Seroepidemiology of respiratory (group 2) canine coronavirus, canine parainfluenza virus, and Bordetella bronchiseptica infections in urban dogs in a humane shelter and in rural dogs in small communities

Abstract

This prospective study evaluated seroepidemiologic features of canine respiratory coronavirus (CRCoV), canine parainfluenza virus (CPIV), and Bordetella bronchiseptica infections in dogs in an urban humane shelter and in rural/small community dog populations in western Canada. Seroprevalence of CRCoV and CPIV was low compared with other countries; seroprevalence of B. bronchiseptica was moderate to high in most populations examined. Rural dogs were 0.421 times (P ≤ 0.0001) less likely to be positive for CRCoV than dogs admitted to the shelter. There were no statistical differences in prevalence of antibodies to B. bronchiseptica and CPIV between urban and rural populations. Dogs from Fort Resolution, NWT were significantly (P < 0.05) less likely to have moderate or high antibody titers to the 3 agents than dogs in the shelter. Seroconversion to CRCoV was common in dogs in the shelter, but was not associated (P = 0.18) with respiratory disease. Antibodies to CRCoV, CPIV, or B. bronchiseptica on arrival were not significantly (P > 0.05) associated with disease-sparing after entry into the shelter.

Introduction

Canine infectious tracheobronchitis (CITB, “kennel cough),” is a respiratory disease that occurs frequently when dogs are kennelled or brought together for reasons such as dog shows or sporting events (1). The disease is irritating but self-limiting and can progress to bronchopneumonia (1). In addition to infectious agents, this disease has multiple co-factors including crowding with high concentrations of pathogens and frequency of exposure, physiological stresses, and differences in host susceptibility, including immune status. Well-recognized infectious agents associated with CITB include canine adenovirus type 2 (CAV-2) (2), canine parainfluenzavirus (CPIV) (3), and Bordetella bronchiseptica (4). Less-frequently recognized agents include canine herpesvirus (5), and Mycoplasma spp. (6). In the last decade, canine influenza, thought to be derived from equine influenza, was implicated in respiratory disease cases in dogs in the United States (7), but a recent survey indicated a low seroprevalence of that agent in dogs in eastern Canada (8).

A group 2 canine respiratory coronavirus (CRCoV), that is closely related genetically and antigenically to bovine coronavirus (BCV) and human (respiratory) coronavirus strain OC43, but is distinct from canine enteric coronavirus (9,10), has been implicated in the last few years as an additional infectious co-factor in CITB (9,11–14). However, there are few data concerning its prevalence in dogs in North America (13,15), especially as it relates to the “lifestyle” or husbandry of dogs.

The objective of this longitudinal seroepidemiological study was to examine the prevalence of CRCoV in a population of dogs in a large urban humane shelter versus dogs in rural and small communities, and compare this to evidence of exposure to/infection with CPIV and B. bronchiseptica, which are thought to be the major etiologic agents in CITB (1). Our hypothesis was that there would be differences in prevalence of responses to respiratory pathogens between dogs from an urban humane shelter setting and dogs from rural settings.

Materials and methods

Study populations and sampling

One hundred and forty two dogs that were routine entrants to the Winnipeg Humane Society (Manitoba; WHS) from November 2004 to December 2005 were the subjects of this study (control population; population C). These dogs were mainly young adults brought to the WHS from Winnipeg and the surrounding area, most often for roaming at large. During the study, approximately 50% of the dogs came from within the city limits of Winnipeg, and 50% were from surrounding communities. The WHS is a large “no-kill” facility that keeps records of vaccinations, treatments, and signs of clinical disease.

For comparative purposes a total of 216 dogs were enrolled. Banks of sera were obtained at the time of routine physical examinations on 1 or more consecutive days from 7 other defined populations in rural western Canada. In August, 2006, 58 sera from Fort Chipewyan, Alberta (population A) and 56 sera from Fort Resolution, Northwest Territories (population B) were collected from dogs of a variety of breeds, many of which were sled dogs that were housed outdoors, usually in group pens, and fed game meat and fish (16). Five other sets of sera were obtained from dogs in small coastal communities in British Columbia in May or September 2007 (17), including 44 from Bella Bella (population BB), 6 from Ocean Falls (population OF), 16 from Klemtu (population KL), 23 from River Islet (population RI), and 13 from Hartley Bay (population HB). The ages of the enrolled dogs were only available for British Columbia (BC). Median age of dogs in coastal BC communities was 3 y (range 3 m to 15 y). Many dogs in these communities were kept outside and/or allowed to roam free, providing opportunities for close interactions with other dogs and wildlife.

