Use of combined conventional and real-time PCR to determine the epidemiology of feline haemoplasma infections in northern Italy

Abstract

Although knowledge of feline haemotropic mycoplasmas (haemoplasmas) has dramatically improved in recent years, some issues still remain to be elucidated. The aim of the current study was to evaluate the prevalence of feline haemoplasma infections in blood samples collected from cats in northern Italy. A convenience-sample of 307 cats (40 anaemic; 258 non-anaemic; nine with unknown haematocrit [HCT]) was investigated using polymerase chain reaction assays. Furthermore, the date of blood collection, signalment and clinicopathological data were retrospectively evaluated to assess predictors and risk factors for infection. Haemoplasma infections were highly prevalent in the sample investigated with an overall prevalence of 18.9% (95% confidence interval: 14.5–23.3%). The prevalence for the three feline haemoplasmas was 17.3% for ‘Candidatus Mycoplasma haemominutum’ (CMhm), 5.9% for Mycoplasma haemofelis (Mhf) and 1.3% for ‘Candidatus Mycoplasma turicensis’ (CMt). Feline immunodeficiency virus-positive status represented a risk factor for infection with an odds ratio of 4.19 (P=0.02). Moreover, a higher prevalence was observed in summer (odds ratio 1.78; P=0.04) which may be consistent with arthropod-borne disease transmission. Cats infected with Mhf showed significantly lower HCT (P=0.03), haemoglobin values (P=0.02) and red blood cell counts (P=0.04), lower mean corpuscular haemoglobin concentration (P<0.01) and higher white blood cell counts (P<0.01) when compared with non-infected cats.

Feline haemoplasmas, formerly known as Haemobartonella felis, are epierythrocytic parasites of domestic and feral felids. ^1,2 Until now, three different species of feline haemoplasmas have been characterised: Mycoplasma haemofelis (Mhf formerly H felis large form, Ohio variant), ‘Candidatus Mycoplasma haemominutum’(CMhm, formerly H felis small form, California variant) ^3–7 and the newly recognised ‘Candidatus Mycoplasma turicensis’ (CMt). ⁸ Recently, a fourth haemoplasma, called ‘Candidatus Mycoplasma haematoparvum-like’ (Mhp-like) has been identified in cats in the US but has not yet been fully characterised. ⁹

The inability to culture these pathogens in vitro has limited the possibilities of investigating the epidemiology and pathogenesis of these agents. However, the development of specific molecular techniques, such as polymerase chain reaction (PCR), has improved our ability to detect haemoplasma infections. Furthermore, conventional PCR (cPCR) and real-time PCR have enabled us to ascertain the prevalence of feline haemoplasmas, to investigate pathogenetic issues and to assess the effectiveness of antibiotic therapy. ^10–12 Unfortunately, study designs may vary greatly among the different epidemiological studies making it difficult to directly compare the results achieved. For example, the cat populations investigated may consist of a convenience sample, of healthy and/or sick cats, only anaemic cats or even anaemic cats suspected of having a haemoplasma infection. ¹³ In addition, the diagnostic sensitivity of the tests applied may vary; the sensitivity of PCR assays is superior to microscopical detection methods which may, in turn, lead to higher reported prevalences than previously published. ^4,10,14,15

Feline haemoplasmas have been reported in domestic cats in many different regions of the world. ^15–24 For the most part, the prevalence data refer to Mhf and CMhm while limited data have been published for CMt. In all cases, CMhm was more prevalent than Mhf, with co-infections being rare or very rare (ranging from 0.2 to 4.5%). ^9,15,24 The CMt sample prevalence observed varied greatly among countries ranging from 1.1% in Switzerland to 26% in South Africa. ²⁵ In recent studies, there seems to be widespread agreement that risk factors for haemoplasma infections in pet cats are represented by male gender, outdoor access and old age. ^{15,20,24,26,27} Conversely, contradictory results have been reported on whether concomitant feline immunodeficiency virus (FIV) and feline leukaemia virus (FeLV) infections could predispose to haemoplasma infections, favour the chronic carrier state or worsen the clinical course. ^{9,11,12,17,28–30} Overall, Mhf seems to exhibit the highest pathogenic potential whereas CMhm and CMt infections do not induce severe anaemia, although it has been suggested that co-infections, eg, with feline retroviruses, may result in significant anaemia. ^8,29

