Correlation between fluorodeoxyglucose positron emission tomography and magnetic resonance imaging findings of non-suppurative meningoencephalitis in 5 dogs

Abstract

This study characterized the [¹⁸F]2-deoxy-2-fluoro-D-glucose positron emission tomography (FDG-PET) findings of encephalitis in dogs and assessed the role of FDG-PET in the diagnosis of meningoencephalitis. The medical records, magnetic resonance (MR), and FDG-PET images of 3 dogs with necrotizing meningoencephalitis (NME), 1 dog with granulomatous meningoencephalitis (GME), and 1 dog with meningoencephalitis of unknown etiology (MUE) were reviewed. On the FDG-PET, glucose hypometabolism was identified in the dog with NME, whereas hypermetabolism was noted in the dog with GME. The T2-weighted images (WI) and fluid attenuated inversion recovery (FLAIR) images were characterized by hyperintensity, whereas the signal intensity of the lesions on the T1-WI images was variable. The metabolic changes on the brain FDG-PET corresponded well to the hyper- and hypointense lesions seen on the MR imaging. This type of tomography (FDG-PET) aided in the differentiation of different types of inflammatory meningoencephalitis when the metabolic data was combined with clinical and MR findings.

Introduction

Granulomatous meningoencephalitis (GME), necrotizing meningoencephalitis (NME), and necrotizing leukoencephalitis (NLE) are relatively common non-suppurative inflammatory disorders of the canine central nervous system (CNS) (1,2). The antemortem diagnosis of these disorders is clinically challenging due to their nonspecific neurological signs and the extensive list of differential diagnoses associated with the presenting symptoms. The tentative diagnosis of a specific form of meningoencephalitis has been made by signalment, clinical symptoms, cerebrospinal fluid (CSF) analysis, diagnostic imaging, and negative infectious disease titers. Results of CSF analysis and imaging may be inconclusive and histopathology is often required for a definitive diagnosis. When there is no definitive histopathological diagnosis, the terminology meningoencephalitis of unknown etiology (MUE) may be used (2).

Magnetic resonance imaging (MRI) is useful for the diagnosis of intracranial inflammatory conditions, and there have been some reports of MRI in dogs with meningoencephalitis (3–6). However, dogs with inflammatory disease may have a normal MRI of the brain. Histopathological and CSF abnormalities have been more commonly reported than MRI abnormalities in dogs (6–8), cats (9), and humans (10) with encephalitis.

Positron emission tomography (PET) is a new imaging technique that enables qualitative and quantitative measurement of tissue metabolism and physiology in vivo with positron-emitting radionucleotides (11). [¹⁸F]2-deoxy-2-fluoro-D-glucose (FDG) is the metabolic tracer most widely used for clinical PET. This tracer acts as a glucose molecule in that it is transported into a tissue by the same mechanisms of glucose transport and is trapped in the tissue as FDG-6-phosphate, which is an inappropriate substrate for the enzyme glucose-6-phosphatase to continue glycolysis (12). Thus FDG accumulation reflects the rate of glucose utilization in a tissue, and mapping of FDG uptake within the body is known to be altered in various inflammatory and neoplastic diseases (13). Especially, the ability of FDG-PET to detect even mild inflammation has been well-established and makes the FDG-PET a potential candidate for the detection of encephalitis (14). In human medicine, FDG-PET data increase the diagnostic confidence and allow for the unequivocal identification of intracranial inflammation in patients whose MR imaging findings are subtle (15,16). Thus, we thought that FDG-PET could be used for the evaluation of canine encephalitis because of its ability to assess the metabolic changes in the CNS. Previously we reported on glucose hypometabolism in areas of necrosis and cavitation associated with NME (17). The potential role of FDG-PET in the diagnosis and the differentiation of encephalitis, however, has not been fully established in veterinary medicine.

The purpose of this study was to characterize the FDG-PET findings of encephalitis in dogs and to assess the role of FDG-PET in the diagnosis of meningoencephalitis.

Materials and methods

Case selection

The medical records of dogs presented to the Veterinary Medical Teaching Hospital, Konkuk University, between November 2006 and May 2008, with MR and FDG-PET images of the brain and inflammatory CSF findings or a histological diagnosis of a non-suppurative inflammatory brain condition were reviewed. Five criteria were evaluated for the MUE cases: 1, focal or multifocal CNS signs; 2, negative blood and/or infectious disease titers (distemper, neosporosis, toxoplasmosis, and ehrlichiosis); 3, CSF mononuclear pleocytosis (> 3 mononuclear cells/μL) and/or a protein concentration > 0.3 g/L assessed by the urine strip test; 4, MRI changes characterized by single or multiple T2 and fluid attenuated inversion recovery (FLAIR) hyperintense lesions within the brain; and 5, postmortem histopathological confirmation of meningoencephalitis. Dogs were included in this study if they fulfilled the first 3 criteria and at least 1 of the others. The breed, sex, age, type, and onset of neurological signs, CSF analysis, therapeutic drugs, survival time, diagnostic imaging, and histopathological findings were reviewed for each dog.

