Diagnostic utility of ultrasound-guided fine-needle aspiration and needle-core biopsy sampling of canine splenic masses

Abstract

This prospective clinical study sought to determine the accuracy of cytopathologic examination and needle-core biopsy (NCB) against diagnoses obtained by excisional histopathology (EH) for canine splenic masses. Twenty-five masses were evaluated ex vivo by ultrasound-guided fine-needle aspiration (FNA) and NCB tissue sampling. Each spleen was placed in a container and artificial skin placed over its surface. Ultrasound-guided FNA using a 22-gauge needle and 2 NCB samples [14-gauge (NCB-14), 16-gauge (NCB-16)] were obtained and submitted for analysis. Results were compared to results obtained by splenic excisional histopathology (EH). There was no difference noted between FNA, NCB-14, or NCB-16 analyses. In addition, there was no difference in accuracy between FNA and NCB-14 or between FNA and NCB-14 versus NCB-16. Reported accuracy of FNA was 0.72, NCB-14 was 0.72, and NCB-16 was 0.64, respectively. Both FNA and NCB-14 displayed a sensitivity of 71% and NCB-16 a sensitivity of 53%. Both FNA and NCB-14 displayed a specificity of 75% and NCB-16 a specificity of 88%. The results demonstrated that NCB had no advantage clinically over FNA at diagnosing splenic pathology. This study further demonstrates that preoperative diagnostic evaluation of the spleen is not highly accurate and cannot be recommended prior to splenectomy.

Introduction

Canine splenic masses present a high risk for spontaneous rupture and metastasis. The varied survival time for dogs after splenic mass removal is determined based on the diagnosis obtained. Unfortunately, the diagnosis is often not obtained until surgical excision of the splenic mass is performed. As a histopathologic diagnosis is most often not obtained until after the splenectomy (1), pet owners are faced with a challenging decision on whether to pursue surgical intervention in a dog that may have a metastatic neoplasm with a poor prognosis before a definitive diagnosis is obtained. Although benign lesions such as a splenic hemangioma or hematoma are cured by splenectomy, the median survival time associated with splenic hemangiosarcoma (HSA) with surgical excision alone has been shown to be as low as 1.6 mo (2). Although HSA is the most often encountered malignant tumor of the spleen, incidence reporting varies greatly between studies, ranging from 17 to 76%, depending on the population under investigation (3–7).

Several studies have evaluated for pre-operative risk factors of malignancy, including the presence of hemoperitoneum, mass rupture, anemia, thrombocytopenia, splenic mass-to-splenic volume ratio, and the use of contrast-enhanced computed tomography (3,5,8–10). Currently, ultrasound-guided fine-needle aspiration (FNA) of the spleen represents one of the most frequently performed diagnostic tests to rule-in malignancy, despite demonstrating only 61.3% agreement between cytopathologic and histopathologic diagnoses (11). Splenic needle-core biopsy (NCB) is an established technique with complication rates from 0 to 12.5% (12–14). An initial study investigating the addition of NCB during sampling of the spleen revealed an enhanced ability to distinguish neoplastic from non-neoplastic lesions as well as aid in subclassification (14); however, continued exploration into this combined approach has remained limited, despite validated safety and efficacy of NCB in both humans and veterinary patients (13,14).

The purpose of this study was to determine whether obtaining a splenic mass FNA and/or NCB would potentially provide dog owners with the information necessary to make challenging decisions pre-operatively. Another purpose was to compare the diagnostic accuracy, sensitivity, and specificity of both FNA and 2 needle gauge NCB sampling techniques (14-gauge and 16-gauge) for splenic masses with the diagnoses obtained through the gold standard diagnostic method of excisional histopathology (EH). We hypothesized that NCB histopathologic diagnosis, regardless of needle gauge size, would be a more accurate method for diagnosis that of FNA. In addition, we hypothesized that NCB histopathologic diagnosis would have excellent sensitivity and specificity, whereas FNA cytopathologic diagnosis would have poor sensitivity and high specificity for accurate diagnosis of splenic masses in dogs.

