Near-Infrared Imaging Using a Cathepsin-Targeted, Quenched Activity-Based Probe Identifies Naturally Occurring Canine Appendicular Osteosarcoma

Abstract

Purpose:

Osteosarcoma is the most common primary malignancy of the skeleton. Despite advances in imaging modalities, neoadjuvant and adjuvant chemotherapy, and limb-sparing surgery over the past decades, there has been a general lack of improvement in survival rates. Tumor recurrence and metastases are associated with a worse prognosis, and complete tumor excision is critical. The use of real-time imaging could facilitate the acquisition of tumor-free surgical margins. Spontaneous osteosarcoma in dogs is an established translational model of human osteosarcoma.

Patients and Methods:

In this study, twelve dogs with spontaneous appendicular osteosarcoma received a cathepsin-targeted quenched activity-based probe (VGT-309) intravenously 16–20 hours before amputation. The limb was imaged at 4 levels of dissection, and the margins of near-infrared fluorescence were marked for comparison with histopathology and preoperative magnetic resonance imaging.

Results:

All appendicular tumors fluoresced on cross-section of the bone, and the extent of fluorescence coincided with MRI and histopathologic margins, with variability associated with necrosis. All tumors had visible fluorescence through cortical bone, although this did not consistently reflect intramedullary extension of tumor. Necrotic tumor regions did not fluoresce.

Conclusions:

This data supports the safety and feasibility of VGT-309 as a tool for evaluating the tissue extent of canine osteosarcoma and spurs investigation for intraoperative detection of osteosarcoma during limb-sparing surgery in humans. Due to frequent intramedullary tumor extension and inconsistent fluorescence through unaffected cortical bone, this modality is likely most appropriate in assessing residual tumor following planned resection. Caution should be exercised when significant tumor necrosis is present.

Introduction

Osteosarcoma (OS), a malignant tumor of primitive bone cells producing osteoid matrix, is the most common primary malignancy of the skeleton. OS accounts for approximately 20% of primary malignant bone tumors and has an incidence of 4–5 per 1 million individuals in the United States, representing 0.2% of all malignant tumors^1,2. Osteosarcoma has a bimodal age distribution and affects males more frequently than females, with most tumors occurring in the metaphysis of appendicular long bones, including the distal femur, proximal tibia, and proximal humerus¹.

Advances in diagnostic imaging, neoadjuvant chemotherapy, and limb-salvage surgery, which can be performed in 80–90% of patients, have improved OS treatment results³. However, there has been little progress in the treatment of recurrent or metastatic OS, with long-term survival of only 20–25% of these patients⁴. The incidence of local recurrence is related to the response to preoperative chemotherapy and the margins obtained at surgery, highlighting the importance of aggressive and thorough excision at the time of primary tumor removal.^2,5 Tumors are removed by limb amputation or via limb-sparing surgery. The latter is generally guided by pre-operative MRI and includes surgical removal of the tumor with a border of unaffected tissue margins; the bone is then stabilized with a metallic prosthesis, bone graft, or both. The rate of local recurrence is 2–3% after amputation and 5–7% after limb-sparing surgery.² The 5-year survival rate is approximately 60% in patients with localized OS and 20% in patients with metastasis or recurrent disease.^5,6 Long-term survival in patients with recurrent tumors was only seen in cases where the tumor was completely removed at a second surgery.⁷ This underscores the importance of accurate tumor detection and complete tumor extirpation at the time of surgery.⁷

Over the past 3 decades, there has been a general lack of improvement in OS survival rates.⁸ The risk of tumor-positive margins, local recurrence, and subsequent metastasis in a number of cancer types has prompted the development of new methods of intraoperative assessment of surgical margins in many tumors.^9–14 Clearly defining tumor boundaries is vital to avoid both under- and overestimation of tumor margins, which could lead to local tumor recurrence or increased surgical morbidity and impaired post-operative function, respectively.^4,15

Intraoperative near-infrared (NIR) imaging with systemically injected NIR contrast agents is a novel method for tumor localization and margin detection that is being investigated in various solid malignancies.^16,17 In veterinary medicine, studies of spontaneously occurring tumors in client-owned animals demonstrate that NIR imaging accurately detects tumors, tumor margins, and residual local disease in canine primary lung tumors, sarcomas, and mammary tumors.^11,18,19

