Corresponding author: Dr B.K. Sturges, Department of Surgical and Radiological Sciences, School of Veterinary Medicine, University of California-Davis, Davis, CA 95616-8747l; e-mail: firstname.lastname@example.org.
Background: Intracranial meningiomas are the most common primary brain tumors in dogs. Classification of meningiomas by tumor grade and subtype has not been reported, and the value of magnetic resonance imaging (MRI) characteristics for predicting tumor subtype and grade has not been investigated.
Hypothesis: Canine intracranial meningiomas are a heterogenous group of tumors with differing histological subtypes and grades. Prediction of histopathological classification is possible based on MRI characteristics.
Animals: One hundred and twelve dogs with a histological diagnosis of intracranial meningioma.
Methods: Retrospective observational study.
Results: Meningiomas were overrepresented in the Golden Retriever and Boxer breeds with no sex predilection. The incidence of specific tumor grades was 56% benign (Grade I), 43% atypical (Grade II), and 1% malignant (Grade III). Grade I histological subtypes included meningothelial (43%), transitional (40%), microcystic (8%), psammomatous (6%), and angiomatous (3%). No statistically significant (P < .05) associations were found among tumor subtype or grade and any of the MRI features studied.
Conclusions and Clinical Importance: Meningiomas in dogs differ from their counterparts in humans mainly in their higher incidence of atypical (Grade II) tumors observed. MRI characteristics do not allow for prediction of meningioma subtype or grade, emphasizing the necessity of histopathology for antemortem diagnosis. The higher incidence of atypical tumors in dogs may contribute to the poorer therapeutic response in dogs with meningiomas as compared with the response in humans with meningiomas.
Meningiomas are common intracranial tumors of the central nervous system in dogs, comprising approximately 40% of all primary spontaneous tumors diagnosed at necropsy.1–5 These tumors share striking similarities to human meningiomas in neuroimaging characteristics, gross and histological appearance, and expression of growth factors and receptors.6–9 The current World Health Organization (WHO) histological classification system of canine meningiomas10 categorizes them into 2 major groups: benign, slow-growing tumors of various subtypes (meningothelial, fibroblastic, transitional, psammomatous, angiomatous, papillary, granular, myxoid), and anaplastic tumors. The microcystic subtype was later identified independently.11 In humans, a more detailed classification system has been applied, allowing consistent correlation of histological findings with biological behavior. A major component of this classification system is the grouping of tumors into 1 of 3 histological grades (benign, atypical, anaplastic) that is critical for predicting clinical behavior and outcome in humans.12,13 In dogs, the current classification of meningiomas is based on tumor subtype alone and does not distinguish grades of malignancy other than anaplastic. No data are available correlating biological tumor behavior with the current or any other classification system in dogs with meningiomas.
The magnetic resonance imaging (MRI) features of canine meningiomas are variously described in small case series and case reports,5,7,9,14–23 and predilection sites have been proposed based on the dural site of origin. In humans, the predictive accuracy of diagnosing meningiomas based on their MRI characteristics is between 65 and 96%.24–26 In addition, the potential to predict tumor subtype or grade, and therefore biological behavior, has been investigated in studies correlating neuroimaging features and anatomical location with gross and histological classification of human meningiomas according to the scheme of Russell and Rubinstein27 or the latest, most widely accepted WHO international histological classification of tumors of the nervous system.12,13 Although reliable prediction of specific tumor subtypes is not reported, the use of adjunctive MRI features in conjunction with signal characteristics may provide data predictive of general histological patterns, vascularity, tumor consistency, and tumor grade. 28–31
Because canine intracranial meningiomas are histologically very similar to their human tumor counterparts, clinical predictive and prognostic indicators used in people may be valuable in the management of dogs with meningiomas. Application of this approach in veterinary neuro-oncology is limited by lack of a comprehensive classification system for canine meningiomas based on both tumor histological subtype and grade. The aims of this study were to (1) develop a histological classification scheme for grading canine meningiomas (based on the human WHO international classification scheme), (2) describe a subtype of meningioma not reported previously, (3) present an anatomical classification scheme based on tumor predilection sites, and (4) correlate MRI findings with subtype and grade of tumor.
Materials and Methods
The medical records of all dogs referred between 1995 and 2005 for evaluation of intracranial disease were reviewed. Criteria for inclusion of dogs in the study were (1) cranial MRI and (2) histological diagnosis of an intracranial meningioma.
