Using Nuclear Medicine Imaging Wisely in Diagnosing Infectious Diseases.

In recent years, there has been an increasing emphasis on efficient and accurate diagnostic testing, exemplified by the American Board of Internal Medicine's "Choosing Wisely" campaign. Nuclear imaging studies can provide early and accurate diagnoses of many infectious disease syndromes, particularly in complex cases where the differential remains broad. This review paper offers clinicians a rational, evidence-based guide to approaching nuclear medicine tests, using an example case of methicillin-sensitive Staphylococcus aureus (MSSA) bacteremia in a patient with multiple potential sources. Fluorodeoxyglucose-positron emission tomography (FDG-PET) with computed tomography (CT) and sulfur colloid imaging with tagged white blood cell (WBC) scanning offer the most promise in facilitating rapid and accurate diagnoses of endovascular graft infections, vertebral osteomyelitis (V-OM), diabetic foot infections, and prosthetic joint infections (PJIs). However, radiologists at different institutions may have varying degrees of expertise with these modalities. Regardless, infectious disease consultants would benefit from knowing what nuclear medicine tests to order when considering patients with complex infectious disease syndromes.

A 66-year-old man was admitted to the hospital with fever. He had a past medical history significant for diabetes (complicated by peripheral neuropathy, Charcot foot, and nephropathy), an aortic aneurysm surgically repaired with an endovascular graft, a right total knee arthroplasty, and chronic low back pain attributed to degenerative disc disease of the lumbar spine. His exam was notable for a slightly swollen right knee without erythema or warmth and a deep nonpurulent ulcer on the plantar surface of his left foot, which did not probe to bone. He had slight paraspinal tenderness without vertebral tenderness on palpation of the lumbar spine. He was found to have methicillin- sensitive Staphylococcus aureus (MSSA) on multiple blood cultures. Transthoracic echocardiography and transesophageal echocardiography were negative for vegetations or valvular abnormalities. A plain radiograph of his left foot did not demonstrate bony erosions to suggest osteomyelitis. A magnetic resonance image (MRI) of the same foot showed some mild soft tissue swelling and marrow edema, but the differential included osteomyelitis versus Charcot foot.

He was started on an oxacillin infusion. His blood cultures became negative after 3 days of therapy. Determining the source of this infection is necessary to guide decisions regarding further management. Diagnostic delays for some infections (such as endovascular graft infections) may have serious consequences. However, choosing diagnostic tests that will yield the most fruitful and timely results remain challenging.

Nuclear medicine imaging can be an excellent resource in trying to diagnose complicated infections such as the case presented here. In this review, we discuss the utility of different nuclear medicine imaging modalities for each of 5 common infectious disease syndromes.

DIABETIC FOOT OSTEOMYELITIS

Although his clinical exam and imaging findings are equivocal given his underlying Charcot arthropathy, our patient is at significant risk of having diabetic foot osteomyelitis (DFO). Historically, a triple-phase bone scan with technetium-99m-labeled bisphosphonates (99mTc) was used to diagnose DFO, but this has been replaced by MRI as the preferred first imaging test. Bone scanning still may have a role in DFO, particularly when MRI is contraindicated (ie, metal implants, significant renal failure) or inconclusive [1]. The degree of radiopharmaceutical uptake in bone scans depends on blood flow and rate of new bone formation. Osteomyelitis will show focal hyperperfusion in phase 1 (arterial phase), soft tissue inflammation in phase 2 (caused by increased vascular permeability or what is known as the blood pool phase), and focal bone localization in phase 3 (osteoblastic activity). A triple-phase bone scan can demonstrate the presence of OM within 2–3 days of the onset of infection, compared with a plain radiograph which can take 2–3 weeks [2].

Reported sensitivity of a bone scan ranges from 80% to 90%, but specificity is less than 50% because it cannot distinguish from other conditions such as Charcot arthropathy, bony metastasis, fracture, trauma, or recent surgery [3]. Given the poor specificity of the bone scan, labeled leukocytes (often referred to as a tagged WBC scan) can be paired with the bone scan, resulting in an improved specificity of 80%–90% [3].

