• Open Access

Diagnostic Yield and Adverse Effects of MRI-Guided Free-Hand Brain Biopsies through a Mini-Burr Hole in Dogs with Encephalitis


  •  Brain biopsies were taken at the University of Leipzig (Germany). Histopathological examinations of biopsy specimens were performed at the Ludwig-Maximilians-University Munich, Germany. Postmortem samples were examined at the Ludwig-Maximilians-University Munich, Germany, and at the University of Bern, Switzerland

Corresponding author: Thomas Flegel, Department of Small Animal Medicine, University of Leipzig, An den Tierkliniken 23, 04107 Leipzig, Germany; e-mail: flegel@kleintierklinik.uni-leipzig.de.



The diagnosis of encephalitis is usually presumptive based on MRI, cerebrospinal fluid analysis, or both. A definitive diagnosis based on histopathology, however, is required for optimizing treatment strategies.


To investigate the diagnostic yield and adverse effects of minimally invasive brain biopsies in dogs with encephalitis.


Seventeen dogs with suspected encephalitis, based on MR imaging and cerebrospinal fluid analysis.


Retrospective study. Minimally invasive, free-hand brain biopsy specimens were taken from forebrain lesions through a 4-mm burr hole using a Sedan side-cutting needle. Routine histopathological examination was performed. The adverse effects were assessed by MRI evaluations after biopsy procedure (12/17) and by sequential neurological examinations.


The overall diagnostic yield with regard to a specific type of encephalitis was 82%. Encephalitis was evident in an additional 12%, but a specific disease could not be determined. There were no deaths caused by the biopsy procedure itself, but the indirect case fatality rate was 6%. Morbidity was 29%, including stupor, seizures, tetraparesis, hemiparesis, ataxia, and loss of conscious proprioception. All these signs resolved within 3–14 days.

Conclusions and Clinical Importance

Minimally invasive brain biopsy in dogs with suspected encephalitis leads to a definite diagnosis in the majority of dogs, allowing for a specific treatment. The advantages of a definite diagnosis outweigh potential case fatality rate and temporary neurological deficits.


central nervous system


cerebrospinal fluid


computed tomography


fluid attenuation inversion recovery


granulomatous meningoencephalomyelitis


magnetic resonance imaging


minimally invasive brain biopsy


necrotizing leukoencephalitis


necrotizing meningoencephalitis

Brain biopsies have been proposed in veterinary medicine ever since advanced imaging modalities such as computed tomography (CT) and magnetic resonance imaging (MRI) became available for the visualization of intracranial pathologies intravitam. To avoid gross procedure-related brain damage, minimally invasive techniques have been developed for tissue sampling from deep-seated lesions. These techniques include stereotactic CT-based systems, optical tracking system, ultrasound-guided biopsy, endoscopic-guided bopsy, and free-hand CT-guided biopsy.[1-11]

Studies on the clinical use of minimally invasive biopsy systems have thus far focused on the diagnosis of neoplastic brain lesions.[2, 4, 7] There is only limited information available on the diagnostic feasibility of minimally invasive brain biopsies (MIBBs) for inflammatory brain lesions. One study included 8/50 dogs with encephalitis but without further specification.[2] Furthermore, there is 1 case report on Cladophialophora bantiana encephalitis in a dog diagnosed via stereotactic brain biopsy.[12] In addition to minimally invasive procedures, surgical biopsies were used in a diagnostic work-up of 7/42 dogs with granulomatous meningoencephalomyelitis (GME).[13] In a clinical setting, the diagnosis of encephalitis in dogs is still mainly based on imaging characteristics (CT, MRI), CSF analysis results, or both. These diagnostic modalities, however, are known to have limited sensitivity and specificity in the detection of inflammatory brain lesions.[14-19] For a more specific treatment in the individual dog, a definite diagnosis is desirable. In addition, the implementation of routine MIBB in dogs to achieve a confirmed diagnosis of encephalitis would allow the reliable assessment of different treatment options.