The transmission routes of feline haemoplasmas are still poorly understood despite the increasing availability of epidemiological data and data from experimental studies that have been carried out to specifically address this matter. Experimental transmission via intravenous, intraperitoneal and oral routes using infectious blood has been reported. ³¹ In addition, iatrogenic transmission by blood transfusion has been documented. ^24,32 Blood-sucking arthropods have been proposed as natural vectors for haemoplasma transmission between cats. Evidence of haemoplasma DNA in ticks and fleas collected from infected cats and in a species of unfed ticks (Ixodes ovatus) in Japan together with a higher prevalence of haemoplasma infections in warmer climates support the arthropod-borne transmission hypothesis. ^24,33–36 Nevertheless, attempts to experimentally transmit the infections by feeding cats infected fleas failed ³⁷ while transmission via haematophagous activity of fleas was not conclusive. ³⁵ A direct transmission of feline haemoplasmas is supported by the identification of CMhm and CMt DNA in the saliva of experimentally infected cats. ^8,36,38 Nevertheless, it has also been demonstrated that CMt is only transiently excreted in saliva during the early phase of infection. ³⁵ Furthermore, risk factors such as male gender, outdoor access of cats and a previous history of bite abscesses indirectly support the hypothesis that haemoplasma infection could be transmitted through cat bite. ^21,24,25,27

Presently, data regarding European prevalence of the three feline haemoplasmas are limited to the United Kingdom, Switzerland and Germany. ^15,24,25,30 With the exception of one study, which investigated Mhf and CMhm infections in a small sample of cats clinically suspected of haemobartonellosis in Spain and Portugal, ¹⁸ the present study investigates all three feline haemoplasma infections in cats in southern Europe (northern Italy) for the first time. The aim of this cross-sectional survey was (1) to establish the prevalence of the three feline haemoplasmas in blood samples collected from a convenience-sample of cats using cPCR and real-time PCR assays; (2) to consider any association between haemoplasma infections, patient characteristics and haematological value with statistical analysis and (3) to retrospectively identify potential risk factors, in particular seasonal variation, in order to add epidemiological information.

Materials and Methods

Case material

The present study used feline blood samples (ethylene-diamine-tetra-acetic acid [EDTA]-anticoagulated) from a convenience-sampled population. They had been obtained from cats seen at the Veterinary Teaching Hospital, University of Bologna, Italy, and underwent routine blood testing at the Veterinary Clinical Pathology Service, between 1 January and 31 December 2006. The samples were frozen daily at −20°C and stored until further analyses. Most of the samples had been collected from sick cats requiring haematological evaluation as part of a diagnostic profile. A minority of samples were collected from apparently healthy cats which were at the hospital for pre-surgical testing or a pre-anesthetic check. The samples collected from cats during repeated presentation were discarded (n=192). Only those samples with a minimum blood volume of 100 μl after routine haematological testing were included.

DNA extraction

DNA was extracted from whole blood samples after thawing at room temperature; the extraction was accomplished by using the QIAamp DNA blood mini kit (Qiagen, Milan, Italy) according to the manufacturer's instructions. DNA was eluted into 200 μl buffer AE and stored at −20°C until use.

