Management of vertebral osteomyelitis over an eight-year period: The UDIPROVE (UDIne PROtocol on VErtebral osteomyelitis)

Management

of

vertebral

osteomyelitis

over

an

eight-year

period:

The

UDIPROVE

(UDIne

PROtocol

on

VErtebral

osteomyelitis)

Alessandro

Russo

,

Elena

Graziano

,

Alessia

Carnelutti

,

Massimo

Sponza

,

Barbara

Cadeo

,

Assunta

Sartor

,

Elda

Righi

,

Matteo

Bassetti

*

a_{InfectiousDiseasesClinic,DepartmentofMedicine,UniversityofUdineandAziendaSanitariaUniversitariaIntegratadiUdine,Udine,Italy} b

DivisionofVascularandInterventionalRadiology,UniversityofUdineandAziendaSanitariaUniversitariaIntegratadiUdine,Udine,Italy

c

MicrobiologyUnit,AziendaSanitariaUniversitariaIntegratadiUdine,Udine,Italy

d

InfectiousDiseases,DepartmentofDiagnosticsandPublicHealth,UniversityofVerona,Verona,Italy

e

InfectiousDiseasesClinic,DepartmentofHealthSciences,UniversityofGenoa,GenoaandHospitalPoliclinicoSanMartino-IRCCS,Genoa,Italy

ARTICLE INFO Articlehistory:

Received13August2019

Receivedinrevisedform7October2019 Accepted11October2019

CorrespondingEditor:EskildPetersen, Aar-hus,Denmark Keywords: Vertebralosteomyelitis Vertebralbiopsy C-reactiveprotein FDG-PET/CT MRI ABSTRACT

Objectives:Vertebral osteomyelitis(VO)is acompellingclinical entityforclinicians because ofits insidiousandindolentcourse,whichmakesdiagnosisdifficult.

Methods:AllpatientswithasuspecteddiagnosisofVOwereanalyzedoveran8-yearperiod(January 2009toJanuary2017).TheUDIPROVEprotocol(UDInePROtocolonVErtebralosteomyelitis)wasapplied inallcases.TheprimaryendpointwastheperformanceoftheUDIPROVEprotocoltoobtainthecausal bacteriaofinfection.

Results:Duringthestudyperiod,133episodesofconfirmedVOwereobserved.Theetiologyofinfection wasobtainedin73.6%ofcases:70.5%weregram-positive,16.3%weregram-negative,and13.2%were mycobacteria.18_{F-Fluorodeoxyglucosepositronemissiontomography/computedtomography(FDG-PET/}

CT)showedthatfortubercularVO,themedianstandarduptakevalue(SUV)washigherwhencompared withVOcausedbyotherbacteria.Clinicalcureattheendoftherapywasreportedin85.7%ofpatients. Previousantimicrobialtherapyandadelayofmorethan5daysinperformingbiopsywereassociated withanundiagnosedetiologyofVO.Targetedantibacterialtherapyandfollow-upwithFDG-PET/CTwere associatedwithclinicalcureattheendoftherapy,whiletheinvolvementofmorethantwovertebraeand inadequatedrainagewereassociatedwithfailure.

Conclusions:RigorousapplicationoftheUDIPROVEprotocolallowedthecausativepathogensofVOtobe obtained–atabouttwicetheratereportedintheliterature.TheuseofFDG-PET/CTforthefollow-upof infectionwasmorereliablewhencomparedtomagneticresonanceimaging(MRI).

©2019TheAuthors.PublishedbyElsevierLtdonbehalfofInternationalSocietyforInfectiousDiseases. ThisisanopenaccessarticleundertheCCBY-NC-NDlicense(

http://creativecommons.org/licenses/by-nc-nd/4.0/).

