Prognostic Indicators and Clinical Outcome in Dogs with Subcutaneous Mast Cell Tumors Treated with Surgery Alone: 43 Cases
ABSTRACT
The purpose of this study was to determine if clinical findings, histologic grade, or other histologic features were associated with clinical outcome in dogs with subcutaneous mast cell tumors (MCTs). Medical records of 43 client-owned dogs were retrospectively reviewed, and follow-up information was gathered via phone or follow-up examination. Progression-free survival (PFS), disease-free interval (DFI), and overall survival were calculated. Forty-two and twenty-two dogs, respectively, had grade 2 (Patnaik grading system) or low-grade tumors (two-tier grading system). Median PFS was 1474 days. Median DFI was not reached at >1968 days. Overall median survival time was not reached at >1968 days. In univariate analysis, argyrophilic nucleolar organizer regions (AgNORs), proliferating cell nuclear antigen, and mitotic index were negatively prognostic for PFS whereas Ki-67, proliferating cell nuclear antigen, and microvessel density were negatively prognostic for DFI. In multivariate analysis, AgNORs remained negatively prognostic for PFS. Results suggest that proliferation indices, especially AgNORs, may be useful in predicting the rare poor outcomes in dogs with subcutaneous MCTs. The vast majority of subcutaneous MCTs appear to be low or intermediate grade with excellent outcomes from good local tumor control.
Introduction
Mast cell tumors (MCTs) represent a common form of cutaneous neoplasia in the dog, accounting for 7–21% of all canine skin tumors.1 Many prognostic factors have been identified for cutaneous MCTs, with histologic grade being the most widely used.2–4 Despite histologic grade being a consistently reliable prognostic factor, variation among pathologists’ interpretation exists.5 Additionally, intermediate-grade (grade 2) cutaneous tumors can exhibit heterogeneous behavior, leading to variable prognoses.5 Markers of cellular differentiation and proliferation have been applied as a potentially more objective means of evaluating canine MCTs. Argyrophilic nucleolar organizer regions (AgNORs), proliferating cell nuclear antigen (PCNA), and Ki-67 have each been shown to be prognostic in canine cutaneous MCTs.6–8 cKIT staining pattern, as well as cKIT mutation status, have also been shown to be prognostic in canine cutaneous MCTs.9–11 Microvessel density (MVD), an indicator of angiogenesis, has also been demonstrated to be prognostic and to correlate to histologic grade in canine MCTs.12 Mitotic index (MI), an evaluation of mitotic activity within a tumor and a component of the Patnaik grading scheme, has been shown to be prognostic in multiple studies as well.3,13–15
Two grading schemes exist for evaluation of cutaneous MCTs. The Patnaik grading scheme represents the most commonly used grading system.3,4 This grading scheme consists of several components, including extent of involvement, cellularity and cellular morphology, MI, and stromal reaction.3 These components have been evaluated individually in several studies. Invasiveness, a component of the Patnaik grading scheme, and depth of the tumor have been examined alone, with varying results.3,13,16 In a study by Preziosi et al., invasiveness (i.e., how deeply through the dermis and subcutaneous tissues a tumor extends) was found to be negatively associated with both survival time and local recurrence.13 A study by Kiupel et al. had differing results, with tumor depth (interpreted as invasiveness) having no prognostic significance in their data set.16 More recently, a study by Kiupel proposing a two-tier grading scheme found high-grade tumors to have at least one of the following: seven or greater mitotic figures in 10 high-power fields (hpfs); at least three multinucleated (three or more nuclei) cells in 10 hpf; at least three bizarre nuclei in 10 hpf; and karyomegaly.3 Tumor depth was not included in their analysis.
Subcutaneous MCTs (tumors arising only in the subcutaneous tissue) may represent a unique clinical subset of canine MCTs. Despite a breadth of knowledge regarding the characteristics, applicability of published grading schemes, and behavior of cutaneous MCTs, these aspects are less clearly defined for subcutaneous MCTs. These tumors may be more invasive and are certainly located deeper than the majority of cutaneous MCTs. As stated above, it is unknown if the Patnaik or two-tier grading scheme can be appropriately applied to this subset of tumors.3,4 Only a handful of reports examine subcutaneous MCTs as a specific entity.17–19 The first published report by Newman et al. found a mean survival time of 1199 days in dogs with subcutaneous MCTs. Proliferation indices (AgNORs, PCNA, and cKIT staining) were not found to be associated with survival.17 A more recent report by Thompson et al. found a similarly prolonged survival time in dogs with subcutaneous MCTs with median survival times not yet reached. Infiltrative growth pattern, presence of multinucleation, and MI were found to be negatively associated with survival.18 In a second paper by Thompson et al., Ki-67, the combination of Ki-67 and AgNORs, cKIT staining pattern, and MI were negatively associated with recurrence and metastasis.19 Thompson et al. also evaluated a small number of dogs with subcutaneous MCTs for KIT, vascular endothelial growth factor, and platelet-derived growth factor phosphorylation status and showed that these may be valuable prognostic variables. Only five dogs with subcutaneous MCTs were included in the data set, however.20 Additional information regarding these tumors, as provided here, may provide valuable information further characterizing subcutaneous MCTs.
