Visual Assessment of CT Findings in Smokers With Nonobstructed Spirometric Abnormalities in The COPDGene Study

Transcription

1 88 Chronic Obstructive Pulmonary Diseases: Journal of the COPD Foundation Original Research. Visual Assessment of CT Findings in Smokers With Nonobstructed Spirometric Abnormalities in The COPDGene Study Song Soo Kim, MD, 1,2 Kunihiro Yagihashi, MD, 1 Douglas S. Stinson, MS, 1 Jordan A. Zach BA, 1 Alexander S. McKenzie BA, 1 Douglas Curran-Everett, PhD, 3 Emily S. Wan, MD, 4 Edwin K. Silverman, MD, PhD, 4 James D. Crapo, MD, 5 David A. Lynch, MB 1 Abstract WithintheCOPDGeneticEpidemiology(COPDGene )studypopulationofcigarettesmokers,9%werefoundto beunclassifiablebytheglobalinitiativeforchronicobstructivelungdisease(gold)criteria.thisstudywasto identifythedifferencesincomputedtomography(ct)findingsbetweenthisnonobstructed(goldu)groupand acontrolgroupofsmokerswithnormallungfunction.thisresearchwasapprovedbytheinstitutionalreviewboardof eachinstitution.ctimagesof400participantsinthecopdgene study(200goldu,200smokerswithnormal lungfunction)wereretrospectivelyevaluatedinablindedfashion.visualctassessmentincludedlobaranalysis ofemphysema(type,extent),presenceofparaseptalemphysema,airwaywallthickening,expiratoryairtrapping, centrilobularnodules,atelectasis,non-fibroticandfibroticinterstitiallungdisease(ild),pleuralthickening,diaphragmatic eventration,vertebralbodychangesandinternalthoracicdiameters(inmm).univariatecomparisonsofgroupsfor eachctparameterandmultiplelogisticregressionwereperformedtodeterminetheimagingfeaturesassociatedwith GOLDU.Whencomparedwiththecontrolgroup,GOLDUparticipantshadasignificantlyhigherprevalenceofunilateral diaphragmeventration(30%vs.16%),airwaywallthickening,centrilobularnodules,reticularabnormality,paraseptal emphysema(33%vs.17%),linearatelectasis(60%vs.35.6%),kyphosis(12%vs.4%),andasmallerinternaltransverse thoracicdiameter(255±22.5[standarddeviation]vs.264.8±22.4,mm)(allp<0.05).withmultiplelogisticregression,all ofthesectparameters,exceptnon-fibroticildandkyphosis,remainedsignificantlyassociatedwithgoldustatus (p<0.05).incigarettesmokers,chestwallabnormalitiesandparenchymallungdisease,whichcontributetorestrictive physiologicimpairment,areassociatedwithgold-nonobstructedstatus. Funding Support: Thisstudywassupportedbyagrant(NIH/NHLBIR01HL089856andR01HL089897)from:COPDGeneticEpidemiology (COPDGene )study. Abbreviations: GlobalInitiativeforchronicObstructivePulmonaryDisease,GOLD;computedtomography,CT;interstitiallungdisease, ILD;forcedexpiratoryvolumein1second,FEV1;forcedvitalcapacity,FEV;GOLD-nonobstructed,GOLDU;bodymassindex,BMI;total lungcapacity,tlc;diabetesmellitus,dm;groundglassopacity,ggo;qualitativect,qct;hounsfieldunits,hu;emphysemaindex,ei%; wallareapercentage,wa%;luminalarea,la;standarddeviation,sd;pulmonaryfunctiontest, PFT Date of Acceptance: March18,2014 Citation: KimSS,YagihashiU,StinsonD,etal.VisualassessmentofCTfindingsinsmokerswithnonobstructedspirometricabnormalities inthecopdgene study. J COPD F. 2014;1(1):88-96doi: 1 DepartmentofRadiology,NationalJewishHealth,Denver,CO 2 DepartmentofRadiology,ChungnamNationalUniversity Hospital,ChungnamNationalUniversitySchoolofMedicine, Daejeon,RepublicofKorea 3 DivisionofBiostatisticsandBioinformatics,NationalJewishHealth, anddepartmentofbiostatisticsandinformatics,colorado SchoolofPublicHealth,UniversityofColorado,Denver,CO 4 ChanningDivisionofNetworkMedicine,DivisionofPulmonary andcriticalcaremedicine,brighamandwomen shospital, HarvardMedicalSchool,Boston,MA 5 DivisionofPulmonary,CriticalCareandSleepMedicine,National JewishHealth,Denver,CO

