In experimentally induced Staph A infection in an animal model, Kumar et al. (2009, 2012) demonstrated avid uptake of ⁶⁸Ga-citrate at infected lesions within 10 min post injection, but focal intense uptake (SUV_max) at 30 min (Fig. 3a). They also demonstrated that the SUV_max at the target increased in a

time-dependent manner for up to 6 h post injection, with concomitant decrease in cardiac blood pool activity (Fig. 3b). Liver and bowel activity decreased up to 90 min, then stabilized. Bone marrow and growth plate uptake at joints showed moderately high uptake over the period of study. Muscular uptake decreased steadily with increased Target/Muscle (T/M) ratios over the entire period of the study from 3.7 to 8.5.

Available literature indicates a number of parameters for imaging infection using different imaging modalities such as bone scanning, labeled white blood cell scanning, and MRI with sensitivity and specificity of 82–25%, 84–80% (21–60% for axial skeleton), and 84–60%, respectively (Ma et al. 1997; Termaat et al. 2005). According to results presented by Nanni et al. (2010), the performance of ⁶⁸Ga-citrate PET/CT was not really superior to that of a conventional imaging diagnostic flow chart (reporting slightly higher sensitivity and comparable specificity, but with data from limited evaluation). Nevertheless, the following advantages were clearly indicated: simple and fast diagnostic procedure, no false positivity in cases of bone implants, and low dosimetry due to short half-life. The short half-life of ⁶⁸Ga-citrate, although an advantage from a dosimetric point of view, could be considered a drawback at the same time, because it does not allow the long uptake time typical of ⁶⁷Ga-citrate scintigraphy. However, their final results showed that a short uptake time was long enough to visualize a pathologic process, although a longer time would have guaranteed higher contrast due to reduction of background (Gelrud et al. 1974). In most patients with bone infection, they reported relatively low SUV_max, in comparison with background. Positive ⁶⁸Ga PET/CT scans presented mean SUV_max of 4.4 ± 1.8. SUV_max was 3.9 ± 1.8 (range 1.7–8.0) for acute osteomyelitis, 5.5 ± 2.0 (range 4.0–7.0) for chronic osteomyelitis, and 5.8 ± 2.0 (range 3.7–8.6) for diskitis (Nanni et al. 2010). Although they used visual criteria to decide on all positive findings (tracer uptake significantly higher than background), most patients had SUV_max between 2 and 4, a value that certainly does not clearly highlight the infected area at first glance on the maximum intensity projection image. Therefore, they reviewed each scan carefully, slice by slice, and had one false-positive result from study of a total of 31 patients, which on further examination turned out to be a tumor lesion. This event was predictable because ⁶⁷Ga-citrate has long been used as a marker for imaging tumors (Bombardieri et al. 2003).

The importance of functional imaging is not limited to diagnosis of infection but extends also to surgical planning. Three-dimensional reconstruction images obtained with ⁶⁸Ga-citrate PET/CT was shown to be very useful in a patient affected by acute osteomyelitis before and after surgical curettage, confirming complete response to antibiotic therapy (Fig. 4). Similarly, ⁶⁸Ga-citrate PET/CT has shown avid uptake of the agent before surgery, which resolved completely after surgical curettage and local antibiotic therapy in a patient affected by acute osteomyelitis of left tibial implant (Fig. 5). Fusion of ⁶⁸Ga-citrate PET images and CT images provides the surgeon with an accurate and detailed basis to better plan the operation and possibly improve patient outcome. They studied biopsy samples in all patients, but reported that only 11 of 30 biopsy samples (37%) were diagnostic. They found 4 false-positive scans, 23 true-positive scans, 13 true-negative scans, and no false-negative scans, resulting in sensitivity of 100%, specificity of 76%, positive predictive value of 85%, negative predictive value of 100%, with overall accuracy of 90%. They did not find significant tracer uptake in uninfected bone implants (Nanni et al. 2010).

It is widely believed that ¹⁸F-FDG is the more standard in vivo marker of bone infection. ¹⁸F-FDG is sensitive but has the great limitation of giving positive results in patients with bone prosthesis, even if there is no infection or mobilization (Gravius et al. 2010). In the population study by Nanni et al. (2010), many patients had different types of prostheses or bone implants, but ⁶⁸Ga-citrate was positive only in cases of infection. However, direct comparison of the performance of the two tracers is important, and no more comments can be made without such a comparison.