Therefore, the objectives of this study were to investigate the diagnostic yield determined as the percentage of dogs for which biopsies allowed a specific histological diagnosis to be established and to determine the adverse effects of MIBBs in dogs with encephalitis.

Materials and Methods


All dogs included in this study were presented to the Neurology Service of the Department of Small Animal Medicine of the University of Leipzig between January 2007 and December 2010 for neurological signs of intracranial disease. They underwent brain biopsies if brain MRI alone or in combination with cerebrospinal fluid (CSF) analysis was suggestive of an inflammatory brain disease, based on the criteria mentioned below, and if the owners consented explicitly to the diagnostic procedure after being informed of the possible risks.

The MRI was performed using a 0.5-T superconducting magnet.1 The following MR imaging sequences were obtained in all dogs: T2-weighted transverse view, fluid attenuation inversion recovery (FLAIR) transverse view, and T1-weighted transverse view before and immediately after an intravenous bolus of 0.1 mmol/kg gadopentetate dimeglumine.2 In addition, every dog underwent at least 1 sagittal scan, either T2-weighted or T1-weighted post contrast. A turbo spin echo inversion recovery sequence transverse view was obtained in 15/17 dogs. MR images were considered suspicious for encephalitis if they displayed multifocal or diffuse regions of T2 or FLAIR hyperintensity. Focal intraaxial lesions with homogenous or ring-like contrast enhancement were biopsied if the CSF analysis after the cisternal puncture revealed the following findings suggestive of encephalitis: nucleated cell count >5/μL or a positive Pandy reaction (semiquantitative immunoglobulin test).

A brain biopsy was performed only if at least 1 lesion was located in the forebrain. Those dogs in a comatose or stuporous state were excluded because of an increased procedural risk.

Anesthesia for the brain MRI and CSF tap was performed using diazepam3 (0.5 mg/kg) and levomethadone4 (0.5 mg/kg) IV as the premedication, followed by induction using propofol5 (2–4 mg/kg) IV. Dogs were intubated, and anesthesia was maintained using 1–2% isoflurane6 in 100% oxygen.

Biopsy Procedure

The biopsy was performed after diagnostics using the same anesthesia in 5/17 dogs, whereas in the other dogs, the biopsy was performed a mean of 2.6 days (range: 1–10) after the diagnostics. In the latter, the same anesthetic protocol was used as for the diagnostics. In addition, the following medication was given before every biopsy: methylprednisolone sodium succinate7 (10 mg/kg) and amoxicillin (9.7 mg/kg) plus clavulanic acid8 (1.9 mg/kg) given IV. Dogs were positioned in sternal recumbency with the head slightly elevated. The head was prepared and approached similar to a routine rostrotentorial approach to the brain.[20] Once the lateral skull was exposed, the biopsy site was identified based on the presurgical measurements from the MRI images. The following coordinates were obtained: 1. rostral distance from the occipital protuberance taken from midsagittal images; 2. lateral deviation from the dorsal midline taken from transverse images; and 3. depth as the distance between the skull surface and target point plus 5 mm attributable to the biopsy needle confirmation. The entry point and biopsy trajectory were planned considering the following criteria: shortest distance between the skull and target point, entry point in the middle of the gyrus avoiding the sulci, avoidance of the sinuses, and ventricular system. The angle of the needle trajectory was measured on the MRI, and the needle was accordingly positioned. Whenever possible, an angle perpendicular to the skull or to the surgery table or a 45° angle to the skull or table was used.

For tissue collection, a 4-mm opening was drilled into the skull using a pneumatic drill with a round burr head. The dura mater was incised with a number 11 scalpel blade. A closed blunt Sedan side-cutting biopsy needle9 with an outer diameter of 2.5 mm and a lateral opening of 10 mm was inserted up to the calculated depth. The side window of the needle was opened once the correct position was achieved, and negative pressure was applied via 0.5-mL suction using a syringe that was attached to the outer end of the needle. Then, the inner part of the biopsy needle was twisted 180° against the outer part, and the needle was removed from the brain. The core biopsy specimen was removed from the needle by flushing with a physiological saline solution. The procedure was repeated 1–2 times, depending on the material yield obtained. The biopsy specimen was fixed in 4% formaldehyde, and the surgical opening of muscle and skin was closed in a routine fashion.

