Quantitative evaluation of bone-resorptive lesion volume in osteonecrosis of the femoral head using micro-computed tomography

(1)

九州大学学術情報リポジトリ

Kyushu University Institutional Repository

Quantitative evaluation of bone-resorptive lesion volume in osteonecrosis of the femoral head using micro-computed tomography

馬場, 省次

http://hdl.handle.net/2324/4060038

出版情報：九州大学, 2019, 博士（医学）, 課程博士バージョン：

(2)

Availableonlineat

ScienceDirect

www.sciencedirect.com

Original article

Quantitative evaluation of bone-resorptive lesion volume in

osteonecrosis of the femoral head using micro-computed tomography

Shoji Baba , Goro Motomura

^∗

, Satoshi Ikemura , Yusuke Kubo , Takeshi Utsunomiya , Hiroyuki Hatanaka , Koichiro Kawano , Yasuharu Nakashima

DepartmentofOrthopaedicSurgery,GraduateSchoolofMedicalSciences,KyushuUniversity,3-1-1,Maidashi,Higashi-ku,Fukuoka,812-8582,Japan

a r t i c l e i n f o

Articlehistory:

Accepted4September2019 Availableonline12September2019

Keywords:

Osteonecrosisofthefemoralhead Boneresorption

Micro-CT Collapse

a b s t r a c t

ObjectivesToquantifythevolumeofbone-resorptivelesionsinpost-collapseosteonecrosisofthefemoral head(ONFH)usingmicro-computedtomography(micro-CT)andassesstheircharacteristicsinpost- collapseONFH.

MethodsWeinvestigated35femoralheadsresectedfrom35patientswithONFH(20menand15 women;meanage,47.2years).Oneachofsevencoronalhigh-resolutionmicro-CTslicesofthefemoral head,thebone-resorptiveareaswereextractedusingbonemicrostructuremeasurementsoftware.Next, thetotalbone-resorptivevolumeratio,deﬁnedastheratioofallbone-resorptivecross-sectionalareasto allfemoralheadcross-sectionalareasinallsevenslices,wascalculated.Associationsbetweentotalbone- resorptivevolumeratioandsex,age,ONFH-associatedfactors,patientworkloadlevels,ONFHstage,ONFH type,necroticvolumeonmagneticresonanceimaging,anddurationfromtheonsetofpaintosurgery wereanalyzed.Lesionlocationandtheassociationbetweenbone-resorptivelesionandcollapsewere alsoevaluated.

ResultsThemeantotalbone-resorptivevolumeratiowas7.0±6.0%,whichvariedsigniﬁcantlybyONFH stage(ARCOcollapsequantitation3A,3.5±2.1%;3B,6.8±3.0%;and3C,13.6±8.8%).ONFHstagewas independentlyassociatedwithtotalbone-resorptivevolumeratio(P<0.05).Furthermore,highbone- resorptivevolumeratioswerefoundintheanteriorfemoralheadandwereassociatedwithcollapse.

ConclusionsThisstudydemonstratedthatbone-resorptivevolumeinpost-collapseONFHwassignif- icantlyassociatedwiththediseasestage,whichwasmorewidespreadintheanteriorportionofthe femoralheadthanintheposteriorportion.

1. Introduction

Nontraumatic osteonecrosis of the femoral head (ONFH) is characterizedby femoralheadcollapseseveralmonthstoyears aftertheoccurrence ofosteonecrosis [1].Previous studieshave demonstratedthatacollapsedfemoralheadwithONFHinevitably involvesasubchondralfracture[2],whichmightbethemechanism ofcollapse initiation.Aftercollapse(ora subchondralfracture), somefemoralheadsmightpreservearelativelysphericalshape, whileothersmightbecomeseverelydeformed.Theshapeofthe femoralheadaftercollapseisessentialtoconsiderwhenselecting atherapeuticstrategy;however,thepathomechanismsofcollapse progressionremainunclear.

∗ Correspondingauthor.

E-mailaddress:[email protected](G.Motomura).

