o 12-yea ollow-up isi and a e u he discussed in his a icle’s Online Reposi o y a www.jaci-inp ac ice.o g. Clus e analysis and disc iminan analysis Clus e analysis was ca ied ou by using a 2-s ep p ocess. Fi s , Wa d hie a chical clus e analysis was pe o med o p ee alua ion o J ALLERGY CLIN IMMUNOL PRACT JULY/AUGUST 2017 968 ILMARINEN ET AL he numbe o clus e s. Then, K-means analysis was ca ied ou by using he p especified numbe o clus e s (5). S epwise disc iminan analysis was pe o med o iden i y a iables disc imina ing be ween he p especified clus e s. S a is ically significan esul s we e expec ed o mos o he compa isons because he objec i e o he clus e analysis was o di e en ia e he pa icipan s in o dis inc pheno ypes o adul -onse as hma. O he s a is ical analyses Con inuous da a a e exp essed as mean SD o median and in e qua ile ange. G oup compa isons we e pe o med by 1-way analysis o a iance wi h he Tukey pos hoc es , he K uskal- Wallis es , o he chi-squa e es . S a is ical analyses we e pe o med by using SPSS so wa e, e sion 22 (IBM Co po a ion, A monk, NY) and MATLAB, e sion 8.6 (Ma hwo ks, Na ick, Mass). RESULTS Pa ien s’cha ac e is ics Cha ac e is ics o he o al coho a baseline and ollow-up a e p esen ed in Table E3 in his a icle’s Online Reposi o y a www. jaci-inp ac ice.o g. Pa ien s we e mos ly emales (58.5%) and nona opic (63.5%), wi h age a as hma onse anging om 15 o 77 yea s. A diagnosis, mos pa ien s we e s e oid-nai e. A he 12-yea ollow-up poin , 78.9% we e daily use s o inhaled co icos e oid (ICS) and 50.9% we e daily use s o ICS and add-on medica ion. Clus e analysis By pe o ming Wa d hie a chical and K-means clus e ana- lyses, 5 clus e s we e iden ified. The basic cha ac e is ics o hese clus e s a e shown in Figu e 2. Clus e 1 was cha ac e ized by low p e alence (10.5%) o hini is (non hini ic as hma), whe eas clus e 2 had he highes smoking his o y (smoking as hma). Clus e 3 consis ed mainly (98%) o women ( emale as hma), and mos pa ien s in clus e 4 we e obese om diagnosis o 12-yea ollow-up isi (obesi y- ela ed as hma). Clus e 5 mos ly included a opic pa ien s wi h he ea lies onse o as hma (ea ly- onse a opic adul as hma). Basic and clinical cha ac e is ics o he clus e s a baseline and a ollow-up a e p esen ed in Tables I-IV. The e we e no majo di e ences in he main diagnos ic ea u es be ween g oups (see Table E1 in his a icle’s Online Reposi o y a www.jaci-inp ac ice.o g). Clus e 1: Non hini ic as hma. Clus e 1 (n ¼38 [22.2%]) was cha ac e ized by lack o hini is, male p edomi- nance (60.5%), onse o as hma a middle age, and second highes smoking his o y (Figu e 2). P opo ion o pa ien s wi h pe manen b onchial obs uc ion (pos -BD FEV 1 / o ced i al capaci y [FVC] <0.7) inc eased om 10.8% o 39.5% om diagnosis o 12-yea ollow-up isi . This clus e also showed he highes weigh gain, wi h he p opo ion o obese pa ien s inc easing om 18.4% o 39.5%. As hma was uncon olled in only 5.3% o he pa ien s, e en hough mos (55.9%) we e ea ed wi h low-dose ICS o no daily ICS (Table I). Clus e 1 showed mode a e loss o FEV 1 du ing he ollow-up and he lowes use o heal h ca e (Figu es 3 and 4). Pa ien s had 1 co- mo bidi y on a e age a he 12-yea ollow-up isi , he mos p e alen being ch onic obs uc i e pulmona y disease (COPD) and hype ension (Table IV). Clus e 2: Smoking as hma. Clus e 2 was he smalles clus e (n ¼19 [11.1%]) and was p edomina ed by olde males. Almos 80% o pa ien s had smoking his o y and 44.4% showed b onchial obs uc ion a diagnosis and could be cha ac e ized as ha ing as hma-COPD o e lap synd ome (ACOS). 