Use of a three-dimensional-printed spine model for surgical planning in a feline vertebral chondroblastic osteosarcoma

Abstract

Case summary

A 2-year-old, male neutered domestic shorthair cat presented with a 1-month history of non-specific pain and inappetence, left-sided Horner syndrome, pelvic limb proprioceptive ataxia, upper motor neuron paraparesis and a plantigrade stance. Advanced imaging revealed a heterogeneous contrast-enhancing extradural mass dorsal to the T1–T2 intervertebral disc space, causing severe left-sided spinal cord compression. Medical image processing software was used to generate a patient-specific three-dimensional spine model based on MRI and CT data. A biomechanical three-dimensional-printed replica of the T1–T3 region allowed preoperative simulation of a left-sided hemilaminectomy and mass removal, confirming contiguity of the lesion with the T1 vertebral body and extension along the vertebral canal to T2–T3. The model aided intraoperative landmark identification and facilitated safe and effective spinal cord decompression. Histopathology from surgical biopsy was most consistent with either chondroblastic osteosarcoma or chondrosarcoma. Postoperative radiation therapy was initiated but discontinued owing to difficulties establishing reliable intravenous access. Progressive neurologic deterioration eventually led to euthanasia. Post-mortem CT and necropsy confirmed a diagnosis of chondroblastic osteosarcoma without evidence of metastatic disease. The cat had a survival time of 356 days after surgery (387 days from initial presentation).

Relevance and novel information

 This report describes the first known vertebral chondroblastic osteosarcoma in a cat. It highlights the value of biomechanical three-dimensional-printed spine models for preoperative planning in complex vertebral surgery and demonstrates that tumor debulking, even when combined with incomplete radiotherapy, can provide meaningful palliation and prolonged survival. Further studies are needed to define optimal treatment strategies and prognosis for feline vertebral osteosarcoma subtypes.

Plain language summary

Using a three-dimensional-printed spine model to help plan surgery for a rare spinal tumor in a cat

A young cat was referred for investigation of ongoing pain and difficulty walking. Advanced imaging showed a mass pressing on the spinal cord in the upper chest region of the spine. To better understand the shape and position of the lesion, a patient-specific three-dimensional-printed model of the affected spinal area was created using MRI and CT scans. This model allowed the surgical team to practice the planned procedure in advance and to identify important landmarks, helping to safely remove part of the mass and relieve pressure on the spinal cord. The cat improved after surgery and later received radiation therapy, although treatment could not be completed. Further testing after death confirmed a rare type of bone cancer of the spine called chondroblastic osteosarcoma. This is the first reported case of this tumor type affecting the vertebrae in a cat. This report shows how three-dimensional-printed spine models can be a valuable tool for planning complex spinal surgery and may help improve quality of life and survival time, even when a cure is not possible.

Introduction

Primary bone and cartilaginous tumors are rare in cats.^{1 –4} This report describes a monostotic vertebral chondroblastic osteosarcoma (OSA) in a young cat, a subtype not previously reported as a primary vertebral tumor in cats, and the first use of a patient-specific biomechanical three-dimensional-printed spinal model to assist surgical planning in feline vertebral tumor debulking. Diagnostic challenges, treatment and outcome are presented.

Case description

A 2-year-old, male neutered domestic shorthair cat was referred with a 1-month history of non-specific pain and inappetence. The cat was strictly indoor, fully vaccinated and tested negative for feline immunodeficiency virus/feline leukemia virus.

Neurologic examination revealed ambulatory upper motor neuron (UMN) paraparesis, mild pelvic limb proprioceptive ataxia, a plantigrade stance, left-sided Horner syndrome and mild cranial thoracic spinal pain. Neuroanatomical localization was consistent with the cervicothoracic spinal cord, more specifically the T1–T3 segments. The absence of thoracic limb deficits suggested focal or subclinical involvement of the cervicothoracic intumescence. The plantigrade posture was considered compensatory; however, a second lesion could not be completely excluded based on the neurologic examination alone.

Routine hematology, serum biochemistry and Toxoplasma gondii serology results were within reference intervals. Thoracic radiographs showed no evidence of metastatic disease and no visible vertebral abnormalities.

Spinal MRI (3.0 T, MAGNETOM Skyra; Siemens) revealed an extradural mass extending from mid-T1 to mid-T2, causing marked left-sided spinal cord compression. The lesion was heterogeneously isointense on T1-weighted (T1W), heterogeneously hypointense on T2-weighted (T2W) and proton-density (PD) sequences, and showed strong, heterogeneous contrast enhancement (Figure 1).

