ABSTRACT

Positron emission tomography (PET) with F-18 fluorodeoxyglucose (FDG) is an invaluable tool in the diagnostic workup of inflammatory and infectious diseases. FDG, however, is nonspecific, and is concentrated by malignant and benign neoplasm, as well as a variety of noninfectious inflammatory conditions. PET studies have important advantages over studies performed with single photon emitting radiopharmaceuticals and considerable effort has been devoted to the development of PET radiopharmaceuticals that are specific, or at least more specific than FDG, for infection. An early attempt at creating a more specific PET radiopharmaceutical was the development of a procedure for labeling autologous leukocytes with FDG. Although investigations have demonstrated improved specificity of FDG labeled leukocytes compared to FDG alone, with results comparable to those obtained with In-111 labeled leukocytes, there are issues with this test. Variability of labeling efficiency and stability of the FDG label, the short (110 minute) half -life of F-18, and the in vitro labeling procedure itself are obstacles to the widespread adoption of this procedure. The use of Cu-64 as the radiolabel improves labeling efficiency and label stability, in vitro, but no clinical investigations have been conducted. The Cu-64 neutrophil specific peptide, has shown promise in preclinical studies, but there have not been any clinical investigations of this agent. The radioiodinated thymidine analogue fialuridine (FIAU) has been investigated as an infection specific imaging radiopharmaceutical. Preclinical investigations were very promising, and an early pilot study using I-124 FIAU PET/CT yielded very encouraging results. Results of subsequent investigations for diagnosing prosthetic joint infection and diabetic pedal osteomyelitis were disappointing and the future of this agent is uncertain. The role of Ga-68 for imaging infection also is under investigation. Ga-68 citrate overcomes many of the disadvantages of Ga-67, but still suffers from a lack of specificity, and investigators have begun to focus their attention on complexing Ga-68 with siderophores and peptides in attempts to develop a more specific imaging agent. Preliminary results with these agents are very encouraging, but clinical trials are needed before their value can truly be ascertained.

Introduction

Positron emission tomography (PET) with F-18 fluorodeoxyglucose (FDG) has become an invaluable tool in the diagnostic workup of patients with a variety of inflammatory and infectious diseases. In spite of its usefulness, however, FDG is a nonspecific radiopharmaceutical that is concentrated by noninfectious conditions, including malignant and benign neoplasms, and a variety of noninfectious inflammatory conditions.

PET studies have important advantages over studies performed with single photon emitting radiopharmaceuticals. PET resolution is superior, which facilitates precise localization of abnormalities. Semi quantitative analysis, which is readily available with PET, but less feasible with conventional gamma camera imaging, potentially could be used to differentiating infectious from non-infectious conditions and for monitoring response to treatment. It is not surprising, therefore, that a considerable amount of effort has been devoted to the investigation and development of PET radiopharmaceuticals that are specific, or at least more specific than FDG, for infection.

Fluorodeoxyglucose Labeled Leukocytes

Perhaps the earliest attempt at creating a PET radiopharmaceutical more specific than FDG was the development of an in-vitro method for labeling autologous leukocytes. Nearly 25 years ago, Osman and Danpure demonstrated the feasibility of in vitro labeling of human leukocytes with FDG (1). They observed that FDG uptake was dependent on the concentration of glucose in the labeling medium. Labeling efficiency decreased from 80%, when the glucose concentration was 15 ug/mL to 2% when the glucose concentration was 1 mg/mL. Cellular glucose retention depended on the extracellular concentration of glucose. In the absence of extracellular glucose, 91% of the FDG remained in the leukocytes at one hour. When the extracellular glucose concentration was 1 mg/mL, however, only 73% of the FDG remained in the cells at one hour.

