PBWO

Aoc

*

†

‡

A

0d

ET and SPECT in Osteomyelitis and Prostheticone and Joint Infections: A Systematic Reviewouter van der Bruggen, MD,* Chantal P. Bleeker-Rovers, MD, PhD,†,‡

tto C. Boerman, PhD,*,‡ Martin Gotthardt, MD, PhD,* and Wim J.G. Oyen, MD, PhD*,‡

Objective To review the literature on diagnostic accuracy and clinical value of single-photonemission computed tomography (SPECT) and positron emission tomography (PET) forimaging of bone and joint infections.Methods The PubMed/MEDLINE and Embase (OvidSP) literature databases were systemat-ically searched for publications on SPECT and PET on osteomyelitis and prosthetic boneand joint infections using specific guidelines with MeSH-terms, truncations, and comple-tion using cross-references.Results In 44 original articles (15 for SPECT and 29 for 18F-fluorodeoxyglucose [FDG]-PET)on osteomyelitis and prosthetic bone and joint infection, 1634 patients were included (580patients SPECT, 1054 patients FDG-PET). Level of evidence (Oxford criteria) was 2-3b. ForSPECT, the highest diagnostic accuracy of 95% for diagnosis of bone and joint infectionsis achieved with combined 111In-WBC and 99mTc-sulfur colloid. Acceptable diagnosticaccuracy was also obtained with 99mTc-WBC or 111In-WBC combined with 99mTc-methylenediphosphonate (99mTc-MDP). FDG-PET is useful for diagnosis of osteomyelitis with asensitivity and specificity generally over 95%. In patients with orthopedic implant infec-tions, sensitivity varies widely from 28% to 91% and specificity from 9% to 97%. Thisvariation in FDG-PET performance in orthopedic implant infections depends largely on the(use of different) criteria to diagnose infection. Determination of the best criteria is still amatter of debate.Conclusions SPECT/computed tomography (CT) with 111In-WBC combined with 99mTc-MDP or99mTc-sulfur colloid seems to be the best imaging technique for diagnosis of bone and jointinfections. FDG-PET is also useful for diagnosis of osteomyelitis with improved spatialresolution over SPECT imaging, allowing more accurate localization. Localization can befurther improved by adding CT. Diagnosis of orthopedic implant infections with FDG-PETdepends strongly on the localization of the implant and the criteria used to diagnoseinfection. Confirmation of well defined criteria to diagnose infection on FDG-PET in patientswith metallic implants is thus of paramount importance for optimal diagnosis.Semin Nucl Med 40:3-15 © 2010 Elsevier Inc. All rights reserved.

lmsbfspa

iita

cute and chronic osteomyelitis, spondylodiscitis, and or-thopedic implant infections are a heterogeneous group

f infectious disease entities. A common denominator is theomplex diagnostic and therapeutic challenge that is regu-

Department of Nuclear Medicine, Radboud University Nijmegen MedicalCentre, Nijmegen, The Netherlands.

Department of Internal Medicine, Radboud University Nijmegen MedicalCentre, Nijmegen, The Netherlands.

Nijmegen Institute for Infection, Inflammation and Immunity (N4i), Rad-boud University Nijmegen Medical Centre, Nijmegen, The Netherlands.

ddress reprint requests to Wouter van der Bruggen, MD, Department ofNuclear Medicine (internal postal code 444), Radboud University Ni-jmegen Medical Centre, PO Box 9101, 6500 HB, Nijmegen, The Neth-

merlands. E-mail: W.vanderBruggen@nucmed.umcn.nl

001-2998/10/$-see front matter © 2010 Elsevier Inc. All rights reserved.oi:10.1053/j.semnuclmed.2009.08.005

arly encountered by clinicians, radiologists, and nuclearedicine physicians. Although these disease entities share

ome common features, they present different problems inoth diagnostic procedures as in therapy. Therefore, the dif-erent infectious diseases and the respective performance ofingle-photon emission computed tomography (SPECT) andositron emission tomography (PET) will be discussed sep-rately.

The patient’s history, subjective symptoms, and biochem-cal and physical findings are often inconclusive, particularlyn the early stages. First abnormalities because of osteomyeli-is may not become visible on a plain x-ray until 10-21 daysfter onset of osteomyelitis, as 30%-50% loss of bone density

ust occur before radiographs can detect the disease.1 De-

3

totddftpraotc

o6

fltdcuRivioan

sSwptgpwwiaopic

ORiMuHtd

9

9

m

upsmttml

6

6

flccslbdorwiow

RSptlorcpfpooWibgb9

psas9

RMRoanp

4 W. van der Bruggen et al

ailed anatomical imaging with computed tomography (CT)r magnetic resonance imaging (MRI) is often unable to de-ect osteomyelitis at an early stage of the disease as well. Earlyetection is crucial in this disease for adequate treatment.1 Toetect osteomyelitis before anatomical changes are present,unctional imaging could have some advantages over ana-omical imaging. Diagnosis of infection in patients with jointrostheses or metallic implants is difficult for more than oneeason. On CT and MRI, metallic implants cause troublesomertifacts. On functional imaging, however, the mere presencef metallic implants can result in increased focal accumula-ion of the radiopharmaceutical uptake, not necessarily indi-ating infection or clinically relevant inflammation.

A variety of nuclear medicine techniques have been devel-ped for evaluation of infection of the locomotor system.7Ga-citrate has been used for imaging of infection and in-ammation ever because of its discovery in the early seven-ies.2 Multiphase skeletal scintigraphy with 99mTc-methyleneiphosphonate (99mTc-MDP) and radiolabeled white bloodells (WBC) have become the most widespread clinicallysed agents for the imaging of bone and joint infections.adiolabeled leukocytes have been the “gold standard” imag-

ng technique for infection for a long time. Nowadays, aariety of new radiopharmaceuticals is also used in infectionmaging. A wide variety of other radiolabeled probes, basedn antibodies, cytokines, and other receptor-binding ligands,re under (preclinical) investigation. Each tracer and tech-ique has its own inherent strongholds and shortcomings.After years in which planar scintigraphy used to be the

tandard, the need for improved localization was met byPECT. This allows more detailed 3D localization, comparedith planar imaging, which can provide crucial information,articularly in patients with osteomyelitis, but also in pa-ients suspected of infected joint prostheses. Recently inte-rated SPECT/CT was developed. Furthermore, the utility ofositron emitting radiopharmaceuticals for PET and PET/CTas explored, allowing acquisition of high-resolution imagesith good anatomic localization. Therefore, these 3D imag-

ng techniques (with or without CT for accurate localization)re more widely used. For PET, nearly all interest has focusedn 18F-fluorodeoxyglucose (18F-FDG). This review aims torovide a concise, structured overview of SPECT and PET

maging in patients suspected of osteomyelitis, spondylodis-itis, and orthopedic implant infections.

verview ofadiopharmaceuticals Used

n Bone and Joint Infectionsany radiopharmaceuticals have been studied and have been

sed for imaging of bone and joint infections in patients.ere, a concise overview of the performance of most impor-

ant tracers for SPECT and PET imaging of bone and jointisease is described.

9mTc-Bisphosphonates9mTc-labeled bisphosphonates (such as MDP and hydroxy-

ethylene diphosphonate [HDP]) are the most commonly t

sed radiopharmaceuticals to image osteoblastic activity. Inatients with osteomyelitis and infected orthopedic prosthe-es, increased osteoblastic activity occurs. Detection of osteo-yelitis with 99mTc-MDP is highly sensitive (�90%). When

he bone has not been violated by other pathologic condi-ions, specificity is high as well (around 90%). In post-trau-atic patients and after surgery, specificity is dramatically

ower (circa 35%).1

7Ga-Citrate7Ga-citrate has been used for imaging of infection and in-ammation ever since the early seventies.2 Although 67Ga-itrate scintigraphy has a high sensitivity for both acute andhronic infection and noninfectious inflammation, there areeveral shortcomings: the need for delayed imaging (�48 h),ow specificity, limited spatial resolution, physiologicalowel excretion, and the high radiation dose of the proce-ure. The physiological bowel uptake of 67Ga-citrate can bevercome by delayed imaging.3 67Ga-citrate has been usedoutinely, mainly to diagnose or to exclude osteomyelitis,2

ith reasonable sensitivity (73%) and relatively low specific-ty (61%).4 67Ga-citrate is now less frequently used, becausef these drawbacks and because compounds and techniquesith more favorable characteristics have been developed.