18 Howe e , he smoking clus e did no di e om o he clus e s on he basis o di using capaci y. The pa ien s had poo lung unc ion, high symp oms, and uncon olled as hma despi e 94.7% being unde daily ICS he apy, 73.4% wi h mode a e- o high-dose ICS, and 78.9% wi h long-ac ing b 2 agonis a ollow-up. E en hough lung unc ion significan ly imp o ed a e s a o he apy, he annual decline in FEV 1 was s eep (78 mL on a e age) om he maximum poin o lung unc ion o he 12-yea ollow-up isi (Figu e 4;Table II). This was he only g oup wi h no dec ease in blood eosinophils o symp oms om diagnosis o ollow-up (Figu e 4). The pa ien s had 3 como bidi ies on a e age (Table IV) and equen heal h ca e use (Figu e 3). O he pa- ien s, 36.8% had been hospi alized o as hma (Figu e 3,B) and his clus e accoun ed o 35.8% o all hospi al ea men pe iods. Clus e 3: Female as hma. This g oup was he la ges (n ¼ 50 [29.2%]) and consis ed o women wi h a wide ange in he age o as hma onse . Clus e 3 con ained mo e (44%) pa ien s wi h no mal weigh (body mass index <25) compa ed wi h o he clus e s and showed he lowes smoking his o y. Fo y pe cen epo ed being symp oma ic al eady du ing childhood e en hough hey we e no diagnosed as ha ing as hma. P e-BD FEV 1 was no mal (>80% p edic ed) in 78% a diagnosis and in 90% a ollow-up, and he annual decline in FEV 1 was he lowes (31 mL). Blood eosinophils we e he second highes and he AQ20 symp om sco e was a a a he mode a e le el o 6 ou o 20 a diagnosis, and bo h educed du ing he ollow-up (Figu e 4). E en hough lung unc ion measu ed by spi ome y and inflamma ion we e wi hin no mal ange and 78% we e unde - going ICS ea men a he 12-yea ollow-up isi , 64% o he pa ien s we e pa ially con olled o uncon olled and heal h ca e use was ela i ely high (Table I;Figu e 3). Female clus e accoun ed o 29.0% o all as hma- ela ed isi s. Clus e 4: Obesi y- ela ed as hma. This clus e (n ¼25 [14.6%]) mos ly con ained nona opic emales wi h he oldes age Yea s 1999-2002 Baseline isi Pa ien s wi h new diagnosis o as hma (age ≥ 15 yea s) we e included n = 260 Pa ien s excluded n = 32 missing in o ma ion on inpu a iables o clus e analysis Yea s 2012-2013 12-yea ollow-up isi n = 203 Res p onse a e=79 % Pa ien s excluded n = 1 consen wi hd awn n = 2 childhood as hma Pa ien s los du ing ollow-up n = 22 dead n = 9 could no be eached n = 5 signi ican como bidi ies n = 18 o he easons Final s udy popula ion n = 171 Visi s o heal hca e Hospi aliza ions FIGURE 1. Flow cha o he s udy. J ALLERGY CLIN IMMUNOL PRACT VOLUME 5, NUMBER 4 ILMARINEN ET AL 969 o as hma onse . On a e age, pa ien s we e obese om diagnosis o he 12-yea ollow-up poin wi hou gaining mo e weigh du ing his ime. The pa ien s we e mul imo bid a ollow-up (Table IV), wi h he mos p e alen como bidi ies being hype - ension, diabe es, and psychia ic diseases. As hma was uncon- olled in 48% o he pa ien s a ollow-up e en hough 92% we e unde going ICS ea men , 50% used high-dose ICS, and 72% we e on add-on medica ion (Table I). A diagnosis, 44% showed p e-BD FEV 1 o mo e han 80% p edic ed. Lung unc ion imp o ed on s a o he apy and emained ela i ely s able h oughou ollow-up (Figu e 4,A). Respi a o y symp- oms, use o o al s e