To perform a WBC scan, a total WBC count of at least 2000 cells/μL is required [2]. Leukocytes are removed from the patient and tagged with a radioactive tracer, usually either 99mTc or 111-indium. Next, the radiolabeled cells are re-infused into the patient where they then accumulate at a focus of inflammation or infection. Given its poor resolution, a tagged WBC scan alone cannot distinguish bone from soft tissue, so it must be combined with another study such as a bone scan to evaluate for OM. Other limitations of the tagged WBC scan include the time-consuming nature of the study, as well as the cost of tagging the WBCs, which can amount to over $1000 [3].

Over the years, studies have shown promising results for the diagnosis of musculoskeletal infections by combining 99mTc-sulfur colloid imaging (referred to as marrow imaging) plus a tagged WBC scan [4, 5]. Although both the radiotracer 99mTc-sulfur colloid and leukocytes travel to the bone marrow, 99mTc-sulfur colloid does not accumulate in areas where the normal marrow has been replaced by infection. Therefore, in OM, one would expect to see accumulation of tagged WBC but no activity on the sulfur colloid marrow imaging [6]. This would be helpful in our patient with Charcot arthropathy and surrounding soft tissue infection, where both the bone scan and tagged WBC scan may yield an incorrectly positive result. A study in patients with Charcot arthropathy found that this combination was indeed superior to a triple-phase bone scan either alone or in combination with a tagged WBC scan for the diagnosis of DFO [4]. In a patient without Charcot foot, colloid imaging may not be required, and a triple-phase bone scan combined with a tagged WBC scan would be a good adjunct when MRI is contraindicated or inconclusive.

Single-photon emission CT (SPECT) is a nuclear medicine technique that can generate three-dimensional information and can be used with various radioisotopes. It is still being evaluated for its application in DFO. One recent study found similar sensitivity and specificity of 99mTc-labeled WBC scan plus SPECT imaging compared with traditional MRI for the diagnosis of DFO [7].

A final modality that can be used for the diagnosis of DFO is PET, in which radioactive fluorine is attached to fluorodeoxyglucose (18F-FDG). This radiotracer will accumulate in locations of increased glucose metabolism in infectious (ie, macrophages), inflammatory, and malignant processes [1, 3]. Clinical studies on F-FDG-PET comparing diabetic with nondiabetic patients have suggested no difference in accuracy, although one study demonstrated a decreased uptake of FDG in ovarian cancer cells grown in media containing 300 mg/dL glucose, suggesting an effect of acute hyperglycemia [8, 9].

A recent meta-analysis of 9 studies evaluating 18F-FDG-PET, either alone or combined with CT scanning, demonstrated a pooled sensitivity and specificity for diagnosing DFO of 74% and 91%, respectively [10]. However, there was a wide range of sensitivities, including 1 small study reporting low sensitivity (29%) in patients with diabetic foot ulcer without clinical signs of osteomyelitis (OM) who ultimately had biopsy-proven OM [10-12]. A major limitation of this meta-analysis was the wide variation in the gold standard used. Most studies used only clinical monitoring as the gold standard, and only a minority used bone biopsy [10]. A large multicenter study using bone biopsy as the gold standard would help clarify the accuracy of 18F-FDG-PET imaging in diabetic foot infections.

VERTEBRAL OSTEOMYELITIS

A spinal or paraspinal infection is also a concern in this patient with back pain and high-grade bacteremia. Although MRIs are quite reliable for the diagnosis of epidural abscesses, there may be clinical situations in which an MRI can give misleading results for the diagnosis of vertebral osteomyelitis (V-OM) (Figure 1). Degenerative disc disease can sometimes be mistaken for infectious endplate abnormalities on MRI (resulting in a false positive). Similarly, MRIs may on occasion miss the diagnosis of V-OM [13]. Clinical features such as the presence or absence of fever or leukocytosis combined with MRI results often will lead to the correct diagnosis, but if the clinical suspicion is high and MRI features are not definitive, or if MRI is contraindicated in situations such as in the presence of some pacemakers or severe renal insufficiency, nuclear medicine studies may be considered as an alternative diagnostic modality.