adiolabeled Leukocytescintigraphy with 111In-labeled leukocytes has an excellenterformance in diagnostic accuracy; a sensitivity of morehan 95% has been reported.5 After successful labeling ofeukocytes with 99mTc using hexamethylpropylene aminexime, 99mTc-labeled white blood cells (99mTc-WBC) haveeplaced 111In-labeled leukocytes for most indications, be-ause of more favorable imaging characteristics.1 Scintigra-hy with 99mTc-WBC shows high sensitivity and specificity

or imaging of infection. Uptake in acute osteomyelitis (sup-osedly due to the enhanced influx of leukocytes in acutesteomyelitis) is generally higher, compared with chronicsteomyelitis. The main drawbacks of 111In-WBC and 99mTc-BC are their laborious preparation, requirement of special-

zed equipment, and the handling of potentially infectiouslood.2 Therefore, a radiopharmaceutical with at least asood clinical performance as radiolabeled leukocytes that cane prepared off-the-shelf is warranted.1 Unlike 111In-WBC,

9mTc-WBC is reliably unstable, and 99mTc-hexamethylpro-ylene amine oxime can elute from the WBC, and can sub-equently be excreted through the kidneys, the gallbladder,nd the intestine.6 Low sensitivity for osteomyelitis of thepine has been reported in studies using 111In-WBC and9mTc-WBC, resulting in nonspecific “cold spots.”7

adiolabeled Specific Antigranulocyteonoclonal Antibodies and Nonspecific IgG

adiolabeled antigranulocyte antibodies have been devel-ped for in vivo labeling of white blood cells. Antibodiesccumulate in infectious and inflammatory foci mainly byonspecific extravasations, because of the enhanced vascularermeability in combination with specific targeting of infil-

rated granulocytes. Intact antibodies (whole IgG) show rel-

aFtdtaetpit9

tf

tcsstdiHc

1

FgfdFmtniaifies

MSTettits

AOpe

icifErurcw

DTpmr

T(

B

OOPSSPP

*†‡

PET and SPECT: Systematic review 5

tive slow blood clearance, while antibody fragments [Fab’ or(ab’)2] and IgM antibodies clear much faster. Uptake of in-act IgG antibodies is higher in the liver compared with ra-iolabeled leukocytes and lower in the spleen. Overall sensi-ivity for infection detection with radiolabeled antibodies ispproximately 80%-90%. Peripheral bone infections are ad-quately visualized, but sensitivity decreases when the infec-ion is located closer to the spine.8 This is mainly a result ofhysiological uptake in normal bone marrow and limited

nfiltration of leukocytes, comparable to diagnostic difficul-ies for 111In-WBC and 99mTc-WBC to diagnose spondylitis.9mTc-anti-CD15 IgM and 99mTc-anti-NCA-95 IgG are amonghe most effective antibodies to diagnose bone and joint in-ections.9,10

Radiolabeled nonspecific human IgG accumulates in infec-ious and inflammatory foci by nonspecific extravasation fa-ilitated by enhanced vascular permeability.7,11 111In-IgG hashown excellent performance in the localization of musculo-keletal infection and inflammation.7 99mTc-HYNIC-lgG scin-igraphy is equally effective as111 ln-lgG scintigraphy, for theetection of infection and inflammation. The apparent phys-

cal and logistic advantages of 99mTc over111In make 99mTc-YNIC-IgG in this respect a more attractive radiopharma-

eutical for imaging infection and inflammation.12

8F-fluorodeoxyglucose (FDG)DG is transported into cells by glucose transporters (glut-1,lut-3) and phosphorylated by hexokinase inside the cell toorm fluorodeoxyglucose-6-phosphate. The phosphorylatedeoxyglucose cannot be further metabolized, and thereforeDG accumulates in activated lymphocytes, neutrophils, andacrophages with minimal decrease over time.13,14 FDG

herefore accumulates in sites of infection, although it is aonspecific tracer that also accumulates in regions of aseptic

nflammation, as well as in malignant lesions.15 These char-cteristics make FDG-PET suitable for imaging of variousnflammatory and infectious diseases, mostly nonosseous in-ections but also osteomyelitis.16,17 Several groups have stud-ed the role of FDG-PET in the diagnosis of chronic osteomy-litis. Acute osteomyelitis and spondylitis can also beuccessfully diagnosed with FDG-PET.18

ethodsearch Strategyhe PubMed/MEDLINE and EMBASE (through OvidSP) lit-rature databases were systematically searched for publica-ions from January 1980 up to December 2008 on conven-ional scintigraphy, SPECT, and PET in bone and jointnfections according to the guidelines provided in “Update ofhe FDG-PET search strategy” by Mijnhout et al19 The searchtrategy is summarized in Table 1.

rticle Selectionriginal articles in English were included. Case reports, re-orts from meetings, editorial comments, or letters to the

ditor were excluded. Articles not concerning human bone

nfections, preclinical animal experiments, and articles dis-ussing results of less than 5 patients with bone or jointnfections were excluded. The remaining 39 full text articlesrom PubMed/MEDLINE and the 38 full text articles fromMBASE (through OvidSP) were further evaluated. After cor-ection for duplicates from both databases, this resulted in 43nique original articles. In addition to these original articles,eview articles were included for cross-referencing. Identifi-ation of 18 cross-references resulted in 61 articles, whichere analyzed and included in this review.

ata Interpretation and Analysisest performance was derived from the various studies andut into perspective. The following criteria and outcomeeasures were regarded as the major topics of interest for this

eview:

● level of evidence of the study (Oxford Centre for evi-dence-based medicine levels of evidence)

● number of patients with bone or joint infection● radiopharmaceutical of choice● patient selection: acute osteomyelitis, chronic osteomy-

elitis, spondylodiscitis, and orthopedic implant infec-tion

● sensitivity, specificity, and accuracy of conventional ra-dionuclide imaging, SPECT, and PET for the differententities of bone and joint infections

● prognostic value of these techniques in patients with(prosthetic) bone and joint infections

● change in management of patients with bone and joint

able 1 Systematic Query in PubMed/MEDLINE and EMBASE*Through OvidSP)

Bone and JointInfections Imaging Techniques

one diseases,infectious (MeSH)steomyelitis (Mesh)steitis (Mesh)eriostitis (Mesh)pondylitis (Mesh)pondylodisc†

rosthesisrosthetic

PET (MeSH)PETPET/†

Petscan†

PET/CT†

PET-CT†

CT/PET†

CT-PET†

(Positron† AND emission,† ANDtomograph†)

Tomography, emission-computed,single-photon (MeSH)‡

SPECT†

SPECT/CT†

SPECT-CT†

CT/SPECT†

CT-SPECT†

In EMBASE terms were used with “include all subheadings”.Truncation.In EMBASE (OvidSP) the corresponding term was “Single Photon

Emission Computer Tomography”.

infections due to FDG-PET and SPECT

CSIoTtsHaoms

O9

Ipiahwwcr9

s(sal

RPrpnif4tW

hvPa(awtMn(bMp

gratabpnpiaw9

icAtwsa

APrwppi9

tnasifihr0SISpacrb

ttowtr1

fp

6 W. van der Bruggen et al

onventional,PECT, and SPECT/CT

maging in the Diagnosisf Bone and Joint Infections

he role of SPECT in bone and joint infections has been inves-igated using a series of radiopharmaceuticals. The qualifiedtudies for this review include 67Ga-citrate, 99mTc-MDP, 99mTc-DP, 99mTc-MDP, 111In-WBC and 99mTc-WBC, 99mTc-labeled

ntigranulocyte antibodies, and 99mTc-sulfur colloid. The levelf evidence of the studies (Oxford Centre for evidence-basededicine Levels of Evidence) ranges from 2 to 3 b. The included

tudies regarding SPECT imaging are summarized in Table 2.

steomyelitis9mTc-MDPn 11 patients, Palestro et al9 present excellent sensitivity forlanar bone scintigraphy with 99mTc-MDP, with a limited spec-

ficity (11 true positives in 11 patients with proven osteomyelitisnd 5 true negatives in 13 patients without osteomyelitis). Thisigh sensitivity and poor specificity for 99mTc-MDP in patientsith osteomyelitis was confirmed by studies by Hakim et al20

ith 34 cases of osteomyelitis and Horger et al with 9 relevantases.21 Therefore, more specific radiopharmaceuticals are war-anted in patients suspected of osteomyelitis. SPECT imaging of9mTc-MDP alone20 or 99mTc-HDP alone21 resulted in acceptableensitivity (78%-84%), but specificity remained inadequate33%-50%).20,21 In hybrid 99mTc-HDP-SPECT/CT, however,pecificity was appreciably higher, when compared with SPECTlone, ie, 86% compared with 50% due to better anatomicalocalization by adding CT.21

adiolabeled WBCalestro et al9 reported that planar scintigraphy with 111In-WBCesulted in good sensitivity and reasonable specificity (10 trueositives in 11 patients with proven osteomyelitis and 8 trueegatives in 13 patients without osteomyelitis), being accurate

n 18 out of 24 patients. Rini et al15 reported an insufficient resultor diagnosing osteomyelitis with 111In-labeled leukocytes, with

true positives in 7 patients with proven osteomyelitis and 2rue negatives in 6 patients without osteomyelitis, using 111In-

