Feasibility of Hybrid 67Gallium-Citrate SPECT/CT Fusion Imaging for Identifying Patients with Osteomyelitis Awaiting Ostectomy

Purpose Co-existence of ischemia and osteomyelitis strongly increase the risk of limb amputation, however, there is no established non-invasive assessment. This single-center cohort study used propensity score-matched analysis. We evaluated the clinical feasibility of hybrid 67 Ga-citrate single photon emission computed tomography and X-ray computed tomography (SPECT/CT) for early diagnosis, severity assessment, and prognosis determination in these patients. Method We enrolled consecutive patients with suspected osteomyelitis with mean follow-up of 5.5 years. All patients underwent 67 Ga SPECT/CT before and 4 weeks after treatment. Osteomyelitis diagnosis was based on histopathology and bacteriology of surgical sample, bone probing, or imaging follow-up. The diagnostic accuracy of bone resection rates and long-term prognoses were determined based on the target-to-background ratio (TBR). demonstrates the usefulness and feasibility of fusion TBR for improving diagnostic quality among patients with osteomyelitis awaiting bone resection even in ischemic conditions. Coupling of 67 Ga-citrate hybrid SPECT/CT fusion imaging by TBR analysis revealed severity thresholds (TBR 10.1) for bone resection and aided in anatomical localization;


Background
Antibiotics or bone resection are the recommended approaches for foot osteomyelitis; however, major limb amputation is often necessary in cases complicated by peripheral artery disease (PAD). The coexistence of ischemia and infection strongly increases the risk of limb amputation, especially in patients with diabetes. Indeed, these factors often interact to decrease the rate of limb salvage. In such patients, the faster rate of progression observed due to wound healing is prohibited by reduced blood ow, and underlining complications such as diabetes show poor prognosis even after limb amputation [1,2]. The European Society of Cardiology guidelines indicate these combined conditions as chronic limbthreatening ischemia (CLTI) [3]. To reduce major limb amputation in CLTI, immediate and accurate decision-making for bone resection is important. Furthermore, the severity of infection in uences both prognosis and the duration of bone resection procedures. However, previous studies offer limited information regarding diagnostic technique, as they have excluded patients with ischemic osteomyelitis [4,5] or have utilized relatively short observation periods [6]. In addition, although soft tissue cultures can be an inaccurate and invasive method, bone biopsies are rarely performed. While noninvasive imaging modalities are suitable, the speci city of magnetic resonance imaging (MRI) is relatively low under ischemic conditions [7,8]. The appropriate delivery of antibiotics is also di cult in CLTI, and precise tools for determining the need for early surgical intervention have yet to be developed [9,10].
Previous research has indicated that bone scanning with 67 Ga-citrate can be used to evaluate the extent of in ammation; however, this method is associated with poor resolution when a recommendation is rated as Grade 2B [11]. In the present study, we aimed to determine the usefulness and feasibility of hybrid 67 Ga-citrate single photon emission computed tomography and X-ray computed tomography (SPECT/CT) fusion imaging and target-to-background ratio (TBR) scores for improving diagnostic quality in patients with suspected osteomyelitis including type 2 diabetes and CLTI.

Study population and protocol
From January 2012 and July 2017, we enrolled consecutive patients with suspected osteomyelitis with foot gangrene or ulcers. The physicians suspected osteomyelitis based on the following conditions and then performed a comprehensive diagnostic evaluation: (a) histopathology, bacteriology of surgical samples, (b) imaging follow-up through bone erosion and/or osteolysis on X-rays, and (c) morphological bone destruction detected by probing. Surgical bone resection was performed following bone probe analysis when patients experienced severe, intractable pain at rest or rapid serological worsening of in ammatory changes despite treatment. Those patients were followed up at outpatient in same hospital. Exclusion criteria were as follows: (a) active systemic infection requiring immediate amputation or softtissue infection exist above foot joint, (b) inability to participate in initial radionuclide imaging studies, (c) any type of blood circulation re-constructive therapy within the preceding month.