In post-collapse ONFH, bone-resorptive lesions can be seen withinthefemoralhead[3,4].BoneresorptioninONFHisconsid- eredtobecausedbyosteoclastsduringtheinitialphaseoftherepair process[5],whileboth theextentandlocation ofbone resorp- tionseem tobedifferentamong patientsin ONFH.Plenketal.

reportedthatthereareseveraltypesofrepairprocessesinONFH,in whichfemoralheadswithpredominantboneresorptionundergo moredestructionwithtime[4],suggestingthepossibleassociation betweenboneresorptionandcollapseprogression.

Ontheotherhand,tothebestofourknowledge,therehave been no studies that have quantitatively characterized bone resorptioninONFH.Withconventionalclinicalcomputedtomog- raphy(CT),detailedanalysisofbone-resorptivelesionsincluding their distribution and dimensions has been difﬁcult. However, usingmicro-CTallowsustoevaluatethethree-dimensionalbone structureoftheentirefemoralheadindetailandquantitativelyat themicro-level[6–8].Therefore,thepurposesofthisstudywere https://doi.org/10.1016/j.jbspin.2019.09.004

(3)

76 S.Babaetal./JointBoneSpine87(2020)75–80

toquantifybone-resorptivelesionsusingmicro-CTandassessthe characteristicsofbone-resorptivelesionsinpost-collapseONFH.

2. Methods 2.1. Patients

Thisstudy was approved by the institutional review board.

Informedconsentwas obtainedfrom allindividualparticipants includedinthestudy.BetweenApril2012andAugust2017,77hips in67patientswithONFHunderwentprimaryjointreplacement surgery(totalhiparthroplastyorbipolarhemiarthroplasty)atour institution.Ofthese77hips,27wereexcludedfromthisstudyfor thefollowingreasons:21hipshadsevereosteoarthriticchanges, fourhipshadseverefemoralheaddeformityduetocollapse,and twohipshadeccentrically-locatednecroticlesions.Inaddition,five hipsinfivepatientstreatedwithabisphosphonateintakeforosteo- porosis wereexcluded.In 10 patientswho underwentbilateral jointreplacementsurgery,onehipineachpatientwasrandomly selectedforanalysistofulfillthestatisticalassumptionofindepen- dentobservation.Thus,weinvestigated35entirefemoralheads resectedfrom35patientswithONFH(20menand15women)with ameanageof47.3years(median46years,range18–81years).

FactorsassociatedwithONFHincludedcorticosteroiduse(>2gof prednisoloneoritsequivalentwithina3-monthperiod)in20hips andalcoholabuse(>320g/week)in15hips,whichwereclassified basedontheAssociationResearchCirculationOsseous(ARCO)eti- ologicclassificationcriteria[9,10].Inaddition,weclassifiedpatient workloadlevelsintothreegroups:lowactivity(unemployed)for 12hips,moderateactivity(workinanofficeenvironmentorthe hospitalityindustry,etc.)for17 hips,andhighactivity(nursing, workingforadeliveryserviceorinconstruction,etc.)forsixhips.

BasedontheARCOinternationalclassificationofosteonecrosis,all 35femoralheadswereclassifiedintostage3[11].Inthecurrent study,15femoralheadswith≤2mmofcollapseweredefinedas stage3A,12femoralheadswith2–4mmofcollapseasstage3B, andeightfemoralheadswith≥4mmasstage3C,byreferenceto theARCOquantitationofadomedepression[11].Eveninpatients

withnoobviousseveredeformityofthefemoralhead,weconsid- eredjointreplacementsurgerytobeindicatedwhentherewasan extensivenecroticlesionwithasubchondralfractureconfirmedon CTorbonemarrowedemasuggestiveofasubchondralfractureon magneticresonanceimaging(MRI)[12].Inaddition,thelocationof necroticlesionwasclassifiedusingtheJapaneseInvestigationCom- mitteeclassificationsystem[13]:typeC1(morethanthemedial two-thirdsoftheweight-bearingportionbut notextendinglat- erallytotheacetabularedge)in14hips,andtypeC2(extending laterallytotheacetabularedge) in21 hips.The meanduration fromtheonsetofpaintosurgerywas11.7±9.3months(median 7months,range 1–36months).Furthermore,thevolumeofthe necroticlesiononMRIwasdefinedasacombinednecroticangle usingthemodifiedKerboulmethod[14].Briefly,theanglesofthe femoralsurfacewithnecrosisonamid-sliceinT1-weightedcoronal andaxialviewsweremeasuredandsummed.Themeancombined necroticanglewas277.4±63.4(median262,range170–465).

