|Home | About | Journals | Submit | Contact Us | Français|
The primary objective of this study was to determine the prevalence of subclinical hemotropic mycoplasma (HM) infections in 2 distinct feline populations: cats from a local shelter and client-owned cats presented for elective procedures (vaccination, ovariohysterectomy, orchiectomy) at the Western College of Veterinary Medicine — Veterinary Teaching Hospital (WCVM-VTH). The second objective of this study was to evaluate the inter-test agreement of 2 independent conventional polymerase chain reaction (PCR) assays used for the diagnosis of feline HM-infections.
Fifty-eight clinically healthy shelter cats and 57 clinically healthy client-owned cats were screened for subclinical HM-infection using a conventional PCR assay to detect the 16S rRNA of Mycoplasma haemofelis and “Candidatus M. haemominutum.” All cats in both groups had normal physical examinations. Sex, age (estimated for shelter cats), breed, reproductive status and the presence or absence of ectoparasites were determined. Packed cell volume (PCV), total protein, retroviral status, and blood smear evidence of HM-infection were evaluated. Subclinical HM-infection was identified by PCR assay in 12% (7/58) of the shelter cats and 4% (2/57) of the client-owned cats. M. haemofelis was found in 3/7 HM-infected shelter cats and 2/2 of the HM-infected client-owned cats; “Candidatus M. haemominutum” was found in 4/7 of the HM-infected shelter cats. There was no significant difference in prevalence of HM-infection between the populations (OR 3.8, 95% CI 0.75 to 19, P = 0.16), and no risk factors for infection were identified in either population.
Blood samples from 44 cats with known PCR results (26 cats sampled in the prevalence study and 18 clinical cases) were submitted to a second independent laboratory for HM PCR assay to assess inter-laboratory agreement. There was substantial, but not complete agreement between the 2 independent laboratories for PCR detection of M. haemofelis (κ = 0.66) and “Candidatus M. haemominutum” (κ = 0.70).
L’objectif premier de la présente étude était de déterminer la prévalence d’infection sousclinique associée au mycoplasme hémotropique (HM) dans 2 populations félines distinctes : des chats provenant d’un refuge local et des chats appartenant à des clients et présentés pour des procédures électives (vaccination, ovario-hystérectomie, orchiectomie) au Western College of Veterinary Medicine-Veterinary Teaching Hospital (WCVM-VTH). Le deuxième objectif de l’étude était d’évaluer l’accord intertest de deux épreuves indépendantes conventionnelles de réaction d’amplification en chaîne par la polymérase (PCR) utilisées pour le diagnostic d’infections félines à HM.
Cinquante-huit chats de refuge cliniquement en santé et 57 chats cliniquement en santé appartenant à des clients ont été éprouvés pour une infection à HM sous-clinique à l’aide d’une épreuve PCR conventionnelle pour détecter l’ARNr 16S de Mycoplasma hemofelis et «Candidatus M. haemominutum». L’examen physique de tous les chats dans les deux groupes ne révéla rien d’anormal. Le sexe, l’âge (estimé pour les chats de refuge), la race, l’état reproducteur et la présence ou l’absence d’ectoparasites ont été déterminés. L’hématocrite (PCV), les protéines totales, l’état rétroviral et une évidence d’infection par HM au moyen d’un frottis sanguin ont été évalués. L’infection sous-clinique à HM a été identifiée par épreuve PCR chez 12 % (7/58) des chats de refuge et 4 % (2/57) des chats de propriétaire. M. haemofelis a été retrouvé chez 3/7 des chats de refuge infectés par HM et 2/2 des chats de clients infectés par HM; «Candidatus M. haemominutum» a été trouvé chez 4/7 des chats de refuge infectés par HM. Il n’y avait aucune différence significative dans la prévalence d’infection par HM entre les populations (OR 3,8, 95 % CI 0,75 à 19, P = 0,16), et aucun facteur de risque pour l’infection n’a été identifié dans les deux populations.
Des échantillons sanguins provenant de 44 chats avec des résultats connus de PCR (26 chats échantillonnés dans l’étude de prévalence et 18 cas cliniques) ont été soumis à un deuxième laboratoire indépendant pour une épreuve PCR pour détecter HM afin d’évaluer l’accord inter-laboratoire. Il y avait un accord marqué mais incomplet entre les deux laboratoires indépendants pour la détection par PCR de M. hemofelis κ = 0,66) et «Candidatus M. haemominutum» (κ = 0,70).
