Utility of Spleen and Liver Cytology in Staging of Canine Mast Cell Tumors
ABSTRACT
Abdominal ultrasound with spleen and liver cytology is part of routine staging for canine mast cell tumors (MCTs). However, such tests are associated with increased morbidity and cost. Therefore, the objectives of this study were to determine if spleen cytology was predictive of liver cytology in canine MCTs and if any patient or tumor variables were associated with spleen and/or liver metastasis. Records of dogs with MCTs and cytology of spleen and liver were reviewed. Two hundred five patients were included. Overall, 22 (10.7%) patients had metastasis, with 21 (10.2%) and 13 (6.3%) having spleen and liver metastasis, respectively, and 12 (5.9%) having both. For patients with a positive (or negative) spleen cytology, the odds ratio of having a positive (or negative) liver cytology was 233.49. However, a negative spleen cytology had a higher predictive value (0.99) than a positive cytology (0.54). Finally, the presence of local and systemic signs and tumor size were associated with spleen, liver, and/or spleen or liver metastasis. These results suggest that spleen cytology is predictive of liver cytology in staging of canine MCTs, and increasing tumor size and presence of local or systemic signs are associated with an increased risk of visceral metastasis.
Introduction
Mast cell tumors (MCTs) are a common skin tumor in the dog, representing 16–21% of cutaneous tumors that occur in this species.1,2 MCTs have variable biologic behavior, with some patients experiencing long-term tumor control with wide surgical excision and others dying of aggressive local and/or metastatic disease.3,4 Identifying prognostic factors in patients with MCT is essential for developing an effective diagnostic and treatment plan. Although no single prognostic factor can be used to predict biologic behavior, a number of prognostic factors have been identified, including anatomic location, grade (Patnaik and Kiupel grading systems), and other histologic parameters. A more extensive review of prognostic factors in canine MCTs can be found elsewhere.2,5
Although histologic grade is strongly predictive of outcome in dogs with MCTs, this diagnostic requires a surgical biopsy. As such, this information may not be readily available at the time of presentation, or obtaining a biopsy may not be in the best interest of the patient. Therefore, other prognostic indicators must be relied upon to determine an appropriate diagnostic and treatment plan. Clinical stage has also been shown to be prognostic in dogs with MCTs. Although there is some debate regarding the impact of multiple cutaneous MCTs (World Health Organization stage III) on outcome, it is well documented that dogs with stage 0 or I (without lymph node or distant metastasis) disease can experience long-term survival or cure following surgery.6,7 On the other hand, distant metastasis to the spleen and liver (stage IV) has been associated with a poor prognosis.8,9
Complete staging of canine MCTs entails cytologic assessment of local lymph nodes as well as abdominal ultrasound with cytology of the spleen and liver. Given the poor sensitivity of ultrasound for detecting mast cell metastasis in the spleen and liver, cytology may be pursued, regardless of ultrasonographic appearance, particularly in high-risk tumors.9,10 Historically, complete staging also included buffy coat examination and bone marrow aspiration and cytology, although more recent reports do not support the utility of these diagnostics.11,12
Interpretation of cytology can be hindered by the presence of resident mast cells, an immune cell present in a variety of normal tissues, including the spleen and liver. More recently, studies have suggested that mast cells in the liver can frequently be nonpathologic or associated with pathologic processes other than metastatic mast cell disease, such as hepatic fibrosis.13,14 Furthermore, adequate visualization and aspiration of the liver can be technically challenging because of a number of factors, including patient conformation, presence of gas in the stomach, and/or patient compliance.15 In one study, over one-fourth of liver cytology samples were considered nondiagnostic or poor, leading to a misdiagnosis, demonstrating the challenge of obtaining diagnostic liver cytology.16 In addition, ultrasound-guided fine-needle aspirations (FNAs) of visceral organs is associated with increased cost and morbidity.17 Therefore, the aim of this retrospective study was to determine if spleen cytology was predictive of liver cytology, suggesting that adequate staging could be achieved in the majority of canine patients with MCT with spleen cytology alone. A secondary objective of this study was to determine if any patient or tumor variables were associated with presence of spleen or liver metastasis.
