FDG in MSK Infection.pdf

This article appeared in a journal published by Elsevier. The attachedcopy is furnished to the author for internal non-commercial researchand education use, including for instruction at the authors institution

and sharing with colleagues.

Other uses, including reproduction and distribution, or selling orlicensing copies, or posting to personal, institutional or third party

websites are prohibited.

In most cases authors are permitted to post their version of thearticle (e.g. in Word or Tex form) to their personal website orinstitutional repository. Authors requiring further informationregarding Elseviers archiving and manuscript policies are

encouraged to visit:

http://www.elsevier.com/authorsrights

Author's personal copy

FDG-PET in Musculoskeletal InfectionsChristopher J. Palestro, MD*,

Diagnosing musculoskeletal infection is challenging and imaging procedures are part of thediagnostic workup. Although the most commonly performed radionuclide procedures includebone, gallium-67, and labeled leukocyte imaging, FDG-PET (PET/CT) is assuming anincreasingly important role in the diagnostic workup of musculoskeletal infection. FDG offersadvantages over conventional radionuclide techniques. PET, a high-resolution tomographictechnique, facilitates precise localization of abnormalities. Semiquantitative analysis poten-tially could be used to differentiate infectious from noninfectious conditions and monitorresponse to treatment. FDG is a small molecule entering poorly perfused regions rapidly; theprocedure is completed in hours not days. Degenerative changes usually show faintlyincreased FDG uptake. FDG uptake usually normalizes within 3-4 months following traumaor surgery. Sensitivities higher than 95%and specicities ranging from75% to99%have beenreported in acute and subacute bone and soft tissue infection. The test is also useful fordiagnosing chronic and low-grade infection because FDG accumulates in activated macro-phages. No one tracer is equally efcacious in all regions of the skeleton and the utility of FDGvaries with the indication. One area in which FDG imaging clearly is useful, and should be theradionuclide study of choice, is in the evaluation of spinal osteomyelitis. The test has a highnegative predictive value and is a useful adjunct to MRI for differentiating degenerative frominfectious end plate abnormalities. The role of FDG imaging in the evaluation of diabetic footinfection has yet to be claried, with some investigators reporting high accuracy and othersreporting just the opposite. Although initial investigations suggested that FDG accuratelydiagnoses lower extremity joint-replacement infection subsequent studies indicate that thistest cannot differentiate aseptic loosening from infection. This is not surprising becauseaseptic loosening and infection both can be accompanied by an intense inammatory reaction.A recent meta-analysis found that the sensitivity and specicity of FDG-PET for diagnosinglower extremity prosthetic joint infectionwas 87% and 82%, respectively, lower thanwhat hasbeen reported for combined leukocyte-marrow imaging over the past 30 years. Data aboutFDG-PET in septic arthritis are limited. FDG accumulates in inammatory arthritis and its rolefor diagnosing septic arthritis likely would be limited.Semin Nucl Med 43:367-376 C 2013 Elsevier Inc. All rights reserved.

Osteomyelitis, or infection of the bone, is caused byviruses, bacteria, and fungi. The infection may belocalized or may involve periosteum, cortex, marrow, andcancellous tissue. Acute hematogenous osteomyelitis is causedby seeding of organisms within the bone that are transportedthrough the bloodstream from a remote source. Acuteosteomyelitis also can be due to the spread of organisms fromdirect trauma, a contiguous focus of infection, and post-operative sepsis.1

The diagnosis of osteomyelitis can be challenging andimaging procedures routinely are included in the diagnosticworkup of this entity. The most commonly performed radio-nuclide procedures include bone, in vitro labeled leukocyte,and more recently, FDG.FDG is transported into cells via glucose transporters and is

phosphorylated by hexokinase to 18F-2-18F-FDG-6 phos-phate but is not metabolized further. FDG accumulates inneutrophils, macrophages, and activated lymphocytes and itsuptake in these cells is related to their metabolic rate and thenumber of glucose transporters. Increased FDG uptake ininammation presumably is due to several factors. There is anincreased number of glucose transporters and an increasedexpression of these glucose transporters by activated inam-matory cells. There is an increased afnity of the glucosetransporters for FDG in inammation, which probably is

0001-2998/12/$-see front matter & 2013 Elsevier Inc. All rights reserved. 367http://dx.doi.org/10.1053/j.semnuclmed.2013.04.006

*Hofstra North Shore-LIJ School of Medicine, Hempstead, NY.Division of Nuclear Medicine and Molecular Imaging, North Shore Long

Island Jewish Health System, Manhasset & New Hyde Park, NY.Address reprint requests to Christopher J. Palestro, MD, Division of Nuclear

Medicine andMolecular Imaging, Long Island JewishMedical Center, 270-05 76th Ave, New Hyde Park, NY 11040. E-mail: [email protected]


secondary to the effects of circulating cytokines andgrowth factors. It is important to remember that althoughFDG accumulates in infection, it is a nonspecic tracer thatalso accumulates in aseptic inammation and malignantlesions.2

