Materials and Methods

Criteria for selection of cases—Medical records of the University of Minnesota Veterinary Medical Center were searched to identify dogs examined between August 2000 and November 2007 in which GA had been diagnosed. Dogs were eligible for inclusion in the study if the diagnosis had been confirmed on the basis of cytologic findings of cytoplasmic morulae within granulocytes (Figure 1). The study protocol was approved by the University of Minnesota Institutional Review Board.

View Full Size

Photomicrograph of a blood smear from a dog infected with Anaplasma phagocytophilum. Notice the morula within the cytoplasm of a neutrophil (arrow). Wright-Giemsa stain; bar = 10 μm.

Citation: Journal of the American Veterinary Medical Association 234, 12; 10.2460/javma.234.12.1559

• Download Figure
• Download figure as PowerPoint slide

Medical records review—Information obtained from medical records of dogs included in the study included signalment, history, county of residence, physical examination findings, concurrent diseases, clinicopathologic findings, results of serologic testing, treatment, and outcome. Blood samples were tested at the University of Minnesota with a commercial, nonquantitative ELISA,^a performed in accordance with the manufacturer's directions, for Dirofilaria immitis antigen and antibodies against Ehrlichia canis and B burgdorferi. In addition, serum samples were submitted to a commercial laboratory^b for determination of IFA titers against A phagocytophilum, Rickettsia rickettsii, E canis, and B burgdorferi.

Telephone survey—Owners of dogs included in the study were contacted and asked to participate in a telephone survey. At least 3 attempts were made to contact an adult owner of each dog. Owners were contacted between 2 and 4 months after the diagnosis of GA had been made. The survey contained 14 questions regarding tick exposure, treatment, and outcome and took 5 to 10 minutes to complete. Questions were also included regarding the dog's travel in the month prior to the diagnosis; whether the dog participated in hunting, camping, hiking, or fishing trips; whether the owner used tick preventatives on the dog and, if so, when and what kind; whether the dog had completed the prescribed course of treatment; how long it took for the dog to resume normal activities after the initiation of treatment; and whether there were any residual clinical signs. Referring veterinarians were contacted for information regarding results of clinicopathologic testing and follow-up information that was not included in the medical record.

Statistical analysis—Descriptive statistics were calculated. The z test was used to compare the proportion of each dog breed in the study to that of the hospital population over the study period. All analyses were performed with standard software.^c Values of P < 0.05 were considered significant.

Results

Dogs—Thirty-four dogs met the criteria for inclusion in the study. There were 4 mixed-breed dogs and 30 purebred dogs, including 7 Labrador Retrievers, 4 Golden Retrievers, 2 American Cocker Spaniels, 2 Brittany Spaniels, and 1 each of 15 additional breeds. Labrador Retrievers were not significantly (P = 0.49) overrepresented, compared with the population of dogs examined at the University of Minnesota Veterinary Medical Center during the study period.

Median age of the 34 dogs included in the study was 8 years (range, 1 month to 13 years). Age distribution was bimodal, with 8 (24%) dogs ≤ 1 year old and 17 (50%) dogs ≥ 8 years old. There were 12 castrated male dogs, 5 sexually intact male dogs, 14 spayed female dogs, and 3 sexually intact female dogs.

All 34 cases were diagnosed between March and November (Figure 2). Distribution of month of diagnosis was bimodal with 15 (44%) cases diagnosed during May and June and 11 (32%) cases diagnosed during October and November.

View Full Size

Distribution of month of diagnosis for 34 dogs from Minnesota and Wisconsin in which GA was diagnosed.

Citation: Journal of the American Veterinary Medical Association 234, 12; 10.2460/javma.234.12.1559

• Download Figure
• Download figure as PowerPoint slide

History and physical examination findings—Duration of illness prior to diagnosis ranged from 1 to 60 days, with 21 of 28 (75%) dogs sick for ≤ 7 days before the diagnosis of GA was made. Initial complaints included lethargy (n = 25/34 [74%]), anorexia (21 [62%]), lameness (11 [32%]), vomiting (8 [24%]), and diarrhea (3 [9%]). One dog was a 4-week-old puppy that was initially examined because of a corneal laceration and did not have any clinical signs of GA. Twenty-seven of the 32 (84%) dogs in which rectal temperature was recorded had a fever (rectal temperature > 39.2°C [102.5°F]) at the time of initial examination. For dogs with a fever, median rectal temperature was 40.6°C (105°F; range, 39.3° to 41.5°C [102.7° to 106.7°F]). Abnormalities identified during the initial physical examination included enlarged peripheral lymph nodes (n = 11/34 [32%]), high respiratory rate (10 [29%]), and signs of abdominal pain (3 [9%]). Splenomegaly was detected during abdominal palpation in 4 of the 34 (12%) dogs, and swelling of the joints was evident in 2 (6%). Three dogs had increased lung sounds on thoracic auscultation, and 2 had a cough.