Figure 1.

99mTc-hydroxymethylene diphosphonate bone SPECT scan of patient with osteomyslitis of fifth lumbar vertebra (arrows). This research was original published in JNMT. Martin Gotthardt et al. J. Nucl. Med. Technol. 2013; 41:157–69. ©by the Society of Nuclear Medicine and Molecular Imaging, Inc.

In contrast to the good sensitivity in diagnosing DFO, a triple-phase bone scan is neither sensitive nor specific for V-OM (Figure 1). False negatives have been reported in the elderly, and although the exact mechanism is unknown, it has been postulated to be from atherosclerosis-induced ischemia. Another major limitation is that bony remodeling may persist for months after an inflammatory or infectious event, leading to false positive results on bone scan. Therefore, bone scanning should not be the only radionuclide modality used in V-OM. A bone scan can be combined with a Gallium – 67 scan to improve the specificity for the diagnosis of V-OM [6] (Figure 1).

Tagged WBC scans are not helpful for diagnosing V-OM. For unknown reasons, V-OM may demonstrate “decreased” uptake (photopenia) on tagged WBC scans compared with healthy bone, but this finding is not specific for infection and can occur with tumor, infarction, Paget’s disease, or even previously treated OM. On the other hand, if uptake is increased, it is virtually diagnostic of V-OM, but this occurs in less than 50% of proven V-OM [14].

Gallium (67Ga) scintigraphy used with planar or SPECT with CT imaging can be used adjunctively for the diagnosis of V-OM (Figure 1). Gallium does not need to be attached to leukocytes, as is the case with technetium, because gallium has a unique mechanism of uptake at sites of inflammation. Gallium binds transferrin and is extravasated at sites of inflammation due to increased vascular permeability. It then binds lactoferrin, which is secreted by leukocytes at sites of inflammation [15]. There can be increased uptake in trauma and tumors, limiting the specificity. In addition, there is a delay between injection of the radiotracer and the imaging test (up to 3 days) [6, 16].

Indium-111 biotin scintigraphy (111indium-biotin) is a newer method that may become a useful diagnostic test for V-OM in the future. However, 111indium-biotin currently is not commercially available. Unlike gallium tracers, it does not accumulate in the normal bone or bone marrow, and it is postulated to be more specific for bacterial infection because biotin seems to be passively transported into bacteria [17]. Single-photon emission CT/CT can be combined with the 111indium-biotin scintigraphy to improve anatomic localization as well as to help differentiate soft tissue infection from bone infection. By using SPECT/CT with 111indium-biotin, one study demonstrated a sensitivity and specificity of 92% in diagnosing V-OM [17].

The 18F-FDG-PET is the nuclear medicine technique most strongly supported by the literature for the diagnosis of V-OM. It can be especially helpful if the MRI is indeterminate due to degenerative disc disease. One study evaluated 18F-FDG-PET in 30 patients with vertebral endplate abnormalities on MRI and found that 18F-FDG-PET was 100% sensitive and specific for V-OM, and it was excellent at differentiating infection from degenerative changes [13]. Another prospective study evaluated the use of 18F -FDG-PET with the addition of CT scanning and demonstrated a sensitivity of 89%, specificity of 88%, and overall accuracy of 88% for diagnosing V-OM [18].

One advantage of 18F -FDG-PET scans is that they normalize within 3–4 months after trauma or surgery. Foreign implants (within 1 year of implantation) and malignancy can demonstrate increased uptake, limiting the specificity in those cases [6].

PROSTHETIC JOINT INFECTIONS

The definitive diagnostic test for PJIs are a cell count and culture obtained from synovial fluid via a joint aspiration. The decision to perform a joint aspiration in the setting of known hardware is complicated by the small risk of introducing bacteria to the joint space. In theory, a good clinical exam and history, as well as nonspecific laboratory findings such as an elevated C-reactive protein, will guide this decision, but certain imaging studies may provide additional evidence. Plain radiographs are of course the simplest and easiest study and may demonstrate joint space widening, radiolucency, migration, and osteolysis, but such findings can indicate an aseptic process as well. Moreover, x-rays are less sensitive in early-onset infections [19].