FDG-PET (and PET/CT) offers advantages over conven-tional radionuclide techniques. PET, intrinsically a high-reso-lution tomographic technique, facilitates precise localizationof abnormalities, especially when performed as PET/CT.FDG is a small molecule that enters poorly perfusedregions rapidly and the procedure is completed within 1-2 hours. Semiquantitative analysis, more feasible with PETtracers than with single photon emitting tracers, potentiallycould be used to differentiate infectious from noninfectiousconditions and to monitor response to treatment. FDG uptakein the normal bone marrow is low, which may facilitate thedistinction of inammatory cellular inltrates from marrow.Degenerative bone changes usually show only faintly increasedFDG uptake. FDG uptake normalizes fairly quickly followingtrauma or surgery, usually within 3-4 months3,4 (Figs. 1and 2).Sensitivities in excess of 95% and specicities ranging from

75% to 99% have been reported for FDG-PET in acute andsubacute bone and soft tissue infection (Fig. 3). The test also isuseful for diagnosing chronic and low-grade infection becauseFDG accumulates in activated macrophages, the predominantcell type present in the chronic infection.3 Guhlmann et al5

reported sensitivity and specicity of 100% and 92%, respec-tively, for FDG-PET for diagnosing chronic osteomyelitis in theaxial and peripheral skeletons. These investigators found thatthe test was especially useful for diagnosing spinal osteomye-litis, an area where labeled leukocyte imaging is of limitedvalue. de Winter et al4 reported sensitivity, specicity, andaccuracy of 100%, 88%, and 93% respectively, for FDG-PETin patients suspected of having chronic musculoskeletalinfection. Zhuang et al6 reported similar results.

FDG-PET accurately diagnoses axial and appendicularskeleton metallic implantassociated infections in traumapatients.4,7,8 Kalicke et al8 reported that among 15 patientswith osteomyelitis, including 7 with acute and 8 with chronicosteomyelitis, FDG-PET was 100% sensitive and was notaffected by the metallic implants used for fracture xation.Schiesser et al9 reported that FDG-PETwas 100% sensitive and93% specic for diagnosing metallic implantassociatedchronic infections in trauma patients. Surgeons in this inves-tigation retrospectively assessed the inuence of FDG-PET ontreatment decisions and found that the test inuenced thedecision-making process in more than half the patients.

Spinal InfectionSpinal osteomyelitis or discitis, which has a predilection for theelderly, accounts for 2%-7% of all cases of osteomyelitis andmay result from bacteremia or direct inoculation of bacteriainto the spine. Staphylococcus aureus is the most commoncausative organism. The infection usually is conned to thevertebral body and intervertebral disc; the posterior elements,however, can be involved in up to 20% of cases. Softtissue abscesses often accompany spinal osteomyelitis. Inpatients with cervical spine involvement, these abscessestypically are retropharyngeal. When the infection involvesthe thoracic spine, the abscesses are paraspinal or retrospinal.Lumbar spine infections are associated with psoasabscesses.10,11

Patients with spinal osteomyelitis often have prolongedsymptomatology before diagnosis. Most patients present withneck or back pain. Up to 20% present with neurologic defectssecondary to nerve root or spinal cord involvement. Althoughthe erythrocyte sedimentation rate is elevated in more than90% of cases, laboratory tests, except for Gram stain andculture, usually are not helpful.10

Figure 1 Degenerative spinal arthritis. Although there are extensive degenerative changes (arrows) on the CT component ofthe examination, there is uniform distribution of FDG throughout the lumbar spine.

C.J. Palestro368


MRI is the imaging procedure of choice for diagnosingspinal osteomyelitis. MRI, which has an accuracy of about90%, provides direct visualization of the spinal column,extradural soft tissues, subarachnoid space, and spinal cord,without intrathecal contrast. However, this test is sensitive tomotion degradation, and individuals who cannot remain stillfor the examination may not be suitable candidates forimaging. Metallic implants such as prosthetic heart valves,pacemakers, and orthopedic hardware may be contraindica-tions to the test. Severe degenerative disc disease, withedemalike granulation changes in the vertebral end platesand in the superior and inferior aspects of the disc, can mimicinfection andMRI cannot always distinguish between them. Inpatients in whom MRI cannot be performed or is notdiagnostic, radionuclide imaging is a useful alternative.11,12

Bone and gallium imaging are the most frequentlyperformed radionuclide procedures in patients suspectedof having spinal osteomyelitis. Neither of these tracers isideal, however. Bone scintigraphy frequently is used as a

screening test for spinal osteomyelitis. False-negative bonescan results have been reported in the elderly, possibly asa result of regional ischemia secondary to arterioscleroticdisease. The study may remain abnormal even after theinfection has resolved, because of the ongoing bonyremodeling that is part of the healing process. Bonescintigraphy is not useful for detecting the soft tissueinfections that often accompany, and at times mimic, spinalosteomyelitis.13

Gallium imaging also has limitations. The physical charac-teristics and normal biodistribution of the agent degrade imagequality. Although positive results have been reported as early as4 hours after injection, imaging for infection typically isperformed 24-72 hours after injection, necessitating multiplevisits to Nuclear Medicine. There is nothing specic aboutgallium uptake in infection; it also accumulates in cases ofaseptic inammation, traumatic foci, and tumors. Data on therole of gallium imaging in postoperative spinal infection, inpatients with and without hardware, are limited.12