Introduction

Vertebralosteomyelitis(VO)isacompellingclinicalentityfor clinicians because of its insidious and indolent course, which makesdiagnosisdifficult.ThediagnosisofVOcanoftenbedelayed byseveralmonthsandthediseasemayinitiallybemisdiagnosed andmismanagedasadegenerativeprocess.Consequently,patients oftendevelopdestructivelesionsorneurologicalcomplications.In thissetting,anelevatedC-reactiveprotein(CRP)levelinpatients

withbackpain,although notspecific,hasasensitivitythat can rangefrom94%to100%(Zimmerli,2010;Guptaetal.,2014).

Overall,theetiologyofinfectioninVOisobtainedinlessthan 40% of cases, making the choiceof antimicrobial regimenand durationoftherapychallenging(ChewandKline,2001).Withthe exception of septic patients and patients with neurological compromise, empiricantimicrobial therapyshould bewithheld wheneverpossibleuntilamicrobiologicaldiagnosisisconfirmed. Animage-guidedaspirationbiopsyisrecommendedforpatients withsuspectedVOwhenamicrobiologicaldiagnosishasnotbeen establishedbybloodcultures (BC)orserologicaltests;however, therateofdiagnosiswithanimage-guidedbiopsy,asreportedin theliterature,isabout30_–40%(Berbarietal.,2015).

Noindicationsforfollow-upwithmagneticresonanceimaging (MRI) in patients with VO who have a favorable clinical and

*Corresponding authorat:InfectiousDiseases Clinic,Department ofHealth Sciences,UniversityofGenoa,GenoaandHospitalPoliclinicoSanMartino-IRCCS, Genoa,Italy.

E-mailaddress:Matteo.Bassetti@hsanmartino.it(M.Bassetti).

https://doi.org/10.1016/j.ijid.2019.10.010

1201-9712/©2019TheAuthors.PublishedbyElsevierLtdonbehalfofInternationalSocietyforInfectiousDiseases.ThisisanopenaccessarticleundertheCCBY-NC-ND license(http://creativecommons.org/licenses/by-nc-nd/4.0/).

ContentslistsavailableatScienceDirect

International

Journal

of

Infectious

Diseases

laboratoryresponsetoantimicrobialtherapyarerecommendedin theinternationalguidelines(Berbarietal.,2015),consideringthe limits of the method. Thus, the role of positron emission tomography(PET)hasbeenassessedoverrecentyears,considering itshighsensitivityfordetectingosteomyelitis.AnegativePETscan excludesthediagnosisofosteomyelitis,includingVO(Ohtorietal., 2010;Hungenbachetal.,2013).

Theaimofthisstudywastoinvestigatetherigorousapplication oftheUDIPROVEprotocol(UDInePROtocolonVErtebral osteomy-elitis) to obtain the causative pathogens of VO with vertebral biopsy, and the role of 18_{F-fluorodeoxyglucose PET/computed}

tomography (FDG-PET/CT) in the diagnosis and follow-up of infectionandcomparisonwithMRI.

Materialsandmethods Studydesign

Thiswasaprospectiveobservationalstudyconductedinalarge tertiarylevelhospitalinItalyoveran8-yearperiod(January2009 toJanuary2017).Duringthestudyperiod,allconsecutive adult patients(age18years)withasuspecteddiagnosisofVOwere analyzed. The UDIPROVE protocol was applied in all cases, as reportedinFigure1.Thestudywaspreapprovedbythelocalethics committeeandwasconductedinaccordancewiththeprinciplesof theDeclarationofHelsinki.

Datacollectionanddefinitions

Patient data were collected prospectively from the medical records, computerized hospital databases, and/or clinical charts usingastandardform.Theseincludeddemographiccharacteristics, clinicalandlaboratoryfindings,baselinecomorbidities,radiological findings, relapseof infection, microbiological data,the Charlson ComorbidityIndex,dateandtypeofsurgery(ifperformed),sourceof infection, persistent bacteremia, development of septic shock, durationofhospitalstay,durationofdefinitiveantibiotictherapy, clinicaltreatmentfailure,and30-daymortality.

