ABSTRACT
Diabetic foot infection, which is defined as an infection of the soft tissues or bone below the malleoli, is a common clinical problem. Diabetic patients can have a significant foot infection with few or no signs or symptoms and the diagnosis of osteomyelitis can easily be overlooked. As a result, imaging studies have become an integral part of the diagnostic workup of these patients. Although not especially sensitive or specific radiographs are routinely performed because they provide an anatomic overview of the area of interest and any pre-existing conditions that could potentially influence the selection and interpretation of subsequent procedures. Bone scintigraphy is sensitive but not specific, and if used at all, should be used as a screening procedure. In-vitro labeled leukocyte scintigraphy is, at the moment, the nuclear medicine gold standard for diagnosing diabetic pedal osteomyelitis, with sensitivity and specificity ranging from 72% to 100% and 67% to 100%, respectively for In-111 labeled leukocytes and 86% to 93% and 80% to 98%, respectively, for Tc-99m labeled leukocytes. Results obtained with leukocytes labeled in-vivo with antigranulocyte antibodies and antibody fragments, have been more variable. Furthermore, these agents are not widely available. In-vitro and in-vivo labeled leukocyte imaging studies are limited by a combination of poor image resolution and the small size of the structures being evaluated. Incorporating single photon emission computed tomography/computed tomography (SPECT/CT) into the procedure improves the accuracy of the test by precisely localizing foci of labeled leukocyte accumulation, facilitating the differentiation of soft tissue and bone infection, and improving confidence of interpretation. Furthermore, the CT component of the examination may have prognostic value. Recent developments suggest that SPECT/CT also may be useful for determining the success of antibiotic treatment in diabetic patients with pedal osteomyelitis. F-18 fluorodeoxyglucose (FDG) positron emission tomography (PET) has several advantages over single photon emitting radiopharmaceuticals. PET is a high resolution tomographic technique. FDG is a small molecule that enters poorly perfused areas quickly. The procedure is completed in hours rather than days and no laborious labeling procedures are involved. Unfortunately, the available literature on the role of FDG-PET and PET/CT in the diagnostic workup of diabetic foot infections is inconclusive and at times contradictory. Large multicenter trials, including a uniform reference standard and more extensive comparison with labeled leukocyte imaging are needed to establish the role of this test in diabetic foot infections.
Introduction
Diabetic foot infection, which is defined as an infection of the soft tissues or bone below the malleoli, is a common clinical problem. Most of these infections occur at a site of skin trauma or ulceration. The estimated lifetime risk of a person with diabetes mellitus developing a foot ulcer is 15% to 25%, with an annual incidence of 3% to 10%. Major predisposing factors include peripheral arterial disease, peripheral neuropathy, and compromised immunity. More than 50% of all nontraumatic lower extremity amputations are related to diabetic foot infections, and 85% of all lower extremity amputations in patients with diabetes are preceded by an ulcer (1,2,3).
Any breach in the cutaneous integument of the foot in a person with diabetes has the potential to develop into severe soft-tissue and/or osseous infection. Osteomyelitis should be considered in all patients with a chronic non healing wound, especially when the wound extends beyond the dermis. The exposed bone or probe to bone test, which is easy to perform, has a sensitivity and specificity of 60% and 91%, respectively, for diagnosing osteomyelitis underlying an open wound (4). Diabetic patients, however, can have a significant foot infection with few or no signs or symptoms and without mounting a systemic inflammatory response and the diagnosis of osteomyelitis in these individuals can easily be overlooked (5). As a result, imaging studies have become an integral part of the diagnostic workup of the diabetic patient with a foot infection.
In spite of a sensitivity of only about 54% and a specificity of about 68%, radiographs, which are nearly universally available and relatively inexpensive, should be the first imaging test performed. Even when not diagnostic, radiographs provide an anatomic overview of the area of interest and any pre-existing conditions that could influence both the selection and interpretation of subsequent procedures (4). Although magnetic resonance imaging plays an important role in the diagnostic evaluation of the diabetic patient with a foot infection, a discussion of this technique is beyond the scope of this review, which will focus on molecular imaging studies.