Forstrom et al. reported their results for in vitro FDG labeling of human leukocytes (2). The labeling efficiency, approximately 80%, was similar to what Osman and Danpure had reported (1). Eighty percent of the activity was in the granulocyte fraction, 14% in the mixed lymphocyte-plasma fraction and 6% was in plasma. FDG labeled leukocytes were stable in platelet-poor plasma for up to four hours. The trypan blue dye test indicated excellent cell viability after labeling. Forstrom et al. subsequently reported the results of biodistribution and dosimetry studies performed on four normal volunteers (3). The FDG-labeled leukocytes, as expected, were distributed primarily in the reticuloendothelial system. Cerebral and urinary tract activity also were present, however, presumably due to elution of FDG from the leukocytes. Whole body and major organ dosimetry estimates for FDG labeled leukocytes, with administered activity between 222- 252 MBq (6-6.8 mCi) were comparable to the reported results for In-111 labeled leukocytes.

Pellegrino et al. compared the uptakes of FDG labeled leukocytes and FDG in a rodent model of acute inflammation, with sterile turpentine, and bacterial infection, with Escherichia coli and Pseudomonas aeruginosa (4). They found markedly higher uptake of both radiopharmaceuticals in infected muscle compared to uninfected muscle. The average FDG labeled leukocyte infected to uninfected muscle ratios were approximately twice as high as the FDG ratios and FDG labeled leukocytes was superior to FDG for differentiating inflamed from normal tissue. The investigators concluded that FDG labeled leukocyte accumulation at sites of inflammation is not due only to accumulation of free FDG released from leukocytes.

Pio et al. compared FDG and FDG labeled leukocytes in mice and human subjects and found that the normal distribution pattern of the two radiopharmaceuticals was different (5). They demonstrated that FDG labeled leukocyte PET could serve as a quantitative marker for identifying both the presence and the severity of intestinal inflammation. In both murine and human subjects, they reported that there was little accumulation of FDG labeled leukocytes in the normal healthy gastrointestinal and urinary tracts. Intestinal foci of FDG-labeled leukocyte accumulation corresponded to areas of histopathologically or colonoscopically confirmed areas of inflamed bowel. Intensity of uptake correlated well with the degree of inflammation. These investigators concluded that FDG labeled leukocytes could provide noninvasive quantitative assessment of bowel inflammation quickly and accurately.

Dumarey et al. prospectively investigated 21 patients with FDG labeled leukocyte PET/computed tomography (CT) (6). The mean labeling efficiency was 75±21% (range 24-96%). The mean stability of the label, in vitro, up to about four hours post labeling was 90%. Imaging was performed approximately three hours after reinfusion of labeled cells. The FDG labeled leukocytes accumulated primarily in the reticuloendothelial system, which is similar to the distribution of other radiolabeled leukocytes. Sensitivity, specificity, and accuracy of FDG labeled leukocytes all were 86%.

Rini et al. compared FDG labeled leukocyte imaging, using a gamma camera coincidence detection system, to In-111 labeled leukocyte imaging in 43 patients. Imaging was performed approximately two to six hours after reinfusion of 196-315 MBq FDG labeled leukocytes and approximately 24 hours after reinfusion of 17-25 MBq In-111 labeled leukocytes (7). The mean FDG labeling efficiency among the 43 patients, was 75±21%, significantly lower (p<0.0001) than the mean labeling efficiency for In-111, which was 90±5%. Six additional patients were excluded from reinfusion of FDG labeled leukocytes because the labeling efficiency was less than 35%. In contrast, the labeling efficiencies for In-111 in these six patients ranged from 89-93% (mean 90+2%). Mean cell viability of FDG-labeled leukocytes was 98%, which was comparable to the mean cell viability of 97% for In-111 labeled leukocytes. The accuracy of FDG labeled leukocytes (84% [36/43]) was similar to that of In-111 labeled leukocytes (81% [35/43]) and there was a high degree of concordance between the two tests.    

Aksoy et al. investigated the role of FDG-labeled leukocyte PET/CT for diagnosing prosthetic joint infection (8). These investigators performed FDG labeled leukocyte PET/CT on 46 patients with positive FDG PET/CT studies. The mean labeling efficiency was 75±17%. Labeled leukocyte viability testing was not performed. Patients underwent imaging 60-90 minutes after reinfusion of 296-703 MBq (8-19 mCi) FDG labeled leukocytes. Since only patients with periprosthetic FDG uptake were included in the investigation, only the positive predictive value of the test, which was 27%, could be calculated. In contrast, the positive predictive value of FDG labeled leukocyte PET/CT was 93.3%. The sensitivity, specificity, and negative predictive values of FDG-labeled leukocyte PET/CT were 93.3%, 97.4%, and 97.4%, respectively.

