ABSTRACT
Domestic cats have several features that make them ideal vehicles for interspecies transmission of influenza viruses; however, they have been largely overlooked as potential reservoirs or bridging hosts. In this study, we conducted serological surveillance to assess the prevalence of novel pandemic H1N1 as well as seasonal human influenza virus infections in domestic cats in Ohio. Four hundred serum samples collected from domestic cats (September 2009 to September 2010) were tested using a hemagglutination inhibition (HI) test. The seroprevalences of pandemic H1N1, seasonal H1N1, and H3N2 were 22.5%, 33%, and 43.5%, respectively. In addition, a significant association between clinical feline respiratory disease and influenza virus infection was documented. In this sample of cats, the prevalence of pandemic H1N1 did not follow the seasonality pattern of seasonal H1N1 or H3N2 influenza, similar to observations in humans. Pandemic H1N1 seroprevalence did not vary in relation to ambient temperature changes, while the seroprevalence of seasonal H3N2 and H1N1 influenza viruses increased with the decline of ambient temperature. Our results highlight the high prevalence of influenza virus infection in domestic cats, a seasonality pattern of influenza virus infection comparable to that in humans, and an association of infection with clinical respiratory disease.
INTRODUCTION
Influenza A viruses (IAVs) belong to the family Orthomyxoviridae and infect both mammalian species (including humans) and avian species (29). Influenza A viruses are subtyped on the basis of the antigenic properties of their hemagglutinin (HA) and neuraminidase (NA) glycoproteins expressed on the viral surface. To date, 16 HA and 9 NA subtypes have been identified (6). Although IAVs demonstrate host restriction (9), interspecies transmission among avian and mammalian hosts has been reported (31).
Due to frequent cohabitation and interactions with humans and other animals, domestic cats are uniquely positioned to serve as reservoirs for influenza virus infection both within a household and within the larger farm or suburban environment. However, for many years felids were thought to be naturally resistant to influenza virus infection (8), and their role in influenza virus epidemiology has been somewhat overlooked.
Experimental infection studies in the early 1970s demonstrated that cats are susceptible to human H2N2 and H3N2 (17, 18, 22), seal H7N7, and avian-origin H7N3 viruses (8). However, influenza virus infection of cats was not further described until 2003, when anecdotal reports from east Asia attributed disease in cats, humans, and birds to highly pathogenic H5N1 (20). The susceptibility of cats to highly pathogenic H5N1 has been confirmed by experimental infections via intratracheal inoculation and ingestion (21). Recently, more attention has been brought to the epidemiology of IAVs in domestic cats and other companion animals after reports of natural infection of cats with pandemic H1N1 (25) and further confirmation of the susceptibility of cats to the virus by experimental infection (28).
Serological surveys of influenza virus infection in domestic cats are limited. In the 1970s, serological evidence of H3N2 subtype influenza virus infection in cats was reported (14, 18), and it was suggested that cats might be exposed to the virus during the human epidemics in the countries (India and Japan) where the studies were performed (7). Analysis of cat sera collected from 1999 to 2005 (16) and sera collected from 2004 to 2008 (19) in Italy using competitive enzyme-linked immunosorbent assay (ELISA) showed no evidence of IAV antibodies. However, in Japan, analysis of cat sera collected between 1997 and 2008 using hemagglutination inhibition (HI) and neuraminidase inhibition (NI) tests for the human and equine H3 influenza virus subtype showed a relatively low prevalence (3.8%) of human H3N2 influenza virus antibody (23). More recently, a serosurvey of cat sera collected during the 2009-2010 influenza season from four different states in the United States showed a higher prevalence of pandemic H1N1 (21.8%), seasonal H3N2 (25.6%), and H1N1 (41.9%) human influenza viruses (12).
Together, these earlier reports of feline infections with different influenza virus subtypes raise concerns that this animal species may play a role in the transmission of IAVs if viruses become endemic in feline populations (7), indicating the necessity to further monitor IAVs in cats. In this study, we conducted a year-long (September 2009 to September 2010) serological surveillance to determine the prevalence of the human influenza virus infections in domestic cats in Ohio. The data were analyzed on the basis of the sample collection month and the sex, age, and health status of the tested cats.