2.2. Quantiﬁcationofbone-resorptivelesions

All35 sampleswerescanned withhigh-resolutionmicro-CT (RmCTT1,Rigaku,Tokyo,Japan)immediatelyaftersurgery.Micro- CTwasperformedatavoltageof60kV,currentof60␮A,voxelsize of133×133×133␮m³,andslicethicknessof0.4mm.Inthecur- rentstudy,abone-resorptiveareawasdeﬁnedasaspace≥1mm indiameterandcoronalimagesofmicro-CTwereinvestigated.On eachcoronalmicro-CTslice,bone-resorptiveareaswereextracted usingbonemicrostructuremeasurement software(TRI/3D-BON, RatocSystemEngineering,Tokyo,Japan) followinga previously reportedmethod(Fig.1A–F)[8].Brieﬂy,boneregionsincludingthe bonetrabeculaeand thecorticalshellweredepictedbybinariz- ingCTimagesusingabonethreshold(Fig.1B).Next,theentire regionwasdepicted(Fig.1C).Emptyregionsnotinvolvingbone weredepictedby subtracting thebone regionsfrom theentire region(Fig.1D). Inordertoidentifyemptyregionsof≥1mmin diameter,emptyregions<1mmindiameterwereclearedtoerode theallemptyregionsthree-dimensionallyby0.5mmdecrements fromtheirsurface(Fig.1E).Torestoretheoriginalshapeofthe

Fig.1.Quantiﬁcationofbone-resorptiveareasdeﬁnedasspaces≥1mmindiameterusingbonemicrostructuremeasurementsoftware(TRI/3D-BON,RatocSystemEngi- neering,Tokyo,Japan).A.Acoronalmicro-CTimage;B.Boneregions,includingthebonetrabeculaeandthecorticalshell,aredepictedafterbinarizingCTimagesusinga bonethreshold;C.Theentireregionofafemoralheadisdepicted;D.Emptyregionsnotinvolvingbonearedepictedbysubtractingtheboneregionsfromtheentireregion;

E.Inordertoidentifyemptyregionsof>1mmindiameter,emptyregionslessthan1mmareclearedtoerodefromthesurfacethree-dimensionallyin0.5mmdecrements;

F.Torestoretheoriginalshapeoftheremainingemptyregionsintheaboveprocess,theirsurfacesaredilatedby0.5mmincrementsfromtheirsurface.Oftheremaining emptyregions,abone-resorptiveareaisdeﬁnedasaregionconnectedtoanecroticlesion.

(4)

Fig.2. Distinguishingbetweenbone-resorptiveareasandbonedefectareas.

Bone-resorptiveareas(whitearrows)aredistinguishedfrombonedefectareas (whitearrowheads)causedbyasubchondralfractureorcollapsebasedonCTvalues.

remainingemptyregionsintheaboveprocess,theirsurfaceswere dilatedby0.5mmincrements(Fig.1F).Oftheremainingempty regions,abone-resorptiveareawasdefinedasaregionconnected toanecroticlesionanddistinctfromareasofbonedefectcausedby asubchondralfractureorcollapsebasedonCTvalues(Fig.2).The definitionofeachbone-resorptiveareawasbasedonconsensus amongthreeobservers(S.B.,T.U.,andH.H.);theyalsoconfirmed theareaofnecrosiswithMRI.Aftermicro-CT,allsampleswere madeaconfirmeddiagnosisofONFHbyahistopathologicaleval- uationbasedonpreviouslyreportedhistopathologicalcriteriaof ONFH[15,16].

2.3. Assessmentofbone-resorptivelesioncharacteristics

Inordertoquantitativelyassessthebone-resorptivevolumein theentirefemoralhead,atﬁrst,thecoronalslicesoffemoralhead weredividedequallyintosevenslices(Fig.3).Ineachslice,the cross-sectionalareaofthefemoralheadwasdeﬁnedasthearea ofaconcentriccirclecenteredonthecenterofthefemoralhead.

