Detection of Osteomyelitis in theDiabetic Foot by ImagingTechniques: A Systematic Reviewand Meta-analysis ComparingMRI, White Blood CellScintigraphy, and FDG-PETDiabetes Care 2017;40:1111–1120 | https://doi.org/10.2337/dc17-0532

OBJECTIVE

Diagnosing bone infection in the diabetic foot is challenging and often requiresseveral diagnostic procedures, including advanced imaging. We compared the di-agnostic performances of MRI, radiolabeled white blood cell (WBC) scintigraphy(either with 99mTc-hexamethylpropyleneamineoxime [HMPAO] or 111In-oxine),and [18F]fluorodeoxyglucose positron emission tomography (18F-FDG–PET)/computed tomography.

RESEARCH DESIGN AND METHODS

WesearchedMedline and Embase as ofAugust 2016 for studies of diagnostic tests onpatients known or suspected to have diabetes and a foot infection. We performed asystematic review using criteria recommended by the Cochrane Review of a data-base that included prospective and retrospective diagnostic studies performed onpatients with diabetes in whom there was a clinical suspicion of osteomyelitis of thefoot. The preferred reference standard was bone biopsy and subsequent patholog-ical (or microbiological) examination.

RESULTS

Our review found 6,649 articles; 3,894 in Medline and 2,755 in Embase. A total of27 full articles and 2 posters was selected for inclusion in the analysis. The perfor-mance characteristics for the 18F-FDG–PET were: sensitivity, 89%; specificity, 92%;diagnostic odds ratio (DOR), 95; positive likelihood ratio (LR), 11; and negative LR,0.11. For WBC scan with 111In-oxine, the values were: sensitivity, 92%; specificity,75%; DOR, 34; positive LR, 3.6; and negative LR, 0.1. For WBC scan with 99mTc-HMPAO, the values were: sensitivity, 91%; specificity, 92%; DOR, 118; positive LR,12; and negative LR, 0.1. Finally, forMRI, the valueswere: sensitivity, 93%; specificity,75%; DOR, 37; positive LR, 3.66, and negative LR, 0.10.

CONCLUSIONS

The various modalities have similar sensitivity, but 18F-FDG–PET and 99mTc-HMPAO–labeled WBC scintigraphy offer the highest specificity. Larger prospective studieswith a direct comparison among the different imaging techniques are required.

1Department of Nuclear Medicine and MolecularImaging, University of Groningen, University Med-ical Center Groningen, Groningen, the Netherlands2Nuclear Medicine Unit, Faculty of Medicine andPsychology, Department of Medical-Surgical Sci-ences and of Translational Medicine, SapienzaUniversity of Rome, Rome, Italy3Department of Nuclear Medicine, University ofPisa, Pisa, Italy4Department of Orthopedics, University of Gro-ningen, University Medical Center Groningen,Groningen, the Netherlands5Division of Medical Sciences, Green TempletonCollege, University of Oxford, Oxford, U.K.6Department of Medicine, University of Wash-ington, Seattle, WA7Department of Health Technology and ServicesResearch,MIRA Institute for Biomedical Technol-ogy & Technical Medicine, University of Twente,Enschede, the Netherlands8Department of Biomedical Photonic Imaging,University of Twente, Enschede, the Netherlands

Corresponding author: Riemer H.J.A. Slart,r.h.j.a.slart@umcg.nl.

Received 15 March 2017 and accepted 3 May2017.

C.L. and M.T. contributed equally to this work.

© 2017 by the American Diabetes Association.Readers may use this article as long as the workis properly cited, the use is educational and notfor profit, and the work is not altered. More infor-mation is available at http://www.diabetesjournals.org/content/license.

Chiara Lauri,1,2 Menno Tamminga,1

Andor W.J.M. Glaudemans,1

Luis Eduardo Juarez Orozco,1

Paola A. Erba,3 Paul C. Jutte,4

Benjamin A. Lipsky,5,6

Maarten J. IJzerman,7 Alberto Signore,1,2

and Riemer H.J.A. Slart1,8

Diabetes Care Volume 40, August 2017 1111

META

-ANALYSIS

Most persons with longstanding diabetesdevelop peripheral neuropathy that, to-gether with peripheral vascular disease(and microvascular dysfunction), oftenleads to foot complications. Patientswith diabetes with these complicationshave an;25% lifetime risk of developinga foot complication (1,2), and these nowappear to be themost common diabetes-related reason for hospitalization. At pre-sentation,.50% of diabetic foot woundsare clinically infected (3).Most are initiallysoft-tissue diabetic foot infections (DFIs),but these often spread contiguously tounderlying bone, resulting in diabeticfoot osteomyelitis (DFO). DFO is nowthe most frequent cause of nontraumaticlower-extremity amputations that are as-sociated with a 5-year mortality of;50%(4). Therefore, prompt identification andoptimal treatment of DFO are importantto help avoid poor outcomes (5).Infection in the diabetic foot is defined

by the presence of classic signs of inflam-mationand sometimes so-called “second-ary” signs (3). Identifying DFO, whichoccurs in ;20% of mild infections and.50% of severe cases (3,6,7), can bemore difficult. This is related to the factthat: 1) DFO can occur in association withuninfected as well as infected ulcers; 2)radiographic changes in bone may benonspecific and delayed for a few weeksafter infection; and 3) patients with dia-betes are also at risk for developing neuro-osteoarthopathy of the foot (Charcot foot[8,9]). The approach to treatment de-pends onproper diagnosis, as DFOusuallyrequires antibiotic and surgical treat-ment, whereas Charcot disease requiresproper offloading, sometimes with latersurgical correction. Thus, diagnosingDFO requires a systematic approach thatincludes clinical, imaging, microbiological,and histopathological methods.The most widely accepted criterion

standard for diagnosing DFO is the pres-ence of characteristic findings on histo-pathological examination and growth onculture of an aseptically obtained speci-men bone (10). Bone biopsy, however, isan invasive procedure, and histologyand culture are relatively expensive andtime-consuming. Thus, it is important todetermine which of the other availablediagnostic tests for DFO might be appro-priate in selected patients. Clinical exam-inations, such as the probe-to-bone test(11), inflammatory markers (especiallythe erythrocyte sedimentation rate)

(12), and plain X-rays are nearly alwaysthe first steps in diagnosing DFO. Insome cases, however, they fail to providediagnostic results. In these situations,more advanced imaging techniques areoften needed.