oids, and heal h ca e use we e all a high le els (Figu es 3 and 4). Clus e 5: Ea ly-onse a opic adul as hma. This g oup was he second la ges (n ¼39 [22.8%]) and consis ed o he younges pa ien s wi h mean age o as hma onse a 33 11 yea s. Almos hal had su e ed om espi a o y symp oms du ing childhood, 59% we e a opic, and 90% had nonalle gic o alle gic hini is (Table I). O his clus e , 59% showed p e- BD FEV 1 o mo e han 80% a diagnosis and 84% could be e e sed o FEV 1 o mo e han 80% p edic ed by BD. Re e sibili y was in gene al he highes . Lung unc ion ini ially showed a good esponse o s e oids, e en hough he loss o lung unc ion a e he maximum poin was also he second s eepes . A diagnosis, hese pa ien s showed he highes blood eosinophils (Table III), which educed un il he 12-yea ollow- up isi (Figu e 4). As hma was con olled in 56% o he subjec s and use o medica ion was he lowes , because 56% we e using low-dose ICS o no medica ion and only 17.9% we e ea ed by long-ac ing b 2 agonis (Table I). Use o s e oid bu s s was in equen and use o heal h ca e was among he lowes (Figu e 3). Valida ion Fo alida ion, we ca ied ou K-means algo i hm 10 imes by he lea e-one-ou me hod o ensu e s abili y and epea abili y o he model. This me hod showed 94.4% epea abili y. Disc iminan analysis By using a s epwise me hod o disc iminan analysis, 12 ou o 17 a iables we e ound o significan ly disc imina e be ween he clus e s: he diagnos ic a iables we e pos -BD FEV 1 /FVC, FEV 1 e e sibili y, and maximal change in FEV 1 ( om diagnosis o Max 0-2.5 ), whe eas he ollow-up a iables included hini is, numbe o d ugs in use o ea como bidi ies, p e-BD FEV 1 , pack-yea s, body mass index, limi a ion o ac i i ies (none/any), and basic a iables o sex, age a as hma onse , and symp oms o as hma o less han 16 yea s, o which hini is, pos -BD FEV 1 / FVC, numbe o d ugs in use o ea como bidi ies, and sex we e ound o be he s onges disc imina ing a iables. Du a ion o symp oms be o e diagnosis, blood eosinophils and neu o- phils, e e sibili y a ollow-up, and ACT sco e we e no ound as s a is ically significan disc iminan s. The pe cen age o co ec classifica ion on he basis o he 12 disc imina ing a iables was 94.7% (da a no shown). DISCUSSION In his s udy, we iden ified pheno ypes o adul -onse as hma by using longi udinal da a and basic and clinical a iables anging om he diagnos ic phase o he 12-yea ollow-up isi . Ou coho included smoke s and pa ien s wi h como bidi ies. The ollowing 5 pheno ypes we e iden ified: (1) non hini ic con olled o pa ially con olled as hma wi h low use o medi- ca ion and heal h ca e; (2) smoking as hma o ACOS wi h poo lung unc ion, high symp oms, and high use o medica ion and heal h ca e; (3) emale as hma wi h no mal clinical pa ame e s FIGURE 2. Basic cha ac e is ics o clus e s. In A-F, C1 o C5 e e o he clus e numbe s. O e all P alues a e shown. In Band F, he ed lines shown a e means. In D, means a e shown. BMI, Body mass index; DG, diagnosis. J ALLERGY CLIN IMMUNOL PRACT JULY/AUGUST 2017 970 ILMARINEN ET AL TABLE I. Gene al ea u es o clus e s Fea u es Clus e 1: non hini ic (n [38) Clus e 2: smoking (n [19) Clus e 3: emale (n [50) Clus e 4: obese (n [25) Clus e 5: a opic (n [39) P alue be ween clus e s P alue be ween clus e s Baseline Follow-up Baseline Follow-up Baseline Follow-up Baseline Follow-up Baseline Follow-up Baseline Follow-up Demog aphic