cPCR and PCR sensitivity

The samples were initially screened with a cPCR assay using a primer pair previously described (outer forward: 5′-ATACGGCCCATATTCCTACG-3′; outer reverse: 5′-TGCTCCACCACTTGTTCA-3′). ¹⁸ The expected amplicon sizes were 595 bp for Mhf and CMt and 618 bp for CMhm, respectively. To confirm the results, PCR products from samples which were positive at the first round of PCR were diluted 1:10 and re-assayed with a semi-nested PCR using an inner reverse primer (5′-AGCTGCTGGCACATAGTT-3′); the latter primer was designed to anneal within a conserved sequence of the three feline haemoplasmas. The expected amplicon sizes of the semi-nested PCR products were 199 bp for CMhm and 176 bp for Mhf and CMt, respectively. The PCR mixture for both PCR assays included 2.5 μl 6X PCR buffer (Invitrogen, Milan, Italy), 2 mM magnesium chloride, 500 nM each primer, 250 nM Deoxynucleotide triphosphates (Promega, Milan, Italy), 1 U recombinant Taq Polymerase (Invitrogen, Milan, Italy) and 1 μl of DNA template to which molecular biology grade water (Eppendorf, Milan, Italy) was added for a final volume of 25 μl. The presence of amplifiable DNA and the absence of PCR inhibitors were confirmed by amplifying the feline 28S rRNA gene on all specimens (forward: 5′-AGCAGGAGGTGTTGGAAGAG-3′; reverse: 5′-AGGGAGAGCCTAAATCAAAGG-3′). Each PCR run included positive and negative controls represented by a plasmid containing the 595 bp PCR product of the Mhf 16S rRNA gene and molecular biology water, respectively. PCR was performed using an EP-gradient S thermalcycler (Eppendorf, Milan, Italy). The first PCR cycle included an initial denaturation at 94°C for 3 min followed by 40 cycles of 94°C for 30 s, 58.6°C for 30 s and 72°C for 45 s, and a final extension step at 72°C for 5 min. The semi-nested PCR differed only in the annealing temperature at 50°C and the extension step of 30 s. The 28S PCR cycle consisted of an initial denaturation at 94°C for 2 min, followed by 40 cycles of 94°C for 20 s, 50°C for 25 s and 72°C for 20 s, and finally 72°C for 5 min. The PCR products were evaluated after electrophoresis on a 1.5% agarose gel and gel staining with ethidium bromide.

For establishing the sensitivity of the PCR assays applied in this study, the first round PCR product was cloned into a plasmid vector using a commercial kit, following the manufacturer's instructions (StrataClone PCR cloning kit, Stratagene). After cloning, the plasmid DNA was quantified photometrically at 260 nm and the target number was estimated by using Oligonucleotide Properties calculator software. ³⁹ The plasmid standard was then 10-fold serially diluted in a canine genomic DNA solution tested to be negative for haemoplasmas by PCR. Serially diluted plasmids were assayed by cPCR in duplicate. The last dilution yielding a positive result was assumed to be the limit of detection.

Real-time Taqman PCR

An aliquot of each genomic DNA sample which had tested positive by semi-nested PCR was sent to the University of Zurich, Switzerland in cold packs for real-time PCR analyses. Real-time PCR assays for CMhm, Mhf and CMt were performed as described. ^8,24

Risk factors

The date of blood collection, signalment and clinicopathological data were retrieved by querying the electronic database (medical records) of the Veterinary Clinical Department, University of Bologna, Italy. Data were saved as an Excel dataset and managed accordingly.

Complete blood count (CBC)

CBCs were performed on 298 out of the 307 EDTA-anticoagulated blood samples using a Cell-Dyn 3500 system (Abbott Diagnostics, Milan, Italy). By using a haematocrit (HCT) cut-off value of 24%, 40 out of the 298 tested samples (13%) originated from anaemic cats and 258 (87%) were from non-anaemic cats.

FeLV–FIV test

The retroviral status was assessed by retrospective evaluation of the medical records from 1 January 2004 to 31 December 2006. As these analyses were only conducted if requested by the attending clinicians, FeLV/FIV data were not available for all the samples. Retroviral infections were assessed by using a commercial immunochromatographic test (Snap Combo Plus, Idexx, Milan, Italy) that detects specific viral antigen p27 for FeLV and specific antibodies raised against FIV gag and env proteins.

