Clinical evaluation of outdoor cats exposed to ectoparasites and associated risk for vector-borne infections in southern Italy

Background Cats can be carriers of infected arthropods and be infected with several vector-borne pathogens (VBP) but there is limited knowledge about their pathogenic role in cats. Results A cross-sectional controlled study investigated the clinical status and antibody (Bartonella henselae, Rickettsia conorii, Ehrlichia canis, Anaplasma phagocytophilum, Babesia microti and Leishmania infantum) and/or blood PCR (Mycoplasma spp., Bartonella spp., Rickettsia spp., Ehrlichia/Anaplasma spp., piroplasmids, L. infantum, Hepatozoon felis) prevalence in 197 cats. Outdoor cats lacking ectoparasiticide treatment or hosting ectoparasites (study group [SG], n = 134) and indoor cats treated against ectoparasites (control group [CG], n = 63) were enrolled. Clinical data and retroviral co-infections were compared between the two groups. Multivariable analysis tested associations between variables and VBP exposure. Lymphadenia, stomatitis, and various haematological abnormalities were statistically more frequent in SG. Antibodies against R. conorii, B. henselae, A. phagocytophylum, B. microti, E. canis and L. infantum were detected. Bartonella henselae, Bartonella clarridgeiae, Mycoplasma haemofelis, “Candidatus Mycoplasma haemominutum” and “Candidatus Mycoplasma turicensis” DNA were identified. Very high antibody (87.8%) and PCR (40.1%) positivity to at least one pathogen were detected and were significantly higher in SG. Co-infections were confirmed in about one-third of the cats and were more frequent in SG cats. Molecular and overall (antibody and PCR) positivity to Bartonella and antibody positivity to R. conorii were higher in SG. Multivariable analysis found significant associations of Bartonella spp. infection with Feline Immunodeficiency Virus (FIV) infection and increased globulins, and of Mycoplasma spp. infection with adult age, FIV infection, anaemia, and increased creatinine. Conclusions A very high prevalence of exposure to zoonotic VBP was found in cats, with Rickettsia and Bartonella infections being most prevalent. Some risk factors were documented namely for Mycoplasma spp. and Bartonella spp. The lifestyle of cats is clinically relevant and requires specific preventative measures to protect their health. Electronic supplementary material The online version of this article (10.1186/s13071-018-2725-8) contains supplementary material, which is available to authorized users.


Background
Vector-borne infections (VBI) are caused by parasites, bacteria or viruses transmitted by hematophagous arthropods, and many of them are of zoonotic concern [1][2][3][4][5][6]. Cats have a high likelihood of ectoparasite exposures when living an outdoor lifestyle and there is a lack of preventive treatment with acaricides. Consequently, these animals can be carriers of infected arthropods and be infected with several vectorborne pathogens (VBP), as observed in dogs [1,[3][4][5][7][8][9][10][11]. The lack of knowledge on the pathogenic role of most of these VBP in cats may limit the diagnosis of vector-borne diseases (VBD). Also, clinical signs and laboratory abnormalities associated with VBD are widely variable and nonspecific [1,3,4]. Moreover, concurrent VBI or retroviral infections can be found, which may influence the clinical course and outcome of VBD in cats [3,12].
Recent literature highlighted some risk factors associated with cat positivity to VBI such as multi-cat household, outdoor access, male gender, FIV positivity, and abortive FeLV infection [3,4,6,12]. Preventative control measures against ectoparasites infestation, i.e. regular individual use of ectoparasiticide formulations, seem the most effective tool to prevent infection in cats, and other hosts [3,4,12]. The present controlled study evaluated the prevalence and risk factors for some VBP in cats exposed to ectoparasites in southern Italy and assessed the impact of the infections on their health status. . Cats aged > 6 months and experiencing at least an intere vector-season since birth (April-October) were recruited irrespective of breed and gender. Most cats (n = 144; 73%) were admitted for elective surgery or annual health check. They were enrolled when the following information was available: type of housing and lifestyle and individual application of ectoparasiticides. According to this information and the occurrence of ectoparasites at physical examination, two groups of cats were considered. The study group (SG, n = 134) included cats with a greater chance of exposure to ectoparasites, i.e. outdoor cats with a lack of regular individual ectoparasiticide treatment and having ectoparasites at enrolment. The control group (CG, n = 63) was composed of indoor cats with no evidence of ectoparasites at enrolment, receiving the appropriate ectoparasiticide treatment, and therefore with a low risk for ectoparasites. Cats living in rescue catteries were excluded from this study.