Vaccines and vaccination

Unless otherwise stated, all dogs entering the humane shelter received a combination parenteral vaccine containing canine distemper, canine parvovirus, canine parainfluenza, and canine adenovirus 2 (Vanguard 5 Plus; Pfizer Animal Health, Kirkland, Quebec) on arrival. Most of the dogs entering during the study period also received 1 of 3 vaccines containing B. bronchiseptica; Coughguard (Pfizer Animal Health); Bronchicine CAe, (Pfizer Animal Health); Bronchishield 3 (Wyeth Animal Health, Guelph, Ontario). The previous vaccination history of dogs entering the humane shelter was not known in most cases. Approximately half of the dogs from rural areas and small communities in BC had been vaccinated for the common “core” canine viruses and rabies, but none, to our knowledge, had received vaccines for B. bronchiseptica. The vaccination history of the dogs from Alberta and the Northwest Territories was not known.

ELISA for CRCoV-reactive antibodies

An indirect enzyme-linked immunosorbent assay (ELISA) to detect IgG antibodies to CRCoV was performed using the closely related bovine coronavirus (BCV; 9,14) as antigen. The BCV antigen was obtained from human rectal tumor (HRT-18) cells infected with a BCV field isolate. Ninety-six-well flat-bottomed microtitration plates were coated overnight in antigen at 4°C, then washed in double-distilled water containing 0.05% Tween 20. Plates were blocked in carbonate coating buffer with 0.2% gelatin for 30 min at 37°C, and then rinsed in double-distilled water containing 0.05% Tween 20. Control and sample sera were diluted 1:50 in phosphate buffered saline (PBS) containing 0.05% Tween 20 and 0.2% gelatin and plated alongside blank control wells containing the diluent solution. The positive control (high reactivity in the assay) was a convalescent serum sample from a dog with respiratory disease from a respiratory outbreak in Calgary, Alberta; a negative control (no reactivity in the assay) was serum from a clinically normal young adult dog. Samples were incubated for 1 h at 37°C, and then washed. Plates were then incubated with horse radish peroxidase (HRP)-conjugated Protein A (Zymed, San Francisco, California, USA), diluted as per manufacturer’s instructions in a PBS solution containing 0.05% Tween 20, 0.2% gelatin and 4% PEG 8000 for 1 h at 37°C. Following this incubation, plates were washed and the enzymatic reaction visualized with ABTS (Kirkegaard and Perry Laboratories, Gaithersburg, Maryland, USA) stopped after 10 min with 1% SDS. Optical density (OD) values for test samples were expressed as ELISA units that were calculated as follows: 100 × [sample OD — blank (buffer only) OD]/[positive standard OD — blank (buffer only) OD].

