A Novel Method of Core Aspirate Cytology Compared to Fine-Needle Aspiration for Diagnosing Canine Osteosarcoma

Introduction

Osteosarcoma (OSA) is the most common primary osseous neoplasm in dogs, accounting for up to 85% of primary bone tumors.¹ OSA is an extremely aggressive tumor, with as many as 90% of patients having micrometastatic disease present at diagnosis.¹ It is important to differentiate OSA from other less common primary bone tumors, such as fibrosarcoma, chondrosarcoma (CSA), and hemangiosarcoma, as the prognosis for these tumors can vary greatly. Median survival time for dogs with OSA ranges from 5 mo with amputation alone to 10–14 mo when combined with other treatment modalities.^2–5 Amputation alone can be curative for fibrosarcoma and can extend median survival time for CSA to 32.5 mo.^1,6 An early and accurate diagnosis obtained by the least invasive method possible is important to decrease patient discomfort and allow owners to make informed treatment decisions.

There is a paucity of information regarding the utility and accuracy of aspirate cytology of bone lesions in dogs compared with human literature.^7–10 Diagnostic accuracy rates for fine needle aspiration (FNA) of bone lesions in dogs range from 69% to 92%.^7–9 Fine needle aspirate cytology of bone lesions has been extensively examined in human medicine, with diagnostic accuracy rates of 64–94%.^11–19 When diagnosing OSA specifically, FNA was diagnostic in 65% and 75% of human patients in two studies.^11,12 Reasons for inconclusive results in these studies included poor bone penetration and misclassification of cytologic samples from OSA variants.^11,12

Histopathology remains the gold standard for diagnosing OSA. Histopathologic samples can be obtained by open surgical biopsy, trephine biopsy, or Jamshidi biopsy with reported accuracy rates of 82–94%.^20,21 The advantage of the sample size obtained by large bore needle biopsy may be offset by the risk of hematoma formation, wound breakdown, infection, local seeding of tumor, or pathologic fracture. The benefits of FNA cytology of bone include decreased morbidity and pain, minimal disruption of bone integrity, short procedural time, and rapid diagnosis.¹² Limitations of cytology may include hypocellular samples, difficulty in differentiating reactive bone from neoplastic bone, and difficulty differentiating between various bone tumors.^11,12

Recently, an alkaline phosphatase (ALP) stain was developed to aid in the cytologic diagnosis of canine OSA.¹⁰ ALP is a membrane bound enzyme that is found in most mammalian organs.¹⁰ Bone is the only connective tissue that produces ALP in dogs; however, the ALP stain cannot differentiate between normal and neoplastic osteoblasts.¹⁰ Therefore, careful cytologic analysis of the sample for evidence of malignancy and lack of inflammatory cells is essential for appropriate interpretation. ALP staining is performed and interpreted after a diagnosis of malignant neoplasia has been confirmed cytologically, as reactive osteoblasts will also stain positive. The high sensitivity and specificity of the ALP stain for OSA make it a useful adjunct for diagnosing OSA cytologically.¹⁰

To the authors’ knowledge, there have been no reports detailing the use of cytology obtained with a bone marrow biopsy needle for the diagnosis of osseous lesions in dogs. In an effort to avoid the problems associated with intact cortical bone and increase diagnostic yield, this study design incorporated a novel diagnostic technique, termed core aspirate cytology (CA), using a bone marrow biopsy needle to penetrate the cis-cortex and acquire cytologic aspirates with a larger bore needle than FNA. The purpose of this study was to determine the diagnostic accuracy of CA and compare it with FNA using ALP staining for the diagnosis of canine OSA. The authors hypothesized that CA would result in significantly more diagnostic samples than FNA and that the results would compare favorably with the reported diagnostic rates of core biopsy in dogs.