Concurrent diseases were identified in 19 of the 34 (58%) dogs. Conditions that had been identified prior to the diagnosis of GA included hypothyroidism (n = 3), osteoarthritis (3), atopy (2), seizure disorders (2), laryngeal paralysis (2), subclinical giardiasis (2), chronic bronchitis (1), immune-mediated hemolytic anemia (1), immune-mediated polyarthritis (1 dog with joint swelling), and hypoadrenocorticism subsequent to treatment for hyperadrenocorticism (1). Conditions identified at the same time as GA included urinary tract infection (n = 2), corneal laceration (1), and systemic inflammatory response syndrome¹⁵ of undetermined cause (1).

Clinicopathologic findings—The most consistent clinicopathologic abnormality, other than intracytoplasmic morulae, was thrombocytopenia (Table 1). Twenty-one of the 22 dogs in which platelet counts were measured had thrombocytopenia, with 11 of the 21 (52%) having platelet counts < 50,000 platelets/μL, and 16 (47%) had mild anemia. White blood cell count was variable (median, 7,500 WBCs/μL; range, 3,500 to 20,200 WBCs/μL; reference range, 4,100 to 13,300 WBCs/μL). Three (9%) dogs had leukopenia (WBC count between 3,500 and 4,000 WBCs/μL), and 6 (19%) had leukocytosis (WBC count between 14,000 and 20,200 WBCs/μL). Six of 31 (19%) dogs had neutrophilia, and only 1 (3%) had neutropenia. Neutrophil count ranged from 1,300 to 18,080 neutrophils/μL (median, 6,950 neutrophils/μL; reference range, 2,100 to 11,200 neutrophils/μL). Only 2 dogs had band neutrophils evident on blood smears. Twenty of 31 (65%) dogs had lymphopenia, with lymphocyte counts ranging from 0 to 2,280 lymphocytes/μL (median, 510 lymphocytes/μL; reference range, 780 to 3,360 lymphocytes/μL). Coagulation testing was performed in 3 dogs, and results were within reference limits in 2 of the 3. The remaining dog had a prolonged partial thromboplastin time; this was the dog with severe inflammatory response syndrome.

Serum chemistry profiles were available for 27 dogs. Twelve (44%) had hypoalbuminemia, 14 (52%) had high serum alkaline phosphatase activities, and 8 (30%) had high serum alanine transaminase activities. Ten (37%) dogs had high serum total bilirubin concentrations.

Results of a urinalysis were available for 13 dogs. One dog had isosthenuria, and 2 had hyposthenuria. Two dogs had proteinuria, with urine protein-to-creatinine ratios of 2.2 and 1.5.

Serologic findings—Nineteen dogs were tested with an IFA assay for antibodies against A phagocytophilum, and 15 (79%) were seropositive (ie, titer ≥ 1:80), with titers ranging from 1:80 to 1:10,240. Twelve of the 15 seropositive dogs had titers ≥ 1:320. One additional dog tested with a point-of-care assay^d by the referring veterinarian was also negative for antibodies against A phagocytophilum. Overall, therefore, 5 of 20 (25%) dogs were seronegative for antibodies against A phagocytophilum at the time morulae were observed in neutrophils. Duration of clinical signs prior to initial examination in the 5 seronegative dogs ranged from 2 to 7 days; all were febrile (rectal temperature, 40.3° to 40.6°C [104.6° to 105.1°F]), and all 3 dogs for which platelet counts were available had thrombocytopenia (59,000 to 83,000 platelets/μL).

In all but 1 dog in which serologic testing was performed, samples used for testing were collected at the time of initial examination. The remaining dog did not have any serologic or microscopic evidence of GA at the time that immune-mediated hemolytic anemia was diagnosed. However, 6 weeks later, while the dog was receiving immunosuppressive doses of corticosteroids, morulae were identified within circulating neutrophils and the A phagocytophilum antibody titer was 1:2,560. Convalescent serologic testing was not performed in any of the other dogs.

