68Ga-DOTAVAP-P1 PET imaging capable of demonstrating the phase of inflammation in healing bones and the progress of infection in osteomyelitic bones

ORIGINAL ARTICLE

68Ga-DOTAVAP-P1 PET imaging capable of demonstratingthe phase of inflammation in healing bones and the progressof infection in osteomyelitic bones

Petteri Lankinen & Tatu J. Mäkinen &

Tiina A. Pöyhönen & Pauliina Virsu & Satu Salomäki &Antti J. Hakanen & Sirpa Jalkanen & Hannu T. Aro &

Anne Roivainen

Received: 1 June 2007 /Accepted: 9 October 2007 / Published online: 24 November 2007# Springer-Verlag 2007

AbstractPurpose Differentiation between bacterial infection andnonbacterial inflammation remains a diagnostic challenge.Vascular adhesion protein 1 (VAP-1) is a human endothelialprotein whose cell surface expression is induced underinflammatory conditions, thus making it a highly promisingtarget molecule for studying inflammatory processes in vivo.We hypothesized that positron emission tomography (PET)with gallium-68-labeled 1,4,7,10-tetraazacyclododecane-N′,N″,N′′′,N″″-tetraacetic acid-peptide targeted to VAP-1 (68Ga-DOTAVAP-P1) could be feasible for imaging the earlyinflammatory and infectious processes in healing bones.

Materials and methods Thirty-four Sprague–Dawley ratswith diffuse Staphylococcus aureus tibial osteomyelitis and34 rats with healing cortical bone defects (representing theinflammation stage of healing) were PET imaged using68Ga-DOTAVAP-P1 as a tracer. In addition, peripheralquantitative computed tomography and conventional radiog-raphy were performed. Bone samples for quantitativebacteriology and specimens were also processed for histo-morphometry of inflammatory and infectious reactions.Results PET imaging showed an uptake of 68Ga-DOTA-VAP-P1 in both the osteomyelitic bones and the healingcortical bone defects during the first 36 h after surgery.Thereafter, only the osteomyelitic tibias were delineated byPET. The osteomyelitic and control animals showed asimilar uptake of the 68Ga-DOTAVAP-P1 at 24 h, whereas asignificant difference was observed at 7 days (p<0.0001).Conclusions The current study showed that PET imagingwith the new 68Ga-DOTAVAP-P1 is capable of accuratelydemonstrating the phase of inflammation in healing bonesand the progress of bacterial infection in osteomyeliticbones. Consequently, this novel imaging agent allowed forthe differentiation of bone infection due to S. aureus andnormal bone healing as soon as 7 days after onset.

Keywords Infectious disease . PET. Osteomyelitis .

Bone healing . 68Ga-DOTA-peptide

Introduction

Positron emission tomography (PET) using 2-[18F]fluoro-2-deoxy-D-glucose (18F-FDG) has been introduced as apromising imaging modality for the evaluation of various

Eur J Nucl Med Mol Imaging (2008) 35:352–364DOI 10.1007/s00259-007-0637-5

P. Lankinen : T. J. Mäkinen :H. T. AroOrthopaedic Research Unit, Department of Orthopaedic Surgeryand Traumatology, University of Turku,Turku, Finland

T. A. Pöyhönen : P. Virsu : S. Salomäki :A. Roivainen (*)Turku PET Centre, Turku University Hospital,Turku, Finlande-mail: [email protected]

S. SalomäkiDepartment of Chemistry, University of Turku,Turku, Finland

A. J. HakanenDepartment of Bacterial and Inflammatory Diseases,National Public Health Institute,Turku, Finland

S. JalkanenMediCity Research Laboratory, University of Turku,Turku, Finland

S. JalkanenNational Public Health Institute,Turku, Finland

infectious conditions of the skeleton [1–3], but the diagnosisof infection in postoperative states remains a challenge. Themethod is reported to be of limited value for differentiationbetween infectious bacterial and nonbacterial inflammatoryprocesses [4, 5]. The problem is that early bone healinginvolves an inflammatory phase that represents a highlyactivated state of cell metabolism and glucose consumption,mimicking an infection on 18F-FDG PET imaging [6, 7].The initial inflammatory response after a trauma involvingbone lasts approximately 48 h. In a recent study utilizing arat osteomyelitis model, bone infection could be distin-guished from bone healing by means of PET using gallium-68 (68Ga) chloride as soon as 2 weeks after onset [8].However, under clinical conditions, more sensitive tracersare needed to describe the evolution of infection. Earlydetection of infection is important because progressiveosteomyelitis can cause devastating changes in bonestructures and it has a tendency to become chronic.Therefore, novel more specific tracers are needed for theimaging of infections. We also lack adequate imagingmodalities for the evaluation of treatment response inantimicrobial studies of bone infection. During the pasttwo decades, several radiopharmaceuticals have beendeveloped for imaging the infection and inflammatoryprocesses, but as yet, none of them has been found to haveoptimal characteristics [9].

Small synthetic peptides are compounds that accumulaterapidly at the site of inflammation and so form a group ofpromising radiopharmaceuticals for the detection of infectiousor inflammatory foci [10–12]. We have evaluated a new68Ga-labeled peptide inhibitor of vascular adhesion protein1/semicarbazide sensitive amine oxidase (VAP-1/SSAO) as aPET tracer for the assessment of inflammatory reaction(Pöyhönen et al. unpublished data). VAP-1/SSAO is a dualfunction endothelial protein with adhesion properties andamine oxidase activity [13, 14]. As an adhesion molecule,VAP-1 coordinates immune response, i.e., the extravasationcascade of white blood cells to inflammatory areas upononset [15]. Several studies with both clinical and experi-mental models have revealed that VAP-1 is upregulated onvasculature at various sites and diseases [16, 17]. Cellsurface expression of VAP-1/SSAO is induced only at sitesof inflammatory reaction extending into non-affected tissues,which makes it a highly promising target molecule to studyinflammatory processes in vivo by PET.

This study aimed at evaluating the feasibility of the new68Ga-labeled 1,4,7,10-tetraazacyclododecane-N′,N″,N′′′,N″″-tetraacetic acid conjugated synthetic peptide (68Ga-DOTAVAP-P1) targeted to VAP-1/SSAO for PET imagingof early inflammatory and infectious processes in healingbones.