Statistical analysis

Potential risk factors were evaluated by calculating the odds ratio with a 95% confidence interval (CI). For categorical variables the χ² test (cell frequencies of >5) or Fisher's exact test (cell frequencies of ≤5) was used. In particular, for seasonal variation, the monthly prevalence data were grouped as follows: January, February and March: winter; April, May and June: spring; July, August and September: summer; October, November and December: autumn. For statistical purposes, seasons were considered as potential risk factors and odds ratios with 95% CIs were calculated and compared with the overall annual prevalence. Clinicopathological findings were found not to be normally distributed by Shapiro–Wilks' W test and were thus analysed by non-parametric statistical tests (Mann–Whitney U test). A P value <0.05 was considered statistically significant.

Results

Overall, 307 feline blood samples were included in the study. DNA extracted from blood samples were analysed by semi-nested PCR for the presence of feline haemoplasmas. The sensitivity of this assay was found to be 52 target copies/reaction. All positive and negative PCR reaction controls yielded positive and negative results, respectively. Of the 307 blood samples, 58 (18.9%; 95% CI: 14.5–23.3%) tested positive by semi-nested PCR; they were subsequently analysed using real-time PCR assays specific for CMhm, Mhf and CMt, respectively (Table 1). Fifty-three samples tested positive for CMhm (17.3%; 95% CI: 13.0–21.5%), 18 for Mhf (5.9%; 95% CI: 3.2–8.5%) and four for CMt (1.3%; 95% CI: 0–2.6%). Co-infections were observed with CMhm and Mhf and with CMhm and CMt, respectively, but not with Mhf and CMt (Table 1). No triple infections were observed. Complete PCR results are presented in Table 1.

Table 1.

Number and percentage (sample prevalence) of cats that tested PCR-positive for feline hemoplasmas

PCR-result/infection (n=307)	Number of infected cats	Observed prevalence (%)	95% CI
Semi-nested PCR	58	18.9	14.5–23.3
Real-time PCR
CMhm	53	17.3	13.0–21.5
Mhf	18	5.9	3.2–8.5
CMt	4	1.3	0.0–2.6
CMhm alone	36	11.7	8.1–15.3
Mhf alone	4	1.3	0.0–2.6
CMt alone	1	0.3	0.0–1.0
CMhm+Mhf	14	4.6	2.2–6.9
CMhm+CMt	3	1.0	0.0–2.1
Mhf+CMt	0	0.0
CMhm+Mhf+CMt	0	0.0

For all 307 cats included in the study, data on breed, gender, desexing status, age and date of blood collection were retrieved from the medical records (Table 2). Breed, gender, desexing status and age did not represent significant risk factors although a tendency was found in that male cats were somewhat more at risk of being infected than female cats (male: odds ratio 1.80, P=0.05; male entire: odds ratio 2.45, P=0.07; Table 2). Complete odds ratios and the respective P values are shown in Table 2. Some seasonal fluctuation of haemoplasma infections was observed. Infections were not uniformly distributed throughout the year. In particular, the observed prevalence was 29.3% and 8.5% during the summer and autumn months, respectively (Table 2). Cats which had been seen during the summer were significantly more frequently infected than those which had been seen in autumn. The monthly observed prevalence figures are shown in Fig 1

Table 2.

Sample characteristics: breed, gender, age, time point of blood collection, anaemia status and FeLV and FIV status of all cats and of cats testing positive or negative for haemoplasmas by semi-nested PCR, respectively