Published data indicate that FDG is an excellent alternativeto conventional radionuclide imaging for diagnosing spinalosteomyelitis.5-9,14 In one of the earliest reports, Guhlmannet al5 observed that FDG-PET correctly diagnosed all 3 cases ofinfection and was true negative in the 1 patient withoutinfection. In another series, Guhlmann et al7 reported thatFDG-PET was signicantly more accurate than a radiolabeledantigranulocyte antibody for diagnosing spinal osteomyelitis.Schmitz et al15 in a study of 16 consecutive patients, reportedthat FDG-PETwas true positive in all 12 patientswith infectionand true negative in 3 of 4 patients without infection. Gratzet al16 found that 18F-FDG-PET was superior to MRI fordetecting low-grade spondylitis or discitis, was superior togallium imaging for identication of paraspinal soft tissueinfection and was superior to bone scintigraphy for differ-entiating advanced degenerative arthritis from infection.Stumpe et al17 comparedFDG-PETwithMRI for diagnosing

spinal osteomyelitis in thirty patients (38 sites) with lumbar

Figure 2 Lumbar spine compression fracture. A 3month-old compression fracture of the third lumbar vertebra (arrow) canbe appreciated on the CT component of the study. There is uniform distribution of FDG throughout the lumbar spineincluding the third lumbar vertebra.

Figure 3 Right femoral osteomyelitis. A 12-year-old patient withbacteremia and right knee pain was referred for an FDG-PET scanrather than labeled leukocyte imaging because of leukopenia. There isfocal hypermetabolism (SUV max: 2.5) in the lateral condyle of theright femur (arrow) on the coronal (left) and axial (right) images.

FDG-PET in musculoskeletal infections 369


spine vertebral end plate abnormalities. FDG-PET was truepositive in all 5 infected sites and true negative in all 33uninfected sites (100% sensitivity and 100% specicity). Thesensitivity and specicity of MRI, in comparison, were 50%and 96%, respectively. The authors concluded that FDG-PETmay be a useful adjunct to MRI for differentiating degenerativefrom infectious end plate abnormalities (Figs. 4 and 5).de Winter et al18 reviewed the results of FDG-PET

performed on patients suspected of having spinal osteo-myelitis. Of the 57 patients who had undergone previousspinal surgery, 27 had spinal implants. The medianinterval between surgery and imaging was 10 months.Fifteen patients, including 10 with and 5 without hard-ware had spinal osteomyelitis. The sensitivity, specicity,and accuracy of the test were 100%, 81%, and 86%,respectively. Sensitivity, specicity, and accuracy were notsignicantly different between the subgroup that hadsurgery within 6 months before FDG-PET and the sub-group that had undergone surgery more than 6 monthsbefore FDG-PET. The specicity of the test, however, wasconsiderably lower (65%) in the group with, than in the

group without (92%) spinal implants. The overall accuracyof FDG-PET was 86% and the negative predictive valuewas 100%. These data indicate that a negative 18F-FDG-PET study excludes spinal osteomyelitis with a high degreeof certainty (Fig. 6). In patients with spinal hardware,however, a positive study is less reliable and should beinterpreted cautiously.In most investigations of FDG imaging of spinal infection,

PET imaging alone has been used. In more recent studies,however, PET/CT has been investigated. Hartman et al19 aspart of larger series, reported that FDG-PET/CT wastrue positive in all 7 cases of spinal osteomyelitis, and truenegative in both patients without spinal osteomyelitis. Theauthors reported that the precise anatomical localizationprovided by PET/CTwas especially useful for planning surgicalintervention and for differentiating soft tissue from boneinvolvement, thereby guiding treatment. Kim et al20 performeddual time point FDG-PET/CT in 22 patients with suspectedspinal infection. Patients were imaged at 1 and 2 hoursafter injection. Although the test was very sensitive fordetecting infection, neither visual nor semiquantitative

Figure 4 Lumbar spine osteomyelitis. On the coronal CT image (left), there is loss of disc space with destruction of thevertebral end plates at L1-L2 (arrow). There is a corresponding area of hypermetabolism (arrow) on the coronal FDG-PETimage (right).

Figure 5 Vertebral end plate destruction. On the coronal CT image (left) there are degenerative changes (upper arrow) andloss of disc space with destruction of the vertebral end plates at L3-L4 (arrow). There is normal FDG accumulationthroughout the lumbar spine on the PET image (right). Bone biopsy and cultures showed negative results for infection.FDG-PET is useful for distinguishing degenerative disc disease and even severe end plate destruction from infectiousspondylodiscitis, a differentiation, which as this case illustrates, is not always possible with anatomical imaging. Comparewith Figure 4.

C.J. Palestro370


analysis could reliably differentiate pyogenic from tuberculousinfection.The results of FDG-PET and PET/CT for diagnosing spinal

osteomyelitis are encouraging; nevertheless there are limita-tions to the test. It is likely that the test would be less reliable fordifferentiating infection from, and detecting infection super-imposed on, tumor, but data comparing FDG uptake in spinalinfection with uptake in spinal neoplasm are lacking. Thepresence of a foreign body can incite an intense immuneresponse and increased FDGuptake, in the absence of infectionis a well-known phenomenon.21,22 Thusmore data are neededto determine the value of FDG imaging in patients with spinal

hardware. FDG accumulation in degenerative changes gener-ally is thought to be relatively low. However, in a retrospectivestudy of 150 patients, Rosen et al23 found signicant focal FDGuptake in degenerative spinal disease in more than half of thepatients (Fig. 7). It is important to note that although the facetjoints are involved in up to 20%of cases of spinal osteomyelitis,FDG uptake localized to the posterior elements of the vertebracannot be equated automatically with infection.Although specicity is an issue, as it is with gallium and

bone imaging, FDG is sensitive, is completed in a matter ofhours rather than days, and image quality is vastly superiorto that obtained with single photon emitting tracers, even

