Prognostic histopathologic features of canine glial tumors

Abstract

Gliomas are relatively common tumors in aged dogs (especially brachycephalic breeds), and the dog is proving to be useful as a translational model for humans with brain tumors. Hitherto, there is relatively little prognostic data for canine gliomas and none on outcome related to specific histological features. Histologic sections of tumor biopsies from 33 dogs with glioma treated with surgical resection and immunotherapy and 21 whole brains obtained postmortem were reviewed. Tumors were diagnosed as astrocytic, oligodendroglial, or undefined glioma using Comparative Brain Tumor Consortium criteria. Putative features of malignancy were evaluated, namely, mitotic counts, glomeruloid vascularization, and necrosis. For biopsies, dogs with astrocytic tumors lived longer than those with oligodendroglial or undefined tumor types (median survival 743, 205, and 144 days, respectively). Dogs with low-grade gliomas lived longer than those with high-grade gliomas (median survival 734 and 194 days, respectively). Based on analysis of tumor biopsies, low mitotic counts, absence of glomeruloid vascularization, and absence of necrosis correlated with increased survival (median 293, 223, and 220 days, respectively), whereas high mitotic counts, glomeruloid vascularization, and necrosis correlated with poor survival (median 190, 170, and 154 days, respectively). Mitotic count was the only histological feature in biopsy samples that significantly correlated with survival (P < .05). Whole-brain analyses for those same histologic features had similar and more robust correlations, and were statistically significant for all features (P < .05). The small size of biopsy samples may explain differences between biopsy and whole-brain tumor data. These findings will allow more accurate prognosis for gliomas.

Dogs are affected by primary intracranial neoplasms at a prevalence of 2.3%.¹⁴ Gliomas, neoplastic tumors arising from the glial cells of the central nervous system, are the second most commonly reported primary intracranial tumor in the dog.¹⁴ Humans are affected by such neoplasms at similar rates.¹ Prognosis is poor for both dogs and humans diagnosed with glial tumors. In humans, glioblastoma, the most common and most malignant glioma, has a median survival of 14 to 15 months.² Sunol et al reported a median survival time of 66 days for dogs with glioma treated with surgical resection alone.¹⁵ Currently, the dog is proving useful as a translation model for humans with brain tumors.⁴

Hitherto, canine glioma histopathologic diagnosis and prognosis have been largely based on World Health Organization (WHO) criteria developed for human glioma.^6–8 This technique has proven less than ideal given our increasing awareness of differences in this disease between the 2 species. In an attempt to bridge the gap between human and canine glial tumors, the National Cancer Institute–led Comparative Brain Tumor Consortium (CBTC), a team consisting of human and veterinary pathologists, recently devised a new diagnostic scheme for these tumors in dogs.⁵ The new CBTC scheme allows for the diagnosis of glial tumors as either oligodendroglioma, astrocytoma, or undefined glioma depending on cell morphology and allows for them to be graded as either low grade or high grade depending on histopathologic features present within the tumors. While the new diagnostic scheme simplifies the diagnosis of glial tumors in the dog, there is still a dearth of knowledge on prognosis of these tumors.

The current literature pertaining to clinical outcome of canine glioma is scant and incomplete. Previous studies have discussed canine brain tumors and their prognoses, but most fail to delineate tumors by specific type and/or grade.^3,12,15 When reporting median survival times for intracranial brain tumors, some studies group together different types of brain tumors such as meningioma and glioma.^3,12 Another study reports survival time of dogs with glioma following surgical removal but did not report the impact of tumor grade.¹⁵ An important limiting factor is the difficulty in obtaining significant case numbers. The primary motivation behind the current study is to provide additional insight into histological features that relate to prognosis of canine gliomas.

The objective of the present study is to determine how specific histopathologic features of malignancy in canine glioma relate to long-term clinical outcome as defined by survival time of the patient.