Materials and Methods
Criteria for Selection of Cases
Medical records were retrospectively evaluated for dogs with a diagnosis of cutaneous or subcutaneous MCTs that presented to the Ohio State University Veterinary Medical Center between August 2009 and August 2019. Dogs were included if they had the following: (1) a cytologic or histologic diagnosis of a cutaneous or subcutaneous MCT, (2) ultrasound-guided FNA of the spleen and liver, and (3) cytologic evaluation of spleen and liver aspirates as part of their initial diagnostic workup. Patients receiving previous chemotherapy or radiation therapy were excluded, although steroid use was permitted.
Case Data
Data were collected through review of the complete medical record. Patient variables including age at diagnosis, sex, and breed were recorded. Dogs were identified as having or not having local and systemic signs at the time of their diagnostic workup. Systemic signs were based on the presence or absence of inappetence, vomiting, and/or diarrhea. Local signs were defined as the presence or absence of tumoral/peritumoral edema, ulceration, and/or hemorrhage. Patients in whom the tumor had been previously excised or local signs were deemed to be related to postoperative complications were not evaluated in this category. Treatment with steroids was also recorded and defined as administration of any dose of exogenous corticosteroids in the 2 wk before evaluation.
Tumor variables were also recorded, including date of diagnosis, site and size of the primary tumor, whether the tumor was cutaneous or subcutaneous, presence of one or more MCTs, KIT mutational status (exon 8 and/or 11 internal tandem duplication), and histopathologic grade (Patnaik and Kiupel), when available. Tumor locations were grouped into anatomical regions as follows: limbs, trunk, head/neck, muzzle, urogenital (penis/prepuce/vulva), inguinal, perineal, or multiple tumors. Patients with multiple tumors that had at least one tumor occurring on the muzzle or in the urogenital, inguinal, or perineal region were grouped into that corresponding anatomic location. Patients with multiple tumors in which none of them occurred in the aforementioned regions were grouped under multiple tumors. Tumor size was determined by using the largest recorded diameter. For patients with multiple tumors, the tumor with the largest diameter was recorded. Tumors were classified as cutaneous or subcutaneous based on descriptions in the medical record and/or histopathology reports. Patients that had multiple MCTs occurring in both the cutaneous and subcutaneous tissues were denoted as such. For patients with multiple MCTs, the tumor with the highest histopathologic grade was recorded. The results of regional lymph node cytology or histopathology was also recorded, when available.
Percutaneous ultrasound-guideda,b,c FNAs of the spleen and liver were obtained using a 25-G, 1.5 in. needle, regardless of the ultrasonographic appearance. For patients with focal lesions, samples were taken from both the focal lesion and general organ parenchyma. At least two samples were obtained per site.
Cytology samples from spleen, liver, and lymph node aspirates, when applicable, were stained with Wright-Giemsa and interpreted by a boarded clinical pathologist. Cytology reports for which the report was definitive or stated “was concerning for” were considered positive for metastasis. Other statements such as “cannot definitively rule out” were considered negative for metastasis. For patients with lymph node biopsies, histopathology samples were reviewed by an anatomic pathologist and categorized as described for cytology samples. After preliminary data collection, a single, board-certified clinical pathologist (J.A.H.), blinded to all history, clinical information, and diagnostic results, reviewed cytologic preparations from the spleen and liver aspirates from each case that was positive for spleen and/or liver metastasis. A corresponding number of preparations from patients with negative spleen and liver samples were also randomly selected for review. All spleen cytology samples were reviewed first before review of all liver cytology samples to avoid interpreting liver cytology samples with knowledge of the spleen cytology interpretation. Additionally, cases were reviewed in a randomized order. Definitive cytologic interpretations of positive or negative were assigned to each case using criteria similar to what has been previously described,9,18 further outlined below. A mean number of 2.2 (range: 1–5) and 1.8 (range: 1–5) slides were considered of diagnostic quality and underwent cytologic review for spleen and liver submissions, respectively. Lymph node cytology and histopathology were not retrospectively reviewed.