Figure 6 Uninfected spinal hardware. (A) Anterior and lateral scout images reveal a considerable amount of orthopedichardware in the lumbar spine of a patient with low back pain. (B) Coronal and sagittal PET/CT images reveal normal FDGuptake. FDG imaging is very reliable for excluding infection in patients with spinal hardware.



when using Single Photon Emission Computed Tomography,SPECT/CT.24 In our institution, when a radionuclide test forspinal osteomyelitis is requested, we perform FDG-PET/CT,rather than gallium SPECT/CT, whenever possible.

Diabetic Foot InfectionsDiabetic foot infection is dened as an inframalleolar infectionin a person with diabetes mellitus. These infections, which areassociated with considerable morbidity, are responsible for thelargest number of diabetes-related hospital bed days and arethe most common cause of nontraumatic lower extremityamputations. Themajor predisposing factor to these infectionsis the mal perforans ulcer, which results from trauma orexcessive pressure on a foot lacking protective sensation. Oncethe cutaneous integument is breached, the wound maybecome actively infected and, by contiguous extension,infection can extend into the bone.25

The diagnosis of osteomyelitis often is overlooked becausediabetics, even in those with a signicant foot infection, canpresent without pain and not mount a systemic inammatoryresponse. Imaging studies are therefore, an essential part of thediagnostic evaluation of these individuals. Labeled leukocyteimaging is the radionuclide gold standard for diagnosingpedal osteomyelitis in diabetics. The sensitivity of planarimaging, using 111In-labeled leukocytes, has ranged from72% to 100% and the specicity from 67% to 100%. Thesensitivity and specicity of 99mTc-exametazime-labeled leu-kocyte planar imaging for diagnosing diabetic pedal osteo-myelitis have ranged from 86% to 93% and from 80% to 98%,respectively.25 More recent data suggest that that labeledleukocyte SPECT/CT imaging improves the accuracy of thetest. Nevertheless the intrinsically low resolution of thesetechniques, together with the small size of the structures beingevaluated and the limitations inherent in the in vitro labelingprocess, have motivated investigators to seek alternatives tolabeled leukocyte imaging.

The intrinsically high resolution of PET is a signicantadvantage over single photon emitting tracers, especially whenaccurate diagnosis depends on the precise localization ofradiotracer accumulation in structures as small as the distalforefoot, where the majority of diabetic foot infections occur.Not surprisingly, the role of FDG-PET and PET/CT in theevaluation of diabetic foot infections has been investigated byseveral groups.Hopfner et al26 studied the role of FDG-PET forpreoperative identication of neuropathic joints in diabeticpatients. The test correctly identied 95% (37/39) of thelesions, including 22/24 bone lesions and all 15 joint/soft tissuelesions. Themean SUVmax in these lesionswas 1.8 (range 0.5-4.1). These investigators reported that the sensitivity of the testwas not affected by blood glucose levels, even though imagequalitywas better in patientswith blood glucose levels less than200 mg/dL than in patients with blood glucose levels above200 mg/dL. Although none of the patients in this investigationhad osteomyelitis, these investigators suggested that, becauseof the relatively low SUV max in the uninfected neuropathicjoints, and because of the high SUV max expected inosteomyelitis, FDG-PET could differentiate osteomyelitis fromneuropathic disease.Basu et al27 used FDG-PET to differentiate osteomyelitis and

soft tissue infection from the uninfected neuropathic joint indiabetics. These investigators reported that the mean SUVmaxin uninfected neuropathic jointswas 1.3 0.4. Themean SUVmax in pedal osteomyelitis was 4.38 1.39 and the SUVmaxin the 1 case of osteomyelitis superimposed on a neuropathicjoint was 6.5. The sensitivity and accuracy of FDG-PET fordiagnosing osteomyelitis in this investigation were 100% and94%, respectively. The investigators concluded that FDGuptake in the neuropathic joint was distinct from that inosteomyelitis and that FDG-PET had a high negative predictivevalue for excluding osteomyelitis in the presence of theneuropathic joint. It is worth noting that only 5 of the 63patients in this investigation had foot ulcers. Pedal osteomye-litis in diabetics without ulcers infrequently occurs, as thisseries illustrates. Given the lack of ulcers in this study group, it

Figure 7 Lumbar spine facet joint arthritis. There is increased FDG uptake in both facet joints (arrows) of a lower lumbarvertebra. The SUVmax of the right facet joint was 4.4 and the SUVmax of the left facet joint was 5.6. The facet joints can beinvolved in up to 20% of spinal osteomyelitis cases and it may not always be possible to differentiate infection from facetjoint arthritis with FDG.