MR imaging

Magnetic resonance imaging was performed with a 1.5-tesla MR system (Magnetom Avanto; Siemens AG, Germany). T1-weighted, T2-weighted, and FLAIR images of the brain were obtained in the transverse, sagittal, and dorsal planes. The imaging protocol consisted of T1-weighted sequences (TR, 420 ms; TE, 4.57 ms), T2-weighted turbo spin echo sequences (TR, 6980 ms; TE, 86 ms), and FLAIR sequences (TR, 8920; TE, 115 ms). The MR images were assessed according to the following subjective criteria: lesion site (telencephalon, diencephalon, mesenchephalon, metencephalon, myelencephalon, and/or meninges); distribution (focal, multifocal and diffuse); margins (distinct or indistinct); presence of ventricular dilatation (mild, moderate and marked); signal intensity of lesions (hypointense, isointense or hyperintense relative to cerebral gray matter).

FDG-PET scanning

In all cases, we performed a FDG-PET of the brain, with the owner’s permission. The dog was fasted for 12 h, to ensure a stable FDG uptake, and then injected with FDG [0.4 mCi/kg body weight (BW)]. The FDG was produced immediately before injection using the onsite cyclotron (Eclipse HP Cyclotron; CTI Molecular Imaging, Knoxville, Tennessee, USA) at the Neuroscience Research Institute of Gachon University. The dog was placed on a heating pad and kept at 38°C for 1 h; the scanning was performed at the MicroPET (Focus 120, Concorde Microsystems, Knoxville, Tennessee, USA) under general anesthesia maintained with isoflurane (Attane; Minrad, Orchard Park, New York, USA). The metabolic imaging was performed with a 1.18 mm (radial), 1.13 mm (tangential) and 1.45 mm full width at half maximum resolution at the imaging center. The scanning time was 1 h. MicroPET images were reconstructed using the ordered subset expectation maximization algorithm with 10 iterations and a pixel size of 0.43 × 0.43 × 0.81 mm³. The glucose metabolism of the neuroanatomical locations that were related to CNS signs and abnormal MR images, were subjectively assessed by comparison with the corresponding areas of the contralateral hemisphere. Hypermetabolic and hypometabolic lesions were a strong reddish color and bluish or yellowish to greenish color, respectively.

Results

Signalments and clinical findings

Five dogs (3 females and 2 males) satisfied the inclusion criteria. The signalments, clinical, and diagnostic findings in these dogs are summarized in Table 1. The affected breeds were predominantly small breeds. The age at presentation varied from 2 to 14 years (mean age: 6.4 y). The most frequently observed neurological signs were seizures in 5 dogs, central blindness in 2, and forebrain dysfunction including circling and head turn in 2. The onset of the neurological signs was insidious in 1 dog (having multiple lesions) and acute in 4 dogs (2 with a single lesion and 2 with multiple lesions). At the time of writing, 4 dogs had died from progression of the meningoencephalitis and 1 dog, treated with prednisolone alone, was euthanized due to the poor response to therapy. The median survival time for 4 dogs with combination therapy was 236 d (range: 153 to 401 d).

Table 1.

The signalments and clinical characteristics of 5 dogs with meningoencephalitis

Dog	Breed	Sex	Age	Neurological signs	Onset of signs	CSF findings	Survival (d)	Therapy
1	Yorkshire terrier	Ma	10y	Status epilepticus, blindness, circling, head turn, hemiparesis	Acute	Monocytic pleocytosisc, elevated protein level	236	Pred.e + Cyclosporin
2	Maltese	Fb	2y	Seizures, depression, blindness, circling, head turn, recumbency	Insidious	Monocytic pleocytosis, elevated protein level	153	Pred. + Cyclosporin
3	Miniature poodle	F	14y	Status epilepticus	Acute	Monocytic pleocytosis, elevated protein level	155	Pred.+ Cyclosporin
4	Monglish	F	4y	Status epilepticus	Acute	Monocytic pleocytosis, elevated protein level	401	Pred.+ Cyclosporin
5	Mininature pinscher	M	2y	Seizures, ataxia, head tilt, nystagmus	Acute	Monocytic pleocytosis, elevated protein level	40d	Pred.