In 12/17 dogs, a control MRI was performed using the sequence where the initial lesion was best visualized to confirm that the biopsy was performed correctly and to check for major hemorrhage. In cases where the lesion was not biopsied in the correct way, another single biopsy specimen was taken as described before considering the deviation of the biopsy from the anticipated trajectory (n = 3). In 5/17 dogs, a MRI after biopsy could not be performed because of limited availability of the MRI machine.


All brain biopsies were examined by a single investigator (KM). In 3 cases, necropsy results of the entire brain were available for direct comparison (AO, KM). The postmortem procedures followed routine protocols. The descriptions below focus on the biopsy specimen processing.

Upon arrival at the laboratory, the fixed biopsy specimen cylinders were split lengthwise (along the trajectory axis) with a razor blade to provide an insight into the laminar architecture of the cerebral cortex and the white–gray transition. The trimmed samples underwent routine central nervous system (CNS) processing using an automatic tissue processor, followed by paraffin embedding. For each sample, serial sections were performed at 5-μm slice thickness starting at the convex surface in one half and the plane surface in the other half. Each 5th slide was taken for routine hematoxylin-eosin staining, whereas the other 4 serial sections were stored for special staining and immunohistochemistry. Those samples showing granulomatous inflammation underwent gram staining, periodic acid Schiff staining, Grocott staining, and Ziehl-Neelsen staining. Immunohistochemistry for leukocyte markers was performed to elucidate the composition of the monomorphic lymphoid infiltrates and evaluate the invasion of white versus gray matter in cases of diffuse parenchymal hypercellularity. Mononuclear infiltrates underwent phenotyping for the T-cell marker CD3 (anti-CD3 antibody, mouse monoclonal10); the B-cell markers CD79a (anti-CD79a, mouse monoclonal10), and CD20 (anti-CD20 antibody, rabbit polyclonal11); and the macrophage/histiocyte marker MAC 387 (anti-MAC 387, mouse monoclonal12), lysozyme (anti-lysozyme antibody, rabbit polyclonal13), or combinations. All tissues showing inflammatory changes were screened for the presence of Toxoplasma gondii (anti-Toxoplasma gondii antibody, goat polyclonal14); Neospora caninum (anti-Neospora caninum antibody, goat polyclonal14); and canine distemper virus (CDV) antigen (anti-CDV antibody, mouse monoclonal14). Immunolabeling was performed via an autostainer using polymere technology15 or via indirect immunohistochemistry using rabbit anti-goat secondary antibodies16 and a horseradish peroxidase (HRP)-diaminobenzidine visualization system and counterstaining with Mayer's hematoxylin.17 With the mentioned exceptions, all cases were investigated the same way.

All samples were investigated at a magnification ranging from ×25 to ×1000 on a light microscope equipped with a CCD camera.18 The histopathological evaluation involved common algorithms for brain biopsy reading supplemented with data on specific canine disease entities.[21, 22] The histopathological determinants for the distinction of GME, NLE, and NME were in accordance with the current literature.[23]

Hemorrhages were regarded as biopsy-related if they (1) were extraparenchymal or in artificial tissue clefts, covering the surfaces of the tissue only; (2) consisted of free red blood cells with a proportionate admixture of inconspicuous peripheral blood leukocytes; (3) did not intermingle with necrotic tissue and abnormal blood vessels; or (4) did not accompany erythrophagocytosis. In the absence of histological consensus criteria, the degree of hemorrhaging was graded empirically as mild (<25% of the total sectioned area), moderate (25–50% of the total sectioned area), or marked (>50% of the total sectioned area).