Imaging offers a complementary andless invasive, although often expensive,approach to diagnosing DFO, with awide panel of modalities including: MRI,scintigraphy with 99mTc-hexamethyl-propyleneamineoxime (HMPAO) or111In-oxine–labeled white blood cells(99mTc-HMPAO–WBCs or 111In-oxine–WBCs) with single-photon emission com-puted tomography (SPECT/computedtomography [CT]), [18F]fluorodeoxy-glucose positron emission tomography (18F-FDG–PET/CT), or 99mTc-antigranulocyteantibody scintigraphy (13–18). Choosingthe most appropriate advanced imagingmodality must be based on not onlythe patient’s clinical presentation butalso the equipment and expertise avail-able at the treating center. Key factorsinclude any recent or ongoing antibiotictherapy, the presence of neuropathicdisease of the foot, the financial costs ofvarious tests, the waiting time beforeimaging can be performed, any possiblecontraindications to the tests, and patientpreference and likely adherence.

Of course, another key issue is which ofthe tests is most diagnostically useful. Todetermine the performance characteris-tics of the currently available advancedimaging tests for diagnosing DFO, weperformed a systematic review and meta-analysis of the literature. The results ofthis review should help clinicians and or-ganizations in preparing guidelines forthe multimodality approach required fordiagnosing DFO.

RESEARCH DESIGN AND METHODS

We conducted this systematic review inaccordance with methods outlined in theCochrane Handbook for Systematic Re-views of Diagnostic Test Accuracy (19)and have presented it following thePreferred Reporting Items for System-atic Reviews and Meta-analysis (PRISMA)guidelines (20).

Eligible StudiesWe only included studies that: providedoriginal data; were designed to provideinformation on diagnosis; were conductedin patients known or suspected to haveboth diabetes (type 1 or 2, regardless of

method of glycemic treatment) and a footinfection; and were published in English.We considered studies for inclusion witheither a prospective or retrospective de-sign and blinded or nonblinded. Becauseof the risk of introducing bias, we excludedcase-control studies, case reports, case se-ries, and animal studies.We also excluded:reviews; articles on topics not germane toour study question (e.g., Charcot neuro-osteoarthropathy,magnetic resonancean-giography for arterial disease, orthopedicimplants, inflammatory markers, thera-peutic studies, and studies on non–footinfections); articles focused on other in-dex tests or using other less specific ra-diopharmaceuticals (67Ga-citrate, labeleddiphosphonates, labeled IgG, labeledciprofloxacin, or labeled ubiquicidin);and articles that used currently outdatedmethodologies or protocols.

Literature Sources and SearchWe searched the databases of Medlineand Embase for studies published throughAugust 2016. We used a combination ofMedical Subject Headings terms and free-text words to define: our population ofinterest (persons with diabetes); the path-ologic process of interest (osteomyelitis orinfections of bone); and the specified im-aging techniques used (MRI, 111In-oxine–WBC SPECT/CT, 99mTc-HMPAO–WBCSPECT/CT, or 18F-FDG–PET/CT). No stud-ies with antigranulocyte antibodies couldbe included, because of the very limitedapplication in diabetic foot diagnostics.Table 1 shows the extended query andcomplete search terms.

Screening and Selection of LiteratureThree reviewers (C.L., M.T., and R.H.J.A.S.)independently screened all retrievedstudies obtained based on their titleand abstract. In a subsequent secondaryscreening, we evaluated the full text ofthe selected articles for eligibility. Wejointly discussed, and resolved by con-sensus, any discrepancy among the re-viewers that arose in study screening andselection. We also conducted a search ofthe references included in the retrievedarticles seeking any additional potentiallyrelevant articles.

Diagnostic Criterion StandardOur preferred reference standard for thediagnosis of osteomyelitis (against whichwe compared the diagnostic performanceof the target imaging techniques) wasevaluation of a specimen of affected

1112 Detection of Osteomyelitis in the Diabetic Foot Diabetes Care Volume 40, August 2017

bone (collected by surgical or percutane-ous biopsy) by histopathological reviewand/or culture. We only included studiesthat used this standard in our pooled es-timation of diagnostic performance met-rics (sensitivity and specificity). However,as it is common in clinical practice to usefollow-up (history and physical examina-tion, blood tests, and plain X-rays) to di-agnose osteomyelitis, we also includedstudies that used this approach for justthe calculation of positive and negativepredictive values.

Data Extraction and QualityAssessmentTwo reviewers (M.T. and L.E.J.O.) inde-pendently extracteddata fromeach studyand evaluated them for quality using theQUADAS-2 method (19). They jointly dis-cussed and resolved any discrepanciesby consensus. The domains of interestin standardized data extraction were:population characteristics; imagingmethods; reference standard for diag-nosing osteomyelitis; descriptive andquantitative results (we generated atwo-by-two contingency table for eachimaging modality); and frequencies for

final diagnoses (using RevMan v.5 forextraction).

For quality assessment, QUADAS-2considers four main domains: risk ofbias in patient selection (low, high, orunclear); index test; reference test; and,study flow and timing. We assessed therisk of bias and applicability concerns perimaging modality per study, then overallas follows: 1) low, if therewas a low risk ofbias in all key domains; 2) unclear, if wecould not assess the risk of bias in one ormore key domains; and 3) high, if the riskof bias was high for one or more keydomains (20).