Materials and methods

Animals

Sixty-eight adult male Sprague–Dawley rats (Harlan, Horst,The Netherlands) weighing a mean of 434 g (SD 61 g) wereused. In addition, ten healthy non-operated rats were usedfor harvesting of control bone samples for histology. Beforesurgery, the rats were acclimated to their new environmentand fed a standard laboratory diet. The local AnimalWelfare Committee and the Provincial State Office ofWestern Finland approved the study protocol (Reg. No.1436/04 and 1539/05).

In each animal, the left tibia was operated and the rightcontralateral tibia served as the intact control. The studyconsisted of two parts. First, time series analyses of PETuptake using 68Ga-DOTAVAP-P1 as a tracer were per-formed at 12, 24, and 36 h and 7, 14, and 28 days aftersurgery. At each time point, two osteomyelitic and twocontrol animals with normal bone healing were examined.Based on the results of the time series analysis, two timepoints were chosen for the statistical comparison of PETuptake between the two groups. In the second part of thestudy, PET imaging with 68Ga-DOTAVAP-P1 was per-formed 24 h and 7 days after surgery in animals withinduced osteomyelitis (n=10 for each time point) andanimals with normal bone healing (n=10 for each timepoint). Peripheral quantitative computed tomography(pQCT) and radiography were performed after PETimaging, followed by harvest of samples for bacteriologicalanalysis and histology.

Staphylococcus aureus (strain 52/52A/80, kindly pro-vided by Dr. Jon T. Mader) was used as the pathogen.Originally, the strain was isolated from a child withosteomyelitis, and it has been used extensively inexperimental models of osteomyelitis [18]. Bacterial cellswere suspended in the sterile saline until the final opticaldensity of 0.18 was achieved, as shown by measurementof the absorbance at 600 nm using a spectrophotometer(Smart Spec 3000, Bio-Rad, USA). Based on the opticaldensity, a suspension containing approximately 3×108

colony-forming units (CFU)/ml of S. aureus was used asthe inoculum. The bacterial suspension was stored at 4°Cand used on the day of preparation. To confirm the actualnumber of bacteria in the suspension, a tenfold dilutionseries was prepared, and 100 μl of each dilution wasplated on blood agar plates for the calculation of CFU/ml.

The diffuse rat osteomyelitis model (stage IVA in theCierny–Mader classification; osteomyelitis secondary to acontiguous focus of infection in the Waldvogel classifica-tion) was adopted [19, 20]. The rats were anesthetized

Eur J Nucl Med Mol Imaging (2008) 35:352–364 353

with a mixture of fentanyl plus fluanisone (Hypnorm,Janssen Pharmaceutica, Beerse, Belgium) and midazolam(Midazolam Hameln 5 mg/ml, Hameln Pharmaceuticals,Hameln, Germany). Left hind leg was shaved, disinfected,and covered with sterile sheets. Using sterile surgicalconditions, a small cortical bone defect (diameter 1.0 mm)was created in the proximal medial metaphysis of the righttibia using a high-speed dental drill. Local bone marrowwas removed with saline lavage. As described earlier,osteomyelitis was induced by administering a volume of0.05 ml of 5% w/v sodium morrhuate (Scleromate,Glenwood, Englewood, NJ, USA), and immediately afterthis, the bacterial inoculum (3×108 CFU/ml of S. aureus)was injected in a 0.05-ml volume into the medullarycavity. The drilling hole was sealed with bone wax (Braun,Aesculap AG & Co., Tuttlingen, Germany) to preventbacterial leakage and also to provide a foreign body forinfection [21]. Finally, the skin wound was cleaned withsterile 40-ml saline lavage without antibiotics and closedin layers.

In control animals, a cortical defect of equal size wasdrilled, but neither sodium morrhuate, bacterial suspension,nor bone wax was used. Before wound closure, the surgicalfield was lavaged with 40-ml sterile saline containing150 mg of cefuroxime sodium (Zinazef, GlaxoSmithKline,Verona, Italy). Bacitracin and neomycin sulphate powder(Bacibact, Orion-yhtymä Oyj, Espoo, Finland) was appliedon the sutured skin wound and covered with an aerosol-based plastic film (HansaPlast, Beiersdorf AG, Hamburg,Germany).

Anesthesia was reversed by a subcutaneous injection ofnaloxone (Narcanti, Du Pont Pharmaceuticals, UK). Aftersurgery, the animals were closely monitored, and standardpostoperative pain medication was given for the first threepostoperative days (0.5 mg/kg of buphrenorphin subcuta-neously every 12 h, Temgesic® 0.3 mg/ml, Schering-Plough, Brussels, Belgium). The animals were housedindividually for 2 days, after which they were returned totheir normal housing in groups of two. All the animals haduneventful postoperative recovery and survived the follow-up period.

PET imaging

Linear nine amino acid peptide (GGGGKGGGG) con-taining DOTA-chelate was synthesized and labeled with68Ga (68Ga-DOTAVAP-P1). Briefly, a DOTA moiety wasadded to the N-terminal amino acid of 9-mer by directcoupling of DOTA-tris(tert-butylester) (Macrocyclics,Dallas, TX) in an automated peptide synthesizer with astandard Fmoc solid-phase peptide synthesis protocol. The

peptide was purified using high-performance liquid chro-matography (HPLC) with a reversed phase column, andthe identity of the peptide was verified with a PE SCIEXAPI 150EX mass spectrometer equipped with a Turbo IonSpray ionization source (Perkin-Elmer Sciex instruments,Toronto, Canada). 68Ga was obtained in the form of68GaCl3 from a 68Ge/68Ga generator (Cyclotron, Obninsk,Russia) by elution with 0.1 M HCl. 68GaCl3 eluate(0.5 ml) was mixed with sodium acetate (18 mg; SigmaAldrich) to give a pH of approximately 5.5. Then, DOTA-peptide (30 nmol) was added, and the mixture wasincubated at 100°C for 20 min. No further purificationwas needed. The radiochemical purity was determined byreversed phase radio-HPLC.