(Traduit par Docteur Serge Messier)
Hemotropic mycoplasmas (HM), a subset of the Mycoplasma genus, are gram-negative, uncultivable, epierythrocytic parasites (1). Infection of cats with HM can result in subclinical disease or can cause severe hemolytic anemia (1). The development and severity of clinical disease in cats depend on host factors, the infecting species of HM, and potentially the infecting strain (1–4). There are currently 4 species known to infect cats: Mycoplasma haemofelis, “Candidatus Mycoplasma haemominutum,” “Candidatus Mycoplasma turicensis,” and “Candidatus Mycoplasma haematoparvum” (2). Testing is not available on a commercial basis for the latter 2 newly recognized species. The proposed routes of natural transmission include arthropod vectors (fleas and ticks) as well as saliva and feces (5–7). Currently, the diagnosis of HM-infection is based on either observation of the coccoid bacteria on erythrocytes during blood smear evaluation or polymerase chain reaction (PCR) assay. Polymerase chain reaction technology has been used to demonstrate a global distribution of HM-infection in domestic and wild felids (2–4,8–11). The prevalence of HM-infection in cats in Canada has not been reported. In addition to aiding in the diagnosis of feline HM-infection, PCR assays may also be useful for monitoring response to therapy.
Polymerase chain reaction assays are reported to offer increased diagnostic accuracy over traditional microscopic visualization and provide a valuable diagnostic option, particularly given the uncultivable nature of HM (12). Many consider PCR assay to be the “gold standard” for HM diagnosis; however, this diagnostic test has not been standardized among laboratories. Diagnostic sensitivity and specificity data have not been reported, and inter-laboratory agreement has not been evaluated. Polymerase chain reaction assays performed at different laboratories are often, incorrectly, assumed to be equivalent. Different primer sets may be used to identify the same organism by different diagnostic laboratories and this, combined with varying protocol, can lead to discordant results.
The primary objective of this study was to determine if there was a difference in the prevalence of subclinical HM-infection in cats from 2 distinct populations: shelter cats and client-owned cats presented for elective procedures (vaccination, ovariohysterectomy, or ochiectomy). We hypothesized that cats from the shelter population may be at higher risk of infection based on the presence of risk factors which have been previously identified for feline HM-infection (time spent outdoors, external parasites, positive retroviral status) (2,3,9). A secondary objective of our study was to evaluate the inter-laboratory agreement of 2 independent validated conventional PCR assays used at separate diagnostic laboratories for the diagnosis of HM-infection.
A convenience sample of 58 clinically healthy cats from the local animal shelter (Society for the Prevention of Cruelty to Animals — SPCA) and 57 clinically healthy, client-owned cats were selected for the study. The client-owned cats had been presented to the Veterinary Teaching Hospital at the Western College of Veterinary Medicine (WCVM-VTH) over a 6-month period during 2006 for routine physical examination and vaccination. The protocol for this study was approved by the Animal Care Committee, University of Saskatchewan, and written consent was obtained before sampling all client-owned cats. The following information was recorded for all cats: age (estimated based on dentition for the shelter cats), sex, neuter status, breed, and the presence or absence of ectoparasites determined following flea combing and otoscopic examination. Cats were grouped by age into 2 categories: < 2 years and ≥ 2 years. Cats were also categorized as purebred or as a domestic breed. For client-owned cats, information on their origin, access to outdoors, and number of cats within the household was also collected. Client-owned cats that originated from the SPCA or from households that had more than 3 cats were excluded from the study. All cats were assessed as clinically healthy, based on a routine physical examination.
Blood was collected from the jugular or saphenous vein for each cat. Fresh blood smears were made immediately to decrease the chance of detachment of mycoplasma organisms from the red blood cells (RBC) following exposure to ethylenediamine tetra-acetic acid (EDTA). Blood samples were collected into EDTA tubes and serum tubes. Blood in the serum tubes was spun within 2 h of collection, and the serum was harvested and stored frozen at −20°C in plastic tubes. Blood collected into EDTA tubes was used to determine packed-cell volume (PCV) and plasma total protein for each cat using microhematocrit tubes and refractometry, respectively. Serum (or plasma when serum was not available) from each cat was tested for feline leukemia virus (FeLV) antigen and feline immunodeficiency virus (FIV) antibody, using a commercially available enzyme-linked immunosorbent assay (ELISA) kit (Snap FIV Antibody/FeLV Antigen Combo; IDEXX Laboratories, Westbrook, Maine, USA) according to the manufacturer’s instructions.