Atthetimeofdiagnosis,thepatientsunderwent MRIand/or FDG-PET/CT(firstvisit).AllpatientswhounderwentFDG-PET/CT were on a low carbohydrate–fat allowed diet for 24h before theFDG-PET/CTwas performed;theyfastedforthe6hprior to

18_{F-FDGinjection,accordingtothelocalprotocol.}

The following definitions were established prior to data analysis:‘relapse’wasdefined asanewdiagnosisofVOcaused bythesameorganismafterclinicalandmicrobiologicalresolution ofapreviouslytreatedepisodeofVO;‘timetobiopsy’wasdefined asthenumberofdaysthatelapsedbetweenthediagnosisofVO andthevertebralbiopsy(whenperformed);‘persistentinfection’ wasdefinedinthecaseofapatientstillbeingtreatedforVOatthe 3-month follow-up, withoutresolution of thedescribed symp-toms;‘clinicaltreatmentfailure’wasdefinedasalackofresponse to the definitive antimicrobial regimen, as reflected by the presence of any of the following after 45 days of therapy: ongoingfever,leukocytosis,orotherclinicalsignsofinfectionthat couldnotbeattributedtocausesotherthanVO;‘30-daymortality’ wasdefinedasdeathfromanycausewithinthe30daysfollowing thediagnosisofVO.

Synchronousormetastaticfociofinfectionwereidentifiedby meansofclinicalandimagingexaminations.Whenpresent,fluid collections and abscesses were drained and thefluid cultured. Coagulase-negativestaphylococci(CoNS)andotherskin commen-sals were considered etiologically irrelevant unless they were isolatedfromtwoormoresetsofBCandtheircausativerolewas consistentwithclinical data.Adequatecontrol of thesourceof infectionwasdefinedastheremovalofanypreexisting contami-natedcentralvenouscatheter,aswellasthedrainageofvertebral abscessesorotherfluidcollections, withthesebeingperformed within 24h after the onset of infection. Patients who were diagnosed withVObasedonatleastone imagingstudyand/or blood or tissue culture results were treated according to the Infectious Diseases Society of America (IDSA) guidelines for vertebralosteomyelitis(Berbarietal.,2015).Threemonthsafter inclusion(secondvisit),allsurvivingpatientsunderwentMRIand/ orFDG-PET/CT.

Endpointsandstatisticalanalysis

TheprimaryendpointwastheperformanceoftheUDIPROVE protocol to obtain the causal bacteria of infection. Secondary endpointsweretheroleofFDG-PET/CTindiagnosisandfollow-up andtheanalysisofclinicalcureattheendoftherapy.

Between-groupdifferenceswereassessedwiththeChi-square testorFisher’sexacttest(forcategoricalvariables)andthe two-tailedt-testorMann–Whitneytest(forcontinuousvariables),as appropriate. Univariate and multivariate analyses were usedto

Figure1.TheUDIPROVEalgorithm.

determinetheeffects of differentvariables on an undiagnosed etiologyofVO,clinicaltreatmentfailure,andclinicalsuccessinthe studypopulation. Possibleconfounding factorsand interactions were weighted during analyses with a backward stepwise selectionandconsideringp 0.05forallvariables,todetermine theeffectsofallanamnestic,clinical,andtherapeuticvariableson theoutcomes.Allreportedp-valuesare two-tailed.Waldcon fi-denceintervals(CI)andoddratios(OR)werecomputedbasedon estimatedstandarderrors.AllanalysesweredoneusingIBMSPSS Statisticsversion20.0(IBMCorp.,Armonk,NY,USA).

Results

During thestudyperiod,133episodesofconfirmedVOwere observed.As reportedin Table 1, theetiology of infection was obtained in 98 cases (73.6%) (BC+vertebral biopsy/drainage) according to the UDIPROVE algorithm (Figure 1). The most frequentlyisolatedbacteriaweregram-positive(70.5%),followed bygram-negative(16.3%)and mycobacteria(13.2%).Amongthe gram-positivebacteria,Streptococcussppaccountedfor24.4%of cases, followed by methicillin-resistant Staphylococcus aureus (MRSA) (19.4%). Biopsy results with or without antimicrobial washoutarereportedinFigure2.