Variable	Total number (%)	PCR		Odds ratio	95% CI	P value
		Negative (%)	Positive (%)
Breed (n=307)
Non-pedigree	246 (81)	198 (64)	48 (16)	1.24	0.59–2.61	0.58
Pedigree	61 (19)	51 (17)	10 (3)	1.00
Gender (n=307)
Male	166 (54)	128 (42)	38 (12)	1.80	0.99–3.26	0.05
Female	141 (46)	121 (39)	20 (7)	1.00
Entire female	47 (15)	41 (13)	6 (2)	1.00
Neutered female	94 (31)	80 (26)	14 (5)	1.20	0.43–3.34	0.73
Neutered male	94 (31)	75 (24)	19 (6)	1.73	0.64–4.68	0.28
Entire male	72 (23)	53 (17)	19 (6)	2.45	0.90–6.63	0.07
Neutered (m+f)	188 (61)	155 (50)	33 (11)	1.00
Entire (m+f)	119 (39)	94 (31)	25 (8)	1.24	0.70–2.23	0.45
Age (n=307)
0–5 years	70 (23)	57 (19)	13 (4)	1.19	0.47–3.03	0.71
5–10 years	81 (26)	62 (20)	19 (6)	1.60	0.66–3.85	0.29
10–15 years	100 (33)	83 (27)	17 (6)	1.07	0.44–2.59	0.88
>15 years	56 (18)	47 (15)	9 (3)	1.00
Season (n=307)
Winter (Jan–March)		44 (83)	9 (17)	0.87	0.41–1.90	0.74
Spring (Apr–Jun)		82 (81)	19 (19)	0.99	0.56–1.79	0.99
Summer (Jul–Sep)		58 (71)	24 (29)	1.78	1.02–3.10	0.04
Autumn (Oct–Dec)		65 (92)	6 (8)	0.40	0.16–0.95	0.03
Anaemia status (n=298)
Anaemic	40 (13)	29 (72)	11 (28)	1.75	0.87–3.75	0.15
Non-anaemic	258 (87)	212 (82)	46 (18)	1.00
FeLV–FIV status (n=91)
FeLV-negative	79 (87)	57 (63)	22 (24)	1.00
FeLV-positive	12 (13)	10 (11)	2 (2)	0.52	0.10–2.56	0.41
FIV-negative	78 (86)	61 (67)	17 (19)	1.00
FIV-positive	13 (14)	6 (7)	7 (8)	4.19	1.24–14.12	0.02

P value in bold are statistically significant (P < 0.05).

Fig 1.

Monthly prevalence of feline hemoplasma infections. Left scale and right scale indicate observed prevalence and case numbers, respectively. Prevalence data categorized based upon months were graphed as bars. Dotted line represents the overall annual prevalence. Joined points represent case number (samples) tested in the corresponding month. Summer presentation (June, July and August) showed significantly higher odds ratio (OD 1.78; 95% CI: 1.02—3.10; P=0.04) when compared to the overall prevalence. Conversely, autumn presentation (October, November and December) showed significantly lower odds ratio (OD 0.40; CI: 0.16—0.95; P=0.03).

Results from CBCs were available from 298 blood samples. Forty cats were categorised as being anaemic (HCT<24%). The presence of anaemia was not significantly associated with haemoplasma PCR-positive status (P=0.15); among the 40 anaemic cats, 11 cats (27%) tested positive for at least one haemoplasma while 46 out of the 258 (18%) non-anaemic cats were positive (Table 2). A comparison of CBC data between PCR-negative and PCR-positive cats was carried out involving singly infected cats only, co-infected cats only or the total number of infected cats. Cats infected with Mhf (either alone or as co-infection) showed significantly lower red blood cell (RBC) counts, haemoglobin (HB), HCT and mean corpuscular haemoglobin concentrations (MCHCs) and higher white blood cell (WBC) counts than PCR-negative cats (Table 3). Cats infected with Mhf alone were compared with Mhf co-infected cats to investigate the effect of co-infections on haematological findings. The four cats infected with Mhf alone had lower, but not significantly different, HCT, HB, RBC counts and MCHC values when compared to Mhf co-infected cats (Table 3). CMhm-infected cats did not show haematological differences when compared with PCR-negative cats except for lower MCHC values (Table 3).

Table 3.