BC SPECT imaging.Theoretically, specificity of imaging with 111In-WBC is

ampered, because of uptake in normal bone marrow, whicharies between patients and is not specific for disease.29

alestro et al9 showed that combination of planar 111In-WBCnd 99mTc-MDP resulted in good sensitivity and specificity11 true positives in 11 patients with proven osteomyelitisnd 10 true negatives in 13 patients without osteomyelitis)ith an accuracy of 87%. Besette et al23 also investigated the

est parameters of combined 111In-WBC SPECT and 99mTc-DP in diagnosing 25 patients with proven postsurgery ster-

al osteomyelitis.23 In this study they found a good sensitivity84%) and excellent specificity (100%). In 26 patients, Sea-old et al24 showed that combined 111In-WBC and 99mTc-DP SPECT seems to be the best technique for follow-up of

ostoperative patients with osteomyelitis in the mastoid re- s

ion, when compared with CT or MRI. This combinationesulted in a sensitivity of 95% and a specificity of 93%.24 Instudy by Weber et al,25 including 20 patients suspected of

emporal and facial osteomyelitis, 30 combined 111In-WBCnd 99mTc-MDP scans were performed. Performance of com-ined 111In-WBC and 99mTc-MDP was adequate with 15 trueositives in 16 scans with proven osteomyelitis and 13 trueegatives in 14 scans without osteomyelitis. Palestro et al29

roposed the use of combined 111In-WBC and bone marrowmaging performed with 99mTc-sulfur colloid. 111In–WBCnd 99mTc-sulfur colloid accumulate in the bone marrow,here 111In-WBC also accumulate in sites of infection and

9mTc-sulfur colloid does not.29 Activity on the 111In-WBCmage without corresponding activity on the 99mTc-sulfurolloid image was therefore considered positive for infection.ny other pattern was considered a negative study for infec-

ion.29 This combined 111In-WBC and bone marrow imagingith 99mTc-sulfur colloid showed a 100% sensitivity and 98%

pecificity in 50 patients with suspected infected total-hiprthroplasty.30

ntibodiesalestro et al9 reported that scanning with 99mTc-labeled mu-ine anti-CD15 IgM antigranulocyte antibodies in patientsith suspected osteomyelitis showed 10 true positives in 11atients with proven osteomyelitis and 9 true negatives in 13atients without osteomyelitis, when imaging 2 hours after

njection. With an accuracy of 91%, this combination of9mTc-anti-CD15 IgM and 99mTc-MDP, seems to approachhe diagnostic accuracy of radiolabeled leukocytes. Unfortu-ately, this radiopharmaceutical is no longer commerciallyvailable. With 99mTc-anti-NCA-95 IgG, Horger et al10

howed in 27 patients that sensitivity for detection of relaps-ng posttraumatic osteomyelitis was excellent and identicalor SPECT and hybrid SPECT/CT (100%), whereas specific-ty improved from 78% with SPECT to 89% when usingybrid SPECT/CT. Analysis of interobserver agreement withegard to localization of infectious foci resulted in kappa �.68 for immunoscintigraphy alone and kappa � 1.0 forPECT/CT, demonstrating the high reliability of this method.10

n addition, they concluded that in 27.5% of the casesPECT/CT imaging led to a change in diagnosis when com-ared with SPECT, which also resulted in a change of man-gement in these patients.10 Wuest et al31 also showed ahange in clinical diagnosis in 7 out of 11 patients with pe-ipheral orthopedic disorders (4 patients with osteomyelitis)ecause of adding CT to SPECT.An interesting feature of nuclear medicine techniques is

he ability to monitor therapy response. Weber et al28 inves-igated the response after treatment of facial and temporalsteomyelitis. They found that 111In-WBC in combinationith 99mTc-MDP bone SPECT imaging revealed normaliza-

ion after successful treatment of osteomyelitis much moreapidly than 67Ga-citrate or CT scans, and concluded that11In-WBC/99mTc-MDP bone SPECT is most useful duringollow-up.28 However, Hakim et al20 reported that FDG-PETossibly performs even better than 99mTc-MDP-SPECT,

howing normalization 1 month after onset of therapy (albeit

i2

iadtota

SPs

oiostTo9

iddut

T

9

6

W

A

*

† ence-b

PET and SPECT: Systematic review 7

n only 2 cases), where the first normalization was seen aftermonths on SPECT.20

The additional role of SPECT/CT vs planar imaging wasnvestigated for 67Ga-citrate and 111In-WBC by Bar-Shalom etl3 in 32 patients with osteomyelitis. The main diagnosticifficulty for both 67Ga-citrate and 111In-WBC-SPECT washe inability to localize an infectious focus within a specificrgan. Contribution of SPECT/CT to diagnosis and localiza-ion was 39% for 67Ga-citrate and 47% for 111In-WBC inssessment of osteomyelitis.3

pondylitis and Spondylodiscitiserformance of scintigraphy with 111In-WBC for spondylitis and

able 2 Studies Regarding SPECT Imaging

AuthorNumber ofPatients* Pathology

9mTc-MDP and99mTc-HDP

Hakim et al(2006)20

42 (34) Suspected chronicosteomyelitis of themandible

Horger et al(2007)21

31 (9) Suspected osteomyelitiprosthesis)

7Ga-citrate Gratz et al(2002)4

16 (11) Suspected spondylitis (prosthesis)

Bar-Shalom et al(2006)3

82 (32) Suspected osteomyelitiprosthesis)

Palestro, et al(1990)30

92 (23) Suspected infected totaarthroplasty

Palestro et al(1991)22

71 (28) Suspected vertebralosteomyelitis

Besette et al(1993)23

32 (32) Suspected sternalosteomyelitis postsu

BC99mTc-WBC111In-WBC

Seabold et al(1995)24

26 (24) Suspected cranialosteomyelitis (noprosthesis)

Weber et al(1995)25

20 (20) Suspected temporal anfacial osteomyelitis (prosthesis)

Van Acker et al(2001)26

21 (6) Suspected infected totaknee arthroplasty

Vanquickenborneet al (2003)27

17 (17) Suspected infected hipprostheses

Filippi et al(2006)28

28 (28) Suspected osteomyelitiand metallic implants

Rini et al(2006)15

51 (47) Suspected osteomyeliti(17) prosthesis (21),various infections (5)

ntigranulocyteantibodies

Palestro et al(2002)9

24 (24) Suspected osteomyelitiand metallic implants

Horger et al(2003)10

27 (27) Suspected relapsingposttraumaticosteomyelitis

Number of patients with suspected osteomyelitis or suspected infeparentheses.

Level of evidence and design of the study (Oxford Centre for evid

pondylodiscitis is drastically lower compared with detection of s

steomyelitis in peripheral bone, as reported in several stud-es.5,22 On planar images acquired after injection of 111In-WBCr 99mTc-WBC, up to 50% of all patients with spondylodiscitishow photopenic lesions because of encapsulation of the infec-ion4 and hence relatively hampered migration of leukocytes.hese photopenic lesions are not specific for infection.32 An-ther disadvantage is physiological uptake of 111In-WBC and9mTc-WBC in normal bone marrow.4 Both mechanisms resultn low sensitivity for 111In-WBC and 99mTc-WBC for infectionetection in the spine. Palestro et al22 showed 15 patients withecreased focal activity of 111In-WBC, 11 patients with increasedptake, and 2 cases with normal uptake in 28 confirmed pa-ients with spondylitis, exemplifying the diagnostic difficulty of

Radiopharmaceutical Gold StandardStudy Design andLevel of Evidence†

99mTc-MDP Histology, radiographs,clinical, andlaboratoryparameters

Prospective cohortdesign, 2b

99mTc-DPD (comparisonwith hybrid CT)

Surgery, biopsy, orfollow-up, includingmicrobiology andradiology

Prospective cohortdesign, 2b

67Ga-citrate Culture, surgery, andhistopathology

Prospective cohortdesign, 2b

67Ga-citrate or 111In-WBC SPECT, bothwith CT

Culture, surgery,clinical follow-up (24mo), and correlativeimaging

Retrospective casecontrol design, 3b

111In-WBC/99mTc-sulfurcolloid

Surgery Retrospective casecontrol design, 3b

111In-WBC Culture and biopsy Retrospective casecontrol design, 3b

Combined 111In-WBCwith 99mTc-MDP(compared with CT)

Surgically and“clinically proven”osteomyelitis

Retrospective casecontrol design, 3b

111In-WBC and 99mTc-MDP

Culture and clinicalfollow-up

Retrospective casecontrol design, 3b

111In-WBC/99mTc-MDP Biopsy/culture resultsin 18 patients andby endoscopic andclinical evaluation

Retrospective casecontrol design, 3b

99mTc-HMPAO-labeledleukocytes

Operative findings,culture, and clinicaloutcome

Prospective cohortdesign, 2b

111In-WBC Culture or clinicalfollow-up

Prospective cohortdesign, 2

99mTc-HMPAOleukocytes (hybridwith CT)

Surgery or cultures Prospective cohortdesign, 2b

111In-WBC Histopathology,microbiology,surgery, or clinicalfollow-up

Prospective cohortdesign, 2b

99mTc-labeled murineIgM monoclonalantigranulocyteantibody and111In-WBC

Pathology and clinicaloutcome

Prospective cohortdesign, 2b

99mTc-labeledantigranulocyteantibodies (with CT)

Surgery, biopsy,cultures, orfollow-up >6 mo

Prospective cohortdesign, 2b

osthetic bone and joint disease, with number of proven infection in

ased medicine Levels of Evidence).

s (no

no

s (no

l hip

rgery

dno

l

s (15)(13)

s,and

s (12)(12)

cted pr

pondylitis.