Bhattacharya et al. studied the role of FDG labeled leukocyte PET/CT for diagnosing infected fluid collections in patients with acute pancreatitis (9). The mean labeling efficiency was 81±10% (range: 31-97%). Labeled leukocyte viability was more than 99% in all patients. Imaging was performed about two hours after reinfusion of the labeled cells. They reported that the test was 100% accurate for diagnosing infected fluid collections.

While the published results suggest that FDG labeled leukocytes is a useful infection imaging radiopharmaceutical, there are issues that need to be addressed if this procedure is to be incorporated into routine clinical use. One important issue is the variability of labeling efficiency for FDG labeled leukocytes, which has ranged from less than 25% to more than 90% (6,7,8,9,10). Given this variability, how much activity should be used to label the cells? Does one assume a worst-case scenario, for example, a 25% labeling efficiency? What happens if the labeling efficiency is 90%? Is the amount of activity reinfused reduced accordingly, and if so will the number of labeled cells reinfused be sufficient to provide diagnostically useful data?

Another issue is the stability of the FDG label. Published results vary. Dumarey et al. reported a mean stability of 90% up to about 4 hours after labeling (6). Bhargava et al. in contrast, reported a mean stability of 85±4% at one hour, which decreased to 68±7% at four hours (10). In comparison, the stability of In-111 labeled leukocytes, in the same subjects, was significantly higher at both time points, 94±2% at one hour and 91±3% at four hours.

Even assuming that the issues of labeling efficiency and label stability can be overcome, the 110-minute half-life of F-18 presents logistical problems, not easily overcome. The in vitro labeling process takes approximately two three hours, so this time needs to be accounted for when determining the amount of activity used to label the cells. The 110-minute half-life of F-18 makes it impractical for labeling to be performed off-site, which means that the test would be limited to those sites capable of performing labeling. This is a significant limitation in the United States where the vast majority of leukocyte labeling procedures are performed in outside radiopharmacies. Delayed (e.g. 24 hours post injection) imaging may occasionally be desirable. The short half-life of F-18, however, precludes imaging much later than 4-5 hours after injection.

Cu-64 Labeled Leukocytes

The ideal radionuclide for labeling leukocytes should have a consistently high labeling efficiency while preserving cell viability. The radiolabel should be stable, with as little elution of radioactivity from the cells as possible. The physical half-life of the radionuclide should be sufficiently long to make the in vitro labeling procedure practical and to allow for delayed imaging. Cu-64 is an intermediate half-lived positron emitting radionuclide with a half-life of approximately 13 hours. Bhargava et al. investigated the potential of this radionuclide for labeling leukocytes in vitro (10). They labeled human leukocytes in vitro with Cu-64 and compared labeling efficiency, cell viability and stability of Cu labeled leukocytes with those of In-111 labeled leukocytes and FDG labeled leukocytes in the same subjects. The mean labeling efficiency for Cu-64 labeled leukocytes, 87±4%, was similar to that for In-111 labeled leukocytes (86±4%). Labeling efficiencies for both were significantly higher than the labeling efficiency for FDG labeled leukocytes (60±19%). Cell viabilities for all three were similar at 1 hour, and significantly higher for Cu labeled leukocytes at three and 24 hours. Label stability was always significantly higher for In-111 labeled leukocytes than for Cu-64 and FDG labeled leukocytes. These results suggest that Cu-64 labeled leukocytes might be useful for imaging infection, but there have been no clinical investigations to date.

Regardless of the radionuclide used, the in-vitro labeling procedure is labor intensive, technically demanding, requires skilled personnel, not always available, and requires direct contact with blood. It is not always possible to label enough leukocytes to obtain diagnostically useful images in the profoundly leukopenic or very young patient. Locke et al. labeled a neutrophil specific peptide, cFLFLFK-PEG, with Cu-64 (11). This peptide is an antagonist to the neutrophil formyl peptide receptor. In vitro, the radiolabeled peptide had a high binding affinity for human neutrophils and, importantly, did not exert any biologic effects on the cells themselves. They also studied this agent in a murine model of Klebsiella pneumonia. Imaging was performed approximately 18 hours after injection of the labeled peptide. Average lung SUV for Klebsiella-infected mice was 0.142±0.054 versus 0.028±0.003 for controls (p<0.003). Although the preclinical data were encouraging, to date no human studies with this agent have been conducted.

