1. Introduction
Ehrlichiosis is caused by bacteria classified within the group of the alpha-proteobacte- ria, order Rickettsiales, family Anaplasmataceae, genus Ehrlichia [1] . This genus consists of obligate intracellular Gram-negative bacteria that mainly infect leukocytes such as monocytes, macrophages, and granulocytes [2] .
Monocytotropic canine ehrlichiosis is caused by Ehrlichia canis, which is found in the form of morulae in the cytoplasm of lymphocytes, monocytes, and macrophages [3] . The incubation period ranges from 8 to 20 days [4] followed by the acute, subclinical, and chronic stages of the disease.
The most important clinical signs of infection are fever, depression, lethargy, anorexia, weight loss, cardiac arrhythmias, hemorrhagic diathesis, central nervous system signs, lymphadenomegaly, splenomegaly, polyarthritis, uveitis anterior, ocular signs and blindness from retinal edema or bleeding and detachment. The most significant laboratory findings are thrombocytopenia, anemia, leukopenia, hyperglobulinemia, proteinuria, lymphocytic pleocytosis, and bone marrow plasmacytosis [4] .
The acute phase lasts between two to four weeks [5] and is characterized by fever, weight loss, anorexia, depression, lymphadenomegaly, splenomegaly, vasculitis, and ocular and musculoskeletal signs [6] [7] [8] . Thrombocytopenia is the most common abnormality in naturally or experimentally infected dogs in this phase of the disease [8] .
The subclinical phase has a duration that varies from months to years [5] [9] . Dogs in this stage are known to be carriers of E. canis and remain clinically healthy for months or years [10] until they spontaneously recover from infection or develop severe illness. During this phase, the animal may show high anti-E. canis antibody titers [11] , persistent thrombocytopenia, and leucopenia with no other clinical signs [9] [10] [11] . Generally, bacterial DNA is not found in blood samples in this phase; spleen and bone marrow are the most appropriate tissues to analyze [3] [12] . Severe pancytopenia, hemorrhagic diathesis, and organism debilitation may characterize the chronic phase [5] . Although the conditions that can lead the animal to this stage are not totally clear, it has been suggested that factors such as breed, immune system deficiency, stress conditions, co-infections, virulence of the strain, and the geographical region could have some influence on the outcome of infection [10] .
E. canis was reported for the first time in dogs of Costa Rica by Meneses [13] . Romero et al. [14] analyzed blood samples from dogs with clinical signs or suspected of suffering ehrlichiosis, and found highly significant differences in hematocrit and hemoglobin values between PCR positive and negative dogs. In this population, 26.6% (8/30) were determined to have acute (seronegative and PCR positive) E. canis infections, while 20.0% (6/30) had subclinical (seropositive and PCR negative) and 6.6% (2/30) had chronic (seropositive and PCR positive) E. canis infections [14] . Another study carried out in four regions of Costa Rica found 58.0% (29/50) of dogs with subclinical E. canis infections [15] . The present study describes and analyzes, in a more representative sampling frame, the clinical and hematological presentation of E. canis infection in dogs of Costa Rica.
2. Materials and Methods
2.1. Study Design, Sample Size and Analyzed Population
A cross-sectional, observational, descriptive study was conducted to determine the presence of, or exposure to, Ehrlichia spp. in blood samples from dogs, using molecular and serological assays, respectively. The total sample size was estimated to be 385 individuals (50% prevalence, 95% confidence) for a population of more than 40,000 dogs, calculated using Win Episcope 2.0. More details of the population studied, sampling methodology and period of study are described in Barrantes-González et al. [16] .