The purpose and hypothesis of this study were to determine if Patnaik and two-tier histologic grading scheme3.4 results, other histological results (AgNORs, MI) and immunohistochemical (IHC) results (PCNA, Ki-67, cKIT, and MVD), and clinical findings were inversely associated with progression-free survival (PFS), disease-free interval (DFI), and survival in dogs with subcutaneous MCTs treated with surgery.
Materials and Methods
Criteria for Case Selection
The pathology database at the Animal Medical Center was searched for dogs with a diagnosis of MCT. The word “subcutaneous” was also included in the search fields for histological description and morphological diagnosis. To select cases in which tumors were subcutaneous, the biopsy reports were reviewed for and confirmed to (1) list the tumor as subcutaneous within the diagnosis and/or (2) provide a histologic description of the tumor as located solely within the subcutaneous tissue. The history collected by the veterinarian of record and the gross description of the tumor location were also reviewed. These cases were reviewed by a single pathologist (S.M.). Dogs were included if results of slide review confirmed a subcutaneous MCT. Inclusion criteria also included treatment with surgery alone and the medical record being available for review and follow-up. If dogs were not dead as a result of disease or other causes in a shorter period of time, follow-up of at least 1 yr was required. Exclusion criteria were treatment with adjuvant therapy (chemotherapy or radiation therapy) and/or concurrent disease (i.e., concurrent neoplasia other than MCT).
Procedures
Medical records of eligible cases were reviewed, and follow-up information was gathered by examination at the Animal Medical Center or telephone conversations with referring veterinarians or owners.
Data obtained from the medical record included age at diagnosis, breed, sex, neuter status, the presence of previous MCTs, the development of additional MCTs (distant to the site of the original MCT), the presence of multiple, synchronous MCTs at the time the subcutaneous MCT was diagnosed, tumor recurrence (local/metastatic), location of the subcutaneous MCT, presence of clinical signs, hematologic or biochemical abnormalities (anemia, neutrophilia, thrombocytopenia; any abnormality demonstrated on the biochemistry panel), results of staging tests (lymph node aspirates, thoracic radiographs, abdominal ultrasound, liver/spleen aspirates, bone marrow aspirate), and evaluation of surgical margins. Of the 56 cases identified from the database with slides or tissue blocks available for review, 43 cases were identified that met the inclusion criteria. Eight cases were excluded based on review of the biopsy specimen. These cases were found to involve the dermal layer. Three cases were excluded because the patient received adjuvant therapy (two chemotherapy, one radiation therapy). Two cases were excluded owing to concurrent disease (one splenic hemangiosarcoma, one malignant mammary gland tumors).
PFS was defined as the duration from the date of diagnosis to either the date of local or metastatic recurrence of the original tumor or development of a second primary tumor. DFI was defined as the duration from the date of diagnosis to the date of local or metastatic recurrence of the original tumor. Survival time was defined as the duration from the date of diagnosis to the date of death (attributable to disease or attributable to other causes) or to last known follow-up. For statistical analysis, PFS, DFI, and survival were considered attributable to disease or censored if the patient was alive, if death was attributable to a cause other than mast cell disease, or if the patient was lost to follow-up.
Histologic Examination and Tumor Grading
For histologic analysis, 10% neutral-buffered formalin-fixed sections stained with hematoxylin and eosin were examined. Samples were evaluated and tumor grade was assigned using the Patnaik and two-tier grading schemes.3,4 All samples were evaluated using the Patnaik grading scheme,3 and 23 samples were evaluated using the two-tier grading scheme.4 For the Patnaik grading system, grade 1 was never assigned, as grade 1 tumors are defined as residing only within the dermis.3 To apply the Patnaik grading scheme to subcutaneous MCTs, cellular characteristics were evaluated (pleomorphism, cytoplasmic granulation, etc.); depth of invasion was not. The two-tier grading system4 was developed for cutaneous mast cell tumors; however, it is based solely on cellular characteristics and does not include any consideration of depth of invasion. Therefore, no modifications were deemed necessary for its application to subcutaneous tumors and it was applied as previously described.4 Twenty-three samples were also separately evaluated for invasiveness and multinucleation based on a study by Thompson et al. showing these factors to be prognostic for poor outcome in dogs with subcutaneous MCTs.18 Fewer samples were evaluated using the two-tier grading scheme4 and for invasiveness and multinucleation specifically, as this was performed at a later date and tumor blocks/slides could not be located for all cases. Using the Patnaik grade scheme, grade 1 was never assigned, as grade 1 tumors reside only within dermis.3 Margins were assessed as either complete (lateral and deep margins >1 mm) or incomplete (cells extending to lateral and/or deep margins). A second margin examination was performed in which margins were assessed as complete (lateral and deep margins >5 mm), equivocal (lateral and deep margins 1–5 mm), or incomplete (cells extending to lateral and/or deep margins). Finally, for the 23 cases from which blocks and slides were available for re-evaluation, the deep margin was qualitatively defined as including a fascial plane or adipose tissue only. Slides were reviewed by a single pathologist (S.M.).