ELISA for CPIV-reactive antibodies

An indirect ELISA to detect IgG antibodies to CPIV was performed as previously described (11), with minor modifications. The CPIV was obtained from a double component intranasal vaccine (Intra-Trac II; Intervet/Schering-Plough, Kirkland, Quebec) using a 0.46 millimicron filter to remove the B. bronchiseptica. The filtrate was inoculated at approximately 0.01 multiplicity of infection onto monolayers of Vero cells at 70% to 80% confluence. Seven days later, when viral cyptopathic effect was prominent, the culture medium was discarded and 3 mL of lysis buffer pH 8.4 [0.01 M Tris, 0.002 M ethylenediaminetetraacetic acid (EDTA), 0.2 M sucrose, 2% (v/v) Triton-X100] were added to an infected flask of cells and an uninfected flask of cells for control antigen. After a freeze and thaw, the cultures were sonicated for 30 s (50% duty cycle) and then clarified by 2 consecutive centrifugations (10 min at 12000 × g). Optimized dilutions of CPIV and control antigens were coated in 96-well microtiter plates at 4°C overnight. After a blocking step, serum samples were added to the plate, in duplicate, at a 1:50 dilution. CPIV-specific IgG was detected using peroxidase-conjugated protein A (Invitrogen; Carlsbad, California, USA) at a 1:5000 dilution. The substrate for the enzyme reaction was ABTS 1-component (Kirkegaarde and Perry Laboratories). Absorbance was read at 405 nm. A positive control serum was obtained from a dog that had been routinely vaccinated against CPIV, was kenneled several times a year, and had been boosted intranasally and then parenterally approximately 1 mo prior to serum collection. A negative control serum was from a young adult dog with an unknown vaccination history that had low optical density values in the ELISA. Optical density (OD) values for test samples were expressed as “units” as described for the ELISA for CRCoV.

ELISA for Bordetella bronchiseptica-reactive antibodies

An ELISA to detect IgG antibodies to B. bronchiseptica was performed as previously described (18). Briefly, 96-well flat-bottomed microtitration plates were coated overnight at 4°C with washed, sonicated B. bronchiseptica, in carbonate coating buffer (7.5 μg/well). The bacterial antigen was prepared from confluent 24-h cultures of the Regina-1 isolate; bacteria were suspended in saline solution, and aliquots were frozen at −70°C until they were used. The optimal dilution of antigen had been determined in a standard checkerboard design, using serum from immune and non-immune dogs. The coating antigen was removed from the wells, and the plates were washed by immersion in double-distilled water containing 0.05% Tween 20. The wells were filled with PBS solution containing 0.05% Tween 20 and 1% gelatin, and plates were incubated for 30 min at 37°C. Test serum, diluted 1:50 with PBS solution containing 0.05% Tween 20 and 0.2% gelatin, was added to replicate wells. Serum samples were diluted 1/50 in PBS and added to plates. Plates were incubated at 37°C for 1 h and washed, and peroxidase-conjugated goat anti-canine IgG (Cappel Laboratories, Cochranville, Pennsylvania, USA) diluted in PBS solution with 0.05% Tween 20, 0.2% gelatin and 4% normal goat serum, was added to the wells. Plates were again incubated for 1 h and washed, and peroxidase substrate was added to the wells according to the manufacturer’s instructions. Controls included serum from unvaccinated dogs (negative controls) and serum from dogs vaccinated against and challenged with B. bronchiseptica (positive controls); blank wells containing PBS solution with 0.05% Tween 20 and 0.2% gelatin were also used. Optical density (OD) values for test samples were calculated as a percentage of the positive control and expressed as “units,” as per the ELISA for canine coronavirus.

Polymerase chain reaction (PCR) for detection of Bordetella spp

Nasal swabs were placed in 1 mL of transport medium (DMEM with antibiotics and fungicide), and stored in liquid nitrogen until tested. The swabs and medium were thawed at room temperature, and vortexed for 1 min. The swabs were removed and the medium was spun at 5000 × g for 10 min to pellet bacteria and cellular debris, and the pellet was resuspended in 60 μL d2H₂O. DNA was extracted according to the QIAamp DNA Mini Kit (Qiagen, Valencia, California, USA). Positive references are B. bronchiseptica culture 95837 and strain Regina-1 (18). Primers B13615F, 5′-ATCGCTGGGATTACCCCTAC-3′ (sense orientation, located at position 13615 to 13634 on the B. bronchiseptica whole genome) and B13930R, 5′-TACTTGAGCGGTTCGAAGGT-3′ 3′ (antisense orientation, located at position 13911 to 13930 on the B. bronchiseptica whole genome) were designed from B. Bronchiseptica strain RB50 and were expected to amplify a 316 bp region (19). The PCR reaction mixture was composed of 45 μL of 1.1 × PCR Supermix Buffer (Invitrogen), 0.5 μL each primer (50 pmol/μL) and 4 μL template DNA. A cycle of pre-amplification was performed for 3 min at 95°C, followed by 40 cycles of amplification at 95°C for 30 s, primer annealing at 55°C for 1 min, extension for 1 min at 72°C. A final extension was performed at 72°C for 10 min.