Intracranial MRI studies were acquired with either a 0.34-Ta (55 dogs) or a 1.5-Tb (58 dogs) imaging system. Noncontiguous transverse slices with a 5-mm slice thickness were generated using standard spin echo pulse sequences for T2-weighted (T2W) images and pre- and postcontrastc T1-weighted (T1W) images. Pre- and postcontrast T1W sagittal images were also acquired.
Intepretation of MRI
MRI features were classified based on the consensus opinion of 2 ACVIM board-certified neurologists (BKS, GDK) and 1 ACVR board-certified radiologist (PDK). The interpretation was based on the majority of the tumor area. Each meningioma was classified on the basis of signal intensity relative to cortical gray matter (hypointense, isointense, or hyperintense) and the signal uniformity (homogenous or heterogenous) on T1W and T2W images. On T1W postcontrast images, signal uniformity was evaluated and classified as either heterogenous or homogenous, and the borders of the tumor were defined as either sharply or poorly defined. Fluid accumulations associated with the tumor were categorized based on T2W and T1W images and were interpreted as representing either a cystic structure or an intratumoral accumulation of fluid (ITF). A cyst was defined as a prominent well-defined accumulation of fluid, eccentrically located in the tumor (Fig 1). Intratumoral fluid accumulation was defined as ≥1 indistinct pockets of fluid contained within the main tumor mass (Fig 2). Peritumoral edema was classified as present or absent based on T2W images. If present, it was graded as peritumoral if it extended ≤10 mm beyond the tumor margin on T2W images (Fig 3) or as diffuse if it extended for >10 mm beyond the tumor margin (Fig 4). When either a dural tail sign was seen or meningeal enhancement was present adjacent to or distant from the main tumor mass, each of these features was classified as meningeal involvement (MI).
Selected tissue from each meningioma was immersed in 10% buffered formalin and processed by routine paraffin embedding, and sections were cut at 5 μm and stained with hematoxylin and eosin before consensus histological classification and grading by 1 veterinary (RJH) and 1 human medical (AWB) neuropathologist according to the criteria in the latest WHO international histological classification of tumors of the nervous system of humans (Figs 5–12). 12,13 Tumors were graded as benign (Grade I), atypical (Grade II), or anaplastic (malignant or Grade III). The benign histological subtypes recognized included meningothelial, transitional, microcystic, psammomatous, and angiomatous. Diagnostic criteria for atypical meningiomas were either mitotic activity (>4 mitoses per 10 high power fields [HPF] with each field representing 0.16 mm2) or ≥3 of the following features: increased cellularity, small cells, nuclear atypia, tumor necrosis, or patternless cell sheets. The chordoid histological subtype was included in this group. Anaplastic meningiomas had cytological features of overt malignancy that were far more accentuated than were those in the atypical tumors. These features included the following: a high mitotic index (≥20 mitoses per 10 HPF of 0.16 mm2), obviously malignant cytology (sarcoma or carcinoma like), and frank anaplasia. According to the current WHO criteria for human meningiomas, tumor histological subtypes were not defined for Grades II and III meningiomas. When brain invasion was seen, or could be assessed from the tissues, it was recorded for the meningioma in question but not used as a criterion for grading.
Anatomical Location of Tumors
Meningiomas were described, according to their anatomical site, as olfactory, parasellar, parasagittal, falcine, basilar, tentorial, over the cerebral or cerebellar convexity, in the cerebellopontine (CP) angle, or extending through the foramen magnum. The anatomical criteria defining each location are summarized in Table 1.
Table 1. Location of intracranial meningiomas in dogs.