Figure 1

MRI scans of the thoracic spine: (a) parasagittal T2, (b) transverse T2-weighted and (c) transverse T1-weighted fat-saturated post contrast. A heterogeneous extra-dural mass (yellow arrow) is visible at the level of the T1 vertebra, causing severe left-sided spinal cord compression (blue arrow).

CT (64-slice Somatom Perspective; Siemens) confirmed a heterogeneously hyperattenuating osseous mass arising from the left aspect of the T1 vertebral body, resulting in severe vertebral canal narrowing (Figure 2a-1). Cisternal cerebrospinal fluid analysis was within normal limits.

Figure 2

Serial CT images showing the progression of the vertebral tumor in a cat. (a) Preoperative parasagittal and (a-1) transverse, (b) immediate postoperative parasagittal and (b-1) transverse, (c) 112 days postoperative parasagittal and (c-1) transverse, and (d) post-mortem (performed 356 days postoperatively) parasagittal and (d-1) transverse images are shown. A hyperattenuating extradural mass (yellow stars and arrowheads) is centered at the T1–T2 vertebrae, causing vertebral canal narrowing and left foraminal stenosis. Progressive regrowth of the bony mass is evident after surgical debulking

Based on signalment and imaging features, vertebral angiomatosis was considered the most likely diagnosis, with OSA and chondrosarcoma (CSA) as differentials.

Initial treatment with gabapentin (10 mg/kg PO q8h) and methylprednisolone (0.8 mg/kg PO q24h), later switched to liquid prednisolone (1 mg/kg PO q24h) for easier administration, improved appetite, activity and spinal pain. Spinal pain later recurred, supporting surgical decompression as a treatment option.

To aid preoperative planning, a three-dimensional-printed spine model was generated by merging MRI and CT data sets using Mimics (Materialise) and printed with a Stratasys J850 Digital Anatomy 3D Printer (Stratasys). The model reproduced both osseous and lesion density, allowing detailed rehearsal of the surgical approach (Figure 3a,b). Simulation confirmed that optimal exposure required a wide left-sided T1–T3 hemilaminectomy and that the lesion was continuous solely with the T1 vertebral body. A pneumatic drill was then used on the model, using the intact vertebral floor as a landmark to accurately determine drilling depth and width while minimizing the risk of iatrogenic spinal cord injury (Figure 3c).

Figure 3

A biomechanical three-dimensional-printed spine model was used to simulate a left-sided hemilaminectomy. (a,b) The tumor is visible on the ventral aspect of the T1 vertebra and within the vertebral canal (red arrow and dashed circle). (c) After surgical simulation, the mass was confirmed to be continuous solely with the T1 vertebral body, allowing for precise planning of the required drilling length and width to minimize the risk of iatrogenic spinal cord injury. The spinal cord is represented within the model (c), providing a realistic reference for surgical access and trajectory

Approximately 1 month after the initial presentation, a left T1–T3 hemilaminectomy was performed. The mass displaced the T1–T2 spinal nerve dorsally and was carefully drilled free from its T1 attachment, crushed ventrally and removed while preserving the nerve root (Figure 4). Postoperative CT confirmed satisfactory spinal cord decompression (Figure 2b-1).

Figure 4

Intraoperative image after a left-sided T1–T3 hemilaminectomy, demonstrating spinal cord decompression with preservation of the T1 nerve root

Perioperative management included intravenous fluid therapy (0.45% NaCl 2 ml/kg/h with KCl 0.05 mEq/kg/h), ketamine (2 mg/kg/h IV constant rate infusion [CRI]), fentanyl (3 µg/kg/h IV CRI), maropitant (1 mg/kg IV q24h), ondansetron (0.54 mg/kg IV q8h), dexamethasone sodium phosphate (0.08 mg/kg IV q24h) and methocarbamol (29.2 mg/kg IV q8h).

Postoperative ketamine and fentanyl CRIs were subsequently tapered over 48 h. Spinal pain resolved and neurologic status improved, with a marked reduction in plantigrade stance, ataxia and Horner syndrome. The cat was discharged after 48 h on gabapentin (10 mg/kg PO q8h, as needed) and a tapering course of prednisolone (0.5 mg/kg PO q24h).