MATERIALS AND METHODS
Serum sampling.Serum samples (n = 400) were passively collected using excess material from clinical samples derived from domestic pet cats presented to The Ohio State University Veterinary Medical Center between September 2009 and September 2010. Both sexes were represented (61% male). The majority of cats were aged between 4 and 12 years (56.8%), and both cats of young age (0 to 4 years) and geriatric cats (over 12 years) were represented (22.2% and 21%, respectively). Most of the cats were presented for wellness care (67.8%), while the remaining cats had various presenting complaints, including signs of respiratory disease (10.8%). Sera collected from individual cats (n = 355) represented 88.8% of total samples tested, while some cats were sampled several times during the year. For cats yielding more than one sample, each specimen was considered independent if the samples were negative and taken at least 1 month apart. In cases of seroconversion, subsequent samples were not counted.
Viruses.Seasonal H1N1 (A/Ohio/K1130/06), H3N2 (A/human/Ohio/06), and pandemic H1N1 (A/OH/0925-1/09) viruses were obtained from the Ohio Department of Health (Reynoldsburg, OH), and the vaccine seed strain for the 2009 pandemic H1N1 (IDCDC-RG15, a pandemic H1N1 vaccine strain developed from A/TX/5/09 H1N1 virus) was provided by the Centers for Disease Control and Prevention, Atlanta, GA. Stocks of the viruses were prepared in eggs or Madin-Darby canine kidney (MDCK) cells and used in the experiments described below.
Pandemic H1N1- and seasonal H1N1- and H3N2-specific control serum production.Hemagglutinin subtype-specific sera were produced in cats by immunization. Seasonal H1N1 (A/OH/K1130/06) and pandemic H1N1 (A/OH/0925-1/09) viruses propagated in MDCK cells were inactivated using 0.1% beta-propiolactone (Sigma, St. Louis, MO). The inactivated viruses (0.5 ml cell culture fluid) were mixed with adjuvant (0.3 ml TiterMax Gold; Sigma, St. Louis, MO) and prepared as an emulsion. Three- to 4-year-old female, influenza virus-negative, specific-pathogen-free cats were inoculated intramuscularly with 0.8 ml of each adjuvanted inactivated antigen containing approximately 256 hemagglutination units of virus. Cats were booster immunized 3 weeks after the priming dose of antigen. Sera were collected 2 weeks following the booster. In addition, seasonal human H3N2-specific serum was obtained from an ongoing infection study and used at 106 50% tissue culture infective doses (TCID50s) of seasonal H3N2 virus (A/human/Ohio/06) to infect cats by the intratracheal route. Sera were collected at 8 days postinfection.
HI test.Serum samples were heat inactivated at 56°C for 30 min. Initially, a set of serum samples (n = 30) was treated with receptor-destroying enzyme (RDE; Denka Seiken Co., Ltd., Tokyo, Japan) as described previously (15) and compared side by side with untreated samples. Since we did not find a significant difference in HI results between enzyme-treated and untreated serum, the remaining samples were prepared by heat inactivation without RDE treatment to avoid dilution of the serum. Positive-control sera, prepared as described above, also showed no nonspecific cross-reactivity without enzymatic treatment.
Hemagglutination inhibition testing was carried out according to the World Organization for Animal Health manual with minor modification (13). Briefly, 2-fold serial dilutions of sera were mixed with 8 hemagglutination units of each virus. The HI reactivity was determined using a 1% suspension of turkey red blood cells. The reciprocal dilution of the sera that showed complete inhibition of hemagglutination was recorded as the HI titer. All negative-control sera showed a less than 2-log2-unit titer (or 1:4 in endpoint serum dilution) against all three specific antigens used in this study. Positive-control sera from the infected and vaccinated cats showed HI antibody titers ranging from 3 to 7 log2 units (1:8 to 1:256) and 6 to 11 log2 units (64 to 4,048), respectively. On the basis of the data from control sera, we used a conservatively high cutoff value, 4 log2 units (1:16) or greater, for a sample to be considered seropositive in surveillance.