PET imaging was performed with an Advance PETscanner (General Electric Medical Systems, Milwaukee,WI, USA) operated in two-dimensional mode (highresolution). The scanner has 18 rings of bismuth germa-nate detectors, and the axial length of the imaging field ofview (FOV) is 152 mm. Dynamic acquisition consistingof 4×5-min frames was started 40 min after the injectionof 68Ga-DOTAVAP-P1. A 5-min transmission scan forattenuation correction was obtained after the emissionimaging using two rod sources containing germanium-68.All 35 transaxial image slices were reconstructed with anordered subsets expectation maximization algorithm (OS-EM), and the central 200-mm-diameter transaxial FOVwas used. The image pixel size was 1.56×1.56 mm in a128×128 matrix. Random counts and dead time werecorrected by the system, and scatter correction wasincorporated into the reconstruction algorithm.

The animals fasted for 4 h before tracer injection. ForPET imaging, the animals were anesthetized by subcutane-ous injection of midazolam and fluanisone plus fentanylcitrate. Mean dose of 25.21 MBq (SD 5.37 MBq; 29.10 μg[SD 6.87 μg]) of 68Ga-DOTAVAP-P1 was injected in thetail vein of the animal in a volume of 0.5–1.0 ml usingsaline as a vehicle.

Quantitative analysis of PET tracer uptake was per-formed on standardized circular regions of interest (ROIs,diameter 3.0 mm) in the operated left and intact right tibiasusing transaxial slices. The size of the ROI was chosen torepresent the minimum diameter of healthy rat tibia. On thetransaxial images, the ROIs included the whole tibial boneand intramedullary cavity. Analysis of PET images wascarried out unaware of other imaging results. The pQCTimages of each tibia were used as the reference for theconstant anatomical positioning of the ROIs. The traceraccumulation was expressed as standardized uptake value(SUV), i.e., [(average radioactivity within the ROI)/(injected dose/rat body weight)]. The SUV ratios between

354 Eur J Nucl Med Mol Imaging (2008) 35:352–364

the operated and non-operated sides were calculated andused for inter-group comparison [3].

Peripheral quantitative computed tomography

Immediately after PET imaging, each animal underwentpQCT scanning. Under fentanyl–fluanisone sedation, thehind limbs were placed in a holder for standard positioning.Imaging was performed using a Stratec XCT Research MpQCT device with software version 5.20 (Norland StratecMedizintechnik, Birkenfeld, Germany). After an initialscout view, the proximal tibias were imaged with sixconsecutive cross-sectional images using a slice distance of0.50 mm. A voxel size of 0.07×0.07×0.50 mm3 was used.The density (mg/cm3) and area (mm2) of trabecular andcortical bone were measured at the cross-sectional plane atthe midlevel of the bone defect in the operated tibia and atthe corresponding plane of the contralateral intact tibia. Thedata analysis was based on intra-animal comparison ofpQCT values between the two tibias. In inter-groupcomparison, the pQCT values of the osteomyelitic tibiawere compared with those of the operated tibia in thecontrol group.

Radiography

The animals were killed after pQCT with an overdose ofsodium pentobarbital (Mebunat®, Orion, Finland). Anterior–posterior and lateral radiographs of the operated limbs weretaken on digital image plates (Fuji IP cassette, type C,Fuji Photo Film, Japan) using a standard stationary X-ray unit (Philips, The Netherlands). The radiographicpresence of osteomyelitic bone changes was classifiedaccording to the Rissing osteomyelitis system [22]. Twoindependent observers evaluated the radiographs withrespect to four parameters (presence of periosteal eleva-tion, architectural distortion, widening of the bone shaft,and new bone formation), and their consensual interpre-tation was used for data analysis. The radiographicpresence of destructive bone changes in the lateral viewof the tibia was quantified using a computer-based imageanalysis system (LabVIEW 6.1, National Instruments,Austin, TX, USA). The area of bone destruction wasmeasured and expressed as a percentage of the total areaof the tibia on the lateral view. The average value of twoindependent observers was used in data analysis.

Microbiological analysis and histology

Using sterile techniques, the bone defect area was exposed,and swab cultures were taken from the subfascial softtissues. The swab specimens were cultured for 20 h at 35°Cin blood agar plates. The tibia was harvested and aseptically

cross-sectioned into three segments (proximal, middle, anddistal) with a high-speed circular saw under sterile salinecooling. The distal bone segment was sent for quantitativebacterial analysis to confirm the presence of osteomyelitis.After snap freezing in liquid nitrogen and homogenizationwith mortal and pestle, bone chips were vortexed in salinefor 5 min, and serial (tenfold) dilutions were withdrawn.The samples were analyzed for the determination of CFU ofS. aureus per gram of bone. The Slidex Staph Plus latexagglutination test (bioMérieux, Marcy l’Etoile, France), asdescribed by van Griethuysen et al. [23], was used for theidentification of isolated pathogens. S. aureus (ATCC29213) was used as the positive control and Enterococcusfaecalis (ATCC 29212) as the negative control. Pulse-fieldgel electrophoresis (PFGE) was used to confirm that theisolated strain of S. aureus was identical to the inoculatedstrain (52/52A/80).

For histopathological examination, the proximal bonespecimens, representing the top level of bone defect, werefixed in 70% ethanol, embedded in isobornylmethacrylate(Technovit 1200 VLC, Kulzer, Germany) and stained with amodified van Gieson method. The middle bone segment wasdecalcified, embedded in paraffin, and stained with hematox-ylin and eosin for scoring the presence of inflammatory cells.Stage of inflammation was semi-quantitatively scored from −(score 0) to +++ (score 3), based on the extent ofinflammatory cell infiltrates in the bonemarrow. Furthermore,in the second part of the study, the level of VAP-1 expressionat the site of infection or healing bone was determined fromdecalcified, paraffin-embedded bone tissue samples as de-scribed below.

Osteomyelitic changes in the periosteum, cortical bone,and medullary canal were classified according to thehistopathological scoring system presented by Petty et al.[24]. Two independent observers classified the histologicalsections, and the results were presented as the averagevalues of their interpretation.

The polyclonal rabbit antiserum used in immunoperox-idase stainings was raised against recombinant human VAP-1, but it also recognizes well rat VAP-1. Staining of paraffinsections was done using microwave treatment in a citratebuffer for antigen retrieval and subsequent staining with theavidin-peroxidase method using the Vectastain kit accord-ing to the manufacturer’s instructions. All samples werestained with rabbit anti-rat antibody and normal rabbitserum as a negative control. The sections were counter-stained with hematoxylin–eosin.