Blood smear evaluation for all 115 study cats was performed by a senior clinical pathologist (board certified by the American College of Veterinary Pathologists) and a senior clinical pathology resident who were both blinded to the population source and PCR assay results for each cat. Samples were classified as positive on smear evaluation if there were ≥ 3 conclusive HM organisms present on examination of 10 to 100 × oil immersion fields (hpf), negative if no HM organisms were visible, and equivocal if non-conclusive inclusions were seen or if there were < 3 HM organisms present on all 10 hpf. The reported sensitivity and specificity of blood smear evaluation for detection of M. haemofelis are 0% and 98%, respectively, and for “Candidatus M. haemominutum” 10% and 87%, respectively (13).
Aliquots of EDTA blood from each cat (n = 115) were stored at − 20°C. Samples were submitted in batches to Prairie Diagnostic Services (PDS) (Saskatoon, Saskatchewan) for evaluation by a previously validated conventional HM PCR assay for both species (14).
Evaluation of the inter-laboratory agreement of 2 independent conventional HM PCRs involved submission of 44 aliquots of feline EDTA blood to a second independent laboratory, Animal Health Laboratory (AHL) (Guelph, Ontario) for analysis using a previously validated, comparable conventional HM PCR assay (15,16). The 44 samples submitted to AHL included: 26 samples collected from cats enrolled in the subclinical prevalence study where adequate sample volume remained to permit a second PCR assay, and samples collected from 18 additional cats. The 26 cats from the prevalence study included 6 cats that were positive on HM PCR assay at PDS (3 for M. haemofelis and 3 for “Candidatus M. haemominutum”), and 20 negative cases. The 18 additional cat samples included 5 cats that were positive for M. haemofelis and 6 cats that were positive for “Candidatus M. haemominutum” on PCR at PDS and 7 cats that were negative for both species on PCR at PDS.
The association between subclinical HM-infection (as determined by the PCR assay from PDS) and each of the following parameters was evaluated using a series of 2-tailed Fisher’s exact tests (Statistix 8; Analytical Software, Tallahassee, Florida, USA): population represented by the cat (shelter or client-owned), age (< 2 y or ≥ 2 y), breed (domestic or purebred), sex, neuter status, presence of external parasites, and retroviral infection status. The associations between HM-infection and PCV as well as total protein were evaluated using Wilcoxon rank sum tests (Statistix 8; Analytical Software). All associations where P < 0.05 were considered statistically significant. The inter-laboratory agreement for the 2 independent qualitative HM PCR assays as well as the agreement between blood smear evaluation and PCR assay were assessed using the kappa statistic (κ) (STATA 9.0; StataCorp, College Station, Texas, USA). A kappa value > 0.6 was interpreted as substantial agreement and kappa > 0.8 was interpreted as almost perfect agreement (17). The difference in the odds of a sample being classified as positive on 2 different tests was assessed using exact McNemar’s chi-squared tests (STATA version 9.0; StataCorp).
A summary of the signalment, hematologic, and diagnostic test results for the 115 cats (58 shelter and 57 client-owned) is presented in Table I. Based on PCR assay, the overall prevalence of subclinical HM-infection when both populations were considered together was 8% [95% confidence interval (CI) = 3.6% to 14.3%; 9/115]. The prevalence of subclinical HM-infection in the shelter cat population was 12% (7/58), compared with 4% (2/57) in the client-owned cat population. Three of the HM-infected shelter cats were infected with M. haemofelis and the other 4 HM-infected shelter cats were infected with “Candidatus M. haemominutum.” Both of the client-owned HM-infected cats were infected with M. haemofelis. The overall prevalence of M. haemofelis in both populations (based on PCR assay) was 4.3% (95% CI = 1.4 to 9.9%; 5/115).
There was no difference between the prevalence of HM-infection in the shelter cat population versus the client-owned cat population [n = 115, odds ratio (OR) = 3.8; 95% CI = 0.75 to 19, P = 0.16). There was also no significant difference in the prevalence of HM-infection between cats ≥ 2 y and cats < 2 y (OR = 3.3; 95% CI = 0.78 to 13.9; P = 0.15). None of the other factors evaluated, including sex (P = 0.48), neuter status (P = 0.99), breed (P = 0.23), presence of external parasites (P = 0.99), or retroviral status (P = 0.99) were associated with HM-infection. The median PCV was significantly lower in the HM-infected cats (median: 35%, range: 29% to 49%) than in the HM negative cats (median: 46%; range: 29% to 60%) (P = 0.01). However, none of the HM-infected cats was anemic (based on the PCV reference interval at PDS: 24% to 45%).