2 89 CT Findings in Nonobstructed Spirometric Smokers Address correspondence to: ASongSooKim,MD DepartmentofRadiology,NationalJewishHealth,Denver,CO anddepartmentofradiology,chungnamnationaluniversity Hospital,ChungnamNationalUniversitySchoolofMedicine, ,Daesadong,Jung-gu,Daejeon,RepublicofKorea Tel ,Fax Keywords: lungdiseases;obstructive;lungdiseases;classification;computed tomography Introduction Chronic obstructive pulmonary disease (COPD) is a progressiveanddebilitatinglungdiseaseandanimportant causeofmorbidityandmortality,characterizedbyairflow limitationthatisnotfullyreversibleafteradministration ofbronchodilators,accordingtotheglobalinitiativefor chronic Obstructive Lung Disease(GOLD) and other guidelines. 1-3 ThespirometricdefinitionofCOPDtypically requiresaratioofforcedexpiratoryvolumein1second to forced vital capacity (FEV1/FVC) of less than 0.7. However,ithasbeendeterminedthatwithincohortsof cigarette smokers, about 8-14% of individuals present withanormalfev1/fvcratio,butareducedfev A recent study comparing this group, called GOLDnonobstructed(GOLDU),withspirometricallynormal smokersdemonstratedthatgoldustatuswasassociated with increased body mass index (BMI), reduced total lungcapacity(tlc),higherratesofnon-whiteindividuals anddiabetesmellitus(dm),aswellasincreasedbronchial wallthickness;however,substantialheterogeneitywithin thisgroupwasobserved. 7 GOLDUindividualsareusually regardedashavingrestrictivelungdisease,butmanyindividualshavetotallungcapacityvalueswithinthenormal range.individualswithnonobstructedspirometricabnormalitieshavenotbeenfullycharacterizedbyimaging,and itremainsunclearhowtheseindividualsareradiologically different from smokers with normal FEV1 values. 4,6,8,9 Therefore,thepurposeofourstudywastoidentifythedifferencesinCTfindingsbetweenthisGOLDUgroupof individualsandacontrolgroupofsmokerswithnormal lungfunction. Methods Clinical Participants IndividualsforthisstudywereselectedfromparticipantsintheCOPDGene Study,whichrecruited10,300 currentandformercigarettesmokers,withandwithout COPD, from 21 institutions. 10 Non-Hispanic whites or African-Americansbetweentheagesof45and80years withaminimumof10pack-yearssmokinghistorywere includedinthestudy.exclusioncriteriaincludedpregnancy, history of significant fibrotic lung disease or diffuse bronchiectasis,previoussurgicalexcisionofatleastone lunglobe(orlungvolumereductionprocedure),active cancerundertreatment,knownlungcancer,metalinthe chest, and recent exacerbation of COPD. Two hundred smokingcontrolparticipantsand200golduparticipants wererandomlyselectedforthisstudy,stratifiedsothatthe numbersofmaleandfemaleparticipantswererepresentative ofthemaincohortforeachgroup.theinstitutionalreview boardsfromeachclinicalcenterapprovedtheanalysesof theclinicalandimagingdata.individual,informedwritten consentwasobtainedfromallpatients. CT Examination, Acquisition and Data Analysis VolumetricCTscanswereobtainedforallparticipants using a standardized technique on multidetector CT scanners. 10,11 Thescanswereobtainedduringdeepinspiratorybreath-holdatastandardCTdose,andattheendof a normal expiration (functional residual capacity) at reducedctdoseinacraniocaudaldirectionofthesupine position. Standardized breathing instructions were provided. The typical CT parameters were as follows: submillimetercollimation( mm)andsubmillimeter reconstruction(thickness: mm,interval: mm)withbothstandardandhigh-frequencyprotocols,120 kvpand200masfortheinspirationscanand50masfor theexpirationscan. 10,11 Nopatientreceivedintravenous contrast medium. Anonymized DICOM CT data were transferredtotheimagingcoreforarchiving,qualityassuranceandquantitativeanalysis. Lobe-Based Quantitative Visual Assessment of CT Images Lobe-based visual assessment of CT images was performedbyusingstandardreferenceimages. 12 Two radiologistsperformedvisualassessment:reader1and reader2werechestradiologistswith8yearsand10years ofexperience,respectively.thereaderswereblindtoany clinical and functional information. Image data sets

3 90 CT Findings in Nonobstructed Spirometric Smokers reconstructed with high spatial frequency algorithms were presented to each reader using AquariusNET (TeraRecon, Inc.). Images were presented in random order to the readers in a blinded fashion regarding disease stage. Expiratory volumetric CT images were evaluatedaftertheinspiratoryvolumetricctdataand therewasnotimelimitationforvisualassessment.the CTimageswerevisuallyinspectedusingawindowwidth of1500hounsfieldunits(hu)andalevelof-700hu.for thelobe-basedvisualemphysemaassessment,readers were asked to determine the type of emphysema as normal,centrilobular,panlobular,ormixed.theextent ofemphysemaineachlobewasalsoassessedusinga 