Mitotic Index
Mitotic index value was generated by evaluating the region of the tumor sample with the highest overall mitotic activity (field selection method), as previously described.14 A single count of 10 hpfs was obtained for each tumor sample. Mitotic index was defined as the number of mitotic figures per 10 hpf (400×).14
AgNORs and IHC Staining
For AgNORs, histochemical assays, and IHC expression assays, 5 µm thick archival tissue sections fixed in 10% neutral-buffered formalin and embedded in paraffin were mounted on positively charged fluid barrier glass slidesa. Twenty-six cases were stained previously with cKIT, Ki-67, PCNA, and AgNORs as part of a diagnostic panel. For those cases in which a tissue block was unavailable, archived unstained slides were retrieved. Only hematoxylin and eosin–stained slides were available for three cases. These slides were destained prior to IHC staining. All adherent tissue sections were dried in a 60°C slide dryer, except the three cases that required destaining. For cKIT, Ki-67, PCNA, and AgNOR assays, slides were immersed in a combination deparaffinization and epitope retrieval solutionb, heated to 120°C for 3 min in a pressure cookerc, cooled to 25°C, and rinsed. Adherent tissue sections for factor VIII staining were manually dewaxed.
To stain AgNORs, sections were assayed by use of a previously described silver-nitrate-formic acid colloid method,21 with a slight modification. A 2% solution of gelatin-formic acid (100 mg of gelatin in 5 mL of distilled water and 50 µL formic acid) was mixed with a 50% silver nitrate solution (5 g AgNO3 /10 mL distilled water). Four hundred to seven hundred microliters of the resulting solution was pipetted onto each slide to cover the tissue. Slides were incubated for 25 min in a dark humidified chamber. Slides were then rinsed three times in distilled water. Results for AgNORs were expressed as mean number of AgNORs (per nucleus)/100 cells counted.6
Immunostaining was performed at room temperature either manually or on an automated immunostainerd with a phosphate-buffered saline–based wash buffer between each step. Endogenous peroxidase activity was quenched with 3% hydrogen peroxide for 10 min. Sections stained for factor VIII were enzymatically treated with proteinase Ke for 5 min. Sections were incubated with the primary antibody for either 30 min (for MVD, PCNA, and cKIT) or 2 hr (for Ki-67). Detection was performed using a biotin-streptavidin amplified IHC systemf. Detection was performed by use of a modified biotin-streptavidin peroxidase technique. A universal biotinylated secondary antibody was applied for 20 min, followed by a 20 min incubation with streptavidin-horseradish peroxidase. Sections were incubated with 3,3’-diaminobenzidine for 10 min and counterstained with hematoxylin. All slides were manually dehydrated in increasing concentrations of alcohol and cleared in a non-xylene clearing agentg, and a cover slip was added with permanent mounting mediah.
For MVD analysis, polyclonal rabbit factor VIII-RAi was used at a dilution of 1:2000. For PCNA, monoclonal murine antibodyj was used at a dilution of 1:6000. For cKIT, polyclonal rat antibodyk was used at a dilution of 1:500. For Ki-67, monoclonal murine antibodyk was used at a dilution of 1:25.
Scoring of tissue sections was performed by capturing a digital microscopic image of an hpf (×40 objective lens) that was representative of the entire MCT. Digital images were imported into a digital image-processing packagel. A total number of immunopositive nuclei were counted from each image. The total number of immunopositive nuclei was counted for PCNA and Ki-67 markers.8 For AgNORs counts, the total number of AgNORs dots/100 nuclei were counted and an average number of dots per cell was calculated.6 For cKIT, location of expression was determined as described in Kiupel et al.9 Patterns of staining were assessed as follows: KIT pattern I, or predominantly perimembrane staining with minimal cytoplasmic KIT protein staining, KIT pattern II, or focal to stippled cytoplasmic KIT staining, or KIT pattern III, or diffuse KIT cytoplasmic staining. Classification was based on the highest staining pattern in at least 10% (per 100 neoplastic cells) of the neoplastic cell population.9 For MVD, a previously described method from Luong et al. was used.22 For quantification of intratumoral MVD, vascular “hot spots” within the sections were identified at ×100 magnification. Photomicrographs of three nonoverlapping hotspots were taken at ×200 magnification with a digital microscope camera systemm and imported into digital imaging softwaren. Mean MVD was determined for each case and expressed as the number of vessels per square millimeter.22
Positive and negative controls included a canine normal tissue array for PCNA, Ki67, and factor VIII and a previously labeled MCT for cKIT. For negative control sections, the primary antibody was replaced with antibody diluent.