Tumor Location and Subgroups
Site(s) Tumor Origin
Cribriform plate Presphenoid bone Cranioventral extent of frontal bone
Cranial to a transverse line drawn at the level of the optic canals
Basisphenoid bone Hypophyseal fossa Tuberculum sellae Cranial/caudal clinoids Cavernous sinus
Begins caudal to olfactory location and extends caudally to include dorsum sellae
Parasaggital angle (the space formed by the convexity dura mater and the lateral wall of the falx cerebri)
Cribriform plate is cranial extent (dorsally); tentorium ossium is the caudal extent (dorsally)
Falx cerebri (tumors completely concealed by overlying cerebral cortex)
Cerebral convexity without involvement of parasaggital, falcine, or parasellar locations
Cerebral convexity without attachment to tentorium or sinus
Infratentorial masses in caudal fossa; supratentorial masses in middle fossa
Cerebellopontine or cerebellomedullary angle
Arising laterally in caudal fossa
Cerebellar vermis or hemispheres without attachment to tentorium or sinus
Basilar region of brainstem; rectangular area coning down toward foramen magnum
Pontine impression of the basioccipital bone; ventral midline in caudal fossa
Same as basilar
Same as basilar but extending through the foramen magnum
For statistical analysis, χ2 tests of homogeneity were used to evaluate all bivariate associations between categorical variables. Each histological grade and subtype (of Grade I tumors) were compared with each of the following MRI characteristics: T1W and T2W signal uniformity and intensity, signal uniformity and border definition on postcontrast images, type of edema, presence of cysts, ITF, or both, MI, and tumor location. In addition, the pattern of edema was compared with the tumor location and the presence of cysts, ITF, or both. Likewise, the presence of cysts, ITF, or both and MI each was compared with tumor location. Tumor histology (grade and subtype) was compared in each of the commonly represented dog breeds, as were MRI characteristics, tumor location, type of edema, and presence of cysts, ITF, or both. Finally, the distribution of the various purebred dogs in the study was compared with the distribution of these breeds in VMTH accessions during the same time period. Differences in age distribution between tumor grade groups were evaluated using Kruskal-Wallis analysis of variance. Statistical analysis was carried out using a statistical software programd and statistical significance was defined as P < .05.
One hundred and twelve dogs fulfilled the inclusion criteria: 20 intact males, 44 neutered males, 5 intact females, and 43 neutered females. The dogs ranged in age from 1 to 16 years with a mean and median age of 10 years. There were 105 purebred dogs, representing 34 different breeds, with the most frequently represented breeds including Golden Retriever (n = 26), Labrador Retriever (n = 14), Boxer (n = 14), and German Shepherd dogs (n = 8); the other 8 dogs were of mixed breed origin. A significantly higher number of Golden Retrievers and Boxers were represented as compared with the general hospital population during this time (P= .025).
Anatomical Location of Meningiomas
One hundred and fourteen tumors occurred at various sites: the olfactory (41), over the cerebral (16) or cerebellar convexity (2), parasagittal (11), CP angle (11), parasellar (9), basilar (8), tentorial (7), falcine (7), and foramen magnum (2) sites. Two of the dogs in the study had 2 meningiomas each (a total of 114 tumors in 112 dogs). Both tumors in each dog were of the same histological grade and subtype, and therefore were counted as the same tumor in all analyses except tumor location. No significant association was found between anatomical location and signalment, tumor grade, subtype, presence of cysts, or MI. Tumors located in the cerebral convexity, olfactory, and parasagittal regions were significantly more likely to have diffuse edema (P= .035), whereas tumors in the CP angle and basilar regions were most likely to have either focal edema or none at all (P= .018). In addition, supratentorial meningiomas were more likely to have ITF accumulation than were infratentorial meningiomas.
Histological grading identified 63/112 benign, 48/112 atypical, and 1/112 malignant meningiomas. Histological subtypes of the benign tumors consisted of 27/63 meningothelial, 25/63 transitional, 5/63 microcystic, 4/63 psammomatous, and 2/63 angiomatous patterns. There were 2 chordoid meningiomas (atypical). The occurrence of brain invasion could not be determined for all tumors because of sampling limitations, but invasion occurred in some of each of the benign and atypical tumors.
The distribution of the 63 benign tumors between sexes of dogs was 13 male, 2 female, 24 male neutered, and 24 female neutered. The sex distribution of 48 atypical meningiomas was 7 male, 2 female, 19 male neutered, and 20 female neutered dogs. The malignant tumor occurred in a female dog.
Evaluation of the signal uniformity on T1W images indicated that 100/112 meningiomas were heterogenous, 78/112 were isointense, 21/112 were hyperintense, and 13/112 were hypointense. Contrast enhancement was homogenous in 74/112 tumors, whereas the other 38 were heterogenous. On T2W images, signal intensity was homogenous in 77/112 tumors and heterogenous in 35/112. The majority of meningiomas were hyperintense (83/112) or isointense (27/112) with respect to cortical gray matter. Other MRI findings included diffuse edema in 57/112 tumors, peritumoral edema in 48/112, and no edema in 7/112. Twenty-nine meningiomas contained large cystic structures; 18 had ITF accumulation. The tumor border was sharply defined in 97/112 tumors and poorly defined in the remainder. MI was seen in 30 tumors.
Correlation between Features of MR Imaging and Histological Classification and Grading
No significant associations were found between tumor grade and any of the MRI features, sex or age of the dog, location of tumor, pattern of edema, presence of cystic structures, presence of ITF, or meningeal enhancement (Tables 2 and 3). Additionally, no significant associations were found between any of the tumor subtypes (Grade I tumors) and the same variables. Edema was present in the majority of meningiomas (106/113), although no association could be found with tumor grade or subtype nor was an association with ITF or cysts identified.
