Introduction

Osteosarcoma is the most common primary malignant bone neoplasia in dogs and in children and adolescents, affecting mainly the appendicular skeleton1,2,3. Certain breeds, such as Rottweilers and Great Danes, and body weight > 40 kg are factors associated with the occurrence of osteosarcoma that have been identified previously3,4.

Prognosis is generally guarded for both humans and dogs with osteosarcoma due to its aggressive biological behavior. The 5-year event-free survival (EFS) for humans is reported to be 27.4% with metastatic lung disease detected at the time of diagnosis5,6. The median survival times in dogs with surgery alone and with adjuvant chemotherapy are reported to be on average five months and 10–14 months, respectively7,8,9,10,11. In one study, 21% of dogs were alive at two years after diagnosis, compared to only 2% in an older study; this difference likely reflects the use of multimodal therapy in the more recent cohort versus surgery alone in the earlier study10,12.

Negative prognostic indicators have been assessed and identified in many studies, but not necessarily replicable, including younger or older age (< 5 years old or > 10 years old)10, higher body weight (> 40 kg)11, and the presence of regional lymph node metastasis13. Studies focusing on histopathological characteristics also did not reach a consensus14,15,16,17,18. In studies including only canine appendicular osteosarcomas, the fibroblastic subtype was shown to have longer survival compared to osteoblastic or chondroblastic osteosarcomas14. Another smaller study of canine osteosarcoma did not find a correlation between histological subtypes and prognosis15. In a meta-analysis study on negative clinicopathological prognostic factors, only elevated serum alkaline phosphatase (ALP) levels and proximal humeral location reached statistical significance1.

In humans, microvascular invasion has been studied in recent years as a potential prognosticator in osteosarcoma19,20. In pediatric patients who had received chemotherapy prior to surgical resection, the presence of peritumoral microvascular invasion was associated with a lower two-year event-free survival (EFS) and overall survival time (OST)19. Microvascular invasion was again shown to be associated with lower overall survival and higher recurrence or metastasis, especially in those with poor response to pre-operative chemotherapy in another human osteosarcoma study20. In dogs, approximately 70% of osteosarcoma samples were found to have microvascular invasion16,21. In one of these studies, microvascular invasion in canine osteosarcoma was found to have a significantly higher hazard of death (hazard ratio: 8.3)16.

Given the scarcity of studies investigating the prognostic significance of microvascular invasion in canine osteosarcoma and the lack of consensus as to whether histological subtypes have prognostic significance in osteosarcoma, the goal of this study was to identify histopathological prognostic factors in canine appendicular osteosarcoma. We hypothesize that the presence of microvascular invasion is associated with negative prognoses for disease-free progression and survival and that histological subtypes can provide prognostic data.

Materials and methods

Case selection

Electronic medical records from the Virginia Tech Animal Cancer Care and Research Center (VT-ACCRC) and from the Virginia Tech Veterinary Teaching Hospital (VT-VTH) were retrospectively reviewed for client-owned dogs with appendicular osteosarcoma from May 15, 2014, to March 9, 2022. Informed consent was obtained from all pet owners. All procedures were carried out in accordance with relevant guidelines and regulations. Dogs were included in the study if they had received thoracic radiographs or CT scan for staging prior to limb amputation surgery (partial or full), had a histopathological diagnosis of primary osteosarcoma arising from the appendicular bones. Dogs that received adjuvant treatment with chemotherapy and/or radiation therapy after amputation were included. Dogs were excluded if they did not receive surgical tumor resection, had non-appendicular osteosarcoma, had presence of metastasis at initial diagnosis via thoracic radiograph or CT scan, or lack of complete follow-up information.

Medical record review

Historical data at the time of diagnosis was obtained from the electronic medical records and included age, sex, neuter status, breed, body weight, serum ALP levels, anatomic location of primary tumor, date of osteosarcoma diagnosis, tumor histopathological evaluation, the presence of regional or distant metastases as determined via CT scans or thoracic radiographs, and abdominal ultrasound, the presence of ulcerated tumor or pathological fracture. Patient follow-up information after surgical treatment was acquired from medical records from recheck visits at the VT-ACCRC, VT-VTH and/or referring veterinarians, and from phone calls to owners requesting information about their pets. Follow-up information obtained included the presence of surgical site infection, the use of adjuvant therapy (chemotherapy, radiation therapy, nonsteroidal anti-inflammatory drugs [NSAIDs]), development of local recurrence or distant metastases, anatomic location of metastases, date of last follow-up if the patient was alive, or date and cause of euthanasia or death.

Histopathological review

All primary tumor specimens were reviewed by a single boarded pathologist (SCO), who was blinded to original histopathological results and clinical outcomes. Subtypes of osteosarcoma (chondroblastic, osteoblastic, fibroblastic, giant cell, telangiectatic, and poorly differentiated) were assigned based on the World Health Organization classification of tumors of soft tissue and bone22. The authors categorized tumors with distinct histologic features of two or more of the existing subtypes as a combined subtype, when there is no one subtype that predominates more than the other in the examined section. The following histopathological results were recorded in addition to tumor subtype: presence of microvascular invasion, mitotic counts performed in an area of 2.37 mm2 in 10 contiguous fields at 400× magnification avoiding areas of the tumor that are cell-poor from hemorrhage, edema, necrosis, cysts, presence of lymphatic invasion, and status of histologically tumor-free margin (HTFM). Lymphovascular invasion was defined by neoplastic cells within a space lined by lymphatic or blood vascular endothelium on H&E staining. HTFMs were determined on microscopic examination to be negative if no tumor cells contacted the surgical edges of the resection, and positive if tumor cells contacted the surgical edges of the resection. For patients who had lymph node extirpation performed at the time of primary tumor removal, the presence of nodal metastasis was evaluated.

Outcome definitions

Local recurrence was defined as tumor regrowth at or near surgical sites of tumor resection, as determined by the attending clinician based on physical examination, radiographic or CT findings, and/or confirmation via histopathology. The presence of distant metastases was defined as presumptive metastatic lesions noted on radiographic/CT or ultrasonographic findings and/or confirmed metastatic lesions obtained via cytology or histopathology. The date of osteosarcoma diagnosis was defined as the date of surgical resection. Progression-free survival time (PFS) was defined as the median time from the date of osteosarcoma diagnosis to the date of documented local recurrence or distant metastases. Overall survival time (OST) was defined as the median survival time from the date of diagnosis to the date of last follow-up if patient was still alive or the date of death due to disease. Death was attributed to disease if local tumor recurrence or metastasis was determined to be the cause. Unknown causes of death were attributed to death due to disease.

Statistical analysis

Dogs were censored in the PFS analysis if they had no documented local recurrence or distant metastasis or were alive or lost to follow up at the time of study completion. Dogs were censored in the OST analysis if they had documented death of causes other than osteosarcoma or were alive or lost to follow up at time of study completion. Kaplan-Meier survival function was used to calculate PFS and OST with 95% confidence intervals. Cox proportional hazard model was used to evaluate clinicopathological (serum ALP, primary tumor location) and histological data (subtype and the presence of microvascular invasion) and outcome measures (PFS and OST) using statistical software SAS version 9.4.

Results

Fifty-seven dogs met inclusion criteria during the study period. There were 23 spayed females, 1 intact female, 28 castrated males, and 5 intact males. The median age at the time of osteosarcoma diagnosis was 8.5 years (range: 1.3–12.1 years) and the median weight was 34.8 kg (range: 4.2–66.5 kg). Twenty-three breeds were represented. The most common were mixed-breed dogs (n = 18) and Labrador Retriever (n = 6). Other breeds included Golden Retriever (n = 4), Greyhound (n = 4), German Shepherd Dog (n = 3), Great Pyrenees (n = 2), Bulldog (n = 2), Doberman Pinscher (n = 2), and Great Dane (n = 2), and 1 each of Akita, Alaskan Malamute, Australian Shepherd, beagle, Borzoi, boxer, Chihuahua, Newfoundland, German Shorthaired Pointer, Chesapeake Bay Retriever, Saint Bernard, English Springer Spaniel, Russell Terrier, and Scottish Terrier.

Out of 57 dogs, 16 dogs (28.1%) were documented to have elevated serum ALP. The median elevation above the reference range was 204 U/L (range: 113–1737). The primary tumor locations included femur (n = 16), tibia (n = 13), radius (n = 13), humerus (n = 10), scapula (n = 2), metatarsus/tarsus (n = 1), ulna (n = 1), and fibula (n = 1). Full limb amputations were performed for surgical tumor resections in 56 dogs and a partial limb amputation was performed for one dog. Regional lymph node extirpation was performed in 43 out of 57 (75.4%) dogs who underwent surgical tumor resection. Ulcerated tumor and pathological fracture at time of diagnosis were present in 2/57 (3.5%) and 8/57 (14%) dogs, respectively. No dogs received palliative radiation therapy pre-operatively. Superficial surgical site infection was present in 5/57 (8.8%) dogs. NSAIDs were administered in 56/57 (98.2%) dogs post-operatively. No post-operative chemotherapy information was available in 4 dogs. Of the remaining 53 dogs, 41 (77.3%) dogs received chemotherapy post-operatively. Chemotherapy protocols utilized included single-agent carboplatin for 4 to 6 doses (n = 40) and single-agent lomustine daily for three weeks (n = 1). Toceranib phosphate, pamidronate, and rapamycin was administered in 5 of 40 (12.5%), 1 of 40 (2.5%), and 2 of 40 (5%) dogs that received carboplatin, respectively. One dog had rapamycin only and another dog had toceranib phosphate only.

All 57 tumor resections achieved negative HTFMs. Microvascular invasion was evident in 5/57 (8.8%) dogs (Fig. 1), and none had lymphatic invasion. Of the histological subtypes in the microvascular invasion group, there was one (20%) chondroblasts, two (40%) osteoblastic, and two (40%) telangiectatic (Table 1). Histopathological assessment of the regional lymph node was performed in all 43 dogs where the lymph nodes were extirpated at time of surgery. 1 of 43 (2.3%) had evidence of metastasis. Of the histological subtypes, there were 43/57 (75.4%) osteoblastic, 5/57 (8.8%) combined subtype, 6/57 (10.5%) telangiectatic, 2/57 (3.5%) chondroblastic, and 1/57 (1.8%) giant cell. There were no fibroblastic or poorly differentiated subtypes. The median mitotic count was 12 per 2.37 mm2 (range: 0–84). The median mitotic count with and without microvascular invasion was 20 and 11.5 per 2.37 mm2, respectively. Serum ALP elevation did not demonstrate a significant association with PFS (p = 0.16) or OST (p = 0.21). Similarly, the anatomic location of the primary tumor was not significantly associated with PFS (p = 0.40) or OST (p = 0.67).

Fig. 1
figure 1

Microvascular invasion with viable tumor cells. Muscularis layer of arteriole is outlined with arrows. Osteoid (asterisks) is shown within a vascular space. Hematoxylin and eosin stain, 200x magnification.

Full size image
Table 1 Progression-free survival time and overall survival time in 57 dogs with appendicular osteosarcoma.
Full size table

One dog in this population had ulnar osteosarcoma, received partial radioulnar amputation, and developed local recurrence at mid-radius 76 days post-operatively. 34/57 (59.6%) dogs developed known or suspected distant metastasis, in which 6/34(17.6%) were diagnosed via histopathology or necropsy, 1/34 (2.9%) was diagnosed via cytology, and 27/34 (79.4%) were diagnosed based on CT or thoracic radiograph. The median PFS was 209 days (range: 141–323 days). PFS for each histological subtype is summarized in Table 1. There was no significant difference in PFS between histological subtypes (p = 0.68). As depicted in Fig. 2, there was no significant difference in PFS in dogs with tumoral microvascular invasion (median 271 days; range: N/A) or without microvascular invasion (median 209 days; range: 140–383 days) (p = 0.81).

Fig. 2
figure 2

Kaplan-Meier curves for overall survival (A) and progression free survival (B) in dogs with or without histologic osteosarcoma vascular invasion.

Full size image

Of the 57 dogs, 19 (33.3%) were censored in the OST analysis, with 4/57 (7%) dogs still alive at the time of study completion (median follow-up time 650 days; range: 356–726). Median OST was 241 days (range: 168–498). OSTs for each histological subtype are summarized in Table 1. There was no significant difference in OST between subtypes (p = 0.28). The presence of microvascular invasion was not significantly associated with OST, with a median of 168 days (range: N/A) for dogs with tumoral microvascular invasion and 262 days (range: 189–517) without tumoral microvascular invasion (p = 0.78) (Fig. 2). One of 5 dogs with microvascular invasion had chondroblastic subtype and was alive at time of study completion at 726 days with no evidence of disease progression. Four of 5 dogs with microvascular invasion (2 osteoblastic and 2 telangiectatic) died from the disease. Clinical data are summarized in Table 2.

Table 2 Clinical data, histological features, treatment received, and outcome for 5 dogs with osteosarcoma microvascular invasion.
Full size table

Discussion

The presence of microvascular invasion and histological subtype were not associated with PFS or OST in this population of dogs.

Since microvascular invasion can provide a route for metastases and 90% of canine osteosarcoma have undetectable micrometastases at time of diagnosis, identifying the presence of microvascular invasion can be crucial as anti-angiogenesis target therapies may be beneficial, which is currently being explored in human medicine23,24,25. Only 5/57 (8.8%) tumors in this study had microvascular invasion, so its prevalence in various subtypes could not be determined. No difference was detected in PFS and OST in dogs with or without microvascular invasion. It is possible that this finding is a result of type 2 error due to the low number of tumors with microvascular invasion resulting in the study being underpowered to reveal the association between microvascular invasion and survival. The prevalence of microvascular invasion in this study was 8.8%, which is lower compared to a larger study where microvascular invasion was noted in 117 out of 152 (70.5%) appendicular and axial tumors16. It was shown to be associated with shorter OST but not PFS16. The percentage of appendicular tumors with microvascular invasion was not specified, which could contribute to the difference in prevalence reported. Additionally, the lack of standardization of tissue processing and sectioning may have resulted in discrepancies compared to the current study findings. In Coomber et al., increased microvascular density was significantly higher in canine primary osteosarcoma with pulmonary metastasis at time of diagnosis23. The significance of microvascular invasion in human osteosarcoma is still being investigated, with studies demonstrating decreased survival and poorer response to treatment19,20,26,27.

The majority of osteosarcomas in this population were of the osteoblastic subtype (75.4%), similar to the proportions reported in two recent studies of appendicular osteosarcoma, which found osteoblastic subtypes in 49% of 87 tumors and 56.9% of 153 tumors, respectively14,15. No significant difference in survival could be detected among histological subtypes in this study. However, only the osteoblastic subtype included an adequate number of cases for 95% confidence intervals to be calculated for both PFS and OST. Due to the low number of cases, there was only one giant cell subtype (alive at time of study completion at 356 days) with no fibroblastic or poorly differentiated subtype present. Given the small number of cases in most osteosarcoma subtypes, the authors postulated that the study was underpowered for any statistical significance to be detected. Therefore, interpretation must be made carefully with the current results. To this date, there is no consensus on the impact of histological subtype on survival in canine appendicular osteosarcoma. One study with 87 tumors showed longer survival in dogs with fibroblastic subtype; however, the comparison was only made between osteoblastic, chondroblastic, and fibroblastic subtypes14. Another study that included 22 tumors in the survival analysis did not find histological subtypes (osteoblastic, telangiectatic, chondroblastic, fibroblastic, poorly differentiated, giant cell-rich) to have an influence on survival outcome15. In osteosarcoma of the flat/irregular bones, telangiectatic subtype had the highest prevalence of metastases28 and had an increased hazard of progression29. Meanwhile, histological subtype did not impact survival in other osteosarcoma studies30,31.

Limitations of this study included rarity of disease resulting in a small cohort of dogs with low numbers of some histological subtypes. Additionally, post-mortem examination to confirm the cause of death and incidence of metastasis was also not available in all cases. The retrospective design of the study led to variability in treatment protocols and follow-up imaging schedules. While most dogs received postoperative chemotherapy, there was substantial variation in the chemotherapy protocols, thereby limiting the ability to robustly evaluate chemotherapy as a prognostic factor. Neither microvascular invasion nor histopathologic subtype were significant prognostic indicators in this population. However, median OST was shorter in dogs with microvascular invasion compared to those without. This finding suggests that the prognostic impact of microvascular invasion may have been missed due to type II error. Exclusion of dogs with metastasis at diagnosis may also have resulted in exclusion of that subset dogs with vascular involvement. However, dogs with metastasis at diagnosis have a different survival prognosis, and this dataset did not have a sufficient number of dogs with metastasis at presentation to conduct meaningful statistical analysis. Similarly, the small number of cases within most histological subtypes restricted clinically relevant comparisons. Although neither microvascular invasion nor histopathologic subtype were prognostic in this study population, identification of prognostic indicators in osteosarcoma is important for guidance toward future development of differential treatment strategies32,33,34,35.

This study has provided valuable information on the prevalence of microvascular invasion and histopathologic subtypes for planning future prospective studies. Prospective studies or evaluation of available datasets from the Comparative Oncology Trials Consortium control group can provide additional data to evaluate the prognostic significance of microvascular invasion and histological subtypes in canine appendicular osteosarcoma, with standardization of procedures such as pre-operative and post-operative imaging, use of adjuvant therapy, and routine lymph node extirpation.

Conclusion

Based on the results of this study, microvascular invasion is not significantly associated with decreased survival in dogs with appendicular osteosarcoma potentially due to type II error. Histological subtype does not provide additional prognostic data.