6-pointscalesystem:0%,1-5%,6-25%,26-50%,51-75%,and above75%.thelingulawasregardedasadifferentlobe, resultingin6lobesforeachcase.thepresenceofairway changes, including bronchial wall thickening and bronchiectasis,wasalsodeterminedineachlobe.the presenceofcentrilobularnodules,paraseptalemphysema, bullae,andmosaicattenuationpatternatinspirationct wasassessedandthepresenceofexpiratoryairtrapping was also assessed by comparison of inspiration and expiration CT images. Bronchial wall thickening was visuallydefinedbyaratiooftheinnertoouterlumenof lessthan Bronchiectasiswasconsideredpresent when the diameter of the inner bronchial lumen was greater than the diameter of the accompanying pulmonaryartery,orwhenabronchuswasvisiblewithin 1cmofthepleuralsurface.Centrilobularnoduleswere defined if more than 50% of a segment was affected. Mosaicattenuationpatternwasdefinedwhenmorethan 25%ofthelobeshowedpatchworkofregionsofdiffering attenuationoninspiratoryct,excludingareasofemphysema. 13 Expiratoryairtrappingwasdefinedonasideby sidecomparisonofinspiratoryandexpiratoryctasthe presence of parenchymal area with less than normal increaseinattenuationandlackofvolumereduction, whichinvolvesmorethan25%ofthelobe,excludingareas ofemphysema. 13 Toevaluatepulmonaryparenchymal andchestwallfeatureswhichmightcontributetorestrictive lungabnormality,thefollowingfeatureswereevaluated: presenceofatelectasis(linear,segmental,lobar),fibrotic ornonfibroticinterstitiallungdisease(ild),including groundglassopacity(ggo),reticularopacity,honeycomb cyst, presence of pleural thickening (diffuse, focal), diaphragmchange(eventration,hernia),andvertebral body change (kyphosis, scoliosis). Diaphragmatic eventrationwasdefinedasafocalabnormalcontourof the diaphragmatic dome. Maximal transverse and anteroposteriorinternalthoracicdiameters(mm)were measured.thenumberofwedgedvertebralbodieswas counted. Presternal subcutaneous fat thickness (mm) was measured at the level of pulmonary trunk; if fat extended outside the field of view, the measurements weremadetotheedgeoffieldofview.diffusepleural thickening was defined as involving more than 180 degreesofchestwallandmorethan5cmincraniocaudal direction.focalpleuralthickeningwasdefinedasinvolving lessthan180degreesofchestwallandlessthan5cmin craniocaudaldirection. Quantitative CT Measurements WholelungvolumetricCTimagesforquantitative analysiswerereconstructedusingastandardalgorithm. Using commercial software (Pulmonary Workstation, VIDADiagnostics,Inc.),quantitativeCT(QCT)measurements for the emphysema index (EI, %), wall area percentage(wa%),totallungcapacity(tlc CT,L),and gas trapping (%) were obtained. 10,14-18 The software automaticallycalculatedtheei(%),whichisdefinedas thepercentageoflungvoxelsatorbelow-950huon inspiratory CT. Quantitative assessment of airway dimensionswasperformedinthesegmental,subsegmental, and subsubsegmental generations of the following bronchial pathways: RB1(apical segment of RUL), RB4 (lateralsegmentofrml),rb10(posterobasalsegmentof RLL),LB1(apicoposteriorsegmentofLUL),LB4(superior segmentoflingula),andlb10(posterobasalsegmentoflll). Ineachsegmentalpathway,airwaydimensions,including WA,luminalarea(LA),andWA%weremeasured.Thewall areapercentagewasdefinedaswa%=wa/(wa+la)x Themeanvalueofthemeasureddimensionsofthe segmental,subsegmentalandsubsubsegmentalgenerations of bronchi was used for statistical analysis. TLC CT was calculatedasthevolumeofvoxelsofthelungandthegas trapping(%),whichisdefinedasthepercentageoflung voxels at or below-856 HU on expiratory CT, was also calculatedautomatically. 10 Pulmonary Function Testing Spirometry was conducted as recommended by the American Thoracic Society (ATS) by highly trained pulmonaryfunctiontechniciansandallspirometrydata wascollectedusingthenddeasyonespirometer(zurich, Switzerland). 10, 20 Each instrument was calibrated daily, andusualassuranceprogramforequipmentchecksand maintenancewasstrictlyfollowed.fev1,fvc,andratioof FEV1/FVC were evaluated, and were expressed as a percentageofpredictedvalue.pulmonaryfunctiontests (PFTs)wereperformedonaveragewithin2days(mean, 1.71±5.9)ofobtainingvolumetricCTscans(median0 day).