Macroscopically, the mass was yellow-tan and mineralized. Histologically, the biopsied tissue consisted of an unencapsulated, poorly demarcated, densely cellular mass replacing bone (Figure 5a). It was composed of sheets and streams of round to spindloid neoplastic cells, frequently surrounded by variably sized islands of basophilic amorphous matrix (chondroid) and, less frequently, small aggregates of eosinophilic homogenous matrix (osteoid). Neoplastic cells had variably distinct cell borders with scant lightly eosinophilic cytoplasm and an eccentric round to oval hyperchromatic nucleus encompassing one to two variably prominent nucleoli. Moderate anisocytosis and anisokaryosis were noted, with seven mitotic figures per 10 high-power fields (2.37 mm²). In several areas, neoplastic cells surrounded variably sized irregular deposits of basophilic mineral (presumed mineralized chondroid).

Figure 5

Photomicrograph of a vertebral chondroblastic osteosarcoma in a cat. (a) Histopathologic examination of the biopsied tissue revealed neoplastic cells surrounding and embedded within variably sized islands of basophilic amorphous matrix (chondroid, asterisk) and scant eosinophilic homogenous matrix (arrowhead). Hematoxylin and eosin (H&E), bar = 20 µm. (b) Post-mortem examination revealed an irregular, hard, pale tan bony mass arising from the T1 vertebral body and crossing the joint space to the T2 vertebrae. The mass protrudes into the vertebral canal, laterally compressing the spinal cord. (c) Histopathology of the affected vertebrae reveals an infiltrative and lytic mass effacing T1 (arrowhead) and extending into the adjacent vertebrae (asterisk). H&E, bar = 2 mm. (d) A higher magnification of the neoplastic population reveals neoplastic cells have variably distinct cell borders with scant lightly eosinophilic cytoplasm and an eccentric round to oval hyperchromatic nucleus encompassing one or two variably prominent nucleoli. Neoplastic cells frequently surround foci of eosinophilic fibrillar material (osteoid). H&E, bar = 20 µm.

Approximately 75% of the neoplastic cells exhibited nuclear immunoreactivity to the special AT-rich sequence-binding protein 2 (SATB2) antibody, supporting an osteoblastic or chondroblastic origin. Alkaline phosphatase staining could not be performed because of the unavailability of a cytologic specimen. ⁵

The absence of well differentiated and proliferative vascular profiles ruled out angiomatosis, while the high cellularity and absence of well-differentiated islands of chondroid with surrounding osteoid made osteochondromatosis unlikely.

After review by four pathologists, including specialists in osteopathology (Dr Brian Murphy) and neuropathology (Dr Martí Pumarola), the findings were most consistent with either a chondroblastic OSA or CSA with endochondral ossification.

The cat continued to improve, with only mild pelvic limb ataxia. Approximately 4 weeks postoperatively, fractionated radiation therapy was initiated (planned total dose 50 Gy in 20 daily fractions of 2.5 Gy), targeting the original tumor bed, surgically disrupted field, and one vertebral body cranial and caudal to the lesion to account for possible microscopic extension (ie, clinical target volume) and daily setup uncertainties (planning target volume). Treatment was discontinued after seven fractions because reliable intravenous access could not be established and the owner declined vascular access port placement.

Follow-up CT at 112 days postoperatively (143 days from initial presentation) showed tumor regrowth with vertebral canal stenosis and left-sided foraminal stenosis at C7–T1 and T1–T2 (Figure 2c-1), although spinal cord compression remained less severe than preoperatively. No pulmonary metastases were detected. Spinal pain and episodic miosis gradually recurred and were managed medically. Further surgery was declined, and gabapentin (5–10 mg/kg PO q8–12h, as needed) and escalating doses of methylprednisolone (up to 1.2 mg/kg PO q24h) were used to provide palliation.

Progressive UMN paraparesis and pelvic limb ataxia developed at 331 days postoperatively, and euthanasia was elected at 356 days postoperatively (387 days from initial presentation).

Post-mortem CT revealed extensive progression of the lesion involving T1 vertebral body with extension into C7 and T2, resulting in severe foraminal stenosis (Figure 2d-1). No metastatic disease was identified. Necropsy confirmed an infiltrative vertebral neoplasm compressing the spinal cord with associated malacia (Figure 5b).

Histopathology revealed a locally infiltrative and destructive mesenchymal neoplasm producing abundant eosinophilic matrix, consistent with osteoid and chondroid matrix (Figure 5c,d), with associated reactive callus formation. The spinal cord at T1–T2 showed mild-to-moderate axonal degeneration, correlating with the gross findings of compression and displacement. Increased osteoid production compared with the initial biopsy supported a final diagnosis of chondroblastic OSA.