Statistical Analysis
Outcome (PFS, DFI, survival) was assessed using Kaplan-Meier (KM) life table analysis and log-rank analysis. A P value of <.05 was considered significant. Patients were censored if their death was attributable to a cause other than MCT, if they were alive at the end of the study period, or if they were lost to follow-up. Factors evaluated to determine whether they were associated with PFS, DFI, and survival were age, gender, neuter status, breed, location of tumor (trunk, extremity, head), the presence of previous MCTs, the development of additional MCTs (distant to the site of the original MCT), the presence of multiple, synchronous MCTs at the time the subcutaneous MCT was diagnosed, tumor recurrence (local recurrence, distant metastasis), presence of clinical signs, hematologic or biochemical abnormalities (anemia, neutrophilia, thrombocytopenia; any biochemistry abnormality), results of histologic grading (Patnaik grading scheme, two-tier grading scheme),3,4 results of staging tests (lymph node aspirates, thoracic radiographs, abdominal ultrasound, liver/spleen aspirates, bone marrow aspirate), evaluation or surgical margins, MI, AgNOR score, Ki-67 score, PCNA score, cKIT staining pattern, and MVD score. For statistical purposes, surgical margins were evaluated separately for each case as complete (>1 mm) or incomplete (<1 mm) and as complete (>5 mm), equivocal (>1 mm to < 5 mm), or incomplete (<1 mm).
Univariate Cox proportional hazards regression model analyses were performed to evaluate whether variables were associated with PFS, DFI, and overall survival and as a screen for stepwise multivariate analyses. If univariate variables with a P <.2 were identified, they were placed into the stepwise multivariate model. Simple regression analysis was performed to compare continuous variables, with P <.05 considered significant. Computer softwareo was used to perform all statistical analyses.
Results
Following evaluation for exclusion criteria, 43 dogs were included in this study. The study population included 30 females (70%) and 13 males (30%). One dog (2%) was a sexually intact male, 12 (28%) were neutered males, 3 (7%) were sexually intact females, and 27 (63%) were spayed females. Age of dogs at treatment time ranged from 3.4 to 12 yr (mean 8.1 yr; median, 8.3 yr). The most common breeds were mixed-breed dogs (n = 8, 19%), golden retrievers (n = 5, 12%), Labrador retrievers (n = 4, 9%), and pugs (n = 4, 9%). Nineteen other breeds were represented with <3 dogs each.
Lesions were located on the extremity (n = 20), trunk (n = 14), and head (n = 9). Nine dogs (21%) had multiple MCTs at the time of diagnosis. Eight dogs (19%) had previous MCTs and 10 dogs (23%) developed additional MCTs. Four of eight dogs (50%) that had previous MCTs went on to develop additional MCTs following the diagnosis of a subcutaneous MCT. No dogs had clinical signs (i.e., vomiting, diarrhea, anorexia) associated with the presence of a subcutaneous MCT.
Information regarding surgical margins was available for all dogs. Twenty-nine dogs (67%) had surgical margins reported as complete and 14 dogs (33%) had surgical margins reported as incomplete when only complete and incomplete designation were used. When re-evaluated with the criteria that a complete margin must be >5 mm, 20 dogs (47%) had surgical margins reported as complete, 9 dogs (21%) had surgical margins reported as equivocal, and 14 dogs (33%) had surgical margins reported as incomplete. When the 23 cases with slides available for evaluation at a later date had deep margin qualification performed, the deep margin was qualified as inclusive of a fascial plane in 13 of the 22 (59%) completely excised tumors, inclusive of adipose tissue in 6 cases (27%), and unable to be determined in 3 cases (14%). Fourteen dogs (32%) had surgical procedures considered to be a scar revision or second surgery at the primary site. Of the scar revision or second surgery procedures, 11 dogs (79%) had surgical margins reported as complete, 2 dogs (14%) had margins reported as equivocal, and 1 dog (7%) had margins reported as incomplete. None of the dogs that had scar revisions or second surgeries had gross disease present at the time of the second surgery. The timing of the scar revision/second procedure was not uniformly reported.