4 91 CT Findings in Nonobstructed Spirometric Smokers Statistical Analysis Resultswereexpressedasmean±SDforcontinuous variablesandasfrequenciesandpercentagesforcategoricalvariables.withunivariateanalysis,differencesin continuous data between the 2 groups were analyzed witht-testandcategoricaldatawereanalyzedwiththe χ²testandfisher sexacttest.forthebinaryvariables,if thereweredisagreementsbetween2independentreaders, finalassignmentsweredeterminedthroughasubsequent casereviewandadjudication.forthecontinuousvariables, themeanof2readingswasused.forextentofemphysema, themedianof2readingswasobtained.multiplelogistic regressionanalysiswasusedtodeterminethevisualand quantitative radiologic variables associated with the GOLDUstatus,afteradjustmentforage,race,gender, packyearsandbmi.thetotalvisualextentofemphysema wascalculatedbyaddingthe6-pointscalescoresforall 6 lobes. The inter-reader agreement of the visual CT parameterswasmeasuredwithcohen skappastatistics. Ap-valueoflessthan0.05wasconsideredstatistically significant.allstatisticalanalyseswereperformedusing JMP9.0(JMP;Cary,NC). Results Thebaselinecharacteristicsofthe400participantsare shownintable1.participantswereclassifiedasgoldu (n=200)andacontrolgroupofsmokerswithnormallung function(n=200).themeanageofthegolduparticipantswas58.9years±8.2standarddeviation(sd),including 71male(meanage,58.8years±7.8)and129female(mean age,58.9years±8.5)participants.themeanageofthe controlgroupparticipantswas60.1years±9.9,including 101 male (mean age, 59.5 years ± 10.3) and 99 female (meanage,60.7years±9.6)participants.comparedto thecontrolgroupparticipants,golduparticpantshad significantly lower mean values of FEV1 percent predicted,fvcpercentpredicted,fev1/fvcratioand hadsignificantlygreaterpackyearsofsmokinganda highermeanbmi(allp<0.0001). Comparison Between GOLDU and Control Group Regarding Parenchymal Abnormalities The results of comparison between the GOLDU groupandthecontrolgroupforctparametersrelatedto lungparenchymalabnormalitiesareshownintable2. Comparedtothecontrolgroup,GOLDUindividualshad significantlygreaterprevalenceofparaseptalemphysema (Figure1),linearatelectasis,andnon-fibroticILDand hadagreaternumberoflobesinvolvedbybronchialwall Table 1. Participant Characteristics and Physiologic Measurements Characteristics All GOLDU Smoking Control No.ofpatients Male:n,(%) 172(43) 71(36) 101(51) Age(y):mean± 59.5± ± ± SD(range) (45-80) BMI(kg/mm 2 ) 30.1± ± ±5.5 <.0001 FEV1%pred 85.5± ± ±11.4 <.0001 ( ) FEV1,pre,L 2.5±0.7( ) 2.0± ±0.7 <.0001 FEV1,post,L 2.5±0.7( ) 2.1± ±0.7 <.0001 FEV1, 35(8.8) 21(10.6) 14(7.1) 0.22 reversibility, a n,(%) FVC%pred 86.3± ± ±11.3 <.0001 ( ) FVC,pre,L 3.2±0.9( ) 2.8± ±1.0 <.0001 FVC,post,L 3.2±0.9( ) 2.8± ±0.9 <.0001 FVC, 21(5.3) 13(7.0) 8(4.0) 0.26 reversibility, a n,(%) FEV1/FVC 0.76± ± ±0.05 <.0001 ( ) TLC,L 5.3±1.1( ) 5.1± ±1.1 <.0001 Packyears: 41.5± ± ±20.2 <.0001 mean±sd ( ) (range) African 104(26) 48(24) 56(28) American race:n,(%) Currently 210(53) 109(55) 101(51) Smoking: n,(%) Dataarepresentedasmean±standarddeviation(SD)orpercentage a Reversibilityof200mlineitherFVCorFEV1 thickening,bronchiectasis,centrilobularnodules(figure 2),mosaicattenuation,atelectasis,GGO,andreticular opacities(allp<0.05).however,theprevalenceofcentrilobularemphysema,bullae,fibroticild,honeycombing cystsandextentofemphysemaandexpiratoryairtrapping wasnotsignificantlydifferent. Comparison Between GOLDU and Control Group Regarding Chest Wall Abnormalities Regardingthemeasurementofthesubcutaneousfat thickness,inonecasewithinthecontrolgroupand3 casesingoldu,measurementswererestrictedbythe edgeoffieldofview.theresultsofcomparisonbetween GOLDU and the control group regarding chest wall abnormalities are shown in Table 3. Compared to the