The immunoperoxidase staining results were semi-quanti-tatively analyzed in a blinded manner using the followingscoring: − (score 0), no positivity; + (score 1), faint positivitywith a few positive vessels detected; +++ (score 3), strongpositivity of many vessels. The rating of ++ (score 2), wasgiven to samples falling between categories + and +++.


Statistical analyses

All the results are expressed as means (SD). The signifi-cance of differences in SUV ratios, pQCT values, and bonedestruction areas between osteomyelitic bones and healingbones were estimated using linear models. The intra-animalcomparison of SUVs and pQCT values between theoperated and non-operated sides was estimated usingrepeated measures analysis of variance. In non-parametricanalyses, Fisher’s exact test was used for the evaluation ofdifferences in radiographic scores, and the Mann–WhitneyU-test was used for the evaluation of differences inhistological scores. A p value less than 0.05 was consideredsignificant. All statistical analyses were conducted usingSAS 9.1.3 statistical software (SAS Institute, Cary, NC,USA).

Results

68Ga-DOTAVAP-P1 PET differentiates bone infectionfrom normal bone healing as soon as 7 days after onset

The radiochemical purity of 68Ga-DOTAVAP-P1 was >95%throughout the study as analyzed by radio-HPLC, and thespecific activity was 940±288 MBq/mg.

Results of the initial time series analysis are presented inFig. 1. Uptake of 68Ga-DOTAVAP-P1 was seen in bothosteomyelitic (bacterial infection) and control (healing bonedefect, i.e., sterile inflammation) tibias during the first 36 hafter surgery. Thereafter, only the osteomyelitic tibias weredelineated by PET (Fig. 1a).

In radiography, animals with induced osteomyelitisshowed progressive increase in the percentual area of bonedestruction. After 7 days, the infected bones showed

Fig. 1. Time-dependent changesin the models used. The linesrepresent the means (n=2/group/time point). a Development ofSUV ratio, calculated as thedifference in the measured ac-tivity between the operated andthe intact contralateral tibia.b, c Time-related changes of thepQCT measurements in corticalbone density (b) and in corticalbone area (c). d Quantitativebacterial cultures from osteo-myelitic animals, represented asCFU per gram of bone. e Totalhistological infection gradeevaluated from cross-sectionalhistological samples (modifiedvan Gieson stain)


periosteal reaction, architectural distortion, widening of thebone shaft, and, at 14 days, new bone formation. Incontrast, animals with healing bone defects showed onlydiminution in defect size, representing cortical bonehealing.

In pQCT imaging, osteomyelitic animals and animalswith healing bone defects did not show any difference inthe percentual comparison of operated tibia to contralateralintact tibia until 36 h after operation. Thereafter, osteomye-litic bones showed progressive decrease in cortical bonedensity and increase in cortical bone area up to 14 days. At28 days, a chronic state of infection resulted in massive newbone formation (Fig. 1b,c).

The inoculated pathogen was cultured from the homog-enized bone specimens in all animals with inducedosteomyelitis. No bacteria could be cultured from homog-enized bone specimens retrieved from the control animals(Fig. 1d). Infection could be confirmed histologically fromvan Gieson stains after 36 h from operation. Animals withhealing bone defects showed no signs of infection (Fig. 1e).

PET imaging with 68Ga-DOTAVAP-P1 showed uptakein healing bones at 24 h, but not at 7 days

Based on the results of the time-series analysis, two timepoints were chosen for the statistical comparison of PETuptake between the two groups. In this pivotal experiment,ten osteomyelitic animals and ten animals with normal bonehealing were examined by PET at 24 h and 7 days aftersurgery. pQCT and radiography were performed after PETimaging, followed by harvest of samples for bacteriologicalanalysis and histology.

PET imaging showed accumulation of 68Ga-DOTAVAP-P1 in both the osteomyelitic (bacterial infection) andcontrol (healing bone defect, i.e., sterile inflammation)tibias, as compared with the contralateral intact tibia, at24 h after surgery (Fig. 2a,c). In contrast, at 7 days, only theosteomyelitic tibias showed accumulation of the tracer, ascompared with the contralateral intact tibia (Fig. 2b,d). Themean SUVs of the infected bone area in osteomyelitic

animals were 1.17 (SD 0.74) and 2.28 (SD 0.93) at 24 hand 7 days after surgery, respectively (Fig. 3a). The meanSUVs of the operated bone area in control animals were1.37 (SD 0.70) and 1.08 (SD 0.54) at 24 h and 7 days aftersurgery, respectively (Fig. 3a).

In osteomyelitic animals, the mean SUV ratios were 1.50(SD 0.16) and 1.87 (SD 0.41) at 24 h and 7 days aftersurgery, respectively (Fig. 3b). The observed increase inSUV ratios between the two time points was significant(p=0.0016). In the control animals with healing bonedefects, there was a significant decrease in the uptakebetween the two time points (p = 0.0008). Thecorresponding mean SUV ratios at 24 h and 7 days aftersurgery were 1.48 (SD 0.18) and 1.09 (SD 0.08),respectively (Fig. 3b). The SUV ratios of osteomyeliticanimals imaged 7 days postoperatively differed significant-ly (p<0.0001) from those of the corresponding controlanimals with healing bone defects (Fig. 3b). In contrast, thedifference in SUV ratios between osteomyelitic and controlanimals imaged 24 h postoperatively was not statisticallysignificant (p=0.8020; Fig. 3b).

In radiography, animals imaged at 24 h showed no signsof infection or healing of cortical bone on plain radiographs(Fig. 4a,c). In the osteomyelitic group, plain radiographs at7 days showed localized osteomyelitis with moderate bonedestruction (Fig. 4b). Animals with cortical bone defectsshowed uncomplicated bone healing by endosteal callusformation at 7 days after surgery (Fig. 4d). At 7 days, themean area of bone destruction was 19.8% in the osteomye-litic animals.

pQCT was applied to characterize the time-relatedchanges in the cortical and trabecular bone density andarea. In the osteomyelitic group, pQCT imaging demon-strated signs of localized osteomyelitis with moderate bonedestruction already at 7 days (Fig. 5a,b). The controlanimals showed endosteal callus formation at the defect sitewithin 7 days after surgery (Fig. 5c,d). The pQCT revealedminor cortical bone changes in animals at 24 h postoper-atively (Fig. 6). The major changes occurred in the corticalbone of the osteomyelitic animals at 7 days (Fig. 6). The