ISapiwrdwfita

WrfibFsp

sfdtwtap9

ffimiitFap1

s9

1

i1

ri1

eiTaiCd

FIBIatmssb

OTlFtcpssistw

tutmashiicte

aafia

ct

rntcibeij

8 W. van der Bruggen et al

nfection of Metallic Implantscintigraphy with 111In-WBC was able to reliably differenti-te between infected and sterile orthopedic metallic im-lants.18 Although the number of patients with suspected

nfected metallic implants scanned with 111In-WBC SPECTas limited (17 patients) in the study of Rini et al,15 they

eported 3 true positives in 3 patients with proven orthope-ic implant infection and 14 true negatives in 14 patientsithout orthopedic implant infection. Palestro et al30 con-rmed good performance of 111In-WBC in 92 patients withotal hip arthroplasties (23 infected), with 87% sensitivitynd 94% specificity.

Love et al33 concluded that planar imaging with 111In-BC/99mTc-sulfur colloid in case of prosthetic joint infection

esults in excellent diagnostic value: sensitivity 100%, speci-city 91%, and diagnostic accuracy 95%. This was significantlyetter than FDG-PET in the same group of patients (�0.001).DG-PET showed good sensitivity (95%), but inadequatepecificity (35%), resulting in a worse accuracy (61%) com-ared with 111In-WBC/99mTc-sulfur colloid.33

The study by Van Acker et al,26 including 19 patients withuspected orthopedic implant infection, reported sensitivityor 99mTc-WBC-SPECT of 100% with a specificity of 53% foriagnosis of infection.26 They managed to improve specificityo 93% when only lesions that were identified on a bone scanith 99mTc-MDP were taken into account. This did not affect

he sensitivity, which remained at 100%. The study by Filippind Schillaci28 including 15 patients with suspected ortho-edic implant infection investigated SPECT/CT imaging with

9mTc-WBC for diagnosis of metallic implant-associated in-ection. Sensitivity and specificity were 100%. This exempli-es 2 different methods (ie, combining 2 different radiophar-aceuticals or adding CT) being able to reach high specificity

n diagnosing infection in patients with suspected orthopedicmplant infection.26,28 Vanquickenborne et al27 showed thathe SPECT 99mTc-WBC provided better lesion contrast thanDG-PET in these patients, resulting in an 88% sensitivitynd a 100% specificity on 4 hours images, while 24 hourslanar images were of no additional value. The analysis of

8F-FDG-PET alone resulted in 88% sensitivity and 78%pecificity.27 Therefore, they concluded that scanning with9mTc-WBC leukocytes is sensitive and more specific than8F-FDG-PET for detecting infection in patients with metallicmplants.27 Test performance of combined 99mTc-MDP and11In-WBC SPECT in suspected bone or joint infection waseported by 2 studies.24,25 This resulted in a sensitivity rang-ng from 95% to 97% and specificity ranging from 93% to00% in 44 patients.27,28

Regardless of the radiopharmaceutical used or the diseasentity (osteomyelitis or orthopedic implant infection), specific-ty of scintigraphic imaging increases when CT is added.10,21,31

his was most strikingly demonstrated in the study of Horger etl21 where specificity of 99mTc-HDP improved from 50% to 86%n 31 patients with suspected osteomyelitis by the addition ofT. A second important improvement by the addition of (low

ose) CT is the enhanced interobserver agreement.21 (

DG-PET and FDG-PET/CTmaging in Diagnosingone and Joint Infections

n most studies investigating the performance of PET in bonend joint infections, FDG was used. The level of evidence ofhe included studies (Oxford Centre for evidence-basededicine levels of evidence) ranges from 2b to 3b. Table 3

ummarizes the studies with a gamma camera in coincidenceetting, the studies with full ring dedicated PET and com-ined FDG-PET/CT.

steomyelitishe detection of osteomyelitis with FDG-PET is generally excel-

ent, with sensitivities ranging from 94% to 100%.16,18,34-36,53

DG dual head gamma camera coincidence imaging in pa-ients with chronic osteomyelitis of the mandible, yieldedonsiderably lower sensitivity of 64%, recognizing only 16atients out of 25 with chronic osteomyelitis, well below theensitivity of 3-phase bone scintigraphy.20 Although there is atudy with coincidence camera that performs well in detect-ng osteomyelitis,34 using a gamma coincidence camera in-tead of a full-ring PET camera apparently reduces the sensi-ivity of the method from 95% to less than 70% for patientsith osteomyelitis of the mandible.Specificity of FDG-PET for osteomyelitis ranges from 87%

o 100%16,18,34-36,53 when using a full ring PET scanner. Whensing a coincidence camera specificity was lower (78%) inhe study, including patients suspected of osteomyelitis of theandible.20 Specificity can be adversely affected by lack of

dequate clinical information, as was demonstrated in thetudy of Zhuang et al16: 2 false positives in 16 patients couldave been avoided when critical information at the time of

nterpretation of the images had been available.16 The lack ofntrinsic high specificity of FDG-PET therefore demands aomplete relevant medical history for the interpreter. Sensi-ivity and specificity for FDG-PET in patients with osteomy-litis in the included studies is summarized in Figure 1.

Rini et al15 showed good sensitivity of FDG-labeled humanutologous leukocytes, where specificity, however, was dis-ppointing (6 true positives out of 7 patients having con-rmed osteomyelitis and 3 true negatives and 3 false positivesmong 6 patients without osteomyelitis).

Sahlmann et al44 investigated the relationship between thehange in the standarduptakevalue (SUV)over timeand theabilityo discriminate between malignant and benign osseous lesions.

The hypothesis was that in osteomyelitis, lesions exhibitelatively stable SUVs during dynamic imaging, where malig-ant bone lesions generally show an increase in SUV overime. The authors hypothesized that when choosing the rightut-off values for the SUV these characteristics may be used tomprove the specificity of FDG-PET for the differentiationetween malignant and benign bone lesions.44 In osteomy-litis, the SUVmax and SUVmean remained stable or decreasedn 16 out of 17 patients between 30 and 90 minute postin-ection. In these patients, a median decrease of 6% for SUVmax

range 1%-31%) and a median decrease of 8.5% for SUVmean

(SrmS

OBpFhntMcmnbaotpdtdnrfb

STPeispisapa1sspFFp(Wpsa

fd

tSip6

OICrcpbectmssswrlpwms

mchswtsdgsfabmktmo

wtaowpsPs

PET and SPECT: Systematic review 9

range 0%-24%) were observed. In 1 patient, SUVmax andUVmean increased over the time. The histology of this patientevealed multiple foreign body granulomas in addition to aononuclear infiltrate. In malignant lesions the SUVmax and

UVmean between 30 and 90 minute postinjection increased.44

steomyelitis in Patients with Diabetesasu et al49 described a perfect accuracy in 6 patients withroven osteomyelitis or neuropathic osteoarthropathy withDG-PET in diabetic patients. FDG-PET was predominantlyelpful in patients with a concomitant foot ulcer with a highegative predictive value in ruling out osteomyelitis in 5 pa-ients (pathology confirmed absence of osteomyelitis, where

RI resulted in 2 false-positives).49 Keidar et al51 found ex-ellent performance of FDG-PET/CT for diagnosing osteo-yelitis in 9 patients with diabetic feet, all true positives ando false negatives.51 In 7 patients with known diabetic ulcers,ut without clinical suspicion of osteomyelitis, Schwegler etl50 found very low sensitivity of FDG-PET for detection ofsteomyelitis (2 true positives in 7 patients with proven os-eomyelitis). MRI (which detected 6 out of 7 cases of unsus-ected osteomyelitis) seems to be superior to FDG-PET inetecting foot ulcer-associated osteomyelitis, and might behe preferred imaging modality in patients with nonhealingiabetic foot ulcers without signs of osteomyelitis.50 There iso satisfactory explanation for the discrepancy between theesults in these studies, thus the determining factor of successor FDG-PET in detecting osteomyelitis in patients with dia-etic foot ulcers remains unclear.

