Gliomatosis Cerebri in Two Dogs
A 3.5 yr old Saint Bernard was evaluated for nonambulatory tetraparesis and cranial nerve dysfunction, and a 7 yr old rottweiler was evaluated for progressive paraparesis. Clinical signs of left-sided vestibular and general proprioceptive ataxia and cranial nerve VII dysfunction in the Saint Bernard suggested a lesion affecting the brain stem. Signs in the rottweiler consisted of general proprioceptive/upper motor neuron paraparesis, suggesting a lesion involving the third thoracic (T3) to third lumbar (L3) spinal cord segments. MRI was normal in the Saint Bernard, but an intra-axial lesion involving the T13–L2 spinal cord segments was observed in the rottweiler. In both dogs, the central nervous system (CNS) contained neoplastic cells with features consistent with gliomatosis cerebri (GC). In the Saint Bernard, neoplastic cells were present in the medulla oblongata and cranial cervical spinal cord. In the rottweiler, neoplastic cells were only present in the spinal cord. Immunohistochemistry disclosed two distinct patterns of CD18, nestin, and vimentin staining. GC is a rarely reported tumor of the CNS. Although GC typically involves the cerebrum, clinical signs in these two dogs reflected caudal brainstem and spinal cord involvement.
Introduction
Gliomatosis cerebri (GC) in humans is defined by the World Health Organization (WHO) as a diffuse glial neoplasm with pronounced cerebral infiltration, often extending to infratentorial structures and spinal cord.1 GC is generally classified as either type I or type II. Type I lacks mass formation and is characterized by diffuse neuroparenchymal infiltration with relative preservation of normal architecture, and type II features distinct mass formation.1–5 GC is infrequently reported in humans, and has only rarely been reported in the dog, cat, and goat.6–12 Distribution of GC in dogs has largely been restricted to the brain, but concurrent involvement of the spinal cord has been observed.7–10
This report describes two unusual canine cases of GC. In one, GC was limited to the caudal brain stem and cranial cervical spinal cord. In the other, GC was limited to the thirteenth thoracic (T13) through to the second lumbar (L2) spinal cord segments. Although microscopically the neoplastic cells were identical, marked differences were demonstrated by immunohistochemistry (IHC), highlighting the uncertainty regarding definitive identification of the cell of origin for GC.
Case Report
Case 1
A 3.5 yr old castrated male Saint Bernard weighing 83.3 kg was initially evaluated by the referring veterinarian for an inability to walk, a head tilt and tongue deviation, and hyperthermia (40.5°C). Despite antibiotic therapy (dosage unknown), the dog became depressed and was referred 5 days later. Upon presentation to the Lloyd Veterinary Medical Center at the College of Veterinary Medicine, Iowa State University, the dog was nonambulatory, had a left-sided head tilt, a heart rate of 200 beats/ min with weak femoral pulses, and a capillary refill time <2 sec. The dog was panting, but thoracic auscultation was normal. Rectal temperature was 41.1°C. The remainder of the physical examination was normal with the exception of a cranial drawer sign in the right stifle. Given the initial physical examination findings, initial assessment and treatment were focused on the cardiovascular system. Packed cell volume and total solids were 44% and 7.2 mg/dL, respectively. Systolic blood pressure was 210 mmHg, and an electrocardiogram disclosed sinus tachycardia. After an IV bolus (90 mL/kg) of lactated Ringer’s solutiona, the elevated heart rate, abnormal pulse quality, elevated systolic blood pressure, panting, and hyperthermia resolved.
Neurologic examination was subsequently performed. The dog’s mental state, proprioceptive placing, and spinal reflexes were normal. There was a left-sided vestibular ataxia, and cranial nerve (CN) exam revealed ventral strabismus, a decreased menace response, and a palpebral reflex in the left eye. There was a left-sided head tilt with spontaneous rotatory nystagmus, with the fast phase directed to the right. The neuroanatomic diagnosis was left-sided dysfunction of CN VII (facial) and VIII (vestibular branch). Otoscopic examination and thoracic and abdominal radiographs were normal. Differential diagnoses included otitis media/interna, endocrine-related neuropathies, or neoplasia involving the structures of the tympanic bulla or petrous temporal bone.