Fig. 2. Representative transaxialPET images with 68Ga-DOTAVAP-P1 of the hind legs atthe site of induced infection(a, b) or healing cortical bonedefect (c, d) at 24 h (a, c) and7 days (b, d) postoperatively.The operated left tibia is markedwith a red arrow and the con-tralateral intact right tibia witha white arrow


cortical bone density decreased by 27% (p<0.0001) inthe infected tibia, as compared with the contralateral tibia.The cortical bone area increased by 53% (p<0.0001) in theinfected tibia, as compared with the contralateral tibia(Fig. 6). These changes in the density and area of corticalbone were also significant as compared with the operatedtibia of the control group (for both, p<0.0001; Fig. 6). The

two sides of the control group showed minor (±7%)differences in both the cortical bone density and the corticalbone area (Fig. 6). The density of trabecular bone waslower in the infected and healing tibias than in thecontralateral tibia (p=0.0117, p=0.0023, respectively).The area of the trabecular bone did not show a similardifference.

The inoculated pathogen was cultured from the homog-enized bone specimens in all animals with inducedosteomyelitis. According to PFGE analysis, the strain(S. aureus, strain 52/52A/80) of the isolated pathogenmatched the inoculated one. The mean quantity of CFU pergram of bone was 2.8×107 (SD 3.0×107) and 1.3×109 (SD

Fig. 4. Lateral X-ray images of osteomyelitic (a, b) and control (c, d)rat tibias at 24 h (a, c) and 7 days (b, d) postoperatively. At 24 h,neither closure of the cortical defect, signs of infection nor visualdifference can be seen between the study groups. In the osteomyeliticanimals killed at 7 days, more prominent periosteal reaction,architectural distortion, and, in some animals, also widening of thebone shaft is observed. The animals with normally healing tibia showsan uneventful healing of the cortical defect

Fig. 3. 68Ga-DOTAVAP-P1 PET imaging. The bars represent themeans (±SD; n=10). a Activity of 68Ga-DOTAVAP-P1 at the level ofsurgery expressed as SUV values. The animals imaged 24 h afteroperation show an increased tracer uptake, as compared withcontralateral intact tibia, in both study groups. In the osteomyelitictibias imaged 7 days postoperatively, a focally high tracer accumula-tion at the site of infection is seen in comparison with the intact righttibia. In contrast, in animals with healing bone defects imaged at7 days postoperatively, no statistically significant difference in uptakeis observed between the operated and the intact contralateral legs.b 68Ga-DOTAVAP-P1 SUV ratios between the operated and thecontralateral intact tibia. In animals with sterile inflammation, theregion of uncomplicated bone healing showed a marked decrease ofthe tracer activity by 7 days. In contrast, in the animals with inducedbacterial infection, the osteomyelitic region showed a significantincrease of the peptide accumulation by 7 days. There was asignificant difference in the SUV ratio between the osteomyelitic andthe control animals already at 7 days after surgery


3.8×109) in animals killed 24 h or 7 days after surgery,respectively. In 27 osteomyelitic animals out of 32 (84%),the swab cultures taken from subfascial soft tissues werepositive for the inoculated S. aureus, indicating theextension of infection outside the bone, but only threeanimals had visible soft tissue abscesses. None of theanimals had an infected draining sinus. No bacteria couldbe cultured from homogenized bone specimens retrievedfrom the control animals. Similarly, all swab cultures fromthe soft tissues of control animals were negative.

The infected animals killed 24 h after operation showedminor osteomyelitic changes in histology. The meaninfection grade at this time point, evaluated by the methoddescribed by Petty et al. [24], was 0.45 (SD 0.22). Theosteomyelitic animals killed at 7 days showed histologicallya severe osteomyelitis of the tibia in all cases (mean 2.68,SD 0.61). There was a significant difference in the meanhistological score between the two time points (p<0.0001),reflecting the progression of infection over time. Thecontrol group showed healing of the defect by endostealnew bone formation, with no signs of infection.

In osteomyelitic animals, the mean grade of inflamma-tory cells was 2.61 (SD 0.49) and 2.35 (SD 0.41) at 24 h or7 days, respectively (Fig. 7a,b). The corresponding mean

inflammatory cell grade in animals with healing bonedefects at 24 h and 7 days after surgery was 1.25 (SD0.35) and 1.20 (SD 0.42), respectively (Fig. 7c,d). Thedifference in the number of inflammatory cells in hema-toxylin and eosin histological stains between osteomyeliticand control animals was statistically significant at each timepoint; p=0.0002 and p=0.0003 at 24 h and 7 days,respectively (Fig. 7e). No inflammatory cell infiltrate couldbe detected in samples retrieved from healthy non-operatedcontrols.

The mean grade of VAP-1 expression in osteomyeliticanimals was 2.39 (SD 0.49) and 2.85 (SD 0.34) at 24 h and7 days, respectively (Fig. 8). The increase in the level ofVAP-1 expression in osteomyelitic animals between the twotime points was statistically significant (p=0.0330). Themean grade of VAP-1 expression at the site of healing bonearea was 1.10 (SD 0.39) and 1.00 (SD 0.47) at 24 h and7 days, respectively (Fig. 8). The expression levels inosteomyelitic animals differed significantly from those ofthe corresponding control animals with healing bonedefects at both time points (for both, p<0.0001; Fig. 8e).

Discussion

Molecular imaging is revolutionizing basic research onskeletal tissues [25]. Combined with PET imaging, molec-ular imaging provides us with means to understand thebasic mechanisms of tissue repair. Many of the signals thatinitiate and dictate the pattern of bone healing are part ofthe initial inflammatory process [26]. The role of pro-inflammatory cytokines in controlling cellular response ofhealing bones has become fully appreciated [27]. The stageof inflammatory response after any bone injury has beenestimated to last approximately 48 h.

This experimental study was designed to evaluate theapplicability of 68Ga-DOTAVAP-P1, a new PET radiophar-maceutical, for the assessment of physiological inflammationprocess in bone healing and osteomyelitis in standardizedanimal models. Of particular interest for clinical significancewas to determine whether and how soon after operation thedifferent processes (an infectious process due to microbialcontamination and inflammatory reactions associated withthe early phases of bone healing) could be differentiated.