pondylitis and Spondylodiscitishree studies4,18,32 specifically investigated the value of FDG-ET in patients with spondylitis and spondylodiscitis. Gratzt al4 used a double-headed gamma camera operated in co-ncidence detection mode in 16 patients with suspectedpondylitis (with 12 confirmed cases) and reported gooderformance with 12 true positives out of 12 patients with

nfection and 14 true negatives out of 16 patients withoutpondylodiscitis. The performance of FDG-coincidence im-ging was superior to MRI, 67Ga-citrate, and 99mTc-MDP, es-ecially in patients with low-grade spondylitis.4 Kalicke etl18 showed that spondylitis was detected by FDG-PET in00% of the patients (n � 7). Schmitz et al32 reported aensitivity of 100% and specificity of 75% in 16 patients withuspected spondylodiscitis. Figure 2 shows an example of aatient with biopsy proven Brucella spondylodiscitis onDG-PET/CT. Guhlmann, et al. found good performance ofDG-PET for detection of infection in 28 patients with sus-ected central skeleton infections, with an accuracy of 96%5 patients had proven spondylodiscitis). The study by de

inter et al37 confirmed the performance of FDG-PET in 17atients with suspected central skeleton infections, with aensitivity of 100% and a specificity of 90%, resulting in 94%ccuracy for central skeleton infections.

Diagnostic accuracy of FDG-PET was compared with per-ormance of SPECT/CT with 67Ga-citrate and MRI in spon-

ylitis in 16 patients by Gratz et al4 They reported a diagnos- i

ic accuracy of 96% for FDG-PET vs 80% for 67Ga-citratePECT and 81% diagnostic accuracy for MRI.4 Especially innfected bone regions with diminished vascularity, FDG-PETroved to be superior to anatomical imaging with MRI and

7Ga-citrate SPECT or 99mTc-MDP SPECT.4

rthopedic Implant Infectionsn a group of 53 patients with hip prostheses, studied byhacko et al42 with 12 confirmed infections, FDG-PET cor-

ectly diagnosed 11 out of 12 infections. In 41 noninfectedases, FDG-PET was correct in all cases except one.42 Foratients with knee prostheses the same sensitivity was found,ut in only 18 cases out of 24 an infection could be correctlyxcluded, indicating lower specificity for FDG-PET to ex-lude infection in knee prostheses, compared with hip pros-heses.42 According to Mumme et al,46 diagnostic perfor-ance of FDG-PET is adequate for differentiation between

eptic and aseptic hip arthroplasty with 91% sensitivity, 92%pecificity, resulting in 91% accuracy in 50 patients. It isuperior to three-phase bone scintigraphy with 99mTc-HDP,ith a 78% sensitivity, a 70% specificity, and a 74% accu-

acy.46 Of critical importance is their conclusion that calcu-ation of SUV is unsuitable as a sole criterion for image inter-retation. FDG-PET results in high focal uptake in patientsith polyethylene and metal wear-induced chronic inflam-ation followed by periprosthetic osteolysis,46 lowering

pecificity for detection of infection of metallic implants.Stumpe et al45 showed that FDG-PET was significantlyore specific (P � .035), but less sensitive (P � .016) than

onventional radiography for the diagnosis of infected totalip prostheses. In a study population of 21 patients withuspected infected total hip implants, sensitivity of FDG-PETas below 30% with approximately 80% specificity, where

hree-phase bone scintigraphy with 99mTc-MDP resulted in aensitivity of approximately 50% with a 90% specificity foretecting infected hip implants.45 Two years later, the sameroup52 concluded that for painful knee arthroplasty, diffuseynovial and focal extrasynovial FDG uptake is commonlyound in patients with malrotation of the femoral componentnd is not related to pain location. The information providedy FDG-PET does not contribute to diagnosis and manage-ent of individual patients with persistent pain after total

nee replacement. The overall performance of FDG-PET inotal knee replacement was worse than in total hip replace-ent.52 Sensitivity and specificity for FDG-PET in patients with

rthopedic implants infections are summarized in Figure 3.In the study by Delank et al48 no false negatives were seen

ith FDG-PET in 5 patients with proven infected joint pros-heses. In patients with inflammation due to periprostheticseptic foreign-body reactions (ie, polyethylene abrasion), 9ut of 20 scans were false positive for infection.48 In patientsith not much abrasion, 2 out of 11 patients were false-ositive for infection.48 This illustrates 1 cause for the limitedpecificity in detection of infected joint prostheses with FDG-ET. Van Acker et al26 concluded that FDG-PET is highlyensitive for detection of metallic implant infection with

nsufficient specificity (73%-80%). However, when only

Table 3 Studies Regarding 18F-FDG imaging

Author Bone* (n) Pathology PET-Tracer† Gold StandardStudy Design and Level

of Evidence‡

18F-FDG with hybrid PETusing coincidencecamera

Gratz et al (2002)4 16 (12) Suspected spondylitis FDG Surgery, histopathology, andfine needle biopsy

Prospective cohort design, 2b

Meller et al (2002)34 30 (7) Suspected chronic osteomyelitis FDG Histology, culture Prospective cohort design, 2bLove et al (2004)33 49 (25) Suspected infection of knee and

hip prosthesesFDG Microbiology Retrospective case control

design, 3bHakim et al (2006)20 42 (32) Chronic osteomyelitis of the

mandibleFDG Histology and radiographs,

clinical and laboratoryparameters

Prospective cohort design, 2b

Rini et al (2006)15 47 (12) Skeletal infection, includingprosthesis

FDG-labeled humanautologous leukocytes

Histopathology, microbiology,surgery, or clinical follow-up

Prospective cohort design, 2b

18F-FDG-PET with fullring PET-scanner (noCT)

Guhlmann et al (1988) rad35 31 (18) Suspected chronic osteomyelitis FDG (no CT) Att. Corr. � Culture of surgical specimens,histopathology

Prospective cohort design, 2b

Guhlmann et al (1988) JNM36 51 (28) Suspected chronic osteomyelitis FDG (no CT) Att. Corr. � Culture of surgery,histopathology, and clinicalfollow-up >2 yr

Prospective cohort design, 2b

De Winter et al (2001)37 60 (25) Suspected chronicmusculoskeletal infection

FDG (no CT) Att. Corr? Histopathology, culture, clinicalfollow-up >6 mo

Prospective cohort design, 2b

Zhuang et al (2000)16 22 (6) Chronic osteomyelitis FDG (no CT) Att. Corr. �

and Att. Corr. �

Surgery, clinical follow-up >1 yr Retrospective case controldesign, 3b

Kalicke et al (2000)18 15 (15) Suspected acute or chronicosteomyelitis or inflammatoryspondylitis

FDG (no CT) Att. Corr. � Histopathology Prospective cohort design, 2b

Schmitz et al (2001)32 16 (12) Spondylodiscitis FDG (no CT) Att. Corr. � Histopathology Prospective cohort design, 2bVan Acker et al (2001)26 21 (6) Suspected total knee

arthroplastyFDG (no CT) Att. Corr. �

and Att. Corr. �

Operative findings and culture,clinical outcome

Prospective cohort design, 2b

Zhuang et al (2001)38 62 (11) Infected hip arthroplasty FDG (no CT) Att. Corr. �

and Att. Corr. �

Surgery, clinical follow-up >1 yr Prospective cohort design, 2b

Chacko et al (2002)39 32 (12) Suspected hip arthroplastyinfection

FDG (no CT) Att. Corr ? Microbiology, histopathology,surgical findings, and clinicalfollow-up >9 mo

Retrospective case controldesign, 3b

Manthey et al (2002)40 23 (4) Suspected infection orloosening of prosthesis

FDG (no CT) Att. Corr. � Operative findings or clinicaloutcome

Retrospective case controldesign, 3b

Zhuang et al (2002)41 27 (27) Infected hip arthroplasty FDG (no CT) Att. Corr. �

and Att. Corr. �

NA Partially prospective andretrospective case controldesign, 3b

Chacko et al (2003)42 56 (34)104 (28)

Chronic osteomyelitisOrthopedic implants

FDG (no CT) Att. Corr. � Surgical pathology, clinicalfollow-up >6 mo

Retrospective case controldesign, 3b

Schiesser et al (2003)43 29 (14) Suspected metallicimplant�associated infectionsafter trauma