Statistical analyses

Statistical analyses were performed using commercial software programs (SPSS 17.0 for Windows; SPSS, Chicago, Illinois, USA). Descriptive analyses were completed and variables were recorded as necessary for statistical modeling. Analyses examining antibody response and population were conducted and reported separately for CRCoV, CPIV, and B. bronchiseptica.

Since there is no standard cutoff available for these tests, the cutoff used for analysis was calculated by converting the mean net OD’s of the positive controls, negative controls, and blanks into an ELISA unit for each pathogen. To get better separation between true positives and negatives, calculated values of the mean net OD plus 3× the standard deviation were determined for the blank, and positive and negative controls (11). These values were then substituted into the formula for determining ELISA units as described. The calculated negative control OD value was used in place of the sample OD in the formula. The resulting ELISA unit was then used as the cutoff for each pathogen; anything equal to or above that value was classified as having a moderate to high antibody titer and anything below was classified as having low or no antibody titer. This dichotomized parameter was used for all subsequent analyses. The ELISA unit cutoff value for CRCoV was 1.9, for CPIV 12.9 and for B. bronchiseptica 14.0.

Logistic regression was performed to assess the relationship between antibody response and population of origin. The outcome variable was the dichotomized antibody response and the only predictor variable included in the model was the population (9 sites) from which the dogs came. The Winnipeg Humane Society (WHS) was the reference group for predictor variables. A chi-square was used to more fully explore if there was a difference between urban and rural dogs. To do this, the populations were dichotomized into rural (all populations except the WHS) and urban (WHS). A chi-square was also used to explore the relationship between antibody response on arrival at the WHS and the development of signs of respiratory disease within the first 10 d at the shelter. The relationship between age and antibody response was examined when both were available from the same animal (n = 102); age was only available for dogs located in the 5 sites in British Columbia. To account for the effect of clustering within site the analysis was performed using generalized estimating equations (GEE) for binary outcomes. The predictor variable, age, was categorized into < 6 mo and ≥ 6 mo. For each of these analyses, statistically significant associations were reported as odds ratios (OR) when P ≤ 0.05.

To investigate the relationship among antibodies to CRCoV, CPIV, and B. bronchiseptica and when antibody response was recorded for each pathogen for the same animal Spearman rank correlation was performed. Results were reported as rho (ρ).

Results

Coronavirus seroprevalence

Although CRCoV seroprevalence varied among the dog populations (n = 370), the vast majority of dogs in all populations had no or low antibodies to this virus (Figure 1). Dogs in Fort Resolution were 0.16 times (P ≤ 0.0001) less likely to have moderate to high CRCoV antibody titers (ELISA unit ≥ 1.9) than dogs in the WHS. Similarly, dogs in Klemtu and dogs in River Islet were 0.082 (P = 0.02) and 0.117 (P = 0.005) times less likely to have moderate or high antibody titers than dogs in the WHS. When the data were divided into simply “rural” versus “urban” (shelter dogs), rural dogs were 0.421 times (P ≤ 0.0001) less likely to be positive for CRCoV than dogs in an urban setting. For the 5 BC populations (n = 102) that had the dog age available, dogs < 6 mo of age were 0.514 times (P = 0.03) less likely to have moderate or high CRCoV antibody titers (ELISA units ≥ 1.89) than dogs that were > 6 mo of age.

Figure 1.

Distribution of antibody responses to CRCoV in the various populations of dogs (n = 370).

Comparison of CRCoV antibodies in dogs with and without respiratory disease

Sequential serum samples collected at 7- to 10-day intervals, when available, from dogs entering the WHS in each mo from November 2004 to August 2005 were examined for seroconversion (a 3-fold or greater increase in baseline ELISA value) to CRoCV. The incidence of seroconversion among dogs that had no or low antibodies to CRCoV during these mo ranged from 33% to 67%, with no apparent differences in seasonality (Figure 2).

Figure 2.