To assess the specificity of HI test results, a virus neutralization (VN) test was performed using a set of samples that had tested positive to pandemic H1N1 (n = 2), seasonal H3N2 (n = 1), and seasonal H1N1 (n = 1) viruses with HI titers of ≥40 and a negative-control serum sample (n = 1). Each serum sample was tested against pandemic H1N1 and seasonal H1N1 and H3N2 viruses to evaluate nonspecific cross-reactivity among subtypes as described previously (27). Briefly, samples were tested in duplicate, where a constant amount of virus (100 TCID50s/25 μl) was mixed with an equal volume of 2-fold serial dilutions of the serum samples for 30 min at 37°C and the mixture was then applied onto MDCK cells grown in a 96-well plate. Plates were observed for cytopathic effect for 4 days postinoculation to determine the virus neutralization titer. Neutralization titers were expressed as the reciprocal serum dilution giving a 50% reduction of the cytopathic effect.
Statistical analysis.Univariable and multivariable linear and logistic regression and multinomial logistic regression analyses were conducted to determine the association between seropositivity (the response variable) and the following variables of interest: cat age, sex, health status, and the ambient temperature changes over the study period. Analyses were conducted using STATA (version 10.1) software (Stata Corp., College Station, TX) as described by the UCLA Academic Technology Services, Statistical Consulting Group (1). To estimate the correlation between the HI titers and VN titers, the Spearman rank correlation coefficient was calculated (5).
RESULTS
In total, 400 serum samples were collected from domestic cats over 1 year (average, 30.2 ± 10.4 samples/month). The HI titers of samples ranged from 0 to 128, and samples showing HI antibody titers of 16 or greater were considered positive to determine the prevalence rate. Of these tested serum samples, 153 samples (38.3%) were negative for all 3 influenza virus subtypes tested and 130 (32.5%) tested positive to only one influenza virus subtype. Eighty-five (21.3%) and 32 (8%) cats concurrently tested positive to two and three influenza virus subtypes, respectively. Of the seropositive cats, 22.5% were seropositive against pandemic H1N1, 33% were positive against seasonal H1N1, and 43.5% were positive against seasonal H3N2. The average log2 HI titers for the pandemic H1N1, seasonal H1N1, and seasonal H3N2 viruses were 4.1 ± 0.6, 4.3 ± 0.5, and 4.8 ± 0.4, respectively.
The subtype-specific VN titers of the selected samples against the tested strains were ≥4 log2 units (range, 4.5 to 5 log2 units), which correlated with the HI titers (correlation coefficient, 0.89; P = 0.01), and minimal nonspecific cross-reactivity was observed among subtypes (VN titer, ≤1 log2 unit).
Univariable analysis indicated that age and sex were nonsignificant predictors for influenza virus seropositivity in cats or any other variables of interest (logistic P > 0.2; Fig. 1); however, those variables were kept in the multiple-regression models due to biological relevance and to control for possible confounding effects. Multivariable analysis of the log2 HI antibody titers controlling for sex, age, sampling time, and health status interactions demonstrated a significant positive correlation between the seasonal H3N2 and the pandemic H1N1 antibody titers (correlation coefficient = 0.48; 95% confidence interval [CI] = 0.39 to 0.56; generalized linear model P < 0.001). To determine the association between predictor variables of interest and the probability of testing positive to influenza virus, multivariable analyses were conducted using the HI titers as binary data (influenza virus positive/negative). The local (city of Columbus, Ohio) average monthly temperature as a continuous surrogate parameter for seasonality and the assigned health status of the cats were strongly associated with the probability of testing HI positive to at least one influenza virus subtype. Comparing pandemic and seasonal influenza viruses, multivariable analysis indicated that temperature had a different association with each influenza virus subtype. Controlling for the other variables, seropositivity against seasonal H3N2 and H1N1 influenza viruses in cats increased as ambient temperature decreased (odds ratio [OR] = 0.95, 95% CI = 0.93 to 0.97, and logistic P < 0.001 and OR = 0.98, 95% CI = 0.96 to 1.0, and logistic P < 0.05, respectively). In contrast, the rate of pandemic H1N1 seropositivity did not vary in relation to the ambient temperature changes (OR = 1.0, 95% CI = 0.99 to 1.0, logistic P = 0.2; Fig. 2).