Molecular Imaging Bone Scintigraphy
Focal hyperperfusion, focal hyperemia, and focal bony uptake are the classic presentation of osteomyelitis on three phase bone scintigraphy. Many of the conditions to which the diabetic patient with foot problems is prone, however, including fracture, the neuropathic joint, and even the pedal ulcer, can mimic osteomyelitis (Figure 1, 2) (6). Sensitivity and specificity of the three phase bone scan range from 75% to 100% and 0% to 59%, respectively (7,8,9,10,11,12,13,14,15,16,17,18). Attempts at improving the specificity of the three phase bone scan, including differentiating arterial from venous hyperperfusion on the flow phase of the study and the addition of delayed, next day, imaging, the so called four phase bone scan, have yielded mixed results and are not widely used (19,20,21,22,23).
Labeled Leukocyte Imaging
In-vitro labeled leukocyte scintigraphy has long been considered the molecular imaging gold standard for diagnosing diabetic pedal osteomyelitis. The sensitivity of the test, using In-111 labeled leukocytes with planar imaging, has ranged from 72% to 100% and the specificity from 67% to 100% [7-10,14,20,21] (Figure 3). The sensitivity and specificity of technetium-99m exametazime labeled leukocyte planar imaging for diagnosing diabetic pedal osteomyelitis range from 86% to 93% and from 80% to 98%, respectively (11,12,13,17,24).
The limitations to in-vitro labeled leukocyte imaging are well known. The labeling procedure requires skilled personnel, is laborious, not always available, and involves direct contact with blood. Considerable effort, therefore, has been devoted to the development of in-vivo leukocyte labeling methods. Most of these investigations have focused on the use of antigranulocyte antibodies and antibody fragments. Besilesomab is a murine monoclonal immunoglobulin G1 (IgG1) antibody that binds to the nonspecific cross-reacting antigen-95, which is expressed on granulocyte and granulocyte precursor cell membranes. About 10% of Tc-99m besilesomab is neutrophil bound within 45 minutes after administration. Another 20% circulates freely and presumably localizes in infection through nonspecific mechanisms (25). Dominguez-Gadea et al. reported that Tc-99m besilesomab was 93% sensitive, 78% specific, and 84% accurate for diagnosing pedal osteomyelitis in diabetics (26). Schwegler et al. reported sensitivity, specificity and accuracy of 29%, 85% and 65%, respectively, for diabetic pedal osteomyelitis (27). The incidence of human antimurine antibody (HAMA) response in patients receiving Tc-99m besilesomab, which ranges from less than 5% in patients receiving a single dose of 125 ug of the antibody to more than 30% in individuals receiving repeated injections is a significant disadvantage to the test. Consequently, patients should be prescreened for HAMA, should not be injected with more than 250 ug antibody and should not undergo repeat administration (25).
Antibody fragments do not induce a HAMA response. Sulesomab is a 50 kDa fragment antigen binding (Fab’) portion of an IgG1 class murine monoclonal antibody that binds to normal cross-reactive antigen-90 present on leukocytes. Approximately 3-6% of the Tc-99m sulesomab injected is associated with circulating neutrophils; by 24 hours after injection, about 35% of the remaining activity is in the bone marrow. Initial investigations suggested that sulesomab uptake in infection includes binding to circulating neutrophils that subsequently migrate to foci of infection and binding to leukocytes already present at the site of infection. There are other data, however, that suggest that sulesomab does not bind to circulating leukocytes, and that it accumulates in infection nonspecifically through increased capillary membrane permeability (25). In a prospective investigation, Delcourt et al. compared combined Tc-99m sulesomab/bone imaging to combined bone/gallium imaging in 25 diabetic patients. The sensitivity, specificity, and accuracy of Tc-99m sulesomab/bone imaging were 67%, 85%, and 74% versus 44%, 77%, and 58%, respectively for bone/gallium imaging (28). Harwood et al. studied 122 diabetic patients with foot ulcers and reported that Tc-99m sulesomab had 91% sensitivity, 56% specificity, and 80% accuracy for diagnosing pedal osteomyelitis (29). The test was significantly more sensitive (92% vs. 79% p<0.05), and slightly less specific (58% vs. 67%) than in-vitro labeled leukocyte imaging and significantly more specific than bone scintigraphy (50% vs. 21%; p<0.05).