Monthly seroconversion to CRCoV in dogs (n = 67) during their stay at the Winnipeg Humane Society humane shelter (white bars). Monthly percentage of CRCoV-seroconverted dogs that developed respiratory disease (black bars).

For 129 dogs that were in the WHS facility for 10 or more days the CRCoV ELISA results from the day of arrival were compared with the development of respiratory disease. Of these, 49 dogs had shown 1 or more signs of respiratory disease typical of CITB, including, cough, nasal discharge, or sneezing, and 80 remained free of respiratory signs. Sixty-four of the dogs had incoming ELISA units of < 1.89 (low or no antibodies); 65 had incoming values of > 1.89. Twenty-eight of the dogs with low antibodies had respiratory disease, whereas 21 of the dogs with moderate to high antibodies to CRCoV developed respiratory disease (Figure 3). There was no statistically significant (P = 0.18) relationship between antibodies to CRCoV and signs of respiratory disease.

Figure 3.

Percentage of dogs with either low (64 of 129) or moderate to high antibody (65 of 129) concentrations to CRCoV on day 1 at the Winnipeg Humane Society humane shelter which developed respiratory disease during their stay. Moderate to high antibody titers were not significantly (at the P < 0.05 level) associated with disease-sparing.

Canine parainfluenza virus seroprevalence

As with CRCoV, CPIV seroprevalence varied among the dog populations (n = 364) and the vast majority of dogs in all populations had no or low antibodies to this virus (Figure 4). Dogs in Fort Resolution were 0.236 times (P = 0.02) less likely to have moderate to high CPIV antibody titers (ELISA units ≥ 12.9) than dogs in the WHS. When the populations were divided into “rural” versus “urban” (shelter dogs) there were no significant differences between the groups (P = 0.39). For the 5 BC populations (n = 102) that had the dog age available, there were no significant differences (P = 0.36) in CPIV antibody titers between dogs < 6 mo of age and dogs ≥ 6 mo of age.

Figure 4.

Distribution of antibody responses to CPIV in the various populations of dogs (n = 364).

Comparison of CPIV antibodies in dogs with and without respiratory disease

Due to the use of vaccines containing CPIV on arrival at the WHS, it was not possible to assess seroconversion to this pathogen, but incoming antibody concentrations were compared to the development of respiratory disease in dogs while at the facility. Among 125 dogs that were evaluated 101 had incoming ELISA units of < 12.9 (low or no antibodies) and 14 had incoming values of > 12.9 in the test using CPIV antigen. Forty-one of the dogs with low antibodies and 2 of the dogs with moderate to high antibodies to CPIV developed respiratory disease (Figure 5). There was no significant (P = 0.06) relationship between antibodies to CPIV and sparing of signs of respiratory disease.

Figure 5.

Percentage of dogs with either low (101 of 115) or moderate to high antibody (12 of 115) concentrations to CPIV on day 1 at the Winnipeg Humane Society humane shelter which developed respiratory disease during their stay. Moderate to high antibody titers were not significantly (at the P < 0.05 level) associated with disease-sparing.

Bordetella bronchiseptica seroprevalence

Most dogs in all populations (n = 358) had moderate to high antibody titers to this pathogen (Figure 6). Dogs from population A (Fort Chipewyan) and dogs from population B (Fort Resolution) were 0.499 (P = 0.03) and 0.437 (P = 0.01) times less likely to have moderate or high B. bronchiseptica antibody titers (ELISA units ≥ 14.0) than dogs at the WHS. There was no significant difference between the titers in rural and urban populations (P = 0.14). For the 5 BC populations (n = 102) that had the dog age available, dogs < 6 mo of age were 0.10 times less (P ≤ 0.001) likely to have a moderate to high antibody titer to B. bronchiseptica compared with dogs ≥ 6 mo of age.

Figure 6.

Distribution of antibody responses to B. bronchiseptica in the various populations of dogs (n = 358).

Comparison of anti-Bordetella antibodies in dogs with and without respiratory disease

Due to the use of Bordetella vaccines, it was not possible to assess seroconversion to this pathogen, but incoming antibody concentrations were compared to the development of respiratory disease in dogs while at the WHS. Thirty of the 129 dogs assessed for respiratory disease had incoming ELISA units of < 14 (low or no antibodies); 99 had incoming values of > 14 in the test using B. bronchiseptica antigen. Fifteen of the dogs with low antibodies had respiratory disease, whereas 34 of the dogs with moderate to high antibodies to B. bronchiseptica developed respiratory disease (Figure 7). There was no statistically significant (P = 0.12) relationship between antibodies to B. bronchiseptica and signs of respiratory disease. The circulation of B. bronchiseptica in the facility during the study period was confirmed by positive PCR tests on 6 of 12 nasal swabs that were collected from dogs with clinical respiratory disease.

Figure 7.

Percentage of dogs with either low (30 of 129) or moderate to high antibody (99 of 129) concentrations to B. bronchiseptica on day 1 at the Winnipeg Humane Society humane shelter which developed respiratory disease during their stay. Moderate to high antibody titers were not significantly (at the P < 0.05 level) associated with disease-sparing.

Bordetella bronchiseptica-reactive antibody responses in vaccinated dogs

During the period of study several different vaccines for B. bronchiseptica were in use, or none were given to some dogs. Manufacture of one of the vaccines (Coughguard; Pfizer) was discontinued during the study period. Therefore, responses to a single dose of an injectable vaccine (Bronchicine; Pfizer), were examined in a subpopulation of 17 young adult dogs that entered the facility in September and October 2005, and for which longitudinal data were available. All except 1 of these dogs had increasing antibody titers (usually greater than 2-fold increase in ELISA values) by 7 to 10 d after infection (Figure 8). Three of the 17 vaccinated dogs, 2 of which had no detectable antibodies to B. bronchiseptica on arrival, developed signs of respiratory disease; the remainder were free of respiratory disease during and immediately after their tenure at the shelter.

Figure 8.

Antibody responses to B. bronchiseptica in dogs that received a single parenteral dose of B. bronchiseptica cellular extract in dogs at the Winnipeg Humane Society humane shelter (n = 17). Dogs were bled and vaccinated on day 1 and re-bled at 7 to 10 intervals. ^a indicates no third blood collection.

Relationship among antibodies to CRCoV2, CPIV, and Bordetella

When ELISA units for CRCoV, CPIV, and Bordetella antigens were compared there was no apparent relationship between antibody titers for the 3 of them (Table 1).

Table 1.

Bordetella bronchiseptica, CRCoV, and CPIV Spearman’s rho correlation coefficients

	B. bronchiseptica	CRCoV	CPIV
B. bronchiseptica	—	0.185 (n = 354)	0.284 (n = 334)
CRCoV	0.185 (n = 354)	—	0.094 (n = 360)
CPIV	0.284 (n = 348)	0.094 (n = 360)	—

Discussion

The data do not support our hypothesis that there would be differences in prevalence of responses to respiratory pathogens between dogs from urban and rural settings. This may be due to the fact that the dogs that entered the humane shelter were not representative of well-cared for animals from single-pet households having had limited contact with other animals.

Nevertheless, there were significant differences in seroprevalence of responses to these 3 important respiratory pathogens among the populations which, in the absence of vaccination, reflected differences in natural exposure, and/or population-based differences in microbial dynamics. Thus, differences in “lifestyle” may significantly impact the frequency of natural exposure to pathogens that may be endemic at the overall population level. There are few data that address this issue at the population level. In one recent seroepidemiologic study (20) it was reported that vaccinated dogs with a “high risk” lifestyle (increased opportunities for commingling) were more likely to maintain higher titers of CPIV-neutralizing antibodies. Alternatively, observed differences could be the result of unidentified geographical factors, such as climate, that could affect pathogen prevalence or transmission. Whatever their basis, differences in the exposure could impact immune responses, resulting in differences in duration of immunity, especially for pathogens such as CRCoV, CPIV, and B. bronchiseptica that have short incubation periods and cause disease at the mucosal surface at the site of entry. These types of pathogens may require higher levels of effector responses, antibodies and/or T-cells, versus memory responses at the time of exposure to achieve protection (21).

