Outcomes of Dogs with Grade 3 Mast Cell Tumors: 43 Cases (1997–2007)
This study reports the outcomes of dogs with grade 3 mast cell tumors (MCTs). Clinical and histopathological data were available for 43 dogs. Median progression-free survival (PFS) and overall survival (OS) were 133 and 257 days, respectively. Tumor size, lymph node (LN) status, and mitotic index (MI) significantly influenced PFS in univariate analysis. Tumor size and LN status remained significant in the multivariate analysis. Lymph node status, local tumor control, LN treatment, and MI significantly influenced OS in univariate analysis but only LN status remained significant in multivariate analysis. These results confirm that locoregional control improves outcomes in patients with grade 3 MCTs.
Introduction
A grading system described by Patnaik is currently used to classify cutaneous mast cell tumors (MCTs) based on histomorphologic features. Grade 3 tumors are the most poorly differentiated and invasive of the MCTs.1 Grade 3 tumors account for 29%–40% of all canine MCTs and the reported metastatic rate of this subset of tumors ranges from 55% to 96%.1–4 Early reports evaluating patients with grade 3 MCTs treated with surgery alone showed these patients to have a poor long-term prognosis.1,5,6 More recent literature has shown variable improvement in survival times for dogs diagnosed with high-grade MCTs.7,8 One study found that patients with grade 3 tumors treated with prednisone and vinblastine (VBL) had a median survival of 331 days. With only a small number of dogs in the published study treated in the adjuvant setting, it is difficult to compare the reported median survival time to studies where larger populations of dogs were treated with chemotherapy following complete surgical excision.9 Evaluation of a larger number of dogs with grade 3 MCTs treated with prednisone and VBL in the adjuvant setting reported a median survival of 1,374 days, and a 1 yr survival probability of 0.71 was reached for yet another group of dogs with grade 3 tumors treated with prednisone and VBL in the adjuvant setting.10,11 A protocol consisting of VBL, cyclophosphamide, and prednisone (VCP) was recently evaluated and dogs with grade 3 MCTs were reported to have a median survival of 145 days. Again, few of these dogs were treated in the adjuvant setting.12
Several studies have shown a considerable degree of interobserver variation between pathologists when grading MCTs.13 Assessment of tumor proliferation markers such as Ki-67, agyrophilic nucleolar organizer regions (AgNORs), and mitotic index (MI) have all been evaluated in canine MCTs for prognostic significance. MCTs with a Ki-67 index of 23, defined as the number of positive cells/grid area, were significantly associated with an increased incidence of recurrence and tumor-related mortality.14 The MI, defined as the number of mitotic figures/10 high-power fields (HPFs), has also been shown to be a predictor of overall survival (OS). One report showed significantly prolonged survival in dogs with grade 3 MCTs with a MI ≤5 and median survival was not reached. In contrast, median survival time was <2 mo for those with a MI >5.15
Also shown to be associated with outcome in canine MCTs was the location of KIT, a tyrosine kinase receptor important for mast cell proliferation and differentiation.16,17 Tumors that have abnormal cytoplasmic KIT protein localization (staining patterns 2 and 3) have been reported to carry a worse prognosis than tumors with normal perimembrane protein localization (staining pattern 1).18,19
The purpose of this retrospective study was to evaluate the outcomes and prognostic characteristics of a large group of dogs with grade 3 MCTs.