Themeanstandarduptakevalue(SUV),CRP,andwhiteblood cellcount(WBC)valuesatthetimeofdiagnosis(firstvisit)andat follow-up (second visit) are reported in Figure 3. Statistically

significantdifferenceswereobservedforSUV(p< 0.001)andCRP (p< 0.001)values.

Finally,theSUV values in subgroups ofpatients withVO at follow-up(secondvisit)arereportedinFigure4:patientswithor without a tubercular etiology, the presence or absence of endocarditis, Staphylococcus aureus etiology or others, and two or more than two vertebrae involved in infection. Statistically significantdifferencesinSUVvalueswereobservedfortubercular etiologyornot(p<0.001)atfollow-up.

Table2reportsthebaselinecharacteristicsandoutcomesofthe studypopulationalongwiththeunivariateanalysisofriskfactors for clinical success or failure. Vertebral biopsy/drainage was performed in 101 patients (75.9%), and 14-day antimicrobial treatmentdiscontinuationbeforevertebralbiopsywasobtainedin 98patients(73.7%).Concomitantendocarditiswasobservedin18 cases(13.5%).ThemeanSUVvaluewas7.23.3,anddiagnosisand follow-upwithFDG-PET/CTwasperformedin59patients(44.3%). AllpatientsunderwentMRIatthetimeofdiagnosisandat follow-up.Clinicalcureattheendoftherapywasreportedin114patients (85.7%),and30-daymortalitywas10.5%.

Univariateanalysisofriskfactorsforclinicalsuccessorfailure showedthattheinvolvementofmorethantwovertebrae(94.7% vs.24.5%,p<0.001),meanSUVvalue(7.63.8vs.73.6,p=0.03), and inadequate source control of infection (84.2% vs. 7.9%, p<0.001)weremorefrequentlyobservedinpatientswithfailure of therapy, compared with the clinical cure group. Conversely, targeted antibacterial therapy (78.9% vs. 42.1%, p<0.001) and follow-upwithFDG-PET/CT(50%vs.10.5%,p<0.001)weremore frequentlyreportedintheclinicalcuregroup.Nodifferencesin 30-daymortalitywereobserved.Follow-upwithFDG-PET/CTshowed thatadecreaseintheSUVof>30%fromthebaselinevaluewas associatedwithclinicalcureattheendoftherapy.Fortubercular VO, the mean SUV at the time of diagnosis was higher when comparedwiththoseofVOcausedbyotherbacteria(11.43.7vs. 72.3,p<0.001).

Logisticregressionanalysisshowedthatpreviousantimicrobial therapyandadelayofmorethan5daysinperformingbiopsywere associated with an undiagnosed etiology of VO, as reportedin

Table3A.Targetedantibacterialtherapyandfollow-upwith FDG-PET/CTwereassociatedwithclinicalcureattheendoftherapy, whiletheinvolvementofmorethantwovertebraeandinadequate drainage of infection were associated with failure of therapy (Table3B).

Table1

Etiologyofvertebralosteomyelitis.

Etiology N=98(%) Gram-positive 69(70.5) MRSA 19(19.4) MSSA 5(5.1) CoNS 13(13.3) Streptococcusspp 24(24.4) Enterococcusspp 4(4.1) Others 4(4.1) Gram-negative 16(16.3) Escherichiacoli 8(8.1) Klebsiellapneumoniae 4(4.1) Others 4(4.1) Mycobacteriumtuberculosis 13(13.3) MRSA,methicillin-resistant Staphylococcus aureus; MSSA, methicillin-sensitive Staphylococcusaureus;CoNS,coagulase-negativestaphylococci.