VAP-1/SSAO is an inflammation-inducible dual functionendothelial glycoprotein with adhesion properties and amineoxidase activity. The VAP-1/SSAO molecule is expressedboth in membrane-associated and in soluble form circulatingin blood stream. It plays a critical role in cellular traffickingrecruiting lymphocytes, CD8+ T lymphocytes in particular,from blood into lymphoid organs and inflamed tissues [14].In inflamed tissues, VAP-1 is rapidly translocated to the cellsurface, whereas it is absent from the endothelial cell surface

Fig. 5. Cross-sectional pQCT images of osteomyelitic (a, b) andcontrol (c, d) rat tibias at 24 h (a, c) and 7 days (b, d) postoperatively.In animals killed 24 h after surgery, neither closure of the corticaldefect (large arrow) nor visual difference can be seen between thestudy groups. In the osteomyelitic animals killed 7 days postopera-tively, there is cortical bone destruction with circumferential periostealreaction (small arrows), reactive new bone and sequestrum formation(large arrow). In the control animals, a small amount of endosteal newbone formation can be seen, associated with uneventful healing of thecortical defect (large arrow)


of normal tissues. Expression of VAP-1 is induced onendothelial cells in man at sites of inflammation [28],including dermal blood vessels in various skin inflammatorydiseases (e.g., psoriasis and atopic eczema), gut bloodvessels in inflammatory bowel disease (Crohn’s disease andulcerative colitis), and synovial blood vessels in inflamedjoints (e.g., patients with rheumatoid arthritis). In endothelialcells, the enzyme activity of VAP-1 can directly regulatelymphocyte rolling [29]. Animal studies have shown that,during inflammation, VAP-1 is rapidly translocated to thecell surface from intracellular sources [30]. According to ourresults, a significant difference was observed in the immuno-histological evaluation of VAP-1 expression between theosteomyelitic and control animals already 24 h after surgery.In contrast, such difference was not seen in PET imaging. In

PET imaging, the used tracer is able to bind only to theluminal VAP-1. This finding suggests that the amount ofVAP-1 translocated onto the luminal surface is almost equalin both groups of rats at this time point. Furthermore,leukocytes enter the adjacent tissues partly via a VAP-1/SSAO-mediated mechanism, and its activity is a key sourcefor hydrogen peroxide generation locally at sites ofleukocyte entry in tissue [31, 32]. A considerable body ofscientific data supports the theory that VAP-1 is an adhesionmolecule critically involved in the process of leukocyteadhesion, and that inhibiting its activity with monoclonalantibodies leads to a significant decrease in inflammatoryresponse [32].

Previous studies by Yegutkin et al. [33] have revealedthat the linear GGGGKGGGG peptide specifically and

Fig. 6. pQCT density (a, c) and area (b, d) of the cortical (a, b) andtrabecular (c, d) bone at the defect area on the proximal part of rattibia at 24 h and 7 days. The bars represent the means (±SD; n=10) ofthe density and area of the cortical bone (expressed as mg/cm3 andmm2, respectively). In animals imaged 24 h postoperatively, asignificant decrease in cortical bone density was seen in bothosteomyelitic and control animals, as compared with the contralateralintact tibia. Furthermore, cortical bone area was decreased in controlanimals at 24 h. Bone infection caused a significant reduction of thecortical bone density in osteomyelitic animals imaged 7 days

postoperatively. In contrast, osteomyelitis resulted in a significantincrease in cortical bone area imaged at 7 days. In control animalsimaged at 7 days postoperatively, a significant, but modest, decreasein cortical bone density and an increase in cortical bone area wereobserved. The values of the density and area of the osteomyeliticanimals differed significantly from those of the corresponding controlanimals, at both time points. The density of trabecular bone wasdecreased in the infected and healing tibia as compared with thecontralateral tibia at 7 days. The area of the trabecular bone did notshow a similar difference


dose-dependently inhibits the VAP-1/SSAO enzyme activ-ity by binding to the corresponding groove of the VAP-1molecule. Based on the fact that the luminal VAP-1 isdetected only at the scene of inflammation, we hypothe-sized that this peptide could be selectively targeted to theinflammatory foci. Indeed, a study performed by Pöyhönenet al. (unpublished data) showed that the peptide, afterbeing conjugated with DOTA and labeled with 68Ga, is stillable to inhibit SSAO activity. Furthermore, we for the firsttime showed that intravenously administered 68Ga-DOTA-VAP-P1 delineates infection/inflammation foci in vivo byPET imaging.

In addition to PET, we used pQCT and conventionalradiography for the evaluation of bone changes. CT is auseful tool for the evaluation of deep bone infections and afeasible alternative to magnetic resonance imaging (MRI)in the assessment of osteomyelitic changes [34]. In thepresent study, pQCT was applied to determine the area anddensity of both cortical and trabecular bone and also for theevaluation of anatomical bone changes. According topQCT, drastic infectious changes were seen especially inthe area and density of cortical bone.

For the present study, the rat model of diffuse tibialosteomyelitis due to S. aureus was adopted. This model is a

Fig. 7. Histologicalhematoxylin and eosin stainsof osteomyelitic animals(a, b) and animals with uncom-plicatedly healing tibias(c, d) killed at 24 h (a, c) or7 days (b, d) postoperatively. Inanimals with healing tibias, theinflammatory change was char-acterized by a modest infiltra-tion of polymorphonuclearleukocytes, at 24 h and 7 days.However, the osteomyelitic ani-mals killed already 24 hpostoperatively showed in-creased infiltration of polymor-phonuclear leukocytes withoccasional formation of micro-abscesses. The osteomyeliticanimals killed 7 days postoper-atively showed drastic changesassociated with the progressionof infection. The differencein the mean grade of inflamma-tory cells was statistically sig-nificant between the two groupsat 7 days (e)


widely accepted experimental model for the research ofbone infections and also for the research of novel PETtracers [8, 19, 20, 35]. The applied experimental model hasimportant benefits. Standardized rat tibial infection isrelatively easy to produce. Quantitative bacteriology con-firmed infection in all osteomyelitic animals in our study.Induced infection is primarily localized in the medullaryarea and its adjacent bone, thus minimizing the impact onthe affect for the general well-being of the animal.