Seroprevalence of pandemic H1N1, seasonal H1N1, and seasonal H3N2 influenza viruses in different cat age groups.
Seroprevalence of human influenza virus infections in domestic cats over 1 year in relation to monthly average ambient temperature.
The cats in this study were categorized with one of three different health statuses on the basis of their owners' initial presenting complaint and veterinary assessment. Sera were collected from 271 cats with no presenting complaint, 43 cats with signs of respiratory disease, and 86 cats with other nonrespiratory signs. The seroprevalence of influenza virus among the cats with both respiratory and other disease conditions compared to healthy cats is shown in Table 1. Multinomial logistic regression analysis indicated that cats with respiratory disease and cats with nonrespiratory diseases had significantly higher probabilities than healthy cats of testing positive to at least one of the tested influenza viruses (OR = 7.4 and 95% CI = 2.8 to 19.8 and OR = 2.0 and 95% CI = 1.2 to 3.5, respectively; multivariable logistic P < 0.01; Table 2). No differences between pandemic and seasonal influenza viruses were identified when assessing the probability for each subtype.
Prevalence of influenza A virus infection in domestic cats in relation to sex and health status
Estimated probabilities of influenza virus infection in relation to health status of tested cats
DISCUSSION
After the emergence of highly pathogenic H5N1 and the 2009 pandemic H1N1 viruses, reports of serious respiratory disease in domestic cats have been documented (10, 11). Since domestic cats are in close contact with the human environment, they are subsequently exposed to human influenza viruses (7); therefore, the possibility of cats being infected with influenza viruses exists.
In this study, we conducted serosurveillance of the 2009 pandemic H1N1 and seasonal human influenza viruses over a period of 1 year. The prevalence rates of the pandemic H1N1 and the seasonal H1N1 and H3N2 antibodies were 22.5%, 33%, and 43.5%, respectively. In a recent serosurvey of samples collected from the South and Midwest United States (12) that was conducted for a short period both during and after the wave of pandemic H1N1 in humans, a similar prevalence of the pandemic H1N1 (21.8%) and seasonal H1N1 (41.9%) but a relatively lower prevalence of the seasonal H3N2 (25.6%) compared to the rates in our study was observed. This finding may be explained by differences in sample size, geographical area, and timing of sample collection, where our study included samples collected during the pandemic wave in the Midwest and extended for 1 year after. Another study conducted in Japan detected only a 3.8% prevalence of the human H3 subtype infection in cats; however, the authors acknowledged that the strain used in the study (A/Tottori/45989/97, H3N2) might be irrelevant to the recent H3N2 strains. Moreover, the sampling time was different, where the samples tested in the study were collected between 1997 and 2008 (23). In two other studies where competitive ELISA was used for serosurveillance (16, 19), no influenza virus antibody was detected at all. Although it may be a practical approach to use ELISA as a screening test to detect all type A influenza virus infections, we found that the sensitivities of two commercially available nucleoprotein-specific ELISAs were very low and correctly detected only positive-control sera with HI titers greater than 64 (data not shown). It appears that a competitive format of the ELISA resulted in a compromise between higher specificity and lower sensitivity. Considering the moderate HI titers (range, 0 to 128) that we observed in our surveillance samples, it is possible that failure to detect influenza virus infection in previous studies using competitive ELISA (16, 19) may be due to low sensitivity.
Although the majority of IAV subtypes in humans during the U.S. 2009-2010 influenza season and during the second wave of the pandemic H1N1 were pandemic H1N1 (99.8%), with very few seasonal H3N2 and H1N1 influenza viruses (0.2%) (3), this was not reflected in the cats in our study. Rather, we observed a higher prevalence of both seasonal influenza viruses than H1N1. The degree of exposure and susceptibility of cats to the pandemic H1N1 could be one explanation for the lower prevalence of pandemic H1N1 in cats than humans at the same point of time. Serological analysis of geographically representative samples based on the pandemic H1N1 waves in human populations can provide better information on the prevalence of the pandemic H1N1 in companion animal populations.