cha ac e is ics and an hopome ics Females, n (%) 15 (39.5) 2 (10.5)*49 (98)†16 (64) 18 (46.2) <.001 Age (y), mean SD 50 12 63 12 55 96694312†55 12†57 86983311†45 11†<.001 <.001 BMI (kg/m 2 ), mean  SD 28.1 5.5z29.9 6.5 27.8 4.0 28.1 4.9 26.3 4.3 27.1 4.8 32.7 5.2†32.8 5.3zxjj 24.9 3.5 26.8 3.9 <.001 <.001 Obese (BMI >30), n (%) 7 (18.4) 15 (39.5) 8 (42.1)z7 (36.8) 7 (14.0) 13 (26.0) 17 (68.0)zjj{ 18 (72.0)†4 (10.3) 9 (23.1) <.001 .001 Smoke s, n (%) 20 (52.6) 20 (52.6) 15 (78.9)jj 15 (78.9)jj 17 (34) 18 (36) 11 (44) 11 (44) 19 (48.7) 21 (53.8) .020 .027 Cu en smoke , n (%) 7 (18.4) 7 (18.4) 5 (26.3) 3 (15.8) 5 (10) 8 (16) 4 (16) 0 9 (23.1) 8 (20.5) .425 .225 Pack-yea s o smoke s, median (IQR) 17 (12-23)jj 19 (15-29)zjj 29 (14-34)*zjj 33 (15-38)zjj 5 (2-10) 6 (2-18) 15 (10-20) 15 (10-25) 4 (3-7) 6 (3-15) <.001 <.001 Symp oms o as hma <16 y, n (%) 0 1 (5.3) 20 (40.0)*{1 (4.0) 19 (48.7)*x{ <.001 Symp oms o as hma be o e diagnosis (mo), median (IQR) 18 (9-60) 24 (11-36) 12 (9-36) 24 (24-60)zjj 12 (8-24) .008 A opic, n (%) 9 (27.3) ND 5 (31.3) ND 19 (40.4) ND 2 (9.1) ND 22 (57.9)*ND .003 ND No. o posi i e SPT, median (IQR) 0 (0-1) ND 0 (0-2) ND 0 (0-2)*ND 0 (0-0) ND 1.5 (0-3)*x{ ND .001 ND Rhini is, n (%) ND 4 (10.5)†ND 14 (73.7) ND 44 (88) ND 24 (96) ND 35 (89.7) ND <.001 As hma con ol and quali y o li e AQ20 sco e, median (IQR) 5 (3-7) 2 (1-4) 8 (5-10) 8 (5-11)zjj{ 6 (4-10) 4 (2-6) 10 (7-13)zjj{ 7 (4-9)z{ 4 (2-9) 2 (1-5) <.001 <.001 ACT sco e, median (IQR) ND 23 (21-24) ND 20 (13-21)zjj{ ND 22 (19-24) ND 19 (15-22)z{ ND 23 (21-25) ND <.001 ACT sco e <20, n (%) ND 5 (13.2) ND 9 (47.4)z{ ND 13 (26.0) ND 15 (60.0)zjj{ ND 5 (12.8) ND <.001 Con olled, n (%) ND 16 (42.1) ND 2 (10.5) ND 18 (36) ND 4 (16) ND 22 (56.4)*xND <.001 Pa ly con olled, n (%) ND 20 (52.6)xND 1 (5.3) ND 21 (42)xND 9 (36) ND 11 (28.2) ND .007 Uncon olled, n (%) ND 2 (5.3) ND 16 (84.2)zjj{ ND 11 (22){ND 12 (48)z{ ND 6 (15.4) ND <.001 Exace ba ions O al s e oids, n (%)#2 (5.4) 4 (21.1) 6 (12.2) 7 (29.2) 2 (5.1) .026 T ea men Daily ICS use , n (%)** 1 (2.6) 27 (71.1) 1 (5.3) 18 (94.7) 6 (12.2) 39 (78) 2 (8.0) 23 (92) 2 (5.1) 28 (71.8) .479 .089 ICS dose,†† median (IQR) 900 (800-1600) 800 (400-1000) 800 (700-1600) 900 (700-1400) 800 (400-1000) 800 (575-1000) 1000 (800-1400) 1000 (475-1525) 800 (800-1600) 800 (400-800) .220 .163 Low/none ICS dose, n (%) ND 19 (55.9) ND 4 (26.7) ND 18 (40) ND 6 (30) ND 20 (55.6) ND .109 Medium ICS dose, n (%) ND 7 (20.6) ND 4 (26.7) ND 11 (24.4) ND 4 (20) ND 11 (30.6) ND .871 (con inued) J ALLERGY CLIN IMMUNOL PRACT VOLUME 5, NUMBER 4 ILMARINEN ET AL 971 bu ela i ely high use o heal h ca e; (4) obesi y- ela ed as hma wi h como bidi ies, high symp oms, and high use o medica ion and heal h ca e; and (5) a opic well-con olled as hma wi h onse ea lie in adul hood. Ins ead o cha ac e izing pheno ypes in one poin o disease, we p o ide pheno ypes based on 12-yea ollow-up da a. Ou esul s show bo h simila i ies o and di e ences om hose o p e ious clus e analyses based on c oss-sec ional da a wi h mixed du a ion o as hma. Mos p e ious analyses ha e excluded smoke s o hea y smoke s and only ew ha e iden ified smoking as hma. 6,19 Clus e A in he Coho o Reali y and E olu ion o Adul As hma in Ko ea (COREA) s udy 6 esembled ou smoking clus e in many espec s. Howe e , in ou s udy, he smoking clus e showed an annual decline in FEV 1 ha was he s eepes o