Comparison Between Tc-99m WBC SPECT/CT and MRI for the Diagnosis of Biopsy-proven Diabetic Foot Osteomyelitis

Objective. Magnetic resonance imaging (MRI) is the recommended diagnostic imaging technique for diabetic foot osteomyelitis (DFO), with a reported accuracy of 79%. The gold standard to diagnose osteomyelitis is bone biopsy, with a positive culture and/or histopathology findings consistent with osteomyelitis. The purposes of this study are to assess the accuracy of technetium-99m (Tc-99m) labeled white blood cell (WBC) single-photon emission computed tomography/computed tomography (SPECT/CT) hybrid imaging for diagnosing DFO confirmed by bone biopsy and to compare that to the diagnostic accuracy of an MRI. Materials and Methods. The authors performed a retrospective chart review of 166 patients who received a bone biopsy to confirm the diagnosis of a suspected DFO at a large municipal hospital between 2010 and 2013. Patients were selected on the basis of whether they received an MRI or a SPECT/CT. Patients whose scans were not within a clinically relevant time frame of the biopsy were excluded. Imaging results were correlated with probability of osteomyelitis determined by bone biopsy. Results. For inclusion criteria, 110 patients met the study’s criteria: 52 SPECT/CT patients and 58 MRI patients. The sensitivity, specificity, positive predictive value, and negative predictive value of SPECT/CT were 89%, 35%, 74%, and 60%, respectively; the corresponding values for MRI were 87%, 37%, 74%, and 58%, respectively. There were no significant differences in accuracy of diagnosing DFO between imaging techniques. Conclusion. This data suggests that the diagnostic accuracy of SPECT/CT imaging in DFOs is similar to an MRI. 

The global prevalence of diabetes mellitus is rising. Diabetes was estimated to effect approximately 366 million people worldwide in 2011. Experts predict 552 million individuals will be diagnosed with the disease worldwide by 2030.¹ Infected diabetic foot ulcers (DFUs) are one of the leading causes of hospitalization and amputation.² The economic and social impact of having an infected DFU are tremendous, as foot problems account for the largest number of bed days used by patients with diabetes.¹ The risk of amputation increases significantly if there is osteomyelitis. Compared to patients with a soft tissue infection, patients with diabetic foot osteomyelitis (DFO) have a threefold higher risk of undergoing minor amputations (59.4% vs 13.8%, P < 0.001) and are treated longer with intravenous antibiotics (32.3 days vs 13.2 days, P < 0.001).³ After amputation, patients often experience limited function, mobility, and quality of life. For instance, it has been estimated that 77% of patients older than 75 years of age who undergo amputation do not return to independent living.⁴ 

An accurate diagnosis of DFO is essential to institute appropriate therapy and to avoid unnecessary treatments of patients who do not have a bone infection. The Infectious Diseases Society of America (IDSA) guidelines recommend using magnetic resonance imaging (MRI) to diagnose DFO, although the guidelines acknowledge bone biopsy for pathology and microbiology is the “gold standard.”⁵ Magnetic resonance imaging is considered to be the best imaging technique for the early diagnosis of DFO with a pooled sensitivity of 0.90 (95% CI, 0.82-0.95), a specificity of 0.79 (95% CI, 0.62-0.91), and a diagnostic odds ratio of 24.4.⁶ Unfortunately, MRI has often been reported to have low specificity and positive predictive value when noninfectious changes are present, especially in patients with previous foot surgery, trauma, or Charcot neuroarthropathy.^7-9 In some studies,^10-12 the specificity and positive predictive values (PPV) of MRI are about the same as flipping a coin. 

Among the different types of nuclear imaging techniques, the 3-phase bone scan might be one of the most sensitive to diagnose osteomyelitis.¹³ However, its utility is best when no confounding factors such as degenerative or reparative changes are in the bone. In addition, planar bone images have suboptimal spatial resolution and the specific site and extent of infection in the foot maybe difficult to assign. Technetium-99m (Tc-99m) white blood cell (WBC) single-photon emission computed tomography/computed tomography (SPECT/CT) is a relatively new imaging technology that combines radiolabelled WBCs and high-resolution X-ray CT. This technique has the advantage of the CT-scan, which improves the assessment of the WBC intensity and the extent of infection into the bone. Also, the potential to grade WBC intensity and declining intensity might be used as a prognostic tool predicting success of treatment.¹⁴ The diagnostic accuracy and prognostic value for DFO has not yet been compared to MRI using bone biopsy as the gold standard. It is the authors’ hypothesis that the accuracy of the hybrid image Tc-99m labeled WBC SPECT/CT for diagnosing DFO is not inferior to MRI.