Statistical AnalysisWe used the STATA program (version12.1; StataCorp), SPSS v.21 for Windows(IBM), and RevMan 5 for statistical anal-yses, setting statistical significance at P,0.05. Using a hierarchical random effectsmodel for binary data, we calculated asummary estimate of the diagnosticodds ratio (DOR) and 95% CIs per imagingtechnique. The odds ratio is a measure ofeffect size, describing the strength of as-sociation or nonindependence betweentwo binary data values. The likelihood

ratio (LR) we applied is used for assessingthe value of performing the diagnosticimaging tests. The sensitivity and speci-ficity of the test is used to determinewhether a test result usefully changesthe probability that a condition (such asthe state of the diabetic foot, infected ornot) exists.

We generated the summary receiveroperating characteristic (sROC) curvesand documented the sensitivity, specific-ity, and positive and negative LRs forevery index test studied (i.e., MRI, 111In-oxine–WBC SPECT/CT, 99mTc-HMPAO–WBC SPECT/CT, or 18F-FDG–PET/CT). Weassessed statistical heterogeneityamong included studies using the I2

statistic (21,22), which expresses the per-centage of the variability that might bebecause of heterogeneity rather thansampling error.

RESULTS

Screening and Selection of LiteratureOur search of PubMed and Embase iden-tified 3,894 and 2,755 articles, respec-tively (Fig. 1). After screening eacharticle based on its title and abstract, weassessed the full article for 38 and in-cluded 29 studies (23–51) in our meta-analysis. Among these studies, 13 werefocused on MRI, 9 on 111In-oxine–WBC,10 on 99mTc-HMPAO–WBC, and 6 on18F-FDG–PET/CT. Two of the includedstudies were posters for which we couldextract the data, but the information wasinsufficient to allow us to perform a biasassessment. For one of these posters, weobtained additional information from anauthor and used this to include the studyin our meta-analysis.

Characteristics and MethodologicalAspects of the Included StudiesMost of the included studies provided in-formation on the population characteris-tics and key methodological aspects.Most studies were prospective, butsome were retrospective, and some en-rolled patients did not undergo all tests.Many included studies used the dual ref-erence standard of clinical follow-up fortest-negative patients and bone biopsywith subsequent pathological examina-tion for test-positive patients.Many stud-ies lacked data regarding the use ofantibiotic therapy, anti-inflammatorydrugs, or control of serum glucose levels.Only one reviewer assessed imaging re-sults in most studies; in those with more

Table 1—Search terms used for Medline and Embase

MeSH terms used for search Text words used for search

Diabetes mellitus Diabetic foot

Foot ulcer Diabetic

Radionuclide imaging Pedal

Technetium 99mTc exametazime Diabetic patient*

Infection Diabetes

Tomography Infect*

Multimodal imaging Ulcer*

Inflamm*

Osteomyelit*

Imaging

Tomograph*

CT

PET CT

PETCT

SPECT

99m-tc

Technetium tc 99m

tc 99m

99mtc

MRI

Leukocyte scan*

Leukocyte scan*

Scinti*

Search terms used for finding potential publications in PubMed and Embase. Important searchterms are in thefield of the diagnostic value of different imagingmodalities in the suspected infecteddiabetic foot.MeSH,Medical Subject Headings. *Search for all termsbeginningwith this string of text.

care.diabetesjournals.org Lauri and Associates 1113

reviewers, the authors rarely providedinformation regarding interobserveragreement.In some studies, investigators used dif-

ferent protocols for the several imagingmodalities. In most studies, the imagingprotocols useddid not conform to currentstandard acquisition procedures, andthey were often combined with a bonescan (especially for those using 111In-oxine–WBC), causing preselection ofpatients. Most of the WBC studies in-cluded in our review used only planarimages acquired with fixed, single timesand often failed to explain the acqui-sition protocol. The heterogeneity inimaging protocols of acquisition andinterpretation is evident for WBC scanusing both 99mTc-HMPAO and 111In-oxine.Studies with PET mainly consisted of

visual assessment of 18F-FDG uptakewithout any semiquantitative analysis

(e.g., measuring the standardized uptakevalue). For studies using MRI, most scan-ners were of low magnetic field strength,and, although the articles were often out-dated, they used correct protocols andsequences for identifying infections.

Figures 2 and 3 show the results of ourassessment of methodological quality,based on theQUADAS-2 checklist (patientselection, index test, reference stan-dard, flow, and timing) and our assess-ment on the risk of bias. Because wecould not assess the two included post-ers adequately, we assessed them asunknown (Fig. 2, yellow) for all corre-sponding domains.

Pooled Diagnostic Performance of theImaging Techniques (Meta-analysis)For each test, we have combined thesROC curves, and their correspondingfindings, in Figs. 4 and 5, respectively. Perimaging modality, the findings are as de-scribed below.

18F-FDG–PET/CT

This pooled analysis included 6 studiescomprising 254 patients. The perfor-mance characteristics were: sensitivity,89% (95% CI 68, 97); specificity, 92%(85, 96); DOR, 95 (18, 504); positiveLR, 11 (4.7, 25.0); and negative LR,0.11 (0.03, 0.4).111In-oxine WBC Scintigraphy

This pooled analysis included 9 studiescomprising 206 patients. The performancecharacteristics were: sensitivity, 92% (72,98); specificity, 75% (66, 82); DOR, 34 (6.9,165.7); positive LR, 3.6 (1.9, 6.7); and neg-ative LR, 0.1 (0.03, 0.4).99mTc-HMPAO WBC Scintigraphy

This pooled analysis included 10 studiescomprising 406 patients. The perfor-mance characteristics were: sensitivity,91% (95% CI 86, 94); specificity, 92%(78, 98); DOR, 118 (30, 459); positive LR,12 (3.7, 36.3); and negative LR, 0.1 (0.06,0.16).