In summary, their preliminary work with ⁶⁸Ga-citrate PET/CT evaluated the sensitivity, specificity, positive predictive value, negative predictive value, and overall accuracy of ⁶⁸Ga-citrate PET/CT in a population of patients with suspected bone infection. Their study did not include recently operated patients or patients with recent bone fractures. Patients with bone implants showed significant artifacts on CT images but no false-positive results in the corresponding PET series, which they attributed to a correction of the View-Point iterative algorithm. Interestingly, no false-negative results were found and the negative predictive value of ⁶⁸Ga-citrate PET/CT was therefore quite high. This finding was confirmed by a previous preclinical publication (Makinen et al. 2005), but they indicated that the main limitation of this study is the lack of a gold standard to validate the PET/CT results—a long follow-up—which is the only reliable approach due to the cited limitations of other diagnostic procedures. They concluded “⁶⁸Ga-Citrate PET/CT is a new diagnostic tool that can be considered in the flow chart of patients with bone infection. However, more experience is required to further validate these results” (Nanni et al. 2010).

Diagnosing discitis has been a challenge for a long time. However, vertebral lesions are diagnosed more distinctly as an intense focal uptake of ¹⁸F-FDG than any other agent (Stumpe et al. 2000), although this is expensive to perform. MRI has shown considerable success in diagnosing discitis, as shown by the area of abnormal signal in L5–S1, equivocally consistent with infective discitis, in Fig. 6. However, it is again very expensive to perform on a routine basis. ⁶⁸Ga-citrate PET/CT, which is very cost-effective and uses an easily available agent, has been shown to define discitis as focal area of increased tracer uptake consistent with inflammation (Fig. 6b). In the past, ^99mTc-labeled ciprofloxacin (Infecton) has been successful, as it defined the lesions associated with discitis as hot area with soft tissue extension in L2 and L3 in Fig. 7a. Conventional imaging agents such as ^99mTc-HMPAO-labeled leukocyte study had limited value in imaging discitis, as it defines a large photopenic area of defect which is hard to diagnose (Fig. 7b).

Preliminary findings from our institution in patients with cellulitis and abscess indicated the potential for ⁶⁸Ga-citrate PET/CT for imaging infection. We have performed a preliminary study in 12 patients using this agent, and the results indicated that infection could be identified within 60 min after injecting ⁶⁸Ga-citrate, which is a considerable improvement over ⁶⁷Ga-citrate studies which required 48–72 h post-injection time. However, whether this agent could be used as a first-line agent for imaging infection requires further substantiation. In a patient study, an infected area in the abdomen at the site of recent appendectomy was detected within 30 min post injection of ⁶⁸Ga-citrate (Fig. 8), which was consistent with CT, and subsequently the abscess (43 mL volume) was drained from the corresponding lesion and microbiology results further confirmed the findings. The SUV_max of the lesion increased from 3.3 to 6.7 from 30 to 60 min (Fig. 8a–c) post-injection period. Cardiac blood pool and liver activities decreased during the period of study. Interestingly, there was persistent high vascular activity in the thigh region. The prolonged vascular retention is consistent with low renal and bladder activity. No bowel activity was seen (Kumar et al. 2012). As the biodistribution of the agent was high in thorax and abdominal region, it has limited but meaningful application (low SUV_max) for imaging infection in these areas but may be ideally suited for imaging infection in the extremities (high SUV_max).

The clinical utility of ⁶⁷Ga SPECT is somehow compromised due to the delayed post-injection waiting time of at least 24–48 h. In this context, one of the concerns related to ⁶⁸Ga imaging was the short half-life of ⁶⁸Ga (t _1/2 = 68 min), which may question the fundamental concept of using this PET radionuclide for longer periods of study. We examined whether the ⁶⁸Ga-TF complex, once prepared in vitro, is capable of detecting infection faster than ⁶⁸Ga-citrate (Kumar et al. 2011). This approach might provide circumstantial experimental evidence if the delayed imaging time associated with ⁶⁷Ga-citrate is due to a longer time required for in vivo binding of Ga³⁺ to plasma transferrin. TF is a glycoprotein with high affinity for binding iron on its domain. The capability of TF to bind Ga³⁺ is similar to iron-binding mechanisms (Martinez et al. 1990). Therefore, we prepared ⁶⁸Ga³⁺-TF complex in vitro and examined whether this new PET tracer is capable of detecting bacterial infection much faster than ⁶⁸Ga-citrate. It is important to establish a faster imaging method for ⁶⁸Ga as its half-life is 68 min compared with 78.3 h for ⁶⁷Ga.

⁶⁸Ga-TF was prepared by mixing post-processed pure ⁶⁸GaCl₃ (180 MBq/0.5 mL) with a solution containing 2.0 mg TF in 1.5 mL sodium carbonate (0.1 M, pH 7.0). The reaction vial was incubated in a water bath at 40°C for 1 h. TF used in the experiments was capable of binding up to 1.0 GBq ⁶⁸GaCl₃ under the reaction conditions described. The RCP was >99%. When ⁶⁸Ga-TF was prepared under identical conditions, except saline was used instead of sodium carbonate solution, the RCP was <15% (Kumar et al. 2011). Therefore, it is mandatory to use sodium bicarbonate instead of saline for preparing ⁶⁸Ga-TF. The solutions were sterilized for animal injection by final filtering with 0.22-μ filter. The RCP of ⁶⁸Ga-TF was determined by TLC/alumina chromatography, using sodium citrate (0.1 M) as solvent (free ⁶⁸Ga, R _f = 1; bound ⁶⁸Ga, R _f = 0). Stability study was performed by mixing 0.3 mL ⁶⁸Ga-TF solution with 1.5 mL human serum at 37°C. RCP was measured at 1 h interval.