Radionuclide detection of osteomyelitis
All patients underwent 67 Ga SPECT/CT using a SPECT/CT hybrid system, which combined a dual-head gamma camera with a two-row multi-section computed tomography (CT) scanner: Symbia T2 (Siemens Healthineers Japan, Tokyo, Japan). SPECT images were acquired at rest, 24 hours after intravenous injection of 67 Ga-citrate 148 MBq (the half-life of 67 Ga is 3.26 days). The 67 Ga SPECT images were acquired over 10 min per bed position (30 projections over an orbit of 180˚, 6˚ per step, and 20 steps per projection). The acquisition range included the entire femur, lower leg, and foot. A middle-energy general purpose collimator was used for image acquisition (matrix: 128 × 128 pixels). The Flash 3D iterative image reconstruction algorithm was used to reconstruct 67 Ga SPECT images. The reconstruction parameters for the numbers of subsets and iterations were 6 and 8, respectively. Non-contrast-enhanced CT images (tube voltage: 110 kVp; tube current time product: 10-40 mA; detector con guration: 2 × 4 mm; matrix: 256 × 256 pixels; reconstruction thickness: 5 mm from femur to feet and 3 mm from ankle joint to toe) were also acquired to enable CT attenuation correction and create 67 Ga SPECT/CT images.
When accumulation of 67 Ga-Citrate was found adjacent to chronic ulcers/gangrene, we performed assessments to determine whether such changes re ected osteomyelitis or connective tissue in ammation, depending on the anatomical information acquired via CT during the same exam sequence. When the area of 67 Ga-Citrate accumulation and bone sites matched, the patient was diagnosed with osteomyelitis. To analyze the data acquired from hybrid 67 Ga SPECT/CT images, regions of interest (ROI) were drawn around the bone(s) with suspected osteomyelitis. The radionuclide count within the ROI was determined, along with the intact femoral uptake radionuclide count (background).
The TBR was de ned as the count of 67 Ga-citrate accumulation per pixel in the ROI divided by the count per pixel in the femur [12]. Reconstruction and fusion imaging were managed by using the Symbia T2 workstation (Siemens Healthineers Japan). The count of 67 Ga-citrate accumulation per pixel was analyzed by using a medical image viewer (Natural Viewer, Hitachi Co., Ltd., Tokyo, Japan). Examinations were performed at the initial stage of treatment and approximately 4 weeks after the rst examination as a follow up.

Ischemia evaluation
Tissue ischemia was de ned by transcutaneous oxygen tension (TcPO 2 , less than 50 mmHg).

Patient outcomes
To determine the clinical turning point using an appropriate scoring system, we adopted the incidence of bone resection (minor amputation) as the primary end point. Secondary endpoints included limb survival, which was de ned as retention of the bilateral calcaneus bones, and the occurrence of major adverse events including cardiovascular events, cerebrovascular events, and all causes of mortality.

Statistics
Within-treatment analyses of changes were performed using Wilcoxon rank-sum tests. Cumulate survival was determined via Cox hazard analysis with Kaplan-Meier analysis. We also evaluated the diagnostic accuracy of 67 Ga SPECT/CT based on the results of sensitivity, speci city, and receiver-operating characteristic (ROC) curve analyses. The area under the ROC curve and standard deviation were obtained by assuming a nonparametric distribution. ROC curves were constructed for the occurrence of bone resection due to osteomyelitis. Propensity score-matched multivariate regression analysis was carried out using the incidence of bone resection as the dependent variables, such as sex, age, hemodialysis, type 2 diabetes, history of cardiovascular disease, and TcPO 2 , in Cox regression analysis. The hazard ratio (HR) is expressed as a 95% con dence interval (CI). The level of statistical signi cance was set at p < 0.05. All statistical analyses were performed using SPSS statistics software (version 25, IBM Corporation, Armonk, NY, USA).