Bothbone-resorptiveandfemoralheadcross-sectionalareaswere calculatedusingtheImageJsoftwareprogram(NationalInstitutes ofHealth,Bethesda,MD,USA).Themeanratioofbone-resorptive areastofemoralheadcross-sectionalareaswascalculatedforeach slice.Inaddition,thetotalbone-resorptivevolumeratioineach femoralheadwasdeﬁnedastheratioofallbone-resorptiveareas toallfemoralheadcross-sectionalareasinsevenslices.Univariate

associationsbetweenthedegreeoftotalbone-resorptivevolume ratio and sex,age, associatedfactors of ONFH(steroidor alcohol),stageandtypeofONFH,patientworkloadlevels,combined necroticangle,anddurationfromtheonsetofpaintosurgerywere analyzedfirst.Subsequently,afterconfirmingtheareaofnecrosis withMRI,theratioofthebone-resorptiveareatofemoralhead areainbothintra-necroticandextra-necroticlesionswereinves- tigated for each slice. Furthermore, thepresence or absenceof collapseinthesevensliceswasrecordedforall35femoralheads, andtherelationshipbetweenbone-resorptiveareasandfemoral headcollapsewasinvestigatedbycomparingbone-resorptiveareas betweensliceswithandwithoutcollapse. Inthecurrent study, collapsewasdefinedasabreakageofthefemoralheadcontour, includingasubchondralfracture.Finally,thecorrelationbetween collapseprogression rateandbone-resorptivevolumeratiowas investigatedusing31of35femoralheadsthatunderwentclinical CTforpreoperativeplanningatanaverageof72.7days(median 13days,range 1–837days)beforesurgery.Briefly,theprogres- sionofcollapseperdaywasdefinedasthedifferencebetweenthe collapsedepthoncorrespondingmid-coronalCTslicesbeforeand aftersurgerydividedbythenumberofdaysbetweenthetwoscans.

Collapse depthaftersurgery wasmeasuredusingthe micro-CT imagesofsurgicallyresectedfemoralheadsscannedimmediately aftersurgery.OneachofCTandmicro-CTimages,collapsedepth wasmeasuredasthedistancebetweentheconcentriccirclethat ﬁtsthefemoralheadandthemaximumincursionintothefemoral headonthelinepassingthroughthecenterofthefemoralhead.

2.4. Statisticalanalysis

Dataareexpressedasmeans±SD.Allvariablesweretestedfor normalitywith theShapiro-Wilk test and for homoscedasticity withLevene’stest.Associationsbetweensex,associated factors, type of ONFH, and bone-resorptive volume ratio, respectively, were compared using Wilcoxonrank sumtest. The association between bone-resorptivevolume in slices with versus without collapsewasalsoinvestigatedusingWilcoxonranksumtest.The bone-resorptivevolumeratioforeachstageorpatientworkload levelwerecomparedusingSteel-Dwassmultiple-comparisontest.

Thecorrelationsbetweenage,durationfromtheonsetofpainto surgery, combined necrotic angle, and bone-resorptive volume ratio, respectively, were analyzed using Spearman’s correlation coefﬁcient.The correlationbetweentheprogressionof collapse per day and bone-resorptive volume ratio was also analyzed

Fig.3. Methodsusedtoquantifythebone-resorptivevolumeoftheentirefemoralhead.

Inordertoquantitativelyassessthebone-resorptivevolumeintheentirefemoralhead,thecoronalslicesofafemoralheadaredividedequallyintosevenslices.Thetotal bone-resorptivevolumeratioineachfemoralheadisdeﬁnedastheratioofallbone-resorptiveareastoallfemoralheadcross-sectionalareasinsevenslices.

(5)

usingSpearman’scorrelationcoefficient.Significantfactorsasso- ciatedwithbone-resorptivevolume ratio wereidentified using multivariateanalysiswithstepwiseregressionthat includedall parameters. All statistical analyses were performed using the JMPsoftwareprogram(version13.0,SASInstituteInc.,Cary,NC, USA).P-valueslessthan0.05wereconsideredtobestatistically significant.

3. Results

Themeantotal bone-resorptivevolumeratiowas7.0±6.0%, whichwassigniﬁcantlydependentonONFHstage(3A,3.5±2.1%;

3B,6.8±3.0%;and3C,13.6±8.8%;P<0.05),whiledurationfrom theonset of pain tosurgery and patientworkload levelswere notassociatedwithbone-resorptivevolumeratio(Table1).Based onmultivariateanalysis,ONFHstagewasidentiﬁedtobeinde- pendently associated with total bone-resorptive volume ratio (P<0.01).