Cathepsin K (CatK) is the primary protease secreted by activated osteoclasts and is intrinsically involved in bone resorption through collagenolytic activity against collagens I and III.^20,21 Overexpression of cathepsins has been documented and proposed as a therapeutic target in canine OS.²² Inhibition of CatK has been studied for its role in osteolytic pathologies in human trials.^23–25 CatK expression in OS makes it a potential target for NIR imaging using a cathepsin-based fluorophore. The imaging agent VGT-309 (Vergent Bioscience, Minneapolis, MN) covalently and irreversibly binds cathepsins in vivo, becoming fluorescent upon displacement of the quencher, QC-1.²¹ Indocyanine green (ICG) acts as the fluorophore and allows visualization of covalently bound VGT-309, allowing for real-time intraoperative tumor detection with commercially available NIR surgical imaging systems.¹³

Dogs are an established spontaneous translational model of human pediatric osteosarcoma.²⁶ Canine OS represents over 80% of malignant bone tumors. As with human osteosarcoma, a majority (64%) of canine OS lesions arise in the metaphyseal appendicular skeleton, including the proximal humerus, distal radius, distal femur, and proximal tibia. Treatment of canine OS is centered around the removal of the primary tumor through limb amputation or limb-sparing surgery paired with adjuvant chemotherapy. ²⁷

The purpose of this study was to evaluate NIR imaging with a cathepsin-targeted quenched activity-based probe, VGT-309, to visualize spontaneous appendicular osteosarcoma tumors and tumor margins in dogs. We hypothesized that VGT-309 would cause canine OS tumor fluorescence, and that fluorescent margins correlate with those defined by magnetic resonance imaging and histopathology.

Materials and Methods

Canine Subjects

Dogs (n=12) diagnosed with appendicular OS were presented to the Matthew J. Ryan Hospital at the University of Pennsylvania School of Veterinary Medicine for treatment and were prospectively enrolled in the study. The study was approved by the University of Pennsylvania Institutional Animal Care and Use Committee (IACUC). Consent documentation was approved by the Privately Owned Animal Protocol Committee of the School of Veterinary Medicine, with informed consent obtained from all owners.

Dogs with evidence of a mixed lytic and proliferative osseous lesion of the metaphyseal appendicular skeleton on orthogonal radiographs were included in the study. Fine needle aspiration or bone biopsy was obtained for diagnosis at the discretion of the attending veterinarian. Dogs with pulmonary metastasis on thoracic radiographs and/or CT were excluded. Additional exclusion criteria included the presence of a pathologic fracture at the primary tumor site and the presence of severe orthopedic and/or neurologic abnormalities, making the dog a poor candidate for limb amputation. The owners elected full limb amputation for local disease treatment (standard veterinary practice) before enrollment in the study.

Cathepsin-targeted contrast agent

VGT-309, a NIR cathepsin-targeted quenched activity-based probe, drug substance and drug product were manufactured in compliance with Good Manufacturing Practices by Chinese Peptide Company (Hangzhou, Zhejiang, China) and UI Pharmaceuticals(Iowa City, IA), respectively. The ICG fluorophore has an excitation peak of 789 nm and an emission peak of 814 nm. VGT-309 drug product was supplied to the University of Pennsylvania School of Veterinary Medicine by Vergent Bioscience, Inc (Minneapolis, MN).

Sixteen to twenty hours before surgery, vials containing 11mg of lyophilized VGT-309 were reconstituted with 2 mL of sterile water and diluted using sterile saline (NaCl) to a final concentration of 2 mg/mL. The solution was protected from light during all steps of the procedure. Dogs received VGT-309 (0.2 mg/kg) intravenously over 10 minutes, and respiratory rate, heart rate, mentation, and blood pressure were monitored for 60 minutes post-injection. The VGT-309 dose was determined based on a previous study of pulmonary tumors in dogs, wherein an initial 0.3 mg/kg dose was successfully reduced to 0.1 mg/kg based on strong tumor fluorescence.²⁸

Magnetic resonance imaging (MRI)

An MRI of the affected bone was performed (1.5T GE SIGNA EXPLORER; Software version 25.1) under general anesthesia prior to surgery. Images of the affected bone were obtained in the sagittal and dorsal planes, and the field of view was set to include the adjacent joints. Images were acquired using a T1-weighted pulse sequence, with 2 NEX and TR/TE variables depending on the patient, and the image slice thickness of 2mm with an interslice gap of 0.2mm.