5 92 CT Findings in Nonobstructed Spirometric Smokers Table 2. Prevalence of Lung Parenchymal Abnormalities in GOLDU and Smoking Control CT parameters (visual) GOLDU Smoking (N=200) Control (N=200) PresenceofEmphysema a (%) TypeofEmphysema (centrilobular,%) ExtentofEmphysema a 0(0,2) 0(0,2) 0.37 (scale,median) ParaseptalEmphysema(%) Bullae(%) BronchialWallThickening 2.87± ±2.41 <.0001 Bronchiectasis 0.57± ± CentrilobularNodules 1.86± ±2.54 <.0001 ExpiratoryAirTrapping 1.00± ± MosaicAttenuation 0.35± ±0.88 <.0001 Atelectasis(linear,%) <.0001 Atelectasis 1.04± ±0.83 <.0001 FibroticILD(%) Non-fibroticILD(%) GGO 0.39± ± Reticular 0.40± ± Honeycomb(%) Table 3. Prevalence of Chest Wall Abnormalities and Quantitative CT Measurement in GOLDU and Smoking Control CT parameters GOLDU Smoking (N=200) Control (N=200) Pleural Thickening (%) Diffuse Focal Diaphragm Changes UnilateralEventration(%) BilateralEventration(%) Hernia-Rt.(%) Hernia-Lt.(%) Internal Thoracic Diameter (mm) Transverse 255.0± ±22.4 <.0001 Anterior-Posterior 188.3± ± SubcutaneousFatThickness(mm) 23.4± ±7.78 <.0001 Kyphosis(%) Scoliosis(%) VertebralBodyWedging(%) CT Measurements GOLDU Smoking (quantitative) (N=200) Control (N=200) EmphysemaIndex(%) 2.02± ± GasTrapping(%) 11.16± ± TLC(CT,liters) 5.06± ±1.14 <.0001 TLC(CT,%predicted) 91.15± ±10.54 <.0001 WA(%) 61.91± ±2.50 <.0001 Dataarepresentedasmean±standarddeviation(SD)orpercentage Dataarepresentedasmean±standarddeviation(SD)ormedian(min,max)orpercentage a Emphysemaisexpressedascentrilobular,panlobularormixedtypeemphysema Figure 1. Paraseptal Emphysema (Arrows) on Coronal CT Reconstruction in a GOLDU Participant Figure 2. Prominent Centrilobular Nodules (White Circles) on Axial CT in A GOLDU Participant

6 93 CT Findings in Nonobstructed Spirometric Smokers Figure 3. Right Diaphragmatic Elevation in a GOLDU Participant. control group, GOLDU participants had a significantly higher prevalence of diaphragmatic eventration (Figure 3) and kyphosis (Figure 4) (all P < 0.01). GOLDU participants had narrower transverse internal thoracic diameter and thicker subcutaneous fat than those of control group participants (both P < ). There were no prevalence differences in scoliosis, vertebral body wedging, and total involved number of vertebral body wedging. Comparison Between GOLDU and Control Group Regarding QCT Measurements The results of comparisons between GOLDU and the control group for QCT parameters are shown in T able 3. Compared to the control group, GOLDU participants had significantly lower TLCCT (L) and higher WA% (all P < ), but none of the individuals in either group showed TLCCT less than 80% predicted. There were no differences in EI (%) and gas trapping (%) between the GOLDU and control groups. Significant Imaging Features Associated With GOLDU After adjusting for age, race, gender, pack years, and BMI, multiple logistic regression analysis showed that significant features associated with GOLDU were the presence of paraseptal emphysema, linear atelectasis, diaphragmatic eventration, greater number of lobes involved in bronchial wall thickening, bronchiectasis, centrilobular nodules, mosaic attenuation, atelectasis, narrower transverse internal thoracic diameter, and thicker subcutaneous fat (all P < 0.05, Table 4). Presence Figure 4. Moderate Kyphosis in a GOLDU Participant. of non-fibrotic ILD, kyphosis, extent of GGO and reticular opacities, were not independently associated with the GOLDU group. Multivariate analysis suggested that the number of lobes involved in bronchial wall thickening, centrilobular nodules, linear atelectasis, and diaphragmatic eventration all were associated with reduction of TLCCT (all P < 0.05). However, paraseptal emphysema, bronchiectasis, and mosaic attenuation did not show significant associations. One hundred and fifty-eight participants (79%) among the GOLDU group had one radiologic finding that could contribute to pulmonary restriction such as fibrotic or non-fibrotic ILD, GGOs, reticular abnormalities, atelectasis, diffuse or focal pleural thickening, or eventration of diaphragm compared with 103 participants (52%) in the control group. Inter-rater agreements for visual assessment in GOLDU and smokers with normal lung function are shown in Table 5. Discussion In this study, we demonstrated that in the GOLDU group, radiologic features including parenchymal lung journal.copdfoundation.org I JCOPDF 2014 Volume 1 Number