FDG (no CT)§ Att. Corr.� and Att. Corr. �

Microbiology of surgicalspecimens and intraoperativefindings

Prospective cohort design, 2b

Vanquickenborne et al (2003)27 17 (8) Suspected infected hipprostheses

FDG (no CT) Culture or clinical follow-up forup to 6 mo

Prospective cohort design, 2b

Sahlmann et al (2004)44 17 (17) Histology proven osteomyelitis FDG (no CT) Att. Corr. � Histology Prospective cohort design, 2bStumpe et al (2004)45 21 (9) Suspected infection related

loosening of prosthesisFDG (no CT) Att. Corr. �

and Att. Corr. �

Microbiology of surgicalspecimens, joint aspiration �

clinical follow-up >6 mo

Prospective cohort design, 2b

Mumme et al (2005)46 50 (42) Suspected infection orloosening of prosthesis

FDG (no CT) Att. Corr. � Operative findings, microbiologyand histology, clinical followup >9 mo

Prospective cohort design, 2b

Pill et al (2006)47 92 (21) Suspected infected hipprostheses

FDG (no CT) Att. Corr. � Intraoperative histology andcultures

Prospective cohort design, 2b

10W

.vander

Bruggenet

al

lsCaapl

piPPti

CFIripmocbcrtcwoirirca

wdStwcmpsaeifts

tale

3C

ontin

ued

Del

ank

etal

(200

6)48

27(5

)S

eptic

join

tar

thro

plas

tyin

fect

ion

FDG

(no

CT)

Att

.C

orr.

�S

urge

ry,

hist

opat

holo

gy,

cultu

res

Pro

spec

tive

coho

rtde

sign

,2b

Bas

uet

al(2

007)

496

(6)

Pro

ven

oste

omye

litis

seco

ndar

yto

com

plic

ated

diab

etic

foot

FDG

(no

CT)

Att

.C

orr.

�

and

Att

.C

orr.

�

His

topa

thol

ogy

and

clin

ical

follo

w-u

pR

etro

spec

tive

case

cont

rol

desi

gn,

3bS

chw

egle

ret

al(2

008)

5020

(7)

Sus

pect

edos

teom

yelit

isin

diab

etic

foot

ulce

rsFD

G(n

oC

T)B

iops

yP

rosp

ectiv

eco

hort

desi

gn,

2b

FDG

-PET

/CT

Kei

dar

etal

(200

5)51

10(4

)S

uspe

cted

oste

omye

litis

indi

abet

icfe

etFD

G-P

ET/C

TH

isto

path

olog

y,cu

lture

ofsu

rgic

alsa

mpl

e,bi

opsy

,an

dcl

inic

alan

dim

agin

gfo

llow

-up

Pro

spec

tive

coho

rtde

sign

,2b

Stu

mpe

etal

(200

6)52

28(9

)S

uspe

cted

knee

arth

ropl

asty

infe

ctio

n18

FDG

-PET

/CT

10FD

G(n

oC

T)M

icro

biol

ogy

ofsu

rgic

alsp

ecim

ens,

clin

ical

follo

w-u

p>

6m

o

Pro

spec

tive

coho

rtde

sign

,2b

Har

tman

net

al(2

007)

5333

(18)

Sus

pect

edos

teom

yelit

isaf

ter

trau

ma

FDG

-PET

/CT

His

topa

thol

ogy

orba

cter

iolo

gic

cultu

reR

etro

spec

tive

case

cont

rol

desi

gn,

3b

mbe

rof

patie

nts

with

susp

ecte

dos

teom

yelit

isor

join

tpr

osth

esis

infe

ctio

n,w

ithnu

mbe

rof

prov

enin

fect

ion

inpa

rent

hese

s.ve

lof

evid

ence

ofth

est

udy

(Oxf

ord

Cen

tre

for

evid

ence

-bas

edm

edic

ine

leve

lsof

evid

ence

).t.

Cor

r.�

Att

enua

tion

corr

ectio

n.o

patie

nts

wer

esc

anne

dw

ithFD

G-P

ET/C

T.

PET and SPECT: Systematic review 11

esions also seen on bone scan were taken into account,pecificity increased to 93% without loss of sensitivity.ombined reading of these 2 studies could therefore bedvantageous in patients with suspected infection of hipnd knee prostheses. In Figure 4, an FDG-PET scan of aatient with extensive infection of a megaprosthesis of the

eft knee is shown.Rini et al18 reported 3 true positives in 3 patients with

roven orthopedic implant infection with 14 true negativesn 14 patients without orthopedic implant infection usingET with18 FDG-labeled human autologous leukocytes.18

erformance of this radiopharmaceutical needs further inves-igation to determine its value in patients with suspectednfected orthopedic implants.

riteria to Conclude Infection onDG-PET in Case of Metallic Implants

t is known that the mere presence of metallic implants canesult in increased focal uptake of FDG, not necessarily indicat-ng infection or clinically relevant inflammation.33,39,41,46,54 Thishenomenon of false-positive FDG-uptake in patients withetallic implants might be explained by high glucose metab-

lism in the joint capsule and around the prosthesis neck,aused by inflammation because of granulomatous foreignody reaction against polyethylene debris particles.55 Be-ause of this lack of specificity, in 4 studies33,38,39,41 the crite-ia to conclude infection on FDG-PET for hip and knee pros-heses were specifically investigated. Love et al33 investigatedriteria in both knee and hip prostheses. Not surprisingly,hen considering any periprosthetic FDG activity, regardlessf location or intensity as positive for infection, this resultedn a sensitivity of 100% with a poor specificity of 9% (accu-acy 47%). None of their constructed criteria (including sol-tary bone–prosthesis interface activity) led to satisfactoryesults, and they concluded that FDG-PET cannot replaceonventional radiolabeled leukocyte/marrow imaging for di-gnosing infection of failed prosthetic joints.33

Chacko and coworkers44 confirmed that specificity is poorhen using increased FDG-uptake as the sole criterion foriagnosing infection for hip prostheses, even when using aUVmax threshold of 7. Therefore, they evaluated 2 criteria forhe assessment of hip prostheses. Their first criterion inhich only uptake at the bone—prosthesis interface was

onsidered positive for infection, resulted in good perfor-ance with sensitivity of 92% and specificity of 97% in 41rostheses, of which 12 were proven to be infected. Theirecond criterion defined that periprosthetic infection was di-gnosed if there was any increased FDG uptake (mild, mod-rate, or severe) adjacent to the prosthesis (regardless whethert was localized at the bone or prosthesis interface, the tip of theemoral component or in the soft tissues surrounding the pros-hesis). This second constructed criterion lead to excellent sen-itivity (100%), but poor specificity (45%).39

Zhuang et al38 also formulated criteria for hip arthroplas-ies, in which FDG uptake around the femoral head and necknd around the tip of the prostheses were considered as non-

specific. They hypothesized that the interface between boneTab

18F-

*Nu

†Le

‡At

§Tw

atpfFiopjfla

satpw

ftgd

CiDatctctar

steom

12 W. van der Bruggen et al

nd prosthesis does not display high FDG-uptake in asymp-omatic patients or in those with aseptic loosening and FDG-ositivity in that area was therefore highly suggestive of in-ection.41 Furthermore, Zhuang et al56 demonstrated thatDG uptake normalizes within 3 months unless the process

s complicated by infection or malignancy. This phenomenonf false-positive FDG-uptake in patients with metallic im-lants might be explained by high glucose metabolism in the

oint capsule and around the prosthesis neck, caused by in-ammation due to granulomatous foreign body reactiongainst polyethylene debris particles.55

The best criterion to conclude infection in patients withuspected orthopedic implant infection on a FDG-PET scanltogether seems to be considering uptake at the bone—pros-hesis interface (with exclusion of the head and the tip) asositive for infection.38,41,55 It is still a matter of debatehether satisfactory diagnostic accuracy for diagnosing in-

Figure 1 Test characteristics of FDG-PET for o

Figure 2 Patient with biopsy proven Brucella spond

ection can be obtained with FDG-PET in patients with me-allic implants. Therefore, this last criterion and other sug-ested criteria need further validation, to assess its value inaily clinical practice.

urrent Status and Future Perspectivesn Pet Imaging of Bone and Joint Infectionsifferentiation between osteomyelitis and infection of thedjacent soft tissues may be better obtained with FDG-PEThan with CT or MRI, because of better lesion-to-backgroundontrast and because of prominent artifacts arising from me-allic implants in CT and MRI.57 Strobel et al58 observed thatombined PET/CT was significantly more accurate (86%) inhe differentiation of benign and malignant lesions than PETlone with an accuracy of 68% (P � .039). There might be aole for FDG-PET in differentiation of uneventful bone heal-

yelitis in terms of sensitivity and specificity.

ylodiscitis of the thoracic spine (Th2-Th3).

ita

DBmowt

f

bw9(f(CoaSpTi

PET and SPECT: Systematic review 13

ng from bone healing complicated by localized osteomyeli-is, as was demonstrated in preclinical research by Koort etl59 and later confirmed in patients by Zhuang et al.41

iscussion and Conclusionsecause of considerable heterogeneity in inclusion criteria,ethodology and outcome measures, meaningful calculation

f pooled sensitivity and specificity for detection of infectionith SPECT and PET in patients with osteomyelitis and pros-

hetic bone and joint infections is not possible.When relying on conventional scintigraphy, best results

or detecting osteomyelitis were observed when using com-

Figure 3 Test characteristics of FDG-PET for suspectespecificity.

