Radiography, ^99mTc–HDP, and ¹¹¹In Labeled Vitamin B₁₂ SPECT of Canine Osteosarcoma: A Comparative Study

Introduction

The more common primary bone tumors diagnosed in the long bones of the dog include osteosarcoma and chondrosarcoma. Osteosarcoma is the most common, representing approximately 85% of all primary bone tumors.^1,2 Although radiography plays an important role in the workup of patients with primary bone tumors, the advent of modern imaging modalities, such as scintigraphy, computed tomography, and magnetic resonance imaging, has improved diagnostic capabilities to plan treatment and evaluate outcome.^3–6 Bone scintigraphy using technetium 99m (^99mTc) hydroxyethylene disphosphonate (HDP) as one of the most common radiopharmaceuticals available is a sensitive imaging diagnostic tool for the diagnosis of a variety of skeletal diseases.^5,7,8 Areas of increased osteoblastic and osteoclastic activity in the bone have increased uptake of ^99mTc–HDP. The principle of the “physiologic” image produced by scintigraphy is based on the fact that the radiopharmaceutical binds to the hydroxiapatite crystals. Most bone diseases can, therefore, be detected at a very early stage using scintigraphy. Although scintigraphy is a very sensitive imaging modality for osteoblastic changes in the bone, it is not a very specific diagnostic tool.^5,9 Inflammation, trauma, infection, degenerative joint disease, and neoplasia produce various degrees of uptake of the radiopharmaceutical.⁹ Bone scintigraphy may not be able to detect bone tumors when the osteolytic and osteoblastic activities are minimal.¹⁰ In this regard, F-18-fluorodeoxyglucose positron emission tomography has proven to be a more sensitive technique, as myeloma and its metastasis can be detected by F-18-fluorodeoxyglucose positron emission tomography but not by ^99mTc–HDP scans.¹⁰ However, bone scintigraphy can help identify various primary tumor types^3,5,11,12 and facilitate the determination of intramedullary extent of primary tumors in long bones.^4,6,13,14

The accurate determination of the proximal extent of a primary bone tumor is important for limb-sparing procedures.^1,2 There are descriptions in veterinary medicine and human medicine literature comparing multiple imaging modalities, including radiography, scintigraphy, computer tomography, and magnetic resonance imaging to try to determine which image modality is the best for accurate determination of intramedullary tumor extent of osteosarcoma in long bones, but the results are variable.^{4–6,13–16,} Determination of tumor extent is very important because incomplete surgical excision of the tumor has been related to prognosis and patient survival.^17–19

Diethylene triaminopentaacetate adenosylcobalamin is a vitamin B₁₂ analog that has been labeled with indium 111 (¹¹¹In) and used to evaluate multiple malignancies in human patients.^20,21 Indium 111 labeled vitamin B₁₂ has also been used to depict ciliary adenoma and soft tissue sarcoma in a dog's limb.²² These studies demonstrated increased uptake in malignant tumors compared with surrounding tissues and also increased uptake in transplanted sarcomas in experimental animals.²³ However, to the authors’ knowledge, the potential use of ¹¹¹In labeled vitamin B₁₂ to image primary bone tumors in dogs has not been previously reported.

Elevated plasma concentrations of transcobalamin I (TC I) and transcobalamin II (TC II) and the serum protein transporters that regulate the cellular uptake of vitamin B₁₂ were reported in patients with breast cancer, multiple myeloma, and renal carcinoma.^24,25 High concentration of TC I was also found in amniotic fluid where there was rapidly proliferating fetal tissue.²⁶ Cell membrane receptors for TC II were encountered in high concentration in multiple malignant cell lines. The results presented in these studies suggested that cellular proliferation was the biologic event that triggered the production of TC II and TC I and their cellular receptors; hence, vitamin B₁₂ was found in very high concentration in rapidly dividing cells to meet their metabolic requirements and achieve rapid cell division efficiently.²⁴ Therefore, due to the characteristics of vitamin B₁₂ in terms of its affinity for highly proliferative tissues, it was hypothesized that ¹¹¹In–vitamin B₁₂ scintigraphy (using single-photon emission computed tomography [SPECT]) would be more accurate than ^99mTc–HDP bone scan and radiography to determine the size of osteosarcoma in long bones of dogs.

The purpose of this study was to compare radiography and planar bone scintigraphy using ^99mTc–HDP and ¹¹¹In–vitamin B₁₂ scintigraphy using SPECT as indicators for the intramedullary size of osteosarcoma in long bones in dogs. To the authors’ knowledge, imaging of osteosarcoma with ¹¹¹In–vitamin B₁₂ SPECT has not been previously reported.

Materials and Methods

This project was performed according to the standards of the Colorado State University's Animal Care and Use Committee. Prospectively, 10 dogs presented to the Comparative Oncology Service at the Colorado State University Veterinary Teaching Hospital for primary bone tumors were included in this study from January to April 2003. The 10 dogs included in this trial had histopathologically confirmed osteosarcoma in six radii, two humeri, one tibia, and one ulna. The breeds presented during the period of the study included rottweiler (n=3), golden retriever (n=2), mixed breed (n=2), Alaskan malamute (n=1), Labrador retriever (n=1), and beagle (n=1) (age range, 2–11 yr; mean, 5.7 yr).

Radiographic evaluations of the affected limbs were performed on mediolateral views, although orthogonal images were also available for evaluation. The proximal and distal extent of tumor was determined based on radiographic evidence of osteolysis and periosteal reaction. Whole body bone scintigraphy with ^99mTc–HDP and ¹¹¹In labeled vitamin B₁₂ SPECT were performed in all dogs. Energy settings used were a 140 keV 20% window for ^99mTc images with dual energy settings and 173–247 keV with a 20% window for ¹¹¹In images. All three imaging studies and the surgical procedure were performed within a 1 wk period. All dogs underwent either a limb spare salvage procedure or limb amputation. Limb amputation was performed in dogs with osteosarcoma involving the humerus and tibia because limb-sparing surgery is limited to osteosarcoma of the distal radius and ulna.

^99mTc–HDP was injected at 0.65 mCi/10 kg, and images were acquired 2 hr after injection. Image acquisition time was 2 min/picture. Matrix size used 256×256×16, with a pixel size of 1.48 mm. Images were obtained using a low energy all-purpose collimator γ camera^a. On the following day, 1–2 mCi of ¹¹¹In labeled vitamin B₁₂ was injected intravenously, and images were obtained 1 to 1 and a 1/2 hr after the injection of the radiopharmaceutical. The patients were anesthetized following a standard protocol at Colorado State University. SPECT images were acquired with a medium energy collimator double head γ camera. Matrix size used was 128×128, with a pixel size of 2.9 mm. In 24 hr, about 50% of the injected dose of ^99mTc is urinated, leaving approximately half of the dose in the dog. This lasts about 4 half lives before ¹¹¹In image acquisition, so the amount of ^99mTc–HDP left in the skeleton was not enough to influence the ¹¹¹In scan. Any activity from the lower energy ^99mTc photons would be stopped by the medium energy collimation used to obtain SPECT images. Resolution of the γ camera was 3.8 mm. SPECT image processing was based on iterative reconstruction algorithm with ordered subset expectation maximization (OSEM), and two iterations were applied to eight subsets.

Planar scintigrams (^99mTc–HDP) and sagittal plane images from ¹¹¹In–vitamin B₁₂ SPECT scan were used to measure the tumor length. Image slices in sagittal plane were reviewed, and the slice showing the maximum tumor size was used for measuring purposes. On the scintigraphic images, the edges of the tumor were established using the scintigraphic activity profile postprocessing imaging tool, which measured the maximum number of counts per pixel. The activity profile function was a computerized postprocessing tool carried on by placing two parallel lines across the long axis of the affected bone and tumor.⁷ This generated a graphic representation of the uptake of the radiopharmaceutical and the maximum number of counts per pixel. By this mean, the margin of the tumor was thought to be delineated more accurately from the relatively normal/abnormal bone interface. The length of the tumor was measured on both scintigraphic techniques. Regions of interest were manually drawn in the bone lesion and normal bone to calculate lesion to background ratios from both planar bone scintigraphy and SPECT scintigraphy. The lesion to background ratio was calculated in the SPECT image with all the slices added together. The ratio was obtained using maximum pixel activity values to minimize any potential influence of regions drawn of slightly different size.

The measurements obtained on the imaging modalities and the lesion to background ratio were performed independently by three coauthors (R.C., P.S., J.U.) and were obtained without knowing the measurement on the macroslides. The numbers obtained were averaged. The measurement of the tumor length on the macroslides was determined independently by one author (B.P.) without knowing the measurement obtained on image modalities.

Postprocessing of the images of the ^99mTc–HDP bone scan as well as the ¹¹¹In labeled vitamin B₁₂ SPECT scans were performed using a nuclear medicine dedicated software^b.

Mediolateral radiographs were used to measure tumor size. The radiographic magnification was factored using the following equation:²⁷

where, MF is the magnification factor, SID is the source to image receptor distance, and SOD is the source to object distance.

Then the actual lesion size was calculated using the equation:

Histopathology was used as gold standard. The macroslides for histopathology were prepared immediately after the surgical procedure. The macroslides used measured 5×7.5 cm. If a bone specimen exceeded the size of the macroslides, two macroslides were made and then laid next to each other to reconstruct the entire specimen. In cases of limb amputation, the abnormal bone was excised by the surgeon immediately after amputation using the same technique as in the limb sparing procedure. None of the specimens showed evidence of incomplete margins proximally on macroslide evaluation. The bone specimen was placed in 10% formalin for about 1 wk, and then it was put in resin ion exchange with 10% acid formic for 2–3 wk for decalcification purposes. The sections were obtained on sagittal slices and were cut 7 μm with a microtome, stained with hematoxylin and eosin, and mounted on macroslides. The microscopic evaluation was performed by one author (B.P.) to obtain the definitive diagnosis and to determine the margins of the tumor.

The three images modalities and the macroslides to prepare the histopathology to obtain the tumor measurement were done in a 1 wk period.

The percentage of intramedullary tumor size overestimation was calculated by the equation:

The statistical analysis performed on the data included simple regression analysis to assess the relationship between variables. Spearman's rank correlation was performed for nonparametric data to evaluate for normality. The correlation coefficient (R²) adjusted for sample size was performed to evaluate the strength of the association between variables. The significance level for R² was set at 0.05. Any statistical difference between R² for the different variables studied was compared with the F test. The statistical difference of radiopharmaceutical uptake ratio of ^99mTc–HDP compared with ¹¹¹In–vitamin B¹² was tested with the paired t test. The significance level for the t test was set at 0.01. The statistical analysis of the data was performed with SAS^c.

Results

Radiographic evaluation showed that two osteosarcomas were purely osteolytic, one was considered mainly osteoblastic, and the remaining six osteosarcomas were mixed osteolytic/osteoblastic. Visual evaluation showed adequate uptake of the radiopharmaceutical ¹¹¹In label vitamin B₁₂ by the osteosarcoma. Figures 1–3 show images of the osteosarcoma in the bone seen on radiographs, planar scintigraphy, and SPECT. The activity profile tool can be seen on the scintigraphic images. The measurements of tumor length (proximal to distal) on the imaging modalities and the macroslide and the overall percentage of tumor length overestimation comparing lateral radiography, ^99mTc bone scan, and ¹¹¹In-–vitamin B₁₂ scintigraphy are presented in Table 1. Taking into account the results from all patients as a single group, the mean percentage of tumor size overestimation was 9.29% on mediolateral radiographs, 5.35% on ^99mTc planar bone scintigraphy, and 33.25% on ¹¹¹In–vitamin B₁₂ SPECT. On mediolateral radiographs, tumor size was underestimated in 30% (mean, 4.1 cm; range, 2.4–5 cm) of the cases, whereas on lateral images from planar bone scans, the tumor size was underestimated in 40% (mean, 6.15 cm; range, 4.8–7.7 cm) of the cases. The tumor length measured with ¹¹¹In–vitamin B₁₂ SPECT was underestimated in 30% (mean, 7 cm; range, 6.5–7.8 cm) of the cases. The percentage of cases in which the tumor was overestimated was 70% for mediolateral radiographs (mean, 6.54 cm; range, 3.8–8.7 cm), 60% on planar bone scans (mean, 5.3 cm; range, 2.7–8.0 cm). and 70% on ¹¹¹In–vitamin B₁₂ SPECT (mean, 6.9 cm; range, 3.6–11 cm). Regression analysis revealed a positive linear relationship between variables studied (^99mTc–HDP versus macroslide, radiograph versus macroslide, ¹¹¹In–vitamin B₁₂ versus macroslide). The correlation coefficient adjusted for sample size was significantly higher (P<0.01) for ^99mTc–HDP (75.5%) and radiography (61.3%) compared with ¹¹¹In–vitamin B₁₂ (28.3%). The correlation coefficient for ^99mTc–HDP was higher than the one obtained for radiographs; however, statistical difference between the two variables was not demonstrated (P>0.05).

Table 1 Measurements Obtained on Image Techniques and Histopathology Macroslide, Tumor Location, and Overall Percentage of Tumor Size Overestimation Comparing Lateral Radiograph (Rad), Technetium 99m (^99mTc) Bone Scan, and Inidium 111 (¹¹¹In)–Vitamin B₁₂ Single-Photon Emission Computed Tomography

View Fullscreen

View Figure

• Download as Powerpoint
• Download Figure

View Figure

• Download as Powerpoint
• Download Figure

View Figure

• Download as Powerpoint
• Download Figure

The mean, SD, and range of the ratio of lesion/background uptake obtained on ^99mTc–HDP bone scan and ¹¹¹In–vitamin B₁₂ SPECT (P<0.01) were 3.72, 1.04, and 3.38 and 8.97, 2.7, and 8.5, respectively. The mean of the ratio of radiopharmaceutical uptake obtained on ^99mTc–HDP bone scan was statistically significant higher than ¹¹¹In–vitamin B₁₂ SPECT (P<0.01).

Discussion

The measurements obtained on radiographs and ^99mTc planar bone scans more accurately reflected measurements obtained on the macroslides, which indicated that ^99mTc planar scintigraphy and radiography were good predictors of tumor size, and therefore intramedullary tumor extent, compared with ¹¹¹In labeled vitamin B₁₂ SPECT. The results obtained in this study with planar bone scintigraphy provided a better reflection of true disease extent compared with reports by other researchers.^6,13,14

Leibman et al. determined that planar scintigraphy bone scans tended to overestimate the tumor extent of osteosarcomas in radii.¹⁴ The same finding was reported by Lamb and Bengson and Wallack et al.^6,13 Wallack et al. reported that bone scintigraphy overestimated the intramedullary osteosarcoma extent by 14±28%.⁶ In this study, the authors found overestimation of intramedullary tumor size of 5.35%. The reason for this difference could be that the present study used the activity profile to obtain the maximum number of counts per pixel in the region of radiopharmaceutical uptake to eliminate the brightness “contamination” resulting from activity spillover in neighboring pixels due to partial volume effects, which are inherently connected with scintigraphy. This resulted in a more accurate delineation of the margin of the tumor and consequently a more accurate measurement of the tumor. This was in agreement with a study from King et al., who reported that bone scintigraphy was a better image modality than radiographs to delineate osteosarcoma in rabbits.²⁸

Radiography has been found to be an imaging technique that tends to overestimate tumor extent in long bones.^6,14 In contrast, others have reported that radiography tended to underestimate the extent of these tumors.¹³ Wallack et al. reported overestimation of 17±28%, whereas Liebman et al. did not report a numeric value for the degree of overestimation.^6,14

Clinical concerns about underestimation of extension of intramedullary tumor are the probability of incomplete tumor resection with risk of recurrence. In contrast, overestimation is preferable because of high probability adequate surgical margins and achieving complete tumor resection.²⁹

One limitation of this study was the small number of dogs that were enrolled in the project. Although the correlation coefficient for ^99mTc–HDP was higher than that obtained for radiography, a statistical difference between the two variables was not obtained. The small sample size in this study was the likely reason for the lack of statistical difference between the correlation coefficient obtained for ^99mTc–HDP compared with radiography.

Based on the results obtained in this study, the hypothesis that ¹¹¹In–vitamin B₁₂ SPECT was more accurate than bone scintigraphy to determine tumor size might not be rejected. The authors had this opinion because of the image resolution problems of the system used to obtain the SPECT images, which could have influenced the apparent greater difference in tumor length comparing the two scintigraphic techniques. Recommendations about the technical aspects of SPECT for bone scintigraphy were published.³⁰ The hypothesis in this study was based on the fact that ¹¹¹In–vitamin B₁₂ would accumulate in the rapid proliferative tumor more efficiently than ^99mTc. The authors considered that this was the most important factor to impact the hypothesis test. The authors thought the settings used in the study were appropriate to image bone tumors with SPECT. It was likely that the activity profile tool used in this study to determine the tumor margins could not discriminate areas of inflammation and/or hyperemia bone sclerosis around the tumor. This was thought to be another factor that could have contributed to the overestimation of tumor size. The lesion to background ratio was lower for ¹¹¹In-vitamin B₁₂ SPECT compared with ^99mTc–HDP, which might have contributed to a poorer definition of the lesion. The authors estimated that low uptake ¹¹¹In–vitamin B₁₂ at the lesion was probably due to a lower dose of ¹¹¹In–vitamin B₁₂ administered compared with ^99mTc–HDP. Cost and availability of ¹¹¹In–vitamin B₁₂ SPECT should also be considered limitations for its use in veterinary medicine. Another limitation was the impossibility of using late images for analysis to improve tumor to background ratio. This was not possible due to logistic problems in terms of scheduling surgery and the use of patients from owners that lived out of town. More investigation needs to be done in veterinary medicine to study the accuracy of SPECT to diagnose and evaluate extension of bone tumors. In humans, SPECT bone scintigraphy was demonstrated to be superior compared with planar bone scintigraphy to delineate the location and extension of inflammatory, traumatic, and malignant bone lesions of the spine.³¹

Currently, magnetic resonance imaging is considered the modality of choice to determine the intramedullary extent of osteosarcoma in both humans and veterinary patients selected for limb spare surgical procedure. However, according to the results presented here, when using the activity profile tool in planar bone scan images, the margins of the tumor can be determined with reasonable amount of spared bone, so planar bone scintigraphy can be used when magnetic resonance imaging is not readily available or the cost of the study cannot be covered by owners. Bone scintigraphy is also a good imaging tool to detect bone metastasis from primary osteosarcoma and skip metastasis.

Conclusion

According to the results obtained in this study, radiographs and ^99mTc bone scan are good predictors for determining intramedullary size of osteosarcoma in long bones of dogs. The authors recommend using the activity profile tool in ^99mTc bone scans to estimate the extent of osteosarcomas. In this study, the authors demonstrated a consistent uptake of ¹¹¹In-vitamin B₁₂ by osteosarcomas.