2.2. Interview, Clinical Examination, and Sampling
Each owner was interviewed to obtain information about the place of origin, tick infestation, treatment of ticks, and signs suggestive of ehrlichiosis (fever, weight loss, depression, epistaxis, petechiae, ecchymosis, hematuria, dyspnea, cough, lymphade- nomegaly, ataxia, diarrhea and scrotal edema), observed by the owners at some point in the lives of their pets. Also, other important information was obtained like if their veterinarian had suspected ehrlichiosis in the past, if the dog was treated because of this suspicion, and finally the medications used to treat those dogs was investigated. Veterinarians made sure to ask the owners in an appropriate fashion, adapting their vocabulary to the educational level of the respondent, to ensure that the questions were understood correctly. Clinical examination was performed to determine attitude (weak, depressed, docile, alert, nervous, aggressive), capillary refill time (>2 s was considered as delayed), color of mucous membranes (very pale, pale, pink, icteric), rectal tempe- rature (≥39.5˚C was considered as feverish) and clinical signs suggestive of ehrlichiosis (weight loss, epistaxis, petechiae, ecchymosis, hematuria, dyspnea, cough, lympha- denomegaly, ataxia, lameness, diarrhea and scrotal edema). Dogs that were treated with doxycycline were recorded. From stray dogs living in recreational parks, consent from the administration was obtained. Only clinical exam and sampling was performed.
2.3. Blood Analysis
The HETTICH® microcentrifuge (5 minutes × 18,600 g) and DAMON/IEC hematocrit reader were used to determine hematocrit values. Blood smears were stained with Giemsa as described by Cowell et al. [17] to determine complete blood count (CBC).
2.4. Serological Analysis
Two commercial techniques were used to detect antibodies against E. canis, “Speed Ehrli” Virbac, an Immunochromatography Membrane Assay (IMA) (Bio Veto Test, Rome, Italy; sensitivity 87%, specificity 95%) and “E. canis and A. phagocytophilum Canine IgG Antibody Kit”, an Indirect Immunofluorescence Assay (IFA) (Fuller Lab- oratories, California, USA; sensitivity and specificity 100%). The methodologies recommended by the manufacturers were used. Sera were analyzed in IFA only in one dilution (1:80). Sera that exhibited fluorescence in 1:80 dilution were considered positive in IFA. Results of the two tests were compared. “Parallel testing” methodology was used to determine the seroprevalence.
2.5. Molecular Analysis
Extraction of DNA from blood samples was performed with the “Wizard Genomic” assay (Promega®, Wisconsin, USA). The conventional nested PCR was carried out as described by Romero et al. [14] : Primers ECC and ECB were used in the first round to amplify a segment of the 16S rRNA of Ehrlichia spp. The second PCR was carried out using primers HE3 and ECAN5 for E. canis, primers HE3 and HE1 for E. chaffeensis, and HE3 and EE5 for amplifying E. ewingii. PCR products (389 to 396 bp) were purified using the QIAquick kit (QIAGEN®), proceeding according to the manufacturer’s instructions, and sent to Macrogen (Seoul, Korea) for sequencing. Further detailed molecular analysis methodology is described in Barrantes-González et al. [16] .
2.6. Statistical Analysis
Data obtained from the interview, clinical examination and results of diagnostic tests (serology and PCR) were entered a digital database. A descriptive analysis of the results through measures of central tendency (mean), measures of dispersion (standard deviation and 95% confidence interval) and frequency (%) by factors was performed. The presence or absence of E. canis antigen and presence or absence of antibodies against E. canis was determined, and frequency distributions of positive and negative results were determined. One-way ANOVA was performed to determine the difference of means between positive and negative (serology and molecular detection separately) groups for each hematological variable. Two-way ANOVA was carried out to assess the difference of means between positive and negative groups per infection status (non-infected, acute, subclinical and chronic) to E. canis. The statistical analysis was performed using STATA IC 13 (Stata Corp., USA).
3. Results
Of the total of dogs (n = 441) that participated in the study, 399 had owners, and it was possible to conduct an owner interview, while 42 (9.5%) dogs were residents of the parks and had no known owner. A total of 55.9% of the owners had observed ticks on their pets, only 2.0% (9/441) of the dogs were diagnosed previously by a veterinarian with ehrlichiosis; 77.8% of them had been treated (28.6% with doxycycline and 71.4% did not remember the name of the medication) to: “ehrlichiosis; seven dogs had been treated (2 with doxycycline, and 5 did not remember the name of the medication), and...” in 3.6% of cases (16 dogs) the veterinarians suspected that the pets had ehrlichiosis. Clinical signs suggestive of ehrlichiosis observed by the owners at some point in the lives of their pets were: bleeding (17.0%), petechiae (9.5%), weight loss (4.5%), and hematuria (1.8%) (Table 1).