^aM — male;

^bF — female;

^cmonocytic pleocytosis (> 3 mononuclear cells/μL); elevated protein level (a protein concentration > 0.3 g/L assessed by the urine strip test);

^deuthanized following the owner’s request;

^ePred. — prednisolone.

MRI and FDG-PET findings

The MRI and FDG-PET findings are summarized in Table 2. The images of 4 dogs are shown in Figures 1 to 4. Histopathologically, 3 dogs were diagnosed with NME and 1 dog was diagnosed with GME. Necropsy could not be performed in 1 dog, thus the diagnosis for this dog was categorized as MUE based on signalments, clinical signs, results of the CSF analysis, and imaging findings. On MRI, lesions were detected in all 5 dogs, 4 of which had indistinct margins. The lesion distribution was considered to be focal in 2 dogs and diffuse in 3 dogs. The focal lesions affected the telencephalon or myelencephalon. Diffuse lesions were preferentially distributed throughout the telencephalon. Ventricular dilation was observed in 4 dogs.

Table 2.

Characteristics of the magnetic resonance and [¹⁸F]2-deoxy-2-fluoro-D-glucose positron emission tomography (FDG-PET) images observed in 5 dogs with meningoencephalitis

Dog	Distribution	Margin	Site	Dilated ventricles	T1WIa	T2WI	FLAIRe	FDG-PET	HDf
1	Diffuse	Indistinct	Telencephalon	Moderate	Hypob	Hyperd	Mixed	Hypometabolism	NMEg
2	Diffuse	Indistinct	Telencephalon, diencephalon	Marked	Isoc	Hyper	Hyper	Hypometabolism	NME
3	Diffuse	Indistinct	Telencephalon	Marked	Iso	Hyper	Hyper	Hypometabolism	NME
4	Focal	Indistinct	Telencephalon	Absent	Hypo	Hyper	Hyper	Hypermetabolism	MUEh
5	Focal	Distinct	Myelencephalon	Moderate	Hypo	Hyper	Hyper	Hypermetabolism	GMEi

^aWI — weighted images;

^bHypo — hypointensity;

^cIso — isointensity;

^dHyper — hyperintensity;

^eFLAIR — fluid attenuated inversion recovery;

^fHD — histopathological diagnosis;

^gNME — necrotizing meningoencephalitis;

^hMUE — meningoencephalitis of unknown etiology;

ⁱGME — granulomatous meningoencephalitis.

Figure 1.

Imaging characteristics and necropsy findings in a Yorkshire terrier with NME (dog 1). Multiple lesions are present in the left parietal and left and right temporal lobes. The lesions are hyperintense on T2-WI (A), hypointense on T1-WI (B), and iso- to hypointense on FLAIR images (C). In the internal capsules bilaterally (*) and the dorsal portion of the parietal lobe (arrows), the lesions are hyperintense on FLAIR images (C). (D) FDG-PET image depicts marked hypometabolism, which is shown with the bluish to greenish color. The distribution of hypometabolism corresponds to the abnormal regions noted on the MRI. On the MR and FDG-PET images, both lateral ventricles and third ventricle are enlarged. (E) In the cerebral cortex, diffuse severe atrophy was noted at necropsy. (F) Histopathologically, in the cerebral cortex, inflammatory lesions with cavitation and macrophage infiltration (arrows) are observed. Hematoxylin and eosin (H & E) × 200.

Figure 4.

Image characteristics and necropsy findings of a miniature pinscher with GME (dog 5). In the left myelencephalon (arrows), a focal lesion is characterized by hyperintense on T2-WI (A) and FLAIR (B) images. On the right side, a smaller but similar lesion is observed (*). (C) These lesions are hypointense on T1-WI (*). In all MR images, the fourth ventricle is moderately dilated. (D) On the equivalent regions compared to the MR images, the FDG-PET images depict hypermetabolism, shown with a strong reddish color (*). (E) At necropsy, on the dorsolateral portion of the left myelencephalon (arrows), granulomatous lesions were found. (F) Histopathologically, in the myelencephalon, perivascular (arrows) and parenchymal infiltration of macrophages, lymphocytes and plasma cells is present (H & E, ×200).