Fluorescence imaging

A standard-of-care limb amputation was performed under general anesthesia, consisting of forequarter amputation or coxofemoral disarticulation and amputation of the hindlimb. Fluorescence imaging was performed using a commercially available NIR imaging system (VisionSense Iridium NIR imaging system, Medtronic Inc., New York, NY). The Iridium is a dual-band (white light and NIR), high-definition system. An 805 nm excitation source was used on a free-standing exoscope with a field of view of approximately 19 × 14 cm and a working distance of 40 cm. Fluorescence was detected using a bandpass filter ranging from 825 to 850 nm.

The tumor was imaged in orthogonal views at four timepoints: through intact skin before amputation (Standard 1)(Figure 1A,B), following amputation and after skin removal and surgical approach to the tumor (Standard 2)(Figure 1C,D), after removal of overlying soft tissues from the bone (Standard 3)(Figure 1E,F), and on cross-section along the long axis after sagittal or dorsal sectioning of the bone with a band saw (Standard 4)(Figure 1G,H). A ruler placed at the level of the bone was used to calibrate all image measurements. Fluorescent margins were marked with suture in peri-tumoral soft tissue under fluorescence guidance. The extent of medullary/osseous fluorescence after cross-sectioning was marked by notching the cortical bone using a Hobby saw for correlation with histopathologic margins after decalcification.

Figure 1:

Images demonstrating fluorescence of a distal femoral osteosarcoma in Dog 2 (A,C,E,G) and proximal humeral osteosarcoma in Dog 9 (B,D,F,H). While no fluorescence was visible through skin (A,B) in Standard 1, fluorescence was appreciable in the remaining dissection planes: Standard 2 (C,D) after surgical approach and through surrounding soft tissues; Standard 3 (E,F) through intact bone after soft tissue removal; and Standard 4 (G,H) after hemisecting of the bone. A central region of necrosis resulting in a lack of fluorescence is visible in the cross-section of the humerus of dog 9 (H).

Quantification of signal-to-background ratios (SBRs)

The region-of-interest plug-in of ImageJ (Version 1.54h, RRID:SCR_003070) was used to quantify tumor and normal bone fluorescence and compare SBRs. The fluorescence intensity’s range, mean, and standard deviation in each area of interest were obtained using the ImageJ histogram function in the red color channel. Positive signal was quantified as the average fluorescence from the identified tumor regions, with the background identified as the average fluorescence of normal bone. SBR was calculated by dividing the average tumor fluorescence by background fluorescence. Mean fluorescent intensities (MFI) of the cross-sectioned bone were assessed for normality using the Shapiro-Wilks test and were not normally distributed; hence, they were described using median (minimum, maximum) and box plots. The two groups were compared using the Wilcoxon rank-sum test. A p-value <0.05 was considered significant. A statistical software program (Stat 13.2 for Mac, StataCorp, College Station, TX) was used for all calculations. Lymph nodes were defined as fluorescent or non-fluorescent. Fluorescence was evaluated after lymph node extirpation. A signal-to-background ratio comparison could not be performed for lymph nodes, as there was no surrounding normal tissue to evaluate for background fluorescence.

Measurements

Tumor length was measured on pre-contrast MRI images, either in the sagittal or dorsal plane, depending on how the pathologist sectioned the bone after surgery. The image most aligned with the gross sagittal or dorsal section of the bone was chosen for evaluation. Tumor length was measured in millimeters from the articular surface nearest to the tumor to the tumor margin furthest from the articular surface. Tumor length was determined by identifying a well-defined junction between the low-to-moderate signal intensity of the intramedullary tumor and the hyperintense bone marrow signal on pre-contrast T1-weighted images.

Images obtained of the cross-sections of bone with NIR fluorescence were retrospectively examined. Images were calibrated to a ruler as described above, and measurements were obtained from the joint nearest the tumor to the most distant margin of fluorescence. Osseous landmarks of the articular surface were used to best reflect the landmarks used for the MRI measurement. The length of the tumor was measured in millimeters from the articular landmark nearest the tumor to the extent of visible fluorescence furthest from the articular surface. NIR, MRI, and histopathologic margins were compared. Images obtained from the cortical bone prior to sectioning were measured in a similar fashion.

