• Open Access

Cerebrospinal Fluid PCR and Antibody Concentrations against Anaplasma phagocytophilum and Borrelia burgdorferi sensu lato in Dogs with Neurological Signs


  • This study was performed in Uppsala, Sweden, at the Swedish University of Agricultural Sciences and the National Veterinary Institute.

Corresponding author: Karin Hultin Jäderlund, DVM, DECVN, Department of Clinical Sciences, Swedish University of Agricultural Sciences, PO Box 7054, SE-750 07 Uppsala, Sweden; e-mail: karin.hultin-jaderlund@kv.slu.se.


Background: The tick-borne bacteria Borrelia burgdorferi sensu lato (sl) and Anaplasma phagocytophilum have been suspected to cause neurological signs in dogs. Diagnosis often has been made based on positive antibody titers in serum of dogs with neurological signs, but a high seroprevalence in dogs in at-risk populations makes diagnosis difficult.

Objective: To determine if the neurological signs in dogs examined were caused by any of these bacteria.

Animals: Fifty-four dogs presented to a board-certified neurologist.

Methods: Prospective study. We divided dogs into 2 groups: those with inflammatory diseases of the central nervous system (CNS) and those with neurological signs from other diseases. Blood and cerebrospinal fluid (CSF) from all dogs were analyzed.

Results: Dogs with inflammatory CNS diseases showed no serum antibodies against any of the agents. Among dogs with neurological signs from other diseases, 10.3% had serum antibodies for B. burgdorferi sl and 20.5% for A. phagocytophilum. All blood samples analyzed for bacterial deoxyribonucleic acid (DNA) and all CSF analyzed for antibodies and bacterial DNA for the 2 agents were negative.

Conclusions and Clinical Importance: Based on this study, these bacteria are unlikely causes of neurologic disease in dogs and the presence of serum antibodies alone does not document or establish a definitive diagnosis of CNS disease caused by these organisms. Dogs that have neurologic disease and corresponding serum antibodies against these agents should have additional tests performed to assess for other potential etiologies of the signs.

A. phagocytophilum

Anaplasma phagocytophilum

B. burgdorferi sl

Borrelia burgdorferi sensu lato


central nervous system


cerebrospinal fluid


deoxyribonucleic acid


immunofluorescence assay


magnetic resonance imaging


polymerase chain reaction

Both borreliosis and granulocytic anaplasmosis are tick-transmitted bacterial diseases that may clinically affect dogs, causing fever and lameness.1,2 Extrapolating from human medicine,3,4 these diseases also have been considered to cause neurological signs in dogs.5–8 The association between borreliosis and neurological signs often has been based entirely on seropositivity.5,7 High seroprevalence in the population at risk makes a diagnosis based on serology questionable.9 In Sweden, seroprevalence for a closely related group of Borrelia species (B. afzelii, B. garinii, and B. burgdorferi sensu stricto), herein called B. burgdorferi sensu lato (sl), and Anaplasma phagocytophilum was established in a canine reference population during 1991–1994.10 In a retrospective study of 248 neurologically affected dogs, it was also shown that in almost all seropositive cases, etiologic factors other than these bacteria were found.9 Only in a few cases could symptomatic infections with B. burgdorferi sl or A. phagocytophilum not be excluded. In that study, a apolymerase chain reaction (PCR) technique for detection of bacterial deoxyribonucleic acid (DNA) in blood or cerebrospinal fluid (CSF) was not available, and CSF antibodies were not routinely examined.9 The probability of diagnosing symptomatic nervous system infections most likely would have increased by such evaluations.2,3

The aim of the present prospective study was to determine if the neurological signs in dogs examined were caused by an infection with B. burgdorferi sl or A. phagocytophilum.

Materials and Methods

The study population was selected from dogs admitted during 2002–2005 to the University Clinic at the Swedish University of Agricultural Sciences, Uppsala, Sweden for neurological examination by one of the investigators (KHJ). Dogs that had a CSF tap performed were included. Dogs lived in an area where tick infestations occur in almost all dogs every season. None of the dogs were vaccinated against the agents under study. We recorded breed, sex, age at examination, and whether or not the dog had been referred or was a primary case at the University Clinic. Diagnostics included spinal radiography, myelography, magnetic resonance imaging (MRI), scintigraphy, hematology, serum chemistry, CSF analysis (cytology, cell counts, protein content, glucose content, and culture), and necropsy. Based on the diagnostic results, the dogs were divided into 2 groups; group 1 with primary inflammatory diseases of the central nervous system (CNS), and group 2 with neurological signs caused by other diseases.