Materials and Methods
Patient Selection and Evaluation
Records from client-owned dogs with grade 3 MCTs diagnosed between 1997 and 2007 were retrospectively evaluated. Criteria for entry included a histologically confirmed diagnosis of a grade 3 MCT. Breed, gender, weight, and age at diagnosis were recorded for each dog. Tumor characteristics evaluated included the size of the tumor, whether the tumor was recurrent or not, tissue location (i.e., cutaneous, subcutaneous, or mucocutaneous), and if there were multiple MCTs. Tumors considered to be subcutaneous included those limited to the subcutaneous space without primary dermal involvement and cutaneous masses that extended into the subcutaneous tissue. Staging tests were not standardized but included complete blood counts (CBCs), serum biochemistry profiles, urinalyses, abdominal ultrasounds, thoracic radiographs, and lymph node (LN) evaluations. Lymph nodes were considered to be positive for metastatic disease based on cytologic analysis if they contained clusters or aggregates of mast cells.20 The stage of disease was determined based on World Health Organization criteria and all dogs with subcutaneous masses were considered to have stage III disease.21
Treatment Evaluation
Information regarding surgical margins of the primary tumor, surgical excision or radiation therapy (RT) of a metastatic LN, and administration of chemotherapy or RT was recorded. Dogs were considered to have adequate local tumor control (ALC) if they had complete surgical margins without histologic evidence of tumor cells or had incomplete surgical margins and underwent definitive RT.
Chemotherapy
The standard chemotherapy protocol used at the Matthew J. Ryan Veterinary Hospital, School of Veterinary Medicine at the University of Pennsylvania consisted of IV VBL, oral cyclophosphamide, and oral prednisone (VCP).12 The VBL was administered as an IV bolus (2 mg/m2) on day 1 of the 21 day protocol and cyclophosphamide was administered (200–250 mg/m2 per os [PO]) over days 8–11 of the protocol. A total of eight cycles were scheduled to be given over a 6 mo period. Prednisone was administered at 1 mg/kg PO q 24 hr tapered after 6 mo, and discontinued. All dogs received a physical examination and CBC weekly during the first cycle and prior to each VBL administration for the remaining seven cycles. Additional CBCs were performed at the discretion of the clinician.
Outcome Assessment
Follow-up information was obtained through examinations and communication with owners and referring veterinarians. Diagnostics performed at recheck assessments included evaluation for tumor recurrence, LN palpation, LN aspiration, cytologic examination of any new masses, and abdominal ultrasound. Progression-free survival (PFS) was defined as the time from diagnosis to local tumor recurrence, progression of measurable disease, metastasis, or new tumor development. OS was defined as the time from diagnosis to death.
Immunohistochemical Staining and Evaluation
Immunohistochemical staining and evaluation for Ki-67 and KIT were performed according to methods previously described.14,18 Immunostaining for Ki-67 was performed with the Benchmark staining platforma and KIT immunostaining was performed using a Dako Autostainerb.
AgNOR Histochemical Staining and Evaluation
AgNOR histochemical staining was performed using a previously described modified silver staining technique.22 To determine the average AgNOR count/cell in each tumor, AgNORs were counted in 100 random mast cells throughout the tumor (at the original magnification ×1,000).
Mitotic Index Evaluation
A single pathologist evaluated tissue sections stained with hematoxylin and eosin via light microscopy and determined the number of mitotic figures. The area of the tumor sample with the highest amount of mitotic activity was evaluated.
Statistical Analysis
Curves for PFS and OS were calculated using the Kaplan-Meier product-limit method. Dogs were censored in the analysis if they were alive at the time of statistical analysis, lost to follow-up, or died of causes unrelated to their tumors. Variables examined for predictors of PFS and OS in the univariate analysis included the size of the tumor, whether it was a locally recurrent tumor at the time of diagnosis, tissue location (i.e., cutaneous, subcutaneous, or mucocutaneous), stage, LN status, ALC, LN treatment, treatment with VCP chemotherapy, MI, Ki-67 and KIT immunohistochemical staining, and AgNOR histochemical staining. Differences in PFS or OS were tested using the log-rank test. The variables determined to be significant in the univariate analysis were then evaluated by multivariate analysis. Multivariate analysis could not be performed for two of the variables (LN treatment and MI) because of low sample size. Statistical significance was set at P<0.05. All data were analyzed using SASc statistical software.
