Criteria for Selection of Cases

Medical records from 247 dogs and cats in which ultrasound-guided splenic aspiration was performed at the University of Wisconsin Veterinary Medical Teaching Hospital from August 2002 through June 2005 were retrospectively reviewed. Criteria for inclusion were ultrasonographic examination of the spleen and both cytologic and histologic evaluation of the same lesion or of splenic tissue if no specific lesion was sampled.

Procedures—Cytologic specimens were obtained by fine-needle aspiration, and histologic specimens were obtained via surgical biopsy (n = 10), ultrasound-guided biopsy (2), or necropsy (21). Histologic interpretation was used as the gold standard. Information collected from medical records included signalment, reason for ultrasonographic examination, whether there was agreement between cytologic and histologic descriptions and diagnoses, and time between acquisition of the cytologic and biopsy or necropsy samples.

All ultrasonographic examinations and ultrasound-guided aspirations were performed with an 8-MHz curvilinear or 10-MHz linear-array transducer.^a Aspiration was performed with 22-gauge, 1.5-inch-long needles according to a described procedure.⁵ Static images of examinations were reviewed to determine the ultrasonographic appearance of splenic parenchyma and lesions. All cytology and histology slides were reviewed by a single clinical (RMD) and anatomic (MMS) pathologist, respectively, for confirmation of final diagnoses.

Statistical analysis—Associations between multiple discrete, similar-appearing nodules and malignant disease and between single lesions and benign disease were assessed with the Fisher exact test. Values of P < 0.05 were considered significant.

View Full Size

Cytologic preparation of a fine-needle aspirate (A [modified Wright stain; bar = 25 Mm]) and histologic preparation of a necropsy specimen (B [H&E stain; bar = 50 Mm]) of the spleen of a dog with malignant histiocytosis.

Citation: Journal of the American Veterinary Medical Association 230, 5; 10.2460/javma.230.5.690

• Download Figure
• Download figure as PowerPoint slide

View Full Size

Cytologic preparation of a fine-needle aspirate (A [modified Wright stain; bar = 25 Mm]) and histologic preparation of a surgical biopsy specimen (B [H&E stain; bar = 50 Mm]) of the spleen of a dog with a splenic hematoma with extensive granulation tissue and necrosis.

Citation: Journal of the American Veterinary Medical Association 230, 5; 10.2460/javma.230.5.690

• Download Figure
• Download figure as PowerPoint slide

Results

Thirty-two cases (29 dogs and 3 cats) met the inclusion criteria. One animal was excluded from the study because of an 8-month interval between fine-needle aspiration of the splenic lesion and necropsy. Ages ranged from 4 to 16 years, with median and mean ages of 10 and 9.94 years, respectively. Sixteen animals were spayed females, 15 were neutered males, and 1 was a sexually intact male. Canine breeds included Golden Retriever (n = 7); mixed (5); Labrador Retriever (2); Rottweiler (2); Cocker Spaniel (2); Bassett Hound (2); American Pit Bull Terrier (2); and 1 each of Shih Tzu, Malamute, Miniature Schnauzer, Beagle, English Springer Spaniel, English Setter, and German Shepherd Dog. Feline breeds were domestic shorthair (n = 2) and domestic medium hair (1).

The most frequent reason for abdominal ultrasonographic examination was to screen for neoplastic metastasis. Animals that underwent ultrasonographic imaging for this reason had a known primary neoplasm in another location (n = 8), an abdominal mass detected via palpation or radiography (6), or myelopathy with a possibility of neoplasia (4). Other reasons for examination included signs of abdominal pain or discomfort (n = 3; one of these animals had a palpable abdominal mass); signs of abdominal disease (eg, vomiting and diarrhea) or clinicopathologic findings (eg, high hepatic enzyme activities [8]); or signs of general malaise, lethargy, inappetance, or weight loss (4).

Each of the 32 animals was arbitrarily assigned a number, and cytologic and histologic interpretations of specimens, as well as the method used to obtain the histologic specimen, were summarized (Table 1). Both cytologic and histologic diagnoses were obtained in 31 animals, and in 1 of those, the fine-needle aspirate was inadequate for evaluation. The cytologic diagnosis corresponded to the histologic diagnosis in 19 of 31 (61.3%) cases (Figure 1). Of those, 11 of 19 diagnoses were benign conditions, 8 of 19 were malignant diseases, and 1 of 19 was infectious disease. The diagnoses differed in 5 of 31 (16.1%) cases: 4 of 5 had benign cytologic but malignant histologic interpretation, and 1 of 5 had a malignant cytologic but benign histologic interpretation (Figure 2). In 7 animals, the cytologic specimen contained cells similar to those in histologic sections, but evaluation of tissue architecture was required to distinguish reactive from neoplastic conditions. Thus, these cases were considered to have cytologic and histologic diagnoses that neither agreed nor disagreed. Of those 7 animals, 5 had malignant disease on histologic evaluation, 1 had a reactive lesion, and 1 had infectious or inflammatory disease. On the basis of final histologic diagnosis in the 14 animals with nonneoplastic diseases or conditions, the cytologic diagnosis was correct in 11 animals, not definitive in 2 animals, and incorrect in 1 animal. Of the 17 animals with malignant neoplastic diseases, the cytologic diagnosis was correct in 8 animals, not definitive for but suggestive of neoplasia in 5 animals, and incorrect in 4 animals.