The authors retrospectively reviewed 166 medical charts of patients with diabetes admitted to a tertiary care hospital (University of Texas Southwestern Medical Center, Dallas, TX) between 2010 and 2013 with clinically suspected DFO confirmed by a bone biopsy. Patients were included when they received either a SPECT/CT or MRI within a clinically relevant time frame of biopsy, which the authors determined to be 8 weeks after consensus with the radiology department at the authors’ institution, as no timeframe is determined in the medical literature. Other medical data points were collected from the chart for each subject including demographic information, diabetes type, laterality and location of the foot ulcer, surgical procedures performed within 6 months of imaging, imaging results, and microbiological and histopathology data of the bone specimens. A positive osteomyelitis occurrence was defined as positive culture results and/or histopathology changes consistent with osteomyelitis in the obtained bone samples. Patients were allocated either to the SPECT/CT or MRI group. Patients who received both SPECT/CT and MRI within the relevant time frame were also included. Institutional review board approval was obtained and informed consent was waived prior to initiation of this study. 

Imaging methods and display technique
The authors obtained SPECT/CT images according to Erdman et al.¹⁴ The SPECT/CT images were acquired using a Symbia T Series SPECT/CT (Siemens Medical Systems, Malvern, PA) dual-head mounted with low-energy, high-resolution collimators, 2 hours after injection of 20–25 mCi of radiolabeled autologous Tc-99m WBC. In the reports, WBC activity that abutted the bone cortex or extended into the marrow space was reported as osteomyelitis, whereas WBC activity localized only in soft tissue was reported as no osteomyelitis. 

Magnetic resonance images were performed with 1.5 T MR scanners (Siemens Medical Systems, Malvern, PA). Magnetic resonance imaging parameters were as follows: slice thickness 3 mm, conventional spin echo, repetition time/echo time 460 ms/13 ms; fat-suppressed T2-weighted images, short-tau inversion recovery images, short T1 inversion recovery, and repetition time/echo time/inversion time 3401 ms/80 ms/150 ms. An extremity coil was used for all feet. Depending on the clinical findings either the forefoot or the hindfoot, including the ankle, was imaged separately. Imaging of the forefoot contained a field of view ranging from the digits to the Lisfranc joints (field of view, 12-16 cm), and imaging of the hindfoot encompassed the foot from the Lisfranc joints to the ankle (field of view, 14-16 cm). Images of both regions consisted of T1, T2, and fat-suppressed images. Short T1-inversion recovery and contrast were used as clinically indicated. The affected bone marrow was compared with the adjacent normal fatty marrow: low-intensity signals on T1-weighted images and high-intensity signals on fat-suppressed T2-weighted images were attributed to osteomyelitis. Incomplete or hazy signals or reticulated patterns were attributed to reactive bone marrow edema. Normal bone marrow signals were considered indicative of areas clear of disease. 

Bone biopsy technique
All subjects received a biopsy using a 14-gauge Jamshidi needle at the site of osteomyelitis. The biopsy area was prepared and draped in sterile technique. The Jamshidi needle was introduced at least 2 cm away from the ulcer site to avoid soft tissue contamination. Using fluoroscopy, the bone biopsy needle was then guided inferiorly through the bone until it just penetrated the distal cortex, to avoid exposure to infected soft tissue. Two specimens were obtained: 1 for culture and 1 for histology. Bone specimens from patients who required surgical debridement or amputation were collected at the time of surgery. 

Statistical analysis
Continuous variables were described using mean and standard deviation and categorical variables were described using group size and percentage. The diagnostic performance of the imaging procedures was assessed using sensitivity, specificity, accuracy, PPV, and negative predictive value (NPV). Analysis was performed with SPSS statistical software version 17.0 (SPSS Inc, Chicago, IL).

For this study, 110 patients met the inclusion and exclusion criteria. All subjects had been diagnosed with diabetes mellitus and peripheral neuropathy prior to inclusion. Demographics of the study population are shown in Table 1. There were no significant differences between the SPECT/CT group (n = 52) and MRI group (n = 58) in all reported demographics except in the type of diabetes (P = 0.012). There were 4 negative studies in the SPECT/CT set and 12 negative studies in the MRI set. Five MRI scans had indeterminate results, giving neither a positive or negative diagnosis of DFO. Of those, 2 bone biopsies were negative and 3 were positive. In the authors’ analysis, these indeterminate results were treated as negative MRI studies. 