Complete blood count (CBC) revealed lymphopenia (0.2946 × 103 cells/μL; reference range, 1.0–4.8 × 103 cells/μL) and monocytosis (1.473 × 103 cells/μL; reference range, 0.15–1.35 × 103 cells/μL). Serum chemistry after IV fluid resuscitation disclosed decreased bicarbonate (14.8 mEq/L; reference range, 17–24 mEq/L), hypophosphatemia (2.0 mg/dL; reference ranges 3.2–6.2 mg/dL), hypomagnesemia (1.60 mg/dL; reference range, 1.82–2.49 mg/dL), increased creatinine (1.7 mg/dL; reference range, 0.1–1.2 mg/dL), and hyperglycemia (128 mg/dL; reference range, 75–115 mg/dL). The urine specific gravity was 1.015. These abnormalities likely reflected a combination of stress, previous dehydration, and IV fluid administration. Serum free thyroxine, total thyroxine, and thyroid stimulating hormone levels were normal.
MRIb of the head was performed. The following pulse sequences were obtained in the transverse and dorsal planes: T1-weighted (T1W), T2-weighted (T2W), T2W fluid attenuated inversion recovery (T2W FLAIR), short τ inversion recovery, T2*W, and fast imaging employing steady-state acquisition. After IV administration of gadopentetate dimegluminec (0.1 mmol/kg), transverse T1W images with and without chemical fat saturation images also were obtained. On MRI, asymmetric bilateral ventriculomegaly was observed, with the left lateral ventricle being larger than the right lateral ventricle. The brain stem, cerebellum, CN VII, CN VIII, tympanic bullae, and the petrous temporal bone appeared normal. Cerebrospinal (CSF) collected from the cerebellomedullary cistern was normal. The CSF WBC count was 0 cells/μL (reference range, 0–5 cells/μL), red blood cell count was 4 cells/μL (reference range, 0 cells/μL), and protein was 14 mg/dL (reference range, 10–24 mg/dL). A cause for CN VII and VIII dysfunction was not identified. The asymmetric ventriculomegaly was considered a congenital anomaly and likely not related to the neurologic deficits.
Symptomatic treatment was instituted consisting of famotidined (0.5 mg/kg per os [PO] q 12 hr) and meclizinee (0.25 mg/kg PO q 24 hr) for 10 days. The dog made gradual improvements and was discharged 2 days after admission. Two weeks later, the owner reported that the head tilt and vestibular ataxia continued to improve.
Six weeks later, the dog underwent surgery at a third veterinary hospital for treatment of right cranial cruciate ligament rupture. Postoperatively, the vestibular dysfunction and the ability to stand deteriorated, which prompted reevaluation at the Lloyd Veterinary Medical Center. Neurologic examination revealed severe left-sided head tilt and tetraparesis as well as left-sided postural reaction deficits of the thoracic and pelvic limbs (the pelvic limbs were more severely affected). At this point, the neuroanatomic diagnosis was consistent with left central vestibular disease and continued facial paralysis likely related to dysfunction of the CN VII nucleus. CBC and chemistry profiles were normal. The owner declined further diagnostics. The dog underwent 8 wk of physiotherapy, showed minimal improvement, and was ultimately euthanized after developing severe aspiration pneumonia.
Postmortem examination revealed gross lesions within the CNS, including bilateral hydrocephalus of the lateral ventricles, multifocal osseous metaplasia of the spinal meninges, and a single, poorly defined pale gray/tan region within the left dorsal and lateral funiculi of C1 and C2 spinal cord segments (Figure 1). Histologically, neoplastic cells infiltrated the spinal cord from C1 to C2 and extended into the caudal brainstem. The neoplastic cells had minimal cytoplasm, a single oval or elongate vesicular nucleus, and one to two small basophilic nucleoli. The mitotic rate was 2–3/10 400× fields. Wallerian-type axonal degeneration was observed in the adjacent white matter. Based on the morphologic features of the cellular infiltrate and its diffuse nature, a diagnosis of GC was made.