Selected haematological findings in subgroups of haemoplasma PCR-positive and PCR-negative cats

Variable – Reference range	Number of cats	Mean±SD	Median	Lower quartile	Upper quartile	Number of cats	Mean±SD	Median	Lower quartile	Upper quartile	P
	PCR-positive Mhf alone					PCR-positive Mhf co-infections
HB (gr/dl) – (8.0–15.0)	4	7.5±3.2	7.5	4.0	9.1	13	10.6±3.5	11.0	8.5	12.5	0.17
HCT (%) – (24.0–45.0)	4	22.5±8.5	22.9	12.5	26.9	13	30.1±9.9	30.5	23.1	35.6	0.21
RBC (mill/μl) – (5.0–10.0)	4	4.9±2.0	4.5	3.1	5.3	13	7.2±2.5	6.9	5.4	8.9	0.13
MCV (fl) – (39.0–55.0)	4	46.2±5.9	46.4	40.4	51.0	13	42.2±4.1	43.1	38.9	44.8	0.21
MCHC (%) – (30.0–36.0)	4	33.1±1.7	33.1	31.2	33.9	13	35.2±1.6	35.2	34.7	35.6	0.05
WBC (1000/μl) – (5.0–19.0)	4	12.7±3.1	12.4	9.3	13.2	13	20.6±13.2	21.7	9.9	27.3	0.40
	PCR-positive Mhf (including co-infections)					PCR-negative for all haemoplasma
HB (gr/dl) – (8.0–15.0)	17	9.9±3.6	10.8	7.1	11.5	241	11.9±3.1	12.5	10.2	14.2
HCT (%) – (24.0–45.0)	17	28.3±9.9	30.1	20.8	32.7	241	33.2±8.1	34.5	29.0	39.2	0.03
RBC (mill/μl) – (5.0–10.0)	17	6.6±2.6	6.6	4.4	8.2	241	7.9±2.0	8.2	6.8	9.3	0.04
MCV (fl) – (39.0–55.0)	17	43.2±4.7	43.1	39.3	45.1	241	43.0±6.0	42.1	39.3	45.5	0.74
MCHC (%) – (30.0–36.0)	17	34.7±1.8	35.0	33.7	35.5	241	35.9±1.6	36.2	35.3	36.8	<0.01
WBC (1000/μl) – (5.0–19.0)	17	18.9±12.1	14.9	9.9	25.4	241	12.0±7.6	10.1	6.9	15.1	<0.01
	PCR-positive CMhm alone					PCR-negative for all haemoplasma
HB (gr/dl) – (8.0–15.0)	36	12.4±3.2	13.2	10.2	14.6	241	11.9±3.1	12.5	10.2	14.2	0.20
HCT (%) – (24.0–45.0)	36	34.3±8.6	37.0	29.2	39.8	241	33.2±8.1	34.5	29.0	39.2	0.25
RBC (mill/μl) – (5.0–10.0)	36	7.9±2.0	8.4	6.8	9.5	241	7.9±2.0	8.2	6.8	9.3	0.74
MCV (fl) – (39.0–55.0)	36	43.8±4.8	43.8	40.4	45.6	241	43.0±6.0	42.1	39.3	45.5	0.17
MCHC (%) – (30.0–36.0)	36	36.4±5.6	35.4	34.7	36.2	241	35.9±1.6	36.2	35.3	36.8	0.04
WBC (1000/μl) – (5.0–19.0)	36	13.1±11.3	9.3	6.0	16.0	241	12.0±7.6	10.1	6.9	15.1	0.96

P value in bold are statistically significant (P < 0.05). HB = haemoglobin, HCT = haematocrit, RBC = red blood cell, MCV = mean corpuscular volume, MCHC = mean corpuscular haemoglobin concentrations, WBC = white blood cell.

The retroviral status was known for 91 out of 307 cats (29.6%). Among those 91 cats, 12 (13.2%) tested FeLV-positive and 13 (14.3%) tested FIV-positive. FIV- (P=0.02) but not FeLV-positive status (P=0.41) was significantly associated with haemoplasma infections (Table 2).

Discussion

The current study evaluated the prevalence of haemoplasmas in a convenience-sampled population of cats that were seen at the Veterinary Teaching Hospital, Faculty of Veterinary Medicine, Bologna, Italy. The limitations of convenience versus truly random samples have been discussed previously. ⁴⁰ Indeed, the convenience sample does not reflect the general Italian cat population and the actual prevalence cannot be understood by data from this study. Nevertheless, convenience samples usually represent a feasible way of sampling pet animals and, at the same time, represent the animal population which is seen by veterinarians.

