• Open Access

Isolation or Molecular Detection of Bartonella henselae and Bartonella vinsonii subsp. berkhoffii from Dogs with Idiopathic Cavitary Effusions


Corresponding author: Dr Edward B. Breitschwerdt, Intracellular Pathogens Research Laboratory and the Veterinary Teaching Hospital, Department of Clinical Sciences, College of Veterinary Medicine, North Carolina State University, 4700 Hillsborough Street, Raleigh, NC 27606; e-mail: ed_breitschwerdt@ncsu.edu.

There are a substantial number of pathophysiologic causes for effusions in dogs.1 Previously, using an insect cell culture medium (Bartonella alpha-proteobacteria growth medium—BAPGM), our laboratory isolated Mycobacterium kansasii from a dog with therapeutically intractable pleural effusion.2 Recently, we developed a combined assay incorporating BAPGM pre-enrichment culture, so as to increase Bartonella bacterial numbers, followed by PCR amplification of organism-specific DNA sequences.3 The combined assay has substantially increased the sensitivity of molecular detection of Bartonella infection.3–5 This pre-enrichment culture medium has also isolated numerous other bacterial species of undetermined pathogenicity from human patients in our laboratory.4

In the past decade, several Bartonella species, including B. henselae, B. vinsonii subsp. berkhoffii, B. clarridgeiae, B. elizabethae, B. washoensis, and B. quintana, have been found to infect dogs.6 Currently, case-based evidence suggests that occult infection with these bacteria may contribute to a wide range of disease manifestations in dogs, humans, and potentially other animals.6B. henselae and B. vinsonii subsp. berkhoffii have been reported in association with granulomatous lymphadenitis, hepatic disease,7 and endocarditis in dogs.8 Between July 2004 and December 2007, blood and effusion samples from 20 dogs of varying breeds and ages, were submitted for BAPGM culture by veterinarians at Auburn University, North Carolina State University and Tufts University's Veterinary Teaching Hospitals, of which 5 were found to be infected with a Bartonella spp. Using a previously described approach, DNA was extracted directly from EDTA-anticoagulated blood or thoracic and abdominal effusion samples, from BAPGM pre-enrichment liquid blood and effusion cultures, and from pooled bacterial subculture isolates.3–5,9 Extracted DNA was screened by PCR with Bartonella genus primers, targeting the 16S-23S intergenic spacer region (ITS), which has a detection limit of 2–3 genomic copies per reaction.9 ITS amplicons were sequenced to confirm the Bartonella species and strain.5,9 This study provides the first molecular and microbiologic evidence to support infection with B. henselae and B. vinsonii subsp. berkhoffii in dogs with idiopathic cavitary effusions or constrictive pericarditis.

Case Reports

Case 1

An 8-year-old-female spayed Labrador mix was referred to Tufts for evaluation of recurrent pleural effusion. Thoracic fluid cytology was consistent with a transudate with a specific gravity of 1.019, total protein of 2.6 g/dL and 255 cells/μL. No bacterial growth or neoplastic cells were documented. A CBC, serum biochemical profile, coagulation profile, and urinalysis were unremarkable except for mild hyperbilirubinemia and hematuria. A torsed lung lobe was surgically removed. Postoperatively despite administration of azathioprine, prednisone, and spironolactone, thoracocentesis was required for 10 consecutive months. When immunosuppressive therapy was tapered, pleural effusion rapidly recurred. Antibiotics, including doxycycline, azithromycin, or enrofloxacin, did not decrease the rate of fluid accumulation, which had become a modified transudate. The dog underwent an exploratory thoracotomy and died postoperatively. B. henselae strain Houston-I was isolated and sequenced from the thoracic fluid and B. henselae strain Houston-I and B. vinsonii subsp. berkhoffii genotype II were isolated from a BAPGM blood culture. Bartonella serology was not requested.