7 94 CT Findings in Nonobstructed Spirometric Smokers Table 4. Results of Multiple Regression Analysis for Imaging Features on GOLDU a CT parameters ParaseptalEmphysema(%) BronchialWallThickening <.0001 Bronchiectasis CentrilobularNodules <.0001 MosaicAttenuation <.0001 Atelectasis(linear,%) Atelectasis DiaphragmaticEventration-Lt.(%) TransverseInternalThoracicDiameter(mm) SubcutaneousFatThickness(mm) <.0001 Non-fibroticILD(%) GGO Reticular Kyphosis(%) DiaphragmaticEventration-Rt.(%) a AdjustedR 2 formultipleregressionanalysisongoldu= disease,aswellaschestwallabnormalities,werepresent and could be discriminated from those of the control group of smokers with normal lung function. These findings could be helpful in explaining the unique physiologicfeaturesingoldu. ReducedFEV1orFVCinthesettingofapreserved FEV1/FVC ratio is usually regarded as a restrictive pattern on spirometry. In keeping with this, GOLDU participants in our study showed significantly lower TLC CT whencomparedwiththecontrolgroupparticipants. Despite having lower TLC values than control participants,noneofthegolduparticipantshadtlc valuesbelow80%predicted.ingeneral,reducedtlc CT may be due either to intrinsic lung disease causing increasedlungstiffness,orabnormalitiesofthechest wall,pleura,diaphragm,abdomen,orthoracicbonycage whichaffectventilation.thisstudydemonstratedthat thegolduparticipantsshownotonlyhigherprevalence ofparenchymalabnormalities(paraseptalemphysema, atelectasis,non-fibroticild,bronchialwallthickening, bronchiectasis,centrilobularnodules,mosaicattenuation, GGO,andreticularopacities),butalsoahigherprevalence of non-pulmonary findings, including diaphragmatic eventration,kyphosis,narrowertransverseinternalthoracic diameterandthickersubcutaneousfat.furthermore,all of these findings, except GGO, reticular opacities of non-fibroticild,andkyphosis,remainedassignificant and independent predictors after adjusting for age, race, gender, pack years, and BMI. These covariates Table 5. Mean Inter-Rater Agreements for Visual Assessment in GOLDU and Smoking Control CT parameters GOLDU Smoking control PresenceofEmphysema TypeofEmphysema ExtentofEmphysema ParaseptalEmphysema Bullae BronchialWallThickening a Bronchiectasis a CentrilobularNodules a ExpiratoryAirTrapping a MosaicAttenuation a Atelectasis(linear) Atelectasis a FibroticILD Non-fibroticILD GGO* Reticular a PleuralThickening b DiaphragmChanges c BonyThoracicChanges d a Meankcoefficientforinter-rateragreementin6lobes b Meankcoefficientfordiffuseandfocalpleuralthickening c Meankcoefficientforunilateralandbilateraldiaphragmaticeventration d Meankcoefficientforkyphosisandscoliosis werechosenbecausetheyareknowntoaffectpulmonary function. 7 Inpreviousstudies,decreasedFEV1andFVCwith resultantpreservationofthefev1/fvcratiohasbeen relatedtoobesitywithincreasedbmi Inourstudy, thesignificantlyincreasedsubcutaneousfatthicknessin thegoldugrouprelativetothatofthecontrolgroup may suggest the importance of obesity as a cause of pulmonaryrestriction.however,highbmialonedoes notaccountforthereductionoffev1andfvcinthe GOLDUgroup. 22,23 Several studies have shown that WA% is related to BMIandWA%variessignificantlybyraceandsex. 25,26 Inourstudy,evenafteradjustmentforage,sex,andBMI, thegoldugrouphadahighernumberoflobesinvolved inbronchialwallthickeningonvisualassessmentand increasedwa%onqctthanthecontrolgroup.thismay reflectahigherprevalenceofbronchialinflammationor remodelinginthegoldugroup.thefindingofparaseptalemphysemaasasignificantindependent,associated factorwithgoldustatusisinteresting.wespeculate thatparaseptalemphysemamayactasaspace-occupying lesion in the lung parenchyma, maybe with resultant