Discussion

The studydata showedthat therigorous applicationof the UDIPROVEprotocolallowedthecausativepathogensofVOtobe obtainedinalmost75%ofcases,ahigherratewhencomparedwith datareportedintheliterature(Berbarietal.,2015).Ofimportance, the useof FDG-PET/CT was more reliable for the follow-up of infection when compared to MRI. As a consequence, targeted antibacterialtherapywas independentlyassociatedwithclinical cureofVO.Finally,previousantimicrobialtherapyandadelayof

morethan5daysinperformingbiopsywereassociatedwithan undiagnosedetiologyofVO,whiletheinvolvementofmorethan two vertebrae and inadequate drainage of infection were associatedwithtreatmentfailure.

The UDIPROVE protocol was associated witha high rate of diagnosisinpatientswithVO.Intheliterature,thesensitivityof image-guidedbiopsyinevaluatedstudiesisabout40%(Sehnand Gilula, 2012). Bhavanet al. showed that theyield of causative pathogens increased with open biopsy (93%) compared with needleaspirationbiopsy(48%)(Bhavanetal.,2010).Ofnote,VOis

Figure3.MeanSUV,CRP,andWBCvaluesatthetimeofdiagnosis(firstvisit)andatfollow-up(secondvisit). SUV:standarduptakevalue;CRP:C-reactiveprotein;WBC:whitebloodcellcount.

Figure4.SUVvaluesinsubgroupsofpatientswithvertebralosteomyelitisatfollow-up(secondvisit). SUV:standarduptakevalue;VO:vertebralosteomyelitis;TB:tuberculosis.

aninfectionthatisverydifficulttodiagnoseandtreat,considering alsothehighrateofpatientsonantibiotictherapyatthetimeof diagnosis.ThelatterresultsinalowrateofpositivityforBCand vertebralbiopsytodeterminetheetiologyofinfection(Saeedetal., 2019). As reported in Figure 2, vertebral biopsy without antimicrobial discontinuation, according to the protocol, was associatedwithahighrateofnegativeresultsinthepresentstudy. No studieshavebeenperformedthathavehelpedguidethe clinicianmanagingpatientswithsignificantclinicalor radiograph-icevidenceofVOwhoalsohavesterilebloodculturesandaninitial non-diagnosticspinalaspirationbiopsyresult.Obtainingsystemic inflammatorymarkerlevelsanddiagnosisandfollow-upwithMRI may confirm the infection and help monitor the response to therapy,clarifythepresenceofabscessinneedofdrainage,and identify spinal instability that could benefit from surgical correction. In previous studies, patients with evidence of progressive epidural and/or paraspinal soft tissue infection on follow-upMRIhaveappearedtobeatgreaterriskof treatment failure (Kowalski et al., 2007). The frequency and utility of obtainingfollow-upinflammatorylaboratorymarkers(likeCRP) whilepatientsarereceivingantimicrobialtherapyforVOhavenot beenestablished.MRIhassomevalueinassessingtheresponseof softtissue infections,but boneand diskabnormalitiestypically appeartoworsen eveninsuccessfullytreatedpatients(Zarrouk etal.,2007;Carragee,1997).Inthepresentstudy,thecombination ofFDG-PET/CTplusCRPvaluesatthetimeofdiagnosisand follow-upseemstobeassociatedwithhigheraccuracyinthemonitoring of the response to therapy. FDG-PET/CT could be used in association with MRI for the detection of VO, but could also

contributetotheearlydeterminationoftheresponsetotherapy. Thestudydataareinlinewiththoseofpreviousattemptstoassess thetreatmentresponsethathavereliedonquantificationofthe SUVvaluebeforeandaftertreatmentinpatientswithVO(Nanni etal.,2012;Kimetal.,2009).