all g oups in con as o he esul s o he COREA smoking clus e in 1-yea ollow-up. Many nega i e ou comes p e iously associa ed wi h smoking as hma we e e iden in ou smoking clus e , including lowe as hma- ela ed quali y o li e (based on AQ20 sco e), equen heal h ca e use, and se e e/uncon olled as hma. 20-22 E en hough he pa ien s in he smoking clus e ypically show i e e sible ai flow limi a ion and como bidi y p ofiles esembling ha o COPD, no mal a e age di using ca- paci y in each clus e sugges s ha he main diagnosis is as hma and no emphysema. In addi ion, he diagnosis o as hma was made by a espi a o y specialis and each pa ien ulfilled he diagnos ic c i e ia o as hma including objec i e lung unc ion measu emen s showing b onchial a iabili y. This smoking g oup, al hough being he smalles one, was esponsible o mo e han a hi d o all as hma- ela ed hospi aliza ions, highligh ing he sig- nificance o his g oup o heal h ca e cos s. Ob iously, much e - o s should be ocused on ad ising pa ien s o s op smoking as ea ly as possible, e en be o e onse o as hma and be o e as hma u ns om a milde o m o di ficul - o- ea smoking as hma wi h high bu den o bo h indi idual and heal h ca e. Obesi y- ela ed emale-p edominan as hma is a clus e iden- ified in ou s udy as well as in some p e ious s udies, 2,5,7,10 and ou esul s add by p o iding da a on p ognosis o he long- e m obesi y. Ou esul s on his clus e suppo p e ious findings indica ing equen symp oms and exace ba ions, high use o heal h ca e, high medica ion, and a nona opic, noneosinophilic disease cha ac e is ic. This clus e is also p one o c ea e a high bu den o heal h ca e, as e idenced by he mos equen use o o al co icos e oids and o heal h ca e se ices. To u he add on p e ious s udies, we ound he highes numbe o como bidi ies in he obese clus e and especially a high p e alence o psychia ic como bidi y (40%). Consis en ly, in a p e ious s udy, he highes dep ession sco e was shown in he la e-onse obese clus e in a coho o se e e as hma. 7 Howe e , in e ac ion be ween obesi y, psychia ic diseases, and as hma equi es u he s udies. 23,24 Recen ly, we showed ha mul imo bidi y is associ- a ed wi h inc eased symp oms o as hma, which may be pa ly ela ed o sys emic inflamma ion in hese pa ien s. 25 Sys emic inflamma ion has been associa ed wi h high s e oid dose in he ea men o adul -onse as hma. 25 The e o e, mul imo bidi y and sys emic inflamma ion may be ele an in u ning as hma in o symp oma ic and s e oid- esis an in obese pa ien s. Howe e , weigh loss has esul ed in imp o ed symp oms, lung unc ion, as hma con ol, and heal h s a us, 23 sugges ing ha i would benefi his subg oup. In addi ion o he obesi y- ela ed emale-p edominan g oup, we iden ified a nonobese clus e o emales wi h good lung TABLE I. (Con inued) Fea u es Clus e 1: non hini ic (n [38) Clus e 2: smoking (n [19) Clus e 3: emale (n [50) Clus e 4: obese (n [25) Clus e 5: a opic (n [39) P alue be ween clus e s P alue be ween clus e s Baseline Follow-up Baseline Follow-up Baseline Follow-up Baseline Follow-up Baseline Follow-up Baseline Follow-up High ICS dose, n (%) ND 8 (23.5) ND 7 (46.7) ND 16 (35.6) ND 10 (50)zND 5 (13.9) ND .022 LABA, n (%) ND 16 (42.1) ND 15 (78.9) ND 25 (50) ND 18 (72) ND 7 (17.9) ND <.001 LTRA, n (%) ND 5 (13.2) ND 1 (5.3) ND 8 (16) ND 11 (44)zjj ND 2 (5.1) ND <.001 Theophylline, n (%) ND 0 ND 0 ND 2 (4) ND 1 (4) ND 0 