Due to the retrospective nature of this study aimed at gaining information on haemoplasma epidemiology, the combined use of cPCR for screening and real-time PCR for typing isolates was adopted accordingly. The epidemiological data gained from the quantification of the haemoplasma load are not compelling in this context due to the demonstration that higher blood loads do not necessarily indicate recent infections ²⁴ while real-time PCR indeed saves extensive sequencing work. Conversely, real-time PCR as a screening tool in epidemiological surveys could be restricted with regard to its limitation of picking up slightly divergent haemoplasma strains which exhibit mutations in the probe target region ⁹ hampering the identification of new strains. Recently, a fourth feline haemoplasma, namely Mhp-like with a very low prevalence of 0.4% has been identified in the US. A discrepancy in results from cPCR and real-time PCR assays ⁹ led to its identification. A similar experimental design which included both conventional and real-time PCR was also carried out in Europe and enabled identification of CMt ^8,24 but not the Mhp-like organism. ²⁴ Furthermore, in the present study, we also found that all samples testing positive on cPCR, yielded positive results on real-time PCR. Thus, the presence of Mhp-like haemoplasmas is very unlikely in the sample investigated. Both techniques are similarly sensitive, although real-time PCR showed a slightly lower detection limit of 1–10 copies/reaction. ^9,24,41 However, in a recent haemoplasma study which compared results from cPCR and real-time PCR assays, all samples which tested positive in real-time PCR also gave positive results in cPCR. ⁹

Our study showed that haemoplasma infections are highly prevalent in a population of not only diseased cats but also healthy cats that were seen at a veterinary facility in northern Italy. CMhm showed the highest prevalence, followed by Mhf and CMt. Co-infections with Mhf and CMhm as well as with CMt and CMhm were common. Co-infections were more frequently detected than in other studies. ^9,15,24 When comparing prevalence data from studies with a similar sampling procedure, CMhm was similarly prevalent while the Mhf prevalence was clearly higher in this study than had been reported for other European countries and the US, and was comparable to the Mhf prevalence recently reported in Australian pet cats. ^{9,14,15,20,22,24} The CMt prevalence in this study was low similarly to that as reported for Swiss and UK pet cats, and clearly lower than in South African and Australian cats. ²⁵ Besides these continental differences, microclimate variations within the same climate area could influence the epidemiology of haemoplasma infection. Significantly, varying haemoplasma prevalences were reported in different areas within Switzerland and California. ^9,24 To explain such differences, a vector-mediated route of transmission has been suggested. Vector-borne diseases usually show seasonal variations. For instance, the flea-transmitted protozoan Trypanosoma microti showed seasonal fluctuations with peak prevalence in the field vole (Microtus agrestis) during the summer, strictly associated with the ectoparasite load. ⁴² Similarly, also in our study, a significantly higher prevalence of haemoplasma infections was found during the summer months. Conversely, in autumn, haemoplasma infections were detected less frequently. Our findings are consistent with the hypothesis that one of the natural means of transmission of haemoplasma infection is arthropod-borne transmission. Indeed, the DNA of haemoplasmas has been detected in fleas and the corresponding cats ²³ although experimental attempts to transmit the infection through the hematophagous activity of fleas have not been conclusive. ³⁵ Other vectors, ie, ticks, have been associated with the epidemiology of haemoplasma infection in Japan ³⁴ but these results could not be confirmed for Europe. ³⁶ Unfortunately, in epidemiological studies following the advent of PCR, prevalence data categorised according to seasonal variation are scarce. In Switzerland, seasonal prevalence fluctuations were not found (R Hofmann-Lehmann, personal communication, 2007) indicating that alternative means of transmission (eg, direct transmission resulting from aggressive interactions between hosts) may prevail in different areas. The behaviour responsible for direct transmission could not, however, explain a summer peak prevalence. Clearly, incidence data obtained in prospective case–control cohort studies could clarify this pivotal epidemiological issue.