C.J. Palestro372


is difcult to extrapolate these results to the diabetic populationusually referred for radionuclide imaging: patients with pedalulcers.Nawaz et al28 prospectively investigated 110 diabetic

patients with possible pedal osteomyelitis. Blood glucose levelwas less than 200 mg/dL in all patients. No information aboutthe number of patients with foot ulcers was provided. Usingvisual image analysis only, the investigators reported that FDG-PET had a sensitivity, specicity, and accuracy of 81%, 93%,and 90%, respectively, for diagnosing pedal osteomyelitis.Schwegler at al29 prospectively evaluated FDG-PET for

diagnosing clinically unsuspected osteomyelitis in 20 diabeticpatients, all of whom had pedal ulcers. Information on bloodglucose levels at the time of imaging was not provided. Onlyvisual image analysis was performed. Histopathologic/micro-biologic conrmation of the nal diagnosis was available for7 patients. FDG-PET detected only 2 (29% sensitivity) of7 cases of osteomyelitis. The authors hypothesized that the lowsensitivity of FDG-PET may have been related to a lower levelof inammatory response in their population or to insulinresistance,whichmay have impaired bony uptake of FDG. Theauthors also commented that motion artifacts and limitedspatial resolution created problems that may have contributedto the poor results.Keidar et al30 compared FDG-PET and PET/CT in 18

clinically suspected sites of infection, including 12 with openwounds or ulcers. Blood glucose levels exceeded 200 mg/dLin 7 patients. Histopathologic/microbiologic conrmation ofthe nal diagnosis was available only for 2 sites. PET/CT locali-zed uptake to bone in 9 sites, including 8 sites of osteomyelitis.The accuracy of FDG-PET/CT in this investigation was about94%. Themean SUVmax in infectious foci was 5.7 and rangedfrom 1.7 to 11.1, for both osseous and soft tissue foci ofinfection. There was no relationship between the patients'glycemic state and the degree of FDG uptake.Kagna et al31 investigated FDG-PET/CT in 39 patients

including 38 with type 2 diabetes. Fourteen of the patientshad been included in an earlier investigation. There were 46sites suspicious for infection: 27, forefoot; 6, midfoot; and 13,hindfoot. Infection was suspected based on the presence ofnonhealing wounds, necrotic ulcers, cellulitis, or severe footpain, with or without systemic manifestations. At the time ofthe study, 29 patients were on antibiotic therapy. There were18 sites of osteomyelitis, of which 5 were conrmed byhistology, 4 by inspection of the bones at the time of surgeryand the remaining were diagnosed clinically. The sensitivity,specicity, and accuracy of the test, using lesion analysis was100%, 93%, and 96%, respectively. The mean SUV max insites of osteomyelitis was 6.7 3.7 (1.7-15.7) vs 4.4 2.4(1.5-10) for soft tissue infection (p o 0.05). Blood glucoselevels ranged from 53 to 330 mg/dL and exceeded 150 mg/dLin 23 patients, including 6 patients with osteomyelitis. Therewas no correlation between blood glucose levels and SUVmaxat the sites of abnormal FDG accumulation. The authorsconcluded that FDG-PET/CT is useful for diagnosing pedalosteomyelitis in diabetic patients. The PET component iden-ties FDG-avid foci in sites of acute infection and the CTcomponent precisely localizes foci to bone or soft tissue.

Familiari et al32 compared FDG-PET/CT to planar99mTc-exametazime-labeled leukocyte imaging for diagnosingpedal osteomyelitis in 13 diabetic patients, all of whom had ahigh pretest likelihood of infection, including 7 who presentedwith exposed bone. All patients had a blood glucose levelof less than160 mg/dL. Histopathologic conrmation of thenal diagnosis was available for all cases. FDG-PET/CT imag-ing was performed 10 minutes, and 1 and 2 hours after injec-tion. These investigators found that the highest accuracy forFDG-PET/CT, 54%, was achieved when the SUV max was 2.0 at 1 and 2 hours after injection and increased over time.Accuracy improved to 62% when CT ndings were included.The accuracy of planar 99mTc-exametazime-labeled leukocyteimaging, in contrast, was 92%. The authors concluded thatFDG-PET/CT is not accurate, and cannot replace labeledleukocyte imaging, for diagnosing pedal osteomyelitis indiabetics. The results of this investigation are in stark contrastto those reported by Keidar et al30 and Kagna et al31

Unfortunately, it is not possible to make meaningfulcomparisons among the various studies. Some investigatorsused only visual image interpretation, others used only semi-quantitative (SUV) analysis, one used visual and SUV analysisand in one investigation it is not clear how images wereinterpreted. Details about the number of patientswith type 1 vstype 2 diabetes is lacking inmost series, as is information aboutthe types of treatment. Perhaps the low sensitivity reported bysome investigators was related to more severe vascularinsufciency in their populations than in other populations,yet these critical data are absent.33

Diagnosing pedal osteomyelitis may be the most difcultaspect of managing foot infections in the diabetic patient. Afternearly 10 years of investigation, the role of FDG-PET and PET/CT for this indication is yet to be established.