Fourteen dogs were tested with an IFA assay for antibodies against E canis, and all 14 were seronegative. Fifteen dogs were tested with an IFA assay for antibodies against R rickettsii, and 1 was seropositive (titer of 1:80). Twelve dogs were tested with an IFA assay for antibodies against B burgdorferi, and 10 were seropositive, with titers ranging from 1:640 to 1:20,480. Nine dogs were tested with an ELISA^a for antibodies against E canis and B burgdorferi, and 1 was seropositive for antibodies against B burgdorferi. Overall, 17 dogs were tested for antibodies against B burgdorferi, and 10 (59%) were seropositive. Six of 11 dogs examined because of lameness were seropositive for antibodies against B burgdorferi. Only 4 dogs were simultaneously tested with both the IFA assay and ELISA for antibodies against B burgdorferi. Results of both assays were negative in 1 dog and positive in another dog. Two dogs with positive IFA assay results had negative ELISA results. Both of these dogs had been vaccinated against Lyme disease, and 1 had been vaccinated 4 days prior to testing and had an IFA titer of 1:10,240.

Treatment—Information on treatment was available for 32 of the 34 dogs. Thirty-one dogs were treated with doxycycline, and 1 was treated with tetracycline. The dosage of doxycycline ranged from 5 to 11 mg/kg (2.3 to 5 mg/lb) every 12 hours for the 28 dogs treated twice daily and from 10 to 13 mg/kg (4.5 to 5.9 mg/lb) every 24 hours for the 2 dogs treated once daily. Duration of treatment ranged from 2 to 8 weeks, although most dogs (19/28 [68%]) were treated for 3 to 4 weeks.

Information on response to treatment (ie, resolution of fever and clinically important signs) was recorded in the medical record of 17 dogs. Clinical signs resolved within 1 to 2 days in 7 (41%) dogs, within 3 to 5 days in 7 (41%) dogs, and within 1 to 3 weeks in 3 (18%) dogs. Platelet count was within reference limits within 2 to 14 days (median, 7 days) in 11 of the 14 dogs for which follow-up platelet counts were available. The 3 remaining dogs still had thrombocytopenia 3 to 5 days after initiation of doxycycline treatment; platelet counts beyond this time were not available.

Telephone survey—Twenty-four of the 34 (71%) owners participated in the telephone survey. Of these, 12 (50%) reported that they had found a tick embedded in the skin of their dog in the month prior to the diagnosis of GA. Nineteen of the 24 (79%) dogs participated in activities that could potentially have exposed them to ticks in the month prior to the diagnosis of GA, including camping (n = 11), hiking (10), hunting (5), and fishing (1) trips, with 7 dogs having engaged in both camping and hiking trips in the month prior to the diagnosis of GA. Twelve of the 24 (50%) owners said that they used tick preventative on their dog, but 8 of these 12 reported using tick preventative only during the summer months (ie, June through August). Only 1 owner reported using tick preventative year-round, 3 owners reported using tick preventative from April through November, and 1 owner reported using tick preventative from June through October. All tick preventatives that were used contained fipronil.^e

When survey participants were asked how long it took for the dogs to resume normal activities after the initiation of treatment for anaplasmosis, 7 of 23 (30%) owners responded that their dog resumed normal activities in 1 to 2 days, 7 (30%) responded that their dog resumed normal activities in 3 to 5 days, 5 (21%) responded that their dog resumed normal activities in 1 to 3 weeks, and 4 (17%) responded that it took a month or longer for their dog to resume normal activities. Ongoing clinical signs were reported in 2 dogs, both of which were reported by their owners to be somewhat sedentary and lethargic, although clinical signs were much improved in both dogs, compared with severity of clinical signs at the time of diagnosis. Neither owner had sufficient concerns to have the lethargy investigated by a veterinarian.

County of tick exposure and county of residence—For 20 dogs included in the study, tick exposure apparently occurred during travel to 12 counties (9 in Minnesota and 3 in Wisconsin) in northwestern Wisconsin, east-central Minnesota, and northern Minnesota (Figure 3). Only 3 dogs were reportedly exposed to ticks in their county of residence (Hennepin and Anoka counties in Minnesota and Pepin county in Wisconsin). None of the dogs in the present study reportedly traveled beyond Minnesota and Wisconsin.

View Full Size

Map illustrating counties in which 34 dogs with GA were reportedly exposed to ticks.

Citation: Journal of the American Veterinary Medical Association 234, 12; 10.2460/javma.234.12.1559

• Download Figure
• Download figure as PowerPoint slide

Discussion

Results of the present study suggested that GA should be suspected in dogs in endemic areas examined because of fever, lethargy, or thrombocytopenia, especially in dogs examined during the late spring or early fall. Treatment with doxycycline was successful in resolving clinical signs and thrombocytopenia.