Thirty-four dogs had complete blood count results available for review. One dog was mildly anemic (hematocrit = 33%) and three dogs had a neutrophilia. Thirty-two dogs had serum biochemical results available for review. Five dogs had increased liver enzymes (two dogs with increased alkaline phosphatase, two dogs with increased alanine transaminase, one dog with increased alanine transaminase and aspartate transaminase), two dogs had increased blood urea nitrogen, and one dog had increased cholesterol. None of these findings were clinically significant. Twenty-three dogs had thoracic radiograph results available; these were all interpreted by a board-certified internist who served as a radiologist or a board-certified radiologist as unremarkable. Twenty-nine dogs had abdominal ultrasound results available. No dogs had abnormalities found to be associated with MCT. Ten dogs had ultrasound-guided splenic aspirates performed; six dogs had ultrasound-guided liver aspirates performed. No evidence of mast cell disease was identified in any of the aspirates. Fourteen dogs had a regional lymph node sampled via fine-needle aspirate; lymph node metastasis was diagnosed in one dog. Twenty-six dogs had bone marrow aspirates performed; no evidence of MCT was identified in any dogs.
Using the criteria of the Patnaik grading scheme,3 42 dogs (98%) had grade 2 tumors and one dog had a grade 3 tumor. Using the criteria of the two-tier grading scheme,4 22 dogs (96%) had low-grade tumors and one dog had a high-grade tumor. Using criteria from Thompson et al.,18 15 (65%) dogs had an infiltrative growth pattern, 3 (13%) had a circumscribed growth pattern, and 5 (22%) had a combined growth pattern. Six dogs (26%) had presence of multinucleation whereas 17 (74%) did not have multinucleation. The mean and median AgNORs scores were 1.61 and 1.52, respectively (range 0.95–3.8). The mean and median PCNA scores were 21.75 and 19, respectively (range 4–57). The mean and median Ki-67 scores were 5.4 and 4, respectively (range <1 to 23). The most frequent cKIT staining pattern was KIT pattern 2 (24 dogs, 56%). Thirteen dogs (30%) had tumors with cKIT staining pattern 1, and six dogs (14%) had tumors with cKIT staining pattern 3. The mean and median MVD scores were 51.4 and 48, respectively (range 20.45–105.75). The mean and median MI scores were <1 and 0 (range 0–6; Table 1). Only one dog had an MI >4 as outlined as prognostic for subcutaneous MCTs in Thompson et al.18 This dog (MI = 6) died at 42 days as a result of progression of mast cell disease. One dog had an MI of 4. This dog died at 24 days as a result of gastric dilatation and volvulus.
Three dogs (7%) had recurrence of local disease, with one dog (2%) developing both local recurrence and distant metastasis. All three of these dogs had incomplete surgical margins, and none of them had a scar revision or second surgical procedure. All three dogs had grade 2 tumors using the Patnaik grading scheme.3 Two of the three had tissue available for evaluation using the two-tier grading scheme4; both of these dogs had low-grade tumors using the two-tier grading scheme. Twenty-one percent of dogs (3/14 dogs, reported above) with incomplete surgical margins had recurrence. As reported above, 10 dogs (23%) developed additional MCTs (Table 2).
KM median PFS (local, metastatic recurrence of primary or a second primary tumor) was 1474 days (Figure 1). Proliferation indices values corresponded to PFS. In univariate analysis, higher AgNORs score was significantly associated with a shorter PFS (P = .0005, hazard ratio [HR] = 7.056, confidence interval [CI] = 2.364–21.059). Similarly, higher PCNA score was associated with a shorter PFS (P = .0174, HR = 1.069, CI = 1.012–1.130). Higher MI was also significantly associated with a shorter PFS (P = .0302, HR = 1.852, CI = 1.061–3.235). Univariate analysis results are summarized in Table 3. In Cox stepwise multivariate analysis, AgNORs remained significant (P = .0160, HR = 7.854, CI = 1.467–42.044).
Citation: Journal of the American Animal Hospital Association 56, 4; 10.5326/JAAHA-MS-6960
KM median DFI (local, metastatic recurrence of primary) was not yet reached at >1968 days (Figure 2). In univariate analysis, higher PCNA score was associated with a shorter DFI (P = .0081, HR 1.145, CI = 1.036–1.265). Higher Ki-67 and MVD scores were also associated with a shorter DFI (P = .0364, HR = 1.175, CI = 1.010–1.367 and P = .0155, HR = 1.069, CI = 1.013–1.128, respectively). Mitotic index was not significantly associated with DFI; however, it approached significance (P = .0522). Analysis of AgNOR scores could not be performed owing to a statistical estimation problem with the software (all patients censored). Univariate analysis results are summarized in Table 4. No variables were significant in Cox stepwise multivariate analysis.