8 95 CT Findings in Nonobstructed Spirometric Smokers decreaseintlcincontrasttotheairwayobstruction foundincentrilobularandpanlobularemphysema.the presenceofcentrilobularnodulesandmosaicattenuation, both markers of small airways disease, in GOLDU participants suggests that small airways disease may sometimes be associated with restrictive physiology. Although smoking and COPD would more likely be associatedwithflatteningofdiaphragm,inourstudy, unilateralandbilateraldiaphragmaticeventrationswere significantlyassociatedwithgoldustatusonunivariate analysis,andtheypersistedasanindependentlyassociated factor on multivariate analysis. This finding suggests thateventrationmaycontributeimportantlytopulmonary restrictionandshouldbereportedwhenfound.thenovel findingthattheinternaltransversethoracicdiameter,but nottheanteroposteriordiameter,wasassociatedwith GOLDUstatussuggeststhattransversenarrowingofthe chestwallmayalsocontributetopulmonaryrestriction. Physiologically, decreased FEV1 with preservation of FEV1/FVC in GOLDU may be caused by volume derecruitment in alveoli during forced expiration. However, there were no significant differences in gas trappingindex(%)betweenbothgroups,althoughwedid notcomparethegastrappingindex(%)lobe-by-lobein 6regionsofthelung. Our study has some limitations. First, inter-rater agreementsforvisualassessmentingolduandsmokers withoutcopdwerevariableandrelativelylow(kappa coefficientrange: ),especiallyinfactorsfound to be different between the 2 groups, identified in adjustedanalysis.however,whenweconsiderthenature oflesionssuchasbronchiectasis,bronchialwallthickening, centrilobular nodules, and mosaic attenuation, these lesionswererelativelysubtlelesionscomparedtothose infaradvancedcopdandwereliabletobesubjectivein radiologists readings.despitetherelativelylowinter-rater agreements,thevalueswereverysimilarinbothgroups, suggesting that the suggested significant imaging featuresassociatedwithgoldumightreflectthetrue differencesbydiseasestateratherthanbiasofradiologist s adjudication.second,althoughstandardizedbreathing instructionswereemployed,itispossiblethatparticipantswereunableorunwillingtoperformanadequate inspiration, which may have contributed to decreased TLCintheGOLDUparticipants.Third,inabilitytomeasure subcutaneousfatthicknessinallcases(becauseoffield ofviewrestriction)isafurtherlimitation. Inourstudy,wedemonstratethattherewerevariables influencingdiseaseprocessesidentifiableonct,which mayalsoaffecttheresultsinfev1andfvc.underthe current COPD criteria staging system, based on the resultsofspirometry,theheterogeneityofgoldumay notbeappreciatedwellinthediagnosisandtreatment of COPD patients. Even though our results showed various meaningful CT parameters in the GOLDU participants,anewgoldstagingsystemwillbeneeded for allowing the identification of GOLDU individuals havingmorethanoneconcurrentdiseaseprocess. 27 Conclusion In cigarette smokers, chest wall abnormalities and parenchymallungdisease,whichcontributetorestrictive physiologic impairment, are associated with GOLDnonobstructedstatus. Acknowledgements The authors wish to gratefully acknowledge the volunteerswhoparticipatedinthecopdgene study. Declaration of Interest Noconflictsofinterest