In the present study, detection of morulae in the cytoplasm of granulocytes was considered diagnostic of GA, although most dogs were also seropositive for antibodies against A phagocytophilum. Although Ehrlichia ewingii is also a tick-borne, granulocyte-tropic bacterium morphologically identical to A phagocytophilum, it is unlikely that dogs included in the present study were infected with this agent because E ewingii is most common in the southern and southeastern United States^16–18 and none of the dogs in the present study had traveled outside of Minnesota or Wisconsin. When serologic testing for both A phagocytophilum and E canis is performed, a combination of positive A phagocytophilum results and negative E canis results increases the likelihood that the infecting agent is A phagocytophilum because antibodies against E ewingii can cross-react with E canis. However, this combination of results does not definitively rule out E ewingii infection or rule in A phagocytophilum infection. Although antibodies against A phagocytophilum can sometimes cross-react in serologic tests with E canis, the higher titer typically represents the true infecting agent.

The 34 dogs in the present study were apparently exposed to ticks in 9 of the 87 counties in Minnesota and 3 of the 72 counties in Wisconsin. These counties likely are not the only counties in these states where exposure to A phagocytophilum occurs but simply reflect the referral population of the University of Minnesota Veterinary Medical Center. Of note is the fact that the Minnesota counties in which tick exposure occurred paralleled those reported as high-risk areas for tick-borne disease by the Minnesota Department of Health.¹⁹

Many dog owners in the present study reported that their dogs were involved in activities (eg, camping, hiking, and hunting trips) that would increase exposure to ticks. However, dogs without any travel history or any history of similar outdoor activities were also included in the study. Thus, any dog living in an area where this pathogen is endemic should be considered at risk for GA. Twelve of 24 owners reported using topical tick preventatives, but 8 of the 12 used them only during the summer months. This may be 1 reason that most cases of GA occurred during the late spring and early fall, in that many dogs may not be protected against ticks at these times. Other reports^4,14 have shown a similar seasonality of GA, which likely reflects times of peak tick host-seeking activity.²⁰ We recommend therefore that dog owners in the upper Midwest use topical tick preventatives from at least March through November.

For dogs in the present study, infection with A phagocytophilum most often resulted in fever, lethargy, anorexia, and lameness, which is consistent with findings in a previous report⁴ of dogs with GA in the same geographic area. Because these clinical signs are similar to those associated with borreliosis and other tick-borne diseases, it is important to test dogs with these signs that are living in areas where the organism is endemic for GA. Vomiting, splenomegaly, and diarrhea were uncommon clinical signs among dogs in this and previous reports.^4,13

Clinicopathologic abnormalities in dogs in the present study were consistent with those reported previously for dogs and humans.^4,21,22 Although thrombocytopenia is the most consistent clinicopathologic finding of dogs and humans with GA,^4,21,22 the mechanism of thrombocytopenia is unknown. In a previous study²³ in which mice were experimentally infected with A phagocytophilum, splenectomized and nonsplenectomized mice as well as mice with intact immune systems and those with severe combined immunodeficiency had equivalent levels of thrombocytopenia. Thus, immune-mediated destruction and splenic sequestration are unlikely mechanisms of thrombocytopenia in the acute phase of disease. Endothelial cells can become infected with A phagocytophilum,^24,25 suggesting that thrombocytopenia may be caused by platelet activation and consumption. Although megakaryocytes can become infected with A phagocytophilum, infection does not alter platelet production in cell culture.²⁶

Fourteen of 27 (52%) dogs in the present study had slightly high serum alkaline phosphatase activities. This was lower than the percentage of dogs with high alkaline phosphatase activity (100%) in a previous study⁴ from the upper Midwest but higher than the percentage (7%) in a study² of 14 dogs in Sweden with GA. Although 8 of 27 (30%) dogs in the present study had slightly high alanine transaminase activities, none of the dogs in the previous studies^2,4 did. Concurrent diseases in these dogs could have been responsible for these clinicopathologic abnormalities. In experimentally infected mice, the innate immune system appears to be responsible for hepatic injury regardless of pathogen burden.²⁷

Doxycycline is the standard treatment for A phagocytophilum infection,^22,28–30 and treatment with doxycycline in 31 dogs in the present study coincided with resolution of clinical signs, fever, and thrombocytopenia. Clinical improvement occurred 1 day to 3 weeks after initiation of treatment. All dogs were treated for at least 2 weeks with doxycycline at a minimum dosage of 10 mg/kg/d. To the authors' knowledge, there have been no controlled studies to evaluate the optimal dose of doxycycline or the optimal duration of treatment in dogs or humans. Only general recommendations can be made on the basis of our findings, but we recommend treating dogs with clinical GA with doxycycline at a dosage of 5 to 10 mg/kg every 12 hours for 2 to 4 weeks.