Clinical examination encountered abnormalities in only a minority of the dogs, such as attitude (weak 0.2%, depressed 0.7%, docile 16.1%), capillary refill time > 2 seconds (5.2%), mucous membranes (very pale 1.1%, pale 11.1%), and rectal temperatures ≥ 39.5˚C (19.5%). Clinical abnormalities included weight loss (4.5%), lymphadenomegaly (2.0%), petechiae (1.6%), cough (0.9%), scrotal edema (0.9%), ataxia (0.5%) and hematuria (0.2%) (Table 1).
Hematocrit and complete blood count (CBC) results from the total dog sample are shown in Table 2.
Analysis of hematological values of dogs that were seropositive and seronegative to E. canis determined a highly significant difference (p < 0.001) in mean hematocrit, hemoglobin and platelet count values, and a weak difference (p = 0.05 - 0.10) in mean corpuscular hemoglobin concentration (MCHC), white blood count (WBC), specifi- cally, lymphocytes and monocytes values. In all cases, mean values for all the above variables were lower in seropositive than in seronegative dogs.
Also, analysis of dogs that were PCR positive and PCR negative to E. canis determined a highly significant difference in the mean values of lymphocytes and basophils, and a weak difference in the hematocrit and mean values between PCR positive and PCR negative dogs. In all cases, mean values for all the above variables were lower in positive than in negative dogs (Table 2).
Table 1. Clinical signs observed by the owner in the past and clinical signs observed by the veterinarian at the moment of sampling.
Table 2.Hematological values between seropositive and seronegative,PCR-positive and PCR-negative dogs to E.canis determined by One-Way ANOVA and overall values of the total dog population analyzed in Costa Rica.
Using parallel testing methodology, 32.1% (131/408) of the dogs were determined as seropositive to E. canis: 23.2% (95/408) by IFA and 30.0% (121/403) by IMA. The DNA of E. canis was detected in 3.2% (13/407) of the analyzed blood samples, DNA of E. chaffeensis and DNA of E. ewingii was not found. A total of 10 (77.0%) E. canis PCR positive dogs showed also antibodies against E. canis.
The results of serological and molecular tests found 29.7% (121/407) of dogs with subclinical infection (seropositive and PCR-negative), 2.5% (10/407) with chronic infection (seropositive and PCR-positive), and only 0.7% (3/407) with acute E. canis infection (seronegative and PCR-positive).
Suggestive clinical signs reported by the owners of dogs with subclinical E. canis infection were bleeding (26/121), petechiae (8/121), weight loss (9/121), hematuria (2/ 121). Clinical signs detected by the veterinarians at the time of sampling of these dogs were fever (43/121), depressive attitude (1/121), pale mucous membranes (21/121), very pale mucous membranes (2/121), bad body condition (72/121), very bad body condi- tion (38/121), petechiae (3/121), lymphadenomegaly (3/121), and ataxia (1/121).
Two-way ANOVA analysis of these seropositive but PCR-negative dogs showed significant differences (p < 0.05) in mean hematocrit, hemoglobin and platelet count and weak significant difference (p < 0.10) in mean WBC count values. The group with subclinical infection showed always lower values than non-infected dogs in all the above values, and platelet values were lower than in dogs with acute infection (p < 0.05), but the mean hematocrit and WBC count (p = 0.05 - 0.10) were higher than in dogs with chronic infections (Table 3(a) and Table 3(b)).
Suggestive clinical signs reported by the owners of dogs suffering chronic E. canis infection were bleeding (2/10) and petechiae (2/10), whereas the veterinarians detected, at the time of sampling in this group, dogs with depressive attitude (1/10) and very pale mucous membranes (4/10).