Results
We enrolled 90 patients with suspected osteomyelitis. The baseline characteristics of the included participants are shown in Table 1. A representative 67 Ga SPECT/CT fusion image is shown in Fig. 1. The fusion image demonstrated the location of osteomyelitis with better resolution than the 67 Ga SPECT image alone. The mean duration of follow up was 5.5 ± 0.3 years. The overall rate of limb salvage was 80%, and the rate of survival was 80% (Fig. <link rid=" g2">2</link>-A and 2-B). Among patients who had undergone hybrid 67 Ga SPECT/CT in the initial stage, 27 (30%) underwent bone resection (Table 1). TBR values were signi cantly higher in the bone resection group (average TBR: 20.5 ± 12.5) than in the nonbone resection group (average TBR: 4.3 ± 3.7) (p < 0.05). The relationship between TcPO 2 and TBR is shown in Fig. 2-C. Major limb amputation was necessary in six (22%) and nine patients (14%) in the bone resection group and the non-bone resection group, respectively (p = 0.26). Death occurred in four (15%) and 10 cases (16%) in the bone resection group and the non-bone resection group, respectively (p = 0.59).
The average pre-treatment TBR of the entire group was 9.1 ± 12.3, while the average follow-up TBR was 4.3 ± 3.8 (p < 0.01, Wilcoxon test, Fig. 3-A), which indicates that medical interventions signi cantly reduced 67 Ga-citrate accumulation. In the treatment base analysis, the pre-treatment TBR cutoff for the incidence of bone resection due to osteomyelitis was 10.1 (sensitivity: 0.85, speci city: 0.82, area under the curve: 0.87, predictive accuracy: 0.86). Bone resection was performed in 25 of the 36 patients in the TBR ≥ 10.1 group (69.4%), and in two of the 52 patients in the TBR < 10.1 group (3.8%) (p < 0.001, chisquare test; Fig. 3-B). The incidence of bone resection was expressed via Kaplan-Meier analysis (p < 0.01, log-rank test, Fig. 3-C). Regarding the radiographic reliability of the TBR assessment, the linear weighted κ values indicated a moderate interobserver agreement of 0.71 and an intraobserver agreement of 0.67, based on previous ndings [13]. A multivariate regression analysis was performed using the incidence of bone resection as a dependent variable. After the propensity score matching, TBR of more than 10.1 was a strong risk factor for bone resection (p < 0.001, HR 9.17, 95% CI 3.14-27.0). However, hemodialysis, type 2 diabetes and lower TcPO 2 did not show statistical signi cance (Fig. 3-D). Additionally, the Cox hazard analysis indicated that this cutoff shows signi cant difference in the detection limit both in CLTI (p < 0.001, odds ratio 4.88, 95% CI 1.35-17.7) and non-CLTI (p < 0.001, odds ratio 5.15, 95% CI 1.81-14.6). Representative Case Figure 4 is a successful representative case without bone resection, which shows the changes in TBR and the location of osteomyelitis before and after treatment. The patient was a 68-year-old man with type 2 diabetes (HbA1c 7.4%, National Glycohemoglobin Standardization Program). He was referred to our hospital with a foot ulcer that was suspected osteomyelitis, induced by a low-temperature burn, which had been present for 4 months. He originally lost four left toes during childhood, although this was not related to diabetes. There was no evidence of limb ischemia (ABI = 1.23). Pre-treatment 67 Ga SPECT/CT fusion imaging revealed osteomyelitis at the distal end of the left fourth metatarsal bone. We opted for non-invasive treatment involving hyperbaric oxygen therapy (HBO). After 20 sessions of HBO without antibiotics or bone resection, his TBR had decreased from 8.16 to 6.74, and his foot ulcer had healed. We speculated that osteomyelitis had prevented ulcer epithelialization, and that the effect of HBO on the enhancement of leukocyte activity and osteogenesis and augmentation of antibiotic action may have alleviated osteomyelitis, thereby enabling complete epithelialization. This case indicated that the noninvasive procedure could be an alternative for the osteomyelitis with lower TBR cutoff.