Case 2

A 3-year-old-female spayed Leonberger was referred to Auburn for evaluation of lethargy, anorexia, weight loss (11 kg), and cervical pain. The only central nervous system abnormality was mild pain on cervical flexion. A possible abdominal mass was noted. There were no hematologic abnormalities. Serum biochemical abnormalities included increased alkaline phosphatase (130 U/L; reference range, 4–95) and alanine amino transferase activities (385 U/L; reference range, 26–200), increased serum creatinine concentration (1.5 mg/dL; reference range, 0.68–1.45), and hypoglobulinemia (2.1 g/dL; reference range, 2.6–5.0). Urinalysis was within normal limits and urine culture failed to grow bacteria. Thoracic radiographs indicated moderate pleural effusion. Ultrasonographic abnormalities included a thickened, stiff stomach wall, abdominal effusion, and irregular kidney margins. An echocardiogram identified no cardiac abnormalities. Pleural fluid analysis was consistent with a modified transudate with a specific gravity of 1.026, total protein of 3.9 g/dL, and 670 total cells/μL. Serum antibodies were not detected to Neospora caninum, Ehrlichia canis, Rickettsia rickettsii, Toxoplasma gondii, canine distemper virus, and Pythium insidiosum. Abdominal exploratory surgery identified a large volume of grossly pink peritoneal fluid. No histologic lesions were found in stomach, liver, omentum, duodenum, mesenteric, and pyloric lymph nodes, kidney, pancreas, and jejunum. B. henselae and B. vinsonii subsp. berkhoffii antibodies were not detected, but B. vinsonii subsp. berkhoffii genotype II was sequenced from the BAPGM pre-enrichment blood culture and the subculture isolate. Azithromycin 250 mg daily and doxycycline 100 mg PO q12 h were administered for 14 days. Nine days after beginning treatment, the dog was eating and drinking normally but pleural effusion failed to resolve. Antibiotics were continued for 2 more weeks. During the next year, the cervical pain continued. The dog was euthanized because of severe peripheral edema with spontaneous fluid extravasation 2 years later.

Case 3

An insulin-dependent diabetic 12-year-old-male castrated Vizsla was evaluated for coughing, gagging, and lethargy. Idiopathic chylothorax initially was diagnosed by the referring veterinarian based on cytological evaluation, which was confirmed by pleural fluid analysis at NCSU-VTH as chylous with a specific gravity of 1.023, total protein of 3.3 g/dL, and 1,940 total cells/μL. A CBC was normal. Biochemical abnormalities included hyperglycemia (serum glucose concentration 258 g/dL; reference range, 73–116), and increased SAP (270 U/L; reference range, 15–156), and ALT activities (100 U/L; reference range, 16–73). Abdominal and thoracic ultrasonography, an echocardiogram, and a helical computed tomography (CT) scan of the thorax and anterior abdomen with lymphangiography were unremarkable, except for pleural effusion. Thoracic duct ligation and partial pericardectomy were performed. The dog recovered uneventfully but continued to accumulate lesser quantities of thoracic fluid, consistently characterized as a modified transudate. Two months postsurgery, the dog developed lethargy, inability to rise, and edema involving the neck, sternum, and right front leg. Urine and pleural fluid cultures failed to grow bacteria. Pleural fluid again was classified as a modified transudate, containing atypical mesothelial cells. A seroma, accompanied by spontaneous leakage of fluid, formed in the ventral neck region. The owner elected euthanasia before availability of the BAPGM culture results. B. henselae DNA was amplified and sequenced directly from the pleural fluid and from the BAPGM enrichment culture. After subculture, an Arthrobacter spp. was isolated, as defined by sequencing the 16S rRNA gene. Serology was not requested. At necropsy, no cause was identified for the severe emaciation, pleural effusion, and subcutaneous edema.