The diagnostic yield was determined as the percentage of dogs for which the biopsies allowed a specific histological diagnosis to be established.[24]

Follow-Up Examination

A neurological examination was performed before the biopsy procedure and daily after the biopsy until discharging the dog from the hospital. The following parameters were evaluated during each neurological examination: level of consciousness, behavior, gait for signs of paresis, ataxia or circling, conscious proprioception in all 4 limbs, menace response, palpebral reflex, corneal reflex, nasal sensation, and physiological and pathological nystagmus. The follow-up information was based on a recheck examination (n = 5) or owner phone call (n = 10). The mean follow-up time for all dogs was 416 days (range: 2–1328 days).



Fifty-one dogs met the inclusion criteria of suspected encephalitis with at least 1 forebrain lesion between January 2007 and December 2010. The owners agreed to perform the procedure in 17 of these dogs after a thorough introduction by the performing clinician (TF). The following breeds were included: Yorkshire Terrier (n = 4), Chihuahua (n = 4), French Bulldog (n = 3), Jack Russell Terrier (n = 1), Maltese (n = 1), mixed breed dog (n = 1), Papillon (n = 1), Pug (n = 1), and West Highland White Terrier (n = 1). There was a male to female ratio of 1.43 : 1. The mean age was 5.3 years (SD: 3.1, range: 0.7–11.1 years). The mean body weight was 6.1 kg (SD: 3.7; range: 2.1–14.0 kg).

CSF Findings

The results of the CSF analysis were available for 16/17 dogs. The mean nucleated cell count was 106 cells/μL (range: 1–387; normal: ≤5), with 4 dogs having a normal cell count. The mean total protein concentration was 0.49 g/L (range: 0.08–1.76; normal: ≤0.25), with 3 dogs having a normal protein concentration. All dogs with a normal total protein concentration also had a normal nucleated cell count. The Pandy reaction was abnormal, indicating an increased concentration of immunoglobulins, in 1/13 tested dogs.

MRI Findings

All dogs had a forebrain lesion according to the inclusion criteria. In addition, there were brain stem lesions in 4/17 dogs and cerebellar lesions in 3/17. Those lesions were classified according to their spatial distribution as focal in 7/17 dogs, multifocal in 6/17, and diffuse in 4/17. Most of the lesions were hyperintense in the T2 and FLAIR images and had a varying contrast enhancement pattern. Meningeal enhancement was seen in 12/17 dogs, focal homogenous parenchymal enhancement in 4/17, diffuse-patchy parenchymal enhancement in 12/17, and ring enhancement in 1/17.

Diagnostic Yield

Two biopsy specimens were obtained in 6 dogs, and 3 biopsies were taken from 11 dogs. The samples were obtained from the following localizations: frontal lobe (n = 5), piriform lobe (n = 1), parietal lobe (n = 6), temporal lobe (n = 3), and occipital lobe (n = 2).

Independent of any pathology, the size and preservation of the samples allowed for a broad identification of the tackled brain region (cerebrum; granular, motor, and mixed neocortex; subcortical white matter) in all 17 cases (Fig 1). Lesions were observed in all 17 samples, and a morphological diagnosis of encephalitis was achieved in 16/17 (94%) animals. One biopsy revealed rather nonspecific alterations in terms of astrocytosis, astrogliosis, microglial proliferation, and vasogenic edema, suggesting that the periphery of the lesion was sampled. Sixteen dogs (94%) presented with inflammatory changes. In 13 of those, the patterns of inflammation, vascular reaction, and by-standing tissue damage were archetypic for GME (6/17), necrotizing meningoencephalitis (NME; 2/17), and necrotizing leukoencephalitis (NLE; 5/17) (Fig 2). Astrocytosis in 2 NLE cases blurred the exclusion of gray matter involvement. The sparing of gray matter here was confirmed via immunohistochemistry. Two GME cases showed lymphoproliferative changes with monomorphic lymphoid infiltrates occupying the Virchow-Robin spaces of occasional blood vessels. Aside from the appearance of parenchymal lesions, the mixed CD3 and CD20/CD79a pattern favored an inflammatory nature.