Prevalence of HM by blood smear evaluation was 0.9% (95% CI = 0.02% to 4.7%). Blood smear evaluation was significantly less likely to be positive for HM-infection (P = 0.008) than PCR assay [κ = 0.19, standard error (S) = 0.05].
Inter-laboratory results for the 2 PCR assays (n = 44) showed there was substantial, but not complete, agreement between the 2 independent HM PCR assays for M. haemofelis (κ = 0.66, S = 0.15) and “Candidatus M. haemominutum” (κ = 0.70, S = 0.15). The AHL PCR identified 6 positive samples for M. haemofelis and although the PDS PCR agreed for 5 of these 6 samples, an additional 3 samples were positive for M. haemofelis with the PDS assay. Similarly, the AHL PCR identified 7 positive samples for “Candidatus M. haemominutum” and the PDS PCR results agreed with 6 of these samples, but the PDS PCR also identified 3 additional positive samples. Though the PDS PCR assay was significantly more likely to be positive than the AHL PCR (P = 0.03), there was no significant difference in the odds of being positive between the two PCR tests for either M. haemofelis or Candidatus M. haemominutum (P = 0.63). The PCR assay used by the AHL detected co-infection in 2 cats, while the PDS PCR assay did not identify any co-infected cats.
The reported prevalence of HM-infection (4% to 30%) varies widely by geographic location (3,18,19). The prevalence of 8% identified in this study is higher than that reported in Ontario, Canada (4%). The relatively low prevalence in the current study, as well as the study in Ontario, might be expected because both studies looked at subclinical disease in a healthy cat population. The high prevalence of HM-infection at 30% was identified in a Spanish study of clinically ill cats (18). In the Kewish (14) study, used to validate a HM PDS PCR assay, HM-infection was documented in the same geographic region as the current study. In the Kewish (14) study, 38% (23/60) of cats were positive for HM on PCR testing. Of the cats that were tested in the Kewish (14) study, 30% (18/60) were suspected to have HM-infection based on blood smear examination and a regenerative anemia, and 72% (13/18) of those cats were subsequently positive for HM on PCR assay. A subpopulation of cats in the Kewish (14) study group with normal complete blood (cell) counts had a 10% prevalence of HM-infection (2/20), which is similar to the prevalence identified in this current larger study.
There was no detectable difference in the prevalence of infection between the 2 populations of cats (shelter versus client-owned) evaluated in this study. These 2 groups were compared based upon the hypothesis that risk factors for HM-infection were likely to be different between client-owned and shelter cats. The natural mode of transmission for HM is not known but is suspected to involve arthropod vectors as well as direct cat to cat transmission. Shelter cats are expected to have a higher prevalence of ectoparasitic infestations and have a higher rate of exposure to other cats. Client-owned cats acquired from the shelter or belonging to households with > 3 cats were excluded to maximize the difference in risk factors between the 2 populations. Though the percentage of HM-infected shelter cats (12%) was 3 times higher than the percentage of HM-infected client-owned cats (4%), the study did not have sufficient power, likely related to the relatively small sample size, to demonstrate a statistically significant difference.
Outdoor access has been previously associated with an increased risk of HM-infection in cats (3). All of the shelter cats presumably had outdoor access as a risk factor in their history. However, the assumption that all the shelter cats had outdoor access may have been incorrect, as their history was not known and some may have been surrendered indoor cats from private homes. Of the client-owned cats, 33% were known to have outdoor access, including the 2 client-owned cats that tested positive for HM.
Choice of a different breakpoint for age may have identified a significant association between age and HM-infection. In previous prevalence studies, older cats with an average age of 10 y are reported to be more at risk for presentation with clinical HM-infection (2). In the current study, the shelter cats were primarily young adults, which may have diminished the ability to detect age as a significant risk factor. Additionally, sample size may not have been adequate to demonstrate an age association.