MRI

This pooled analysis included 13 studiescomprising 421 patients. The perfor-mance characteristics were: sensitivity,93% (95% CI 82, 97); specificity, 75%(63, 84); DOR, 37 (11.3, 121.3); positiveLR, 3.66 (2.1, 6.4); and negative LR, 0.10(0.04, 0.26).

Comparison of Imaging ModalitiesIn summary, 18F-FDG–PET/CT and 99mTc-HMPAO–WBC scans each had the highestspecificity (92%), followed by MRI and111In-oxine–WBC scan (both 75%). Thesensitivity for all of the imagingmodalitieswere similar, with MRI at 93%; 111In-oxine–WBC scans, 92%; 99mTc-HMPAO–WBC scans, 91%; and 18F-FDG–PET/CT,89%.

CONCLUSIONS

This meta-analysis documents the inde-pendent performance characteristics ofthe four imaging modalities most com-monly used in the diagnosis of osteomy-elitis of the foot in patients with diabetes(i.e., MRI, 111In-oxine–WBC SPECT/CT,99mTc-HMPAO–WBC SPECT/CT, and 18F-FDG–PET/CT). To the best of our knowl-edge, this is the first meta-analysis andsystematic review that included all ofthese imaging modalities, allowing us toassess their comparative diagnostic values.The most relevant finding of our analy-sis was the higher specificity comparedwith other imaging techniques for both

Figure 1—Schematic flow chart explaining the process for selection of articles included in thismeta-analysis.

1114 Detection of Osteomyelitis in the Diabetic Foot Diabetes Care Volume 40, August 2017

18F-FDG–PET/CT and 99mTc-HMPAO–WBC scintigraphy (with either planar orSPECT/CT acquisitions). By contrast, thesensitivity was very similar for WBC-scan,18F-FDG–PET/CT, and MRI.

The sensitivity and specificity of radio-labeled WBC reported in the articles weanalyzed ranged from 75 (29) to 100%(30–32) and from 67 (32) to 100% (33),respectively. Factors that could haveinfluenced the variations in results in-clude: the number of patients imaged;the choice of imaging radiopharmaceuti-cal; the acquisition protocol of the im-ages; and the interpretation criteriafollowed for scan analysis (change of up-take with time and qualitative vs. semi-quantitative analysis). Only two of thearticles on 99mTc-WBC (33,34) used amethodology confirmed and approvedby the European Association of NuclearMedicine (EANM) (52). In four articles(35,37,45,50), the protocols of acquisitionand/or the interpretative criteria werenot explained in the text. Although weare unable to make a direct comparisonamong these articles, the two conductedaccording to EANM procedural guidelinesreported a sensitivity of 86 and 100%and a specificity of 100% (33,34), whereasthe remaining four studies showed alower average sensitivity and specificity(89 and 67%, respectively). If a currentlyapproved protocol of acquisition and in-terpretationof images hadbeenused, it ispossible that the diagnostic accuracywould have been higher. We were un-able to undertake a similar analysis forthe studies of 111In-labeled WBC, as allwere performed with outdated imagingprotocols.

Most of the studies included in our re-view used only planar images acquiredwith fixed, single times and often failedto explain the acquisition protocol. Theheterogeneity in imaging protocols of ac-quisition and interpretation is evident forWBC scan using both 99mTc- HMPAO and111In-oxine.

In only a few studies did the authorsevaluate the value of semiquantitativeanalysis. Nawaz et al. (24) performed ananalysis in a cohort of 110 subjects stud-ied with both 18F-FDG–PET/CT and MRI.They used only visual assessment of 18F-FDG uptake (without any semiquantita-tive analysis) and did not perform CTcoregistration, which might account forthe relatively low sensitivity of the tech-nique compared with MRI (81 and 91%,

Figure 2—Assessment of methodological quality based on the QUADAS-2 method (patient selec-tion, index test, reference standard, flow, and timing) for each study. Risk of bias summary andapplicability concerns assessments: green, low risk; red, high risk; and yellow, unclear risk.

care.diabetesjournals.org Lauri and Associates 1115

respectively). Conversely, Kagna et al.(25) demonstrated the value of perform-ing a maximum standardized uptakevalue evaluation and CT coregistration toprecisely evaluate theextensionof infectioninto bone and soft tissue. This study waslimited, however, by the fact that microbi-ological confirmation of infection was onlyperformed in two cases (Fig. 2, yellow).Familiari et al. (33), in accordance with

EANM recommendations, used bothqualitative and target/background ratioanalysis and acquired images at threetime points (30 min, 3 h, and 20 h post-injection), with time corrected for techne-tium decay. The performance characteristicsfor osteomyelitis (confirmed by histopa-thology or bone culture) using specificinterpretation criteria were: sensitivity,86%; specificity, 100%; positive predictivevalue; 100%, negative predictive value,86%; and diagnostic accuracy, 92%. Simi-larly, Unal et al. (32) performed a target/background ratio for early and late im-ages and found high sensitivity (94%)and specificity (100%). Other groups in-vestigated the role of SPECT or SPECT/CT with different protocols of acquisitionand, consequently, different results(34,35).Compared with a systematic review of

18F-FDG–PET/CT for diagnosing DFO pub-lished 3 years ago (53), we found a highersensitivity with a similar specificity; thisappears mainly related to their using aper-patient analysis, whereas we used aper-study analysis. Nawaz et al. (24) con-cluded that 18F-FDG–PET/CT, althoughless sensitive when compared with MRI(81 vs. 91%), had higher specificity (93 vs.78%) and diagnostic accuracy (90 vs.