Materials and Methods

Selection of Cases

The study was performed using samples obtained from dogs enrolled in brain tumor clinical trials at the University of Minnesota College of Veterinary Medicine. The protocols used for the clinical trials were approved by the University of Minnesota Institutional Animal Care and Use Committee (IACUC) and including signed client consent to participate in the trial. Data from 33 dogs that were previously treated successfully for intracranial glial tumors between 2008 and 2015 were chosen for inclusion into the study. The dogs included in this project were diagnosed with a solitary brain mass by magnetic resonance imaging (MRI) after having clinical signs of central nervous system (CNS) disease. All dogs underwent successful surgical resection of the intra-axial mass for reduction of tumor volume, to reduce compression and intracranial pressure, and to obtain tissue for definitive histopathology and vaccine preparation. An immediate postoperative MRI was performed to assess extent of resection and measure residual tumor volume. After recovery from anesthesia, all dogs were monitored in the intensive care unit for at least 24 hours. Antiepileptic drugs (phenobarbital, levetiracetam, and zonisamide; alone or in combination) were continued after surgery, and the corticosteroid dose was tapered and discontinued within 10 to 14 days. Tumor cell lysates were prepared by culturing single-cell suspensions from minced fresh tumor samples at 37 °C in 5% O₂ as described.⁹ Cultured tumor cells were lysed by multiple freeze-thaw cycles and then irradiated (20 Gy). The first vaccination was given 10 to 14 days after surgery and was repeated every 2 weeks for 6 doses. Each vaccination consisted of intradermal injection of autologous tumor lysate (~500 μg of protein) with a Toll-like receptor agonist as an immune adjuvant.¹⁰ The dogs’ status was assessed by serial physical and neurological examinations and MRI scans that were compared to the immediate postoperative scans. The time to tumor recurrence and death by euthanasia or natural causes were recorded.

Sample Collection and Processing

Biopsy samples were collected at the time of surgical resection, and whole brains were collected postmortem from the same cohort of animals when possible (Table 2). Histopathologic analysis was performed by at least one board-certified veterinary pathologist (MGO’S) at the time of tumor biopsy and following death when whole brain was available. Tumors were initially diagnosed as canine glioma according to WHO criteria,^6–8 and diagnoses were further updated according to the recent 2018 CBTC criteria.⁵ Hematoxylin and eosin (H&E)-stained slides and paraffin-embedded tissue blocks were available for 33 tumor biopsies and for 21 whole brains collected postmortem.

Table 2.

Pathologic data in biopsies and whole-brain samples from 33 dogs with intracranial glial tumors, and relationship to overall survival.