ND .419 LAMA, n (%) ND 0 ND 5 (26.3)zjj{ ND 0 ND 2 (8) ND 0 ND <.001 BMI, Body mass index; LABA, long-ac ing b 2 agonis ; LTRA, leuko iene ecep o an agonis ; ND, no de e mined; SPT, skin p ick es . *P<.05 s clus e 4 a co esponding ime poin (baseline o ollow-up). †P<.05 o all o he clus e s a co esponding ime poin (baseline o ollow-up). zP<.05 s clus e 5 a co esponding ime poin (baseline o ollow-up). {P<.05 s clus e 1 a co esponding ime poin (baseline o ollow-up). xP<.05 s clus e 2 a co esponding ime poin (baseline o ollow-up). jjP<.05 s clus e 3 a co esponding ime poin (baseline o ollow-up). Pai ed compa ison be ween baseline and ollow-up is no shown. #A leas 2 cou ses o o al s e oids du ing he 2 p e ious yea s be o e he ollow-up isi . **Baseline ICS use s e e o hose who used ICS daily be o e diagnosis. ††Baseline ICS dose is he s a ing dose a diagnosis. Low-dose ICS e e s o 400 m g, medium-dose ICS o >400e800 m g, and high-dose ICS o >800 m g budesonide equi alen s. 15 J ALLERGY CLIN IMMUNOL PRACT JULY/AUGUST 2017 972 ILMARINEN ET AL TABLE II. Lung unc ion o clus e s (mean SD) Cha ac e is ic Clus e 1: non hini ic (n [38) Clus e 2: smoking (n [19) Clus e 3: emale (n [50) Clus e 4: obese (n [25) Clus e 5: a opic (n [39) P alue be ween clus e s P alue be ween clus e s Baseline Follow-up Baseline Follow-up Baseline Follow-up Baseline Follow-up Baseline Follow-up Baseline Follow-up Lung unc ion P e-BD FEV 1 , % e 82 15 85 14*53 18†63 19†90 12 96 13 78 13*79 16*81 14*86 12*<.001 <.001 Pos -BD FEV 1 ,% e 8616 89 14*60 18†67 20†94 13 98 13 83 14*81 16*z92 12 92 12 <.001 <.001 P e-BD FVC, % e 90 16 97 14 73 17*zx 90 15*95 12 103 14 84 13*88 14*93 15 98 14 <.001 <.001 Pos -BD FVC, % e 92 16 100 16 81 16*z93 14 96 13 103 14 86 12*z90 14*97 13 99 14 <.001 .007 P e-BD FEV 1 /FVC, % e 0.75 0.08*0.71 0.08*0.57 0.11†0.57 13†0.80 0.07 0.76 0.05 0.76 0.06 0.72 0.10 0.74 0.10*0.72 0.07 <.001 <.001 Pos -BD FEV 1 /FVC, % e 0.77 0.07*0.72 0.08*0.58 0.12†0.58 0.14†0.83 0.07 0.78 0.05 0.78 0.08 0.72 0.10*0.81 0.08 0.76 0.07 <.001 <.001 FEV 1 e e sibili y (mL) 166 137 134 147*267 251 142 172*139 142 47 69 144 211 62 104 407 204*xjj 195143*xjj <.001 <.001 FEV 1 e e sibili y (% change) 5.6 5.3 4.8 5.2*15.0 16.4*zxjj 6.8 7.7*5.9 7.8 1.8 2.8 6.2 8.8 3.2 5.3 14.2 13.6*xjj 6.6 5.6*<.001 <.001 Di using capaci y DL CO /VA (% e ) 104 17 100 15 94 21 88 28 100 20 95 14 99 18 93 13 107 16 97 14 .155 .107 Annual change in lung unc ion om Max 0-2.5 o ollow-up{ D FEV 1 (mL/y) 49 32*78 54*31 24 46 30 59 40*<.001 D FVC (mL/y) 37 32 48 63 24 31 40 39 38 49 .173 Maximal change in FEV 1 ( D om diagnosis o Max 0-2.5 ) D FEV 1 (mL#) 266 365 972 899*zxjj 161 252 184 239 561 459*xjj <.001 DL CO , Di using capaci y o he lung o ca bon monoxide; e , e e ence; VA, al eola olume. *P<.05 s clus e 3 a co esponding ime poin (baseline o ollow-up). †P<.05 o all o he clus e s a co esponding ime poin (baseline o ollow-up). zP<.05 s clus e 5 a co esponding ime poin (baseline o ollow-up). xP<.05 s clus e 1 a co esponding ime poin (baseline o ollow-up). jjP<.05 s clus e 4 a co esponding ime poin (baseline o ollow-up). Pai ed compa ison be ween baseline and ollow-up is no shown. {Annual change in FEV 1 o FVC om poin o maximal lung unc ion wi hin 2.5 y a e s a o he apy o he 12-y ollow-up isi . # D FEV 1 when deduc ing p e-BD FEV 1 alue a diagnosis om p e-BD FEV 1 alue a poin o maximal lung