Figure 4 Patient with an extensively infected mega prCitrobacter freundii infection with pathology results sh

medial side of the left knee.

ined 111In-WBC and 99mTc-MDP SPECT with SPECT/CTith sensitivity in the range of 84%-97% and specificity8%-100%.24,25 99mTc-labeled antigranulocyte antibodies

99mTc-anti-NCA-95 IgG) also showed excellent sensitivityor detection of relapsing posttraumatic osteomyelitis100%) with good specificity (89%) using hybrid SPECT/T, with 100% interobserver agreement.10 The detectionf (chronic) osteomyelitis with FDG-PET is feasible anddequate, with high sensitivity and specificity.16,19,34-36,53

everal studies concerning FDG-PET imaging in diabeticatients with osteomyelitis provide conflicting results.hese differences can be explained by small numbers of

ncluded patients and heterogeneity in inclusion criteria.

opedic implant infections in terms of sensitivity and

is of the knee and osteomyelitis (cultures confirmedactive infection and necrosis). Note the fistula on the

d orth

osthesowing

SF

aiaPiisSostfPdb9ts(wms

wFWa

R

1

1

1

1

1

1

1

1

1

1

2

2

2

2

2

2

2

2

2

2

3

3

3

3

14 W. van der Bruggen et al

pondylitis and spondylodiscitis are well diagnosed withDG-PET.4,18,32

Combined 111In-WBC and 99mTc-sulfur colloid SPECT/CTre adequate tools to diagnose (prosthetic) bone and jointnfections. With a sensitivity of 100%, specificity of 91% andccuracy of 95%, it seems to be significantly better than FDG-ET.33 99mTc-WBC is a very sensitive tool (�95%) for imag-

ng of infection in patients with metallic implants. Specificitys also high (93%-100%) in 99mTc-WBC SPECT/CT,26,28 but iteems dramatically lower (53%) in case of 99mTc-WBCPECT alone.26 The improvement of specificity by additionf CT to SPECT is of substantial importance, as has beenhown in multiple studies.10,15,21,26,29,31 For patients with me-allic implants, FDG-PET has a good sensitivity (91%-100%)or diagnosis of infection.26,27,38,39,43,46,47 Specificity of FDG-ET in patients with metallic implants, however, is stronglyependent on the used criteria to report infection based onoth localization and intensity of FDG-uptake, ranging from% to 97%.33,38,39,41 Specificity is generally higher in hip pros-heses, compared with knee prostheses.42 An adequate criterioneems to be to consider uptake at the bone—prosthesis interfacewith exclusion of the head and the tip) as positive for infection,ith 92% sensitivity and 97% specificity.41 This criterion re-ains to be validated in a prospective study design as another

tudy failed to reproduce this observation.33

For both SPECT and PET, specificity improves considerablyhen the scintigraphic images are fused with CT.10,15,21,26-28,31,58

or SPECT, this holds true for combining 111In-WBC or 99mTc-BC with 99mTc-MDP or 99mTc-sulfur colloid.26,30 Adding CT

lso enhances interobserver agreement.31

eferences1. Elgazzar AH, Abdel-Dayem HM, Clark JD, et al: Multimodality imaging

of osteomyelitis. Eur J Nucl Med 22:1043-1063, 19952. Boerman OC, Rennen H, Oyen WJ, et al: Radiopharmaceuticals to

image infection and inflammation. Semin Nucl Med 31:286-295, 20013. Bar-Shalom R, Yefremov N, Guralnik L, et al: SPECT/CT using 67Ga

and 111In-labeled leukocyte scintigraphy for diagnosis of infection.J Nucl Med 47:587-594, 2006

4. Gratz S, Dorner J, Fischer U, et al: 18F-FDG hybrid PET in patients withsuspected spondylitis. Eur J Nucl Med Mol Imaging 29:516-524, 2002

5. Datz FL: Indium-111-labeled leukocytes for the detection of infection:current status. Semin Nucl Med 24:92-109, 1994

6. Peters AM: The utility of [99mTc]HMPAO-leukocytes for imaging in-fection. Semin Nucl Med 24:110-127, 1994

7. Nijhof MW, Oyen WJ, van Kampen A, et al: Evaluation of infections ofthe locomotor system with indium-111-labeled human IgG scintigra-phy. J Nucl Med 38:1300-1305, 1997

8. Becker W, Bair J, Behr T, et al: Detection of soft-tissue infections andosteomyelitis using a technetium-99m-labeled anti-granulocyte mono-clonal antibody fragment. J Nucl Med 35:1436-1443, 1994

9. Palestro CJ, Kipper SL, Weiland FL, et al: Osteomyelitis: diagnosis with(99m)Tc-labeled antigranulocyte antibodies compared with diagnosiswith (111)In-labeled leukocytes—initial experience. Radiology 223:758-764, 2002

0. Horger M, Eschmann SM, Pfannenberg C, et al: The value of SPET/CT inchronic osteomyelitis. Eur J Nucl Med Mol Imaging 30:1665-1673, 2003

1. Oyen WJ, Claessens RA, van Horn JR, et al: Scintigraphic detection ofbone and joint infections with indium-111-labeled nonspecific poly-clonal human immunoglobulin G. J Nucl Med 31:403-412, 1990

2. Dams ET, Oyen WJ, Boerman OC, et al: Technetium-99m labeled tohuman immunoglobulin G through the nicotinyl hydrazine derivative:

a clinical study. J Nucl Med 39:119-124, 1998

3. Kubota R, Yamada S, Kubota K, et al: Intratumoral distribution offluorine-18-fluorodeoxyglucose in vivo: high accumulation in macro-phages and granulation tissues studied by microautoradiography.J Nucl Med 33:1972-1980, 1992

4. Kubota R, Kubota K, Yamada S, et al: Microautoradiographic study forthe differentiation of intratumoral macrophages, granulation tissuesand cancer cells by the dynamics of fluorine-18-fluorodeoxyglucoseuptake. J Nucl Med 35:104-112, 1994

5. Rini JN, Bhargava KK, Tronco GG, et al: PET with FDG-labeled leuko-cytes versus scintigraphy with 111In-oxine-labeled leukocytes for de-tection of infection. Radiology 238:978-987, 2006

6. Zhuang H, Duarte PS, Pourdehand M, et al: Exclusion of chronic os-teomyelitis with F-18 fluorodeoxyglucose positron emission tomo-graphic imaging. Clin Nucl Med 25:281-284, 2000

7. Bleeker-Rovers CP, de Kleijn EM, Corstens FH, et al: Clinical value ofFDG PET in patients with fever of unknown origin and patients sus-pected of focal infection or inflammation. Eur J Nucl Med Mol Imaging31:29-37, 2004

8. Kalicke T, Schmitz A, Risse JH, et al: Fluorine-18 fluorodeoxyglucosePET in infectious bone diseases: results of histologically confirmedcases. Eur J Nucl Med Mol Imaging 27:524-528, 2000

9. Mijnhout GS, Riphagen II, Hoekstra OS: Update of the FDG PET searchstrategy. Nucl Med Commun 25:1187-1189, 2004

0. Hakim SG, Bruecker CW, Jacobsen H, et al: The value of FDG-PET andbone scintigraphy with SPECT in the primary diagnosis and follow-upof patients with chronic osteomyelitis of the mandible. Int J Oral Max-illofac Surg 35:809-816, 2006

1. Horger M, Eschmann SM, Pfannenberg C, et al: Added value ofSPECT/CT in patients suspected of having bone infection: preliminaryresults. Arch Orthop Trauma Surg 127:211-221, 2007

2. Palestro CJ, Kim CK, Swyer AJ, et al: Radionuclide diagnosis of vertebralosteomyelitis: indium-111-leukocyte and technetium-99m-methylenediphosphonate bone scintigraphy. J Nucl Med 32:1861-1865, 1991

3. Bessette PR, Hanson MJ, Czarnecki DJ, et al: Evaluation of postopera-tive osteomyelitis of the sternum comparing CT and dual Tc-99m MDPbone and In-111 WBC SPECT. Clin Nucl Med 18:197-202, 1993