Case Number	Biopsy Diagnosis	Biopsy Grade	Whole Brain Diagnosis	Whole Brain Grade	Glomeruloid Vascularization		Necrosis		Mitotic score (0–4)		Survival (Days)
					Biopsy	Whole	Biopsy	Whole	Biopsy	Whole
1	Oligodendroglioma	High	Oligodendroglioma	High	Yes	Yes	No	Yes	1	3	27E
2	Oligodendroglioma	High	Oligodendroglioma	High	Yes	Yes	No	Yes	3	4	56E
3	Undefined Glioma	High	Undefined glioma	High	Yes	Yes	Yes	Yes	4	4	60E
4	Undefined Glioma	High	—	—	No	—	Yes	—	1	—	91E
5	Oligodendroglioma	High	—	—	No	—	No	—	0	—	108E
6	Undefined Glioma	High	Undefined glioma	High	No	Yes	Yes	Yes	1	3	117E
7	Oligodendroglioma	High	Oligodendroglioma	High	Yes	Yes	Yes	Yes	3	4	132E
8	Undefined Glioma	High	Undefined glioma	High	Yes	Yes	Yes	Yes	3	4	144E
9	Oligodendroglioma	High	—	—	Yes	—	Yes	—	2	—	153 E
10	Oligodendroglioma	High	Oligodendroglioma	High	No	Yes	No	No	1	1	154 E
11	Oligodendroglioma	High	—	—	Yes	—	Yes	—	3	—	155 E
12	Undefined Glioma	Low	Undefined glioma	High	No	Yes	No	Yes	2	3	162 E
13	Oligodendroglioma	High	Oligodendroglioma	High	Yes	Yes	No	Yes	3	3	185 E
14	Undefined Glioma	High	Undefined glioma	High	No	Yes	No	Yes	0	2	194 E
15	Oligodendroglioma	High	Oligodendroglioma	High	Yes	Yes	No	No	3	3	196 E
16	Oligodendroglioma	High	Oligodendroglioma	High	No	Yes	No	No	1	1	205 E
17	Oligodendroglioma	High	Oligodendroglioma	High	Yes	Yes	Yes	Yes	3	3	212 E
18	Astrocytoma	High	Astrocytoma	High	No	Yes	No	Yes	2	2	214 E
19	Oligodendroglioma	High	—	—	No	—	No	—	0	—	220 E
20	Oligodendroglioma	High	—	—	No	—	No	—	2	—	223 E
21	Oligodendroglioma	High	—	—	Yes	—	Yes	—	3	—	225 E
22	Oligodendroglioma	High	Oligodendroglioma	High	No	No	No	No	1	2	293 E
23	Oligodendroglioma	High	—	—	Yes	—	No	—	2	—	340 E
24	Oligodendroglioma	High	—	—	No	—	No	—	0	—	357 E
25	Oligodendroglioma	Low	Oligodendroglioma	High	No	Yes	No	Yes	0	3	369 E
26	Undefined Glioma	High	Undefined glioma	High	Yes	No	Yes	Yes	2	2	534 E
27	Oligodendroglioma	High	—	—	Yes	—	Yes	—	3	—	2165
28	Astrocytoma	Low	Astrocytoma	Low	No	No	No	No	0	0	522 E
29	Astrocytoma	Low	Astrocytoma	Low	No	No	No	No	0	0	726N
30	Astrocytoma	Low	—	—	No	—	Yes	—	1	—	743 E
31	Astrocytoma	Low	Astrocytoma	Low	No	No	No	No	1	2	801 E
32	Astrocytoma	Low	Astrocytoma	Low	No	No	No	No	0	1	853 E
33	Astrocytoma	Low	—	—	No	—	No	—	1	—	1720N

Abbreviations: E, euthanized; N, natural death.

Tumors were diagnosed as either oligodendroglioma, astrocytoma, or undefined glioma using the updated simplified CBTC classification.⁵ Tumors were also graded as either high or low grade based on the histopathologic features of malignancy, namely, the number of mitotic figures, glomeruloid vascularization, and necrosis. Herein we use the term “glomeruloid vascularization” synonymously with the term “microvascular vascularization” used in the consensus CBTC manuscript.⁵ Suspected features of malignancy, specifically glomeruloid vascularization and necrosis, were recorded as either present or absent in each tumor sample. Another suspected feature of malignancy, the number of mitotic figures, was subjectively and semiquantitatively assessed and scored on a 0 to 4 scale for the whole section as follows: 0 = none, 1 = rare, 2 = low numbers, 3 = moderate numbers, and 4 = high numbers. This approximates to a mitotic figure count per high power field (HPF) as follows: 0 = 0 mitotic figures in >10 HPFs (up to 30 or more HPFs); 1 = 1 mitotic figure in >10 HPFs (11 to 30 or more HPFs); 2 = 1 to 3 mitotic figure in 10 HPFs; 3 = 4 to 10 mitotic figure in 10 HPFs; 4 = >10 mitotic figure in 10 HPFs. Mitotic figures were counted in nonconsecutive fields using a 40× objective lens (diameter of field of view was 0.625 mm).

Kaplan-Meier analyses were performed on the data to produce survival curves, and groups were compared using the log-rank test; P < .05 was considered statistically significant.

Results

The study included 33 cases consisting of both high (n = 27) and low (n = 6) grade tumors as well as 3 tumor types—oligodendroglioma (n = 19), astrocytoma (n = 7), and undefined glioma (n = 7; Table 1). All dogs recovered from anesthesia and were discharged from the hospital within a day or two after surgery. There were negligible side effects of the immunotherapy, and dogs’ quality of life was deemed good by their owners. Seizures were generally well-controlled by antiepileptic drugs until tumor recurrence. Three dogs were euthanized or died due to advanced age (>15 years of age), and 1 dog was euthanized due to cardiac failure at 12.5 years of age without evidence of tumor recurrence. Two dogs with high-grade glioma were euthanized due to uncontrollable seizures and had tumor recurrence. One dog diagnosed with low-grade glioma at 9 months of age was euthanized 2 years after surgery due to poor seizure control. The other 26 dogs died or were euthanized because of tumor recurrence. Clinically, the only factor that affected survival time was tumor grade; tumor location, breed, age, and sex did not affect outcome (unpublished data).

Table 1.

Signalment, final diagnosis, and overall survival in 33 dogs with intracranial glial tumors.

Number	Breed	BW at Surgery (kg)	Age at Surgery (years)	Sex	Diagnosisa	Biopsy Grade	Whole Tumor Grade	Overall Survival (Days Postsurgery)
1	Boston Terrier	8	11	M/C	Oligodendroglioma	High	High	27
2	Shih Tzu	8	11	F/S	Oligodendroglioma	High	High	56
3	Staffordshire Terrier	26	1	FI	Undefined glioma	High	High	60
4	Labrador Retriever	39	7	M/C	Undefined glioma	High	N/A	91
5	Bull Mastiff	43	7	F/S	Oligodendroglioma	High	N/A	108
6	Boxer	40	6	M/C	Undefined glioma	High	High	117
7	Boston Terrier	11	6	F/S	Oligodendroglioma	High	High	132
8	Boxer	29	6	M/C	Undefined glioma	High	High	144
9	Boxer	25	7	M/C	Oligodendroglioma	High	N/A	153
10	Boxer	18	9	F/S	Oligodendroglioma	High	High	154
11	Boston Terrier	15	10	M/C	Oligodendroglioma	High	N/A	155
12	Airedale	31	8	M/C	Undefined glioma	Low	High	162
13	Boxer	40	7	M/C	Oligodendroglioma	High	High	185
14	Staffordshire Terrier	18	9	M/C	Undefined glioma	High	High	194
15	American Bulldog	36	5	M/C	Oligodendroglioma	High	High	196
16	Staffordshire Terrier	20	6	MI	Oligodendroglioma	High	High	205
17	Staffordshire Terrier	50	7	M/C	Oligodendroglioma	High	High	212
18	Wheaten Terrier	20	10	F/S	Astrocytoma	High	High	214
19	Dogue de Bordeaux	75	4	M/C	Oligodendroglioma	High	N/A	220
20	Boxer	39	5	M/C	Oligodendroglioma	High	N/A	223
21	English Bulldog	24	10	F/S	Oligodendroglioma	High	N/A	225
22	Boston Terrier	11	5	F/S	Oligodendroglioma	High	High	293
23	Boxer	29	6	F/S	Oligodendroglioma	High	N/A	340
24	Boxer	40	8	M/C	Oligodendroglioma	High	N/A	357
25	English Bulldog	20	9	F/I	Oligodendroglioma	Low	High	369
26	English Springer Spaniel	18	7	M/C	Undefined glioma	High	High	534
27	Boxer	26	9	F/S	Oligodendroglioma	High	N/A	>2165b
28	German Shepherd Cross	20	11	M/I	Astrocytoma	Low	Low	522
29	Golden Retriever	19	1	F/S	Astrocytoma	Low	Low	726
30	Mixed Breed	24	14	M/C	Astrocytoma	Low	N/A	743
31	Golden Retriever	27	12	F/S	Astrocytoma	Low	Low	801
32	Pembroke Corgi	13	9	F/S	Astrocytoma	Low	Low	853
33	American Bulldog	25	11	F/S	Astrocytoma	Low	N/A	1720

Abbreviations: M/C, male, castrated; F/S, female, spayed; FI, female, intact; NA, not available; MI, male, intact.

^aFinal diagnoses were based on biopsy or whole brain when available.

^bPatient 27 was still alive at 15 years old as of July 2019.