The accuracy of both in-vitro and in-vivo labeled leukocyte imaging is limited by a combination of poor image resolution and the small size of the structures being evaluated. Several investigators have used single photon emission computed tomography/computed tomography (SPECT/CT) to improve the accuracy of the test. Heiba et al. observed that, even when performed as a SPECT/CT study, it was not always possible to differentiate between soft tissue infection and osteomyelitis on indium labeled leukocyte studies (30). They investigated simultaneous dual isotope labeled leukocyte/ bone SPECT/CT, and when necessary, bone marrow SPECT/CT, in 213 diabetic patients. They found that simultaneous dual isotope SPECT/CT was significantly more accurate than planar imaging and single isotope bone and In-WBC SPECT/CT. Performing simultaneous dual isotope labeled leukocyte/bone SPECT/CT facilitated precise localization of labeled leukocyte accumulation, thereby improving confidence of interpretation and test accuracy. In another investigation, Heiba et al. compared simultaneous dual isotope, labeled leukocyte/bone SPECT/CT to conventional imaging for diagnosing and managing foot infections in diabetic patients (31). Compared to conventional imaging studies, dual isotope SPECT/CT resulted in higher reader confidence and more accurate diagnosis of soft-tissue infection, osteomyelitis and other bony abnormalities. Dual isotope SPECT/CT also provided clear guidance on patient management and resulted in more limb salvage procedures instead of amputations. Equally important, from an economic standpoint, compared to conventional imaging, the use of dual isotope SPECT/CT resulted in considerably decreased length of hospitalization.
An alternative to dual isotope SPECT/CT is to use technetium rather than indium labeled leukocytes. Technetium offers superior resolution and labeling and imaging both can be performed on the same day. Filippi et al. performed planar and SPECT/CT technetium labeled leukocyte imaging on seventeen diabetic patients with 19 suspected sites of infection. SPECT/CT was performed six hours after reinfusion of labeled cells (32). Labeled leukocyte imaging was positive in 16 of the 19 sites. SPECT/CT changed the interpretation of planar and SPECT images in 10 (53%) of the sites, by excluding osteomyelitis in six sites, identifying bone and soft tissue infection in three sites and identifying osteomyelitis in one site. These investigators concluded that the accuracy of diagnosing diabetic pedal osteomyelitis with technetium labeled leukocyte imaging is improved by incorporating SPECT/CT into the imaging protocol. Przybylski et al. retrospectively reviewed the results of technetium labeled leukcoyte SPECT/CT performed on 14 diabetic patients, all of whom had pedal ulcers (33). Five of the patients also had peripheral arterial disease. Imaging was performed about three hours after labeled leukocyte reinfusion. The sensitivity, specificity, and accuracy of the test were 87.5%, 71.4%, and 80%, respectively.
Erdman et al. in an effort to classify the severity of diabetic foot infections for prognostic purposes, developed a scoring system, the Composite Severity Index (CSI) for technetium labeled leukcoyte SPECT/CT (34). They hypothesized that analyzing certain key anatomic and molecular imaging parameters on the SPECT/CT images could provide both diagnostic and prognostic information. They retrospectively analyzed and graded the intensity of labeled leukocyte uptake, the number and anatomic location of the lesions, and on the CT component of the test, the presence or absence of adjacent disrupted bone architecture. The CSI was derived from these data and was compared to a binary interpretation (positive or negative for osteomyelitis), of the images. These investigators observed that the likelihood of a favorable outcome varied inversely with the CSI score. As the CSI score increased, the likelihood of a favorable outcome decreased and the likelihood of treatment failure increased. The CSI was significantly more accurate (p=0.016) than binary interpretation for predicting therapeutic outcome.
Over the years, the vast majority of the investigations have focused on the role of labeled leukocyte imaging for diagnosing diabetic foot infections, with scant attention paid to monitoring response to treatment.