9 96 CT Findings in Nonobstructed Spirometric Smokers References: 1. GlobalInitiativeforChronicObstructiveLungDiseasestrategy forthediagnosis,managementandpreventionofchronicobstructive pulmonarydisease:anasia-pacificperspective.respirology.2005; 10(1): IqbalA,SchlossS,GeorgeD,IsonakaS.Worldwideguidelinesfor chronicobstructivepulmonarydisease:acomparisonofdiagnosis andtreatmentrecommendations.respirology.2002;7(3): CelliBR,MacNeeW.Standardsforthediagnosisandtreatmentof patientswithcopd:asummaryoftheats/erspositionpaper. Eur Respir J.2004;23(6): KohlerD,FischerJ,RaschkeF,SchonhoferB.UsefulnessofGOLD classificationofcopdseverity.thorax.2003;58(9): ArneM,LisspersK,StallbergB,etal.Howoftenisdiagnosisof COPD confirmed with spirometry? Respir Med. 2010; 104(4): ZielinskiJ,BednarekM,GoreckaD,etal.IncreasingCOPDawareness. Eur Respir J.2006;27(4): WanES,HokansonJE,MurphyJR,etal;COPDGeneInvestigators. ClinicalandradiographicpredictorsofGOLD-unclassifiedsmokers in the COPDGene study. Am J Respir Crit Care Med. 2011; 184(1): ArneM,LisspersK,StallbergB,etal.Howoftenisdiagnosisof COPD confirmed with spirometry? Respir Med. 2010;104(4): RabeKF,HurdS,AnzuetoA,etal;GlobalInitiativeforChronic Obstructive Lung Disease. Global strategy for the diagnosis, management,andpreventionofchronicobstructivepulmonary disease:goldexecutivesummary.am J Respir Crit Care Med. 2007;176(6): ReganEA,HokansonJE,MurphyJR,etal.Geneticepidemiology ofcopd(copdgene)studydesign.copd.2010;7(1): HanMK,KazerooniEA,LynchDA,etal;COPDGeneInvestigators. Chronic obstructive pulmonary disease exacerbations in the COPDGenestudy:associatedradiologicphenotypes.Radiology. 2011;26(1): KimSS,SeoJB,LeeHY,etal.Chronicobstructivepulmonary disease:lobe-basedvisualassessmentofvolumetricctbyusing standardimages--comparisonwithquantitativectandpulmonary functiontestinthecopdgenestudy.radiology,2013;266(2): HansellDM,BankierAA,MacMahonH,McLoudTC,MullerNL. RemyJ.FleischnerSociety:glossaryoftermsforthoracicimaging. Radiology.2008;246(3): GevenoisPA,deMaertelaerV,DeVuystP,ZanenJ,YernaultJC. Comparisonofcomputeddensityandmacroscopicmorphometry in pulmonary emphysema. Am J Respir Crit Care Med.1995; 152(2): TschirrenJ,HoffmanEA,McLennanG,SonkaM.Segmentation and quantitative analysis of intrathoracic airway trees from computed tomography images. Proc Am Thorac Soc. 2005; 2(6):484-87, PalagyiK,TschirrenJ,HoffmanEA,SonkaM.Quantitativeanalysis of pulmonary airway tree structures. Comput Biol Med. 2006; 36(9): Zhang L, Hoffman EA, Reinhardt JM. Atlas-driven lung lobe segmentationinvolumetricx-rayctimages. IEEE Trans Med Imaging. 2006;25(1): HoffmanEA,SimonBA,McLennanG.StateoftheArt.Astructural and functional assessment of the lung via multidetector-row computed tomography: phenotyping chronic obstructive pulmonarydisease.proc Am Thorac Soc.2006;3(6): OkazawaM,MullerN,McNamaraAE,ChildS,VerburgtL,Pare PD.Humanairwaynarrowingmeasuredusinghighresolution computed tomography. Am J Respir Crit Care Med. 1996; 154(5): Standardizationofspirometry,1994update.AmericanThoracic Society.Am J Respir Crit Care Med.1995;152(3): ManninoDM,HolguinF,PavlinBI,FerdinandsJM.Riskfactors forprevalenceofandmortalityrelatedtorestrictiononspirometry: findingsfromthefirstnationalhealthandnutritionexamination Surveyandfollow-up.Int J Tuberc Lung Dis. 2005;9(6): SinDD,JonesRL,ManSF.Obesityisariskfactorfordyspnea but not for airflow obstruction. Arch Intern Med. 2002;162(13): SchachterLM,SalomeCM,PeatJK,WoolcockAJ.Obesityisa riskforasthmaandwheezebutnotairwayhyperresponsiveness. Thorax.2001;56(1): JonesRL,NzekwuMM.Theeffectsofbodymassindexonlung volumes.chest. 2006;130(3): Ogawa E, Nakano Y, Ohara T, et al. Body mass index in male patientswithcopd:correlationwithlowattenuationareasonct. Thorax.2009;64(1): MairG,MaclayJ,MillerJJ,etal.AirwaydimensionsinCOPD: relationshipswithclinicalvariables.respir Med.2010;104(11): BonJM,WeissfeldJL,SciurbaFC.Auniquespirometricphenotype incopd.am J Respir Crit Care Med. 2008;178(4):