Study sites, cat enrolment and sampling procedures
Clinical history and physical examination findings for cats were registered in a clinical form. Also, information on region, age, sex, breed, lifestyle, ectoparasiticide treatments, flea and tick presence was included. Cats were classified as "young" if they had experienced only one vector-season since birth, and "adult" if they had experienced more than one vector-season.
From each cat, blood, conjunctival and oral swabs were obtained. One millilitre of blood was placed into one tube with EDTA and used within 24 hours for complete blood count (CBC) and subsequently stored at -20°C until further use for molecular investigations. Left-over blood (about 2 ml) was used to perform blood smears (immediately) and to obtain serum after clotting into a dry tube. Blood serum was stored at -20°C until further use for haematological and serological investigations. Urine samples were obtained by cystocentesis when possible and used for urinalysis within 2 hours and urine protein and creatinine ratio (UPC) within 24 h from collection. When enlarged lymph nodes were observed, a fine needle aspiration was performed. Sealed needles and swabs were stored at -20°C until further use for molecular tests.

Hematological investigations and urinalysis
Complete blood count was performed using a laser haematology analyser (IDEXX ProCyte Dx® Hematology Analyser, Idexx Laboratories, Westbrook, Maine, USA). Reference intervals of CBC are listed in Additional file 1: Table S1. Blood smears were stained by May Grünwald-Giemsa staining and examined for haematological abnormalities and presence of hemoparasites [13].
Urinalysis was performed using Combur 9 Test strips (Roche Diagnostics, Indianapolis, Indiana, USA), Vet 360 refractometer (Reichert, Seefeld, Germany) and a microscopic evaluation of urine sediment was performed using the Kova system (Kova International, Garden Grove, CA, USA). Urine protein and creatinine ratio were evaluated with Catalyst Dx® Chemistry Analyzer (Idexx Laboratories, Westbrook, Maine, USA), using 0.4 as the cut-off value for proteinuria [14].

Serological investigations
All cats were tested for feline leukaemia virus (FeLV) antigen and antibodies to feline immunodeficiency virus (FIV) by a rapid enzyme-linked immunosorbent assay (ELISA) (SNAP Combo Plus FeLV ag/FIV ab test, Idexx Laboratories, Westbrook, Maine, USA). Sera from individual cats were also tested for the presence of immunoglobulin G (IgG) antibodies against Bartonella henselae, Rickettsia conorii, Ehrlichia canis, Babesia microti and Anaplasma phagocytophilum antigens by the immunofluorescence antibody test (IFAT) using commercial kits (Fuller Laboratories Fullerton, California, USA). The manufacturer's protocol was followed for all serological tests using a cutoff dilution of 1:64 for B. henselae, R. conorii and B. microti; and 1:50 for E. canis and A. phagocytophilum. The presence of L. infantum IgG antibodies was investigated by IFAT according to Persichetti et al. [15] and the cut-off value was established at 1:80 as previously described [15]. Titres of positive samples were determined for all pathogens, with the exception of B. microti, and considered high when they were at least four times over cut-off value.