unc ion wi hin 2.5 y om s a o he apy; eflec s ea ly esponse o ea men . J ALLERGY CLIN IMMUNOL PRACT VOLUME 5, NUMBER 4 ILMARINEN ET AL 973 unc ion and a wide ange in he age o as hma onse . The p e iously defined clus e s “leas se e e as hma wi h no mal lung unc ion,”“middle-age onse , emale-dominan ,”and “la e-onse mild as hma” 6,7,19 show esemblance o ou emale clus e , including p edominance o nonobese emales, no mal lung unc ion, 6,7,19 and s able lung unc ion in he sho ollow- ups. 6,7 Low smoking his o y, low body mass index, low coun o como bidi ies, and high eosinophils a diagnosis may ha e a ec ed he good o e all p ognosis o he emale clus e . How- e e , he heal h ca e use was ela i ely high in his g oup gi en ha many clinical pa ame e s we e wi hin no mal ange. This may be explained by emales’s onge pe cep ion o symp oms and lowe h eshold o con ac heal h ca e when compa ed wi h males. 26 In addi ion, emale pa ien s wi h simila se e i y o as hma as he co esponding male pa ien s ha e shown be e lung unc ion bu wo se as hma- ela ed quali y o li e. 26 In his clus e , lung unc ion based on spi ome y was mo e s able when compa ed wi h lung unc ion in o he g oups, bu peak expi a- o y flow ollow-up migh ha e shown a iable obs uc ion and disease ac i i y. In addi ion, whe he pa ame e s such as blood eosinophils o ai way hype esponsi eness be e co ela e o disease ac i i y and symp oms emains unclea . In he emale clus e , oughly hal o he pa ien s we e a e ile age, and he o he hal a menopausal o pos menopausal age a as hma onse , sugges ing ha no uni o m sex ho monee ela ed mechanism explains he pa hophysiology o as hma in his clus e . Howe e , ho monal aspec s p obably play a significan ole in he pa ho- genesis and cou se o he disease, 26-31 e en hough he mecha- nisms a e likely mul i ac o ial. P esence o symp oms a childhood in 40% o he pa ien s also sugges s he in ol emen o T H 2- ela ed mechanisms and some o e lap wi h clus e 5. Simila o p e ious findings, 32 mos pa ien s in ou s udy had a coexis ing alle gic o nonalle gic hini is. Howe e , we also iden ified a nona opic mild o mode a e male-p edominan as hma wi hou hini is, which o ou knowledge has no been epo ed p e iously. Despi e he second highes smoking his o y, 40% p e alence o pe manen b onchial obs uc ion, and obesi y a ollow-up, hese pa ien s had significan ly be e p ognosis when compa ed wi h hose in clus e s 2 and 4. Rhini is has been associa ed wi h mo e se e e as hma, 32,33 sugges ing ha lack o hini is is a significan de e minan associa ed wi h he a o able p ognosis in his g oup. Male p edominance, mode a e smoking his o y, and he highes weigh gain sugges ha pa hophysio- logical mechanisms in his clus e a e ela ed o hose ea u es. As hma in obese men has been epo ed o be less o en se e e when compa ed wi h ha in obese women 26 and obesi y- ela ed as hma may ha e impo an sex-specific di e ences conce ning he media o s o he disease. 34,35 Fo example, in a ecen s udy, no simila di e ence exis ed in sys emic inflamma ion be ween nonobese and obese males as seen in emales. 