Withrespecttothelocationofbone-resorptiveareas,ahigh percentage of bone-resorptive cross-sectional areas tended to beseen in the anterior femoralhead (Fig.4). Therewas more bone-resorptive area in slice 1 than in slices 4–7 (P<0.01, respectively)andinslice2comparedwithslice6(P<0.05).Fur- thermore, most bone-resorptive areas in each slice and 81.9%

ofallbone-resorptiveareaswerefoundwithinnecroticlesions.

Meanbone-resorptive areain intra-necrotic lesions(5.9±7.1%) wassigniﬁcantlylargerthaninextra-necroticlesions(1.3±3.2%;

P<0.001).Inaddition,sliceswithfemoralheadcollapsewerealso more likely to befound in theanterior portion. Regarding the correlationbetweenbone-resorptivelesionsandfemoralheadcol- lapse,slices withcollapse hada signiﬁcantlyhigher meanratio ofbone-resorptiveareas(9.4±10.6%)thansliceswithoutcollapse (2.9±3.3%;P<0.001;Fig.5A).Themeanprogression ofcollapse perdaywas87.1±115.9␮m(median26.8␮m,range0–400␮m).

Therewasasigniﬁcantpositivecorrelationbetweentheprogres- sionofcollapseperdayandbone-resorptivevolumeratio,(=0.59, P<0.001;Fig.5B).

4. Discussion

The current quantitative study demonstrated that bone- resorptive volume within the femoral head is significantly associatedwiththepost-collapsestageofONFH.Inaddition,the collapseprogressionrateandbone-resorptivevolumeratiowere positivelycorrelated. Thesefindings suggest that thedegree of boneresorptionmightincreaseascollapseprogresses.Ontheother hand,wefoundnootherfactorsinfluencingbone-resorptivevolume, including the duration from the onset of pain or patient workloadlevels.Althoughitremainsunclearwhetherboneresorp- tionisanantecedentphenomenontocollapseorasubchondral fracture, the current results suggested a cause-and-effect rela- tionship between bone resorption and collapse progression in post-collapseONFH.

Severalstudiessupporttherelationshipbetweenboneresorp- tionandcollapseprogression.Plenketal.reportedthatthereare severaltypesofrepairprocessesinONFH,inwhichfemoralheads withpredominantboneresorptionundergomoredestructionwith time[4].Karasuyamaetal.histopathologicallydemonstratedthat signiﬁcantlymoreosteoclastsarepresentattheboundaryofthe necroticlesionwithcollapse versuswithoutcollapse[17].Inan animalstudy,increasedboneresorptionwasshowntobeassoci- atedwiththelossofstructuralintegrity,leadingtofemoralhead deformity [18]. Considering thesereports and ourﬁndings, we hypothesizethat the failure of mechanical stabilization due to excessiveboneresorptionmayleadtotheprogressionofcollapse.

Bisphosphonates,whichareosteoclastinhibitors,wereusedon atrialbasis topreventtheoccurrenceoffemoralheadcollapse basedontheconceptthatosteoclasticbone resorptioninitiates collapse [5]. Previous reports have indicated that bisphosphonates may have the potential to postpone the progression of femoralheadcollapse[19–21],whichsuggeststhatboneresorp- tionmaybeacauseofcollapseprogression.However,tothebest ofourknowledge,therehavebeennostudiesthathavequanti- tativelycharacterizedbone-resorptivelesionsinONFH.Although wecannotconcludewhetherbone-resorptivelesionsareacause oraconsequenceoffemoralheadcollapseduetothedesignof thisstudy,thisistheﬁrststudythatquantitativelycharacterized

Table1

Univariateanalysisoftherelationshipbetweenbone-resorptivevolumeratioandpatientcharacteristics.