None of the cats in the study had detectable flea or tick infestations, reflecting the low prevalence of these ectoparasites in this region. Ear mites, Otodectes cyanotis, were found in a number of shelter cats. As expected, Otodectes cyanotis infection was not associated with HM-infection as these mites do not engage in hematophagus activity, which is presumed to be needed to transmit HM.
Retroviral infection is an established risk factor for HM-infection and clinical disease (1). There were 2 FeLV positive cats identified in the study (one client-owned cat and one shelter cat) and both were negative on PCR for HM-infection. There was 1 client-owned FIV-positive cat and this cat was also negative on PCR for HM-infection. The lack of association between retroviral infection and HM-infection in this study likely reflects the small sample size and possibly the low prevalence of these infections in this area as well as the low prevalence in the study group related to the selection of clinically healthy cats.
Blood smear evaluation identified only 11% (1/9) of the HM-infections identified by PCR which is very similar to previous reports (13). Blood smear evaluation is reported to have relatively low sensitivity as demonstrated by a recent study that found the sensitivity of this technique to be 0.0% for M. haemofelis and 10.3% for “Candidatus M. haemominutum” compared to quantitative PCR (13). Reliance on blood smear evaluation for a diagnosis of HM will underestimate prevalence and lead to missed clinical cases.
A difference in PCV between HM PCR-positive and negative cats has been reported in some studies (2,13). The biological significance of the trend towards a lower PCV in the HM-positive cats is questionable in the current study as none of the cats had a PCV below the reference interval.
Inter-laboratory agreement for the 2 HM PCR assays was substantial but results were not identical. Polymerase chain reaction assays for the same organism often use different primers and are therefore not equivalent, as they target different non-conserved portions of the 16S rRNA. Prairie Diagnostic Services used the same primers as AHL plus one additional primer each for M. haemofelis and “Candidatus M. haemominutum.” The PDS laboratory was more likely to find positive M. haemofelis samples compared to AHL. The higher number of positive results at PDS may result from the additional primers, subtle differences in the assay, or may represent false positives. At the same time, AHL found dual infections where PDS found none. Once again, the use of additional primers and other subtle differences in the assay may have affected the ability of PDS to detect a dual infection or the dual infections may represent false positives at AHL. This study did not have sufficient data to calculate diagnostic sensitivity and specificity for the 2 PCR tests using either traditional methods or latent class models which do not require a gold standard to estimate test accuracy. Even if we could assume blood smear evaluation is a reasonable proxy for a gold standard, given that there was only one positive sample, meaningful estimations could not be generated. Different detection rates between PCR assays is relevant to patient care, as a practitioner may withhold treatment if HM-infection is not confirmed by testing. Also, in screening blood-donor cats for HM-infection, a positive cat may inadvertently be screened as negative. Different detection rates are also relevant from a research standpoint, as prevalence studies are more difficult to interpret if between test agreement is low and diagnostic accuracy is unknown.
The biggest limitation of this prevalence study was that it lacked power. Preliminary calculations to determine the number of cats to be included in the study were based on a pilot study (using the same PDS PCR assay as the current study) that showed a much higher prevalence of HM-infection in the cat shelter population than was identified in this study. One possible explanation for the difference in prevalence of HM-infections found between this study and the original pilot study would be a problem with the PCR assay. Specifically, it is possible that a problem with the assay may have led to either false positive results in the pilot study or false negative results in the present study. Although the differences were not statistically significant, the odds ratios and their associated confidence intervals for both population source and age do not rule out differences among these groups.
In conclustion, there was no detectable difference observed in the prevalence of subclinical HM-infection between the clinically healthy shelter and client-owned cat populations in this study. The power to detect differences between the 2 populations was low because the prevalence was lower than expected. There was substantial but not perfect agreement between the results of the HM PCR assays for the 2 laboratories accessed in this study. The need for continued attention to standardization of PCR testing is emphasized.
The authors thank Anju Tumber of Prairie Diagnostic Services, University of Saskatchewan PCR lab, Pat Bell-Rogers at Animal Health Laboratory, University of Guelph PCR lab, Dr. Ryan Dickinson of PDS, and Dr. Kathi Ellis of the Department of Veterinary Pathology, University of Saskatchewan for blood smear evaluation, as well as the Saskatoon SPCA, Pfizer, Purina, and Idexx for their support.
Dr. Nibblett was supported by a WCVM Interprovincial Graduate Student Fellowship and Companion Animal Health Fund Fellowship and the project was funded by the WCVM Companion Animal Health Fund.