4. Seabold JE, Simonson TM, Weber PC, et al: Cranial osteomyelitis:diagnosis and follow-up with In-111 white blood cell and Tc-99mmethylene diphosphonate bone SPECT, CT, and MR imaging. Radiol-ogy 196:779-788, 1995

5. Weber PC, Seabold JE, Graham SM, et al: Evaluation of temporal andfacial osteomyelitis by simultaneous In–WBC/Tc-99m-MDP boneSPECT scintigraphy and computed tomography scan. OtolaryngolHead Neck Surg 113:36-41, 1995

6. Van Acker F, Nuyts J, Maes A, et al: FDG-PET, 99mtc-HMPAO whiteblood cell SPET and bone scintigraphy in the evaluation of painful totalknee arthroplasties. Eur J Nucl Med Mol Imaging 28:1496-1504, 2001

7. Vanquickenborne B, Maes A, Nuyts J, et al: The value of (18)FDG-PETfor the detection of infected hip prosthesis. Eur J Nucl Med Mol Imag-ing 30:705-715, 2003

8. Filippi L, Schillaci O: Usefulness of hybrid SPECT/CT in 99mTc-HM-PAO-labeled leukocyte scintigraphy for bone and joint infections.J Nucl Med 47:1908-1913, 2006

9. Palestro CJ, Love C, Tronco GG, et al: Combined labeled leukocyte andtechnetium 99m sulfur colloid bone marrow imaging for diagnosingmusculoskeletal infection. Radiographics 26:859-870, 2006

0. Palestro CJ, Kim CK, Swyer AJ, et al: Total-hip arthroplasty: peripros-thetic indium-111-labeled leukocyte activity and complementary tech-netium-99m-sulfur colloid imaging in suspected infection. J Nucl Med31:1950-1955, 1990

1. Wuest W, Kuwert T, Grunewald M, et al: Skeletal SPECT/CT of theperipheral extremities -interdisciplinary approach in orthopaedic dis-orders-first clinical results. Cent Eur J Med 2:499-510, 2007

2. Schmitz A, Risse JH, Grunwald F, et al: Fluorine-18 fluorodeoxyglu-cose positron emission tomography findings in spondylodiscitis: pre-liminary results. Eur Spine J 10:534-539, 2001

3. Love C, Marwin SE, Tomas MB, et al: Diagnosing infection in the failed

joint replacement: a comparison of coincidence detection 18F-FDG

3

3

3

3

3

3

4

4

4

4

4

4

4

4

4

4

5

5

5

5

5

5

5

5

5

5

PET and SPECT: Systematic review 15

and 111In-labeled leukocyte/99mTc-sulfur colloid marrow imaging.J Nucl Med 45:1864-1871, 2004

4. Meller J, Koster G, Liersch T, et al: Chronic bacterial osteomyelitis:prospective comparison of (18)F-FDG imaging with a dual-head coin-cidence camera and (111)In-labelled autologous leukocyte scintigra-phy. Eur J Nucl Med Mol Imaging 29:53-60, 2002

5. Guhlmann A, Brecht-Krauss D, Suger G, et al: Chronic osteomyelitis:detection with FDG PET and correlation with histopathologic findings.Radiology 206:749-754, 1998

6. Guhlmann A, Brecht-Krauss D, Suger G, et al: Fluorine-18-FDG PETand technetium-99m antigranulocyte antibody scintigraphy in chronicosteomyelitis. J Nucl Med 39:2145-2152, 1998

7. de Winter F, van de Wiele C, Vogelaers D, et al: Fluorine-18 fluorode-oxyglucose-position emission tomography: a highly accurate imagingmodality for the diagnosis of chronic musculoskeletal infections. J BoneJoint Surg Am 83-A:651-660, 2001

8. Zhuang H, Duarte PS, Pourdehnad M, et al: The promising role of18F-FDG PET in detecting infected lower limb prosthesis implants.J Nucl Med 42:44-48, 2001

9. Chacko TK, Zhuang H, Stevenson K, et al: The importance of thelocation of fluorodeoxyglucose uptake in periprosthetic infection inpainful hip prostheses. Nucl Med Commun 23:851-855, 2002

0. Manthey N, Reinhard P, Moog F, et al: The use of [18 F]fluorodeoxy-glucose positron emission tomography to differentiate between syno-vitis, loosening and infection of hip and knee prostheses. Nucl MedCommun 23:645-653, 2002

1. Zhuang H, Chacko TK, Hickeson M, et al: Persistent non-specific FDGuptake on PET imaging following hip arthroplasty. Eur J Nucl Med MolImaging 29:1328-1333, 2002

2. Chacko TK, Zhuang H, Nakhoda KZ, et al: Applications of fluorode-oxyglucose positron emission tomography in the diagnosis of infection.Nucl Med Commun 24:615-624, 2003

3. Schiesser M, Stumpe KD, Trentz O, et al: Detection of metallic implant-associated infections with FDG PET in patients with trauma: correla-tion with microbiologic results. Radiology 226:391-398, 2003

4. Sahlmann CO, Siefker U, Lehmann K, et al: Dual time point 2-[18F]fluoro-2=-deoxyglucose positron emission tomography inchronic bacterial osteomyelitis. Nucl Med Commun 25:819-823, 2004

5. Stumpe KD, Notzli HP, Zanetti M, et al: FDG PET for differentiation ofinfection and aseptic loosening in total hip replacements: comparisonwith conventional radiography and three-phase bone scintigraphy. Ra-diology 231:333-341, 2004

6. Mumme T, Reinartz P, Alfer J, et al: Diagnostic values of positron

emission tomography versus triple-phase bone scan in hip arthroplastyloosening. Arch Orthop Trauma Surg 125:322-329, 2005

7. Pill SG, Parvizi J, Tang PH, et al: Comparison of fluorodeoxyglucosepositron emission tomography and (111)indium-white blood cell im-aging in the diagnosis of periprosthetic infection of the hip. J Arthro-plasty 21:91-97, 2006

8. Delank KS, Schmidt M, Michael JW, et al: The implications of 18F-FDGPET for the diagnosis of endoprosthetic loosening and infection in hipand knee arthroplasty: results from a prospective, blinded study. BMCMusculoskelet Disord 7:20, 2006

9. Basu S, Chryssikos T, Houseni M, et al: Potential role of FDG PET in thesetting of diabetic neuro-osteoarthropathy: can it differentiate uncom-plicated Charcot’s neuroarthropathy from osteomyelitis and soft-tissueinfection? Nucl Med Commun 28:465-472, 2007

0. Schwegler B, Stumpe KD, Weishaupt D, et al: Unsuspected osteomy-elitis is frequent in persistent diabetic foot ulcer and better diagnosedby MRI than by 18F-FDG PET or 99mTc-MOAB. J Intern Med 263:99-106, 2008

1. Keidar Z, Militianu D, Melamed E, et al: The diabetic foot: initial expe-rience with 18F-FDG PET/CT. J Nucl Med 46:444-449, 2005

2. Stumpe KD, Romero J, Ziegler O, et al: The value of FDG-PET inpatients with painful total knee arthroplasty. Eur J Nucl Med MolImaging 33:1218-1225, 2006

3. Hartmann A, Eid K, Dora C, et al: Diagnostic value of 18F-FDG PET/CTin trauma patients with suspected chronic osteomyelitis. Eur J NuclMed Mol Imaging 34:704-714, 2007

4. Love C, Pugliese PV, Afriyie MO, et al: Utility of F-18 FDG imaging fordiagnosing the infected joint replacement. Clin Positron Imaging 3:159, 2000

5. KisielinskiK,CremeriusU,ReinartzP, et al: Fluordeoxyglucosepositronemis-sion tomography detection of inflammatory reactions due to polyethylenewear in total hip arthroplasty. J Arthroplasty 18:528-532, 2003

6. Zhuang H, Sam JW, Chacko TK, et al: Rapid normalization of osseousFDG uptake following traumatic or surgical fractures. Eur J Nucl MedMol Imaging 30:1096-1103, 2003

7. Stumpe KD, Strobel K: 18F FDG-PET imaging in musculoskeletal in-fection. Q J Nucl Med Mol Imaging 50:131-142, 2006

8. Strobel K, Exner UE, Stumpe KDM, et al: The additional value of CTimages interpretation in the differential diagnosis of benign vs. malig-nant primary bone lesions with 18F-FDG-PET/CT. Eur J Nucl Med MolImaging 35:2000-2008, 2008

9. Koort JK, Makinen TJ, Knuuti J, et al: Comparative 18F-FDG PET ofexperimental Staphylococcus aureus osteomyelitis and normal bone

healing. J Nucl Med 45:1406-1411, 2004