Case 4

A 5-year-old-female spayed Pomeranian was referred to the NCSU-VTH for diagnostic evaluation of panhypoproteinemia. The owners reported infrequent bouts of diarrhea and poorly localized abdominal pain. Serum total protein concentration (2.6 g/dL; reference range, 5.1–7.4), albumin (1.3 g/dL; reference range, 2.8–4.0), and globulin concentrations (1.3 g/dL; reference range, 2.0–4.1) were low. Hematologic abnormalities included lymphopenia (283 lymphocytes/μL; reference range, 480–3,762) and neutrophilia (13,745 neutrophils/μL; reference range, 2,529–12,876). With the exception of bilirubinuria, urinalysis was unremarkable, urine protein/creatinine ratio was normal, and urine culture was negative. Radiographs confirmed thoracic and abdominal effusion. Ultrasonography identified bicavitary anechoic effusion, consistent with a transudate, a collapsed right lung lobe, and portal vein thrombosis. The liver was hypoechoic, abdominal lymph nodes slightly enlarged, and the intestinal lumen was distended with fluid. Echocardiography did not identify pericardial effusion or support a diagnosis of right-sided heart failure. By abdominocentesis, a clear transudative fluid with a specific gravity of 1.005 was obtained. Aerobic and anaerobic culture of the abdominal fluid failed to grow bacteria. Endoscopic gastric biopsies were unremarkable, whereas the duodenum was moderately infiltrated with lymphocytes and plasma cells, accompanied by multifocal lymphangiectasia. A heartworm antigen test was negative. Antibodies to B. henselae and B. vinsonii subsp. berkhoffii, R. rickettsii, E. canis, and Babesia canis were not detected by IFA testing. B. vinsonii subsp. berkhoffii genotype II was isolated from pleural fluid and B. henselae from the blood. Protein-losing enteropathy, accompanied by a portal vein thrombosis, was diagnosed. Treatment consisted of a high protein, low residue diet and metronidazole 10.5 mg/kg PO q8h for 4 weeks. Within 2 weeks, the portal vein thrombosis was no longer visible by ultrasonography. Serum protein concentrations increased progressively and were within reference ranges (total protein, 5.8 g/dL; albumin, 3.6 g/dL; globulin, 2.2 g/dL) by 3 months after initial evaluation. Effusion resolved and did not recur during a 3-year follow-up period.

Case 5

A 6-year-old-female spayed Labrador Retriever was referred for diagnostic evaluation of abdominal effusion. Historically, the dog had remained alert and active. Six liters of serosanguineous, modified transudate with a specific gravity of 1.026, total protein of 4 g/dL and 770 total cells/μL was aspirated from the abdomen. There were no hematologic or coagulation abnormalities, except hypocholesterolemia (134 mg/dL; reference range, 138–317). Central venous pressure was slightly increased (8–9 mmHg). Abdominal and thoracic CT scans identified distention of the hepatic veins and the cranial and caudal vena cava, indicative of right-sided heart failure. By echocardiography, minimal pericardial effusion and hypokinetic left ventricle free wall were documented. Catheterization of the right side of the heart detected a cranial vena cava pressure of 16.1 mmHg, right atrial pressure of 17.4 mmHg, right ventricle pressure at the end of the diastole of 15.6 mmHg and wedge pressure of 17.4 mmHg. The tracing of the right atrial pressure, with prominent x and y descents, was consistent with constrictive pericarditis. Subtotal pericardectomy was performed. On histopathology, the pericardium was variably thickened (up to 0.5 cm) with densely packed collagen fibers, containing few blood vessels and rare hemosiderin-laden macrophages. No etiologic agents were seen, no fungal species were isolated and aerobic and anaerobic blood cultures were negative. On day 8 postpericardectomy, the dog became febrile (T 104°F), vomited, and had diarrhea. A CBC was unremarkable. Two liters of serosanguineous fluid, removed by thoracocentesis, was classified cytologically as an inflammatory sterile effusion containing neutrophils, some phagocytosed by macrophages, and large, atypical mesothelial cells. Anaerobic and aerobic cultures again were negative, potentially because amoxicillin-clavulanate was started by the referring veterinarian 36 hours before sample collection. The treatment regimen was changed to ciprofloxacin, azithromycin, and metoclopramide. Pleural fluid, cultured in BAPGM, resulted in the isolation of Bartonella spp., which was not successfully sequenced. Serum was not available for Bartonella IFA testing. Once culture results became available, treatment with azithromycin (7.8 mg/kg PO q2h) was begun for 5 weeks. At reevaluation 1 month later, the dog was eating normally, active, and alert, and there was no radiographic evidence of thoracic or abdominal effusion. Nine months postpericardectomy, the dog remained healthy with normal exercise tolerance when running and swimming.