Magnetic resonance images are the preferred technique for most bone and joint infections, but the presence of hardware may result in artifact abnormalities. Recent metal suppression techniques such as multiacquisition with variable-resonance image combination or slice encoding for metal artifact correction have been used by specific producers of MRI machines and have greatly reduced the artifact abnormalities seen with older MRI machines, allowing for greater resolution and improved diagnosis of hardware failure with or without infection [20, 21]. However, not all imaging centers and hospitals have adopted such techniques, and nuclear medicine modalities may provide an option that is equally or more accurate.

The preferred nuclear medicine study for PJIs is the tagged WBC bone marrow scintigraphy with sulfur colloid, which has a reported accuracy of 86%–98% [19]. The most common radiotracer used in these studies is indium-111. As mentioned above, a positive result occurs when the signals conferred by the tagged WBCs and the sulfur colloid are discordant, because infections will suppress the uptake of sulfur colloid [22]. The data on 18F-FDG-PET/CT have been mixed and are not useful within the first year of implantation, because postoperative inflammation will result in false-positive results [19].

ENDOVASCULAR GRAFT INFECTIONS

Perhaps one of the more pressing diagnoses to exclude in our patient is the possibility of an endovascular graft infection. Endovascular graft infections are difficult to manage medically and almost always necessitate surgical intervention. If surgical intervention is not a possibility, chronic suppression with antimicrobials is a second-line option. Given the morbidity of the disease as well as the morbidity of the intervention, establishing an accurate diagnosis is imperative.

The gold standard in diagnosing mycotic aneurysms has been CT angiography (CTA). However, the test characteristics of CTAs in diagnosing endovascular graft infections, particularly soon after surgery, are poor. Although air bubbles around the site of the graft may suggest infection, 50% of CTAs will have this finding for weeks to months after surgery. Hematomas and lymphoceles may also reduce the specificity of CT scans, and the sensitivity is particularly poor in low-grade infections [1]. Tagged white blood scans have been evaluated as a diagnostic method with sensitivities ranging from 50% to 100% [23, 24]. However, the limited spatial resolution of tagged WBCs may decrease its sensitivity for low-grade infections considerably.

The 18F-FDG-PET/CT demonstrates excellent test characteristics, with both a sensitivity and specificity of over 90%, provided that strict interpretation criteria are used (Figure 2) [25]. There may be chronic aseptic inflammation around the synthetic graft material that results in mild to moderate uptake, but a focal area of abnormal uptake will improve the specificity and positive predictive value. In one study, focally intense lesions on PET combined with an irregular boundary noted on CT predicted 97% of prosthetic graft infections [26].

Figure 2.

18F-FDG PET/CT scan of patient with provend Escherichia coli infection of vascular graft. This research was original published in JNMT. Martin Gotthardt et al. J. Nucl. Med. Technol. 2013; 41:157–69. ©by the Society of Nuclear Medicine and Molecular Imaging, Inc.

The degree of FDG avidity may predict infection to some extent. For example, mycotic aneurysms alone are in general more FDG avid (greater than 4.5 standardized uptake values) than uninfected aneurysms [27]. Specific cutoff criteria have not been evaluated for endovascular graft infections, and this is an area that merits further research.

FEVER OF UNKNOWN ORIGIN

Our patient does not meet criteria for fever of unknown origin (FUO) because he has positive blood cultures, but there may be some parallels with the workup of FUO given the unclear source of the bacteremia. This section will review the utility of various nuclear medicine studies for the evaluation of an FUO.