I-124 Fialuridine

The radioiodinated thymidine analogue fialuridine (FIAU) was developed for reporter genes, for cells that were transfected with herpes simplex virus thymidine kinase (TK). This enzyme transfers a phosphate group from ATP to pyrimidine deoxynucleoside. The lipophilic agent diffuses into the cell where it is trapped with the TK activity, because the phosphorylated tracer cannot pass the plasma membrane (12). Bettegowda et al. demonstrated that the TK gene of bacteria was sufficiently similar to that of the viral TK and that FIAU could also be phosphorylated by endogenous bacterial TK (13). In their investigation, FIAU inhibited wild-type bacterial growth, but had no effect on TK-deficient bacterial growth. Using the single photon emitting I-125 labeled FIAU, they successfully imaged bacterial infections in mice. Uptake in foci of infection was identified within four hours after radiopharmaceutical administration. The authors speculated that activity remained in infected tissues because I-125 FIAU was incorporated into the bacterial DNA. Activity in uninfected tissue, however, decreased over time, resulting in high target to background ratios by 48 hours after radiopharmaceutical administration.

Pullambhatla et al. investigated the role of I-125 FIAU imaging in murine bacterial pulmonary infection (14). Uptake in infected lungs was present at two hours post injection and was significantly higher than uptake in infected than inflamed uninfected and control lungs. There was a significant decrease in pulmonary uptake following antibiotic treatment. The investigators suggested that radiolabeled FIAU bacterial imaging might be useful to facilitate the development of new antibiotics in preclinical models.

Diaz et al. conducted a pilot investigation of I-124 FIAU PET/CT for diagnosing musculoskeletal infection in nine subjects including eight with suspected infection and one healthy control (15). None of the subjects received antibiotic treatment for at least three weeks prior to imaging. All seven patients with musculoskeletal infection demonstrated radiopharmaceutical uptake at the site of infection within two hours after injection. There was no abnormal uptake in one subject without infection or in the one control. There were no adverse reactions among any of the nine subjects.

Results of subsequent investigations of I-124 FIAU for diagnosing musculoskeletal infection, however, have been disappointing. Zhang et al. prospectively investigated the role of I-124 FIAU for diagnosing lower extremity joint arthroplasty infection in 19 subjects, all of whom underwent surgery within thirty days after imaging (16). Three of the 19 had a final diagnosis of infection. Imaging was performed at two and twenty fours after radiopharmaceutical administration. Interpretation of the PET/CT images was hampered by metal induced attenuation correction artifacts. On the uncorrected PET images, there was diffuse, pronounced muscle uptake. None of the semiquantitative measures used, which included target-to-background ratio SUVmax, SUVmean and SUVpeak, were useful for differentiating infected from uninfected prostheses. The authors concluded that the utility of I-124 FIAU in the detection of prosthetic joint infection is limited by suboptimal image quality due to metal artifact and high nonspecific muscle uptake. The results of an investigation of I-124 FIAU for diagnosing pedal osteomyelitis in diabetics also were disappointing. The study was terminated because of a lack of correlation between I-124 FIAU uptake and bone biopsy results (17).

Ga-68

For nearly fifty years, Ga-67 citrate has been used for imaging infection. There are, however, significant disadvantages to this radiopharmaceutical, which include nonspecific uptake in aseptic inflammation, tumors, and trauma. Image quality, even when performed as a single-photon emission computed tomography/CT study is suboptimal and the interval between administration and imaging usually is 48 to 72 hours. Now that the positron emitter Ga-68 is more widely available, investigators are studying the role of this radionuclide for diagnosing infection. Kumar et al. studied the role of Ga-68 citrate in the diagnosis of a Staphylococcus aureus infection in rodents (18). Moderate radiopharmaceutical uptake was present within five minutes after injection. There was intense focal uptake from thirty minutes to six hours post injection. They also studied one patient with a postoperative intraabdominal infection and observed that the test was positive within thirty minutes after injection. In a pilot study of patients with tuberculosis, Ga-68 accumulated in pulmonary and extra-pulmonary tuberculous lesions and was superior to CT for detecting extra-pulmonary disease. The authors concluded that Ga-68 might be useful for differentiating active from inactive disease and for monitoring treatment response (19).