81%). These results are comparable withthe findings of our systematic review.

In contrast, our results differed fromthose in a meta-analysis by Kapoor et al.

(54) of the diagnostic performanceofMRIcompared with WBC scintigraphy, bonescanning, and plain radiography. Theyconcluded that MRI was superior to

Figure 3—Assessment of methodological quality based on the QUADAS-2 method (patient selection, index test, reference standard, flow, and timing)combining all the studies. Risk of bias summary and applicability concerns assessments: green, low risk; red, high risk; and yellow, unclear risk.

Figure 4—ROC curves of all studies that included each test. Thirteen studies are included forMRI (421 patients), 9 studies for 111In-oxine–WBC (206 patients), 10 studies for 99mTc-HMPAO–WBC (406 patients), and 6 studies for 18F-FDG–PET/CT (254 patients). The dashed coloredlines are the 90% prediction intervals calculated using a bivariate hierarchical model withthe STATA-13 program.

1116 Detection of Osteomyelitis in the Diabetic Foot Diabetes Care Volume 40, August 2017

WBC scans, using the studies of Croll et al.(42) and Levine et al. (43). We also in-cluded these two studies, but we found a

higher specificity for WBC scan labeledwith 99mTc-HMPAO than for MRI, proba-bly because of the very small number of

WBC scans they included against MRI(only 2 out of the 17 included articles ofthe patients received WBC). Also, studies

Figure 5—Forest plots (of sensitivity and specificity) of all of the studies that used each test and pooled the diagnostic performance of the imagingtechniques. For each test, we have combined the sROC curves and their corresponding findings. FN, false negative; FP, false positive; TN, true negative; TP,true positive.

care.diabetesjournals.org Lauri and Associates 1117

of Kapoor et al. (54) are dated (.20 yearsold), using outdated WBC techniques(for instance, no late imaging after 24 hand lacking additional [hybrid low-dose]CT), resulting in moderate accuracy ofWBC for this indication. In the future,we suggest exploring the appealingapproach of hybrid imaging (55) in de-fining an optimal approach to imagingDFIs.A final issue concerns the convenience

of the different imaging modalities to theaffected patient. In general, all modalitiesrequire that the patient be injectedwith acontrast agent or a radiopharmaceutical,followed by imaging with the specifiedscanner (PET, SPECT, or MRI). An advan-tage of MRI is the lack of exposure of thepatient to a dose of radiation. Disadvan-tages of MRI include the fact that somepatients experience claustrophobia in the(long) gantry and that this procedure iscontraindicated in some patients whohave an implanted device (e.g., implant-able cardioverter defibrillator or cardiacpacemakers). Furthermore, the specificityof MRI is decreased by metal scatter inpatients who have metallic hardware(e.g., screws or plates) in situ at the sus-pected site of infection. A key disadvan-tage of PET and SPECT are that theyimpart a radiation dose to the patient.Furthermore, these procedures are ingeneral more expensive than MRI. Thecost of each procedure is an importantconcern, not only for the patient but alsofor the health care system. Cost will, how-ever, vary considerably over time, by thespecific procedures used, and dependingon the health care system and insuranceissues germane to the patient undertreatment. Furthermore, there is a differ-ence between the actual cost and price(or charge) for any diagnostic test. Forexample, in many centers in the U.S., theamount charged for imaging tests is muchhigher than their costs, with the profitsoften used to offset losses related toother aspects of care.

LimitationsUnfortunately, because of the limitednumber of published articles, we couldnot analyze the value of using antibodiesagainst granulocytes (either whole IgG orF[ab] fragment), despite the fact that theyare now routinely used in some nuclearmedicine centers (16–18). In fact, thenumber of patients included in all of theavailable studies was relatively small,

ranging from 6 to 110, with most enroll-ing ,20. This leads to wide CIs, makingcomparisons between different tech-niques difficult and subgroup analysesor meta-regression impossible. Addition-ally, direct comparisons of all availableimaging techniques within the same pa-tient groups were not available, as noneof the studies compared all four of thedifferent techniques in the same study.

Many studies used the dual referencestandard for osteomyelitis of clinical fol-low-up for test-negative patients andbone biopsy with subsequent pathologi-cal examination for test-positive patients.Despite the potential for introducing bias,we included all of these studies, as exclud-ing them would significantly reduce thenumber of articles available, precludingperforming a comparison among tests.We also believe that this approach ismore representative of what is widelydone in clinical practice.

As mentioned earlier, a key problem indiagnosing DFO is differentiating it fromneuro-osteoarthropathy (Charcot foot)(56). Although some studies includedCharcot foot as a separate diagnosisfrom soft tissue infection and osteomye-litis, we elected to leave this issue for afuture review. Another factor is that thepublished articles often failed to provideinformation on the patients’ use of anti-biotics or anti-inflammatory drugs and ontheir control of glucose levels. Thus, wecannot define the optimum time for im-aging following antibiotic therapy or thepossible effects of antibiotic or anti-inflammatory treatment, or glycemiccontrol, onWBC or 18F-FDG–PET/CT scans.However, three prospective studies(57–59) have concluded that WBCscintigraphy performed under (or justafter) antibiotic treatment retains a highsensitivity and specificity, perhaps evengreater than for MRI for detecting resid-ual disease (57). A retrospective study of297 patients with suspected osteomyeli-tis or soft tissue infection by Glaudemanset al. (60), although not focused on thediabetic foot, suggested that there wereno significant differences in performanceresults of WBC scintigraphy results be-tween patients who were and thosewho were not receiving antibiotic ther-apy. We are unable to assess the possibleeffect of hyperglycemia on 18F-FDG–PET/CT findings, as glucose levels wereoften missing in the retrieved articles. Inone study, however, hyperglycemia in a

fasting state did not appear to signifi-cantly influence the quality of 18F-FDG–PET/CT imaging (61). Despite all of theselimitations, including some biases, ingeneral, there will be a preferencefor 99mTc-HMPAO–labeled WBC scin-tigraphy and 18F-FDG–PET/CT, which areboth techniques that proved theirvalue in many other infectious diseasesand offer the best highest specificityfor diagnosing DFO. Sensitivity resultswere comparable among all imagingmodalities.