Age-based serosurveys in humans have shown strong antibody responses to seasonal influenza virus strains in school-age children and no cross-reactive antibodies to the pandemic H1N1. In contrast, preexisting cross-reactive antibodies to the pandemic H1N1 and an increase in prevalence with age were detected in older individuals (>60 years of age) (24). This observation may explain the higher incidence of pandemic H1N1 infection in young children and adults than the elderly (30). Although we observed a high prevalence of the seasonal H3N2 antibodies in the 0- to 2-year-old age group among the cats (65.2%), the analysis showed that neither age nor sex was predictive of infection with the tested influenza virus subtypes. The absence of variability among the cat age groups can be attributed to the shorter life span of cats compared to humans and thus the lack of previous exposure to the influenza viruses that are antigenically related to pandemic H1N1 viruses.
Seasonal human influenza viruses show a typical pattern in humans, where their transmission is dependent on temperature and humidity. However, the pandemic H1N1 influenza in the Northern Hemisphere in 2009 showed a different pattern, where it started earlier in the spring and peaked almost 2.5 months earlier than seasonal influenza (2), although in guinea pig models the virus showed dependence on both temperature and relative humidity (26). In this study, we noticed that the pandemic H1N1 infection in cats was not associated with changes in ambient temperatures, and we observed no significant differences in the prevalence of infection over the period of the study. This finding may be due to the introduction of the virus into a new host population with no previous exposure. Similar to the prevalence of seasonal influenza in humans, both H1N1 and H3N2 infections in cats were significantly associated with the decline of ambient temperatures; thus, a significant increase in prevalence occurred during the winter season (Fig. 1). The pandemic H1N1 seasonal pattern in humans was different during the 2010-2011 season, where its prevalence declined to 38%, and the seasonal H3N2 pattern was predominant (62%) and started to follow the seasonal influenza pattern (4). The possibility exists that a similar change in the pattern of pandemic H1N1 will occur as the virus becomes established in cats. Future seroepidemiological studies will explore whether the pandemic H1N1 outcompetes seasonal human influenza virus infections in the feline population.
The role of influenza viruses in respiratory disease of felids is not well understood (7). A recent report of pandemic H1N1 infection in a cat highlights the potential of influenza virus to be the causative agent of clinical respiratory disease (25). However, experimental infection of cats with the 2009 pandemic H1N1 induced mild to moderate respiratory signs. Our findings documented a strong association between presentation with clinical respiratory disease and seropositivity to at least one influenza virus subtype in cats. Also, the association of influenza virus infection in cats with other disease conditions was strong compared to that in the healthy cats, which may be explained by the immune status of the cats or by an initial influenza virus infection that was complicated by coinfections with other viral and/or bacterial pathogens (or vice versa). However, in healthy cats the 57.2% rate of seropositivity to at least one influenza virus subtype raises the concern that cats may serve as subclinical influenza virus carriers that may shed into the environment virus that can infect humans and/or other species.
In conclusion, we observed a high prevalence of the 2009 pandemic H1N1 as well as seasonal human H3N2 and H1N1 antibodies in domestic cats in Ohio. Moreover, our data show a correlation between influenza virus infection and clinical respiratory disease in cats. Since cats may be exposed to different influenza virus subtypes, including human- and avian-origin influenza viruses, their potential role in the epidemiology of influenza virus should be further investigated.
ACKNOWLEDGMENTS
We thank Kathy Smith and Kevin Sohner (Ohio Department of Health, Reynoldsburg, OH) and Ruben Donis (Centers for Disease Control and Prevention, Atlanta, GA) for providing viruses. We also thank Megan Strother and Sarah Nelson for technical assistance.
This research was partly supported by National Institutes of Health grant R21AI077787 and the Ohio Agricultural Research and Development Center Research Program.
FOOTNOTES
- Received 9 August 2011.
- Returned for modification 26 August 2011.
- Accepted 16 September 2011.
- Accepted manuscript posted online 28 September 2011.
- Copyright © 2011, American Society for Microbiology. All Rights Reserved.