Figure 1.

Histological appearance of brain biopsy samples. (A) Subgross picture of a brain biopsy taken from a NME-affected Chihuahua dog that even on low-power field examination presented with confluent basophilic areas (arrow) corresponding to cellular infiltrates. (B) Astroglial intranuclear inclusion body (arrow) in a noninflamed area of a brain with IHC-confirmed canine distemper virus infection. Hematoxylin-eosin staining; the scale bars reflect the final magnification.

Figure 2.

Typical parenchymatous brain lesions associated with granulomatous and necrotizing types of meningoencephalitis of unknown origin. (A) Epitheloid perivascular infiltrates (arrowhead) are a major distinctive feature in cases of GME. (B): In NME and NLE the inflammation causes tissue destruction as seen in this early-intermediate resorptive stage of NLE with tissue rarifaction (asterisk). Both, Hematoxylin-Eosin staining; scale bars reflecting the final magnification.

Intranuclear Cowdry type A inclusion bodies were seen in a few neuroglial cells of a brain biopsy specimen from a 1-year-old mongrel dog with nonspecific mild lymphohistiocytic infiltrates. The tentative diagnosis of canine distemper virus encephalitis was confirmed through immunohistochemistry. None of the other etiological tests revealed infectious microorganisms or viral antigens in any of the remaining tissues. Therefore, the overall diagnostic yield with regard to a specific histological diagnosis was 82%.

Another biopsy (1/17) of a French Bulldog showed diffuse inflammatory changes confined to the subcortical white matter with remarkable sparing of the adjacent inner isocortical laminae. Tissue necrosis was not evident in this case. Similar infiltrates without necrosis were noted in the brain samples of an 8-year-old female Chihuahua (1/17).

Acute seizure-related gray matter changes, such as neuronal excitotoxic necroses and postictal edema, were seen throughout in more than half of the biopsies, regardless of whether the primary damage was located in the white or gray matter. Mild (4/17) to moderate (2/17) biopsy-related hemorrhage was observed in 6 biopsy samples. Hemorrhage was not observed in all samples from an individual dog, suggesting minimal procedure-related damage. The changes described above were not observed in every single biopsy specimen of a given dog. Altogether, a specific histopathological pattern was observed in 30/45 (67%) of the samples.

The evaluation of the postbiopsy MRI, which was performed in 12/17 dogs, resulted in taking a 3rd biopsy specimen in 4 dogs because correct sampling was questionable. This finding would indicate an error rate of 33% on the 1st attempt based on MR imaging. The diagnostic yield in the subgroups of dogs having 2 or 3 biopsies was 100 and 72.7%, respectively. In 8/17 cases, a conclusive diagnosis could be drawn from each individual biopsy, including the 4 cases in which a 3rd biopsy specimen had been taken because a postoperative MRI could not confirm that the target was correctly biopsied.

In 3 dogs, a necropsy was available for comparison. The diagnosis of canine distemper encephalitis in a mixed breed dog and the diagnosis of NLE in a French Bulldog were confirmed, whereas necropsy revealed NME in a Chihuahua, in which the biopsy had shown inflammatory changes but not the necrotizing component of NME.

Clinical Adverse Effects

Case fatality rate as a direct consequence of brain puncture was 0%, but there was 1 dog that developed generalized tonic–clonic seizures after the biopsy. This dog died 2 days later because of suffocation after aspirating food during a seizure. Hence, considering this secondary effect, the case fatality rate was 6%. The mean survival time of the other 16 dogs was 488 days (SD: 375; range: 6–1519 days). Of those 17 dogs, 6 were still alive at the time of completing this manuscript.