No complications of the aspiration procedure were recorded. In 2 animals, the time from fine-needle aspiration to acquisition of the histologic specimen was 3 and 5 months, but in both instances, the cytologic interpretation matched the histologic interpretation. In the remaining cases, histologic specimens were obtained within days of the aspirates.

Ultrasonographic appearances and histologic diagnoses were summarized (Table 2). Ultrasonographic images sensitively detected splenic lesions (Figures 3 and 4), but no ultrasonographic appearance was specific for any given disease or vice versa.

Single discrete masses were benign diseases in 7 of 10 lesions, malignant in 2 of 10 lesions, and infectious in 1 of 10 lesions. Multiple similar-appearing masses were benign in 2 of 14 lesions, malignant in 11 of 14 lesions, and infectious in 1 of 14 lesions. Results of the Fisher exact test for association between multiple and single lesions and malignant and benign disease, respectively, yielded a value of P = 0.006.

View Full Size

Ultrasonograms of 2 spleens affected with lymphoma. A—Spleen with a generalized heterogeneous appearance. B—Spleen containing multiple, discreet, heterogeneous lymphosarcoma nodules (inside markers). The images reveal the variability in the ultrasonographic appearance of lymphoma.

Citation: Journal of the American Veterinary Medical Association 230, 5; 10.2460/javma.230.5.690

• Download Figure
• Download figure as PowerPoint slide

View Full Size

Ultrasonogram of a spleen with a targetoid nodule. Notice the hypoechoic rim and hyper- to isoechoic center (inside markers). The final diagnosis was metastatic bile duct carcinoma on both cytologic and histologic examination.

Citation: Journal of the American Veterinary Medical Association 230, 5; 10.2460/javma.230.5.690

• Download Figure
• Download figure as PowerPoint slide

Discussion

The increasing use of abdominal ultrasonographic examination as a screening process for veterinary patients at referral institutions has resulted in a substantial increase in detection of splenic lesions that may or may not be related to the clinical course of an animal's disease. The ultrasonographic appearance of a single lesion, whether it is benign or malignant, can be ambiguous, necessitating evaluation of the underlying cellular process. For example, disease may be present in a spleen of normal ultrasonographic appearance and size.^4,14–16 Minimally invasive diagnostic procedures are indispensable in avoiding unnecessary surgical procedures.

Fine-needle aspiration is a safe and effective method for obtaining cells for microscopic examination. In a study¹⁷ involving > 1,000 splenic aspirates from humans, no hemorrhagic complications were associated with needles that were 22 gauge in size or smaller, a result that was corroborated in earlier animal studies^6,13 and in the present study. A low complication rate (5% [15/298]) was reported in another large splenic aspiration study⁸ in humans, although 14 of those complications were defined as mild. Fine-needle aspiration is a good-to-excellent technique for obtaining high-quality specimens for cell evaluation; in an earlier human study,⁸ specimens were successfully obtained in 269 of 298 (90%) splenic aspirations, and diagnostic specimens were obtained in 69 of 74, 95 of 102, and 90 of 97 instances in studies^6,9,11 of nonsplenic lesions in animals. In the present study, 31 of 32 (96.9%) aspiration specimens were diagnostic, and in an earlier study¹³ of splenic disease in animals, 33 of 33 (100%) aspiration specimens were diagnostic. These proportions are dependent on the skill of the aspirator and may improve with ultrasound guidance. In other studies, 3 of 33 (9%) aspiration procedures performed blindly with the intention of sampling the spleen yielded liver tissue,¹³ whereas 1 of 74 (1.4%) ultrasound-guided aspirates intended to sample spleen yielded liver tissue.⁶

The accuracy of a cytologic interpretation of an aspirate, however, is more difficult to assess because of the possibility of missing the complete underlying process as a result of the highly focal sampling. The goal of fine-needle aspiration of a lesion is to obtain adequate numbers of intact cells that are representative of the pathologic process. A high-quality aspirate has an adequate number of intact cells, but the degree to which an underlying disease process is represented by the cells may be unknown. In addition to ease of sampling and rapidity with which results are yielded, another advantage of cytologic evaluation is the individual cellular detail that can be observed. Conversely, an important advantage of histologic evaluation of a lesion is that a larger specimen is typically obtained and the tissue architecture in the lesion can be evaluated, enabling assessment of the invasiveness and extent of the lesion. It may be more difficult to obtain representative samples from certain tissues. For example, in 3 studies^9,10,12 involving the liver in dogs and cats, 34 of 56 (61%), 38 of 97 (39%), and 59 of 124 (48%) fine-needle aspirates were representative of underlying disease, although in the latter study, fine-needle biopsy specimens were compared with wedge-biopsy specimens.¹² However, in a study⁷ of soft tissue sarcomas, 56 of 72 (78%) aspirates accurately reflected the underlying disease process. Cytologic evaluation is often used as a guide to indicate the need for additional testing, including biopsy procedures, when a definite diagnosis cannot be obtained in this manner.¹¹

For comparisons limited to cytologic and histologic evaluation of the spleen, there is little published information on diagnostic accuracy. This may be a result of clinician tendency to perform splenectomy without presurgical aspiration and cytologic evaluation of the lesion or reluctance to perform nonsurgical biopsy of the spleen because of the risk of intra-abdominal hemorrhage. The reported diagnostic accuracy of splenic aspirates is better than that associated with liver aspirates, with 253 of 298 (85%) accurate diagnoses reported in humans⁸ and 14 of 14 accurate diagnoses reported in cats and dogs.¹³ This high accuracy may be related to the predilection for hemolymphatic disease to affect the spleen in dogs and cats and a correspondingly high yield of exfoliating neoplastic cells.¹⁸ Results of the present study indicated a lower rate of accuracy, with 19 of 31 (61.3%) aspirates yielding an accurate diagnosis.

In the 5 animals in which cytologic diagnoses were not in agreement with histologic diagnoses, diseases were typically reported as benign disease when the histologic diagnosis was malignant disease. In 2 of those cases, reactive lymphoid hyperplasia was diagnosed on the basis of cytologic evaluation of a mixed population of lymphoid cells. In 1 animal, there was a predominance of small lymphocytes with small numbers of large lymphocytes and plasma cells. Histologic evaluation revealed small-cell lymphoma. In any lymphoid tissue where small lymphocytes predominate, a cytologic diagnosis of small-cell lymphoma is difficult to make definitively, and in this animal, histologic evaluation of tissue architecture was required to determine that the lymphoid cells were neoplastic. In another animal, the proportions of small and large lymphoid cells were equal, although large-cell lymphoma was diagnosed on the basis of histologic evaluation. In the spleen, reactive lymphoid hyperplasia and lymphoma often cannot be distinguished by cytologic evaluation alone. When a predominance of large lymphoid cells is detected in an aspirate, these cells may represent either a hyper-plastic lymphoid follicle or lymphoma.¹⁹ In such instances, the ultrasonographic appearance of the spleen is an important consideration. If the cells are from a discrete splenic nodule, a conservative interpretation of lymphoid hyperplasia should be rendered until additional diagnostic testing proves otherwise. In contrast, if a homogeneous population of large lymphoid cells is detected in a diffusely large spleen, suspicion of lymphoid neoplasia increases. Histologic evaluation of tissue architecture, immunohistochemical staining, and PCR techniques to assess clonality within a lymphoid population may be important in reaching a definitive diagnosis.¹⁹ Cytologically distinguishing between reactive and neoplastic mastocytosis in the spleen presents similar challenges, and a high suspicion of splenic mast cell neoplasia depends on observing either large numbers of mast cells in dense aggregates or atypical mast cell morphology.

Cytologic evaluation of an aspirate in 1 animal was consistent with malignant disease when the underlying process was identified as a benign lesion on histologic evaluation. In that instance, multiple criteria of malignancy (eg, severe cytoplasmic, nuclear, and nucleolar pleomorphism) were evident in the connective tissue population on cytologic evaluation, and although the histologic appearance of the lesion had similar pleomorphic features, clear architectural evidence of a tumor was not found. This finding highlighted shortcomings of the small sample size obtained when fine-needle aspiration is used and the need to evaluate tissue architecture in certain instances and underscored the importance of undertaking further histologic sampling when clinical suspicion of malignancy warrants it.¹¹ Reactive populations of cells can acquire pleomorphic features, especially in the presence of severe inflammation, and evaluation of tissue architecture is essential in those instances. Although not included in the design of the present study, an important consideration is the availability of alternative diagnostic tests to help confirm a cytologic or histologic diagnosis or to provide a diagnosis when cytologic and histologic methods are inconclusive even when high-quality specimens were available.