ConclusionOur systematic review and meta-analysissuggest that 99mTc-HMPAO–labeled WBCscintigraphy and 18F-FDG–PET/CT offerthe highest specificity for diagnosingDFO while demonstrating comparablesensitivity to the other imaging tech-niques we reviewed (MRI and WBC scin-tigraphy with 111In-oxine). In view of thecontinued lack of consensus on this issue,we believe there is a need for a standard-ization of diagnostic methods and anevidence-based sequential approach. Se-lecting the most appropriate imaging testin any clinical situation depends upon theparticular circumstances of the patient,the expertise, and equipment availableat the treating site and the costs of theprocedures. The goal is certainly to usethe most cost-effective imaging methodto allow accurate diagnosis and prompttreatment of this common, complex,and costly problem. This will requirefurther prospective studies with largernumbers of patients and with a directcomparison between the differentradiological and nuclear medicinetechniques.

Acknowledgments.The authors thankN. Smidtfor assistance and general advice for the review,S. vanderWerf forhelping conduct the literaturesearches, and Y. Takwoingi and M. Leeflang (allfrom the University Medical Center Groningen)for assisting with the data analyses and statisticalevaluations.Funding.R.H.J.A.S. is the recipient of a researchgrant from the University Medical Center Gro-ningen Healthy Aging Pilot Fund.Duality of Interest. No potential conflicts of in-terest relevant to this article were reported.Author Contributions. C.L. analyzed data andwrote the manuscript. M.T. analyzed data andreviewed the manuscript. A.W.J.M.G. designedthe studyandwroteand reviewedthemanuscript.L.E.J.O. analyzed data and edited the manuscript.P.A.E. reviewed the manuscript. P.C.J. designedthe studyandwroteand reviewedthemanuscript.

1118 Detection of Osteomyelitis in the Diabetic Foot Diabetes Care Volume 40, August 2017

B.A.L. reviewed and edited the manuscript. M.J.I.analyzed the data and reviewed the manuscript.A.S. reviewed the manuscript. R.H.J.A.S. designedthe study and wrote, edited, and reviewed themanuscript. R.H.J.A.S. is the guarantorof thisworkand, as such, had full access to all the data in thestudy and takes responsibility for the integrity ofthe data and the accuracy of the data analysis.

References1. Markakis K, Bowling FL, Boulton AJ. The dia-betic foot in 2015: an overview. Diabetes MetabRes Rev 2016;32(Suppl. 1):169–1782. Zhang P, Lu J, Jing Y, Tang S, Zhu D, Bi Y. Globalepidemiology of diabetic foot ulceration: a sys-tematic review and meta-analysis (†). Ann Med2017;49:106–1163. Lipsky BA, Aragon-Sanchez J, Diggle M, et al.;International Working Group on the DiabeticFoot. IWGDF guidance on the diagnosis and man-agement of foot infections in persons with diabe-tes. Diabetes Metab Res Rev 2016;32(Suppl. 1):45–744. BrennanMB, Hess TM, Bartle B, et al. Diabeticfoot ulcer severity predicts mortality among vet-erans with type 2 diabetes. J Diabetes Complica-tions. 2017;31:556–5615. Lavery LA, Armstrong DG, Wunderlich RP,Mohler MJ, Wendel CS, Lipsky BA. Risk factorsfor foot infections in individuals with diabetes.Diabetes Care 2006;29:1288–12936. Prompers L, HuijbertsM, Apelqvist J, et al. Highprevalence of ischaemia, infection and serious co-morbidity in patients with diabetic foot disease inEurope. Baseline results from the Eurodiale study.Diabetologia 2007;50:18–257. Lipsky BA. Osteomyelitis of the foot in diabeticpatients. Clin Infect Dis 1997;25:1318–13268. Ramanujam CL, Zgonis T. The diabetic charcotfoot from 1936 to 2016: eighty years later and stillsrowing. Clin Podiatr Med Surg 2017;34:1–89. Kucera T, Shaikh HH, Sponer P. Charcot neuro-pathic arthropathy of the foot: a literature reviewand single-center experience. J Diabetes Res2016;2016:320704310. BerendtAR, Peters EJ, Bakker K, et al. Diabeticfoot osteomyelitis: a progress report on diagnosisand a systematic review of treatment. DiabetesMetab Res Rev 2008;24(Suppl. 1):S145–S16111. Lam K, van Asten SA, Nguyen T, La Fontaine J,Lavery LA. Diagnostic accuracy of probe tobone to detect osteomyelitis in the diabeticfoot: a systematic review. Clin Infect Dis 2016;63:944–94812. van Asten SA, Jupiter DC, Mithani M, LaFontaine J, Davis KE, Lavery LA. Erythrocyte sedi-mentation rate and C-reactive protein to monitortreatment outcomes in diabetic foot osteomyeli-tis. Int Wound J 2017;14:142–14813. Palestro CJ, Love C. Nuclear medicine and di-abetic foot infections. Semin Nucl Med 2009;39:52–6514. IsraelO, Sconfienza LM, LipskyBA.Diagnosingdiabetic foot infection: the role of imaging and aproposedflow chart for assessment. Q J NuclMedMol Imaging 2014;58:33–4515. Lipsky BA, Berendt AR, Cornia PB, et al.; In-fectious Diseases Society of America. 2012 Infec-tious Diseases Society of America clinical practiceguideline for the diagnosis and treatment of di-abetic foot infections. Clin Infect Dis 2012;54:e132–e173