Citation: Journal of the American Animal Hospital Association 56, 4; 10.5326/JAAHA-MS-6960
KM overall median survival time was not yet reached at >1968 days as 26 dogs (60%) were alive at the end of the study (Figure 3). Overall 1 yr survival rate was 95%, 2 yr survival rate was 95%, and 3 yr survival rate was 86%. In univariate analysis, higher PCNA score was significantly associated with a shorter survival time (P = .0209, HR = 1.106, CI = 1.015–1.205). Local or metastatic recurrence of the primary tumor was significantly associated with an abbreviated survival (P = .0078). MI approached significance with P value of .0578 (HR = 3.612, CI = 0.958–13.598). Analysis of AgNOR scores could not be performed because of a statistical estimation problem with the software (all patients censored). Univariate analysis results are summarized in Table 5. No variables were found to be significant in Cox stepwise multivariate analysis.
Citation: Journal of the American Animal Hospital Association 56, 4; 10.5326/JAAHA-MS-6960
Simple regression analysis demonstrated associations between AgNOR score and PCNA (P = .0077, R2 = 0.173), AgNOR score and Ki-67 (P = .0012, R2 = 0.240), AgNOR and MI (P = .002, R2 = 0.299), PCNA and Ki67 (P = .0122, R2 = 0.154), PCNA and MVD (P = .0016, R2 = 0.243), and Ki67 and MI (P = .0088, R2 = 0.163). Results are summarized in Table 6.
Of the 11 (25%) dogs that died or were euthanized, 3 (27%) died of causes associated with the MCT. Seven dogs (63%) died or were euthanized as a result of other causes, including gastric dilatation volvulus, unspecified, splenic hemangiosarcoma with hemoabdomen, and central nervous system disease. Dogs that died as a result of MCT or other disease did not have necropsies. The three dogs that were evaluated as having died as a result of MCT were presented for euthanasia because of recorded recurrence/progression (local, systemic) of MCT, and the dogs that were recorded as having died as a result of other causes had no documented recurrence/progression of MCT (local, systemic) and were presented for euthanasia owing to causes delineated as other than MCT. Five dogs were lost to follow-up at 409, 418, 432, 451, and 1968 days. Overall median follow-up was 634 days (range 24–1968 days).
Discussion
The goal of this study was to determine if histologic grade, other histological characteristics (MI, AgNOR), IHC results (PCNA, Ki67, cKIT and MVD), and clinical findings would predict the behavior of canine subcutaneous MCTs. Canine MCTs comprise a very heterogeneous population of tumors. By far the majority of MCTs arise in the dermis; however, it has long been recognized that a subset of these tumors arise solely within the subcutaneous tissue without invasion or involvement of the dermal layer.17–19 The Patnaik grading scheme is widely recognized as a common histopathologic grading scheme for canine MCTs, although a specific entity of subcutaneous tumors is not addressed in the description of this grading scheme.3 Because of limitations associated with this grading scheme, a two-tier system was proposed,4 but again, subcutaneous tumors are not specifically addressed.
To date, only three studies examine the specific entity of canine subcutaneous MCTs.17–19 In the first study, by Newman et al., it was found that these tumors have intermediate histological grade, with affected dogs having prolonged mean survival times (1199 days).17 A second study evaluating canine subcutaneous MCTs by Thompson et al. showed a median survival time not yet reached at >1500 days with 1, 2, and 5 yr survival probabilities of 93, 92, and 86%, respectively. This study showed MI, infiltrative growth (i.e., lack of tumor demarcation), and presence of multinucleation to be negatively associated with survival on multivariate analysis. Additionally, MI was negatively associated with local recurrence and metastasis on multivariate analysis.18 Since publication of that study, MI, infiltrative pattern, and presence of multinucleation are frequently reported in lieu of or in addition to grading for patients with subcutaneous MCTs. In the study presented here, we found similar histologic grade results to Newman et al., with the majority (98 and 96%, respectively) of subcutaneous MCTs consistent with grade 2 or low-grade tumors. Additionally, survival times were long, similar to both Newman et al. and Thompson et al., with the median KM survival time not yet reached (>1968 days) with median follow-up time of 634 days. Local recurrence rate reported here (7%) was similar to that reported in Newman et al. (9%)17 and Thompson et al. (8%).18 Metastatic rate was similarly low in this study (2%), as well as in Newman et al.17 (6%) and Thompson et al. (4%).18 Recurrence rate in dogs with incomplete surgical margins was 21%, which is higher than that reported in Newman et al. (8%)17 and Thompson et al. (12%)18 but similar to what is reported in a study of incompletely excised grade 2 cutaneous MCTs (23%).23 Recurrence (local/distant metastasis) was found to be strongly predictive of survival. No dogs that had scar revision/second procedure had local recurrence, possibly advocating for strong recommendation of revision surgery in patients with incompletely excised subcutaneous MCTs.