Bone-resorptivevolumeratio(%) P-value Correlationcoefﬁcient

Mean±SD Median(IQR) Range

Sex^a

Male 5.9±6.7 4.2(1.9–7.0) 0.5–31.2 <0.05

Female 8.4±4.7 6.7(4.7–12.3) 2.0–18.5

Associatedfactor^a

Steroid 7.4±4.8 6.7(3.3–11.2) 0.5–18.5 0.22

Alcohol 6.4±7.5 4.7(2.0–7.2) 1.3–31.2

JICtype^a

C1 6.8±4.3 5.8(3.2–10.9) 1.9–15.2 0.56

C2 7.0±7.0 6.0(2.0–7.9) 0.5–31.2

ONFHstage^b

3A 3.5±2.1 3.1(1.9–5.4) 0.5–7.2 3Avs3B:<0.01

3B 6.8±3.0 6.8(4.2–9.0) 1.6–11.9 3Avs3C:<0.01

3C 13.6±8.8 12.6(7.2–17.7) 2.0–31.2 3Bvs3C:<0.05

Workloadlevels^b

Low 9.1±4.7 7.8(6.1–12.2) 1.9–18.5 LowvsMod.:0.15

Moderate 6.4±7.3 3.4(2.0–7.9) 0.5–31.2 LowvsHigh:0.09

High 4.1±2.2 4.1(1.9–5.9) 1.6–7.2 Mod.vsHigh:0.92

Age^c 0.27 0.19

Durationfromtheonsetofpaintosurgery^c 0.99 <0.01

Combinednecroticangle^c 0.46 0.13

SD:standarddeviation;IQR:interquartilerange;JIC:JapaneseInvestigationCommittee;ONFH:osteonecrosisofthefemoralhead.

aWilcoxonranksumtest.

b Steel-Dwassmultiple-comparisontest.

c Spearman’scorrelationcoefﬁcient.

(6)

Fig.4.Meanratioofbone-resorptivecross-sectionalareatofemoralheadcross-sectionalareaandpercentageofbone-resorptiveareainintra-necroticversusextra-necrotic lesionsbyslice.Themeanratioofbone-resorptiveareas(bargraph)andpercentageofsliceswithcollapse(linegraph)tendtobehigherintheanteriorportionofthe femoralhead.Thereismorebone-resorptiveareainslice1thaninslices4–7(P<0.01,respectively),aswellasinslice2versusslice6(P<0.05).Inaddition,withineach slice,mostbone-resorptiveareasarefoundinintra-necroticlesions.Thetotalbone-resorptiveareainintra-necroticlesions(5.9±7.1%)issigniﬁcantlylargerthanthetotal bone-resorptiveareainextra-necroticlesions(1.3±3.2%;P<0.001).

Fig.5. A.Theassociationbetweenbone-resorptivelesionandfemoralheadcollapse.

Themeanratioofbone-resorptiveareasinsliceswithcollapse(9.4±10.6%)issignificantlyhigherthanthatinsliceswithoutcollapse(2.9±3.3%,P<0.001);B.Thecorrelation betweentheprogressionofcollapseperdayandthebone-resorptivevolumeratio.Thereisasignificantpositivecorrelationbetweenthebone-resorptivevolumeratioand progressionofcollapseperday(=0.59,P<0.001).Thecurvedlineindicatesa95%confidenceellipsethatcontains95%ofthedata.

bone-resorptivelesionsinpost-collapseONFHanddemonstrateda possibleassociationbetweenbone-resorptivevolumeanddegree offemoralheadcollapseaswellasthecollapseprogressionrate.

Therefore,webelievethatthisstudyprovidesvaluableinformation forabetterunderstandingofbone-resorptivelesionsincollapsed femoralheadandservesasabasisforfutureinvestigationstopre- ventfemoralheadcollapseinpatientswithONFH.

Inthecurrentstudy,bone-resorptivelesionsweremorecom- monintheanteriorportionofthefemoralhead,andlesionlocation wassigniﬁcantlyassociatedwiththepresenceofcollapse.Since theanteriorportionofthefemoralheadissusceptibletomechan- icalstressduetothenatureofacetabularcoverage[22],itseems reasonablethatcollapse wasmore commonlyseenintheante- riorportionofthefemoralhead.Ontheotherhand,tothebest ofourknowledge,nostudieshavedemonstratedboneresorption beforecollapseinONFH.Althoughthecurrentstudyshowedsim- plybone-resorptivevolumeaftercollapse,wespeculatedthatbone

resorptionbeingpredominantlyintheanteriorportionmaybedue totheinﬂuenceofcollapse.