We describe a series of 114 histologically confirmed meningiomas with respect to tumor subtype, grade, and location, and examined potential correlations between these findings and MRI features.
Optimal management of patients presenting with intracranial meningiomas is dependent on the ability to make an accurate histological diagnosis and an appropriate choice of therapy based on the likely biological behavior of the tumor. To provide such prognostic indicators, grading of meningiomas in humans now is standard practice, and the grading system considers tumor subtype as well as histological features associated with malignancy (eg, mitotic index, necrosis, cytological atypia) because atypical and malignant variants behave more aggressively and have a higher potential for recurrence. Histological classification of the canine tumors in this study, using criteria of the human WHO international histological classification,12,13 identified an incidence of 56% benign, 43% atypical, and 1% malignant meningiomas. This finding is in contrast to what is observed in human meningiomas, of which approximately 80% are benign, up to 8% are atypical, and < 3% are classified as malignant.32,33 Although direct comparisons cannot be made, response to standard surgical and radiation therapy in human meningioma patients appears to be better than the response in dogs (ungraded), with 5-year recurrence free rates of up to 90%.34 In comparison, median survival times for canine meningioma patients after comparable treatment protocols are 1.5–3 years.8,35,36 This difference most likely reflects many factors, including surgical and technical expertise,37 but the larger percentage of higher grade tumors in dogs may be a clinically relevant factor in the poorer therapeutic response compared with what is observed in human meningioma patients, particularly when reported cases have not been stratified according to tumor grade.
Meningothelial and transitional subtypes of meningiomas were the most common Grade I tumors in this study, similar to other reported studies.11,38 However, conclusions cannot be made because other studies have not used similar criteria for grading canine meningiomas. The 2 chordoid meningiomas (not reported previously) were graded as atypical tumors in addition to the other 46/112 atypical meningiomas reported here. Some of the meningiomas reported previously in the veterinary literature may have met the criteria of atypical tumors had the human WHO classification criteria been applied.
Although Golden Retrievers, Labrador Retrievers, Boxers, and German Shepherd dogs comprised more than over 50% of the dogs in this study, only the Golden Retriever and the Boxer were significantly overrepresented compared within the normal hospital admission breed profile. An increased incidence of meningiomas has been reported in German Shepherd dogs, Collies, Golden Retrievers, and Boxers, but no statistical analysis of data was carried out.2,35–37,39,40 Although 1 large study1 and some small studies have suggested an increased incidence of meningiomas in female dogs, most studies have found no sex predilection2,8,35–37,39,40 and no statistical difference in sex incidence was seen in this study. This finding differs from human meningioma patients where females are overrepresented with low-grade tumors and males with higher grade tumors.32
Although a definitive diagnosis of intracranial meningioma requires histopathological examination of tumor tissue, biopsy or resection of tumors is not possible in all cases. Under such circumstances, the ability to predict tumor type and grade based on MRI findings would be beneficial, and has been investigated for human intracranial meningiomas. Among the meningiomas in this study, there was no apparent correlation between standard MRI findings and histological subtype or tumor grade. Atypical meningiomas had more distinctly defined borders on postcontrast T1W images than did benign tumors, but this difference did not achieve statistical significance (P= .054). The potential clinical relevance of such a difference argues strongly for further study of this question in a larger sample to provide a more definitive answer and a more precise estimate. Studies carried out in humans to determine whether or not the subtype or grade of meningioma can be predicted from MRI have yielded generally inconclusive results. Some studies have found no correlation with tumor classification,41–44 whereas others reported that the signal intensity of T2W images directly correlated with the histologic subtype in the majority of cases.28–31,43 Tumors with hyperintensity on both T1 and T2W images, extensive edema, and central necrosis have been reported to exhibit more aggressive behavior and a higher grade, as have tumors with irregularly defined borders.25
The most consistent MRI feature in the meningiomas reported here was precontrast T1W images that were isointense to gray matter (70% of tumors). This finding is similar to those of 1 other study in dogs7 and several in humans.25,29–31,42,43 Two other studies in dogs9,20 reported that the majority of tumors were hypointense on T1W images (Table 4). This disparity may relate to lower numbers of tumors in these studies or the type of meningiomas studied, but no correlation between T1W signal intensity and histological subtype or grade has been found in humans28–31,43 or in the current study.