Histologic grade (Patnaik or two-tier)3,4 was not associated with a shorter PFS, DFI, or survival. These results may indicate that subcutaneous MCTs as an entity are typically lower grade. Alternatively, this may be a result of a low number of grade 3 or high-grade tumors. This may also be a result of selection bias because patients that were treated with adjuvant therapy (i.e., chemotherapy, radiation therapy) were excluded from the study. It would seem likely that dogs with grade 3 tumors underwent adjuvant therapy based on a high likelihood of metastasis. In our study, though, only three dogs were excluded owing to local or systemic adjuvant therapy, and all three of these dogs had grade 2 tumors. No dogs with grade 3 or high-grade tumors were excluded as a result of receiving adjuvant therapy.
An additional difficulty in regard to grade is the use of grading schemes (Patnaik, two-tier)3,4 developed for cutaneous MCTs. Slight modifications were made to the Patnaik grading scheme (exclusion of grade 1 and exclusion of evaluation of depth of invasion). The two-tier grading scheme was applied as described. It is unclear if it is truly valid to apply these grading schemes to subcutaneous MCTs, although we did show that grade as assigned by either cutaneous grading scheme was not a prognostic factor for subcutaneous MCTs. Thompson et al.18 describes MI, infiltrative growth pattern, and presence of multinucleation as negative prognostic indicators for subcutaneous MCTs separate from the application of cutaneous grading schemes. Infiltrative growth pattern and multinucleation were not found to be prognostic in the study presented here; however, a smaller number of cases (23 rather than 43) were evaluated using these criteria.
Proliferation indices are frequently used as prognostic indicators in human oncology, and many of these indices are becoming more commonplace in veterinary medicine. In veterinary medicine, proliferation indices have been researched most in the field of canine MCTs. Of these, AgNORs, PCNA, Ki-67, cKIT, MI, and, more recently, MVD have each been shown to be prognostic in dogs with cutaneous MCTs.6–12 In this study, AgNORs, PCNA, and MI were found to be prognostic for PFS (defined as either recurrence/metastasis of the original tumor or development of a de novo tumor). Only AgNORs remained significant in multivariate analysis. Unfortunately, AgNORs could not be evaluated in Cox regression analysis for DFI or survival owing to median survival and KM median DFI not yet reached and low number of deaths due to disease. Simple regression analysis, however, showed that AgNORs were associated with PCNA, Ki67, and MI. Bostock et al. first showed the utility of AgNORs in dogs with cutaneous MCTs, demonstrating that dogs with tumors with higher AgNORs counts had higher recurrence rates and shorter survival times.6 Similarly, AgNORs count was found to be prognostic for development of metastasis in dogs with cutaneous MCTs by Simoes et al.7 The previously cited study by Newman et al.17 did not find proliferation markers to be significant (AgNORs, PCNA, Ki-67, cKIT); however, these variables were only evaluated with survival. Thompson et al. examined proliferation markers in dogs with subcutaneous MCTs in a case-control study evaluating 60 dogs with subcutaneous MCTs.19 In that study, increased MI, Ki-67, the combination of Ki-67 and AgNORs, and higher cKIT staining pattern were found to be negatively associated with local recurrence and metastasis on univariate analysis. Proliferating cell nuclear antigen and MVD were not evaluated. Markers were not evaluated with survival and multivariate analysis could not be performed.19 Thompson et al. also evaluated canine subcutaneous MCTs in a study evaluating 306 dogs.18 In that study, increased MI was found to be significantly negatively associated with local recurrence, metastasis, and survival on both univariate and multivariate analysis.18 This differs from our results, in that MI was only found to be significant on univariate analysis for PFS, although it did approach significance for both DFI and survival.
This difference may be a result of a smaller study size in the study reported here; however, it highlights the importance of ideally using a panel of proliferation indices for prognostication. A single test is unlikely to predict every tumor with propensity for more aggressive behavior. Mitotic index evaluates the mitotic phase of cells, whereas other indices evaluate growth fraction and rate.14,19 Interestingly, in Thompson et al.,18 MI >4 was a negative prognostic indicator for subcutaneous MCTs, whereas Romansik et al.14 and Elston et al.15 found MI >5 and >7, respectively, to be negative for cutaneous MCTs. The two-tier system4 uses MI >7 as indicative of a high-grade cutaneous tumor. These differences may indicate that MI cutoffs cannot be uniformly applied to cutaneous and subcutaneous MCTs. Indices that detect cells in the entire growth cycle (AgNORs, Ki-67, PCNA) may be of more prognostic utility. Additionally, uniform methods of evaluation and reporting of MI have not been well established.19
The molecular mechanism of the development of de novo MCTs in dogs is unknown. Although most practitioners consider additional MCTs to be a de novo event, it has been demonstrated in a recent study that de novo tumors may be clonally related to primary tumors.24 The development of additional tumors in this study was not associated with a negative DFI or survival. A total of 21% of dogs in this study had concurrent multiple MCTs, 23% of dogs developed additional MCTs, and 19% of dogs had previous MCTs. This is similar to rates previously reported.23,25 Although the presence of multiple MCTs in patients that stage otherwise negatively has not been negatively associated with survival,26 AgNORs may help predict which dogs are at risk for development of second tumors and/or local/metastatic recurrence. Although no therapies have currently been identified as preventive for the development of canine MCTs, dogs at risk for the development of locally recurrent and/or metastatic disease likely benefit from adjuvant therapy. In patients determined to have subcutaneous MCTs with elevated AgNOR counts, practitioners can counsel owners on options for adjuvant therapy as well as determine a follow-up schedule for monitoring for the development of additional, likely de novo, tumors.