16. Palestro CJ, Caprioli R, Love C, et al. Rapiddiagnosis of pedal osteomyelitis in diabeticswith a technetium-99m-labeled monoclonal anti-granulocyte antibody. J Foot Ankle Surg 2003;42:2–817. Delcourt A, Huglo D, Prangere T, et al. Com-parison between Leukoscan (Sulesomab) andGallium-67 for the diagnosis of osteomyelitis in thediabetic foot. Diabetes Metab 2005;31:125–13318. Harwood SJ, Valdivia S, Hung GL, QuenzerRW. Use of Sulesomab, a radiolabeled antibodyfragment, to detect osteomyelitis in diabetic pa-tients with foot ulcers by leukoscintigraphy. ClinInfect Dis 1999;28:1200–120519. Whiting PF, Rutjes AW,WestwoodME, et al.;QUADAS-2 Group. QUADAS-2: a revised tool forthe quality assessment of diagnostic accuracystudies. Ann Intern Med 2011;155:529–53620. Moher D, Liberati A, Tetzlaff J, Altman DG;PRISMA Group. Preferred reporting items for sys-tematic reviews and meta-analyses: the PRISMAstatement. PLoS Med 2009;6:e100009721. Higgins JPT, Thompson SG. Quantifying het-erogeneity in a meta-analysis. Stat Med 2002;21:1539–155822. Higgins JPT, Thompson SG, Deeks JJ, AltmanDG. Measuring inconsistency in meta-analyses.BMJ 2003;327:557–56023. Enderle MD, Coerper S, Schweizer HP, et al.Correlation of imaging techniques to histopathol-ogy in patients with diabetic foot syndrome andclinical suspicion of chronic osteomyelitis. Diabe-tes Care 1999;22:294–29924. Nawaz A, Torigian DA, Siegelman ES, Basu S,Chryssikos T, Alavi A. Diagnostic performance ofFDG-PET, MRI, and plain film radiography (PFR)for the diagnosis of osteomyelitis in the diabeticfoot. Mol Imaging Biol 2010;12:335–34225. Kagna O, Srour S, Melamed E, Militianu D,Keidar Z. FDG PET/CT imaging in the diagnosis ofosteomyelitis in the diabetic foot. Eur J Nucl MedMol Imaging 2012;39:1545–155026. Keidar Z,MilitianuD,Melamed E, Bar-ShalomR, Israel O. The diabetic foot: initial experiencewith 18F-FDG PET/CT. J Nucl Med 2005;46:444–44927. Shagos GS, ShanmugasundaramP, Varma AK,Padma S, Sarma M. 18-F fluorodeoxy glucosepositron emission tomography-computed to-mography imaging: a viable alternative to threephase bone scan in evaluating diabetic footcomplications? Indian J Nucl Med 2015;30:97–10328. Maurer AH, Millmond SH, Knight LC, et al.Infection in diabetic osteoarthropathy: use of in-dium-labeled leukocytes for diagnosis. Radiology1986;161:221–22529. Keenan AM, Tindel NL, Alavi A. Diagnosis ofpedal osteomyelitis in diabetic patients using cur-rent scintigraphic techniques. Arch Intern Med1989;149:2262–226630. Johnson JE, Kennedy EJ, Shereff MJ, Patel NC,Collier BD. Prospective study of bone, indium-111-labeledwhite blood cell, and gallium-67 scan-ning for the evaluation of osteomyelitis in thediabetic foot. Foot Ankle Int 1996;17:10–1631. Newman LG, Waller J, Palestro CJ, et al. Leu-kocyte scanning with 111In is superior tomagneticresonance imaging in diagnosis of clinically unsus-pected osteomyelitis in diabetic foot ulcers. Dia-betes Care 1992;15:1527–1530

32. Unal SN, Birinci H, Baktiroglu S, Cantez S.Comparison of Tc-99m methylene diphospho-nate, Tc-99m human immune globulin, andTc-99m-labeled white blood cell scintigraphy inthe diabetic foot. Clin Nucl Med 2001;26:1016–102133. Familiari D, Glaudemans AW, Vitale V, et al.Can sequential 18F-FDG PET/CT replace WBC im-aging in the diabetic foot? J Nucl Med 2011;52:1012–101934. Filippi L, Uccioli L, Giurato L, Schillaci O. Di-abetic foot infection: usefulness of SPECT/CT for99mTc-HMPAO-labeled leukocyte imaging. J NuclMed 2009;50:1042–104635. Przybylski MM, Holloway S, Vyce SD, ObandoA. Diagnosing osteomyelitis in the diabetic foot:a pilot study to examine the sensitivity and spec-ificity of Tc(99m) white blood cell-labelled sin-gle photon emission computed tomography/computed tomography. Int Wound J 2016;13:382–38936. Larcos G, Brown ML, Sutton RT. Diagnosis ofosteomyelitis of the foot in diabetic patients:value of 111In-leukocyte scintigraphy. AJR Am JRoentgenol 1991;157:527–53137. Devillers A, Moisan A, Hennion F, Garin E,Poirier JY, Bourguet P. Contribution of technetium-99m hexamethylpropylene amine oxime labelledleucocyte scintigraphy to the diagnosis of dia-betic foot infection. Eur J Nucl Med 1998;25:132–13838. Crerand S, Dolan M, Laing P, Bird M, SmithML, Klenerman L. Diagnosis of osteomyelitis inneuropathic foot ulcers. J Bone Joint Surg Br 1996;78:51–5539. Splittgerber GF, Spiegelhoff DR, Buggy BP.Combined leukocyte and bone imaging used toevaluate diabetic osteoarthropathy and osteomy-elitis. Clin Nucl Med 1989;14:156–16040. Harvey J, CohenMM. Technetium-99-labeledleukocytes in diagnosing diabetic osteomyelitis inthe foot. J Foot Ankle Surg 1997;36:209–214; dis-cussion 25641. Georga S, Manes C, Lo-Presti D, et al. SPECT/CT value in the diagnosis of diabetic foot infec-tions by 99mTc-HMPAO-labeled leucocyte scan.Eur J Nucl Med Mol Imaging 2015;42(Suppl. 1):S790–S79142. Croll SD, Nicholas GG, Osborne MA, WasserTE, Jones S. Role of magnetic resonance imagingin the diagnosis of osteomyelitis in diabetic footinfections. J Vasc Surg 1996;24:266–27043. Levine SE, Neagle CE, Esterhai JL, Wright DG,Dalinka MK. Magnetic resonance imaging for thediagnosis of osteomyelitis in the diabetic patientwith a foot ulcer. Foot Ankle Int 1994;15:151–15644. Remedios D, Valabhji J, Oelbaum R, Sharp P,Mitchell R. 99mTc-nanocolloid scintigraphy for as-sessing osteomyelitis in diabetic neuropathic feet.Clin Radiol 1998;53:120–12545. ErtugrulMB, Baktiroglu S, Salman S, et al. Thediagnosis of osteomyelitis of the foot in diabetes:microbiological examination vs. magnetic reso-nance imaging and labelled leucocyte scanning.Diabet Med 2006;23:649–65346. Kearney T, Pointin K, CunninghamD, GedroycW, Robinson S, Elkeles RS. The detection of pedalosteomyelitis in diabetic patients. Practical Diabe-tes Int 1999;16:98–10047. Cook TA, Rahim N, Simpson HC, Galland RB.Magnetic resonance imaging in the management