CSF was collected from the cerebellomedullary cistern. Blood and CSF were analyzed for the presence of bacterial DNA by the PCR technique and for the presence of antibodies by the immunofluorescence assay (IFA) at the National Veterinary Institute (Uppsala, Sweden) according to their standard methods.9 For the PCR technique, we extracted DNA from 200 μL EDTA blood samples or CSF with the EZ1 DNA tissue kit in the Biorobot EZ1 with the EZ1 DNA bacterial card,a according to the manufacturer's instructions. To prevent contamination when performing PCR analysis, preparation of reaction mixtures, DNA extraction, amplification, and detection of PCR products all were performed in different laboratory rooms. Additionally, aerosol-resistant filter pipette tips were used throughout all experiments. PCR for A. phagocytophilum was performed as described by Goodman et al,11 with the primers described in the erratum of that paper, but converted to a real-time PCR system by the addition of a Taqman probe (5′-6-AMd(CTGTCGTCAGCTCGTGTCGTGAGATGTTG)BHQ-1-3′). PCR for B. burgdorferi sl was performed according to Schwaiger et al12 but substituting the forward primer 5′-AGGTGCTTTCCAAAATAGACTTG-3′ to have a better match with flagellin gene sequences deposited in GenBank after the original PCR method was published.

Because of the introduction of borrelial PCR technique in the lab after the sampling was already in process, the sampled volume was insufficient to perform both tests in some dogs. Antibodies in CSF were analyzed in 71% of dogs in group 1 and in 45% of dogs in group 2, whereas CSF PCR was carried out in all cases. In blood, PCR for A. phagocytophilum was done in all dogs, whereas PCR for B. burgdorferi sl was done in 93% of dogs in group 1 and in 85% of dogs in group 2. Serum antibodies for both agents were analyzed in all but 1 dog. In dogs with CSF pleocytosis, granulocytes were inspected for detection of A. phagocytophilum inclusion bodies.


Fifty-four dogs were included (32 males and 22 females). Median age was 3 years (range, 6 months–11 years). Thirty different breeds were represented, with German Shepherds the most common (n = 6). The numbers of referred and primary cases were 29 and 25, respectively. Seventeen dogs had been treated with antibiotics with no response: enrofloxacin was given to 8, amoxicillin to 5, cephalexin to 2, and doxycycline to 2. Fourteen dogs were classified as group 1, whereas 40 were classified as group 2. Dogs in group 1 had a median age of 1 year (range, 7 months–6 years); dogs in group 2 had a median age of 5 years (range, 6 months–11 years).

Group 1 consisted of 10 dogs diagnosed with steroid-responsive meningitis-arteritis, and 4 dogs with other miscellaneous inflammatory CNS diseases. In group 2 (n = 40), diagnoses consisted of degenerative diseases of the spine (associated with intervertebral disks or ligaments) (n = 17), neoplastic diseases (n = 6), arachnoid cyst (n = 3), back pain of unknown etiology (n = 3), behavioral problems (n = 2), orthostatic tremor (n = 2), degenerative myelopathy (n = 1), spinal cord hamartoma (n = 1), head shaking (n = 1), springer spaniel rage syndrome (n = 1), fibrous osteodystrophy (n = 1), chorioretinitis (n = 1), and left forebrain lesion of unknown etiology (n = 1).

In group 1, seroprevalence for both agents was 0% (n = 14). In group 2, 4/39 dogs (10.3%) were seropositive for B. burgdorferi sl, and 8/39 dogs (20.5%) for A. phagocytophilum. The 4 dogs seropositive for B. burgdorferi were diagnosed with disk herniation (n = 2), neoplasia (n = 1), and behavioral problem (n = 1). The 8 dogs seropositive for A. phagocytophilum were diagnosed with disk herniation (n = 4), neoplasia (n = 2), caudal cervical spondylomyelopathy (n = 1), and left forebrain lesion of unknown etiology (n = 1). Serum titers ranged from 1 : 80 to 1 : 640 for both bacteria. All blood samples (n = 54) analyzed with PCR were negative. Also, all CSF samples analyzed for antibodies (n = 28) and by PCR (n = 54) for the 2 agents were negative. CSF from 12 dogs in group 1 and 4 dogs in group 2 had pleocytosis (> 5 leukocytes/μL). No A. phagocytophilum inclusion bodies were detected in their CSF granulocytes.

Nineteen dogs were euthanized. Two dogs from group 1 and 15 dogs from group 2 were necropsied. No indications of tick-transmitted nervous system infections were found on postmortem examination. Three of these dogs were seropositive for A. phagocytophilum, 2 with neoplasia (1 with a meningeal fibrosarcoma and 1 with a CNS adenocarcinoma) and 1 with a disk herniation. The dog with meningeal fibrosarcoma also was seropositive for Borrelia by IFA.


The dogs in this study were divided into 2 groups, 1 with inflammatory CNS diseases, and 1 with neurological signs from other diseases. The reason for this subdivision was our belief that tick-borne bacterial infections would more likely cause neurologic disease in the inflammatory group than in the group with neurological signs from other diseases. A routine CSF analysis was carried out in all dogs because examination of CSF is fundamental when determining the diagnosis in cases with inflammatory diseases of the CNS, but other variables from the diagnostic work-up also were considered in the assignment of dogs to groups. Two dogs in group 1 eventually were diagnosed by postmortem examination.