For spleen aspirates, a cytologic interpretation of “negative” was assigned if no mast cells were seen at all or if low numbers (fewer than ∼5–10) of well-granulated, uniform mast cells were seen within stromal material only and aggregates were not seen within nonstromal regions. A cytologic interpretation of “positive” was assigned if aggregates of three or more mast cells were present in multiple 40× fields of nonstromal regions or if mast cells with atypical features (poor granularity, moderate anisocytosis, or anisokaryosis) were noted in multiple fields.
For liver aspirates, a cytologic interpretation of “negative” was assigned if no mast cells were seen at all or if low numbers (1–2) of well-granulated, uniform, individualized mast cells were seen within clusters of hepatocytes and did not exceed a rate of 1 mast cell per 100 hepatocytes or if rare well-differentiated mast cells (fewer than 1 per 10 40× fields) were seen in the background. A cytologic interpretation of “positive” was assigned if aggregates of three or more mast cells were present within more than one cluster of hepatocytes, if aggregates of three or more mast cells were present in multiple 40× fields within the background (nonhepatocyte clusters), or if mast cells with atypical features (poor granularity, moderate anisocytosis, or anisokaryosis) were noted in multiple fields.
Statistical Analysis
Logistic regressions accounting for repeated measurements were used to model liver cytology outcomes by spleen cytology results. Logistic regression analysis was also used to calculate predicted probabilities. A receiver operating characteristic (ROC) analysis was performed to determine the discriminatory ability of spleen cytology results to predict liver cytology results. A Fisher exact (categorical data) or Mann-Whitney (ordinal and continuous data) test were used to test for an association between patient and tumor variables and positive spleen, liver, and spleen or liver metastasis. Variables examined included age, sex, brachycephalic breed, presence of systemic and local signs, tumor location and size, presence of multiple tumors, histopathologic grade (Patnaik and Kiupel) and c-kit mutational status. The variables age and tumor size were analyzed as continuous variables. For statistical purposes, dogs with Patnaik grade 1 and 2 tumors were grouped together, as were dogs with exon 11 and/or exon 8 KIT internal tandem duplication mutations. In addition, given evidence that tumors arising from the muzzle, inguinal, urogenital, and perineal region may be associated with a more aggressive biologic behavior, these locations were grouped together under “biologically aggressive locations” for statistical analysis. Standard statistical softwared was used for data analysis. P < .05 was considered significant.
Results
Two hundred twenty-eight patients were diagnosed with a cutaneous or subcutaneous MCT and had ultrasound-guided FNAs of the spleen and liver with cytologic evaluation. Of these 228 patients, 23 had nondiagnostic liver (n = 21; 9.2%) or spleen (n = 2; 0.9%) cytology due to poor cellularity. These patients were excluded from analysis.
Fifty breeds were represented with mixed-breed dog (n = 54), Labrador retriever (n = 30), golden retriever (n = 15), boxer (n = 13), Pit bull terrier (n = 10), and pug (n = 8) being the most common. Dogs ranged from 1 to 16 yr of age, with the median age at presentation being 8 yr. One hundred sixteen were spayed females, 76 were castrated males, 7 were intact males, and 6 were intact females. Forty-nine patients had their tumor(s) surgically excised before presentation. Of the patients with gross tumor present, the mean tumor diameter was 3.4 cm (range: 0.3–23.0). Other patient and tumor variables are summarized in Table 1.