Comparison of hematological values of these seropositive and PCR-positive dogs with non-infected dogs determined significant differences (p < 0.05) in mean hematocrit, hemoglobin and platelet count values and weak significant differences (p < 0.05 - 0.10) in mean WBC count values. Mean values were always lower in the group of dogs with chronic E. canis infection compared to non-infected dogs. Mean hematocrit and WBC values were the lowest from all groups and platelet count values were lower compared with dogs with acute infection (Table 3(a) and Table 3(b)).
Suggestive clinical signs reported by the owner of one dog suffering acute E. canis infection was petechiae (1/3), and clinical signs detected by the veterinarians at the time of sampling in this group of dogs was fever (2/3).
Analysis of dogs with acute infections (seronegative but PCR-positives) showed significant differences (p < 0.05) in mean platelet count values when compared to non- infected dogs. Hematocrit and hemoglobin mean values were higher than in dogs with chronic infections (p < 0.05). Mean platelet count values were the highest of all groups (p < 0.05) (Table 3(a) and Table 3(b)).
(a)(b)
Table 3.(a) Hematological values for red blood indices and platelet counts among non-infected (1), acute (2), subclinical (3) and chronic (4) E. canis infected dogs determined by Two-Way ANOVA test; (b) Hematological values for white blood cells among non-infected (1), acute (2), subclinical (3) and chronic (4) E. canis infected dogs determined by Two-Way ANOVA test.
Two-way ANOVA results, when the model was significant (p < 0.10), must be read in the following order: corrected model, intersection, serology effect, molecular detection effect and interaction; otherwise only the corrected model data was included 1: Non-infected dogs 2: Acute infection 3: Subclinical Infection 4: Chronic infection MCHC: Mean Corpuscular Hemoglobin Concentration: Number of individuals SD: Standard Deviation 95% CI: 95% Confidence Interval SSD: Statistically Significant Mean Differences (means bearing different letters differ significantly (p < 0.05); letters bearing *or + differ by p = 0.05 - 0.10 exclusively from those that also bear the same symbol, otherwise p < 0.05.
Two-way ANOVA results, when the model was significant (p < 0.10), must be read in the following order: corrected model, intersection, serology effect, molecular detection effect and interaction; otherwise only the corrected model data were included 1: Non-infected dogs 2: Acute infection 3: Past infection 4: Persistent infection WBC: White Blood Cells n: Number of individuals SD: Standard Deviation 95%.
Analysis using two-way ANOVA did not determine any association with changes in white blood cell differential values and serology or molecular detection status, nor did it determine any significant difference in white blood cell values between infection stages (Table 3(b)).
4. Discussion
Suggestive clinical signs reported by the owners of dogs suffering subclinical infections (bleeding, petechiae, weight loss, and hematuria) were in accordance with reports in the literature of dogs presenting acute or chronic E. canis infection [18] . Clinical signs detected by the veterinarians at the time of sampling (fever, depressive attitude, pale or very pale mucous membranes, bad or very bad body condition, petechiae, lymphad- enomegaly, and ataxia) were also clinical signs that may be caused by other hemo- parasite infections [19] . Hemopathological findings in these groups of dogs were anemia (low hematocrit and hemoglobin), leukopenia and thrombocytopenia, showing lower values than non-infected dogs, but higher values than dogs with acute or chronic infections. These results are consistent with other reports that suggest that dogs need about 505 days to show normal hematologic values after an acute E. canis infection [20] . It is important to point out that only a very few dogs in this category (3.3%; 4/121) had been treated for ehrlichiosis at some time in their lives with doxycycline; therefore, it seems that infection passed unnoticed by owners" instead of "it appears that infection passed unnoticed by owners and that most of the dogs resolved E. canis infections without any medication. However, it cannot be ruled out that dogs continue to be persistently infected with E. canis. It has been reported that untreated or inadequate treated dogs can recover clinically, but remain infected with E. canis for months or years, showing only abnormal platelet counts and representing a source of infection for other dogs [10] [11] . In these cases, experimental results indicate that the bacteria are located in the spleen, which is the last organ to eliminate this pathogen [21] . Since thrombocyte values were lower than in dogs with acute and non-infected dogs, further studies have to determine if the E. canis is still present in tissues or organs (bone marrow and spleen) of subclinically infected dogs and if the infection was resolved without medication. The above could establish future recommendations about the need for medication of dogs with a positive serological result for E. canis and with hemopathological findings, but without DNA positive results and clinical signs, and for the implementation of antigen detection instead of antibody detection techniques [21] .