The overall morbidity associated with the biopsy procedure was 29%. However, the neurological status 1 day after the procedure was the same as that before the biopsy in 11/17 dogs; 1 dog did neurologically better after the biopsy, whereas 5/17 showed a neurological deterioration after the biopsy. Postoperatively, these dogs developed the following signs one in each dog: stupor and nonambulatory tetraparesis (resolved within 3 days); nonambulatory hemiparesis (resolved within 14 days); seizures, reduced level of consciousness, and ambulatory tetraparesis (resolved within 4 days); mild generalized ataxia (resolved within 4 days); and unilateral loss of conscious proprioception (resolved within 3 days).

A control MRI was performed immediately after the procedure in 12/17 dogs. No significant abnormalities other than those expected after biopsy, such as a skull defect and evidence of a biopsy tract, were seen in 6/12 dogs. However, 1 dog had an epidural/subdural hematoma, 2 had a ventricular pneumocephalus within the lateral ventricles, 1 had an epidural/subdural pneumocephalus, and 3 had gas accumulation visible in the biopsy trajectory (Fig 3). The postbiopsy MRI did not, however, result in an additional intervention to correct any of those complications. The dogs with the hematoma and with gas within the ventricular system were neurologically unchanged after the biopsy compared with their status before the procedure.

Figure 3.

Post brain biopsy MR images with pathological findings. (A) French Bulldog, T1-weighted post contrast: pneumocephalus in the left lateral ventricle (white arrow); (B): Chihuahua, T2-weighted: edema along the biopsy tract (white arrow) and small subdural hemorrhage (white arrowhead).


The overall diagnostic yield of free-hand brain biopsies using a mini-burr hole (percentage of samples resulting in a diagnosis) in dogs with suspected encephalitis based on MRI and CSF analysis was 82%. The brain biopsies thereby allowed for the diagnosis of GME, NME, NLE, and canine distemper infection. In an additional 12% of cases, encephalitis was evident, but a specific diagnosis could not be achieved. The final diagnosis was later possible in 1 case by postmortem investigation.

The case fatality rate and morbidity of the biopsy procedure were 0 and 29%, respectively. The risk of the procedure is evident, but all neurological deficits resolved within 3–14 days. The macroscopic appearance of the biopsy tracts 6 days after the biopsy procedure is shown in Figure 4 and demonstrates some hemorrhage along the biopsy tract only. Therefore, the procedure is associated with a relatively low risk. Given these considerations, performing a biopsy in dogs with suspected encephalitis may lead to a more specific diagnosis in the individual dog and therefore result in more reliable information on survival times using different treatment protocols prospectively.

Figure 4.

Macroscopic appearance of the biopsy tract on external inspection (A) and after transverse sectioning (B). (A) A brownish discoloration is seen at the entry site of the trajectory in the right-sided ectomarginal gyrus (dashed circle) of this dog 6 days after surgery. (B) Two biopsies were taken through the same burr window at different angles (dashed arrows). Thus, the samples were harvested from the cortex and subcortical white matter/semioval center close to the dorsal entry of the internal capsule. Note the rather mild hemorrhages associated with the biopsy puncture.

Inflammatory changes were seen in 16 dogs (94%), but in 2 of 16 of cases (12%), those changes were unspecific and did not result in a definite etiopathogenic diagnosis. There are only limited data on brain biopsies in the veterinary literature available for comparison, and most focus on neoplastic lesions. A study investigating minimally invasive brain biopsies in 50 dogs, including 8 with encephalitis, found an overall diagnostic yield of 100%.[2] Another publication described a diagnostic yield of 95% in 23 dogs with mainly neoplastic lesions.[4] Those results are similar to data from human medicine, which have reported diagnostic yields of stereotactic brain biopsies in neoplastic lesions between 90 and 99%.[25, 26] The results obtained in groups of dogs or humans with predominantly neoplastic lesions are not suitable for comparison because they rarely present with the same distribution pattern and MR features. Whereas in neoplasias, a well-circumscribed lesion is quite often biopsied, clearly differentiating the primary inflammatory focus from secondary edema and seizure-related lesions on MR images might be impossible in encephalitis. Hence, the diagnostician has to address the obstacle that the inflammatory focus is blurred by secondary features and with the possibility that the area chosen for biopsy might not include the primary lesion in all cases.