Table 4. Data of multiple studies on the MR imaging characteristics of canine meningiomas.
MR, magnetic resonance; ITF, intratumoral accumulation of fluid; NS, not studied.
No. of tumors
12.5% mixed iso/hyper
On T2W images, hyperintensity was noted in 75% of tumors whereas only 2% were hypointense, similar to the incidence in 2 other studies of dogs.9,20 T2W signal intensity appears to be more variable in human meningiomas, possibly related to tumor consistency or stiffness. Less than 50% of tumors, however, are hyperintense on T2W images.45,46 Increased T2W signal does not appear to be related to the higher incidence of atypical tumors in dogs, but may reflect differences in consistency between the spectrum of canine and human tumors studied. Histological evaluation of focal areas of hyperintensity indicated that such areas were either foci of low cellularity with presumed ITF accumulation or a consequence of frank intratumoral necrosis. Intratumoral hemorrhage with temporal changes in paramagnetivity may also contribute to focal hyperintensity.
The majority of human meningiomas have an intense pattern of T1W homogenous contrast enhancement and a sharply delineated border similar to what was observed in this study and a previous report in dogs.7 Three other studies in dogs reported that the majority of meningiomas were heterogeneously contrast enhancing,5,9,20 but sample sizes were small (Table 4). Meningeal enhancement, including dural tail sign, was seen in 27% of the meningiomas in this study and was not associated with tumor location, shape, or histological grade. The definition of dural tail sign has not been standardized,19 which may be reflected in the wide variation in the incidence of this sign (23–82%) reported in previous studies in dogs with meningiomas.17,19,20 Approximately 70% of the meningiomas had a spherical, ovoid, or globular shape, similar to previous reports.7,20 Plaque-like tumors were restricted to parasellar and basilar regions corresponding to the well-recognized growth patterns of canine meningiomas in these 2 locations.40
A finding of diagnostic and clinical relevance from this study was the detection of parenchymal edema, primarily affecting white matter, associated with 94% (105/112) of all meningiomas. This edema was extensive in 57 (51%) and restricted to a peritumoral location in 48 (43%) meningiomas. Edema is also a common finding in human meningiomas and can be seen in up to 67% of cases.25 Definitive correlations between edema and other factors, such as tumor grade, tumor size, and cortical disruption, have not been consistently described in humans, but tumors located rostrotentorially, particularly parasagittal, those involving the olfactory groove and frontal convexity, but not those in the posterior fossa, have been reported to be more likely associated with edema.47,48 Similar statistically significant correlations were seen in the present study. Vascular endothelial growth factor (VEGF) and its receptors are reported to play a major role in angiogenesis, vascular permeability, and tumor recurrence in human meningiomas.49,50 Both VEGF and its major receptor VEGFR-2 are upregulated in canine meningiomas.6,8 Survival time for canine meningioma patients was inversely associated with VEGF expression in 1 series,8 although all tumors were benign (Grade I). Corticosteroids are potent inhibitors of edema, potentially acting to inhibit VEGF expression. Thirty-nine dogs in this study were receiving corticosteroid therapy at the time of imaging, but none of the dogs without edema on MRI had received corticosteroids. Given the high prevalence of edema in canine meningiomas, additional studies investigating correlations between tumor grade and VEGF expression and edema are indicated.
Because of the striking pathological, immunological, molecular, and MRI similarities between human and canine meningiomas, this study applied the criteria used in the WHO international histological classification of human meningiomas for histological classification and grading. The advantages of this system over the current domestic animal WHO tumor classification system for meningiomas are that it includes a more comprehensive list of histological subtypes, has a detailed grading system, and is a system designed for prognostication of long-term outcome based on clinical experience. The classification system used in humans has been applied in a few recent studies of intracranial meningiomas in dogs8,11,38 because of the increasingly recognized limitations of the domestic animal classification scheme. From our results, it appears that the incidence of atypical (Grade II) meningioma is higher than in humans, and, as described previously,11,51 many of the benign tumors have mixtures of different histological patterns with the subtype identified based on the dominant cell pattern. However, this unique feature of canine meningiomas does not affect the criteria used for grading. We propose that the system used in humans be adapted universally because of its inherent advantages and we anticipate that it could be modified for canine meningiomas after prospective studies determining tumor classification (grade) and clinical outcome are completed. We have attempted to categorize tumors in dogs using the human WHO classification system to establish a more detailed system of grading for these tumors. As more information becomes available, this approach hopefully will provide a more detailed framework in which to correlate tumor histology with biological tumor behavior and long-term prognosis.