Newman et al. studied serial planar indium labeled leukocyte imaging in 35 diabetic patients with 41 pedal ulcers, including 28 with biopsy proven underlying osteomyelitis (20). They observed that the intensity of labeled leukocyte activity decreased by 16 to 34 days and normalized by 36 to 54 days of appropriate antibiotic treatment. They did not, however, use this information for prognostic purposes. More recently, data are emerging that suggest that technetium labeled leukocyte SPECT/CT may be useful for monitoring treatment response in diabetics with pedal osteomyelitis. Vouillarmet et al. compared radiographs, three phase bone scintigraphy, and technetium labeled leukocyte SPECT/CT for assessing pedal osteomyelitis remission, which was defined as an absence of relapsed osteomyelitis after one year follow up, in 29 patients (35). Radiographs showed no improvement or progression from baseline in 20 patients, 16 of whom had relapsed osteomyelitis. There was radiographic improvement in nine patients, including one with relapsed osteomyelitis. Three phase bone scintigraphy was positive in 26 patients; only five developed relapsed osteomyelitis. The test was negative in three patients, none of whom developed relapsed osteomyelitis. Technetium labeled leukocyte SPECT/CT, performed 20 hours after reinfusion of labeled leukocytes, was negative in 22 patients, none of whom developed relapsed osteomyelitis. Seven studies were positive for osteomyelitis, including five with relapsed infection. Sensitivity, specificity, and positive and negative predictive values for predicting osteomyelitis relapse after completion of antibiotic treatment were 80%, 33%, 20% and 89%, respectively for radiographs; 100%, 12.5%, 15.5% and 100%, respectively for three-phase bone scintigraphy; and 100%, 91.5%, 71.5% and 100%, respectively, for technetium labeled leukocyte SPECT/CT. The authors concluded that a negative test reliably identifies diabetic foot osteomyelitis remission and potentially could be very useful for guiding antibiotic therapy.
Lazaga et al. retrospectively reviewed pre- and post treatment technetium labeled leukocyte SPECT/CT in 20 diabetic patients with pedal osteomyelitis to assess the value of this test for monitoring response to treatment (36). Successful treatment of osteomyelitis was defined as wound healing or lack of re-admission for bone infection at the same site within 1 year. The sensitivity, specificity, and positive and negative predictive values for determining success of treatment were 90%, 56%, 69% and 83%, respectively.
These preliminary results suggest that technetium labeled leukocyte SPECT/CT may be useful for monitoring response to treatment in diabetic patients with pedal osteomyelitis. The test appears to be especially valuable when it is negative, making relapsed infection very unlikely.
F-18 Fluorodeoxyglucose
F-18 fluorodeoxyglucose (FDG) positron emission tomography (PET) has several advantages over single photon emitting radiopharmaceuticals. PET intrinsically is a high resolution tomographic technique that enables precise localization of radiopharmaceutical accumulation. FDG is a small molecule that enters poorly perfused areas quickly. The procedure is completed in hours rather than in days, and in contrast to in-vitro labeled leukocytes, no laborious labeling procedures are required. FDG uptake in traumatized bone usually normalizes within three to four months and degenerative bone changes usually show only mildly increased uptake (6).