Suggestive clinical signs reported by the owners (bleeding and petechiae), and detected by the veterinarians at the time of sampling (depressive attitude and very pale mucous membranes) of dogs suffering chronic infections were in accordance with reports of the literature [18] . These chronic infections could be due to the fact that only 1 out of 10 dogs was treated with doxycycline, although ineffective. It has been docu- mented that dogs fail to clear the systemic infection with the pathogen [19] and the lack of diagnosis and treatment of these dogs could have an impact on the incidence of this event. Hemopathological findings in these group of dogs were anemia (low hematocrit and hemoglobin values) and thrombocytopenia, which is consistent with other reports [11] [14] [22] [23] .
Hemopathological findings of dogs with chronic E. canis infection were anemia (low hematocrit and hemoglobin), leukopenia and thrombocytopenia and values were in all cases lower in the group compared with non-infected dogs, and dogs with acute and subclinical infections, due to the longer presence of E. canis in blood, and probably other tissues and organs, representing a source of infection [11] [24] and a burden for the organism since the infection was not resolved solely by the dog. Medication of dogs with chronic infection is recommended, and follow-up treatment is necessary to ensure elimination of E. canis from the dog’s body.
Suggestive clinical signs reported by the owners (petechiae), and detected by the veterinarians at the time of sampling (petechiae and fever) of dogs suffering from acute infections were in accordance with Gaunt et al. [25] , who found fever as the only clinical sign in dogs with experimental E. canis infection between 21 - 35 days p.i. Harrus and Waner [18] stated that hemorrhagic tendencies, like dermal petechiae, were one of the clinical signs presented during the acute phase. Hematological findings were higher hematocrit, hemoglobin and thrombocyte count values than in dogs with chronic infection. This difference is of particular importance since the course of chronic infec- tions may often be complicated and progressively become worse as a result of bone marrow hypoplasia [23] . Anemia was not seen in these dogs although when present it was classically mild to moderate [26] . Platelet count values were also higher than those in non-infected dogs. The acute phase is characterized by megathrombocytosis [8] . Some of these infections could also be recent acute infections since reports on experimental infection in German shepherds using the E. canis Oklahoma strain show that the detection of anti-E. canis antibodies was possible two days post infection [27] . Other authors observed seroconversion in Beagles infected with the same Oklahoma strain, occurring between days 5 and 15 p.i. [28] .
The present study determined clinical signs and significant hemopathological values of acute, subclinical and chronic E. canis infections in dogs of Costa Rica to guide veterinarians in making treatment decisions. Since serological assays are the only available rapid tests for veterinarians, we recommend treating only those dogs that show sug- gestive clinical signs and hemopathological values, while the respective PCR is being performed. This is because the clear majority of seropositive dogs (92.6%) in the present study showed no bacteria in the blood, although significant differences in the values of hematocrit, hemoglobin, MCHC, platelet count, leukocytes, lymphocytes, basophils, and monocytes were detected. Future studies must determine if dogs with subclinical E. canis infection eliminate the agent without any medication, or if they are still persistently infected and need medication; it must also be determined if they will develop the chronic disease at some point in their lives.
Acknowledgements
This investigation was supported by the Fondo Especial para el Financiamiento de la Educación Superior, Consejo Nacional de Rectores (FEES-CONARE) and Vicerrectoría de Investigación, UNA. We wish to thank the city councils of the recreational parks we visited, and the Club de Leones for their support.