Despite technological advances, PET is characterized bya relatively low spatial resolution. In this study, we used aclinical PET scanner (GE Advance) for the imaging of ratbecause, at the time of the study, we did not have access to

a dedicated small animal PET scanner. The animal wasplaced in the center of the scanner gantry giving the highestpossible spatial resolution, i.e., 4.8 mm. Unfortunately, thestructures we were aiming at were quite small (10 mm indiameter), which might result in partial-volume effect(PVE) and possibly invalidate the in vivo quantitativePET data. PVE means that the apparent pixel values in PETimages are influenced by the surrounding high pixel values[36]. Other factors, such as reconstruction algorithm andfilter, scanner sensitivity, and scan duration, can also affectthe accuracy of the measured radioactivity concentration inan object [37]. In the present study, the central 200-mm-diameter transaxial FOV were reconstructed using OS-EM

Fig. 8. Representative VAP-1immunohistological sections(a, c) and corresponding nega-tive control stains (b, d) showedslight staining of blood vesselwalls at grade 1 (a) and moreprominent staining at grade 3(c). There was a statisticallysignificant difference in theVAP-1 expression between theinfected and the healing bonesalready at 24 h (e)


algorithm with 2 iterations and 28 subsets, without Z-axisfilter, and using segmented attenuation correction [38].Using 128×128 matrix, this leads to 1.56×1.56 mm pixelsize, which is approximately one third of the imageresolution. Boellaard et al. [39] have suggested that betterimage quality may be obtained by using OS-EM instead offiltered back-projection, resulting in fewer PVE. PVEcorrection is theoretically possible if high-resolution struc-tural imaging, such as MRI, can be coupled with theknowledge of the scanner resolution. However, the lack ofanatomic MRI of the rats prevented PVE correction in thisstudy. To minimize the PVE, we used the averageradioactive concentration within the ROI (kBq/ml) andSUV ratios to measure the tracer uptake in infected bone.To minimize the PVE, we used the average radioactiveconcentration within the ROI (kBq/ml). After a fracture, thehealing involves processes both in the bone itself and alsoin the surrounding soft tissues. High-resolution PET wouldallow us to accurately discriminate periosteal inflammationfrom that of surrounding soft tissues.

From the clinical point of view, the current study mighthave benefited from MRI, especially for the evaluation ofthe extent of infection spread to soft tissues. For successfulinduction of deep bone infection, the applied animal modelrequired a large bacterial inoculate injected directly to themedullary cavity [20]. S. aureus is the most frequentpathogen causing any clinical type of bone infection and soa natural choice for a preclinical study of osteomyelitis.

To conclude, the current study showed that PET imagingwith the new 68Ga-DOTAVAP-P1 is capable of accuratelydemonstrating the phase of inflammation in healing bonesand the progress of bacterial infection in osteomyeliticbones. The tracer does not specifically accumulate toinfectious sites, but rather describes the rate of inflamma-tory or infectious reaction. It is anticipated that this tracermay not be able to differentiate aseptic loosening frominfection of the joint prosthesis. Consequently, this novelimaging agent allowed for the differentiation of boneinfection due to S. aureus and normal bone healing as soonas 7 days after surgery. This imaging method could beapplied to delineate the effects of anti-inflammatory agentson bone healing, as well as in the evaluation of treatmentresponse in antimicrobial studies of bone infection. Fromthe clinical point of view, this new PET tracer could bevaluable for the detection of infection at its early stages.

Acknowledgments This work was funded by grants from theNational Technology Agency of Finland (TEKES), the Academy ofFinland (grants No. 205757 and No. 103032), the InstrumentariumFoundation, the Turku University Foundation, the Walter and LisiWahl Foundation, and the Finnish Cultural Foundation. PetteriLankinen is a Ph.D. student supported by the Finnish GraduateSchool for Musculoskeletal Diseases. Tiina Pöyhönen is a Ph.D.student supported by the Drug Discovery Graduate School of the

University of Turku. The authors acknowledge Anni Virolainen-Julkunen, M.D., Ph.D., for conducting the PFGE analysis, and JouniAlin, M.Sc., for statistical consultation.

References

1. Guhlmann A, Brecht-Krauss D, Suger G, Glatting G, Kotzerke J,Kinzl L, et al. Fluorine-18-FDG PET and technetium-99mantigranulocyte antibody scintigraphy in chronic osteomyelitis.J Nucl Med 1998;39:2145–52.

2. Kalicke T, Schmitz A, Risse JH, Arens S, Keller E, Hansis M, etal. Fluorine-18 fluorodeoxyglucose PET in infectious bonediseases: results of histologically confirmed cases. Eur J NuclMed 2000;27:524–8.

3. de Winter F, van de Wiele C, Vogelaers D, de Smet K, Verdonk R,Dierckx RA. Fluorine-18 fluorodeoxyglucose-position emissiontomography: a highly accurate imaging modality for the diagnosisof chronic musculoskeletal infections. J Bone Joint Surg Am2001;83-A:651–60.

4. De Winter F, Vogelaers D, Gemmel F, Dierckx RA. Promisingrole of 18-F-fluoro-D-deoxyglucose positron emission tomogra-phy in clinical infectious diseases. Eur J Clin Microbiol Infect Dis2002;21:247–57.

5. Koort JK, Mäkinen TJ, Knuuti J, Jalava J, Aro HT. Comparative18F-FDG PET of experimental Staphylococcus aureus osteomye-litis and normal bone healing. J Nucl Med 2004;45:1406–11.

6. Einhorn TA. The cell and molecular biology of fracture healing.Clin Orthop Relat Res 1998;355 Suppl:S7–21.

7. Kon T, Cho TJ, Aizawa T, Yamazaki M, Nooh N, Graves D, et al.Expression of osteoprotegerin, receptor activator of NF-kappaBligand (osteoprotegerin ligand) and related proinflammatory cyto-kines during fracture healing. J Bone Miner Res 2001;16:1004–14.

8. Mäkinen TJ, Lankinen P, Pöyhönen T, Jalava J, Aro HT,Roivainen A. Comparison of 18F-FDG and 68Ga PET imagingin the assessment of experimental osteomyelitis due to Staphylo-coccus aureus. Eur J Nucl Med Mol Imaging 2005;32:1259–68.

9. Rennen HJ, Boerman OC, Oyen WJ, Corstens FH. Imaginginfection/inflammation in the new millennium. Eur J Nucl Med2001;28:241–52.

10. Chianelli M, Mather SJ, Martin-Comin J, Signore A. Radio-pharmaceuticals for the study of inflammatory processes: areview. Nucl Med Commun 1997;18:437–55.