Results
Forty-three dogs met the inclusion criteria. Twenty-five were spayed females, 10 were castrated males, and 8 were intact males. Mean body weight was 28 kg (range, 2.9–50.5 kg) and mean age was 9 yr (range, 2–13 yr). Tumor characteristics and treatment information have been summarized in Table 1. All dogs that did not have ALC had measurable disease. Ten of the 21 dogs with LN metastasis had a histopathological diagnosis and the other 11 dogs with LN metastasis had a cytologic diagnosis.
Chemotherapy
Thirty-two dogs received chemotherapy. Twenty-six of these dogs received VCP chemotherapy and had ALC, thus receiving their chemotherapy in the adjuvant setting. The mean number of cycles of VCP chemotherapy was five (range, two to nine cycles). An additional six dogs received treatment with other chemotherapy: lomustine (n=2), VBL alone (n=3), and a combination of VBL and chlorambucil (n=1). Three of these six dogs received adjuvant chemotherapy and the other three received palliative chemotherapy. Ten of the dogs that failed VCP received additional treatments as part of a rescue protocol. These included lomustine (n=9) and the tyrosine kinase inhibitor imatinib (n=1). Six dogs were treated with prednisone alone, three dogs did not receive any systemic treatment, and treatment was unknown for two dogs.
Immunohistochemistry and Mitotic Index
Twenty of the dogs included in the study had tissue blocks available for assessment of MI, Ki-67 expression, AgNOR frequency, and KIT protein localization. The MI ranged from 0 to 41 and the mean MI was 8. Of the tumors evaluated, 13 (65%) were determined to have a MI of ≤5, and 7 (35%) had a MI >5. Of the tumors evaluated, 16 (80%) had a Ki-67 index >23 cells and 19 (95%) showed cytoplasmic KIT staining (either pattern 2 or 3). The AgNOR counts ranged from 1.4–4.6/cell and the average AgNOR count was 3.1/cell. Ki-67 expression, AgNOR frequency, and KIT staining patterns did not correlate with either PFS (P=0.5, P=0.8, and P=0.9, respectively) or OS (P=0.2, P=0.3, and P=0.6, respectively).
Outcome
Seventeen patients were censored from analysis. Nine of these dogs were still alive, three dogs died of reasons unrelated to their tumor, and five dogs were lost to follow-up. The PFS and OS for the entire population were 133 days and 257 days, respectively (Figure 1). Prognostic factors that were predictors of outcome for the entire population of dogs (n=43) in univariate analysis were described in Table 1. A subgroup analysis of dogs with ALC (n=35) was also performed and the univariate analysis results were presented in Table 2.
Citation: Journal of the American Animal Hospital Association 47, 1; 10.5326/JAAHA-MS-5557
NA, not applicable
NA, not applicable
In the univariate analysis, tumor size, LN status, and MI were significantly associated with PFS in the entire population of dogs. In the multivariate analysis, tumor size, and LN status were significantly associated with PFS (P=0.02 and P=0.04, respectively). In the univariate analysis, LN status, ALC, LN treatment, and MI were significantly associated with OS in the entire population. In the multivariate analysis, only LN status remained prognostic (P=0.04).
For all dogs with ALC, tumor size and MI were significantly associated with PFS in the univariate analysis. For dogs with ALC, LN status, LN treatment, and MI were significantly associated with OS in the univariate analysis (Figures 2 and 3).
Citation: Journal of the American Animal Hospital Association 47, 1; 10.5326/JAAHA-MS-5557
Citation: Journal of the American Animal Hospital Association 47, 1; 10.5326/JAAHA-MS-5557
Tumor Progression
Tumor progression for all dogs was presented in Table 3. Seven of the dogs that developed progressive disease in a regional LN did not have evidence of LN metastasis at the time of diagnosis. Another five dogs that initially had LN metastasis and received treatment of that LN then developed LN metastasis in another regional LN. These 12 dogs had progressive disease either during or after chemotherapy. Specifically, four of the seven dogs with tumors >3 cm developed local recurrence, and two of these four dogs developed both local recurrence and LN metastasis. The other three dogs with tumors >3 cm were lost to follow-up. In comparison, only 2 of the 19 dogs that had small tumors (≤3 cm) and ALC developed local recurrence. Twenty-four out of 27 of the dogs that failed due to local recurrence and/or regional metastasis received adjuvant chemotherapy and four out of five of the dogs that developed distant metastasis received adjuvant chemotherapy.