Nanni et al. performed 40 Ga-68 citrate PET/CT on 31 patients suspected of having musculoskeletal infection (20). Imaging was performed approximately one hour after radiopharmaceutical administration. All 23 cases of musculoskeletal infection were positive. Four of 17 cases without infection also were positive. All four false positive results were due to tumor. Sensitivity, specificity, accuracy, positive predictive and negative predictive value were 100%, 76%, 90%, 85%, and 100%, respectively.

Although Ga-68 citrate reduces or eliminates some of the disadvantages of Ga-67 citrate, such as poor image quality and the long interval between administration and imaging, it still suffers from limited specificity (20,21). As a result, other investigators have turned their attention to complexing Ga-68 with peptides in attempts to develop more specific imaging agents (22). Vascular adhesion protein 1 (VAP-1) is a human endothelial protein that is expressed on cell surfaces under inflammatory conditions. Lankinen et al. postulated that Ga-68 labeled 1,4,7,10-tetraazacyclododecane-N′, N″,N′′′,N″″-tetraacetic acid-peptide targeted to VAP-1 (Ga-68 DOTAVAP-P1) could be useful for imaging early inflammation and infection in healing bones (23). These investigators compared Ga-68 DOTAVAP-P1 PET in 68 rats, including 34 with uninfected healing cortical bone defects and 34 with Staphylococcus aureus osteomyelitis. There was no significant difference in the mean grade of VAP-1 expression at 24 hours and seven days (1.10±0.39 and 1.00±0.47, respectively) in animals with healing cortical bone defects. The mean grade of VAP-1 expression in osteomyelitic animals was significantly higher (p=0.0330) at seven days (2.85±0.34) than at 24 hours (2.39±0.49). VAP-1 expression levels were significantly higher (p<0.0001) in infected animals than in the animals with healing bone defects at both time points. Ga-68 DOTAVAP-P1 uptake was similar in osteomyelitis and the healing cortical bone defects during the first 36 hours after surgery. Beyond 36 hours, uptake was observed only in infected bones. Ga-68 DOTAVAP-P1 uptake was significantly higher (p<0.0001) in the infected bones at 7 days. The authors concluded that Ga-68 DOTAVAP-P1 accurately demonstrates the inflammatory phase in healing bones and the progress of infection in osteomyelitic bones, facilitating the differentiation of bone infection due to Staphylococcus aureus and normal bone healing within seven days after onset. Ujula et al. performed Ga-68 DOTAVAP-P1 PET on healthy rats and rats with Staphylococcus aureus osteomyelitis of the tibia (24). Radiopharmaceutical uptake in osteomyelitis was compared with negative control peptide and competition with an excess of unlabeled DOTAVAP-P1. Ga-68 DOTAVAP-P1 was more efficiently bound to VAP-1-transfected cells than to controls. The agent was rapidly cleared from the circulation, excreted quickly in urine and had an in vivo half-life of 26±2.3 minute. Infected bones demonstrated a modest target-to-background ratio. Compared to Ga-68 DOTAVAP-P1, studies with the negative control peptide and competitors revealed significantly lower uptake at the site of infection. The authors concluded that the results of this investigation represented a proof-of-concept that infection-induced VAP-1 can be targeted by Ga-68 DOTA peptide.