Several groups have investigated the role of FDG-PET and PET/CT in the evaluation of diabetic foot infections. Nawaz et al. prospectively compared FDG-PET, plain radiography, and magnetic resonance imaging (MRI) for diagnosing pedal osteomyelitis in 110 diabetic patients, although not all patients underwent all three studies (37). All patients had a blood glucose level below 200 mg/dL. Images were interpreted visually; SUV was not used. Sensitivity, specificity, positive and negative predictive values, and accuracy of FDG-PET were 81%, 93%, 78%, 94%, and 90%, respectively. Sensitivity, specificity, positive and negative predictive values, and accuracy of MRI were 91%, 78%, 56%, 97%, and 81%, respectively. Sensitivity, specificity, positive and negative predictive values and accuracy of radiographs were 63%, 87%, 60%, 88%, and 81%, respectively. The authors concluded that FDG-PET is highly specific for diagnosing diabetic pedal osteomyelitis and is a useful compliment to MRI. The authors also suggested that when MRI cannot be performed FDG-PET can be used after negative or inconclusive radiographs. The vast majority of diabetic patients with pedal osteomyelitis present with ulcers involving the distal forefoot; pedal osteomyelitis in the absence of an ulcer is uncommon. Unfortunately data about the areas of interest (forefoot, mid foot, hind foot) and the number of patients with soft tissue ulcers were not provided, which makes it difficult to assess the significance of the findings in this investigation. The results obtained by Schwegler at al. were very different (27). These investigators, who also used only visual image analysis, prospectively evaluated FDG-PET for diagnosing clinically unsuspected osteomyelitis in 20 diabetic patients with nonhealing pedal ulcers. The ulcers were located in the forefoot in 18 patients and in the hind foot in two patients. Information on blood glucose levels at the time of imaging was not provided. Sensitivity, specificity and accuracy were 29%, 92% and 70%, respectively, for FDG PET and 86%, 92% and 90%, respectively, for MRI. The investigators speculated that the low sensitivity of the test may have been related to a diminished inflammatory response in the study population and/or to impaired bony uptake of FDG, because of insulin resistance. Motion artifacts and limited spatial resolution may have been additional contributing factors. The authors concluded MRI might be the preferred imaging modality for diagnosing osteomyelitis in diabetic patients with non healing pedal ulcers.
Yang et al. studied 48 diabetic patients with suspected pedal osteomyelitis, including 27 with serum glucose levels above 150 mg/dL (38). The accuracy of FDG-PET was 93.8%. Sensitivity was 88.9% for patients with serum glucose levels above 150 mg/dL and 88.3% for patients with serum glucose levels below 150 mg/dL. They concluded that a mildly to moderately elevated serum glucose level does not affect FDG-PET accuracy for detecting pedal osteomyelitis in diabetic patients.
Keidar et al. compared FDG-PET and PET/CT in 14 diabetic patients with 18 clinically suspected sites of infection, including seven in the forefoot, one in the mid foot, and ten in the hind foot (39). Blood glucose levels ranged from 84 mg/dL to 330 mg/dL. Seven patients had blood glucose levels in excess of 200 mg/dL. There were 14 foci of abnormal uptake on the PET images consistent with infection. PET/CT confirmed the diagnosis in 13 of the 14 sites, correctly localizing the abnormalities to bone in 8 sites to bone and to the soft-tissues in five sites. One site, thought to be infection on PET images, was identified as diabetic osteoarthropathy on PET/CT images. Maximum SUV in the sites of abnormal FDG uptake ranged from 1.4 to 11.1. There was no relationship between the patients’ blood glucose level and degree of FDG uptake. The authors concluded that FDG can be used for diagnosing diabetes-related foot infection and that by performing PET/CT it is possible to accurately differentiate osteomyelitis from soft-tissue infection (Figure 5). Kagna et al. in an expansion of this investigation reported sensitivity, specificity, and accuracy of 100%, 93%, and 96%, respectively for the diagnosis of diabetic pedal osteomyelitis (40).
Familiari et al. compared FDG-PET/CT to planar Tc-WBC imaging in 13 diabetic patients with a high pretest likelihood of foot infection (41). All 13 patients had a blood glucose level of less than 160 mg/dL. Histopathologic confirmation of the final diagnosis was available for all cases. Both FDG-PET/CT and technetium labeled leukocyte planar imaging were performed at multiple time points. Sensitivity, specificity, positive predictive value, negative predictive value, and accuracy of technetium labeled leukocyte imaging for diagnosing osteomyelitis were 86%, 100%, 100%, 86%, and 92%, respectively. Sensitivity, specificity, positive predictive value, negative predictive value, and accuracy of FDG PET/CT for diagnosing osteomyelitis were 43%, 67%, 60%, 50%, and 54%, respectively. Accuracy of FDG PET/CT improved to 62%, when CT findings were incorporated into the image interpretation. The investigators concluded that FDG PET/CT has a low diagnostic accuracy and cannot replace labeled leukocyte imaging for diagnosing pedal osteomyelitis in diabetics.