A Patnaik and Kiupel histopathologic grade was obtained for 143 and 119 patients, respectively. In the remainder of the cases, a biopsy was not performed or the tumor was not graded because of subcutaneous location. There were 11 Patnaik grade 1, 101 Patnaik grade 2, and 31 Patnaik grade 3 tumors. Of the MCTs that had both a Patnaik and Kiupel grade, all Pantaik grade 1 and 3 tumors were classified as Kiupel low and high grade, respectively. For Patnaik grade 2 tumors that were also assigned a Kiupel grade, 69 were low grade, and 15 were high grade.
One hundred eleven patients had an evaluation of regional lymph node via cytology or histopathology, of whom 69 (62.2%) had evidence of nodal metastasis. Of these patients, 11 had spleen and/or liver metastasis. All patients that had visceral metastasis and lymph node evaluation had nodal metastatic disease.
Overall, of the 205 patients with a diagnostic spleen and liver cytology, 22 patients (10.7%) had cytologic evidence of spleen and/or liver metastasis, with 21 (10.2%) having splenic metastasis, 13 (6.3%) having liver metastasis, and 12 (5.9%) having both. Nine patients had splenic metastasis in the absence of liver metastasis, whereas one patient had liver metastasis in the absence of splenic metastasis. Logistic regression was performed to determine whether the results of spleen cytology could predict results of liver cytology. For patients with a negative spleen cytology, the odds ratio, adjusted for age and sex, was 233.49, indicating that the odds of having a negative liver cytology in patients with a negative spleen cytology was 233.49 times that for patients with a positive spleen cytology (95% confidence interval [CI]: 26.41–2063.57 P < .0001). Similarly, the odds ratio for patients with positive spleen cytology was also 233.49, indicating that the odds of having a positive liver cytology in patients with a positive spleen cytology was also 233.49 times that for patients with a negative spleen cytology (P < .0001). An ROC curve verified the utility of spleen cytology to predict liver cytology with an area under the ROC curve of 0.949, indicating excellent discrimination (0.9–1.0).
Despite an identical odds ratio, a negative spleen cytology had a higher predictive value (0.99; CI: 0.96–1.00) than a positive spleen cytology (0.54; CI: 0.27–0.67). In other words, the probability of a negative liver cytology was 0.99 in patients with a negative spleen cytology. On the other hand, the probability of a positive liver cytology was 0.54 in patients with a positive spleen cytology.
The association of patient and tumor variables with spleen and liver metastasis is shown in Table 2. Of the variables evaluated, the increasing tumor size, presence of local signs, and presence of systemic signs were associated with a positive spleen, liver, and/or spleen or liver cytology.
Discussion
The results of this study demonstrate a strong association between the results of spleen and liver cytology in canine patients with cutaneous and subcutaneous MCTs. More specifically, patients with a positive (or negative) spleen cytology are more than 200 times more likely to have a positive (or negative) liver cytology. In addition, a negative spleen cytology had a high predictive value (0.99), whereas a positive spleen cytology had a predictive value of 0.54. Although there was not a perfect correlation between spleen and liver cytology, only 1 of 22 (4.5%) patients with visceral metastasis had liver metastasis alone, suggesting that spleen cytology alone is sufficient for staging in the majority of canine patients with MCT. A more cautious interpretation of these results would suggest that omission of liver cytology due to nondiagnostic samples or patient-specific factors would be unlikely to affect a patient’s clinical stage, particularly in patients with a negative spleen cytology.