Prosthetic Joint InfectionMore than 1 million lower extremity joint arthroplasties areperformed annually in the United States. Failures owing tofracture, dislocation, and heterotopic ossication, relativelyuncommon, usually can be diagnosed radiographically. Themost common cause of prosthetic failure is aseptic loosening.More than 25% of all prostheses eventually demonstrateevidence of loosening, often necessitating revision arthroplasty.The most frequent cause of aseptic loosening is an inamma-tory reaction to one or more of the prosthetic components.Particulate debris attracts and activates the tissue phagocytesthat normally accumulate around the device. This debris isimpervious to enzymatic destruction, and stimulates repeated,futile, attempts at phagocytosis. This in turn stimulates thesecretion of proteolytic enzymes and cytokines that damagebone and cartilage and activate immune cells, resulting inprogressive osteolysis and loosening of the prosthesis. Asynoviallike pseudomembrane develops. This pseudomem-brane most often is composed of histiocytes (95% of speci-mens) and giant cells (80%), and less frequently, lymphocytesand plasma cells (25%). Neutrophils rarely are found (10%).34



Infection, a very uncommon cause of prosthetic joint failure,ranges in frequency from about 1%-2% for primary implants,to about 3%-5% for revision implants. The inammatoryreaction that accompanies the infected prosthesis can be similarto that present in aseptic loosening, with one critical difference:neutrophils, rarely present in aseptic loosening, invariably arepresent, and usually in large numbers, in infection.34

Differentiating aseptic loosening from infection of a pros-thetic joint is extremely important because the treatment ofthese 2 entities is very different. Treatment of aseptic looseningusually consists of a single-stage exchange arthroplasty thatis completed in a single hospital admission with 1 surgicalintervention. The treatment of the infected joint replacement,in contrast, typically requires multiple admissions. An exci-sional arthroplasty, or removal of the prosthesis, is performedfollowed by several weeks to months of antimicrobial therapy,followed eventually by a revision arthroplasty. A sensitive butnonspecic test would lead to multiple, expensive, operationsin patients for whom a single intervention may have beensufcient. The specic, but insensitive, test would also result inadditional surgical interventions, because undiagnosed infec-tion would cause any revision implant to fail.34

Differentiating between aseptic loosening and infection canbe challenging. Clinical signs and symptoms of infection oftenare absent. Increased numbers of circulating leukocytes, andelevated erythrocyte sedimentation rate and C-reactive proteinlevels are neither sensitive nor specic. Joint aspiration withGram stain and culture is the denitive preoperative diagnosticprocedure. Although the specicity of the test is about 90%,the sensitivity is more variable, ranging from about 30% to

90%. Among the various imaging studies, plain radiographsare neither sensitive nor specic and cross-sectional imagingmodalities, such as CT and MRI are limited by metallicartifacts.11,13

Radionuclide imaging currently is the imaging method ofchoice for evaluation of suspected joint replacement infection,and combined labeled leukocyte-marrow imaging, which hasan accuracy of about 90%, is the radionuclide study of choicefor diagnosing prosthetic joint infection.34

Diagnosing prosthetic joint infection with FDG-PET hasbeen investigated extensively. Zhuang et al35 in an investigationof 74 prosthetic joints, including 21 infected devices, foundthat the presence of increased FDG activity along the boneprosthesis interface indicated the presence of infection. Theseinvestigators reported a sensitivity, specicity, and accuracy of90%, 89.3%, and 89.5%, respectively, for prosthetic hipinfection, and sensitivity, specicity, and accuracy of 90.9%,72%, and 77.8%, respectively, for prosthetic knee infection.They also reported that the accuracy of the test was dependenton the location, not the intensity, of FDG uptake.Chacko et al36 reported that bone prosthesis interface

activity along the shaft of the femoral component of a hipreplacement was 92% sensitive and 97% specic for infection.Similar to Zhuang et al35, they found that intensity of FDGuptake was not useful for differentiating infected from asepti-cally loosened prostheses. Reinartz et al37 reported that activityaround the acetabular component and proximal aspect of thefemoral component of a hip replacement was not associatedwith infection. They also observed that although periprostheticFDG uptake patterns were useful for differentiating infection

Figure 8 Asymptomatic hip prostheses. There is extensive, heterogeneous hypermetabolism adjacent to the proximal aspectof both hip prostheses, more extensive on the left. Activity adjacent to the right hip prosthesis had an SUV max of 7.6,whereas activity adjacent to the left hip prosthesis had an SUV max of 5.1. The so-called bone prosthesis interface activityalong the medial and lateral aspects of the femoral component of the left hip prosthesis (arrowheads) can be noted, whichsome investigators have described as specic for infection. When evaluating joint prostheses, both the attenuation-corrected andnonattenuation corrected images should be reviewed. Sometimes apparent periprosthetic activity is becauseof an artifact produced by attenuation correction. This was not the case in this individual as the activity is seen on both thecorrected (center) and uncorrected (right) images.

C.J. Palestro374


from aseptic loosening, intensity of FDG uptake was notuseful.Contradictory ndings, however, have been reported by

others. Van Acker et al38 studied 21 patients with suspectedprosthetic knee infection and reported that FDG-PET was100% sensitive but only 73% specic for diagnosing infection.Interpreting the FDG-PET studies together with bone scintig-raphy, improved test specicity to 80%.Manthey et al39 reported that by analyzing the patterns and

intensity of periprosthetic uptake, it was possible to accuratelydifferentiate among aseptic loosening, synovitis, and infection.They also reported that activity around the femoral head andneck indicated the presence of synovitis and infection. Thesendings directly contradict those of Zhuang et al35, Chackoet al36, and Reinartz et al37.Stumpe et al40 compared bone prosthesis interface activity

to urinary bladder activity in patients with 35 painful hipprostheses. These investigators classied studies in whichperiprosthetic activity was intense as positive for infection.They did not analyze the location of bone prosthesis interfaceactivity. They found that, although FDG-PET was reasonablyspecic (81%-85%), the test was not sensitive for diagnosinginfection (33%-56%). The overall accuracy of the test was69%, which was lower than the 80% accuracy of bonescintigraphy alone in the same population. False-positiveresults, not surprisingly, were associated with foreign bodyreactions in aseptically loosened devices.Love et al22 using coincidence detection imaging compared