Discussion
Treatment of the primary tumor in this group of dogs affected outcome. Dogs that had ALC experienced significantly longer survival times versus those that did not, which is not unexpected. Contrary to some of the earlier reports supporting a poor prognosis for dogs with high-grade tumors, these results confirm that definitive local control is worthwhile and can improve outcome. It should be noted that local control was not significant when evaluated with multivariate analysis.
Furthermore, this study showed that dogs with smaller tumors (≤3 cm) experienced longer PFS compared with dogs with larger tumors (>3 cm) using both univariate and multivariate analyses. This difference was also maintained in the subgroup analysis where only the dogs with ALC were evaluated. Despite ALC, over 50% of the dogs with tumors >3 cm developed local recurrence compared with 11% of dogs with tumors ≤3 cm. These findings suggest that dogs with large grade 3 MCTs may require more extensive surgical resection or RT to prevent recurrence compared with smaller grade 3 tumors. These results are consistent with a previous study that showed MCTs >3 cm were associated with a worse prognosis after surgery.23 It is important to mention that exact margin measurements were not noted in many of the histopathology reports and it is possible that smaller tumors were simply amenable to a more aggressive surgery thus resulting in wider and cleaner margins.
LN metastasis in this group of dogs also affected outcome. Dogs that did not have LN metastasis at the time of diagnosis experienced significantly longer PFS and OS than those that did have LN metastasis. This indicates that LN status in dogs with grade 3 MCTs is an important prognostic indicator. Nevertheless, adequate treatment (either with surgery or RT) of the metastatic LN improved survival and the presence of LN metastasis in these patients should not be considered a reason not to treat with curative intent. Not all of the dogs had LN biopsies to confirm metastatic disease; however, the fact that LN metastasis correlated with outcome supports the criteria for a cytologic diagnosis of lymph node metastasis that was presented by Krick, et al. (2009) and used in this study.20
Regional LNs were the most common site of disease progression in this study. This finding may support the use of prophylactic LN irradiation in dogs with high-grade MCTs regardless of cytologic evidence of LN metastasis. This may also partially explain the relatively long remission times reported by Hahn et al. (2004) in dogs with grade 3 tumors. Dogs treated with surgery combined with RT to the surgical site and closest regional LN had a median survival time of 840 days.8 Prophylactic LN irradiation also appeared to improve outcome in a group of dogs with high-risk MCTs that were concurrently treated with prednisone and VBL.10
Perhaps the most intriguing finding of the current study is that 27 of the 43 dogs failed due to local recurrence and/or regional metastasis and only 5 dogs developed distant metastasis. This may contradict the belief that dogs with high-grade MCTs are at high risk for distant metastatic disease and suggests that increased attention should be devoted to improving locoregional tumor control. The low rate of distant metastatic disease may also be a result of the high number of dogs in this study that were treated with chemotherapy.
The fact that so few dogs died due to distant metastatic disease also suggests that even though all dogs were not initially completely staged, few actually had distant metastasis at the time of diagnosis. Consequently, under-staging is not likely to be an issue. Regarding the outcome (i.e., cause of death), it is possible that some of the dogs that failed due to locoregional recurrence might have had subclinical distant metastasis; however, the direct cause of death or euthanasia was secondary to the locoregional disease and the associated clinical side effects. Therefore, these findings suggest that efforts and prophylactic measures to provide better locoregional control should be considered in patients with clean surgical margins and nonmetastatic regional LNs. A meaningful analysis of the impact of chemotherapy on PFS or OS could not be performed due to the fact that so few dogs either received prednisone alone or did not receive any systemic treatment.