Statistical analysis
The Kolmogorov-Smirnov normality test was used to assess the normality of distribution of the continuous variable age. Statistical differences between SG and CG cats were tested for significance by Fisher's exact test for categorical variables and with the Mann-Whitney test for numerical variables using the GraphPad InStat v3.05 for Windows 95 (GraphPad Software Inc., San Diego California, USA, 2000). Differences were considered significant if the P-value was < 0.05. Independent variables that yielded significant differences between the 2 groups were analysed in the overall 197 cats with multivariable logistic regression analysis using the STATA 9.2 software (StataCorp LP, College Station, Texas) to test for possible risk factors associated with investigated VBP. According to the molecular and antibody detection, outcome variables were as follows: L. infantum (molecular and antibody tests), Bartonella spp. (molecular and antibody tests), Mycoplasma spp. (molecular test), R. conorii (antibody test), E. canis (antibody test), B. microti (antibody test), A. phagocytophilum (antibody test). Dichotomous variables (outcome variable and investigated risk factors) were analysed with multivariable logistic regression analysis, which was applied for each pathogen. Also, an ordered logistic regression model was applied to the variable response co-infection (i.e. exposure to more than one VBP) and the same potential risk factors were considered for single pathogens. The variable co-infection was measured on an ordinal scale as it considers the number of infections present simultaneously in a cat under investigation (no infection, one infection, ≥ 2 infections). P-values with an odds ratio (OR) and 95% confidence interval (CI) of multivariable analyses were obtained. An OR value > 1 implies a positive association between independent and dependent variables, while an OR < 1 implies an inverse association.
One-hundred and thirty-four cats (68%) were seropositive to two or more pathogens with a significant difference between the two groups (75.4% in SG and 60.3% in CG) (Fisher's exact test: P = 0.0437, OR = 2.01, 95% CI: 1.06-3.82). Antibody prevalence concerning the pathogens under consideration is reported in Table 4. Prevalence of anti-R. conorii antibodies was significantly higher in SG cats (Fisher's exact test: P = 0.0094, OR = 2.30, 95% CI: 1. 24-4.27). High titers were detected against B. henselae, R. conorii, and E. canis in 10.7, 10.1 and 3.5% of tested cats, respectively, and no significant difference was found between the two groups when comparing the number of cats with high titers. Antibody prevalence to at least one pathogen was 90.8% in young cats (69/76) and 85.9% in adult cats (104/121) with no significant difference between groups.

Molecular assays
Positive PCR tests were obtained for Bartonella spp. (21.3%), Mycoplasma spp. (18.3%) and L. infantum (6.6%) ( Table 5) but they were negative for Ehrlichia/ Anaplasma spp., piroplasmids (Babesia spp. and Theileria spp.), Rickettsia spp. and Hepatozoon felis. The following species were sequenced ( There was no significant difference for molecular positivity rate to at least one pathogen according to gender, age and region. Leishmania infantum DNA was amplified in some cases from two specimens/cat as follows: blood and lymph node (n = 2), blood and urine (n = 1) or conjunctival and oral swabs (n = 1). Parasitic load for L. infantum ranged from 1 to 80,000 Leishmania/ml detected in blood EDTA, from 1 to 11,000 Leishmania/specimen in lymph nodes, from 7 to 120 Leishmania/specimen in conjunctival swabs, from 16 to 92 Leishmania/specimen in oral swabs, and from 1 to 30 Leishmania/ml in urine.

Overall exposure prevalence
An overall prevalence of exposure was calculated for pathogens investigated by both serological and molecular methods. Overall, Bartonella spp. exposure was 48.7% (96/197) and a higher prevalence was observed in FIV positive cats (Fisher's exact test: P = 0.0002, OR = 17.07, 95% CI: 2.20-132.65). Overall, L. infantum positivity rate was 14.7% (29/197), but no significant difference was observed between SG and CG or according to age, gender or FIV positivity.
One-hundred-eighty-two cats (92.4%) were found positive (by IFAT and PCR) to at least one VBP. The SG cats had higher (96.3%) exposure prevalence to at least Table 4 Overall antibody prevalence and titer range in the study (SG) and control (CG) groups  Two or more co-infections were detected in 62 cats (31.5%). In SG, there was a higher prevalence (37.3%) of multiple positivity compared to CG (19%) (Fisher's exact test: P = 0.0132, OR = 2.53, 95% CI: 1.23-5.20]). Conversely, no difference was found in FIV positive cats compared to FIV negative cats.