Histopathology

The specimens were fixed in 10% neutral buffered formalin, and affected bone was processed for histopathologic evaluation by decalcification in 5% formic acid. Sectioning of the bone was performed to include the region of the grossly apparent neoplasm and at the denoted tumor margins, including periarticular soft tissue/periosteum marked with suture and the medullary margin at the level of notched cortical bone. Additional sampling of the grossly normal bone was performed beyond the fluorescence region (beyond the cortical notch). If applicable, tumor subclassification was noted. Available draining lymph nodes were evaluated. Specimens were routinely processed with 5-micrometer-thick sections stained with hematoxylin and eosin and evaluated with light microscopy. A Board-certified veterinary pathologist examined all specimens.

Data Availability

The data generated in this study are available upon request from the corresponding author, as the imaging agent is still undergoing clinical testing. The chemical structure of VGT-309 (Abenacianin Sodium) is available through PubChem (CID: 171390000) and has been previously reported by Suurs et al. (9).

Results

Twelve (12) dogs were enrolled in the study (Table 1). The ages ranged from 3 to 12 years (median, 9 years), and weights ranged from 19.7 to 55.3 kg (median, 32.7 kg). There were 8 female spayed and 4 male castrated dogs. Breeds included mixed breed dog (3), Doberman Pinscher (2), and one each Great Pyrenees, English Bulldog, Rottweiler, Great Dane, Saint Bernard, American Pit Bull Terrier, and Boxer dog. Tumor location included distal tibia (4), proximal humerus (4), distal femur (2), proximal tibia (1), distal radius (1). One dog with a proximal humeral lesion also had a scapular body lesion within the same limb that was evaluated. The VGT-309 injections were well-tolerated with no observed adverse reactions.

Table 1:

Canine demographics and tumor characteristics.

DOG #	WEIGHT (KG)	AGE (YR)	SEX	BREED	LESION SITE	LESION LATERALITY
1	45	9	FS	Great Pyrenees	Distal tibia	Right
2	28	8	MC	English Bulldog	Distal femur	Right
3	55.3	9	MC	Rottweiler	Distal tibia	Right
4	55	8	FS	Great Dane	Distal tibia	Right
5	36	10	FS	Saint Bernard mix	Proximal humerus	Left
6	34.8	6	FS	Doberman Pinscher	Distal tibia	Left
7	30.5	12	FS	American pit bull terrier	Proximal humerus	Right
					Scapula	Right
8	24	9	MC	MBD	Distal radius	Right
9	26.2	10	FS	Doberman Pinscher	Proximal humerus	Left
10	24.5	3	FS	Boxer	Proximal tibia	Left
11	41.5	10	MC	MBD	Proximal humerus	Right
12	19.7	3	FS	MBD	Distal femur	Left

Signalment of dogs with appendicular osteosarcoma enrolled in the study.

Abbreviations: Female spayed (FS), male castrated (MC), mixed breed dog (MBD)

All tumors were histologically defined as osteosarcoma. Veterinary osteosarcomas are not commonly subclassified except for uncommon to rare fibrous/fibroblastic, telangiectatic, chondroblastic, and osteoclastic subtypes.²⁹ All of these tumors would be characterized as conventional [intramedullary/central] based on the WHO classification scheme.³⁰ All evaluated regional lymph nodes (11/12 dogs) were negative for metastatic disease- all were consistent with drainage reaction (sinus edema with histiocytosis, erythrocytosis, and plasmacytosis). Lymphoid cortical hyperplasia was present in the lymph nodes of six cases.

All appendicular tumors (12/12, 100%) displayed NIR fluorescence on cross-sectional imaging of the bone (Standard 4) and through the cortical bone after soft tissue removal (Standard 3). The scapular lesion co-imaged in the case of a proximal humeral osteosarcoma (Case 7) was not included in the analysis. Upon imaging after surgical approach/skin removal (Standard 2), 11/12 (91.7%) tumors demonstrated fluorescence. Fluorescence was noted in 4/11 (36.4%) lesions through intact skin (Standard 1) and not recorded in one case. The scapular lesion (Case 7) demonstrated minimal fluorescence in Standard 4 and no notable fluorescence in the other three imaging standards. Regional lymph nodes were evaluated in 6/12 (50%) cases and were all positive for fluorescence (100%).

NIR fluorescence of all tumors in cross-section (Standard 4) was greater than that of the surrounding normal bone (p= 0.0022) (Figure 2). One tumor could not be evaluated due to an error in image storage. The mean tumor to normal SBR (n=11) was 3.55 (range 1.46–5.56). Median MFI was 140.73 A.U. (range, 86.36–216.42 A.U.), while the median MFI of the background was 37.06 A.U. (range, 20.98–148.49 A.U.).

Figure 2:

A box-and-whiskers plot demonstrating the mean fluorescence (Arbitrary Units, AU) of normal bone and tumor. NIR fluorescence of all tumors was greater than that of the surrounding normal bone (p= 0.0022)

The presence/absence of visible fluorescence and mean tumor to normal tissue SBR in each Standard are summarized in Table 2. The mean tumor to normal SBR of visibly positive fluorescence through skin (Standard 1) was 2.18, whereas the mean SBR of visibly negative fluorescence was 1.25. The mean SBR of positive dogs through soft tissue (Standard 2) was 1.88; one dog did not have visible fluorescence with an SBR of 1.18. All dogs had visible fluorescence through cortical bone (Standard 3), with a mean SBR of 1.71.

Table 2:

Fluorescence and MRI tumor measurements.

DOG #	NIR LENGTH ON CROSS-SECTION (MM)	NIR LENGTH THROUGH CORTICAL BONE (MM)	MRI LENGTH (MM)	STANDARD 1 FLUORESCENCE (SBR)	STANDARD 2 FLUORESCENCE (SBR)	STANDARD 3 FLUORESCENCE (SBR)	STANDARD 4 FLUORESCENCE (SBR)	LYMPH NODE FLUORESCENCE
1	42.5	42	42.3	Yes (1.44)	Yes (1.93)	Yes (N/A)	Yes (1.46)	---
2	54.5	50.2	53.6	No (1.18)	Yes (3.65)	Yes (1.57)	Yes (2.99)	---
3	148.4	126	146.7	Yes (3.29)	Yes (2.61)	Yes (1.65)	Yes (3.30)	---
4	157.013	N/A	154.5	---	Yes (1.49)	Yes (N/A)	Yes (3.30)	yes
5	155.4	N/A	143.5	No (0.90)	No (1.19)	Yes (0.906)	Mild (N/A)	---
6	114.7	64.1	111.7	Yes (1.99)	Yes (1.50)	Yes (1.42)	Yes (4.18)	yes
7 HUMERUS	105.1	N/A	134.9	No (1.18)	Yes (0.92)	Yes (1.24)	Yes (2.23)	yes
SCAPULA	N/A	N/A	134.8	No (N/A)	No (N/A)	No (N/A)	No (N/A)	no
8	42	37.5	41.9	No (1.09)	Yes (1.10)	Yes (1.40)	Yes (4.52)	---
9	123.4	109.5	123	No (0.99)	Yes (2.56)	Yes (1.45)	Yes (5.21)	yes
10	67.4	N/A	57	No (1.04)	Yes (1.79)	Yes (N/A)	Yes (5.56)	---
11	168.8	158.3	171.2	No (1.62)	Yes (1.48)	Yes (3.57)	Yes (3.47)	yes
12	164.1	74	149.1	Yes (1.99)	Yes (1.67)	Yes (2.17)	Yes (2.87)	yes

Measurements of appendicular osteosarcoma tumor length, from the nearest joint, measured using NIR imaging and MRI imaging. The presence of fluorescence in each Standard and available lymph nodes is signified as a subjective yes/no to demonstrate real-time visualization in surgery, with signal-to-background ratio (SBR) reported when possible for objective measure. Standard 1: through intact skin; Standard 2: after skin incision, through soft tissues; Standard 3: after soft tissue removal; Standard 4: after cross-section of bone.

The mean tumor length measured from the sagittal MRI image was 110.8mm (range, 42.3–171.2 mm). The mean tumor length as measured from the fluorescence images was 111.9 mm (range, 42.5–168.8 mm). The scapular lesion was measurable on the MRI image (134.8mm) but did not have sufficient fluorescence for measurement on the fluorescence images. Fluorescence overestimated the tumor length compared to MRI in 10/12 lesions (83.3%) and underestimated the length in 2/12 lesions (16.7%). The mean discrepancy in measured tumor lengths between the two modalities was 6.5 mm (median, 2.4 mm; range, 0.2–29.8 mm). The length of fluorescence through cortical bone was measured when available and is summarized in Table 2. Images were excluded (N/A) if not saved with appropriate calibration, if dissection did not allow visualization of the associated joint to use as a landmark for measurement, or if it was not possible to distinguish the fluorescent tumor margin.

Histology was used to confirm the presence or absence of neoplastic cells at the marked neoplastic margins. These included (notched) cortical margins in the cortical and medullary portions of the longitudinally sectioned bone (Figure 3). No neoplastic cells were noted to extend beyond the margin of fluorescence (notch) in any case and were not identified in additional sections of cortical/medullary bone distant from this margin, away from the primary neoplastic mass with the exception of one dog (Case 7) in which neoplastic cells were identified in a distant piece of medullary bone sampled. Neoplastic cells were not identified beyond the marginal periosteal and periarticular tissues of the primary neoplastic mass (e.g. distal and lateral aspects of fluoresced bone Figure 3).

Figure 3:

Representative images from dog #3 with OS of the distal tibia. a) MRI measurement of the sagittal tibia; extent of tumor was defined by measuring from the nearest joint surface to the most proximal extent of medullary change, b) NIR imaging after surgical approach to the tumor (Standard 2), c) NIR imaging after removal of soft tissues (Standard 3), d) NIR imaging of the hemisected tibia (Standard 4). To demonstrate extent of medullary involvement with measurements corresponding with MRI, white arrow is notch of extent of NIR imaging, e) composite photomicrograph of marginal tissue at corresponding notch (black arrows), tumor in medullary cavity (block arrow) does not extend beyond the notched margin; SP-subperiosteal woven bone; C cortical bone; bar is 500 micrometers; f) osteosarcoma, conventional type with tumor osteoid (astericks); bar is 20micrometers.

Follow-up was available on 11/12 dogs (92%). The date of death was available for 9/12 dogs (75%), with a median time from amputation to death of 10 months (range, 3–19 months). All 11 dogs with follow-up received adjuvant carboplatin post-operatively. 9/11 dogs (82%) had suspected metastasis during the follow-up period; 7 dogs had radiographic pulmonary nodules, one dog had an intra-abdominal mass confirmed as canine OS by fine needle aspiration, and one dog had a lytic lesion consistent with OS of the 4^th lumbar vertebral body diagnosed via MRI. One dog was diagnosed with a hemoabdomen and palpable intra-abdominal mass 10 months post-operatively; the owners declined further diagnostics. The final dog had no pulmonary metastasis noted on radiographs at last follow-up, 15 months post-operatively; this dog was euthanized 19 months post-operatively, but the reason for euthanasia was not available. No dogs had short- or long-term adverse events reported associated with the administration of the study drug VGT-309.

Discussion

In this study, NIR fluorescence imaging using VGT-309, a cathepsin-targeted quenched activity-based probe, accurately identified appendicular OS and tumor margins in a spontaneous canine tumor model. All long-bone-associated tumors imaged in this study fluoresced using near-infrared imaging at the level of the bone, and there were strong correlations to both MRI and histopathologic margins of the canine OS tumor.

Surgical resection is a cornerstone of osteosarcoma treatment. The primary surgery for osteosarcoma is an en-bloc resection of the tumor, biopsy tract, and a perimeter of tumor-free soft tissue. Up to 10% of OS patients develop local recurrence, likely due to residual neoplasia.^31,32 Patients with local recurrence require additional surgery or local disease treatment, and their prognosis is generally poor despite multimodal treatment.^32–34 Therefore, accurate intraoperative identification of the tumor and tumor margins is crucial. Residual cancer is often microscopic and cannot be visualized or palpated by the surgeon. The use of frozen sections of medullary bone is described to ensure tumor-free margins.³² This presents challenges, including operative time, tissue preparation, artifacts from freezing, and sampling errors. The current study showed no tumor beyond the marked fluorescent corticocancellous or soft tissue borders on histopathology. These results support the use of NIR imaging as a viable method of assessing the extent of the tumor during surgery, complementing preoperative surgical planning of an appropriate resection to achieve tumor-free margins.

The majority of tumors in the current study had intramedullary extension of the tumor at its margin, which was detected on cross-sectional NIR imaging. The extent of intramedullary extension was visualized on preoperative MRI but not consistently seen through the intact cortical bone using NIR imaging. The images available that could be comparably measured in Standard 3 (intact cortical bone after soft tissue removal) consistently underestimated the intramedullary extent of the tumor. While the limited case numbers and challenges in histologically evaluating this bone, due to the processing and dissection techniques, limit the interpretation of this finding, it is likely that intramedullary fluorescence will not be detected through dense, unaffected cortical bone. However, NIR imaging can be used as a method to assess surgical excision of tumors by imaging the cut ends of the resected and in situ bone and the surrounding soft tissues, complementing preoperative assessment and surgical planning based on MRI imaging.

Preoperative MRI is used to determine cancer extent and plan surgical excision.^35,36 This imaging modality is considered the most accurate method to determine the extent of tumor involvement in peri-articular and neurovascular structures. It is regarded as the gold standard for determining the feasibility and specifics of limb-sparing surgery.^37,38 This modality was therefore chosen for this study to best reflect surgical planning for a limb-sparing surgery in humans. Histopathology was used to determine the extent of the tumor.³⁸ The fluorescent margins were highly correlated with the tumor margins measured on MRI, with NIR fluorescence revealing a slightly longer tumor length than MRI in 10/12 cases. Though limited by the small number of cases, this data suggests a tendency towards a small over-estimation of tumor extent using NIR imaging. In a previous study in canine fibrosarcoma, NIR fluorescence extended 3–6 mm beyond the histologic margins of the tumors.³⁹ NIR fluorescence microscopy, used to evaluate 3 μm sections around the fibrosarcomas, showed no fluorescence. A subsequent Monte Carlo computer simulation suggested that the fluorescence observed beyond the tumor was likely due to the diffusion of NIR light from the tumor.³⁹ A similar phenomenon may be occurring in the OS cases described here. This may be an advantage during surgery as NIR imaging may provide an additional buffer of fluorescent but tumor-free tissue, ensuring that the surgical margins do not impinge upon the tumor. In this study, NIR imaging was performed at four different levels of tissue dissection: intact skin, skin removed, soft tissue dissected to the tumor, and bone sectioned. The dissection of soft tissues for the different levels of imaging prevented accurate histologic assessment of the soft tissue margins. Further studies would be warranted to specifically assess these soft tissue margins after en-block resection of the tumor (residual tumor) during limb-sparing surgery.

In this case series, dogs with pulmonary metastasis were excluded. However, NIR fluorescent imaging may improve the results of metastasectomy, and further studies would be warranted to explore this potential in OS metastatic lesions. Pulmonary metastasectomy is part of multimodality OS treatment and is associated with improved survival in human patients with resectable metastases.^40,41 The survival benefit for sarcoma metastasectomy is greatest when all macrometastatic lesions are removed.^42,43 However, during thoracotomy, 34–37% of sarcoma patients have additional lung nodules discovered by palpation that were not seen on preoperative computed tomographic imaging.^44,45 Surgeons use both visual and tactile assessment to identify pulmonary nodules during thoracotomy, but tactile sensation is lost when metastasectomy is performed via thoracoscopy. NIR imaging has improved the detection of pulmonary lesions identified on preoperative imaging and those not identified by intraoperative visualization or palpation during thoracotomy and video-assisted thoracic surgery.⁴⁶ Intraoperative NIR imaging using indocyanine green detected an additional 42 lesions in 31 of 52 sarcoma patients compared to a historical, non-imaging control group. The near-infrared imaging group had significantly longer progression-free survivals and median times to recurrence than the control group.⁴⁶ Indocyanine green is a non-specific tumor imaging agent that cannot differentiate between neoplastic and inflammatory pulmonary lesions.¹⁹ Imaging using molecularly targeted agents, including a folate receptor targeted agent and VGT-309, has been described for primary pulmonary tumors,^11,28 and in one case of metastatic OS, where an additional sub-centimeter nodule not seen on CT scan was visualized during thoracoscopic surgery.⁴⁷ Further translation studies and clinical trials are required to evaluate possible advantages of using VGT 309 for intraoperative identification of metastatic pulmonary OS lesions.

One dog in the study (Case 7) had aggressive bony lesions of both the proximal humerus and mid-body scapula, both diagnosed as osteosarcoma by histopathology. The scapular lesion is suspected to be a secondary lesion and may represent bony metastasis, as this is not a common bone to be affected by this disease process; however, primary (or de novo) osteosarcoma cannot be excluded. This scapular lesion was substantially necrotic. Necrosis likely explains the lack of fluorescence noted with this tumor compared to the others evaluated in this study. This same dog had the largest disparity between tumor length measurements on MRI vs. NIR imaging (29.8mm), with the fluorescent imaging measurement lesser than that of the MRI. Histologically, at the level of fluorescence cessation (notched cortical bone) neoplastic cells did not extend to the level of the notched bone, but sections were taken in the more distant region noted on the MRI (approximately 1 to 1.5 cm from notch) and were positive for tumor. Particularly given the scapular lesion in the same case, there is the possibility of extension including discrete islands of tumor within the medullary cavity. Additionally, as with the scapula, this humeral tumor was substantially necrotic. The lack of vascular supply may have prevented the uptake of contrast during the study and affected the sensitivity of NIR imaging. While further evaluation is warranted, caution should be taken to evaluate the margins of particularly necrotic tumors using this imaging modality. Without including this outlier, the mean discrepancy in measurement length for the remaining tumors is 4.4mm. Only one other dog, #11, had an MRI measurement 2.4mm greater than the NIR measurement; NIR measurement was greater than MRI measurement in all other cases.

All regional lymph nodes imaged in this study were positive for near-infrared fluorescence but showed no evidence of metastatic disease on histopathology. Cathepsins expressed by tumor cells and tumor-associated macrophages⁴⁸ draining to local lymph nodes, and cathepsin expression by resident lymphocytes and macrophages likely explains the lymph node fluorescence. The findings of this study suggest that near-infrared imaging using VGT-309 should not be used to guide assessment of lymph nodes for metastatic disease in canine OS.⁴⁸

This study is considered a pilot study and is limited by a small sample size. MRI is not routinely performed in dogs undergoing amputations. The risks associated with prolonged general anesthetic time led the authors to limit this sample size to one that established the feasibility of this imaging modality. The successful use of this imaging agent in dogs should be translated with caution to humans and other species of interest. There is a lack of standardized protocols for calibrating optical imaging devices; this data was collected using a single imaging device, but further study is warranted to evaluate the detection of this fluorophore using other commercially available units. While this study evaluated the margins of fluorescence of primary osteosarcoma lesions, further studies would be warranted to assess the detection of satellite lesions, metastatic lesions, and local recurrence after initial excision. The histogram analysis of this study was limited to the red color channel of each image to standardize objective measures, however, this may confound the estimate of visual fluorescence in the green and blue channels. While OS is not commonly subclassified in dogs, the uptake of VGT-309 may be affected by tumor type and associated metabolic profiles. Human studies using this agent would be warranted to evaluate these variables further.

The results of this pilot study support the use of VGT-309 near-infrared imaging to detect the presence of osteosarcoma tissues and margins intra-operatively. The ability to use real-time imaging to help guide surgical excision may improve completeness of excision while decreasing morbidity in limb-sparing procedures.

Statement of translational relevance.

This study describes the preclinical evaluation of a cathepsin activity-based probe (VGT-309) for the intraoperative imaging of appendicular osteosarcoma in an established, spontaneous canine model of osteosarcoma. Intraoperative near-infrared (NIR) fluorescence using VGT-309 accurately delineated the margins of canine osteosarcomas and correlated with margins seen on preoperative magnetic resonance imaging and postoperative histopathology. Intra-operative NIR tumor imaging during human limb sparing surgery may help to identify neoplastic margins, allowing for complete tumor resection, decreasing the incidence of both local recurrence and subsequent metastatic disease, and improving patient survival. This study supports the feasibility of cathepsin-based intraoperative NIR imaging in limb-sparing surgery for osteosarcoma.