Apart from the target area, the number of pieces retrieved during the biopsy seems to influence the diagnostic yield of the entire procedure. A conclusive pattern leading to a specific diagnosis was observed in only 67% of obtained pieces. On the basis of our results, at least 2 tissue samples should be taken via a slightly changed trajectory. There was no major difference between 2 and 3 samples with respect to the diagnostic yield of the entire intervention, mirrored by diagnostic yields of 100 and 73%, respectively. Three samples were usually taken only in cases for which the MRI after the biopsy procedure raised doubts that the biopsy needle had been in focus during the harvest of the first 2 samples, which was the case in 4/12 dogs. This observation indicates an error rate of 33% on the first attempt. However, a conclusive histopathological pattern could be identified in every single biopsy specimen of those 4 dogs, indicating that the 3rd sample would not have been necessary. Those findings underline the difficulties in identifying the ideal target point, especially in often poorly circumscribed inflammatory lesions because MRI changes might not reflect the histopathology in every case.[14]

A study in 86 human patients with mainly neoplastic brain lesions undergoing stereotactic brain biopsy nicely featured the diagnostic yield variations corresponding to the number of obtained samples. Correspondingly, the yield was 76.5% with 1 sample, 84.0% with 2 samples, 88.2% with 3 samples, and 100% with as many as 5 samples.[27] Interestingly, another study enrolling 270 humans showed that the procedural risk of multiple biopsies does not increase as long as the same biopsy trajectory is used for all attempts.[28]

The diagnostic accuracy (number of samples resulting in a correct and confirmed diagnosis, expressed as a %) could not be assessed in this study because only 3/17 dogs had a postmortem examination. Therefore, no conclusion can be drawn from the comparison between the biopsy and necropsy results in the study population. The diagnostic accuracy remains to be elucidated via adjunct postmortem investigations of the entire brain in the future. In contrast, a mean survival time of 503 days in dogs with a biopsy diagnosis of GME, NME or NLE upon immunosuppressive medication in combination with cephalexin for only 5 days after the biopsy leads to the assumption that an underlying infectious etiology had not been missed.

Veterinary data on the diagnostic accuracy of MIBBs in encephalitis are not broadly available, except for a series of 8 dogs with encephalitis that had a diagnostic accuracy of 100%.[2] The diagnostic accuracy of brain biopsies for mainly neoplastic lesions, however, varies in dogs between 25 and 91%.[2, 11] Comparable human studies, including mainly neoplastic and a smaller percentage of inflammatory lesions, report diagnostic accuracies between 90 and 97.7%.[29, 30]

The diagnostic yield and accuracy may vary depending on the method used to obtain the biopsies. Deep-seated brain lesions are generally approached using a minimally invasive approach to the brain. These procedures allow a lesion to be sampled while minimizing trauma to the surrounding healthy tissue. The disadvantages of MIBBs are a small sample size and limited control of correct sampling and procedure-related hemorrhage. Most MIBB systems used in veterinary medicine apply a rigid frame to the skull to guide the needle along the preplanned trajectory to assure correct sampling.[1-4, 6-8] Free-hand biopsies as described here cannot guaranty needle advancement along the anticipated trajectory, which might be in part responsible for those samples not resulting in a definite diagnosis. Another reason for nondiagnostic samples could be the intrinsic nature of the inflammatory lesions, in which the primary lesion can be difficult to be differentiated from secondary changes in the MR images.

In any case, the diagnostic yield could be increased if correct sampling is checked by immediate cytological evaluation of a smear obtained from the biopsy specimen.[2, 31] This option, however, was not available for the dogs described here.