Pathologic diagnoses and overall survival are shown in Table 1. Histologic scoring and data on biopsy and whole-brain samples are shown in Table 2. Tumor type as diagnosed on both biopsy sample and whole-brain sections correlated with clinical outcome (Fig. 1). A diagnosis of astrocytoma correlated with longer survival times than a diagnosis of either oligodendroglioma or undefined glioma. Based on diagnoses obtained from biopsies, the survival times were 743 days for astrocytoma, 205 days for oligodendroglioma, and 144 days for undefined glioma (P = .0083). Based on diagnoses obtained from whole-brain samples, the survival times were 726 days for astrocytoma, 190 days for oligodendroglioma, and 153 days for undefined glioma (P = .0093).

Figure 1.

(a) Patient survival by diagnosis on biopsy sample. Astrocytoma n = 7, oligodendroglioma n = 19, undefined n = 7, P = .0083. (b) Patient survival by diagnosis on whole-brain sample. Astrocytoma n = 5, oligodendroglioma n = 10, undefined n = 6, P = .0093. Figure 2. (a) Patient survival by tumor grade on biopsy. Low grade n = 8, high grade n = 25, P = .007. (b) Patient survival by tumor grade on whole-brain sample. Low grade = 4, high grade = 17, P = .0017.

Tumor grade as determined on both tumor biopsy and whole-brain samples sections correlated with clinical outcome (Fig. 2). A diagnosis of low-grade glioma correlated with a significantly longer survival time than a diagnosis of high-grade glioma. Based on diagnoses obtained from biopsies, the survival times were 734 days for low-grade glial tumors and 194 days for high-grade glial tumors (P = .007). Based on diagnoses obtained from whole-brain samples, the survival times were 763 days for low-grade glial tumors and 185 days for high-grade glial tumors (P = .0017).

The presence of glomeruloid vascularization in the tumor correlated with poorer clinical outcomes (Fig. 3). Based on tumor biopsy samples, the survival times were 170 days for dogs with tumors exhibiting glomeruloid vascularization and 223 days for those not exhibiting glomeruloid vascularization. Based on whole-brain samples, the survival times were 162 days for dogs with tumors exhibiting glomeruloid vascularization and 630 days for those not exhibiting glomeruloid vascularization. Survival time by presence of glomeruloid vascularization was significant for whole brain sections (P = .0002), but not for biopsy samples (P = .2586).

Figure 3.

(a) Patient survival by glomeruloid vascularization on biopsy. GV not observed n = 19, GV observed n = 14, P = .2586. (b) Patient survival by glomeruloid vascularization on whole-brain sample. GV not observed = 6, GV observed = 15, P = .0002. Figure 4. (a) Patient survival by necrosis on biopsy. Necrosis not observed n = 21, necrosis observed = 12, P = .7189. (b) Patient survival by necrosis on whole-brain sample. Necrosis not observed = 8, necrosis observed = 13, P = .0123.

The presence of necrosis in the tumor correlated with a worse clinical outcome (Fig. 4). Based on tumor biopsy samples, the survival times were 154 days for dogs with tumors exhibiting necrosis and 220 days for those not exhibiting necrosis. Based on whole-brain samples, the survival times were 162 days for dogs with tumors exhibiting necrosis and 407 days for those not exhibiting necrosis. Survival time by presence of necrosis was significant for whole-brain sections (P = .0123) but not for biopsy samples (P = .7189).

The numbers of mitotic figures in tumors correlated with survival times (Fig. 5). Dogs with higher numbers of mitotic figures in both biopsy and whole-brain sections had shorter survival times than those with lower numbers of mitotic figures. Tumor biopsy samples in which the number of mitotic figures were graded as 0, 1, 2, 3, and 4 corresponded to median survival times of 363, 205, 218, 158, and 60 days, respectively (P = .0421). Based on biopsies, those with no or rare mitotic figures (graded as 0 or 1) had median survival times of 293 days compared to 190 days for those with low, moderate, or high numbers of mitotic figures (graded as 2, 3, or 4). The number of mitotic figures was the only feature in biopsy samples that had statistical significance (P < .05). Whole-brain tumor sections in which the number of mitotic figures were graded as 0, 1, 2, 3, and 4 corresponded to median survival times of 624, 205, 293, 185, and 96 days, respectively (P = .0006). Based on whole-brain sections, those with no or rare mitotic figures (graded as 0 or 1) had median survival times of 522 days compared to 189 days for those with low, moderate, or high numbers of mitotic figures (graded as 2, 3, or 4).