The probability of the different imaging techniques to identify osteomyelitis (ie, a positive osteomyelitis occurrence) was determined by comparing the SPECT/CT and MRI results with the results of the bone biopsy. Table 2 demonstrates the outcome measurements of SPECT/CT and MRI. The sensitivity and specificity of SPECT/CT to diagnose osteomyelitis was 89% and 35%, and the PPV and NPV were 74% and 60%, respectively. The sensitivity and specifity of MRI were 87% and 37%, and the PPV and NPV were 74% and 58%, respectively. There was no significant difference in the accuracy of SPECT/CT and MRI (0.65, CI 0.57-0.83 vs 0.71, CI 0.61-0.85). The cross-tabulation of imaging results and biopsy results are shown in Table 3. 

Because the point prevalence of osteomyelitis in the MRI and SPECT-CT subjects, as judged by bone biopsy (67.3% vs 67.2%), is not significantly different according to a 2-sided test of proportions (P = 0.994), the authors feel confident in comparing the accuracy numbers of the 2 tests. Comparing these parameters between MRI and SPECT tested samples with 2-sided test of proportions obtained P values of 0.85, 0.92, 0.99, 0.93, and 0.95, for sensitivity, specificity, PPV, NPV, and accuracy, respectively. 

The results of this study demonstrate SPECT/CT and MRI had similar outcomes in diagnosing biopsy-proven DFO. Bone biopsy is the gold standard for the diagnosis of DFO; however, lack of training to perform bone biopsy, invasiveness of the procedure, and cross contamination are limitations to the procedure. In the present study, both SPECT/CT and MRI had high sensitivity (89% and 87%), low specificity (35% and 37%), modest PPV (74% and 71%), and poor NPV (60% and 58%). The PPV reports the proportions of positive results that are true positives (SPECT/CT 74%; MRI 71%); therefore, when the imaging result is positive for DFO, there is a high likelihood that culture and pathology of the bone will also be positive. The PPV of a test depends on the pretest probability of osteomyelitis.¹⁵ In this study, the pretest probability of having osteomyelitis was high since the study population was an inpatient population with a high index of suspicion for DFO. In fact, 67% of patients in both imaging sets had DFO based on bone biopsy results. Thus, the results of this study are not generalizable to populations with a low pretest probability of DFO such as patients seen in an outpatient setting. The same selection bias is found in other studies that combine SPECT/CT imaging with bone biopsy results. For example, in Aslangul et al,¹⁶ Gallium-67 SPECT/CT imaging was combined with bedside percutaneous bone puncture in 55 patients, resulting into a tool with a PPV of 91.7% and NPV of 90.7%. However, by selecting patients with positive SPECT/CT results, Aslangul and colleagues¹⁶ used a patient population for their study with a high pretest probability of DFO, thus positively affecting the accuracy numbers of the imaging.

When comparing the outcomes of the present study with previous literature, the sensitivity, specificity, PPV, and NPV of MRI and SPECT/CT are quite different (Table 4). One of the main reasons for limited specificity is the presence of reactive bone marrow edema, including acute Charcot neuroarthropathy and recent surgery, which can last 3-6 months.¹⁷ Most published MRI studies have small sample sizes (n < 40), and they use clinical criteria as the referenced diagnostic standard rather than a gold standard of bone culture and/or histopathology (Table 4). The clinical criteria most commonly used include “the probe-to-bone test,”¹⁸ clinical appearance of the site of interest,¹⁹ or wound healing. The methods and results of the present study question the reliability of other studies that use less stringent clinical parameters as their reference standards. 

Technetium-99m labeled WBC SPECT/CT is a relatively new test, and there are only a handful of studies that evaluate this imaging technique in DFO. The validity and reliability of Tc-99m WBC SPECT/CT has been determined in a small pilot study by Przybylski et al.²⁰ Fourteen patients underwent SPECT/CT for suspected DFO, confirmed by histopathology criteria when available (n = 9). The sensitivity and specificity of Tc-99m WBC-labeled SPECT/CT were 87.5% and 71.4%, respectively. Positive predictive value and NPV were 77.8% and 83.3%, respectively, with an accuracy of 80%. Since Przybylski and colleagues²⁰ did not explicitly define the imaging diagnosis of osteomyelitis, the authors of the present study are not able to determine whether a more stringent definition was used. Given the published results²¹ showing increased specificity with delayed time-point imaging in a multicenter trial, the authors realize specificity can be higher.  Also, previous studies have confirmed the value of SPECT/CT to diagnose inflammatory bone lesions,²² but most publications have focused on large osseous structures instead of the type of small lesions typical of DFO.²³ In a study by Meawal et al,²⁴ Tc-99m WBC SPECT/CT was used to refine the diagnosis of osteomyelitis among a cohort of patients with nonspecific and clinically incongruent MRI results. In a subset of 9 patients with nonspecific MRI results, antibiotic treatment was modified in 89% of patients with Tc-99m WBC SPECT/CT. 