Ofinterest,themeanSUVwashigherfortubercularVOwhen comparedwithpyogenicspondylodiscitis.Delayeddiagnosisofa tubercularetiology mayleadtosevereconsequences, including permanent spinal damage, and although spinal biopsies often representthemostaccuratediagnosticmethod,low microbiolog-icalyieldscanbeassociatedwithtubercularVO.Inthissetting, imagingrepresentsakeydiagnostictooltoachieveearlydiagnosis ofthediseaseandtomonitortheresponsetoantibiotictreatment. Asobservedpreviously(Bassettiet al.,2017), theassociationof specificFDG-PET/CTfeatures,suchashighSUV,withtubercularVO has been reported but remains to be determined definitively (Bakheetetal., 1998;Trecarichietal.,2012).Then,theconfirmation ofhighSUVvaluescouldallowphysicianstoconsideratubercular etiologyforVOinamoretimelymanner(Heyselletal.,2013).

Inthisstudypopulation,failureoftherapyforVOwasobserved in19patients(14.3%).ThefailurerateofpatientstreatedforVOin mostclinical studieshasvaried between10% and 30%(Berbari etal.,2015;Ohtorietal.,2010;Hungenbachetal.,2013;Sehnand Gilula,2012;Bhavanetal.,2010;Saeedetal.,2019;Kowalskietal., 2007;Zarrouketal.,2007;Carragee,1997;Nannietal.,2012;Kim etal.,2009;Bassettietal.,2017;Bakheetetal.,1998;Trecarichi etal., 2012;Heyselletal.,2013;Rissing,1997).Thetherapeutic management of VO patients with treatment failure should be tailoredtothesuspectedreasonforfailure.Consultationwitha

Table2

Baselinecharacteristicsandoutcomesofthestudypopulation,andunivariateanalysisofriskfactorsforclinicalcureorfailureoftherapy.

Variables Totalpopulation Clinicalcure Failureoftherapy p-Value N=133(%) n=114(%) n=19(%)

Age(years),meanSD 65.713.9 66.314.1 65.613.4 0.94

Malesex 68(51.1) 58(50.8) 10(52.6) 1.0

CharlsonComorbidityIndex,meanSD 2.51.4 2.41.5 2.51.6 1.0 Previousspinalsurgery 20(15) 18(15.7) 2(10.5) 0.73 Previousantimicrobialtherapy(30days) 112(84.2) 95(83.3) 17(89.4) 0.74 14-dayantimicrobialtreatmentdiscontinuationbeforevertebralbiopsy 98(73.7) 85(74.5) 13(68.4) 0.58 >2vertebraeinvolved 46(34.5) 28(24.5) 18(94.7) <0.001

Endocarditis 18(13.5) 15(13.1) 3(15.7) 0.91

Otherlocalizationofinfection 41(30.8) 30(26.3) 11(57.8) 0.01 Biopsy/drainageperformed 101(75.9) 88(77.1) 13(68.4) 0.34 Bloodculturepositivity 34(25.5) 30(26.3) 4(21) 0.62 SUVvalue,meanSD 7.23.3 73.6 7.63.8 0.03 CRPconcentration(mg/l),meanSD 42.232.9 40.231.7 44.933.9 0.04 WBCcount(109

/l),meanSD 8.35.7 8.45.2 8.15 0.81 Inadequatesourcecontrolofinfection 25(18.8) 9(7.9) 16(84.2) <0.001 Timetobiopsy,meanSD 11.67.8 10.58.2 12.99.2 0.03 Targetedantibacterialtherapy 98(73.6) 90(78.9) 8(42.1) <0.001 Timetodefinitivetherapy,meanSD 10.98.9 11.29.2 10.28.7 0.81 Lengthofdefinitivetherapy,meanSD 41.926.3 43.227 40.124.9 0.07 Lengthofhospitalization,meanSD 3417.2 33.615.1 34.220.5 0.92 UseofFDG-PET/CT 59(44.3) 57(50) 2(10.5) <0.001 30-daymortality 14(10.5) 12(10.5) 2(10.5) 1.0 SD,standarddeviation;SUV,standarduptakevalue;CRP,C-reactiveprotein;WBC,whitebloodcell;FDG-PET/CT,18_{F-fluorodeoxyglucose}

positronemissiontomography/ computedtomography.