In this study, we detected infection with B. henselae, B. vinsonii subsp. berkhoffii or both organisms in 5 dogs ranging in age from 3 to 12 years that were diagnosed with pleural, pericardial (constrictive pericarditis), or abdominal effusion. Clinical signs generally were nonspecific and included lethargy, fever, vomiting, diarrhea, abdominal distention, lameness, and cervical pain. DNA was not amplified directly from 3 of 3 blood samples or from 2 of 3 effusion samples, but was amplified from pre-enrichment BAPGM blood and effusion cultures and from subculture isolates, a finding that further supports the utility of the enrichment process before PCR for more optimal detection of infection with a Bartonella spp.3,5,9

In humans, thoracic effusions have been reported as infrequent sequelae of Cat Scratch Disease, caused by B. henselae and due to infection with B. quintana.10 Unfortunately, BAPGM cultures often were established after administration of antibiotics. If blood and effusion samples had been cultured earlier in the course of illness and before antibiotic therapy, enhanced detection of Bartonella or other fastidious bacteria may have been achieved in other cases. As extravascular accumulation of fluid is always caused by a pathologic process and as dogs can be chronically infected with B. vinsonii subsp. berkhoffii for at least a year,11 microbiologic and molecular documentation of infection with this genus of bacteria may reflect an opportunistic role, a cofactor in disease expression or a primary pathogenic role in various patients with cavitary effusion.

Exudates are the expected inflammatory response induced by bacterial infection, but this paradigm may not apply to Bartonella spp. Recently, the lipopolysaccharide of B. quintana was shown to have anti-inflammatory rather than proinflammatory properties.12 In addition, previous experimental infection studies in dogs suggest that B. vinsonii subsp. berkhoffii is associated with organism-induced immunosuppression.11 These and other unknown factors potentially could contribute to the development of an effusion in the absence of a strong host inflammatory response.

Recent experimental studies using rodent models emphasize the ability of Bartonella spp. to invade vascular endothelial cells.13 In immunocompromised people, endothelial infection with B. henselae induces single or multiple vasoproliferative lesions (peliosis hepatis and bacillary angiomatosis).14,15 Therefore, it is plausible that vascular endothelial infection contributes to increased vascular permeability and aberrant fluid accumulation.

Despite isolation or molecular detection of B. henselae and B. vinsonii subsp. berkhoffii in these dogs, no direct cause and effect association can be implicated. However, our results may be clinically relevant because most idiopathic effusions obtained from dogs generally are considered aseptic based on conventional microbiological culture approaches. Two dogs in this study, for which serum was available for testing, were not seroreactive to B. henselae and B. vinsonii subsp. berkhoffii antigens by IFA testing, as described previously.9 Antigenic variability among B. henselae test strains previously has resulted in false negative B. henselae IFA results in human patients with suspected cat scratch disease,16 and a similar occurrence may explain discrepant serology and PCR results. The concept that bacterial infection in transudates or modified transudates obtained from dogs is an infrequent occurrence should be reassessed in the context of Bartonella infection.


This research was supported by the State of North Carolina and in part through graduate student stipend support provided to NA Cherry by Novartis Animal Health, and salary support provided by IDEXX Laboratories and Bayer Corporation. We thank Mrs Tonya Lee for editorial assistance.