The vast majority of dogs in this study had low, and, most often, no detectable antibodies to CRCoV. These findings are at variance with the only other published data (12) from Canadian dogs that reported a seroprevalence of 59.1%; however, the latter study only included 44 sera that were probably mostly derived from eastern Canada, and the characteristics of the populations from which these dogs came were not reported. Similarly, there was a low prevalence in antibodies to CPIV compared with other studies in the last decade (11,22). Reasons for this dramatic difference are not known, but could be related to geographic differences and/or differences in chances for exposure. Such factors could also have contributed to the observation that the dogs in Fort Resolution consistently had lower antibody titers than those at the shelter. In addition, in the case of CPIV, apparently lower seroprevalence could be due to differences in the prevalence or frequency of vaccination. The high incidence of seroconversion to CRCoV in dogs during their stay at the WHS is consistent with findings in humane shelters and kennels in the United Kingdom (9–12,13), and supports the idea that this acutely infecting virus, much like CPIV (22), can rapidly spread among susceptible dogs when they are commingled. In contrast to a previous study in England (9), antibodies to CRCoV on arrival at the WHS were not associated with significant sparing of clinical respiratory disease.

There are few population-based seroepidemiologic data on naturally occurring B. bronchiseptica infections in dogs (4,22,23). In contrast to a recent study of unvaccinated dog populations in Sweden (22) that reported an overall seroprevalence of 22% for antibodies to B. bronchiseptica, most dogs in this study had moderate to high titers of antibodies to this pathogen. The presence of B. bronchiseptica on nasal swabs of half of the dogs with clinical respiratory disease that were sampled, indicates that this pathogen is still a major etiologic agent in CITB (1,19), although it can also be found in subclinically infected dogs (4). The data do not support the idea that rural dogs have increased titers of antibody to B. bronchiseptica and other respiratory pathogens (compared to urban dogs) due to the greater potential for contact with wildlife reservoirs of these pathogens, as has been documented or suggested in the case of other pathogens and parasites (16,17). In previous studies, crowding of dogs by virtue of “high risk” lifestyle activities was significantly associated with the seroprevalence of CPIV (20,22) but not B. bronchiseptica (22). It was suggested that a higher prevalence of antibodies to CPIV in commingled dogs was due to the acute nature of respiratory viral infection, with an accompanying relatively short period of shedding (22). Interestingly, early studies of experimental transmission of CPIV demonstrated that concentrations of CPIV-neutralizing antibody declined to usually undetectable levels by 3 or 4 mo after exposure (24). The same would likely apply to CRCoV (11), but not to B. bronchiseptica, which induces a carrier state in at least some animals, thereby providing a more prolonged source of exposure, and less dependence on crowding for transmission (4,22).

In prospective studies of parenteral vaccination with B. bronchiseptica bacterins, there was an association between B. bronchiseptica — reactive serum antibodies and disease- sparing in subsequent experimental infections (18,25). In contrast, in this study, and in a previous examination of serum antibody responses to B. bronchiseptica lipopolysac-charide (LPS;22), there was no significant association between B. bronchiseptica-reactive serum antibodies at the time of presentation at a humane shelter and the subsequent development of respiratory disease. In the present study there are at least 5 complicating factors that may explain this apparent disparity, and the apparent lack of a previously reported disease-sparing effect of antibodies to CRCoV. First, some animals in the “moderate to high” antibody groups determined by a “cutoff” value based on very little or no reactivity in the ELISA may not have had antibody concentrations sufficient to affect disease-sparing. Second, there is usually co-circulation of other respiratory pathogens, such as Mycoplasma spp. (1). Third, the level of pathogen exposure, or disease producing “synergy” among co-infecting pathogens, may have “overwhelmed” any disease-sparing effect of antibodies to any 1 pathogen. Fourth, the use of vaccines for B. bronchiseptica and CPIV on arrival may have stimulated disease-sparing responses in dogs with low or no antibodies at the time of presentation to the facility. These responses were consistent with anamnestic responses in previously exposed or vaccinated dogs, as it would be unlikely to have primary responses of this magnitude 7 d after vaccination (26). Finally, the number of animals enrolled in the study, and the observed incidence of disease may not have provided enough statistical power to identify biologically relevant associations.