Multivariable logistic regression analysis
Seventeen variables that showed a significative difference in prevalence of one or more VBP between SG and CG with univariate analysis entered the multivariable logistic regression analysis performed on the 197 cats. Significant associations were found for exposure to L. infantum, Bartonella spp., Mycoplasma spp., B. microti, and A. phagocytophilum (Table 6) and co-infections (Additional file 3: Table S3). The significant associations concerned adult age and hemoplasma infection, FIV positivity and Bartonella spp. or hemoplasma infection, anaemia and hemoplasma infection, elevated ALP serum activity or low albumin concentration and A. phagocytophilum antibody positivity, high globulin concentration and Bartonella spp. infection, and high creatinine concentration with hemoplasma infection. Moreover, adults had a lower risk of B. microti antibody positivity compared to young cats; elevated serum CK activity was less likely in cats with L. infantum, hemoplasma, coinfections, or B. microti antibody positivity; cats with Bartonella spp infection were less likely to display elevated ALP serum activity.

Discussion
This controlled field study examined the prevalence of selected VBP and the clinical and clinicopathological abnormalities in cats exposed to ectoparasites. Moreover, some risk factors for VBP exposure were identified using significant associations detected by multivariable logistic regression analysis between some of the investigated pathogens and independent variables.
The study was based on both antibody and molecular detection of feline VBPs under consideration to increase the possibility of evaluating exposure of cats to tested pathogens. We found that cats were extremely exposed to VBP because of the high antibody (88.3%), molecular (40%), and overall (antibody and PCR) positivity (92.4%) to at least one pathogen. Moreover, cats exposed to ectoparasites, because of their outdoor lifestyle and the lack of regular ectoparasiticide treatments (SG), showed a significantly higher molecular and overall (antibody and PCR) positivity compared to indoor cats with no ectoparasites and subjected to regular application of ectoparasiticides (CG). Exposure to multiple VBP was also very frequent, as about two-thirds of tested cats were antibody positive, 10% PCR positive and 30.1% antibody and PCR positive to more than one VBP at the time of sampling; the difference between SG and CG was significant for antibody or antibody and molecular positivity. Epidemiological similarities shared by some VBP can obviously be responsible for vector-borne co-infections but also other factors, such as pathogenic interactions between them, can delay or prevent clearance of VBI and concur to co-infections. Other studies detected a high overall blood PCR positivity to at least one VBP, i.e. 25-29.9% in Portugal [4,17], 48.9% in northern Italy [18] and 45.4% in Cyprus [3]. Prevalence of co-infections was lower (2.2-9.8%) in those studies [3,4,17,18].
A small percentage of enrolled cats (12%) were antibody negative to all of VBP, and this percentage was significantly higher in the CG. However, we found a significant percentage of CG cats positive to the same VBP detected in the SG, and this may be explained by lack of compliance to ectoparasiticide treatment, low efficacy of the used ectoparasiticide, and/or non- vectorial transmission of the pathogen. For instance, in the case of L. infantum infection, pyrethroids are used in dogs for sand fly bite prevention, but almost all of these compounds are toxic to cats, and only a collar containing flumethrin and imidacloprid was able to reduce the incidence of L. infantum infection in cats [19][20][21]. Blood transfusion is a main non-vectorial transmission route of feline VBI but, at least for hemoplasmas, other routes are strongly suspected [22][23][24]. Moreover, our model of multivariable analysis did not find significant associations between exposure to any individual or multiple VBP and outdoor lifestyle or individual ectoparasiticide treatment. Attipa et al. [3] also used multivariable logistic regression to investigate risk factors of some VBPs detected in cats from Cyprus and, similarly, they did not find any association between positivity to any tested VBP and the lack of ectoparasiticide use while they found an association of outdoor lifestyle only with hemoplasma positivity [3]. Based on antibody detection, R. conorii (or other cross-reacting Rickettsia species) was the most frequent agent circulating among tested cats (48.7%), and it was significantly more common in cats exposed to ectoparasites (SG). Rickettsia conorii is historically the most important zoonotic species of the Rickettsia genus in the Mediterranean area, and was recently confirmed as a possible causative agent of acute febrile illness in dogs showing a transient positive blood PCR and seroconversion [25]. Studies about the infection of cats with Rickettsia spp. of the Mediterranean spotted fever group reported the collection of infected ticks on cats, and obtained positive blood PCR and a high antibody prevalence [10,17,[26][27][28]. These data support the need for prospective investigations about the pathogenic role of spotted fever group Rickettsia spp. in cats.