Citation: Journal of the American Animal Hospital Association 48, 5; 10.5326/JAAHA-MS-5796
Case 2
A 7 yr old male rottweiler weighing 29 kg presented to the Veterinary Medical Teaching Hospital at the College of Veterinary Medicine, University of Georgia for evaluation of a 6-mo history of pelvic limb weakness. Although initially static, 3 wk prior to evaluation, the owner observed progression of the pelvic limb weakness. Physical examination was normal, but neurologic examination revealed ambulatory general proprioceptive ataxia and upper motor neuron paraparesis. Bilateral postural reaction deficits were present in the pelvic limbs. Spinal reflexes, muscle tone, and muscle mass were normal except for the absence of patellar reflexes bilaterally. The dog was nonpainful on palpation along the thoracolumbar vertebral column. Despite the lack of patellar reflexes, neuroanatomic diagnosis was consistent with a lesion involving the T3–L3 spinal cord segments; however, given the lack of patellar reflexes, a lesion affecting L4–L6 spinal cord segments, spinal nerves, or femoral nerves could not be excluded from consideration. Differential diagnoses included intervertebral disc disease, degenerative myelopathy, myelitis (infectious or noninfectious), and neoplasia.
CBC, chemistry profile, and urinalysis were normal. Abdominal and thoracic radiographs were normal except for ventral spondylosis deformans at T12–T13. An MRIf of the vertebral column from the T1 vertebra to the sacrum was performed. The following pulse sequences were acquired in the sagittal and transverse planes: T1W FLAIR, T2W, and T2W FLAIR sequences. Additionally, T1W FLAIR sequences using a chemical fat saturation were acquired after IV administration of gadopentetate dimegluminec (0.1 mmol/kg). MRI revealed an ill-defined intra-axial expansile lesion of the spinal cord with attenuation of the subarachnoid space and epidural fat from the midbody of T13 vertebra to the cranial aspect of the L2 vertebra (Figure 2). The lesion was hyperintense on T2W and T2W FLAIR sequences and isointense on T1W FLAIR sequences. After IV contrast administration, enhancement was not noted on T1W FLAIR fat suppression sequences.
Citation: Journal of the American Animal Hospital Association 48, 5; 10.5326/JAAHA-MS-5796
Analysis of CSF obtained from the subarachnoid space between the L5–L6 intervertebral articulation disclosed a mixed cell pleocytosis (17 cells/μL; reference range, 0–5 cells/μL) and increased protein (42.2 mg/dL; reference range, <24 mg/dl). Pleocytosis consisted of 55% lymphocytes, 40% foamy macrophages, and 4% nondegenerate neutrophils. Differential diagnoses for the spinal cord lesion included meningomyelitis (infectious or noninfectious), neoplasia, or a degenerative process. Ischemic myelopathy was considered unlikely given the chronic progressive history. Antibiotic therapy consisting of clindamycing (10.7 mg/kg PO q 12 hr) and doxycyclineh (10.7 mg/kg PO q 12 hr) was initiated. Serum antibody titers directed against Neospora caninum, Borrelia burgdorferi, Ehrlichia canis, and Rickettsia rickettsii were negative. Serum IgM antibody titers directed against Toxoplasma gondii were 1:64 (reference range, ≥1:64), and the IgG antibody titer was negative. Serum latex Cryptococcus antigen titer was negative.
The dog’s paraparesis remained stable for approximately 3 wk. The owner then noted worsening of the paraparesis. Antibiotic therapy was discontinued, and the dog was administered immunosuppressive doses of corticosteroids (prednisonei, 1 mg/kg PO q 12 hr); however, the dog’s condition continued to decline. Adjunctive therapy of cytosine arabinosidej (50 mg/m2 subcutaneously q 12 hr for 2 days) was administered. The dog’s condition remained unchanged for 4 days. Consequently, the owner elected euthanasia.