Figure 5.

(a) Patient survival by mitotic count grades on biopsy (0 n = 8), (1 n = 9), (2 n = 6), (3 n = 9), (4 n = 1), P = .0421. (b) Patient survival by miotic count grades on whole-brain sample (0 n = 2), (1 n = 3), (2 n = 5), (3 n = 7), (4 n = 4), P = .0006.

Discussion

There exists a gap in the literature concerning long-term outcomes in dogs with glial tumors.^3,12,15 Prior studies fail to differentiate survival times according to specific tumor type and grade, and none discuss specific features of malignancy as seen on histology as they relate to survival. Here, we report specific histopathologic features of canine glial tumors that correlate with poorer survival times, namely, glomeruloid vascularization, increased numbers of mitotic figures, and presence of necrosis.

Our study highlights potential limitations of analyzing brain biopsies to determine prognosis. Specifically, overall survival was significantly correlated with the histological features of glomeruloid vascularization and presence of necrosis based on analysis of whole brains (P = .0002 and .0123, respectively) but not of biopsies (P = .2586 and .7189, respectively). Whereas those features, when present in biopsies, are consistent with a poor outcome, their absence could simply reflect nonrepresentative sampling of the tumors such that they are not captured in the biopsy samples. Glomeruloid vascularization is located in a distinct arcade at the margin of the tumor that could easily be missed in a biopsy. Zones of necrosis are often sporadic and of limited extent. We also cannot rule out, however, that the observed relationship of these survival outcomes with histological features in biopsies versus whole brains might also reflect progression in tumor malignancy in the interval of time between biopsy sampling and postmortem collection of the whole brain. Multiple studies of human glial tumors indicate that these tumors often undergo malignant transformation from low-grade tumors to high-grade malignant tumors.^11,13 Malignant transformation reflects increased rates of replication, angiogenesis, and migration due to genetic mutations. In contrast, higher numbers of mitotic figures were associated with a poor outcome in biopsies (P = .0421) as well as in whole-brain samples, suggesting that this observation is a more robust histological feature in biopsies for prediction of outcome.

The purpose of the current study was to to determine which histologic features may have prognostic value in canine glioma. Currently, for human glioblastoma, molecular diagnostics are required to make a definitive diagnosis including subtype.⁸ Until molecular diagnostics of glioma are validated for dogs, the CBTC recommends diagnosing canine glial tumors based on cellular morphology as determined by histopathology.⁵ Going forward, a combination of histologic and molecular diagnostics may be used to determine the diagnosis of these tumors that can then be related to prognosis once more information is available about response to different therapies.

The current study has some limitations. Relatively small numbers of cases were evaluated, and this precluded the application of more sophisticated multivariate statistical analyses to determine the independent contribution of different histological features to overall survival. Most astrocytomas in our study were diagnosed as low-grade tumors, which may skew the survival curve for dogs with these tumors. The proportions of tumor types in this study are similar to those of previous studies.^5,14 However, only 14% of the astrocytoma tumors in the current study were high grade, as compared to about 55% in a previous study.¹⁴ Furthermore, we note that these animals were treated with immunotherapy which could potentially influence the outcome.

In summary, specific features of malignancy such as glomeruloid vascularization, increased numbers of mitotic figures, and necrosis correlated with shorter survival in dogs with glioma. Dogs with astrocytoma lived significantly longer than those with oligodendroglioma or undefined glioma. In addition, dogs diagnosed with low-grade glioma lived longer than those with high-grade glioma. These findings were significant in whole brain samples, but not always for biopsy samples indicating that some biopsies may not be adequate enough or sufficiently representative for proper determination of diagnosis and prognosis. Thus, this study provides information on how histologic diagnoses and features of malignancy relate to long-term outcome in canine glioma.