care.diabetesjournals.org Lauri and Associates 1119

of diabetic foot infection. Br J Surg 1996;83:245–24848. Yuh WT, Corson JD, Baraniewski HM, et al.Osteomyelitis of the foot in diabetic patients:evaluation with plain film, 99mTc-MDP bone scin-tigraphy, and MR imaging. AJR Am J Roentgenol1989;152:795–80049. Yang H, Zhuang H, Rubello D, Alavi A. Mild-to-moderate hyperglycemia will not decreasethe sensitivity of 18F-FDG PET imaging in thedetection of pedal osteomyelitis in diabeticpatients. Nucl Med Commun 2016;37:259–26250. La Fontaine J, Bhavan K, Lam K, et al. Com-parison between Tc-99m WBC SPECT/CT andMRI for the diagnosis of biopsy-proven dia-betic foot osteomyelitis. Wounds 2016;28:271–27851. Zorkaltsev M, Zavadovskaya V, Udodov V,Zamyshevskaya M, Grigoryev E, Kurazhov A. Ra-diolabeledWBC-scintigraphy versusmagnetic res-onance imaging and bone scintigraphy in diabeticpatients with suspected osteomyelitis of the foot.Eur J Nucl Med Mol Imaging 2014;41(Suppl. 2):S590–S59152. Erba PA, Glaudemans AW, Veltman NC, et al.Image acquisition and interpretation criteria for

99mTc-HMPAO-labelled white blood cell scintig-raphy: results of a multicentre study. Eur J NuclMed Mol Imaging 2014;41:615–62353. Treglia G, Sadeghi R, Annunziata S, et al. Di-agnostic performance of fluorine-18-fluorodeox-yglucose positron emission tomography for thediagnosis of osteomyelitis related to diabeticfoot: a systematic review and a meta-analysis.Foot 2013;23:140–14854. Kapoor A, Page S, LavalleyM, Gale DR, FelsonDT. Magnetic resonance imaging for diagnosingfoot osteomyelitis: a meta-analysis. Arch InternMed 2007;167:125–132.55. Knesaurek K, Kolker D, Vatti S, Heiba S. Pre-cise fusion of MRI and dual energy 111InWBC/99mTc HDP SPECT/CT in the diabetic footusing companion CT: an example of SPECT/MRIimaging. Q J Nucl Med Mol Imaging 2015;59:129–13556. Ertugrul BM, Lipsky BA, Savk O. Osteomyelitisor Charcot neuro-osteoarthropathy? Differentiat-ing these disorders in diabetic patientswith a footproblem. Diabet Foot Ankle 2013;5;4.57. Vesco L, Boulahdour H, Hamissa S, et al. Thevalue of combined radionuclide andmagnetic reso-nance imaging in the diagnosis and conservative

management of minimal or localized osteomyelitisof the foot in diabetic patients. Metabolism 1999;48:922–92758. Lazaga F, Van Asten SA, Nichols A, et al.Hybrid imaging with 99mTc-WBC SPECT/CT tomonitor the effect of therapy in diabetic foot os-teomyelitis. Int Wound J 2016;13:1158–116059. Newman LG, Waller J, Palestro CJ, et al. Un-suspected osteomyelitis in diabetic foot ulcers.Diagnosis and monitoring by leukocyte scanningwith indium in 111 oxyquinoline. JAMA 1991;266:1246–125160. Glaudemans AW, de Vries EF, Vermeulen LE,Slart RH, Dierckx RA, Signore A. A large retrospec-tive single-centre study to define the best imageacquisition protocols and interpretation criteriafor white blood cell scintigraphy with 99mTc-HMPAO-labelled leucocytes in musculoskeletalinfections. Eur J Nucl Med Mol Imaging 2013;40:1760–176961. Rabkin Z, Israel O, Keidar Z. Do hyperglycemiaanddiabetes affect the incidence of false-negative18F-FDG PET/CT studies in patients evaluated forinfection or inflammation and cancer? A compar-ative analysis. J Nucl Med 2010;51:1015–1020

1120 Detection of Osteomyelitis in the Diabetic Foot Diabetes Care Volume 40, August 2017