Materials and methods

Sample selection

Dogs presenting to the Boren Veterinary Medical Teaching Hospital and diagnosed with a splenic mass were identified. After splenectomy was performed, the spleen was placed into a container and submersed in water for FNA and NCB sampling prior to then being submersed in formalin and submitted for EH. Since ex-vivo samples were obtained from each spleen, approval from our Institutional Animal Care and Use Committee was not required.

Experimental design

After identifying a case that would undergo splenectomy, a phantom mold to mimic skin was created to simulate ultrasound-assisted FNA and NCB sample collection in a live dog (http://sonopath.com/resources/ultrasound-resources/phantom-mold-recipe-fna-practice). To create the phantom mold, 8 packets of gelatin in 2 cups of cold water were mixed and allowed to stand for 1 min. Four cups of boiling water and 4 tablespoons of sugar-free psyllium fiber supplement (Metamucil) were added. The mixture was whisked until the gelatin was completely dissolved. The mold was placed into the refrigerator for 1 to 2 h and monitored for firmness. When ready, the mold was then placed over top of the open container holding the excised spleen (Figure 1).

Figure 1.

Image demonstrating a spleen placed into a bucket after removal, with artificial skin placed over the surface of the spleen. An ultrasound probe was then placed onto the surface of the artificial skin over the mass lesion to perform the ultrasound-guided diagnostic sampling.

Ultrasonographic evaluation of each splenic mass was performed by a single Diplomate of the American College of Veterinary Internal Medicine (Internal Medicine) using a Sonosite X-Porte ultrasound machine (FUGIFILM Sonosite, Bothell, Washington, USA) and an 8-5 MHz curvilinear electronic transducer (Figures 2,3). Three separate FNA samples were obtained from the periphery and center of the splenic mass lesion using a 22-gauge, 1.5-inch needle and 6-cc syringe using a standard capillary technique (15). Each sample was obtained employing ultrasound guidance from locations within the splenic mass that were identified as having abnormal splenic architecture or heterogenous architecture, rather than areas of hypoechoic cavitation (Figure 2). Slides made from the FNA samples were air-dried and stained with an aqueous Romanovsky stain and submitted for cytopathologic evaluation by a single Diplomate of the American College of Veterinary Pathology. Although a single clinical pathologist read out the cytological samples, if there was question as to what the pathologist was visualizing on the slide, they did have the ability to consult with a colleague, thus mimicking the clinical scenario of sample evaluation. The clinical pathologist reported a diagnosis for each spleen analyzed. Six total NCB samples were subsequently obtained from the splenic lesion using the same method of ultrasound guidance through the phantom skin. Three of the 6 samples were obtained using a 14-gauge (14-gauge × 9 cm, 11-mm penetration) needle-core biopsy instrument (NCB; Monopty Disposable Core Biopsy Instrument; Becton, Dickinson and Company, Franklin Lakes, New Jersey, USA) and the other remaining 3 samples were obtained using a 16-gauge (16-gauge × 9 cm, 11-mm penetration) NCB (Monopty Disposable Core Biopsy Instrument; Becton, Dickinson and Company). Spleens were randomized so that the NCB samples were obtained using a 14-gauge needle first in some lesions and a 16-gauge needle first in others. Needle-core biopsy samples were batched according to the gauge of the biopsy instrument, labeled as NCB-14 (14-gauge) and NCB-16 (16-gauge), and placed in a specimen jar containing 10% formalin. After completion of FNA and NCB sampling, the spleen was placed in 10% formalin and submitted for EH per protocol for a clinical case. Formalin-fixed NCB and whole tissue EH submissions underwent routine automated histologic processing, were paraffin-embedded, sectioned at 5 μm, and stained with hematoxylin and eosin (H&E).

Figure 2.

Image depicting ultrasound-guided fine needle aspiration of the splenic lesion, with the needle being placed through the artificial skin prior to entry into the spleen.

Histopathological evaluation of the samples (both NCB and EH) were performed by Board-certified pathologists (ACVP) assigned to the surgical pathology service at the Oklahoma Animal Disease and Diagnostic Laboratory. Evaluation of NCB versus EH were conducted by different pathologists to eliminate any bias in the final diagnosis for each sample. Any difficult interpretations were handled by consensus review of the group. All diagnoses were reviewed for accuracy and discrepancies by the study pathologist (JWR). If sample quality was considered inadequate or incomplete, the diagnosis was recorded as non-diagnostic.

Clinicopathologic and histopathologic assessments were completed independently of one another. The diagnoses expressed between samples were reported descriptively and grouped into one of 4 categories: normal splenic tissue, non-diagnostic, non-neoplastic, or neoplastic. Non-neoplastic diagnoses were grouped to include lymphoid hyperplasia, myeloid metaplasia, hemosiderosis, infarction, fibrosis, hematopoiesis, and hematoma formation. Positive inclusion was made in cases in which diagnostic modifiers such as “probable” or “suggestive of” were made.

Statistical analysis

The EH diagnosis obtained for the splenic lesion was considered a reference standard against the accuracy indices formulated between FNA, NCB-14, and NCB-16 sample acquisition techniques. Accuracy of a method was defined as the proportion of times the method resulted in the same diagnosis as the diagnosis obtained by EH. A power analysis was not performed prior to conducting the study. A 2-sample test of equality of proportions was used for comparing the methods with each other. Statistical comparison of each sample group included sensitivity, specificity, and positive and negative predictive values. To evaluate the statistical significance between sample groups in identification of neoplasia, all samples were included. If a sample was deemed normal, non-diagnostic, or displayed signs consistent with inflammation without evidence of neoplastic changes, the category was defined as “non-neoplastic.” In cases that displayed conflicting diagnoses between cytopathologic and histopathologic samples, positive and negative predictive values were obtained using the same predefined category grouping. All results were evaluated by a statistician and significance was set at P < 0.05.

Results

Spleens from 25 dogs that were diagnosed with a splenic mass and underwent a splenectomy were included in the study. Using EH as the reference standard, 68% (n = 17/25) of the splenic masses were diagnosed as neoplastic in origin, and 32% (n = 8/25) were diagnosed as non-neoplastic. Of the 17 spleens that were diagnosed with neoplastic masses, 12% (n = 2/17) were non-angiogenic and 88% (n = 15/17) were diagnosed as HSA (Table I). Of the 8 non-neoplastic diagnoses, multiple lesion types were reported, including lymphoid hyperplasia, myeloid metaplasia, extramedullary hematopoiesis, and infarction (Table II).

Table I.

Results obtained from each method of diagnosis, as compared to excisional histopathology in 25 spleens.

Diagnostic method		Diagnostic result
EH	Neoplasia 68% (n = 17) HSA 88% (n = 15)	Non-angiogenic 11% (n = 2)	Non-neoplastic 32% (n = 8)	Non-diagnostic NA
FNA	Neoplasia (n = 13) 52% (n = 13)		Non-neoplastic 48% (n = 12)	Non-diagnostic NA
NCB-14	Neoplasia 52% (n = 13) HSA 84.6% (n = 11)	Non-angiogenic 15.4% (n = 2)	Non-neoplastic 44% (n = 11)	Non-diagnostic 4% (n = 1)
NCB-16	Neoplasia 36% (n = 9) HSA 77.8% (n = 7)	Non-angiogenic 22.2% (n = 2)	Non-neoplastic 56% (n = 14)	Non-diagnostic 8% (n = 2)

EH — Excisional histopathology; FNA — Fine-needle aspiration; NCB-14 — Needle-core biopsy, 14-gauge; NCB-16 — Needle-core biopsy, 16-gauge; HSA — Hemangiosarcoma.

Table II.

Cytopathologic and histopathologic diagnoses for ex-vivo samples obtained from 25 canine splenic masses.

Sample	EH	FNA	NCB-14	NCB-16
1	Sarcoma	EMH	Sarcoma	Sarcoma
2	HSA	EMH, lymphoid hyperplasia	HSA	HSA
3	HSA	HSA	HSA	Non-diagnostic
4	HSA	HSA	HSA	Hematoma
5	Hematoma	Normal	Hematoma	Hematoma
6	HSA	HSA	HSA	HSA
7	HSA	EMH	HSA	HSA
8	EMH, lymphoid hyperplasia	EMH	Lymphoid hyperplasia, hemosiderosis, myeloid metaplasia	Lymphoid hyperplasia, hemosiderosis, myeloid metaplasia
9	HSA	HSA	HSA	HSA
10	HSA	EMH	HSA	Myeloid metaplasia, hemosiderosis
11	EMH, lymphoid hyperplasia, hemosiderosis	EMH	Myeloid metaplasia, hemosiderosis	Myeloid metaplasia, hemosiderosis
12	HSA	HSA	Hematoma	Hematoma
13	HSA	HSA	HSA	HSA
14	Lymphoid hyperplasia, hematoma, myeloid metaplasia	EMH	Lymphoid hyperplasia, hemosiderosis, myeloid metaplasia	Lymphoid hyperplasia, hemosiderosis, myeloid metaplasia
15	HSA	HSA	Hematoma	Hematoma
16	HSA	HSA	HSA	Necrosis, fibrosis
17	HSA	HSA	No neoplasia	HSA
18	Lymphoid hyperplasia, myeloid metaplasia, hematoma	HSA	Hematoma	Hematoma
19	Hematoma, infarction, hemosiderosis, EMH	EMH	Hemosiderosis, hematoma, myeloid metaplasia, infarction	Hemosiderosis, hematoma, myeloid metaplasia, infarction
20	Hemosiderosis, lymphoid hyperplasia, EMH, hematoma	EMH	Non-diagnostic	Non-diagnostic
21	Sarcoma	HSA	Hematoma, infarction, hemosiderosis	Hematoma, infarction, hemosiderosis
22	Hematoma, lymphoid hyperplasia	EMH	HSA	Hematoma, infarction
23	HSA	HSA	Hematoma	Hematoma, infarction
24	HSA	HSA	HSA	HSA
25	Sarcoma	EMH	Fibrosarcoma	Fibrosarcoma

EH — Excisional histopathology; FNA — Fine-needle aspiration; NCB-14 — Needle-core biopsy, 14-gauge; NCB-16 — Needle-core biopsy, 16-gauge; HSA — Hemangiosarcoma; EMH — Extramedullary hematopoiesis.

Fine-needle aspirate cytopathologic evaluation of each spleen yielded a diagnosis of neoplasia in 52% (n = 13/25) of samples and non-neoplasia in 48% (n = 12/25) of samples. Needle-core biopsy 14-gauge samples yielded neoplasia in 52% (n = 13/25) of samples, with 84.6% (n = 11/13) diagnosed as HSA and 15.4% (n = 2/13) diagnosed as non-angiosarcoma in origin. Non-neoplasia was diagnosed in 44% (n = 11/25) of samples, and 4% (n = 1/25) were non-diagnostic (Table I). Needle-core 16-gauge samples yielded a diagnosis of neoplasia in 36% (n = 9/25) of samples, with 77.8% (n = 7/9) as HSA and 22.2% (n = 2/9) diagnosed as non-angiosarcoma in origin. Non-neoplasia was diagnosed in 56% (n = 14/25), and 8% (n = 2/25) were non-diagnostic (Table I). A list comparing each method of diagnosis and the diagnosis obtained for each splenic mass lesion is provided (Table II).

There was no statistically significant difference noted in each method of analysis (FNA, NCB-14, NCB-16), using EH as the gold standard method of diagnosis. In addition, there was no difference in accuracy between FNA and NCB-14 analysis (P = 1.0) or between FNA and NCB-14 versus NCB-16 analysis (P = 0.7618). Reported accuracy of FNA was 0.72, NCB-14 was 0.72, and NCB-16 was 0.64, respectively. Fine-needle aspiration and NCB-14 displayed a sensitivity of 71% each and NCB-16 displayed a sensitivity of 53%. Fine-needle aspiration and NCB-14 both displayed a specificity of 75% and NCB-16 displayed a specificity of 88%. With a sensitivity of 71% for FNA, the positive predictive value (PPV) was 92% and negative predictive value (NPV) was 55%. With a sensitivity of 71% for NCB-14, the PPV was 92% and NPV was 60%. With a sensitivity of 53% for NCB-16, the PPV was 100% and NPV was 50%.

Discussion

Splenic masses are common in older dogs and are often associated with hemorrhage and hemoperitoneum, forcing pet owners to make a quick decision regarding surgical intervention with a splenectomy, palliative care, or humane euthanasia. This study sought to determine the accuracy of two diagnostic techniques, FNA and NCB, to determine if a reliable pre-operative diagnosis could provide pet owners with the guidance they may desire in the decision-making process for their dog if the possibility of an aggressive neoplasm was present. Although this diagnosis may not change the mind of some owners, obtaining additional information can aid in making an informed decision for further treatment.

Fine-needle aspiration with cytopathologic evaluation using a 22-gauge needle can be associated with less risk than an NCB instrument with histopathologic evaluation (11,16). However, cytopathologic evaluation of an organ has not always correlated to histopathologic evaluation due to a highly focal sampling location (11,17) and poor exfoliation of neoplastic cells, especially those of mesenchymal origin. There is limited information in the veterinary literature comparing the diagnostic accuracy, sensitivity, and specificity of FNA cytological samples and NCB histopathology to EH.

The objective of our study was to evaluate the diagnostic accuracy, sensitivity, and specificity of non-invasive diagnostic methods (FNA and NCB) for the diagnosis of canine splenic masses. The results of this study did not show a significant difference in accuracy between the FNA and NCB-14 or FNA/NCB-14 and NCB-16 methods; therefore, our first hypothesis was rejected. It is interesting to note that although the criteron for statistically significant difference was not met, the accuracy of NCB-16 was much less at 64% compared to 72% with FNA or NCB-14. These data confirm the concern that an accuracy of 72% for FNA and NCB-14 or 64% for NCB-16 leaves open the question of the reliability of performing this diagnostic evaluation prior to surgery as an attempt to identify an aggressive neoplasm. In addition, these results will aid clinicians in making an informed decision that allows for weighing the risks and benefits of performing a diagnostic procedure in a dog with a vascular mass where that action may result in further hemorrhage.

Needle-core biopsy and FNA techniques each demonstrated poor sensitivity, with NCB-16 yielding the lowest at 53%. Fine-needle aspiration did not yield a suspected high specificity at 75%, as hypothesized. Thus, our second hypothesis was also rejected. Interestingly, the NCB-16 method yielded the highest specificity of the 3 methods evaluated. With a low sensitivity identified in FNA and both NCB sampling techniques, these methods are likely to have a higher number of false negatives, thus missing more cases of neoplastic disease. This is counter to the goal of performing this diagnostic evaluation, to have an increased chance of diagnosing an aggressive neoplasm that may change the owner’s decision to take their dog to surgery.

All samples submitted for cytopathological evaluation were read independently without knowledge of the anatomic pathologist’s interpretation of the NCB samples, and vice versa. In addition, the pathologists who evaluated the NCB samples did so with the samples anonymized to the EH sample submission. This allowed the removal of any bias in the results from NCB histopathology based on diagnosis obtained from EH. The diagnosis of HSA can be challenging, especially in a small sample volume of the core biopsies. Hemangiosarcoma has distinct histological features that can be masked in small sample volume or hidden in regions of significant necrosis or inflammation. The diagnosis of HSA (or not) in the current study was straight-forward within each sample as confirmed by evaluation from 2 to 4 pathologists. The only challenging sample was Case #1 that was reviewed by at least 5 pathologists; all agreed that the biopsies contained neoplasia, but there was disagreement on the cell of origin. Non-angiomatous origin was confirmed by immunohistochemical staining (CD31) and thus the diagnosis was sarcoma.