35 Lowe le el o sys emic inflamma ion could also con ibu e o lesse numbe o como bidi ies and be e p ognosis o as hma. Fu he s udies a e needed o e alua e he pa hophysiological mechanisms in his clus e . Clus e 5 in ou s udy fi s well wi h he p e ious findings o he impo an ole o age o onse in defining he pheno ype. 1,3 A opy, childhood symp oms, ea lies onse o as hma, good s e oid- esponsi eness, and la ge FEV 1 e e sibili y suppo he conclusion ha his clus e ep esen s he adi ional ea ly-onse as hma bu s a ing a ea ly adul hood. In addi ion, he good p ognosis s eng hens he iew. 36 A co esponding adul -onse TABLE III. In lamma o y bioma ke s o clus e s, median (IQR) Bioma ke Clus e 1: non hini ic (n [38) Clus e 2: smoking (n [19) Clus e 3: emale (n [50) Clus e 4: obese (n [25) Clus e 5: a opic (n [39) P alue be ween clus e s P alue be ween clus e s Baseline Follow-up Baseline Follow-up Baseline Follow-up Baseline Follow-up Baseline Follow-up Baseline Follow-up Blood eosinophils (10 9 /L) 0.20 (0.11-0.32) 0.14 (0.09-0.25) 0.20 (0.17-0.48) 0.23*(0.13-0.43) 0.30 (0.16-0.44) 0.16 (0.10-0.28) 0.22 (0.18-0.32) 0.13 (0.06-0.25) 0.38†(0.27-0.60) 0.20 (0.12-0.28) .008 .035 To al IgE (kU/L) 95 (28-278) 54 (23-159) 100 (70-552) 95 (24-315) 64 (27-147) 66 (21-138) 62 (22-125) 61 (24-118) 108 (56-409) 67 (30-383) .099 .464 Blood neu ophils (10 9 /L) ND 3.5 (2.9-4.7) ND 4.0 (3.4-4.7) ND 3.8 (2.5-5.2) ND 4.4 (3.2-5.3) ND 3.5 (2.9-4.0) ND .215 FENO (ppb) ND 10 (4-18) ND 10 (5-21) ND 11 (5-18) ND 8 (5-18) ND 16 (6-24) ND .364 FeNO, F ac ional exhaled ni ic oxide; IQR, in e qua ile ange; ND, no de e mined. *P<.05 s clus e 4 a co esponding ime poin (baseline o ollow-up). †P<.05 s clus e 1 a co esponding ime poin (baseline o ollow-up). Pai ed compa ison be ween baseline and ollow-up is no shown. J ALLERGY CLIN IMMUNOL PRACT JULY/AUGUST 2017 974 ILMARINEN ET AL TABLE IV. Como bidi ies o clus e s Como bidi y Clus e 1: non hini ic (n [38) Clus e 2: smoking (n [19) Clus e 3: emale (n [50) Clus e 4: obese (n [25) Clus e 5: a opic (n [39) P alue be ween clus e s P alue be ween clus e s Baseline Follow-up Baseline Follow-up Baseline Follow-up Baseline Follow-up Baseline Follow-up Baseline Follow-up Hype ension, n (%) 4 (10.5) 9 (23.7) 4 (21.1) 13 (68.4)*†z3 (6.0) 7 (14.0) 12 (48.0)*†z19 (76.0)*†z2 (5.1) 6 (15.4) <.001 <.001 Diabe es, n (%) 0 4 (10.5) 0 7 (36.8)†z0 3 (6.0) 3 (12.0) 11 (44.0)*†z0 2 (5.1) <.001 <.001 Co ona y hea disease, n (%) 1 (2.6) 5 (26.3)†z0 2 (8.0) 0 <.001 <.001 COPD, n (%) 2 (5.4) 9 (23.7)†8 (44.4)x10 (52.6)†zjj 0 1 (2.0) 0 2 (8.0) 1 (2.6) 3 (7.7) <.001 <.001 Any psychia ic disease, n (%) ND 3 (7.9) ND 1 (5.3) ND 5 (10) ND 10 (40)*†zND 3 (7.7) ND .001 Dep ession, n (%) ND 1 (2.6) ND 1 (5.3) ND 3 (6.0) ND 7 (28)*ND 2 (5.1) ND .004 Pain ul condi ion, n(%) ND 3 (7.9) ND 2 (10.5) ND 2 (4) ND 5 (20) ND 1 (2.6) ND .090 T ea ed dyspepsia, n(%) ND 1 (2.6) ND 2 (10.5) ND 4 (8) ND 6 (24) ND 1 (2.6) ND .020 To al no. o como bidi ies, median (IQR) ND 1 (0-2)zND 3 (1-4)*†zND 0.5 (0-1) ND 3 (2.5-4)*†zND 0 (0-1) ND <.001 No. o d ugs{, median (IQR) ND 1 (0-3) ND 3 (2-9)*†zND 1 (0-2) ND 6 (4-7)*†zND 0 (0-2) ND <.001 IQR, In e qua ile ange; ND, no de e mined. *P<.05 s clus e 1 a co esponding ime poin (baseline o ollow-up). †P<.05 s clus e 3 a co esponding ime poin (baseline o ollow-up). zP<.05 s clus e 5 a co esponding ime poin (baseline o ollow-up). xP<.05 s all clus e s a co esponding ime poin (baseline o ollow-up). jjP<.05 s clus e 4 a co esponding ime poin (baseline o ollow-up). Pai ed compa ison be ween baseline and ollow-up is no shown. {Numbe o d ugs in use o ea como bidi ies. J ALLERGY CLIN IMMUNOL PRACT VOLUME 5, NUMBER 4 ILMARINEN ET AL 975