When ⁶⁸Ga-TF (10–15 MBq/0.2 mL) was injected into infected rats, the lesion was detectable within 20 min post injection, although intense, focal uptake was seen only after 50 min post injection (Fig. 9a). The accumulation of activity at the lesion increased with time, as shown by significantly increased SUV_max at the target from 20 min for up to 4 h post injection (Fig. 9b). There was a considerable decrease in cardiac activity during the initial 90 min post-injection period, but a small percentage of activity was measurable for up to 4 h post injection of the study. The activity associated with liver and bowel decreased rapidly up to 90 min of the post-injection period. The low level of activity then persisted for up to 4 h post-injection time. There was marginal increase in the activity associated with the spine over the entire period of the study, whereas muscular uptake increased marginally during the initial 60 min post-injection time but then decreased steadily over the period of the study. Since the background activity was washed out in a time-dependent manner, the T/M ratio increased progressively over the entire period of the study from 2.2 to 7.5 from 10 min to 4 h post injection (Fig. 9b) (Kumar et al. 2011).

The results in one particular rat showed avid uptake of the agent, as expected at the site corresponding to Staph A-induced infection in the right thigh. It also showed an additional lesion, with clearly defined intense focal uptake in the lower abdominal area, which was persistent during the entire period of study (Fig. 10). Physical examination of the additional lesion site indicated an infected wound and an abscess on the skin. When biopsied, microbiology identified the organism as Proteus mirabilis. Therefore, the study indicated that ⁶⁸Ga-TF is capable of detecting both Gram-positive and Gram-negative infection caused by both Staph A and Proteus mirabilis, and may be extrapolated to all anaerobic bacteria.

Therefore, with the circumstantial experimental evidence, it may be reasonable to conclude that ⁶⁸Ga-TF complex is able to accumulate at bacterial and inflammatory lesions (Henkin 1978; Tsan 1985). Available literature suggests that the total amount of TF in humans is approximately 240 mg/kg (Bernstein 1998). Therefore, it is reasonable to assume in our study that the mass of injected TF (0.2 mg/rat) in ⁶⁸Ga-apo-transferrin is too small to induce any unexpected pharmacological effect to influence the biodistribution of the agent.

PET using ¹⁸F-FDG is ideal for imaging infection, inflammation, and tumor, which are typically characterized by increased glucose metabolism and increased expression of glucose transporters (Ichiya et al. 1996; Sugawara et al. 1999; Stumpe et al. 2000). ¹⁸F-FDG uptake is elevated in activated inflammatory cells such as leukocytes, granulocytes, and macrophages because of the above phenomenon. Early bone healing also mimics infection, as it involves an inflammatory phase, which represents a highly activated state of cell metabolism and glucose consumption (Einhorn 1998), and therefore findings with ¹⁸F-FDG PET imaging may be complicated by not differentiating inflammation, infection, and bone healing. It has been proposed that an interval of 3–6 months should be allowed to minimize the risk of false-positive findings in patients with postsurgical and traumatic bone healing (de Winter et al. 2002). Recent studies in a rabbit osteomyelitis model indicated that bone infection could be distinguished from bone healing by means of ¹⁸F-FDG PET after only an interval of 3 weeks (Koort et al. 2004). However, the transient ¹⁸F-FDG uptake by healing bones could still create problems in differential diagnosis under these clinical conditions.

The clinical utility of the two agents ¹⁸F-FDG and ⁶⁸Ga-chloride was compared for monitoring normal bone healing and osteomyelitis in rat tibia (Fig. 11). Quantitative uptake of these agents was also estimated to verify the uptake of ¹⁸F-FDG and ⁶⁸Ga, in normal bone healing and osteomyelitis. The results showed, based on 120-min dynamic imaging, that accumulation of ⁶⁸Ga was slower (70 min post injection) than that of ¹⁸F-FDG (40 min post injection) in the infected bone. On the other hand, they showed increased uptake of both tracers (¹⁸F-FDG and ⁶⁸Ga) in osteomyelitic tibia compared with contralateral normal tibia. Control animals with healing bone defects showed slightly increased uptake of ¹⁸F-FDG, but no apparent increase in ⁶⁸Ga uptake was seen as compared with contralateral intact tibia (Makinen et al. 2005).