Five of 20 dogs in the present study were seronegative for antibodies against A phagocytophilum during the acute phase of illness. As in human patients, clinical disease can be manifest and morulae can be found within granulocytes before a measurable antibody response occurs. In a study²⁹ of human patients in the upper Midwest, for instance, all 12 patients tested during the acute phase of the disease were seronegative for antibodies, but 9 were seropositive when convalescent samples were tested 27 to 126 days after the onset of symptoms. Therefore, in areas where the causative organism is endemic, GA must be considered in the differential diagnosis for dogs with unexplained fever and thrombocytopenia, even if results of serologic testing are negative. Detection of morulae in the cytoplasm of granulocytes supports a definitive diagnosis of GA, and testing of convalescent serum samples is recommended for all dogs suspected of having GA on the basis of compatible clinical signs. It is important to remember that not all dogs seropositive for antibodies against A phagocytophilum have GA. Serologic studies^3,5,14 of dogs in areas where the organism is endemic have found a high seroprevalence of antibodies against A phagocytophilum, despite the fact that relatively small numbers of clinical cases have been reported. Indeed, 1 dog in the present study did not have any clinical signs of disease, even though morulae were seen in granulocytes and results of serologic testing were positive.

Recently, a new point-of-care ELISA^d has become available that allows veterinarians to test dogs for antibodies against A phagocytophilum, E canis, and B burgdorferi and for D immitis antigen. Undoubtedly, many veterinarians in endemic areas will find a large percentage of their canine patients test positive for antibodies against A phagocytophilum, as many dogs exposed to infected ticks develop subclinical infection and apparently clear the organism. Therefore, a positive serologic test result is not a sufficient reason to treat a dog with doxycycline. On the other hand, a negative serologic test result does not rule out GA because infected dogs can develop clinical signs before becoming seropositive. If GA is suspected on the basis of clinical signs and positive serologic test results, testing of a convalescent serum sample may be helpful in retrospectively determining whether A phagocytophilum was responsible for the patient's illness.

Granulocytic anaplasmosis is transmitted by ixodid ticks, including Ixodes scapularis in the upper Midwest and Northeast and Ixodes pacificus in northern California. Ixodes ticks also transmit B burgdorferi, and in mice, coinfection with A phagocytophilum and B burgdorferi has been shown to alter immune responses and disease severity.³¹ Both pathogens act synergistically to alter cytokine production, resulting in suppression of interleukin-2 and interferon-J production and enhancement of interleukin-4 production.³¹ Mice experimentally infected with both A phagocytophilum and B burgdorferi developed a heavier bacterial burden and more severe arthritis than those infected with B burgdorferi alone.³² In a study¹⁴ of dogs in northern Minnesota, 25% of those tested had been exposed to both A phagocytophilum and B burgdorferi, and dogs were more likely to be clinically affected if they were seroreactive to both organisms than if they were seroreactive to only 1. In the present study, 10 of 17 dogs tested for antibodies against B burgdorferi were seropositive. However, only 1 of these dogs was tested with the ELISA that tests for the C6 peptide of B burgdorferi, making it specific for natural exposure to the organism, versus the IFA assay, which cannot as readily differentiate between natural and vaccine exposure. Indeed, 2 of the dogs in this report with positive IFA assay results and negative ELISA results were known to have been vaccinated recently. A limitation of the present study was that vaccine history was unavailable for 8 of the 10 dogs with positive IFA assay results for B burgdorferi. Therefore, it is unknown whether these antibody responses represented natural exposure to the pathogen or vaccination. Additionally, because only 9 of the 34 dogs were tested for antibody against the C6 peptide, it is difficult to make any conclusions regarding coinfection. All but 1 dog in the present study was brought to the Veterinary Medical Center with clinical signs consistent with GA, but it is possible that some of these dogs developed clinical illness as a result of coinfection. Because A phagocytophilum and B burgdorferi are both transmitted by the same vector and because coinfection may alter immune responses to these infections, more information is needed regarding the clinical and clinicopathologic changes in dogs with coinfection.

Other limitations of the present study include the nonuniform availability of serologic and clinicopathologic data and variability in the dose of doxycycline and duration of treatment. Long-term outcome of dogs in the study was not assessed. Findings further elucidate the clinical nature of GA in dogs and alert practitioners to its occurrence in areas where A phagocytophilum is endemic.