Discussion

Biomechanical three-dimensional-printed spine models are increasingly used in human neurosurgery for surgical planning and simulation^6,7; however, to the authors’ knowledge, these have not been reported in feline spinal surgery. In this case, the patient-specific model enabled precise spatial localization of the lesion and realistic reproduction of texture and mechanical characteristics of the tumor and surrounding bone, facilitating preoperative planning and safe optimization of the surgical approach. Although more aggressive surgical options such as vertebrectomy could theoretically be considered for vertebral neoplasms, preoperative simulation confirmed that hemilaminectomy would provide adequate spinal cord decompression while minimizing surgical morbidity in this cervicothoracic location.

OSA is the most common primary malignant bone tumor in cats; however, vertebral OSA is rarely reported. It typically affects middle-aged to older animals^1,8 and, compared with canine OSA, has a lower metastatic potential, making surgical excision the treatment of choice.^1,2,4 Nevertheless, axial OSAs carry a guarded prognosis (median survival time 3.7–6.7 months) owing to local invasiveness and the difficulty of achieving complete resection, although surgery may still provide meaningful palliation.^1,2,4,9 Improved survival has been reported when surgery is combined with adjuvant radiotherapy and/or chemotherapy,^{1,2,4,9 –12} and only one previous case report describes vertebrectomy for feline vertebral OSA, ¹¹ documenting survival exceeding 5 years after aggressive multimodal treatment.

Histologic subtypes of OSA analogous to those described in dogs have been reported in cats, although subtype does not appear to carry prognostic significance.^{2,10 –12} Published imaging descriptions of feline vertebral OSAs are limited to isolated reports, including giant-cell, ¹¹ osteoblastic, ¹² fibroblastic ¹² and mixed ¹⁰ variants. In these reports, giant-cell, fibroblastic and mixed variants typically appear as centrally lytic lesions with peripheral mineralization on CT^{10 –12} and T1-isointense, T2-hypointense signal on MRI with strong, uniform contrast enhancement, ¹⁰ whereas osteoblastic OSAs show heterogeneous hyperintensity on T1, T2 and short-TI inversion recovery sequences with diffuse enhancement. ¹²

The cat in this report was younger than typically reported for feline OSA.^1,2 The plantigrade stance observed at presentation resolved after spinal cord decompression, making a primary peripheral neuropathy unlikely. Chronic UMN dysfunction with secondary disuse weakness and compensatory posturing was considered a plausible explanation, although electrophysiologic assessment would be required to confirm this.

Imaging features differed from those described for other feline vertebral OSA subtypes, with the lesion showing minimal osseous lysis and appearing heterogeneously isointense on T1W images, hypointense on T2W and PD sequences, and heterogeneously hyperattenuating on CT with strong contrast enhancement. Differentiation between CSA and OSA was challenging, ¹³ as biopsy samples were dominated by chondroid matrix. SATB2 immunolabelling was positive but not definitive, reflecting a known overlap between CSA and OSA and the limited validation of this marker in cats.^14,15 Alkaline phosphatase staining, which may aid in distinguishing OSA from other mesenchymal tumors, could not be performed because of the absence of a cytologic specimen. Definitive diagnosis was achieved only at necropsy, which demonstrated abundant osteoid production by malignant mesenchymal cells, ruling out CSA and confirming chondroblastic OSA, a subtype that, to the authors’ knowledge, has not previously been reported as a primary vertebral tumor in cats.

No metastatic disease was identified at post-mortem examination, consistent with the low metastatic rate reported in feline OSA. Despite incomplete radiotherapy and the inability to pursue chemotherapy, surgical debulking provided meaningful palliation and prolonged survival. Progressive involvement of adjacent vertebrae ultimately precluded further intervention.

Study limitations include the absence of follow-up spinal MRI, inability to evaluate the full effects of adjuvant radiotherapy or chemotherapy, and the lack of alkaline phosphatase at initial diagnosis.

Conclusions

This report describes a rare monostotic vertebral chondroblastic OSA in a young cat. A patient-specific biomechanical three-dimensional–printed spine model proved valuable for surgical planning and safe tumor debulking. Surgical decompression combined with radiotherapy, although the latter was incomplete, provided meaningful palliation and prolonged survival. Further studies are warranted to clarify optimal treatment strategies and prognosis for feline vertebral OSA subtypes.