Themain limitationofthis studywastheinabilitytoassess time-dependentchangesinbone-resorptivevolumeprospectively.

Since we limited the subjects of the current study to whole femoral heads resected from patients with ONFH in order to assessthedetailedcharacteristicsofbone-resorptivelesionsusing micro-CT,thefindingsofthecurrentstudyreflectonlythestate of bone-resorptivelesionsatthetime ofsurgery. However,the current study demonstrated nosignificant correlation between duration fromtheonset ofpaintosurgeryand bone-resorptive volume. Therefore, we believe that this study provides useful information aboutboneresorptionin post-collapseONFH,even though evaluation occurred at one-time point. Second, regard- ingtherelationshipbetweenbone-resorptivevolumeandcollapse progressionrateperday,bone-resorptivelesionvolumeshouldbe ideallyassessedatbaselinetoevaluateitscausalrelationshipwith

(7)

collapseprogression.However,itisnotpossibletoperformmicro- CT beforesurgery. Alternatively, we utilized clinicalCT images takenforpreoperativeplanning.Sincetheimagequalityofclinical CTscanswastoopoortoquantifythevolumeofbone-resorptive lesions,weassessedthecorrelationbetweencollapseprogression rateand bone-resorptivevolume of theresected femoralhead.

Althoughnotideal,webelievethecurrentresultsprovidevalu- abledatatosupportarelationshipbetweenboneresorptionand collapse progressionin post-collapse ONFH.Finally, thepatient agedistributionwaswideandwedidnotevaluatepatientactivi- tiesofdailyliving,whichmaybeassociatedwiththeprogression offemoralheadcollapse.Althoughthisstudydemonstratedthat patientageandworkloadlevelswerenotsigniﬁcantlycorrelated withbone-resorptivevolume,inafutureprospectivestudy,eval- uationofpatientactivityanddetailedexaminationbyagegroup shouldbeperformedtoelucidatethepathomechanism of bone resorptioninONFHusinghigh-resolutionradiologicalequipment invivo.

5. Conclusions

Thisstudydemonstratedthatbone-resorptivevolumeinpost- collapseONFHwassigniﬁcantlyassociatedwiththediseasestage andcollapseprogressionrate,whichwasmorewidespreadinthe anteriorportionofthe femoralhead thanin theposteriorpor- tion.Webelievethatthisstudyprovidesabetterunderstandingof boneresorptionincollapsedfemoralheadsandpotentialclinical therapeutictargetsforONFH.

Authors’contributions

Studydesign:S.B.andG.M.Acquisitionofdata:S.B,G.M,S.I., andY.N.Studyexecution:S.B.,Y.K., T.U.,H.H.,and K.K.Drafting ofthemanuscript:S.B.Revisionofthemanuscript:GM,SI, and YK.Approvaloftheﬁnalversionofthemanuscript:S.B.,G.M.,S.I., Y.K.,T.U.,H.H.,K.K.,andY.N.Allauthorstakeresponsibilityforthe integrityofthedataanalysis.

Disclosureofinterest

Theauthorsdeclarethattheyhavenocompetinginterest.

Acknowledgements

We thank Junji Kishimoto, a statistician from the Digital MedicineInitiativeatKyushuUniversity,forhisadviceonthesta- tisticalanalysis.ThisworkwaspartiallysupportedbyGrants-in-Aid forScientiﬁcResearch(16K10906,16H07057,and19K09601)from theJapanSocietyforthePromotionofScience.

References

[1]OhzonoK,SaitoM,TakaokaK,etal.Naturalhistoryofnontraumaticavascular necrosisofthefemoralhead.JBoneJointSurgBr1991;73:68–72.

[2]MotomuraG,YamamotoT,YamaguchiR,etal.Morphologicalanalysisof collapsedregionsinosteonecrosisofthefemoralhead.JBoneJointSurgBr 2011;93:184–7.

[3]MurpheyMD,ForemanKL,Klassen-FischerMK,etal.Fromtheradiologic pathologyarchivesimagingofosteonecrosis:radiologic-pathologiccorrela- tion.Radiographics2014;34:1003–28.

[4]PlenkJrH,GstettnerM,GrossschmidtK,etal.Magneticresonanceimaging andhistologyofrepairinfemoralheadosteonecrosis.ClinOrthopRelatRes 2001;386:42–53.