An interesting finding to note is that in two separate dogs, the diagnosis obtained via the NCB or FNA technique was HSA, whereas the diagnosis with EH was a benign lesion. Specifically, in dog #18, HSA was diagnosed on the FNA sample and in dog #22, HSA was diagnosed on the CNB-14 sample; both yielded a benign lesion on EH. This discordant finding is likely a reflection of samples for EH being collected from areas of the mass that were non-representative of the lesion, and further highlights the challenging nature of diagnosing splenic HSA, even when using EH as a gold standard method of diagnosis. Splenic masses with HSA can have an uneven distribution of neoplastic cells, with accompanied inflammation, necrosis, fibrosis, or hematoma formation. The pathology literature indicates that some patients that receive benign diagnoses from splenic lesions go on to develop HSA (2) and recent publications have focused on best practices for submission and sampling of splenic lesions (18). Even though EH is considered the gold standard, it cannot be considered 100% reliable. Subsequently, a consideration is that if EH concludes neoplasia is not present after sampling multiple sections, a patient may unnecessarily be euthanized based on a diagnosis of neoplasia using FNA or NCB.

The difference in sensitivity and specificity between the 14-gauge and 16-gauge NCB instrument was unexpected; however, it is not surprising that using a larger needle-core (NCB-14) would result in a more accurate representation of the splenic mass and therefore improved specificity. Even if this is true, this cannot explain why the FNA method using an even smaller needle (22 gauge) yielded the same sensitivity and specificity of the larger bore NCB technique. One explanation for this finding may be that with an ex-vivo sampling technique, there is less blood flow and blood dilution, thereby increasing the sampling success and cellular concentration present within the tissue, allowing a 22-gauge needle to identify a neoplastic lesion more accurately. Although we were able to identify large blood vessels feeding the neoplastic lesion (Figure 4), these could not be used as representation of blood flow. Ideally, an FNA would be the diagnostic modality of choice in a more unstable patient, because NCB sampling takes time. If a patient is hemorrhaging from the mass or is unstable despite resuscitative efforts, unfortunately waiting for this diagnostic result may not always be an option.

Figure 4.

Ultrasound image of a splenic mass demonstrating a large feeding vessel into the mass (arrow).

There are limitations to this study. The current study evaluated FNA and NCB in ex-vivo splenic masses and therefore risk and rate of hemorrhage associated with these sampling procedures was not evaluated. In addition, an ex-vivo sample collection may reduce blood dilution of samples obtained, leading to a false accuracy of needle techniques. Likewise, necrotic areas may be less identifiable without having active blood flow to the mass. However, in a clinical patient, these potential risks would need to be taken into consideration along with the diagnostic information to be obtained by each sampling method. Sampling ex-vivo could have led to bias in location of sampling within the spleen; however, the sampling method employed with the use of artificial skin and ultrasound guidance to decide on the sample site was designed to mimic an in-vivo sampling technique using ultrasound guidance. In addition, we demonstrated that concern could occur for accuracy when diagnosing HSA with EH. Therefore, relying on this method, or any method for that matter, as the gold standard for diagnosis in splenic lesions provides a challenge due to the nature of the tissue being studied.

The results of this ex-vivo study demonstrate that preoperative evaluation of the spleen is not acceptably accurate for making life-altering clinical decisions on a patient and therefore cannot be recommended as an additional preoperative diagnostic technique to obtain a diagnosis of the splenic mass. Excisional histopathology would still be the recommended method for diagnosis of a splenic mass, assuming the patient has undergone appropriate systemic evaluation for metastasis and owners are made aware of surgical complication risks.

Figure 3.

Ultrasound image of a splenic mass demonstrating visualization of heterogenous or abnormal tissue as well as cavitations (arrow). The star depicts the artificial skin overlying the spleen.