11. Corstens FH, van der Meer JW. Nuclear medicine’s role ininfection and inflammation. Lancet 1999;354:765–70.

12. Signore A, Chianelli M, Bei R, Oyen W, Modesti A. Targetingcytokine/chemokine receptors: a challenge for molecular nuclearmedicine. Eur J Nucl Med Mol Imaging 2003;30:149–56.

13. Salmi M, Jalkanen S. A 90-kilodalton endothelial cell moleculemediating lymphocyte binding in humans. Science 1992;257:1407–9.

14. Smith DJ, Salmi M, Bono P, Hellman J, Leu T, Jalkanen S.Cloning of vascular adhesion protein 1 reveals a novel multifunc-tional adhesion molecule. J Exp Med 1998;188:17–27.

15. Koskinen K, Vainio PJ, Smith DJ, Pihlavisto M, Ylä-Herttuala S,Jalkanen S, et al. Granulocyte transmigration through theendothelium is regulated by the oxidase activity of vascularadhesion protein-1 (VAP-1). Blood 2004;103:3388–95.

16. Salmi M, Kalimo K, Jalkanen S. Induction and function ofvascular adhesion protein-1 at sites of inflammation. J Exp Med1993;178:2255–60.

17. Tohka S, Laukkanen M, Jalkanen S, Salmi M. Vascular adhesionprotein 1 (VAP-1) functions as a molecular brake duringgranulocyte rolling and mediates recruitment in vivo. FASEB J2001;15:373–82.


18. Rissing JP. Animal models of osteomyelitis. Knowledge, hypoth-esis, and speculation. Infect Dis Clin North Am 1990;4:377–90.

19. Mader JT. Animal models of osteomyelitis. Am JMed 1985;78:213–7.20. O’Reilly T, Mader JT. Rat model of bacterial osteomyelitis of the

tibia. In: Zak O, Sande MA, editors. Handbook of animal modelsof infection. Bath, Avon, UK: Academic; 1999. pp. 561–75.

21. Nelson DR, Buxton TB, Luu QN, Rissing JP. The promotionaleffect of bone wax on experimental Staphylococcus aureusosteomyelitis. J Thorac Cardiovasc Surg 1990;99:977–80.

22. Rissing JP, Buxton TB, Weinstein RS, Shockley RK. Model ofexperimental chronic osteomyelitis in rats. Infect Immun1985;47:581–86.

23. van Griethuysen A, Bes M, Etienne J, Zbinden R, Kluytmans J.International multicenter evaluation of latex agglutination tests foridentification of Staphylococcus aureus. J Clin Microbiol2001;39:86–9.

24. Petty W, Spanier S, Shuster JJ, Silverthorne C. The influence ofskeletal implants on incidence of infection. Experiments in acanine model. J Bone Joint Surg 1985;67-A:1236–44.

25. Mayer-Kuckuk P, Boskey AL. Molecular imaging promotesprogress in orthopedic research. Bone 2006;39:965–77.

26. Einhorn TA, Majeska RJ, Rush EB, Levine PM, Horowitz MC.The expression of cytokine activity by fracture callus. J BoneMiner Res 1995;10:1272–81.

27. Gerstenfeld LC, Cullinane DM, Barnes GL, Graves DT, EinhornTA. Fracture healing as a post-natal developmental process:molecular, spatial, and temporal aspects of its regulation. J CellBiochem 2003;88:873–84.

28. Jaakkola K, Nikula T, Holopainen R, Vahasilta T, Matikainen MT,Laukkanen ML, et al. In vivo detection of vascular adhesion protein-1 in experimental inflammation. Am J Pathol 2000;157:463–71.

29. Stolen CM, Marttila-Ichihara F, Koskinen K, Yegutkin GG, TurjaR, Bono P, et al. Absence of the endothelial oxidase AOC3 leadsto abnormal leukocyte traffic in vivo. Immunity 2005;22:105–15.

30. Salmi M, Jalkanen S. VAP-1: an adhesin and an enzyme. TrendsImmunol 2001;22:211–6.

31. Salmi M, Jalkanen S. Cell-surface enzymes in control ofleukocyte trafficking. Nat Rev Immunol 2005;5:760–71.

32. Merinen M, Irjala H, Salmi M, Jaakkola I, Hänninen A, et al.Vascular adhesion protein-1 is involved in both acute andchronic inflammation in the mouse. Am J Pathol 2005;166:793–800.

33. Yegutkin GG, Salminen T, Koskinen K, Kurtis C, McPherson MJ,Jalkanen S, et al. A peptide inhibitor of vascular adhesion protein-1 (VAP-1) blocks leukocyte-endothelium interactions under shearstress. Eur J Immunol 2004;34:2276–85.

34. Ledermann HP, Kaim A, Bongartz G, Steinbrich W. Pitfalls andlimitations of magnetic resonance imaging in chronic posttrau-matic osteomyelitis. Eur Radiol 2000;10:1815–23.

35. Gratz S, Béhé M, Boerman OC, Kunze E, Schulz H, Eiffert H,et al. 99mTc-E-selectin binding peptide for imaging acuteosteomyelitis in a novel rat model. Nucl Med Commun2001;22:1003–13.

36. Hoffman EJ, Huang SC, Phelps ME. Quantitation in positronemission computed tomography: 1. Effect of object size. J ComputAssist Tomogr 1979;3:299–308.

37. Boellaard R, Krak NC, Hoekstra OS, Lammertsma AA. Effects ofnoise, image resolution, and ROI definition on the accuracy ofstandard uptake values: a simulation study. J Nucl Med2004;45:1519–27.

38. Visvikis D, Cheze-LeRest C, Costa DC, Bomanji J, Gacinovic S,Ell PJ. Influence of OSEM and segmented attenuation correctionin the calculation of standardised uptake values for [18F]FDGPET. Eur J Nucl Med 2001;28:1326–35.

39. Boellaard R, van Lingen A, Lammertsma AA. Experimental andclinical evaluation of iterative reconstruction (OSEM) in dynamicPET: quantitative characteristics and effects on kinetic modeling.J Nucl Med 2001;42:808–17.


Documents

68Ga-DOTAVAP-P1 PET imaging capable of demonstrating the phase of inflammation in healing bones and the progress of infection in osteomyelitic bones