[5]KenzoraJE,GlimcherMJ.Pathogenesisofidiopathicosteonecrosis:theubiqui- touscrescentsign.OrthopClinNorthAm1985;16:681–96.

[6]WangC,WangX,XuXL,etal.Bonemicrostructureandregionaldistributionof osteoblastandosteoclastactivityintheosteonecroticfemoralhead.PLoSOne 2014;9:e96361.

[7]MaJ-X,HeW-W,ZhaoJ,etal.Bonemicroarchitectureandbiomechanicsofthe necroticfemoralhead.ScientiﬁcReports2017:7.

[8]ChibaK,BurghardtAJ,OsakiM,etal.Three-dimensionalanalysisofsubchondral cystsinhiposteoarthritis:anexvivoHR-pQCTstudy.Bone2014;66:140–5.

[9]YoonBH,JonesLC,ChenCH,etal.Etiologicclassiﬁcationcriteriaofarcoon femoralheadosteonecrosispart1:glucocorticoid-associatedosteonecrosis.J Arthroplasty2019;34:163–8.

[10]YoonBH,JonesLC,ChenCH,etal.Etiologicclassiﬁcationcriteriaofarcoon femoralheadosteonecrosispart2:alcohol-associatedosteonecrosis.JArthro- plasty2019;34:169–74.

[11]JWM G. ARCOcommitteeon terminology andstaging. ARCO Newsletter 1993;5:79–82.

[12]TheruvathAJ,SukerkarPA,BaoS,etal.Bonemarrowoedemapredictsbone collapseinpaediatricandadolescentleukaemiapatientswithcorticosteroid- inducedosteonecrosis.EurRadiol2018;28:410–7.

[13]SuganoN,AtsumiT,OhzonoK,etal.The2001revisedcriteriafordiagnosis, classiﬁcation,andstagingofidiopathicosteonecrosisofthefemoralhead.J OrthopSci2002;7:601–5.

[14]HaYC,JungWH,KimJR,etal.Predictionofcollapseinfemoralheadosteonecro- sis:amodiﬁedKerboulmethodwithuseofmagneticresonanceimages.JBone JointSurgAm2006;88:35–40.

[15]YamamotoT,IrisaT,SugiokaY,etal.Effectsofpulsemethylprednisolone onboneandmarrowtissues:corticosteroid-inducedosteonecrosisinrabbits.

ArthritisRheum1997;40:2055–64.

[16]MotomuraG,YamamotoT,MiyanishiK,etal.Combinedeffectsofananti- coagulantandalipid-loweringagentonthepreventionofsteroid-induced osteonecrosisinrabbits.ArthritisRheum2004;50:3387–91.

[17]KarasuyamaK,YamamotoT,MotomuraG,etal.Theroleofscleroticchanges inthestartingmechanismsofcollapse:ahistomorphometricandFEMstudy onthefemoralheadofosteonecrosis.Bone2015;81:644–8.

[18]KimHK,SuPH.Developmentofﬂatteningandapparentfragmentationfollow- ingischemicnecrosisofthecapitalfemoralepiphysisinapigletmodel.JBone JointSurgAm2002;84a:1329–34.

[19]GianakosAL,Moya-AngelerJ,DuggalS,etal.Theefﬁcacyofbisphosphonates withcoredecompressionandmesenchymalstemcellscomparedwithbispho- sphonatesaloneinthetreatmentofosteonecrosisofthehip:aretrospective study.HSSJ2016;12:137–44.

[20]AgarwalaS,JainD,JoshiVR,etal.Efﬁcacyofalendronate,abisphosphonate,in thetreatmentofAVNofthehip.Aprospectiveopen-labelstudy.Rheumatology 2005;44:352–9[Oxford].

[21]KangP,PeiF,ShenB,etal.Aretheresultsofmultipledrillingandalendronate forosteonecrosisofthefemoralheadbetterthanthoseofmultipledrilling?A pilotstudy.JointBoneSpine2012;79:67–72.

[22]IwasakiK,YamamotoT,MotomuraG,etal.Commonsiteofsubchondralinsuf- ﬁciencyfracturesofthefemoralheadbasedonthree-dimensionalmagnetic resonanceimaging.SkeletalRadiol2016;45:105–13.