As expected, serum antibodies for the 2 agents were found in many dogs. The seroprevalence for the bacteria in these dogs was not different from that reported in other studies in dogs performed in the same region, where seroprevalence of 5–9% for B. burgdorferi sl and 21% for A. phagocytophilum has been reported.9,10 Actually, the group of dogs with inflammatory CNS diseases had a seroprevalence of 0% for both agents. One contributing factor to the seroprevalence in group 1 may have been the low median age. Previous studies have shown that the seroprevalence for these organisms in dogs tends to increase with age in Sweden.9,10 All dogs with serum antibodies against B. burgdorferi sl and A. phagocytophilum in our study were diagnosed with diseases other than inflammatory disease of the CNS. All CSF samples examined in both groups were negative for B. burgdorferi sl and A. phagocytophilum antibodies and by PCR. These findings suggest that none of the dogs had, or recently had had, symptomatic or subclinical infections of the CNS involving any of these bacteria.

The dogs in this study lived in a region of population at risk where the 2 tick-borne agents have been detected and also have caused other, nonneurological, clinical signs in dogs.2,10 In studies of dogs experimentally infected with B. burgdorferi or A. phagocytophilum, all dogs seroconverted, and dogs that developed clinical signs were seropositive from the beginning of the disease.1,2 Seropositivity for B. burgdorferi lasted for >1 year,1 whereas seropositivity for A. phagocytophilum lasted for at least 6 months.2 Because no dogs included in our study were vaccinated against the bacteria studied, positive titers were thought to be from natural exposure to the agents. Antibodies in CSF, whenever present, could be because of intrathecal production in response to the presence of a provoking agent, or to leakage across the blood-CSF barrier. Our results indicate that some dogs in the present study had been exposed to the agents from weeks to years preceding the sampling procedure, with no detectable concomitant infection of the CNS. In the seropositive dogs that had serum titers of up to 1 : 640, not even leakage of specific antibodies across the blood-CSF barrier could be detected.

A. phagocytophilum is a strictly intracellular bacterium residing in granulocytes of the blood, and occasionally inclusion bodies may be seen. Compared with the search for inclusion bodies, PCR technique on blood is an even more sensitive diagnostic method during active infection.2 Occasionally, positive PCR results have been found several months after experimental infections,2 indicating that chronic longstanding infections may occur. This PCR method has not been evaluated previously for use in CSF from Swedish dogs, but the presence of bacterial DNA probably could be detected in CSF. This method does not, however, differentiate between bacterial DNA from dead and living organisms.3 The antibiotic of choice for treating A. phagocytophilum, doxycycline, possibly may influence results of PCR. Doxycycline was used in only 2 dogs, both of which were seronegative and had negative PCR results. If infection of the CNS occurs, granulocytes with bacteria and bacterial DNA present in CSF, as well as presence of intrathecally produced antibodies, would be expected. Shaw et al,8 have reported an association between unspecified meningitis in 4 dogs and granulocytic anaplasmosis, as confirmed by PCR. Our study could not confirm such an association, nor could this finding be confirmed in our previously reported retrospective study based on antibody test results, which included 30 dogs with inflammatory CNS diseases.9

Because B. burgdorferi sl bacteria migrate in tissue and are not normally found in blood, PCR technique for that bacterium in blood is considered of low diagnostic value. However, it may be possible to detect migrating bacteria in the CNS in canine CSF. The PCR test used in this study previously has not been evaluated by CSF samples. The reason we were unable to analyze both PCR and antibody titers for all dogs was because of the fact that borrelial PCR only became available after initiation of the study. Also, the volume of CSF or blood collected in some dogs was not sufficient to perform both tests. In those cases, PCR of CSF was deemed a more important test. However, in blood we preferred the serological methods.

In humans with neuroborreliosis, neurological signs most commonly are associated with meningitis, encephalitis, or peripheral neuropathies. A powerful diagnostic aid in human cases is the presence of pleocytosis in CSF, together with intrathecally-produced specific antibodies.3 In addition, sensitivity of PCR in CSF has ranged from 12 to 100% in different reports, and high specificity has been reported.3 A few canine cases with Borrelia antibodies in CSF have been reported,6,13 but in no case was there any described concomitant pleocytosis. In addition, in dogs experimentally infected with B. burgdorferi sl, symptomatic neuroborreliosis has not been reported.1

In conclusion, B. burgdorferi sl and A. phagocytophilum appear to be rare causes of neurologic disease in dogs. A positive serum antibody titer against these agents does not prove the presence of active CNS disease. A definitive diagnosis of CNS infection with B. burgdorferi sl or A. phagocytophilum should not be made in dogs that are presented with neurologic signs solely on the presence of serum antibodies against B. burgdorferi sl or A. phagocytophilum. In this situation, other tests always should be performed to determine if other etiologies may be the cause of the signs.


aQiagen, Hilden, Germany


The authors wish to thank the Agria Insurance Foundation for Research for financial support, and Dr Helena Rylander for her critical appraisal of the manuscript.