Materials and Methods

Client-owned dogs presenting for diagnosis or treatment of lytic and/or proliferative bony lesions at Affiliated Veterinary Specialists in Orange Park, FL were prospectively sampled between February 1, 2008 and February 28, 2010. Informed consent was obtained from all owners before sample collection. Dogs were excluded from the study if the owner did not consent to sampling or the primary investigator was not available to perform the sampling procedures. Dogs were premedicated with hydromorphone^a (0.08 mg/kg subcutaneously) and general anesthesia induced with propofol^b (4 mg/kg IV) and maintained with isoflurane (2–3% in oxygen). After aseptic preparation of the skin overlying the lesion, FNA was performed using a 22 gauge, 1.5 inch needle attached to a 6 cc syringe. Aspiration was performed until blood filled the needle hub. Aspirates were spread on glass slides in a routine manner. After FNA was performed, CA samples were obtained using a 16 gauge bone marrow biopsy needle^c. A stab incision was made over the lesion with a number 15 Bard-Parker blade. The biopsy needle was inserted and penetrated the cis-cortex of the affected site with the stylet in place. After stylet removal, a 20 cc syringe was attached to the needle and aspirated until blood entered the syringe. The needle and syringe were removed together and slides prepared as for FNA. A maximum of three attempts were made using each cytologic technique to produce at least four slides for evaluation. After cytologic sampling, histologic samples were immediately obtained by Jamshidi-style needle biopsy with the same needle used for CA or collected after amputation within 2 weeks of initial sampling.

Cytologic Evaluation

All cytologic evaluations were performed by the same board-certified clinical pathologist. Information regarding the breed, age, sex of the patient, and the location of the lesion was provided. Slides were stained with Wright-Giemsa for initial diagnosis (Figure 1). A cytologic diagnosis of sarcoma was made based on the presence of an abnormal population of mesenchymal cells exhibiting multiple criteria of malignancy, including variation in nuclear to cytoplasmic ratio, anisocytosis, anisokaryosis, prominent nucleoli, and cytoplasmic vacuolization. If malignant mesenchymal cells were identified, the remaining slides were coated with nitroblue tetrazolium chloride/5-bromo-4-chloro-3-indolyl phosphate toluidine salt solution (ALP stain) for 8–10 min at room temperature, then rinsed with tap water. A minimum of 10 fields were randomly examined at 50× magnification. Positive staining for ALP was indicated by distinct grayish black to brown staining of the cell surface of >75% of the neoplastic cells (Figure 2, Figures 3A,B). Imprints of canine kidney were used as a positive control. A cytologic diagnosis of sarcoma that stained positive for ALP was considered consistent with OSA. A sarcoma that was negative for ALP was considered a non-OSA sarcoma. Cytologic specimens containing any combination of blood, inflammatory cells, reactive osteoclasts, or reactive osteoblasts were recorded as reactive bone (RB). All cytologic specimens from each patient were evaluated at the same time and the clinical pathologist was aware of the results of FNA, CA, and ALP.

View Figure

• Download as Powerpoint
• Download Figure

View Figure

• Download as Powerpoint
• Download Figure

View Figure

• Download as Powerpoint
• Download Figure

Results

Twenty-seven dogs with lytic and/or proliferative bony lesions were sampled during the study. Their signalment, lesion location, and diagnostic test results are summarized in Table 1. The most common breeds were Labrador retriever (n=6), rottweiler (n=5), mastiff (n=3), greyhound (n=3), and Doberman pinschers (n=2). There were one intact female, one intact male, 18 spayed females, and 7 neutered males. Mean age and body weight were 8.6 yr and 39 kg, respectively. The most common locations sampled were the distal radius (n=7), proximal humerus (n=6), proximal femur, and distal scapula (n=2 each).

TABLE 1 Summary of Results of Fine Needle Aspiration and Core Aspirate Cytology for the Diagnosis of OSA in Dogs

−, alkaline phosphatase stain negative; +, alkaline phosphatase stain positive; dist, distal; FS, female, spayed; MN, male, neutered; OSA, osteosarcoma; prox, proximal; RB, reactive bone; S, sarcoma

View Fullscreen

Twenty dogs were diagnosed histologically as OSA, two as non-OSA sarcomas, and five as RB. Cytologic diagnosis using FNA was sarcoma in 18 samples and RB in 9. The CA diagnosis was sarcoma in 21 samples and RB in 6. After a diagnosis of sarcoma was made cytologically, 17 FNA and 19 CA samples stained positive for ALP.

All sarcomas that stained positive with ALP on FNA and CA were diagnosed histologically as OSA. All sarcomas that stained negative for ALP were diagnosed as non-OSA sarcomas histologically. The sensitivity of ALP for OSA cytologically was 100% in this study. There were too few non-OSA tumors to calculate specificity.

For dogs with histologically confirmed OSA, FNA was diagnostic in 17 of 20 (85%) cases and CA diagnostic in 19 of 20 (95%) cases. The difference in diagnostic accuracy between FNA and CA for the diagnosis of OSA was not significant (P=0.60). When all histologic diagnoses were evaluated, the overall diagnostic accuracy in this study was 23 of 27 (85%) FNAs and 26 of 27 (96%) CAs. The difference in overall diagnostic accuracy between FNA and CA was not significant (P=0.35).

Discussion

The results of this study support the use of cytology with ALP staining for the diagnosis of OSA in dogs. For dogs with OSA in this study, the cytologic diagnosis was accurate in 85% of FNA and 95% of CA. CA allowed for penetration of cortical bone using a larger needle and increased diagnostic accuracy, although the difference between techniques was not significant. ALP staining proved to be a valuable aid in diagnosis with 100% sensitivity for OSA.

Studies in human medicine cite an intact cortex as a technical difficulty associated with FNA of bone.^11–13 The radiographic appearance of certain bony lesions is associated with a high rate of insufficient samples, and the use of FNA is avoided.¹³ Large cystic lesions, lesions with an intact cortex and minimal soft tissue involvement, and lesions with a dense, calcified matrix are considered poor candidates for FNA.^11,13 The authors wanted to determine whether using a bone marrow biopsy needle to penetrate the cis-cortex would consistently obtain diagnostic samples from bony lesions. Although there was an increased diagnostic rate with CA, the difference was not significant. This might be due to small sample size, or the lack of intact cortical bone associated with OSA, so that FNA could easily penetrate the medullary mass. It was also possible that the increased diagnostic accuracy rate with CA was due to the larger needle size, rather than the ability to penetrate cortical bone. Needle size appeared to have little influence on sample quality, as the cellularity of the two sampling methods was similar in most cases. Both techniques were easy to perform, required no specialized equipment, and did not result in any complications. The use of 22 gauge needles for FNA used in this study were also associated with a lower risk of tumor seeding compared with core or open biopsies in people.¹⁸ This might be beneficial when considering treatments such as limb sparing in which residual tumor could be catastrophic. The effect of needle size on bone cytology in dogs warrants further research as this study could not distinguish between the effects of needle size and cortical bone penetration.

Few studies have evaluated the accuracy of cytology for the diagnosis of OSA in dogs without the use of advanced imaging. In a preliminary study comparing FNA cytology and histopathology, a sensitivity of 69% and specificity of 73% were reported for the diagnosis of OSA.⁸ In that report, cytologic findings of inflammation, reactive osteoblasts, or nondiagnostic samples were seen in aspirates later identified histologically as OSA.⁸ Another study found partial or full agreement in 70% of the cytologic and histologic diagnoses when evaluating malignant versus nonmalignant bony lesions.⁹

In a recent retrospective study, the overall diagnostic accuracy of FNA cytology of bone lesions was 71% compared with histopathology.⁷ The cytologic diagnosis correlated with histopathology in 92% of neoplastic lesions, although no specific data were presented for dogs with histologically confirmed OSA.⁷ There were several limitations to that study. These included not differentiating between tumor types, inconsistent use and reporting of imaging assistance, and the exclusion of cytologic samples with poor cellularity, resulting in an increased percentage of agreement between cytology and histopathology.⁷ The authors’ diagnostic rates for OSA compared favorably with that study, and no samples were excluded from analysis. Similarly, the authors defined diagnostic accuracy as correlation between cytology and histopathology. One inherent difficulty in diagnosing OSA was accurate sampling of the tumor and avoiding the region of transitional RB. Although cytology and histopathology might have agreed in cases diagnosed as RB, it was not possible to exclude an underlying neoplastic process in these cases.