There are many published diagnostic algorithms for FUO, many of which include nuclear medicine tests [28]. Nuclear medicine imaging tests may be particularly well suited for imaging in FUO because they can image the entire body at once. Most modalities can identify infections, malignancies, and inflammatory states such as large vessel vasculitides, which together comprise the majority of FUO etiologies. Furthermore, because nuclear medicine tests often detect functional and metabolic changes, they may become positive early in the course of a disease before morphologic changes take place and can be detected on conventional imaging such as plain radiography or CT [29].

Two-dimensional gallium scintigraphy was previously considered the nuclear medicine test of choice for FUO. However, it is time consuming and exposes the patient to a large amount of radiation. It has largely been replaced by 18F-FDG-PET scanning because of the higher sensitivity and specificity compared with a gallium scan, in addition to the better spatial resolution [30].

A major limitation is that 18F-FDG has physiologic uptake in the brain, heart, bowel, urinary tract, and even occasionally in the bone marrow, limiting its usefulness in these areas [1]. The 18F -FDG-PET may be even more useful if combined with CT to allow better anatomic localization. One study retrospectively evaluated 112 patients with FUO who underwent 18F-FDG PET/CT and found that 18F-FDG-PET/CT contributed to the diagnosis 66% of the time, and a final diagnosis was made in 74% of the patients [29]. Other studies have shown that FDG-PET/CT contributed to the diagnosis of FUO 46%–90% of the time, more often than other modalities [29, 31, 32].

The 18F-FDG-PET/CT scanning has also been shown to be very helpful in finding sites of metastatic infections during Gram-positive bacteremia, as in our patient, and has been suggested as a diagnostic tool in all patients with high-risk Gram-positive bacteremia as well as those at risk for disseminated infection [1].

Tagged WBC scanning may not be helpful in cases of FUO because the leukocytes that are labeled are neutrophils, which mainly accumulate in foci of bacterial infections; typical bacterial infections make up a minority of causes of FUO (15%–40%) [1]. The sensitivity and specificity of tagged WBC reported in the literature varies widely from 60% to 100% and 33% to 92%, respectively [33].

If there is strong suspicion for a bacterial infection, a tagged WBC scan may be helpful. A tagged WBC scan can be done with various radiotracers including indium or technetium, and each has different properties; often the decision of which radiotracer to use will be made in conjunction with a radiologist. The physiologic distribution of indium-111 is limited to the liver, spleen, and bone marrow, whereas 99mTc has physiologic uptake in the urinary tract, large bowel, and occasionally gallbladder. The presence of physiologic uptake may make the utility for diagnosing infection in these areas less ideal. A scan using 99mTc is faster, requiring only several hours after the reinfusion of WBC into the patient compared with up to 30 hours using indium-111.

CONCLUSIONS

There are a variety of nuclear medicine techniques available to complement traditional imaging and assist with the diagnostic workup of complicated infectious disease syndromes such as the case presented in this review (Table 1). For this gentleman with multiple potential infectious etiologies of his fever and sources of his MSSA bacteremia, a tagged WBC scan would be a good first diagnostic test, but the limited resolution may make it difficult to diagnose endovascular infections, which arguably is the more pressing diagnosis to make. A tagged WBC scan combined with a sulfur colloid bone marrow imaging may help distinguish his DFO from Charcot, as well as determining his likelihood of a PJI, because both may have implications for treatment duration or the need for further surgical management. However, if the tagged scan is negative, we would recommend trying to obtain an 18F-FDG-PET/CT quickly to assess for the presence of an endovascular infection.

Table 1.