Antibody prevalence for B. henselae was high (45.7%), as was the seroreactivity for A. phagocytophilum (26.9%), B. microti (20.3%) and E. canis (16.2%) antigens. In many previous serological studies performed by IFAT, antibody titer against tested VBP was not reported, or they were low [6,10,29,30]. Interestingly, we detected high titers against B. henselae (10.7%), R. conorii (10.1%) and E. canis (3.5%) antigens; however, we cannot exclude the possibility of serological cross-reactions with other species of the same genus. This is most important for R. conorii and E. canis because we documented the exposure of cats by serology, but we were not able to find the pathogen DNA in blood as shown in other studies [31,32]. Similarly, we did not detect DNA of Anaplasma, Hepatozoon and piroplasmids such as Babesia or Cytauxzoon. This might be a consequence of the lack of exposure, clearance or sequestration of the organisms in other tissues, or technical limitations. However, in a study performed on blood collected from outdoor cats from a confined area of Sicily, only Hepatozoon felis DNA was detected in one cat (0.3%). Therefore, it is likely that some VBPs are not common in Sicily and the Calabria region [33]. Anaplasma/Ehrlichia DNA detection is rare in cats, and more frequently A. phagocytophilum is amplified [4,17,31,34], but A. platys or A. platys-like organisms were also occasionally characterised in cats in southern Europe and the seropositivity detected in this study could be due to different species of Anaplasma [3,35]. However, Ehrlichia spp. and Anaplasma spp. were not found in other studies in Greece and Spain [36,37].
Hepatozoon felis distribution seems to be quite variable in southern Europe. In fact, it is very rare in Italy, but a focus was recently detected in Matera, where single cases of Hepatozoon canis and Hepatozoon silvestris infections were also seen [10,33,38]. Hepatozoon felis is reported in the Iberian Peninsula with a prevalence range of 1.6-13.8% [4,17,31,39] and the highest molecular prevalence (37.9%) was recently found in Cyprus [3].
Feline piroplasmid infection is uncommon in Europe. Piroplasmid infection is caused by Cytauxzoon spp. [39][40][41][42][43], Babesia vogeli, or Babesia canis [4,17]. In Italy, DNA of B. microti was sequenced in cats from Sicily and Milan but, despite 20.3% antibody prevalence for B. microti obtained in this study, we were not able to detect piroplasmid DNA, and we cannot exclude that other Babesia spp. elicited antibody production in cats [44,45]. Interestingly, adult cats were less likely to have antibodies to B. microti compared to those < 1 year of age. Young cats might therefore be more susceptible to Babesia spp. infection than adults but the antibody response would not persist possibly because of clearance of the infection.
Bartonella spp. was VBP with the highest overall (48.7%) and molecular (21.3%) prevalence in this study, and molecular prevalence was significantly higher in SG. A significant association between Bartonella spp. exposure and FIV infections or increased globulin values were found by multivariate analysis as previously reported by univariate analysis only for globulins [46] but not for FIV [3,5]. A very high B. henselae antibody prevalence (45.7%) was also found. Two species were sequenced, and B. henselae was more prevalent (15.2%) than B. clarridgeiae (6.1%). These results confirm data obtained in 42 cats from the same area carrying ticks or fleas at the time of examination, where blood PCR positivity for Bartonella spp. was 38.1%, and B. henselae (21.4%) was more frequently detected than B. clarridgeiae (16.6%) [10]. A higher molecular prevalence was previously found in Sicily in a study using a nested-PCR where positivity was 70.6% in blood, 72.9% in lymph node aspirates, and 60.0% in oral swabs and B. henselae was the only sequenced species [8]. In other European countries, lower blood PCR prevalences were reported ranging 1-22.4%, and this might be due to the exposure of investigated cats to fleas or the assays used [3-5, 28, 31, 36, 37]. However, also in these latter studies, B. henselae was usually more frequent than B. clarridgeiae and only in one cat was B. kholerae DNA was sequenced [37].