Postmortem examination revealed gross lesions in the CNS that were limited to swelling of the spinal cord between the T13–L2 vertebrae. The spinal cord appeared normal on cut section. Histologically, neoplastic cells were limited to spinal cord segments T13–L2. A distinct mass formation was lacking. In some areas, the gray matter-white matter interface was obscured due to infiltrates of plump, polygonal neoplastic cells. In other affected regions, infiltrating cells were less dense, and there was relative preservation of the parenchyma. The neoplastic cells were similar as described in case 1. Based on the morphologic features of the cellular infiltrate and its diffuse nature, a diagnosis of GC was made.
Comparison of Cases 1 and 2
Although neoplastic cells in both dogs were histomorphologically indistinguishable, significant differences were observed with respect to lesion location, gray/white matter distribution, and IHC staining patterns. In case 1, lesions were mostly within the white matter (i.e., the left dorsal and lateral funiculi) of spinal cord segments C1 and C2 (Figure 3A). Extension of the lesion was observed within the caudal brainstem at the level of the dorsal nucleus of the trapezoid body where neoplastic cells infiltrated the left caudal cerebellar peduncle and the area ventral to the fourth ventricle near the medial vestibular nucleus. In contrast, the lesions in case 2 predominantly involved the gray matter of T13–L2 spinal cord segments (Figure 3B). In case 2, low numbers of lymphocytes and plasma cells surrounded blood vessels in both the white and gray matter within and adjacent to the neoplastic infiltrate.
Citation: Journal of the American Animal Hospital Association 48, 5; 10.5326/JAAHA-MS-5796
IHC was performed on formalin-fixed, paraffin-embedded tissues from each case. The following antibodies were used for IHC testing using standard laboratory protocols: vimentink, pan-cytokeratinl, glial fibrillary acidic protein (GFAP )m, neuron-specific enolasen, neurofilamentso, CD18p, nestinq, CD133r, S-100s, CD3t, and CD79au. A Multi-Link secondary antibody (Biogenex Laboratories Inc, San Ramon, CA) was used in all IHC assays except for CD3. For this reaction, the secondary antibody EnVision (Dako North America Inc) was used.
The specifications and results of IHC staining have been summarized in Table 1. Neoplastic cells from both dogs lacked immunoreactivity for the following antigens: CD3, CD79a, neurofilaments, neuron-specific enolase, S-100, GFAP, cytokeratin, and CD133. In both dogs, GFAP immunoreactivity was restricted to reactive astrocytes intermixed within GFAP-negative neoplastic cells. Distinct staining patterns for vimentin, nestin, and CD18 antibodies were observed between the two dogs. Neoplastic cells of case 1 exhibited cytoplasmic immunoreactivity for vimentin (50%–55% of cells, Figure 4A) and nestin (35%–40% of cells, Figure 4B) but not CD18 (Figure 4C). In contrast, neoplastic cells of case 2 were infrequently immunoreactive for vimentin (5%–10%, Figure 4D) but a subpopulation of cells displayed immunoreactivity for nestin (15%–20% of cells exhibited punctate cytoplasmic staining, Figure 4E), and CD18 (25%–30%, Figure 4F).
CNS, central nervous system; GFAP, glial fibrillary acidic protein; IHC, immunohistochemistry; NSE, neuron-specific enolase.
Citation: Journal of the American Animal Hospital Association 48, 5; 10.5326/JAAHA-MS-5796
Samples were deparaffinized and processed for electron microscopy in attempt to further characterize these unusual neoplasms; however, fixation and processing artifacts were significant. Though not definitive, the unremarkable features of cells in these cases suggest that the tumors were composed of either an undifferentiated or immature glial cell population.