This study evaluated cutaneous, subcutaneous, and mucocutaneous MCTs. The Patnaik cutaneous MCT grading system, used by most veterinary pathologists, uses a combination of both cytologic criteria and tumor invasiveness for classification. According to this grading system, grade 1 mast cell tumors are confined to the dermis and contain well-differentiated mast cells. Grade 2 tumors infiltrate or replace the lower dermal and subcutaneous tissue and contain moderately differentiated mast cells. Grade 3 tumors replace the subcutaneous and deep tissue and consist of poorly differentiated mast cells.1 This particular grading system does not include well-differentiated tumors that originate in the subcutaneous tissue or moderately- to poorly-differentiated tumors that are cutaneous and confined to the dermis. In the current study, only eight of the grade 3 tumors were considered to be primarily cutaneous in origin. We found no significant difference in outcome between dogs that had primary cutaneous versus subcutaneous tumors, suggesting that the Patnaik grading system can also be applied to subcutaneous tumors. This also supports the idea that high-grade MCTs, regardless of site of origin, have an aggressive biologic behavior with a high rate of metastasis to regional LNs and a high rate of locoregional recurrence. A recent report evaluating 53 subcutaneous MCTs showed that patients with subcutaneous MCTs experienced extended mean ± standard deviation survival times of 1,199 ± 565 days; however, these subcutaneous tumors were determined to be of intermediate-grade.24
Unfortunately, all samples were not available for evaluation of MI, Ki-67 expression, AgNOR frequency, and KIT protein localization. Available biopsies were from dogs with and without ALC and were representative of the population of dogs evaluated. The MI for the entire population of dogs and the subpopulation of dogs with ALC was significantly associated with both PFS and OS in the univariate analysis. This supports the results of a previous study where patients with grade 3 tumors with a MI >5 had a poor outcome.15 Evaluation of MI is an inexpensive, readily available prognostic tool that should be used for evaluation of patients with grade 3 MCTs.
Staining patterns for Ki-67 and AgNORs were not prognostic for survival; however, the proliferation markers were elevated for most tumors that were evaluated. Furthermore, KIT staining patterns 2 and 3 were observed in all, except one, of the examined tumors. These results support the high-grade nature of tumors included in this study.
Limitations of this study include the nonstandardization of treatment and the relatively small number of dogs included in the study and subgroup analysis.
Conclusion
In conclusion, results of this retrospective study indicate that some dogs diagnosed with grade 3 MCTs can experience extended survival times, especially if they are diagnosed and treated with early stage disease. LN status in dogs with grade 3 MCTs is an important prognostic factor and appropriate treatment of metastatic LNs can significantly improve survival. Our results show that definitive treatment of both the primary tumor and metastatic LNs is warranted for owners interested in pursuing the most effective therapy for their dogs. Furthermore, it is important to note that the majority of dogs failed due to local and/or regional disease despite complete surgical margins, negative LN status at initial staging, and treatment of metastatic LNs. These findings may also support the addition of either RT or a more extensive surgery to both to the primary site and draining LNs in dogs with grade 3 MCTs >3 cm.
Kaplan-Meier curve depicting progression-free survival (PFS) and overall survival (OS) for all dogs (n=43).
Kaplan-Meier curve depicting overall survival (OS) for all dogs with adequate local tumor control (ALC, n=35) with and without lymph node (LN) metastasis at the time of diagnosis.
Kaplan-Meier curve depicting overall survival (OS) for all dogs with adequate local tumor control (ALC, n=35) with and without treatment of a metastatic lymph node (LN) with either surgery or radiation therapy (RT).
Contributor Notes
Carrie Tupper Hume's present affiliation is the Veterinary Specialty Center of Delaware, New Castle, DE.