Table3

Variablesindependentlyassociatedwithanundiagnosedetiologyofvertebralosteomyelitis(A),clinicalcureandfailureoftherapy(B).

A B

Variables OR 95%CI p-Value Variables OR 95%CI p-Value Previousantimicrobialtherapy 2.3 1.98–3.12 <0.001 Targetedantibacterialtherapy 0.34 0.21–0.54 0.01 >5daysdelayinperformingbiopsy 4.2 3.12–7.23 <0.001 Follow-upwithFDG-PET/CT 0.45 0.31–0.78 0.02

>2vertebraeinvolved 2.87 1.89–4.2 <0.001 Inadequatedrainageofinfection 4.21 2.87–8.23 <0.001 OR,oddsratio;CI,confidenceinterval;FDG-PET/CT,18

surgeon and infectious disease physician experienced in the treatmentof spinal infections maybewarranted in those with suspected or proven treatment failure. The decision regarding whether a surgical intervention is warranted needs to be individualized and should incorporate similar principles as to whethertoperformsurgeryat thetimeof VOdiagnosisornot (Valanciusetal.,2013).

Someimportantlimitationsofthisstudyshouldbehighlighted. Firstofall,therewas nopre-interventionperiod,whichreduces thequalityoftheevidence,sincethisisadescriptiveanalysisof patientsmanagedaccording totheUDIPROVE protocol.Second, this was a single-center study with a short follow-up period (3months),reducingthegeneralizabilityandcomparabilityofthe results.Finally,FDG-PET/CTwasnotperformedforallofthestudy populationanditsroutineusecouldbelimitedduetothehigh costs.

Inconclusion,anetiologyofinfectionwasobtainedin73.6%of cases(combinationofBCplusvertebralbiopsy/drainage), accord-ingtotheUDIPROVEalgorithm.Moreover,thecombinationofthe protocolwiththeuseofFDG-PET/CTfordiagnosisandfollow-up wasassociatedwithahighrateofclinicalcure(85.7%)inthisvery difficult-to-treat infection. Finally, antimicrobial therapy at the timeof vertebralbiopsynegatively influencedthepossibilityof obtaininganetiologyofinfection.

Therigorousapplicationofantimicrobialwashoutin combina-tion with vertebral biopsy and the collection of BC greatly increasedtherateofdiagnosisofVOcausativepathogens.Finally, higherSUVvaluescouldbeassociatedwithatubercularetiology, andthecombinationofSUVplusCRPvaluescouldbehelpfulin monitoringtheresponsetotherapyinallpatients.

Authorcontributions

ARand MBdesignedthestudy:EG,AC, MS,BC,ERcollected data;AR,ER,MBwrotethemanuscript.

Funding None. Ethicalapproval

Thestudywaspreapprovedbythelocalethicscommittee. Conflictofinterest

None.

References

BakheetSM,PoweJ,EzzatA,RostomA.F-18-FDGuptakeintuberculosis.ClinNucl Med1998;23:739–42.

BassettiM,MerelliM,DiGregorioF,DellaSiegaP,ScremM,ScarparoC,etal.Higher fluorine-18 fluorodeoxyglucose positron emission tomography (FDG-PET) uptakeintuberculouscomparedtobacterialspondylodiscitis.SkeletalRadiol 2017;46(June):777–83.

BerbariEF,KanjSS,KowalskiTJ,DarouicheRO,WidmerAF,SchmittSK,etal.2015 InfectiousDiseasesSocietyofAmerica(IDSA)clinicalpracticeguidelinesforthe diagnosisandtreatmentofnativevertebralosteomyelitisinadults.ClinInfect Dis2015;61:e26–46.

Bhavan KP, Marschall J, Olsen MA, Fraser VJ, Wright NM, Warren DK. The epidemiologyofhematogenousvertebralosteomyelitis:acohortstudyina tertiarycarehospital.BMCInfectDis2010;10:158.