This finding has important clinical applications given the technical challenges of performing ultrasound-guided liver aspirates, particularly for a less experienced ultrasonographer. Barriers to adequate liver visualization and aspiration include certain patient conformations (e.g., large, deep-chested dogs), presence of gas in the stomach, respiratory motion, and/or poor patient compliance, with only the latter being problematic for aspiration of the spleen.15 In this study, 21 of 228 (9.2%) patients had nondiagnostic liver cytology and were excluded from analysis. This number likely underestimates the true number of nondiagnostic samples, because it is routine practice in the authors’ institutions to evaluate cytologic samples for diagnostic quality before submission. Furthermore, this number is lower than previously reported figures, which have shown nondiagnostic samples in 12 and 28% of liver cytology submissions.16,19 In contrast, a recent study evaluating spleen cytology in dogs with splenic nodules or heterogeneous splenic parenchyma encountered only 1 nondiagnostic spleen cytology in 239 cytology reports available for review, in agreement with the low rate found in our study (2 of 228; 0.8%).20 Finally, although ultrasound-guided FNA and biopsy of the liver is associated with a low complication rate, the procedure still carries a risk of localized hemorrhage and discomfort to the patient as well as an increased cost to the owner.21,22
Although this study found a strong association between the results of spleen and liver cytology, it is important to note that a negative spleen cytology had a higher predictive value (0.99) than a positive spleen cytology (0.54), likely, in part, due to the low prevalence of spleen and/or liver metastasis. Clinically, a high negative predictive value is useful because liver aspirate and cytology would not be warranted in a patient with a negative spleen cytology. On the other hand, patients with a positive spleen cytology only have a 54% likelihood of a positive liver cytology. Despite a lower positive predictive value, this discrepancy has minimal clinical impact because a patient with a positive spleen cytology would be considered to have visceral metastasis, regardless of the presence or absence of liver metastasis.
This study also evaluated several patient and tumor characteristics for their association with a positive spleen and/or liver cytology. Increasing tumor size as well as the presence of local or systemic signs were all associated with positive spleen, liver, and/or spleen or liver cytology. Tumor size has previously been recognized as a negative prognostic factor. Several studies have demonstrated that tumor size influences prognosis with tumors >3 cm being associated with nodal metastasis, progression-free interval, and survival time.8,23–25 In this study, increasing tumor size was shown to be associated with the presence of spleen and/or liver metastasis. The association of larger tumor size with visceral metastasis found in this study suggests that the impact of tumor size on outcome cannot solely be explained by challenges in achieving locoregional control.
The presence of local signs, defined as tumoral/peritumoral edema, ulceration, and/or hemorrhage, was also associated with the presence of spleen and spleen or liver metastasis. This finding is supported by a previous study that showed that tumor ulceration was predictive of survival in dogs with MCTs.26 Another study found that the presence of clinical signs, which included both local and gastrointestinal signs, had a negative influence on survival.24 In the current study, it is possible that local signs associated with MCTs were related to increasing tumor size. Although not investigated in any of the aforementioned studies, it is postulated that larger tumors may be more likely to ulcerate. Finally, the presence of systemic signs, defined as inappetance, vomiting, and/or diarrhea, was significantly associated with the presence of spleen, liver, and spleen or liver metastasis. This is in agreement with previous studies that have found the presence of systemic signs to be associated with visceral or systemic mast cell disease.4,27
Interestingly, neither Patnaik grade 3 nor Kiupel high grade were found to be associated with visceral metastasis in this study. Because histopathologic grade is one of the most consistent prognostic factors for canine MCTs, this finding likely represents a type II error given the large number of tumors that were not graded, with 62 (30.2%) and 86 (42.0%) of dogs not having a Patnaik or Kiupel histologic grade, respectively. A histopathologic grade, usually obtained via surgical excision, may not be obtained because of a multitude of reasons, including subcutaneous tumor location, owners declining surgery, or surgery not being recommended because of tumor size or presence of metastatic disease. In this study, dogs whose tumors were not graded were more likely to have visceral metastasis. This finding likely represents a selection bias because surgery is less likely to be recommended in dogs with visceral metastasis.