Male and, in particular, intact male cats showed a tendency to have an increased risk of infection with an odds ratio of 1.80 (P=0.05) and 2.45 (P=0.07), respectively. Infections were most prevalent in cats 5–10 years of age and were progressively less detected in cats over 10 years of age. This finding differs from other studies which have reported increased prevalence in younger (≤3 years old) ²⁷ or older cats ^15,24 but are in accordance with others. ^26,43 The discrepancies between different studies may be due to different statistical methods (age as a continuous versus a categorical variable), different sample characteristics (the highly variable number of cats included in the studies must be taken into account, different inclusion criteria also influenced the results), different diagnostic techniques (blood smear examination or PCR assay) or a combination thereof. Another important reason for this discrepancy can be explained by the fact that, in cats testing positive, the stage of infection is not known. ¹³

Haemoplasma PCR-positive status was not associated with anaemia (P=0.15) as has also been shown by others. ²⁴ This is likely due to the higher prevalence of low pathogenic haemoplasmas in comparison to Mhf. In fact, CMhm PCR-positive cats without co-infections did not show haematological differences when compared to PCR-negative animals, except for a significantly lower MCHC (P=0.04) which is in agreement with other studies. ¹¹ Nevertheless, the MCHC median values of both groups were within the reference range and the data were widely overlapping. Conversely, statistically lower RBC counts (P=0.04), HB (P=0.02) and HCT values (P=0.03) were observed in the subgroup of all Mhf-infected cats (including singly and co-infected) when compared to PCR-negative cats. This subgroup also showed a significantly higher WBC count (P<0.01) most likely due to co-infections as either Mhf- and CMhm-singly infected cats had a WBC count nearly equal to PCR-negative cats (Table 3). In agreement with previous reports, the overall CBC data confirm the greater potential of Mhf in inducing anaemia when compared to other feline haemoplasmas. ^4,6,10,14,24

Both feline retroviral infections, FIV and FeLV, were similarly prevalent in the sample population of this study. The FIV prevalence in our study was lower than the 30.9% previously reported in an epidemiological survey carried out on sick cats in Italy while the data regarding FeLV prevalence are similar. ⁴⁴ It should be emphasised that our data regarding FIV/FeLV association with haemoplasma infection may have differed due to the retrospective nature of this study. In fact, the retroviral status was not investigated in all subjects as many data were retrieved retrospectively (2004–2005), thus a proportion of cats considered negative may actually have been positive.

Only FIV-positive status was significantly associated with haemoplasma infections. Interactions between the feline retroviruses and haemoplasma infections have been investigated both experimentally and in natural infections with discrepant results. ^{11–13,17,28,29,45,46} Indeed, FeLV/FIV testing in our study was performed on clinical request which was most likely based on a clinical suspicion in cats with advanced clinical stages of disease. A possible favouring effect of FIV-induced immunodeficiency in predisposing to haemoplasma infections or alternatively in favouring the chronic carrier states could be hypothesised. A similar transmission route or exposure to similar risk factors cannot be ruled out. The reasons for the low haemoplasma prevalence in FeLV-positive cats are unknown. FeLV antigenaemia should ideally be repeated after 3 months to detect cats that were only transiently antigenaemic. In all except one cat in our study, FeLV infection was assessed only once. Thus, it cannot be excluded that many cats in our study underwent a regressive infection or, alternatively, less pathogenic FeLV strains may be prevalent in Italy. In both cases, FeLV-positive status itself was not sufficient to predispose to haemoplasma infections.

In conclusion, our study shows that feline haemoplasmas are highly prevalent in a convenience-sample of cats that were seen at a veterinary facility in Italy. Furthermore, we found that the haemoplasma prevalence is higher during the summer, indirectly supporting a potential vector-borne transmission of these agents in Italy. Because of the high prevalence of haemoplasma infections found in the current study and the possibility of efficient treatment if the infection is recognised early, veterinarians should be aware of this infection.