In this study, five dogs were diagnosed histologically as RB. As the cytology agreed in all five, the diagnoses were considered accurate. Of these dogs, two had no evidence of clinical or radiographic disease progression on follow-up examination (3 and 4 mo, respectively). Two dogs were euthanized with progression of clinical and radiographic signs and were suspected of having neoplasia; however, a final histologic diagnosis was not obtained. The final dog was rebiopsied 9 mo later as OSA. It was unknown if the initial diagnosis of RB made by FNA, CA, and biopsy was correct, and the lesion underwent malignant transformation, or if the original tests failed to identify the underlying disease process. The exclusion of these dogs from OSA analysis might have inadvertently increased diagnostic accuracy. Additionally, as samples were submitted on a case-by-case basis, the clinical pathologist was not blinded to the results of FNA, CA, and ALP, which might have introduced a source of bias into the comparison of diagnostic accuracy.

One case of OSA diagnosed by FNA was interpreted as RB by CA. The use of successive cytologic collection techniques might have led to the false diagnosis of RB by disrupting tumor or traumatizing regional bone and soft tissues. Histopathology confirmed the diagnosis of OSA in that case. Four sarcomas (three OSA, one non-OSA sarcoma) diagnosed by CA were interpreted as RB by FNA. Cytologic findings in these cases mainly consisted of blood and platelet clumps with occasional osteoblasts and inflammatory cells. The authors suspected this resulted from lack of cortical bone penetration, although inaccurate needle placement could not be excluded without the use of imaging assistance. Histopathology confirmed the CA diagnosis in all four cases.

The use of advanced imaging techniques has been previously investigated to aid in accurate sampling of bony lesions. In one study, the use of computed tomographic guidance resulted in 5 of 6 (83%) diagnostic aspirates and 17 of 17 (100%) diagnostic biopsies.²² Using ultrasound to guide FNA of bone lesions produced diagnostic samples in 11 of 23 (48%) and 32 of 36 (89%) lesions in two separate studies.^23,24 Ultrasound has the advantage of guiding the needle through a break in cortical bone and directly into a lesion; however, the technique requires specialized equipment and an experienced sonographer. All cytologic samples in this study were obtained without any imaging assistance. The technique of CA avoided the problems associated with intact cortical bone, but also required accurate localization of the lesion. In this study, passage of the bone marrow needle into the medullary cavity was confirmed by palpation only. In most cases it was easy to palpate penetration of the weakened cis-cortex and feel the needle hit the trans-cortex when advanced after stylet removal. However, aspiration from outside the medullary cavity could not be entirely excluded without the use of fluoroscopy or radiography to confirm needle location.

Historically, OSA has been difficult to differentiate from other sarcomas cytologically.²⁵ The cytologic characteristics of canine osteosarcoma have been described.²⁵ These include angular nucleoli, anisonucleosis, macronucleolization, nuclear molding, absent mitoses, cytoplasmic vacuolization, anisocytosis, and anisokaryosis.²⁵ The high sensitivity and specificity of the ALP stain for OSA made it useful for differentiating OSA from other malignant mesenchymal neoplasms.¹⁰ In an initial study to differentiate OSA from other vimentin-positive tumors, ALP staining had 100% sensitivity and 89% specificity for canine OSA.¹⁰ In that study, 33 of 33 OSAs, 1 of 1 amelanotic melanoma, 1 of 1 multilobulated tumor of bone, and 1 of 4 CSAs stained positive for ALP, whereas 3 of 4 CSAs, multiple fibrosarcomas, soft tissue sarcomas, synovial cell tumors, plasma cell tumors, mast cell tumors, and lymphomas did not stain positive for ALP.¹⁰ In this study, ALP staining was 100% sensitive for OSA, which was consistent with the use of the ALP stain in two previous veterinary reports.^10,23

Conclusion

The authors were able to diagnose OSA cytologically using ALP staining in 85% and 95% of FNA and CA samples, respectively. The results of this study compared favorably with the diagnostic rate of FNA for OSA in humans and the diagnostic rate of core biopsies in dogs. In dogs with OSA, the authors were unable to detect a significant difference in diagnostic accuracy between FNA and CA. CA might be useful in cases with intact cortical bone that could not be diagnosed by FNA. In conclusion, the authors recommend cytology obtained by FNA with ALP staining as an initial diagnostic test or in conjunction with needle biopsy for the evaluation of suspected OSA lesions in dogs.