Overall, the prevalence of distant metastasis in this study was 10.7%, which is higher than previously reported. In a large study by Stefanello et al., ∼5% of dogs with cutaneous MCTs were diagnosed with distant metastases at the time of initial presentation.23 Our higher number is likely a result of selection bias because patients with one or more negative prognostic factors were more likely to have complete staging tests performed. On the other hand, this study included any canine patient with MCT that had ultrasound-guided FNA and cytology of the spleen and liver, regardless of tumor grade, cutaneous or subcutaneous location, or other prognostic factors. Similarly, the rate of lymph node metastasis in this study was higher than previously reported.28 Of the 111 patients in this study that had lymph node aspirate and cytology or histopathology, 69 (62.2%) had nodal metastatic disease. However, this number should be interpreted with caution because lymph node cytology and histopathology were not retrospectively reviewed using previously established criteria, and for this reason, lymph node metastasis was not evaluated as a variable in statistical analysis. However, all patients that had spleen and/or liver metastasis as well as regional lymph node evaluation (n = 11) were documented to have nodal metastasis, in accordance with a previous study that showed that no dogs with MCT had distant metastasis in the absence of nodal metastasis.28 The relatively low percentage of patients undergoing regional lymph node evaluation (54%) likely reflects the limitations of lymph node assessment in the absence of sentinel lymph node mapping in addition to the challenges of assessing lymph nodes in patients with multiple MCTs.
One of the limitations of this study is that it was retrospective in nature. As a result, there was not uniform criteria used to determine which patients were staged, and in many cases, staging may have been driven by owner wishes. For example, 28 patients with only subcutaneous tumors were staged with spleen and liver aspirates, despite being a population at low risk for visceral metastasis.28 Given that the primary objective of this study was to determine if spleen cytology was predictive of liver cytology, irrespective of the presence or absence of metastasis, all patients with a cutaneous or subcutaneous MCT and spleen and liver cytology were included.
To improve consistency in cytologic evaluation, we had a single, board-certified clinical pathologist (J.A.H.) review cytologic preparations of spleen and liver aspirates from each case that was previously designated positive for spleen and/or liver metastasis as well as a randomly sampled, corresponding number of negative cases, using criteria similar to what has been previously described.9,18 Although not all negative cases were reviewed, it is important to note that there was no change to any case that was previously designated as negative for spleen and liver metastasis, suggesting consistency in evaluation of negative samples. Finally, no histopathologic correlation was performed to verify the presence of visceral metastasis.
Another limitation is the small number of patients that had spleen and/or liver MCT metastasis. This was particularly relevant when evaluating patient and tumor variables for association with spleen or liver metastasis, because the limited number of cases with metastasis were further subdivided into smaller groups. In order to minimize this impact, patients were grouped together, when clinically relevant, to increase statistical power. In addition, diagnostic tests were not standardized across patients. As such, some patients lacked certain data such as tumor grade, c-kit mutational status, and/or lymph node cytology, which further limited statistical power in finding associations between patient and tumor variables and spleen and liver cytology.
Finally, the inclusion of patients on steroids at the time of staging may have also influenced our results. In our study, 27 (13%) of the patients were on steroids at the time of staging, although length of steroid administration could generally not be determined from review of medical records. Two of these 27 patients on steroids had visceral metastasis, with both having spleen and liver metastasis. It is possible that steroid use resulted in downstaging of patients; however, 11 of 17 (65%) patients on steroids that also had lymph node evaluation had evidence of nodal metastasis, suggesting that steroid use is unlikely to completely obfuscate the results of staging tests.
Conclusion
In conclusion, this study found that spleen cytology was predictive of liver cytology in dogs with cutaneous or subcutaneous MCTs, because dogs with a positive (or negative) spleen cytology were ∼230 times more likely to have a positive (or negative) liver cytology. However, a negative spleen cytology had a higher predictive value (0.99) than a positive spleen cytology (0.54). Finally, increasing tumor size, presence of local signs, and presence of systemic signs were associated with a positive spleen, liver, and/or spleen or liver cytology. These findings demonstrate that spleen cytology is predictive of liver cytology in dogs with cutaneous or subcutaneous MCTs, suggesting that adequate staging may be achieved with spleen cytology alone.
Contributor Notes