CarrageeE.Theclinicaluseofmagneticresonanceimaginginpyogenicvertebral osteomyelitis.Spine(PhilaPa1976)1997;22(7):780–5.

ChewFS,KlineMJ.DiagnosticyieldofCT-guidedpercutaneousaspirationprocedures insuspectedspontaneousinfectiousdiskitis.Radiology2001;218:211–4.

GuptaA,KowalskiTJ,OsmonDR,EnzlerM,SteckelbergJM,HuddlestonPM,etal. Long-termoutcomeofpyogenicvertebralosteomyelitis:acohortstudyof260 patients.OpenForumInfectDis2014;1:8.

Heysell SK,Thomas TA, Sifri CD, Rehm PK, Houpt ER.18-fluorodeoxyglucose positronemissiontomographyfortuberculosisdiagnosisandmanagement:a caseseries.BMCPulmMed2013;13:14.

HungenbachS,Delank KS, DietleinM, EyselP,Drzezga A,SchmidtMC. 18F-fluorodeoxyglucoseuptakepatterninpatientswithsuspectedspondylodiscitis. NuclMedCommun2013;34:1068–74.

KimS-JSJ,KimI-J,SuhKT,KimY-K,LeeJS.Predictionofresidualdiseaseofspine infectionusingF-18FDGPET/CT.Spine(Philadelphia,Pa1976)2009;34:2424–30.

KowalskiTJ,LaytonKF,BerbariEF,SteckelbergJM,HuddlestonPM,WaldJT,etal. Follow-upMRimaging inpatients withpyogenicspineinfections:lackof correlationwithclinicalfeatures.AJNRAmJNeuroradiol2007;28:693–9.

NanniC,BorianiL,SalvadoriC,ZampariniE,RoratoG,AmbrosiniV,etal.FDGPET/CT isusefulfortheinterimevaluationofresponsetotherapyinpatientsaffectedby haematogenous spondylodiscitis. Eur J Nucl Med Mol Imaging 2012;39 (10):1538–44.

OhtoriS, SuzukiM,Koshi T, Yamashita M,YamauchiK,Inoue G, et al. 18F-fluorodeoxyglucose-PETforpatientswithsuspectedspondylitisshowingModic change.Spine2010;35:E1599–603.

RissingJP.Antimicrobialtherapyforchronicosteomyelitisinadults:roleofthe quinolones.ClinInfectDis1997;25:1327–33.

SaeedK,EspositoS,AscioneT,BassettiM,BonnetE,CarneluttiA,etal.Hottopicson vertebral osteomyelitis from the international society of antimicrobial chemotherapy. Int J Antimicrob Agents 2019;(June), doi:http://dx.doi.org/ 10.1016/j.ijantimicag.2019.06.013pii:S0924-8579(19)30166-9.[Epubaheadof print].

SehnJK,GilulaLA.Percutaneousneedlebiopsyindiagnosisandidentificationof causativeorganismsincasesofsuspectedvertebralosteomyelitis.EurJRadiol 2012;81:940–6.

TrecarichiEM,DiMecoE,MazzottaV,FantoniM.Tuberculousspondylodiscitis: epidemiology, clinical features, treatment, and outcome. Eur Rev Med PharmacolSci2012;16Suppl2:58–72.

ValanciusK,HansenES,HoyK,HelmigP,NiedermannB,BungerC.Failuremodesin conservativeandsurgicalmanagementofinfectiousspondylodiscitis.EurSpine J2013;22:1837–44.

ZarroukV,FeydyA,SallèsF,DufourV,GuiguiP,RedondoA,etal.Imagingdoesnot predicttheclinicaloutcomeofbacterialvertebralosteomyelitis.Rheumatology (Oxford,England)2007;46(2):292–5.

Zimmerli W. Clinical practice. Vertebral osteomyelitis. N Engl J Med 2010;362:1022–9.