Conclusion
Results of the study confirm that proliferation indices, specifically AgNORs and MI, are useful in predicting the behavior of the subcutaneous subset of canine MCTs. These objective measurements may be more useful than grade in these tumors. The vast majority of these tumors appear to be low grade or intermediate grade, making histologic grade, regardless of grading scheme used, a potentially less useful prognostic factor. Although an MI >7 equates with a high-grade designation in the two-tier system, when this grading scheme was applied to our set of patients, histologic grade was not prognostic. In our data set, no patients had an MI >6. Proliferation indices, rather than grade, may aid in determining which grade 2 or low-grade subcutaneous tumors necessitate adjuvant therapy.
The idea that a panel of indices and not solely MI are useful in prognostication is supported by data in veterinary and human medicine. A recent study by Krick et al. evaluated the use of proliferation indices and cKIT mutation status to evaluate stage in grade 2 MCTs.27 Only AgNORs and recurrence were significantly associated with higher stage (stage 2 versus stage 1). Mitotic index was not significantly associated with increased stage.27 Further numerous studies show the utility of proliferation indices in human medicine, and panels of proliferation indices have long been used as part of a diagnostic workup in human oncology.28 Two examples include an evaluation of proliferation indices and molecular diagnostics for the diagnosis and prediction of metastasis in melanocytic tumors and an evaluation of proliferation indices for prognostication in meningiomas.29,30 The first study showed both Ki-67 and MI to be useful in demonstrating malignancy and lymph node metastasis, respectively, in melanocytic tumors.29 Similarly, a second study demonstrated the utility of phosphohistone-H3 over MI and Ki-67 (MIB-1) in predicting the likelihood of recurrence in human meningioma. Phosphohistone-H3 is an immunohistochemical marker of mitotic activity.30
There were several limitations to our study. Because of its retrospective nature, there was no standardized protocol for follow-up. There was also no standardized protocol for staging tests, and a higher percentage of patients had abdominal ultrasound than had regional lymph node aspirates (67% compared with 33%). Additionally, cause of death was typically not determined via necropsy. Sample size likely negated meaningful statistical analysis for some variables owing to low statistical power. Lastly, cKIT mutational analysis, a commonly accepted prognostic test for cutaneous MCTs,10,11 was not performed. Interestingly, in the report by Thompson et al., no cKIT mutations (presence of internal tandem duplications in exon 11) were found.19 A second report by Thompson et al. evaluated both cKIT mutation status and expression of phosphorylated KIT in cutaneous and subcutaneous MCTs.20 Of the five subcutaneous MCTs in that report, again none expressed cKIT mutations; however, two of the tumors expressed phosphorylated KIT. cKIT mutational status and phosphorylation status needs to be further evaluated in this subset of tumors.
In closing, it is important to recognize this subset of canine MCT. This study demonstrates the importance of using proliferation indices to help determine malignancy of this subset of tumor and to potentially guide follow-up and treatment.
Progression-free survival (PFS) Kaplan-Meier (KM) curve. Results of KM PFS analysis. PFS was defined as recurrence or metastasis of a primary tumor or development of a second mast cell tumor. Median PFS was 1474 days.
Disease-free interval (DFI) Kaplan-Meier (KM) curve. Results of KM DFI analysis. DFI was defined as local or metastatic recurrence of a primary tumor. Media DFI was not yet reached at >1968 days.
Overall survival Kaplan-Meier (KM) curve. Results of KM overall survival analysis. Median overall survival was not yet reached at >1968 days. Patients were censored if their death was attributable to a cause other than MCT, if they were alive at the end of the study period, or if they were lost to follow-up.
Contributor Notes
AgNOR (argyrophilic nucleolar organizer region); CI (confidence interval); DFI (disease-free interval); hpf (high-power field); HR (hazard ratio); IHC (immunohistochemical); KM (Kaplan-Meier); MCT (mast cell tumor); MI (mitotic index); MVD (microvessel density); PCNA (proliferating cell nuclear antigen); PFS (progression-free survival)