The T1-weighted images (WI) showed hypointense signals in 3 dogs and were isointense in 2 dogs. Of the 3 dogs with NME, 2 had isointensity and 1 had hypointensity. Hypointensities were found in the dogs with GME and MUE. On the T2-WI, the signal was hyperintense in all cases. There was a strong correlation between the signal and the location of lesions visible on the FLAIR images and those visible in the T2-WI. Mixed signals, however, were noted in the FLAIR images of 1 dog (Figure 1C).

All dogs had abnormal FDG-PET scans. The FDG-PET showed hypometabolism in the 3 dogs with NME and hypermetabolism in the dogs with GME and MUE. The distribution of hypo- or hypermetabolism corresponded well to the abnormal regions noted on the MR images.

Discussion

In this study, the metabolic changes of the brain on the FDG-PET corresponded well with the hyper- or hypointense lesions on the MR images. Glucose hypometabolism was identified with NME, whereas hypermetabolism was noted in GME.

Nonsuppurative meningoencephalitis has been described more commonly in small-breed dogs (4,6,8), but large-breed dogs can also be affected (3,8,18,19). The 5 dogs examined in this study were all small-breed dogs. The neurological signs present in these dogs reflected the location of the brain lesion. Four dogs presented with clinical signs suggesting a mixture of prosencephalic (cerebrum or thalamus) abnormalities, whereas 1 dog with GME showed central vestibular signs related to brainstem lesions. In 1 dog with disseminated lesions noted on imaging and histopathology, the onset of the clinical signs was insidious. However, 2 dogs with focal lesions had an acute onset of clinical signs. The rapid clinical course of 1 dog with focal GME is contrary to the more commonly described slowly progressive disease presentation. These findings suggest that lesion distribution may not be associated with chronology of clinical signs and is consistent with the reported variable clinical presentation of canine non-suppurative meningoencephalitis.

Generally, the leptomeningeal inflammation associated with NME and GME results in CSF mononuclear pleocytosis and increase in total protein. These characteristics were present in this study.

There are a few reports on the MRI changes in histologically confirmed NME in dogs (4,6,20,21). In these reports, the lesions appeared hypointense on T1-WI and hyperintense on T2-WI and FLAIR images. In the 3 dogs with NME most lesions were detected in the cerebrum. In these dogs some lesions appeared isointense on T1-WI (Figure 2C) with mixed signal intensities on FLAIR sequences (Figure 1C). In a previous report of NME in a Yorkshire terrier (6), the lesions were described as iso- to hypointense on T1-WI. In dog 1, lesions were hypo- and hyperintense on T1- and T2-WI (similar to CSF) and remained iso- to hypointense on FLAIR images (Figures 1A to 1C). Because FLAIR is a pulse sequence that suppresses the CSF, iso- to hypointensity of FLAIR images likely reflects similar protein content within the lesions compared to the CSF protein levels. These features may be related to the profound inflammation extending from the leptomeninges through the cerebral cortex into the parenchyma. Generally, this inflammation leads to a loss of the distinction between the gray-white matter junctions (22), which was demonstrated on necropsy. The MRI features of 1 dog with GME consisted of a single lesion in the myelencephalon characterized by high signal intensity on T2, FLAIR and low signal intensity on T1. This is in agreement with previous reports (3,23).

Figure 2.

Image characteristics and necropsy findings of a Maltese with NME (dog 2). In the dorsal aspect of the left cerebral cortex and subcortical white matter (thick arrows), piriform lobe (*), and thalamus (thin arrows), multifocal lesions are hyperintense on T2-WI (A) and FLAIR (B) images. (C) These lesions are isointense on T1-WI. (D) On the equivalent regions compared to the MR images, the FDG-PET image depicts marked hypometabolism, which is shown with yellowish to greenish color (*). In all images, both lateral ventricles and the mesencephalic aqueduct are severely dilated. (E) At necropsy, the atrophy of the left cerebral cortex and subcortical white matter (thin arrows), and the dilation of lateral ventricles were identified. (F) Histopathologically, in the cerebral cortex and meninges, malacic foci and mononuclear inflammatory changes are observed (H & E, ×100).