As in other studies performed in Italy [10,18], Cyprus [3], and Portugal [16,47], we frequently detected hemoplasma DNA in cat blood (18.3%) and "Ca. Mycoplasma haemominutum" was more commonly sequenced, compared to "Ca. Mycoplasma turicensis" and M. haemofelis. Other studies reported lower prevalence (range 7.8-14.9%) but similar species were sequenced and represented possible co-infections [12,36,48,49]. We obtained significant associations by multivariate analysis between hemoplasma positivity and anaemia (OR= 2.76), adult age (OR= 4.53), FIV positivity (OR= 5.07), and increased creatinine values (OR=7.68). Interestingly, hemoplasma positive cats have seven times higher risk for an increased creatinine concentration and this association was never found before by multivariate analysis. We performed a cross-sectional study and therefore did not have the possibility to confirm chronic kidney disease (CKD) in cats with high creatinine; however, in two-thirds of cats urinalysis showed inappropriate urine specific gravity that is suggestive of CKD. A causative role for variables cannot be assessed by cross-sectional investigations, and prospective surveys on hemoplasma long carrier cats should be considered. Older age of hemoplasma-positive cats was also found in two other studies using multivariable logistic analysis, and could contribute to the association between hemoplasma positivity and high creatinine concentrations [3,49]. Hemolytic anaemia is the main pathogenic effect of M. haemofelis and less frequently of other hemoplasmas, but subclinical carriers can be found, and this may explain when in other studies multivariate analysis did not find associations between hemoplasmas and anaemia [3,49]. Conversely, significant association with FIV was previously reported by other authors using multivariable analysis, and this comorbidity could be due to epidemiological factors (sharing of the way of transmission) or facilitation of long-term hemoplasma infections in FIV positive cats [3,49,50].
Feline L. infantum infection can be considered an emergent VBI in endemic areas of canine leishmaniosis [20]. Many studies evaluated antibody and or molecular prevalence in southern Italy, and a wide prevalence range was found by both antibody detection (2.4-59%) and blood PCR (7.1-61%) [10,33,[51][52][53]. Epidemiological (endemicity, characteristics of the studied population) and technical (serological cut-off, molecular technique) differences may account for this variability. Antibody prevalence obtained in this study was within the above range (9.6%), but blood DNA detection was as low as 2.0%. Low molecular prevalence was also found by non-invasive samplings as conjunctival (1.5%) and oral (1.5%) swabs, or by lymph node aspirate (1.7%) or urine (2%). Interestingly, high parasite loads were obtained only from blood (up to 80,000 Leishmania/ml), and the clinical relevance of this finding lies in the risk of iatrogenic transmission of L. infantum by blood transfusion as reported in dogs [54,55].
Finally, the univariate analysis pointed out some clinical and clinicopathological abnormalities that were significantly more frequent in outdoor cats exposed to ectoparasites than in indoor cats protected from ectoparasites despite the former being significantly younger and less frequently admitted for health problems. Other infectious and parasite agents in addition to VBP frequently affect outdoor cats and may be responsible for this occurrence [28,36,[56][57][58]. The only biochemical abnormality more frequent in indoor cats was increased creatinine, but these cats were significantly older than those of SG, and this bias could influence the result. Indoor cats were also significantly more overweight compared to outdoor cats and therefore are predisposed to metabolic or urinary problems [59][60][61]. The clinical relevance of these findings is that lifestyle significantly influences the health of cats and adequate preventative measures have to be tailored accordingly.

Conclusions
A very high prevalence of zoonotic VBP exposure was found in cats, with Rickettsia spp. and Bartonella spp. being the most prevalent. Overall, cats exposed to ectoparasites for the lack of preventative measure and/or an outdoor lifestyle had a higher risk for VBI and coinfections and for some clinical and clinicopathological abnormalities. Some risk factors were documented by multivariable logistic regression analysis providing a better understanding of the epidemiology for selected feline pathogens, namely for Mycoplasma spp. and Bartonella spp. infections that were both found associated with FIV.
Also, the lifestyle of cats appeared clinically relevant for diseases other than those associated with VBP exposure and requires specific preventative measures to protect their health.

Additional files
Additional file 1: