Bone infarcts are areas of osteonecrosis that develop as a result of ischemia.1 Typical histologic findings such as focal loss of vascular perfusion, edema, adipocyte necrosis, and proliferation of fibroblasts from bone marrow stromal cells can be seen in marrow spaces within a few days after an ischemic insult.2 By 2 weeks, changes in cancellous bone located in the focal area of infarction, including an absence of bone lining cells and empty osteocyte lacunae, can be seen. By this time, the periphery of the infarct will have evidence of capillary proliferation, fibroplasia, foamy macrophages engulfing lipid from necrotic adipocytes, and osteoclastic resorption of the necrotic bone. Eventually, some surfaces of necrotic bone begin to be covered by woven bone through a process known as creeping substitution, a feature of infarcted bone first described by Phemister3 in the early 1900s.
In the veterinary literature, bone infarcts have been reported in association with various causes, including tibial plateau leveling osteotomy, total hip arthroplasty, and neoplasia (eg, sarcoma).4–10 However, an underlying etiology is not always determined, in which case a nondescript vascular event is presumed.11–13
Two previous reports4,6 described radiographic findings in 9 and 4 dogs with sarcoma-associated bone infarcts. Bone infarcts were characterized as “irregularly demarcated areas of increased radiopacity in the medullary cavities” in one report4 and as “organized areas of increased radiodensity, which usually extended the full length of the bone,” in the other.6 However, the radiographic appearance of benign bone infarcts has not been described previously, and there are no reports comparing the radiographic appearance of benign bone infarcts with the appearance of bone infarcts associated with neoplasia. The objectives of the study reported here were to describe the radiographic appearance of benign bone infarcts and bone infarcts associated with neoplasia in dogs and to determine the utility of radiography in differentiating benign from malignancy-associated bone infarcts.
Materials and Methods
The study was designed as a retrospective cohort study. The database of a referral osteopathology service provided by one of the authors (RRP) was searched for cases submitted between 2007 and 2017 for which a histologic diagnosis of bone infarction had been made. Cases were included in the study if they involved dogs and if corresponding histologic reports and diagnostic-quality radiographic images of the lesion were available for review.
For all cases included in the study, histologic reports were reviewed, and cases were classified as bone infarction with no evidence of local primary bone tumor (benign infarction) or as bone infarction associated with a primary bone tumor (malignancy-associated infarction). Information on patient signalment (ie, age, breed, and sex) was recorded, along with the type of specimen that had been submitted for examination (ie, excisional biopsy specimen, amputated limb, necropsy specimen, or needle biopsy specimen) and the number of samples (if known). For cases classified as benign infarction, the referring veterinarian was contacted for follow-up information if possible.
Radiographs for cases included in the study were anonymized and randomized with a web-based random number generator. Images were then independently reviewed by 2 board-certified veterinary radiologists (LJG and CMR). Reviewers were provided only the radiographs and were blinded to patient signalment and histologic diagnosis. Radiographs were evaluated by each reviewer for evidence of cortical destruction (present or absent), a transition zone (distinct or indistinct), type of bone activity (sclerotic, lytic, or mixed), pattern of medullary lysis (absent, nonaggressive, or aggressive), and pattern of periosteal proliferation (absent, nonaggressive, or aggressive). Geographic medullary lysis was classified as nonaggressive, whereas permeative and punctate lysis were classified as aggressive. A smooth periosteal reaction was classified as nonaggressive, and sunburst, spiculated, or palisading periosteal reactions were classified as aggressive. Classifications were based on classic criteria for differentiating benign from aggressive bone lesions.14
Affected bone, location within the bone (epiphyseal, metaphyseal, diaphyseal, or mixed), and presence or absence of a pathological fracture were also recorded. Reviewers then classified each case as likely benign infarct, likely malignancy-associated infarct, or undistinguishable, as they would for a routine clinical case. This classification was based on the reviewer's experience and radiographic impression. Once all classifications were assigned, the reviewers worked together to resolve any disagreements regarding classifications assigned for the 5 radiographic descriptors; however, no attempt was made to reach agreement on the final radiographic diagnosis assigned by each reviewer.
Statistical analysis
Statistical analyses were performed by one of the authors (SAJ) with commercially available software.a Sensitivity and specificity of each reviewer in correctly classifying lesions, with benign infarction as the condition of interest, were calculated; cases classified as undistinguishable were considered an incorrect classification for this portion of the analysis. Kappa analysis was performed to test for agreement between reviewers. The Fisher exact test was used to determine whether there were significant associations between the histologic diagnosis and each of the 5 radiographic descriptors (cortical destruction, transition zone, bone activity, pattern of medullary lysis, and pattern of periosteal proliferation) or between the histologic diagnosis and presence of polyostotic versus monostotic lesions. For bone activity, the presence of sclerosis was classified as a nonaggressive feature, whereas lysis and a mixed pattern were classified as aggressive features. For both periosteal proliferation and pattern of medullary lysis, the absence of these features was classified with nonaggressive features (smooth and geographic, respectively) for statistical analysis. All tests were 2 sided. Values of P ≤ 0.05 were considered significant.
Results
Forty-nine cases fulfilled the criteria for inclusion in the study. Orthogonal radiographic projections were available for 44 of the 49 cases, and a single radiographic projection was available for the remaining 5. For 35 cases, specimens submitted for histologic examination had been obtained by means of biopsy (needle biopsy, n = 33; core biopsy, 1; and wedge biopsy, 1); for 27 of these 35 cases, biopsies had been performed at multiple sites. One specimen was a fracture fragment, and 13 specimens had been obtained from an amputated limb or at the time of necropsy.
Thirty-three cases were classified histologically as benign infarction, and 16 were classified histologically as malignancy-associated infarction (Figures 1–4). Sixteen of the 33 benign infarcts had histologic evidence of arteriosclerosis and idiopathic capillary bed degeneration or collapse without evidence of thrombosis. No cases of bone infarction associated with osteomyelitis were identified.
Mediolateral (A) and craniocaudal (B) radiographic images of a 2-year-old Labrador Retriever with a benign infarct involving the distal aspect of the femur, and photomicrographs (C and D) of a biopsy specimen from the lesion. On the radiographic images, there is a monostotic lesion with aggressive medullary lysis and periosteal proliferation, cortical destruction, and an indistinct transition zone. Histologically, Haversian vessels and canals were avascular and degenerative and most of the osteocyte lacunae were empty, indicative of chronic ischemia (C). H&E stain; bar = 400 μm. Intertrabecular marrow spaces contained edematous fatty marrow and a few ischemic hematopoietic cells but no inflammatory cell exudate or neoplastic infltrate (D). H&E stain; bar = 400 μm.
Citation: Journal of the American Veterinary Medical Association 256, 12; 10.2460/javma.256.12.1352
Mediolateral (A) and craniocaudal (B) radiographic images of a 3-year-old Airedale Terrier with a benign infarct involving the proximal aspect of the tibia, and photomicrographs (C and D) of a biopsy specimen from the lesion. On the radiographic images, there is a monostotic lesion without cortical destruction or periosteal proliferation and with a distinct transition zone and aggressive medullary lysis. Histologically, the tibia contained a border of periosteal new bone that was continuous with the outer surface of the cortical bone (C); no inflammatory cell exudate was seen. H&E stain; bar = 40 μm. The stroma of the fatty marrow contained edema fluid and a few degenerative, small-caliber arterioles with thickened walls associated with ischemia of the adipose tissue (D); empty osteocyte lacunae and peritrabecular fibrosis were not evident. H&E stain; bar = 400 μm.
Citation: Journal of the American Veterinary Medical Association 256, 12; 10.2460/javma.256.12.1352
Mediolateral (A) and craniocaudal (B) radiographic images of a 5-year-old Great Dane with osteosarcoma involving the distal aspect of the radius, along with photographs (C and D) and photomicrographs (E and F) of the lesion. On the radiographic images, there is a monostotic lesion with aggressive medullary lysis, cortical destruction, an aggressive periosteal reaction, and an indistinct transition zone. In photographs of the lesion, the tumor is fusiform and located entirely within the distal aspect of the radius but has partially entrapped part of the surface of the ulna. Histologically, osteosarcoma tissue has nearly completely replaced the fatty marrow (E and F). H&E stain; bar = 200 μm (E) and 100 μm (F).
Citation: Journal of the American Veterinary Medical Association 256, 12; 10.2460/javma.256.12.1352
Oblique (A) and mediolateral (B) radiographic images of a 3-year-old Doberman Pinscher with an osteosarcoma involving the distal aspect of the femur, and photomicrographs (C and D) of a biopsy specimen from the lesion. On the radiographic images, there is a monostotic lesion without cortical destruction or periosteal reaction but with an indistinct transition zone and aggressive medullary lysis. Histologically, ischemic cancellous and cortical lamellar bone was composed almost entirely of lamellar (mature) bone with almost no formation of woven (immature or reactive) bone (C). H&E stain; bar = 400 μm. A nodule of spindle cells representing the border of a spindle cell tumor can be seen (D). H&E stain; bar = 300 μm.
Citation: Journal of the American Veterinary Medical Association 256, 12; 10.2460/javma.256.12.1352
Sixteen (48%) of the benign infarcts and 6 (38%) of the malignancy-associated infarcts were correctly classified by both reviewers. One of the reviewers classified 7 cases as undistinguishable, whereas the other reviewer correctly classified all 7 of these cases as benign infarcts (Table 1). Interrater agreement was only moderate (κ = 0.438; 95% CI, 0.246 to 0.63). For one reviewer, sensitivity of identifying benign infarcts was 52% (17/33), and specificity was 69% (11/16); for the other reviewer, sensitivity of identifying benign infarcts was 70% (23/33), and specificity was 56% (9/16).
Agreement between classifications assigned by 2 board-certified veterinary radiologists who reviewed radiographs from 49 dogs with histologically confirmed benign (n = 33) or malignancy-associated (16) bone infarcts involving the appendicular skeleton.
| Reviewer 2 classification | |||
|---|---|---|---|
| Reviewer 1 classification | Correct | Incorrect | Undistinguishable |
| Correct | 22 | 6 | 0 |
| Incorrect | 3 | 11 | 0 |
| Undistinguishable | 7 | 0 | 0 |
Benign infarcts
Of the 33 dogs with benign infarcts, 13 (39%) were spayed females, 1 (3%) was a sexually intact female, 17 (52%) were castrated males, and 1 (3%) was a sexually intact male; sex of the remaining dog was not known. Age ranged from 9 months to 12 years (median, 7 years; mean, 6.7 years). There were 6 Labrador Retrievers, 5 mixed-breed dogs, 2 Greyhounds, 2 Siberian Huskies, 2 German Shepherd Dogs, 2 Borzois, 2 pit bull–type dogs, 2 Golden Retrievers, 2 Mastiffs, 1 Great Dane, 1 Saint Bernard, 1 Airedale Terrier, 1 Akbash, 1 Akita, 1 Alaskan Klee Kai, and 1 Yorkshire Terrier; breed of the remaining dog was not known. Lesions were identified in the femur (n = 4), humerus (10), radius (6), tibia (4), ulna (1), and intermedioradial carpal bone (1); the remaining 7 dogs had polyostotic lesions. Twenty (61%) dogs had multifocal lesions (ie, lesions affecting contiguous regions of bone that included > 1 of the epiphysis, diaphysis, and metaphysis), 8 (24%) had diaphyseal lesions, 4 (12%) had metaphyseal lesions, and 1 (3%) had a lesion involving a cuboidal bone. None of the lesions were confined only to the epiphysis. Four dogs had associated pathological fractures. Additional or follow-up information was available for 6 dogs with benign infarcts. In 3 of these, the radiographic appearance of the lesions was unchanged at the time of recheck examination (12 months, 18 months, and 4 years). Two dogs had full resolution of clinical signs (lameness) by 2 and 10 months. One dog did not have any changes in clinical signs but also did not have any radiographic evidence of thoracic metastases after 14 months.
Malignancy-associated infarcts
Of the 16 dogs with malignancy-associated infarcts, 9 were spayed females, 1 was a sexually intact female, and 5 were castrated males; sex of the remaining dog was not known. Age of 2 dogs was not known; for the remaining dogs, age ranged from 18 months to 14 years (median, 5 years; mean, 6.6 years). There were 4 mixed-breed dogs, 3 Golden Retrievers, 2 Great Danes, 1 German Shepherd Dog, 1 Doberman Pinscher, 1 Labrador Retriever, 1 Saint Bernard, 1 pit bull–type dog, and 1 Anatolian Shepherd; breed of the remaining dog was not known. Lesions involved the femur (n = 5), humerus (4), radius (5), tibia (1), and ulna (1); none were polyostotic. Seven lesions were multifocal (ie, affecting contiguous regions of bone that included > 1 of the epiphysis, diaphysis, and metaphysis), 6 were metaphyseal, and 3 diaphyseal. None of the lesions were confined only to the epiphysis. Five dogs had associated pathological fractures.
Radiographic descriptors
Of the 5 radiographic descriptors (cortical destruction, transition zone, bone activity, pattern of medullary lysis, and pattern of periosteal proliferation), only the pattern of medullary lysis (absent or nonaggressive vs aggressive; P = 0.011) and the pattern of periosteal proliferation (absent or nonaggressive vs aggressive; P = 0.044) were significantly associated with the histologic diagnosis (benign vs malignancy-associated infarct; Table 2). All 16 (100%) malignancy-associated infarcts had aggressive medullary lysis, compared with 23 of the 33 (70%) benign infarcts that did. Eight of the 16 (50%) malignancy-associated infarcts had aggressive periosteal proliferation, compared with 7 of the 33 (21%) benign infarcts that did.
Radiographic characteristics identified by 2 board-certified veterinary radiologists who reviewed radiographs from 49 dogs with histologically confirmed benign (n = 33) or malignancy-associated (16) bone infarcts involving the appendicular skeleton.
| Histologic classification | ||||
|---|---|---|---|---|
| Radiographic characteristic | Category | Benign | Malignancy associated | P value* |
| Cortical destruction | 0.310 | |||
| Absent | 10 (30) | 3 (19) | ||
| Present | 23 (70) | 13 (81) | ||
| Transition zone | 0.124 | |||
| Distinct | 5 (15) | 0 (0) | ||
| Indistinct | 28 (85) | 16 (100) | ||
| Bone activity | 0.124 | |||
| Lytic | 1 (3) | 0 (0) | ||
| Sclerotic | 5 (15) | 0 (0) | ||
| Mixed | 27 (82) | 16 (100) | ||
| Medullary lysis | 0.011 | |||
| Absent | 5 (15) | 0 (0) | ||
| Nonaggressive | 5 (15) | 0 (0) | ||
| Aggressive | 23 (70) | 16 (100) | ||
| Periosteal | 0.044 | |||
| proliferation | Absent | 12 (36) | 3 (19) | |
| Nonaggressive | 14 (42) | 5 (31) | ||
| Aggressive | 7 (21) | 8 (50) | ||
Data are given as number of dogs (percentage).
Results of a Fisher exact test for an association between radiographic characteristic and histologic diagnosis. For bone activity, lysis and mixed (aggressive characteristics) were grouped and compared with sclerotic (nonaggressive characteristic). For medullary lysis, absent and nonaggressive (geographic) were grouped and compared with aggressive (permeative or punctate). For periosteal proliferation, absent and nonaggressive (smooth) were grouped and compared with aggressive (sunburst, spiculated, or palisading).
Discussion
Results of the present study suggested that radiography was not particularly helpful in distinguishing benign from malignancy-associated bone infarcts in dogs. Only 16 of 33 (48%) benign infarcts and 6 of 16 (38%) malignancy-associated infarcts were correctly identified by both reviewers, and for both reviewers, the sensitivity of identifying benign infarcts was low, indicating that the discriminatory value of radiography was poor.
Previously published descriptions of the radiographic appearance of bone infarcts in dogs are scarce. However, radiographic markers of aggressive bone lesions have been described, with lesions that have cortical destruction, an indistinct transition zone, or certain types of periosteal proliferation generally considered to be aggressive.14 Malignancy-associated infarcts in the present study generally fit this description, and many were located in sites typically associated with primary bone tumors. However, many of the benign infarcts also had these radiographic characteristics. For example, 5 of the 33 benign infarcts were associated with aggressive osteolysis in the distal aspect of the radial diaphysis or distal metaphysis, 3 were associated with aggressive lysis in the proximal aspect of the tibia, and 3 were associated with aggressive lysis in the distal aspect of the femur.
Periosteal proliferation represents a reaction of cortical bone to various types of insult. Local vascular compromise is a less commonly recognized cause of periosteal proliferation and has been postulated to occur secondary to alterations in the local interstitial fluid pressure secondary to venous stasis.15 Classically, the pattern of periosteal proliferation has been considered to depend on the inciting cause. In the present study, however, even though the pattern of periosteal proliferation (absent or nonaggressive vs aggressive) was significantly associated with the histologic diagnosis (benign vs malignancy-associated infarct), only 8 of 16 malignant infarcts had aggressive periosteal proliferation, and 7 of 33 (21%) benign infarcts did also. Because ischemia is often an acute event, it is not surprising that variations in periosteal proliferation in response to benign infarction could occur, and the pattern of periosteal proliferation may be more related to the chronicity of an infarct or the period of time over which it occurred.
The pattern of medullary lysis (absent or geographic vs permeative or punctate) was also significantly associated with the histologic diagnosis (benign vs malignancy-associated infarct) in the present study. All 16 malignancy-associated infarcts had aggressive medullary lysis; however, 23 of the 33 (70%) benign infarcts also had aggressive medullary lysis. It was unclear to us why some benign infarcts had aggressive medullary lysis, but we postulated that this appearance was related to differential occlusion of the terminal capillary network. The ensuing variable remodeling response could have resulted in medullary heterogeneity that appeared radiographically like punctate or permeative lysis.
There was a broad spectrum of breeds represented in the present study population; however, most of the dogs were larger. Three of the dogs were Great Danes, and 2 of them had infarcts secondary to osteosarcoma. This breed and other large breeds of dogs are overrepresented among dogs with appendicular osteosarcoma.16–21 The only dog in our study that was known to be a small-breed dog was a Yorkshire Terrier, and this dog had a benign infarct. However, there were 9 mixed-breed dogs, all of unknown weight, so the true number of small-breed dogs in the study was not known, prohibiting further discussion regarding their predisposition to infarcts. Previous reports4,6 have suggested that small-breed dogs, particularly Miniature Schnauzers, are overrepresented among dogs with benign infarcts, but this was not the case for the present study, which did not include any Miniature Schnauzers.
Most of the benign and malignancy-associated infarcts in the present study were located in the metaphysis or adjacent diaphysis, locations typically associated with primary bone tumors. Those that were not were all located in the diaphysis, other than 1 benign infarct in a cuboidal bone (intermedioradial carpal bone). No polyostotic malignancy-associated infarcts were identified, whereas 7 of the 33 (21%) benign infarcts were polyostotic. Of these, 4 had histologic evidence of vascular thrombosis. A disseminated cause for benign bone infarction is commonly suspected; therefore, the presence of polyostotic disease in this group was unsurprising. Hematogenous disease, such as canine distemper virus infection, has been previously implicated as a cause for metaphyseal lesions, but is more commonly seen in younger patients.22 Of the 7 dogs with polyostotic disease in the present study, only 1 was young (2 years).
Prior to initiating our study, we had anticipated that pathological fractures would be more commonly seen with malignancy-associated infarcts than with benign infarcts. However, 4 of the 9 pathological fractures were seen in dogs with benign infarcts. Given that cortical destruction was a common radiographic finding in dogs with both benign and malignancy-associated infarcts, it was not surprising that pathological fractures were seen in both groups.
For all cases in the present study, the histologic diagnosis had been made by an experienced orthopedic pathologist (RRP). In most instances, multiple biopsy specimens and radiographs of the biopsy site were available to assist the pathologist in making their diagnosis. However, it is possible that not all biopsy specimens were representative of the lesion, particularly in the few cases for which only a single specimen was available. Thus, some benign cases may have been misclassified, particularly those without follow up information. In addition, it has been postulated that infarction may predispose to later malignant transformation,23 possibly as a result of excessive reparative cell proliferation.24 Longitudinal imaging and histologic studies of presumed benign infarcts would be required to further investigate the possibility of this phenomenon in dogs.
Cases included in the present study had primarily been submitted to the osteopathology service for a second or third opinion, and the cases were worldwide in distribution and thus not representative of specific geographic regions. The proportion of benign infarcts that have an aggressive radiographic appearance is unknown for any given clinical population, and extrapolation of the proportion of benign versus malignancy-associated infarcts with certain radiographic characteristics is discouraged because of biases associated with the referral population. It is recognized also that the reviewing radiologists could not be entirely blinded to signalment owing to inherent radiographic features, such as bone shape and size, typical for specific dog breeds. Additionally, body weight of the dogs was not recorded.
In conclusion, this study has demonstrated the wide variability and considerable overlap in the radiographic appearance of benign and malignancy-associated bone infarcts in dogs. Thus, additional diagnostic testing should be pursued even for osseous lesions with an aggressive radiographic appearance. Patient signalment and lesion location did not appear to be useful in differentiating benign from malignancy-associated infarcts in the present study. However, polyostotic disease was identified only in dogs with benign infarcts, and all malignancy-associated infarcts were monostotic.
Acknowledgments
The authors thank Dr. John F. Griffin for his input regarding the statistical analysis.
Footnotes
SPSS Statistics for Windows, version 24.0, IBM Corp, Armonk, NY.
References
1. Fondi C, Franchi A. Definition of bone necrosis by the pathologist. Clin Cases Miner Bone Metab 2007;4:21–26.
2. Catto M. Pathology of aseptic bone necrosis. In: Davidson JK, ed. Aseptic necrosis of bone. Amsterdam: Excerpta Medica, 1976;3–100.
3. Phemister D. Necrotic bone and the subsequent changes which it undergoes. JAMA 1915;64:211–216.
4. Dubielzig RR, Biery DN, Brodey RS. Bone sarcomas associated with multifocal medullary bone infarction in dogs. J Am Vet Med Assoc 1981;179:64–68.
5. Marcellin-Little DJ, DeYoung DJ, Thrall DE, et al. Osteosarcoma at the site of bone infarction associated with total hip arthroplasty in a dog. Vet Surg 1999;28:54–60.
6. Riser WH, Brodey RS, Biery DN. Bone infarctions associated with malignant bone tumors in dogs. J Am Vet Med Assoc 1972;160:414–421.
7. Newman ME, Johnson KA. Suspected intramedullary bone infarct subsequent to tibial plateau levelling osteotomy in a dog. Aust Vet J 2015;93:255–258.
8. Haney DR, Peck JN. Influence of canal preparation depth on the incidence of femoral medullary infarction with Zurich Cementless Canine Total Hip arthroplasty. Vet Surg 2009;38:673–676.
9. Marsolais GS, Peck JN, Berry C, et al. Femoral medullary infarction prevalence with the Zurich Cementless Canine Total Hip arthroplasty. Vet Surg 2009;38:677–680.
10. Sebestyen P, Marcellin-Little DJ, DeYoung BA. Femoral medullary infarction secondary to canine total hip arthroplasty. Vet Surg 2000;29:227–236.
11. Fenger CK, Bertone JJ, Biller D, et al. Generalized medullary infarction of the long bones in a horse. J Am Vet Med Assoc 1993;202:621–623.
12. Martig S, Lippold BS, Oevermann A, et al. Polyostotic bone lesions consistent with bone infarction in a horse. Vet Rec 2008;162:352–353.
13. Sánchez J, Gonzalo-Orden JM, Ginja MMD, et al. Imaging diagnosis–medullary tibial infarction in a horse. Vet Radiol Ultrasound 2010;51:159–161.
14. Thrall DE. Textbook of veterinary diagnostic radiology. 6th ed. St Louis: Elsevier, 2013.
15. Rana RS, Wu JS, Eisenberg RL. Periosteal reaction. AJR Am J Roentgenol 2009;193:W259–W272.
16. Spodnick GJ, Berg J, Rand WM, et al. Prognosis for dogs with appendicular osteosarcoma treated by amputation alone: 162 cases (1978–1988). J Am Vet Med Assoc 1992;200:995–999.
17. Anfinsen KP, Grotmol T, Bruland OS, et al. Breed-specific incidence rates of canine primary bone tumors–a population based survey of dogs in Norway. Can J Vet Res 2011;75:209–215.
18. Withrow SJ, Vail DM, Page RL. Withrow and MacEwen's small animal clinical oncology. 5th ed. St Louis: Elsevier, 2013.
19. Misdorp W, Hart AA. Some prognostic and epidemiologic factors in canine osteosarcoma. J Natl Cancer Inst 1979;62:537–545.
20. Ru G, Terracini B, Glickman LT. Host related risk factors for canine osteosarcoma. Vet J 1998;156:31–39.
21. Rosenberger JA, Pablo NV, Crawford PC. Prevalence of and intrinsic risk factors for appendicular osteosarcoma in dogs: 179 cases (1996–2005). J Am Vet Med Assoc 2007;231:1076–1080.
22. Baumgärtner W, Boyce RW, Alldinger S, et al. Metaphyseal bone lesions in young dogs with systemic canine distemper virus infection. Vet Microbiol 1995;44:201–209.
23. Bahk WJ, Lee AH, Kang YK, et al. Infarct associated sarcoma: a possible pathogenesis based on histological observation of repair tissue origin in two cases. Acta Oncol 2010;49:868–872.
24. Mirra JM. Secondary malignant fibrous histiocytoma (arising in bone infarcts). In: Mirra JM, Picci P, Gold RH, eds. Bone tumors: clinical, radiologic, and pathologic correlations. Philadelphia: Lea & Febiger, 1989.