FDG-PET and leukocyte-marrow imaging in 59 failed lowerextremity joint replacements. Among the various criteria usedfor FDG image interpretation, the presence of bone prosthesisinterface activity, with a target-to-background ratio greaterthan 3.6 for hip replacements and 3.1 for knee replacementswere the most accurate (71%) for diagnosing infection. Theaccuracy of leukocyte-marrow imaging, 95%, was signicantlyhigher. These investigators concluded that FDG-PET cannotreliably differentiate aseptic loosening from infection of a lowerextremity prosthetic joint.Kwee et al41, in a meta-analysis, reported that the sensitivity

and specicity of FDG-PET for diagnosing prosthetic jointinfectionwere 82% and 87%, respectively. This is considerablylower than what has been reported, by numerous investigatorsfor over more than 30 years, for labeled leukocyte-marrowimaging. Consequently, there appears to be little, if any role forFDG-PET in the evaluation of prosthetic joint infection (Fig. 8).

Septic ArthritisInfectious organisms reach the joint through direct inocula-tion, hematogenous spread, or contiguous spread from anadjacent intra-articular site of infection. Ultrasound and MRIare the principal imaging studies used for evaluating thesuspected septic joint and show changes before radiography.Radionuclide studies are of limited use in the evaluation ofseptic arthritis. The classic presentation of acute arthritis, bothseptic and aseptic, on the 3-phase bone scan is hyperperfusionand hyperemia of the joint on early images, with increasedactivity limited to the articular surfaces of the involved boneson delayed, bone, images. Osteomyelitis and acute arthritis,moreover, are not mutually exclusive, and bone scan ndingsconsistent with arthritis do not exclude and potentially canobscure a superimposed osteomyelitis.11 Neither gallium norlabeled leukocyte imaging reliably separate septic from inam-matory arthritis.42-44

At the present time, data about the role of FDG-PET in septicarthritis are limited. Like gallium and labeled leukocytes, FDGaccumulates in inammatory arthritis45 (Fig. 9). Conse-quently, the role of FDG-PET and PET/CT for diagnosingseptic arthritis likely would be very limited.

References1. Osman DR: Diagnosis and management of musculoskeletal infection. In:

Fitzgerald RH, Haufer H, Malkani RL, (eds): Orthopedics. St. Louis,Mosby; 2002, pp. 695-707

2. LoveC, TomasMB,TroncoGG, et al: Imaging infection and inammationwith 18F-FDG-PET. RadioGraphics 2005;25:1357-1368

3. Strobel K, Stumpe KDM: PET/CT in musculoskeletal infection. SeminMusculoskelet Radiol 2007;11:353-364

4. de Winter F, van de Wiele C, Vogelaers D, et al: Fluorine-18 uorodeox-yglucose positron emission tomography: A highly accurate imagingmodality for the diagnosis of chronic musculoskeletal infections. J BoneJoint Surg Am 2001;83-A:651-660

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

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

7. Guhlmann A, Brecht-Krauss D, Suger G, et al: Fluorine-18-FDG PET andtechnetium-99m antigranulocyte antibody scintigraphy in chronic osteo-myelitis. J Nucl Med 1998;39:2145-2152

8. Kalicke T, Schmitz A, Risse JH, et al: Fluorine-18 uorodeoxyglucose PETin infectious bone diseases: Results of histologically conrmed cases. Eur JNucl Med Mol Imaging 2000;27:524-528

Figure 9 Rheumatoid arthritis. The intense FDG accumulation in both hip joints can be noted. As with single photonemitting tracers, it is not likely that FDG plays a signicant role in the diagnosis of the septic joint.



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

10. Love C, Patel M, Lonner BS, et al: Diagnosing spinal osteomyelitis:A comparison of bone and gallium scintigraphy and magnetic resonanceimaging. Clin Nucl Med 2003;25:963-977

11. Palestro CJ, Love C,Miller TT: Imaging ofmusculoskeletal infections. BestPract Res Clin Rheumatol 2006;20:1197-1218

12. Gemmel F, Dumarey N, Palestro CJ: Radionuclide imaging of spinalinfections. Eur J Nucl Med Mol Imaging 2006;33:1226-1237

13. Palestro CJ, Love C: Radionuclide imaging of musculoskeletal infection:Conventional agents. Semin Musculoskelet Radiol 2007;11:335-352

14. Meller J, Koster G, Liersch T, et al: Chronic bacterial osteomyelitis:Prospective comparison of F-18-FDG imaging with a dual-head coinci-dence camera and In-111-labelled autologous leucocyte scintigraphy. EurJ Nucl Med Mol Imaging 2002;29:53-60

15. Schmitz A, Risse JH, Grunwald F, et al: Fluorine-18 uorodeoxyglucosepositron emission tomography ndings in spondylodiscitis: Preliminaryresults. Eur Spine J 2001;10:534-539