Discussion
In this report, both cases of GC lacked distinct mass formation (type I). In humans, neoplastic cellular infiltration typically involves two or more cerebral lobes, the corpus callosum, diencephalon, basal nuclei, and variable involvement of the brain stem.3 Of particular interest in the current cases, neoplastic infiltration of the cerebrum was absent. Instead, the brain stem and spinal cord (case 1) or spinal cord alone (case 2) was involved. Although the anatomic distribution of the neoplasm in the dogs reported herein were dissimilar to the distribution defined by the WHO for GC, the authors of this report believe that a diagnosis of GC is warranted as the neoplastic cells were histomorphologically indistinguishable from previous reports of GC where the anatomic distribution of the lesions matched the WHO definition.9 In humans, GC preferentially affects white matter; however, gray matter involvement has also been observed.13,14 In dogs, white matter is more commonly affected than gray matter, and tropism for CN nuclei has been reported.9 Within the spinal cord, neoplastic cells in case 1 preferentially involved white matter tracts, whereas neoplastic cells in case 2 preferentially affected gray matter.
Dogs with GC range in age from 3 to 11 yr of age.7–10 Clinical signs generally include mentation changes, postural reaction deficits, and CN dysfunction.7,9 To the authors’ knowledge, the current report is the first to document GC with only spinal cord involvement in the dog.
In humans, computed tomography (CT) and MRI play important roles in the diagnosis of GC. With CT, lesions are iso- to hypodense and do not display contrast enhancement.2–4,15 With MRI, lesions are iso- to hypointense on T1W images, hyperintense on T2W and T2W FLAIR images, and may display contrast enhancement.2–5,15–17 Given the paucity of reported cases in dogs, imaging studies are limited; however, results similar to humans would be expected.7,9 In previous reports, CT in one dog disclosed a mass in the frontoparietal lobe of the right cerebral hemisphere that displayed subtle contrast enhancement.9 Similar to case 2 described herein, MRI lesions reported in dogs have been ill-defined lesions that are hypointense on T1W, hyperintense on T2W and T2W FLAIR images, and some displayed contrast enhancement.7,9 MRI was normal in one dog with signs of caudal brain stem involvement, similar to case 1 in the current report.9 Imaging findings in GC can be nonspecific, making discrimination from other infiltrative diseases difficult; however, unusually widespread involvement of the neuraxis may provide a clue. Definitive diagnosis requires histopathologic evaluation.2,4,5,15–17
The fact that the cell of origin in primary GC remains unclear is reflected in the WHO classification of GC as a glial tumor of uncertain histogenesis.1 Although oligodendroglial differentiation has been reported, neoplastic cells in most human cases are immunoreactive for GFAP, supporting astrocytic differentiation.13,18–23 The histogenesis of GC in dogs is even more uncertain, although one recent report demonstrated distinct oligodendroglial lineage.24 Most canine cases are not astrocytic based on the absence of immunoreactivity for GFAP.9 GFAP staining in the current cases likely identifies reactive astrocytes within the lesion, whereas neoplastic cells lack immunoreactivity. Despite this, a recent report suggests that GC and astrocytoma share a common origin in dogs.8
Although the neoplastic cells in the present cases were histologically indistinguishable, IHC staining patterns were distinct. In case 1, diffuse immunoreactivity for vimentin and nestin (but not for CD18 and GFAP) suggests immature cells of neuroepithelial origin without astrocytic differentiation. Neoplastic cells in case 2 also lacked GFAP immunoreactivity; however, vimentin was infrequently expressed and punctate nestin immunoreactivity was observed in some cells. The significance of distinct nestin staining patterns (diffuse versus punctate) in these cases is unknown. Further, the expression of CD18 in a subpopulation of neoplastic cells in case 2 suggests divergence along the monocyte/macrophage/microglial line. Human GC can have histologic and IHC features of microglial differentiation.25 Therefore, although the neoplastic cells in the two cases presented herein were histomorphologically indistinguishable, “microgliomatosis” or “myelogliomatosis” may be a more qualifying diagnosis for case 2. Microgliomatosis has been reported in the dog, yet debate exists as to whether microgliomatosis actually is a variant of CNS lymphoma.26,27 In one study, cases of microgliomatosis lacked the IHC staining pattern of B cell lymphoma.10 Lack of immunoreactivity to CD3 and CD79a in both dogs presented herein suggests that CNS lymphoma is unlikely; however, the possibility of null cell lymphoma cannot be completely ruled out.