Immunohistochemical and immunocytochemical staining and PCR assays for antigen receptor rearrangement to determine clonality in a population of lymphoid cells are examples of tests that can be added to the diagnostic plan as a result of information obtained from initial cytologic evaluation. It is likely that these tests and newer molecular techniques will be used more frequently in the future.¹⁹ In 5 of the 7 animals in which reactive and neoplastic processes could not be distinguished by cytologic evaluation, the final histopathologic diagnosis was a malignancy, suggesting that follow-up evaluation is warranted when an experienced cytopathologist raises the possibility of neoplastic disease.

The various ultrasonographic appearances of disease were typical of the nonspecific appearances of splenic diseases that have been reported.^{1-4,14,15,20,21} The most common abnormalities identified in the spleens of dogs in the present study were regenerative nodules-reactive lymphoid hyperplasia (n = 7), hematoma (3), lymphoma (3), hemangiosarcoma (2), splenitis (2), mast cell tumor (2), and histiocytic neoplasms (2). If taken as percentages of the 29 dogs involved, the fact that regenerative nodules were most common and were followed in frequency by hematoma, lymphoma, and hemangiosarcoma was similar to findings in splenic specimens that were collected via methods other than splenectomy in an earlier study²² of 1,372 dogs. It has been suggested that hemangiosarcoma is overrepresented¹⁸ in studies based on evaluation of specimens derived surgically or via splenectomy.^23–25 No comment can be made regarding splenic aspirates in cats in the present study because of the small number of cats (n = 3) that were included.

Several ultrasonographic findings in the present study were noteworthy, although they were consistent with previous descriptions of disease. The appearance of gas with the characteristic reverberation artifact associated with clostridial splenitis detected in 1 animal was not a surprising finding,²⁰ nor was the finding that the single targetlike lesion detected in another animal was malignant (metastatic bile duct carcinoma).²⁶ A consistent and diagnostically challenging finding is the ultrasonographically normal spleen that contains mast cell tumor,^4,14 as confirmed at the time of splenectomy and as occurred with 1 animal in the present study. Hyperechoic foci that cast an acoustic shadow ultrasonographically in lesions in 2 animals were suggestive of mineralization; however, no mineralization was found on the histologic section. This may have been a result of incomplete tissue sampling.

The finding of multiple discrete lesions with similar ultrasonographic appearance was significantly associated with malignancy in the present study, and single lesions were more often benign. Although this association may be intuitive, to the authors' knowledge, it has not been reported. In an abstract on lymphoid hyperplastic nodules in dogs,^b 4 of 31 dogs had 2 nodules in this benign condition; in another study,²⁷ benign myelolipomas in the spleens of cheetahs were always multiple, a finding that was contrary to our results. This association between multiple lesions and malignancy does not extend to masses of different ultrasonographic appearances. Discrete nodules with infectious causes were found both singly and in multiples.

As with most retrospective studies, there were sources of bias and other shortcomings in this study. No hyperechoic lesions were included, perhaps because clinicians often associate hyperechoic lesions with a benign condition such as benign myelolipoma.^27,28 Other benign conditions that have been reported as having hyperechoic lesions are lymphoid hyperplasia^1,b and fibrosis.²⁰ However, uniformly hyperechoic nodules have also been detected with malignant mast cell tumor in cats^1,4 and malignant histiocytosis in dogs,³ albeit less frequently than those lesions having a hypoechoic appearance. Ill-defined nodules were also underrepresented in the present study. Although multiple poorly defined nodules were aspirated and included in the original list of aspirates, corresponding histologic specimens were obtained for only 2 nodules, the diagnoses for which were hematopoietic neoplasia and mast cell tumor. Additionally, findings may have been skewed because it is possible that sampling for histologic evaluation was performed more frequently in animals that had diseases with a poor prognosis, whereas aspirates of lesions that were suggestive of benign diseases may not have been followed up with biopsy or splenectomy.

In conclusion, ultrasound-guided splenic aspiration with cytologic evaluation yields an accurate diagnosis in most instances. The procedure is technically easy to perform and is typically safe. However, as with any focal sampling technique, there is the possibility that the underlying disease will be missed or that cytologic evaluation will not distinguish between reactive and malignant disease because architectural features cannot be evaluated. Ultrasonographic appearances of the spleen should not be interpreted as indicative of benign versus malignant disease; however, presence of multiple discrete similar-appearing nodules are often associated with malignant disease.