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

17. Stumpe KD, Zanetti M, Weishaupt D, et al: FDG positron emissiontomography for differentiation of degenerative and infectious endplateabnormalities in the lumbar spine detected on MR imaging. AJR Am JRoentgenol 2002;179:1151-1157

18. de Winter F, Gemmel F, Van de Wiele C, et al: 18-uorine uorodeox-yglucose positron emission tomography for the diagnosis of infection inthe postoperative spine. Spine 2003;28:1314-1319

19. HartmannA, EidK,DoraC, et al: Diagnostic value of 18F-FDGPET/CT intraumapatientswith suspected chronic osteomyelitis. Eur JNuclMedMolImaging 2007;34:704-714

20. Kim SJ, Lee JS, Suh KT, et al: Differentiation of tuberculous and pyogenicspondylitis using double phase F-18 FDG PET. Open Med Imaging J2008;2:1-6

21. de Winter F, Huysse W, De Paepe P, et al: High F-18 FDG uptake in aparaspinal textiloma. Clin Nucl Med 2002;27:132-133

22. Love C, Marwin SE, Tomas MB, et al: Diagnosing infection in the failedjoint replacement: A comparison of coincidence detection 18F-FDG and111In-labeled leukocyte/99mTc-sulfur colloid marrow imaging. J NuclMed 2004;45:1864-1871

23. Rosen RS, Fayad L, Wahl RL: Increased 18F-FDG uptake in degenerativedisease of the spine: Characterization with 18FFDG PET/CT. J Nucl Med2006;47:1274-1280

24. Gemmel F, Rijk PC, Parlevliet T, et al: Expanding role of 18F-uoro-D-deoxyglucose PET and PET/CT in spinal infections. Eur Spine J2010;19:540-551

25. Palestro CJ, Love C: Nuclear medicine and diabetic foot infections. SeminNucl Med 2009;39:52-65

26. Hopfner S, Krolak C, Kessler S, et al: Preoperative imaging of Charcotneuroarthropathy in diabetic patients: Comparison of ring PET, hybridPET, and magnetic resonance imaging. Foot Ankle Int 2004;25:890-895

27. Basu S, Chryssikos T, Houseni M, et al: Potential role of FDG-PET in thesetting of diabetic neuro-osteoarthropathy: Can it differentiate

uncomplicated Charcot's neuropathy from osteomyelitis and soft tissueinfection? Nucl Med Commun 2007;28:465-472

28. Nawaz A, Torigian DA, Siegelman ES, et al: Diagnostic performance ofFDG-PET, MRI, and plain lm radiography (PFR) for the diagnosis ofosteomyelitis in the diabetic foot. Mol Imaging Biol 2010;12:335-342

29. Schwegler B, Stumpe KD, Weishaupt D, et al: Unsuspected osteomyelitisis frequent in persistent diabetic foot ulcer and better diagnosed by MRIthan by 18F-FDG PET or 99mTc-MOAB. J Intern Med 2008;263:99-106

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

31. Kagna O, Srour S, Melamed E, et al: FDG PET/CT imaging in thediagnosis of osteomyelitis in the diabetic foot. Eur J Nucl Med MolImaging 2012;39:1545-1550

32. Familiari D, Glaudemans AWJM, Vitale V, et al: Can sequential 18F-FDG-PET/CT imaging replace WBC imaging in the diabetic foot? J Nucl Med2011;52:1012-1019

33. PalestroCJ: FDGanddiabetic foot infections: The verdict is. J NuclMed2011;52:1009-1011

34. Love C, Marwin SE, Palestro CJ: Nuclear medicine and the infected jointreplacement. Semin Nucl Med 2009;39:66-78

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

36. Chacko TK, ZhuangH, Stevenson K, et al: The importance of the locationof uorodeoxyglucose uptake in periprosthetic infection in painful hipprostheses. Nucl Med Commun 2002;23:851-855

37. Reinartz P, Mumme T, Hermanns B, et al: Radionuclide imaging of thepainful hip arthroplasty. Positron-emission tomography versus triple-phase bone scanning. J Bone Joint Surg Br 2005;87:465-470

38. 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 2001;28:1496-1504

39. Manthey N, Reinhard P, Moog F, et al: The use of [18F] uorodeox-yglucose positron emission tomography to differentiate between synovitis,loosening and infection of hip and knee prostheses. Nucl Med Commun2002;23:645-653

40. 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.Radiology 2004;231:333-341

41. Kwee TC, Kwee RM, Alavi A: FDG-PET for diagnosing prosthetic jointinfection: Systematic review and metaanalysis. Eur J Nucl Med MolImaging 2008;35:2122-2132

42. Uno K, Matsui N, Nohira K, et al: Indium-111 leukocyte imaging inpatients with rheumatoid arthritis. J Nucl Med 1986;27:339-344

43. Palestro CJ, Vega A, Kim CK, et al: Appearance of acute gouty arthritis onindium-111 labeled leukocyte scintigraphy. JNuclMed 1990;31:682-687

44. Palestro CJ, Goldsmith SJ: In-111 labeled leukocyte imaging in a case ofpseudogout. Clin Nucl Med 1992;17:366-367

45. Beckers C, Ribbens C, Andre B, et al: Assessment of disease activity inrheumatoid arthritis with (18)F-FDG PET. J Nucl Med 2004;45:956-964

C.J. Palestro376

Documents

FDG in MSK Infection.pdf