Overview of Nuclear Medicine Studies and Clinical Applications

Modality	Mechanism	Attributes	Applications
Tagged WBC	WBCs are separated from blood and tagged with indium or technetium	Cost of tagging WBCs is high (>$1000) but typically covered by insurance companies if infection is an indication.Requires at least 2 days (harvest of WBC then infusion).May be best used when paired with other modalities	DFOVGIPJI (with sulfur colloid scintigraphy or bone marrow scintigraphy
Positron emission tomography (PET)	Gamma cameras detect positron emitting radioisotope (18F) and create 3D images.Metabolic and functional informationF-FDG (glucose)	Cost of FDG is low (>$100) but currently not covered by insurance companies for infectionFair spatial resolution (3–5 mm).Good contrast resolution.Moderate radiation, depending on radiotracer, but high radiation when combined with CT	DFO- limitedV-OMVGIFUOCan be combined with CT for better spatial resolution
Single-photon emission tomography (SPECT)	Gamma cameras detect gamma emitting radioisotope and create 3D images.Metabolic and functional information	Less expensive than PETLimited spatial resolution compared with PET (8-10mm)Good contrast resolutionLow radiation exposure, depending on radiotracer, but high radiation when combined with CT	Coronary artery diseaseV-OMPossible utility in DFO (over PET due to cost and no need for FDG).Needs to be combined with CT for spatial resolution
Scintigraphy (planar)• Gallium• Technetium-99m• Indium-biotin	Gamma cameras detect radioisotope and create 2D images.Bone scintigraphy is also known as bone scan	Often paired with other imaging modalities to improve test characteristics, ie, triple phase bone scan.High radiation exposure with Gallium.Indium-biotin maybe more specific for infection	DFOV-OM (Ga)FUO (if PET/CT unavailable)

Abbreviations: CT, computed tomography; DFO, diabetic foot osteomyelitis; FDG, fluorodeoxyglucose; FUO, fever of unknown origin; Ga, gallium; OM, osteomyelitis; PJI, prosthetic joint infections; VGI, vascular graft infections; V-OM, vertebral osteomyelitis; WBC, white blood cells; 3D, 3 dimensional.

Like any other diagnostic test, false positives can occur in many conditions such as trauma, autoimmune diseases, or malignancy, and nuclear medicine imaging should always be interpreted in the appropriate clinical context. Other infection-specific nuclear medicine modalities, such as radiopharmaceutical-labeled antibacterials, antimycobacterial drugs, antifungal drugs, and even antimicrobial peptides, are currently under investigation and may at some point offer even greater accuracy [34]. Early studies are promising, but more research is needed, including cost-benefit analyses, before they will be routinely available.

A discussion about radiographic imaging techniques is incomplete without a brief mention of the relative radiation exposure of different modalities (Figure 3). In general, the level of radiation exposure conferred by nuclear medicine techniques is similar to that of CT scans, with a range of 2–15 milliSieverts (mSv). The International Commission on Radiologic Protection recommends limiting artificial radiation to no more than 50 mSv a year. All of the modalities discussed in this paper have radiation doses ranging from 6 to 15 mSv, 18F-FDG-PET and gallium scintigraphy have the highest levels of radiation with 14 and 15mSv, respectively, comparable to a CT done to rule out pulmonary embolism (15 mSv) or a coronary angiography (16 mSv) [35].

Figure 3.

Radiation exposure of various imaging modalitites (computed tomography [CT] vs nuclear medicine). CTA, CT angiography; F-FDG-PET, fluorodeoxyglucose-positron emission tomography; PE, pulmonary embolism; WBC, white blood cells.

Lastly, addressing the issue of the cost of nuclear medicine imaging is important and complex. Different institutions carry different insurance payment structures that may preclude its use in the inpatient setting, where nuclear medicine studies are more often needed. The Centers for Medicare and Medicaid Services has not approved coverage for PET CTs for the indication of diagnosing infectious diseases [36]. Estimated individual out-of-pocket expenses for PET CT scans are approximately $7000 [37], which make it cost prohibitive for most patients. Further research is needed to address the cost-benefit ratio to make such studies more accessible.

In conclusion, when MRIs or CTs are inconclusive or contraindicated, 18F-FDG PET/CT, with its excellent test characteristics and anatomic localization, appears to be an excellent imaging modality that can be used to include or exclude many infectious disease syndromes. Early prosthetic joint or diabetic foot infections may be better diagnosed with tagged WBC scan with sulfur colloid imaging. In our experience, different centers have different comfort levels and expertise in using these various imaging modalities for the purpose of diagnosing infections. Additional research is also needed to create more “standardized” algorithms for using nuclear medicine imaging studies in the diagnosis of infectious diseases.