In the present cases, expression of two progenitor or stem cell antigens (i.e., nestin and CD133) was evaluated. Nestin is widely considered to be a marker of neuroepithelial progenitor cells in many mammalian species.28 Immunoreactivity for nestin has been demonstrated in human and canine GC.8,29 The CD133 antigen is expressed in neuronal and glial progenitor cells within brain neoplasms of humans.30,31 Nestin and CD133 coexpression has been considered to be a characteristic phenotypic marker of neural progenitor stem cells in both humans and dogs. Despite this, both dogs in this report lacked CD133 expression.31,32,33 It is unknown whether the lack of CD133 expression indicates greater differentiation or a nonneural progenitor cell origin.
Conclusion
GC is a rare differential diagnosis for an expansile spinal cord lesion that is hyperintense on T2W images without contrast enhancement. More common considerations for such imaging findings include compressive or noncompressive intervertebral disc herniation and ischemic myelopathies.34–37 In most cases, an accurate presumptive antemortem diagnosis can be established through signalment, a thorough anemesis, observation of an extradural lesion (i.e., intervertebral disc disease), and clinicopathologic data. Lack of an abnormal MRI in a dog with progressive neurologic signs warrants consideration of GC. Ultimately, the diagnosis of GC should be based on widespread CNS infiltration (usually without a mass) and histomorphologic characteristics of the neoplastic cells.9 With increasing knowledge of the IHC staining features of CNS tumors, a predominant pattern of immunoreactivity in GC may emerge. However, given the uncertainty as to the origin of GC, heterogeneity in IHC findings may be expected. At present, IHC staining should be used to eliminate other neoplasms with similar microscopic features from consideration.
The authors would like to thank Ms. Deb Moore (Veterinary Pathology, Iowa State University) for her expertise and assistance with the IHC staining and Ms. Judith Stasko (National Animal Disease Center, Ames, IA) for assistance with electron microscopy sample processing for these cases.
Photograph of the formalin-fixed spinal cord (C1–C2 vertebra) of case 1. A subtle, poorly defined, darkened region within the left lateral funiculus of the C1–C2 spinal cord segments is present (arrow). Bar= 500 mm.
A sagittal T2-weighted MRI of the vertebral column of case 2. Note the intra-axial hyperintensity in the spinal cord segments contained within the T13–L2 vertebrae (arrow). Additionally, the affected spinal cord is enlarged compared with the spinal cord cranial to the hyperintense area.
A: Photomicrograph of sections of the C1–C2 spinal cord segments from case 1 showing the spinal cord white matter was infiltrated by neoplastic polygonal to spindle shaped cells. B: Photomicrograph of sections of T13–L2 spinal cord segment from case 2 showing the spinal cord gray matter infiltrated by neoplastic polygonal to spindle shaped cells surrounding multiple neurons (arrows). Hematoxylin and eosin staining, bar = 20 μm.
Immunhistochemistry results from case 1 (A–C) and case 2 (D–F). The cells have diffuse cytoplasmic immunoreactivity for vimentin and nestin (A, B), but lack immunoreactivity for CD18 (C). Note the gitter cell within a dilated myelin sheath that is positive for CD18 (C, arrow). Cells from case 2 exhibit minimal cytoplasmic immunoreactivity for vimentin (arrowheads), whereas the blood vessels (arrows) are positive internal controls (D). The cells exhibit variable staining in a punctuate perinuclear and cytoplasmic pattern for nestin (E), and variable cytoplasmic immunoreactivity (arrows) for CD18 antigen (F). The asterisk (*) indicates a neuron. Bars = 20 μm.
Contributor Notes
B. Plattner's present affiliation is Department of Pathobiology, Ontario Veterinary College, University of Guelph, Guelph, ON, Canada.
J. Clemans’ present affiliation is Sugar Land Veterinary Specialist, Center Sugar Land, TX.
D. Garcia-Tapia's present affiliation is Pfizer Animal Health, Department of Metabolism and Safety, Kalamazoo, MI.