The most obvious limitations of this study are the retrospective design and the time interval selected between the obtained bone sample and imaging. Ideally, the authors could have done both imaging studies during a defined time frame and then performed surgery to obtain the bone specimens for culture and histopathology. This approach would have eliminated the confounding factors involved in comparing 2 separate groups of patients with DFO. However, the authors feel this study resembles clinical practice. Most practitioners will only perform a bone biopsy once they have the imaging results, most likely to avoid an invasive procedure or possible introduction of bacteria in the bone in the presence of an open wound. Besides that, the current evidence that evaluates the correct time frame between the 2 tests is scarce. In a study by Yuh et al,²⁵ abnormal signal intensity lasted for several months on Tc-99m methyl diphosphonate bone scintigraphy after responsiveness to intravenous antibiotics treatment. 

In this paper, the authors remark on several drawbacks to the above analysis, all circumstances of the fact that osteomyelitis is relatively uncommon, making larger studies difficult. First, it would be ideal to compare the 2 tests in 1 population. That is, ideally, the study would have had 1 group of patients on whom both MRI and SPECT/CT were assessed. The authors could then, directly in paired analysis, compare the sensitivity and specificity of the 2 tests using methods as in Trajman and Luiz²⁶ and Hawass.²⁷ This did not happen in the current study as performing both studies was not standard of care. For the purposes of the present study, it was decided to compare using the unpaired proportion test. Second, once that was decided, the authors also had to decide what to do with those unusual patients who had both tests, as they contribute doubly to the analysis. As there were only 14 of these patients, it was decided to ignore the possible influence they might have and treat their observations under MRI and SPECT/CT as if they were from separate patients. Third, given the comparison with a proportion test, and the inability to show within the limits of this study that SPECT/CT was superior, the authors would have liked to show it was at least noninferior to MRI. Unfortunately, sample size precludes doing this, and the study is underpowered to see such similarities as those observed here as significant. For example, a rough estimate shows that, given a test with 90% sensitivity, in order to see with significance 0.05 and power 80% that a second test has sensitivity no less than 85% requires 445 patients for each of the 2 tests. 

The data gathered from this study suggests Tc-99m labeled WBC SPECT/CT and MRI provide similar test results when compared to bone biopsy as the referenced gold standard.⁵ Both techniques have high sensitivity but only modest PPV in the study’s patient population. Imaging studies provide information that supports or refutes the clinical suspicion with variable certainty. However, defining their exact role remains challenging in the absence of a prospective study comparing the imaging studies to a gold standard and, indeed, the limitations of the gold standard. At the least, the authors hypothesize both imaging techniques will be useful in practices with limited biopsy options, cases in which clinical judgment suggests a pretest likelihood of osteomyelitis or where ruling out osteomyelitis would spare the patient an undesirable therapy. Further, the authors suggest that since bone marrow edema, the chief MRI biomarker for osteomyelitis, may be seen in other sequelae of diabetic foot complications and are slow to resolve, radiolabeled WBC SPECT/CT may be superior in these cases, and this should be the basis for a prospective study. The care of the infected diabetic foot requires a combination approach: clinical assessment, laboratory tests, and imaging methods based on the needs of individual patients. However, laboratory tests and imaging studies should not replace clinical assessment, but rather be used as an integrated modality for the correct diagnosis of DFO. 

Affiliations: Department of Plastic Surgery; Department of Radiology; and Department of Internal Medicine, Infectious Diseases, University of Texas Southwestern Medical Center, Dallas, TX

Correspondence:
Javier La Fontaine, DPM, MS 
Department of Plastic Surgery
University of Texas Medical Center
Dallas, TX
Javier.lafontaine@utsouthwestern.edu

Disclosure: The authors disclose no financial or other conflicts of interest.