The majority of diabetic patients with diabetic foot infections present with an ulcer in the distal forefoot. The neuropathic, or Charcot, joint, which is much less common than, and not always associated with, an ulcer, usually develops in the diabetic mid or hind foot. The tarsal and tarsometatarsal or Lisfranc joints are the most frequently involved locations (60%) followed by the metatarsophalangeal joints (30%) and the tibiotalar joint (10%). Differentiating infection from the neuropathic joint or diagnosing infection superimposed on the neuropathic joint is challenging. Three-phase bone scintigraphy invariably is positive in both conditions. Labeled leukocytes accumulate in both the infected and the uninfected neuropathic joint. Accumulation in the uninfected neuropathic joint is due, at least in part, to hematopoietically active marrow in the involved bones. Performing complementary bone marrow imaging accurately differentiates labeled leukocyte uptake due to bone marrow from that due to infection (Figure 6) (42). Palestro et al. reported an accuracy of 95% for planar In-WBC/bone marrow imaging. Heiba et al. reported similar results using dual isotope SPECT/CT (30,43).
Data on the role of FDG PET and PET/CT in the evaluation of the neuropathic joint are limited. Höpfner et al. performed FDG-PET on 16 diabetic patients to determine the value of the test in the preoperative evaluation of the neuropathic joint (44). None of the patients in this investigation had osteomyelitis. The investigators suggested that, even though none of the patients studied had diabetic foot infection, because of the relatively low uptake in the uninfected neuropathic joint, FDG-PET could be useful for differentiating osteomyelitis from neuropathic lesions. Shagos et al. compared FDG-PET to three phase bone scintigraphy in the evaluation of diabetic foot complications including osteomyelitis (n=36) and the neuropathic joint (n=43) (15). They reported that FDG-PET was more specific than bone scintigraphy for osteomyelitis, while bone scintigraphy was more sensitive than FDG-PET for the neuropathic joint. Basu et al. investigated the ability of FDG-PET to differentiate pedal osteomyelitis and soft tissue infection from the uninfected neuropathic joint (45). The sensitivity and accuracy of FDG-PET for diagnosing osteomyelitis were 100% and 94%. The investigators concluded that FDG-PET had a high negative predictive value for excluding osteomyelitis in the presence of the neuropathic joint. Unfortunately only one patient in this investigation had osteomyelitis superimposed on the neuropathic joint.
Based on available evidence it is not possible to conclude that FDG PET or even PET/CT should be the molecular imaging test of choice for diabetic pedal osteomyelitis. The limitations of the available data include variable methodology, different patient populations, variable image interpretation, (visual versus SUV), lack of information about the presence or absence of vascular insufficiency, inconsistent correlation with MRI, a paucity of comparative studies with labeled leukocyte imaging, and lack of a uniform standard of truth (46,47). Treglia et al. in a meta-analysis, reported a sensitivity and specificity of 74% and 91%, respectively for FDG for diagnosing diabetic pedal osteomyelitis and commented that the literature on FDG imaging in suspected diabetic pedal osteomyelitis remains limited and large multicenter studies using bone biopsy as the reference standard are needed (47).
Conclusion
The diagnosis of pedal osteomyelitis in a diabetic patient with a foot infection is often overlooked and imaging studies have become an integral part of the diagnostic workup of these patients. At the present time, in-vitro labeled leukocyte imaging is the molecular imaging test of choice for this population. Published data suggest that either In-111 or Tc-99m labeled leukocytes can be used, with comparable results. Regardless of the radiolabel used, however, SPECT/CT should routinely be performed. Incorporating SPECT/CT into the procedure improves both reader confidence and test accuracy. SPECT/CT may also have prognostic value and be useful for monitoring response to treatment. The concept of using FDG PET and PET/CT in the diagnostic evaluation of the diabetic foot infection is exciting, but more extensive investigations are needed before any conclusions about its role can be drawn.