In our study, the case fatality rate and morbidity, as defined clinically, were 6 and 29%, respectively. The death of 1 dog did not result directly from the biopsy procedure, but instead was caused by suffocation during a seizure event 2 days after the biopsy. Those seizures did develop after the biopsy procedure; however, whether they were induced by the biopsy or merely reflected the progression of the encephalitis itself remains unclear.

All surviving dogs that neurologically deteriorated after the biopsy returned to the status before biopsy within 3–14 days (median 4 days). Likewise, the case fatality rate and morbidity seemed to be 0% in the 8 dogs with encephalitis of a series of 50 dogs with different brain diseases.[2] Two studies in dogs with predominately neoplastic lesions concordantly reported a procedural case fatality rate of 8% and a morbidity ranging between 12 and 27%.[2, 4] In most of those dogs with neoplastic diseases, the biopsies were taken from a brain with a focal lesion with or without secondary changes, but with the majority of the brain not being affected by the primary disease process. In contrast, encephalitic changes and inflammatory edema mostly appear far more widespread and could therefore potentially affect the brain in a more generalized manner. This spatial pattern might render the brain less resistant to biopsy-induced trauma. Therefore, the morbidity and case fatality rate could potentially be higher in inflammatory than in neoplastic lesions if assessed in a larger collection of affected animals. In addition to the more generalized nature of inflammatory lesions, the use of a free-hand biopsy technique may have added to the temporary morbidity in the dogs presented here. Minor deviation from the anticipated trajectory may, for example, explain the inadvertent penetration of the ventricular system, causing pneumocephalus in 2 dogs.

When evaluating the intrinsic procedural risk against the diagnostic benefit, the fact that all the obtained data of this study were based on biopsies taken from the forebrain should be considered. Sampling from the brain stem or cerebellum in dogs with encephalitis likely will result in a less optimistic/positive outcome. A human study involving 100 patients found a morbidity of 1.1% in nonbrain stem lesions, whereas the morbidity was 25% if samples were harvested from the brain stem.[24] Half of those site-related neurological deficits were permanent.

In conclusion, free-hand biopsies of forebrain lesions using a mini-burr hole in dogs with suspected encephalitis is a relatively safe procedure resulting in a specific histological diagnosis in the majority of dogs (82%). Therefore, biopsy can and should be recommended because developing new specifically tailored therapies based on a definite diagnosis potentially in association with histological or molecular markers may lead toward longer survival times in the future.


The authors thank Mrs Karin Stingl for her excellent technical support regarding the immunohistochemistry.


  1. 1

    Gyroscan NT Compact Plus, Philips Medical Systems, Eindhoven, The Netherlands

  2. 2

    Magnevist, Bayer Health Care, Leverkusen, Germany

  3. 3

    Faustan, Temmler Pharma GmbH Co. KG, Marburg, Germany

  4. 4

    Polamivet, Intervet Deutschland GmbH, Unterschleißheim, Germany

  5. 5

    Narcofol, CP-Pharma Handelsgesellschaft mbH, Burgdorf, Germany

  6. 6

    Isufluran CP, CP-Pharma Handelsgesellschaft mbH

  7. 7

    Prednisolut, Milbe GmbH Arzneimittel, Brehna, Germany

  8. 8

    AmoxiClav Hikma, Hikma Pharma GmbH, Gräfelfing, Germany

  9. 9

    Elekta Instruments AB, Stockholm, Sweden

  10. 10

    Dakocytomation, Glostrup, Denmark

  11. 11

    Labvision, Fremont, CA

  12. 12

    Abcam PLC, Cambridge, UK

  13. 13

    Cellmarque Co, Rocklin, CA

  14. 14

    VMRD Inc, Pullman, WA

  15. 15


  16. 16

    Abcam PLC

  17. 17

    Histoprime, Linaris GmbH, Wertheim-Bettingen, Germany

  18. 18

    Zeiss Axiophot, Carl Zeiss Microimaging GmbH, Göttingen, Germany