The uptake mechanisms of Ga-67 in infection are complex and not well understood. It is believed that siderophores, which are gallium avid low molecular weight chelating agents produced by bacteria, are involved (25). The potential of Ga-68 labeled siderophores for imaging infection has been investigated (26,27,28). Petrik et al. conducted in vitro and in vivo investigations of several Ga-68 labeled siderophores using Aspergillus fumigatus cultures (26). Uptake in the cultures depended on iron load and siderophore type. In mice, there was rapid blood clearance and renal excretion of Ga-68 triacetylfusarinine C (TAFC) and Ga-68 ferrioxamine E (FOXE). Furthermore these two agents bound Ga-68 with high affinity and stability, low protein binding and high affinity and specificity for Aspergillus fumigatus. These results suggest that these two agents may be useful for imaging Aspergillus infection. In a rat model of Aspergillus lung infection, Petrik et al. demonstrated highly selective uptake of both Ga-68 TAFC and Ga-68 FOXE in infected lung tissue and good correlation with disease severity (27).

Petrik et al. investigated the specificity of Ga-68 TAFC and Ga-68 FOXE. In vitro uptake of these two radiopharmaceuticals was examined in various fungal, bacterial and yeast cultures and human lung cancer (H1299) cells (28). In vivo imaging was performed in fungal and bacterial rat infection and inflammation models. There was rapid accumulation of Ga-68 TAFC and Ga-68 FOXE in A. fumigatus cultures, with significantly less uptake in other fungal species and minimal uptake in other microorganisms and H1299 cells, with the exception of Ga-68-FOXE uptake in Staphylococcus aureus. There was rapid uptake of both agents in the lungs of A. fumigatus-infected rats, low accumulation in sterile inflammation and no uptake in bacterial abscesses. Ga-68 FOXE was more sensitive, while Ga-68 TAFC was more specific.

Antimicrobial peptides are an important component of the natural defenses of most living organisms. Most are small, cationic and amphipathic (hydrophilic and hydrophobic) and attack pathogens by multiple mechanisms. They exhibit broad-spectrum activity against Gram-positive and Gram-negative bacteria, yeasts, fungi and enveloped viruses. They also are involved in apoptosis, wound healing, and immune modulation (29). While most investigations of antimicrobial peptides have used Tc-99m as the radiolabel, the potential of Ga-68 labeled antimicrobial peptides as infection specific imaging compounds is now being investigated (30,31,32). Ebenhan et al. labeled the UBI29-41 fragment of ubiquicidin with Ga-68, and compared uptake of this agent in staph  aureus muscle infection, turpentine induced muscle inflammation, and ovalbumin induced lung inflammation in rabbits (31). PET/CT was performed at intervals up to two hours post injection. Infected thigh muscle could be clearly differentiated from sterile inflamed thigh muscle. Uptake in the ovalbumin-induced inflammation of the lungs was insignificant. These results suggest that Ga-68 labeled UBI29-41 potentially could differentiate infection from sterile inflammation.

Recently, Mokaleng et al. investigated the potential of an antimicrobial peptide derivative, TBIA101 that was conjugated to DOTA and radiolabeled with Ga-68 for imaging infection (32). Ga-68 DOTA-TBIA101 micro-PET/CT was performed on E. coli muscle infection in mice. Radiopharmaceutical uptake was significantly higher (p=0.333) in infected thigh muscle compared to uninfected thigh muscle, forearm muscle (p=0.092), and background (p=0.021). Ga-68 DOTA-TBIA101 clearly localized the infection site, with no notable uptake in the contralateral muscle.

Conclusion

The value of FDG-PET in the diagnostic workup of patients with infection and inflammation is now well established. The most significant limitation to this radiopharmaceutical is lack of specificity. Investigators have sought both to capitalize on the advantages that PET offers compared to single photon emitting radiopharmaceuticals, and to develop PET radiopharmaceuticals with improved specificity for infection. Initial investigations focused on in-vitro labeling of leukocytes with PET radionuclides. Although published results have been encouraging, these agents, for a variety of reasons, have not enjoyed widespread use, nor is it likely that they will. Furthermore, while labeled leukocytes are specific for leukocyte migration, they are not truly specific for infection. Initial results with I-124 FIAU, thought to be infection specific, were encouraging, but more recent data have dampened enthusiasm for this agent. Preliminary results with Ga-68 labeled siderophores and antimicrobial peptides are exciting, but clinical trials are needed before their value can truly be ascertained.