In human medicine, diagnosing encephalitis is difficult for the same reasons as in veterinary medicine. Brain biopsy is the sole method for an antemortem definitive diagnosis. However, there is a significant risk including fatal intracranial hemorrhage, sampling error, and nonspecific findings associated with these biopsies; thus, neuroimaging is often of critical importance for establishing a diagnosis of encephalitis, particularly in cases with nonspecific clinical findings (15,16). As a relatively noninvasive, low-risk procedure, FDG-PET has been used in the diagnosis of human encephalitis. In many cases, the FDG-PET data helped to establish the diagnosis and definitively identify the affected side (10,15,16). In this study, the abnormal lesions on the FDG-PET corresponded with the MRI and necropsy findings. Interestingly, the glucose uptake of the inflammatory lesions was opposite for NME and GME. Hypometabolism was identified in NME, whereas hypermetabolism was noted in GME. The accumulation of FDG reflects the rate of glucose utilization in tissues; since FDG is transported into a tissue by the same mechanisms of glucose transport (24). Histopathologically, NME is characterized by inflammatory changes consisting of lymphocytic, plasmacytic, and histiocytic infiltration and apparent parenchymal necrosis preferentially involving the cerebrum (25). Loss of brain tissue from necrotic changes including malacia, liquefaction, and cavitations, may be the cause of apparent glucose hypometabolism detected during FDG-PET. Previously we reported on evidence of glucose hypometabolism in areas of necrosis and cavitation associated with the NME (17); this is supported by the 3 cases with NME reported here. The FDG-PET findings in dogs with NME are similar to those in human cases of Rasmussen encephalitis. In 1 report, Rasmussen encephalitis in a human patient was related to hypometabolism on FDG-PET (15). This encephalitis is characterized by atrophy of the cerebral hemispheres; the etiology remains unknown (15) but a slow viral infection or autoimmune mechanism has been proposed (26,27). Granulomatous meningoencephalitis lacks the necrosis and microcavitation that occur in NME (22). The lesions associated with GME are characterized by perivascular cuffing, varying numbers of macrophages and plasma cells in the parenchyma, multifocal granulomas, and hemorrhage (23,25). These granulomatous inflammatory lesions may be associated with areas of hypermetabolism detected during FDG-PET.

The dog diagnosed with MUE (dog 4) had lesions of glucose hypermetabolism on FDG-PET which is most consistent with GME. The anatomic distribution of the lesions is most consistent with NME but may also be seen in some cases of GME. In 1 histopathological study of NME, the lesions were classified into 3 stages according to the severity of necrosis and the intensity of the inflammatory reaction (25). During the early clinical course, between the clinical onset and death, which ranged from 2 to 68 d, the lesions had moderate to severe inflammatory changes with a few malacic foci. Because necrosis is not well observed during the early stages, glucose hypermetabolism related to severe inflammation may be present in the early lesions of NME. However, we performed MRI and FDG-PET in dog 4 at 288 d after the clinical onset of symptoms. Thus, this dog might have had glucose hypermetabolism related to GME; severe malacic foci or cavitation is well known at this stage of NME. Future studies employing serial FDG-PET scans in cases of suspected canine meningoencephalitis will help determine the chronology of lesion detection and appearance. This will help correlate imaging findings with clinical presentation and resolution of clinical signs.

Recently, new radiotracers related to neuroinflammation, such as ¹¹C-rofecoxib and ¹¹C-PK11195, have been studied in human and laboratory medicine (28,29). ¹¹C-rofecoxib is used to investigate overexpression of cyclooxygenase type 2, which is triggered by inflammatory stimuli (28). ¹¹C-PK11195 can detect areas of neuroinflammation because it specifically binds to the peripheral-type benzodiazepine receptors expressed in activated microglia (29). The value of these markers in veterinary medicine should be investigated.

The main therapy for NME and GME is aggressive immunosuppression; corticosteroids have been widely used for treatment (22). Recently, various treatment protocols using cytosine arabinoside, cyclosporine, and procarbazine have been recommended due to the side effects of the treatment with steroids alone (1,20,30). In this study, the mean survival time of 4 dogs treated with a combination of prednisolone and cyclosporine was 236 d; however, 1 dog treated with prednisolone alone was euthanized 40 d after initiation of the therapy. The usefulness of cyclosporine either alone or in combination with prednisolone has been previously reported (20,30,31).

In conclusion, in this study of 5 dogs, FDG-PET contributed to the diagnosis of meningoencephalitis when the metabolic data was combined with MRI and clinical findings. Additionally the potential role of FDG-PET in differentiating different types of inflammatory encephalitis was identified. However, further studies with larger sample sizes are necessary to confirm the associations.

Figure 3.

Image characteristics of a monglish dog with MUE (dog 4). In the left cerebral cortex and subcortical white matter (thick arrows), lesions are hyperintense on T2-WI (A) and FLAIR (B) images. (C) These lesions are hypointense on T1-WI (*). (D) FDG-PET image depicts diffuse cerebral hypermetabolism, shown with a strong reddish color (*). This high metabolism is also shown in the ventral portion of the occipital lobe (thin arrow).