Discussion
The incidence of limb amputation has increased along with the number of patients with diabetes or PAD [14]. However, only a few studies have investigated the role of early changes in diagnosing and determining the severity of osteomyelitis [15]. Imaging modalities such as X-ray and CT only show osteolysis in the delayed phase. Further, MRI is limited by its low speci city [7,8] and contraindications in certain patients, as well as its inability to precisely re ect severity under ischemic conditions [16].
Although uorodeoxyglucose positron emission tomography (FDG-PET)/CT could also allow for a precise diagnosis while reducing radiation exposure [17,18], it is not yet a common diagnostic method for osteomyelitis in Japan because it is not yet covered by health insurance. In contrast, 67 Ga-citrate scintigraphy allows for radiotracer uptake into the site of local in ammation via diffusion (i.e., phagocytosis into leukocytes) as a chelating agent [19]. Moreover, 67 Ga-citrate is known to enter cells in its ionic form [20]. While such planar images are associated with high speci city, their resolution is low.
However, CT can provide detailed images of toe/tarsal anatomy with acceptable resolution, a hybrid 67 Ga SPECT/CT machine allows for simultaneous examination using a single scanner, avoiding artifacts due to changes in the patient's position, and precise detection of osteomyelitis locations. In addition, ischemia (low TcPO 2 ) and infection (high TBR) can vary in foot ulcer patients; thus, TBR scores determined via 67 Ga SPECT are useful for early diagnosis, decision-making regarding bone resection, severity scoring, and prognostic determination in patients with refractory ischemic osteomyelitis awaiting bone resection.
In the present study, diagnostic accuracy was comparable to that reported for other radionuclide examinations and MRI studies [21]. Furthermore, multi-regression analysis indicated that TBR cutoff is an independent risk factor for bone resection (Fig. 3-D). When TBR is less than cutoff, imaging follow-up and non-invasive treatment such as hyperbaric oxygen therapy may be one of the options [22,23].
As a limitation, rst, 67 Ga SPECT/CT is associated with exposure to radiation [24]. We minimized radiation exposure: the radiation dose for whole-body gallium scintigraphy is 14.8 mSv and the computed tomography dose index (CTDI) in our imaging study was 1.3 mGy. We believe 67 Ga SPECT/CT is not hazardous since it is performed commonly in daily medical care for decades. Second, we did not perform any comparisons using other tracers in this study. Some previous studies diagnosed osteomyelitis by bone scintigraphy and white blood cell scintigraphy [8,25]. However, we think those two tracers are not the optimal choice for quantitative evaluation of osteomyelitis. Bone scintigraphy cannot be used for the quantitative evaluation of in ammation because it merely visualizes the process of bone matrix synthesis followed by osteomyelitis and bone necrosis. White blood cell scintigraphy is seldom performed because of its complicated labeling procedure and the Japanese health insurance coverage regulation. Third, we did not speculate about the correspondence between imaging and histopathological ndings. We performed pathological assessment in 11 of 27 bone-resected patients, and 8 patients were diagnosed with osteomyelitis histopathologically. However, as shown by our representative case, it is di cult to perform histopathological assessment in some cases whose osteomyelitis healed without bone resection. Finally, the sample size of this study was small.

Conclusions
This study demonstrates the usefulness and feasibility of hybrid 67 Ga-citrate SPECT/CT fusion imaging and TBR scores for improving diagnostic quality among patients with osteomyelitis awaiting bone resection even in ischemic conditions. Coupling of 67 Ga-citrate hybrid SPECT/CT fusion imaging by TBR analysis revealed severity thresholds (TBR 10.1) for bone resection and aided in anatomical localization; therefore, this may prevent unnecessary amputation and could be used to assess long-term limb prognosis.

Consent for publication
All patients included in this study provided written informed consent before surgical treatment. Also, all patients included gave written informed consent that their data could be used for scienti c purposes.

Availability of data and material
Please contact the author for data request.

Competing interest
The authors declare that they have no competing interests.    Target-to-background ratio (TBR) changes, receiver-operating characteristic (ROC) curve, and results of Kaplan-Meier analysis. A. Changes in TBR before and after treatment. Open circles indicate the non-bone resection group, while closed circles indicate the bone resection group. TBR improved in the bone resection group (p < 0.01). However, no signi cant differences are observed between the groups (p = not signi cant, repeated-measures analysis of variance). B. participants ow diagram and number from eligibility criteria, and methods of selection. C. Kaplan-Meier analysis for bone resection incidence using a TBR cutoff of 10.1. D. The independent predictors of bone resection. The results of hemodialysis, type 2 diabetes mellitus, transcutaneous oxygen tension (TcPO2) value < 50 mmHg. TBR ≥ 10.1 is matched by propensity score-matched analysis before regression analysis. The circle indicates hazard ratio. Lines represent 95% con dence intervals (CIs).