|Home | About | Journals | Submit | Contact Us | Français|
Swine influenza (SI) is a disease of significance for the swine industry, and vaccination is often recommended as a way to reduce its impact on production. The efficacy of SI vaccines is well established under experimental conditions, but information about field efficacy is scarce. The objective of this study was to evaluate the efficacy of a commercial inactivated bivalent (H1N1/H3N2) vaccine under conditions of natural exposure to a field SI variant. To accomplish our goal we used a randomized, blinded, field trial in 2 cohorts of finisher pigs in a multi-site swine production system located in southern Ontario. During the trial, this herd experienced an outbreak of porcine circovirus associated disease (PCVAD). The efficacy of the SI vaccine was assessed through its effect on average daily weight gain, and serological responses to SI over time. The effect of vaccination on pig growth was different in the 2 cohorts. Weight gain was higher in vaccinated pigs than in control pigs in Cohort 1, but was numerically higher for control pigs than for vaccinated pigs in Cohort 2. Vaccination against swine influenza, in a herd experiencing an outbreak of PCVAD, was of questionable value.
L’influenza porcin (SI) est une maladie importante pour l’industrie porcine, et la vaccination est souvent recommandée comme moyen pour réduire son impact sur la production. L’efficacité des vaccins contre SI est bien établie lors de conditions expérimentales, mais les informations quant à l’efficacité en condition de terrain sont rares. L’objectif de la présente étude était d’évaluer l’efficacité d’un vaccin bivalent inactivé commercial (H1N1/H3N2) dans des conditions naturelles d’exposition à un variant de champs de SI. Afin de réaliser notre objectif nous avons réalisé un essai clinique randomisé, à l’aveugle dans 2 cohortes de porcs en finition dans un système de production porcine en multi-sites dans le sud de l’Ontario. Durant l’essai, ce troupeau a vécu une épidémie de maladie associée au circovirus porcin (PCVAD). L’efficacité du vaccin contre SI a été évaluée par son effet sur le gain de poids quotidien moyen, et les réponses sérologiques envers SI dans le temps. L’effet de la vaccination sur la croissance des porcs était différent dans les 2 cohortes. Le gain de poids était plus élevé chez les porcs vaccinés comparativement aux porcs témoins dans la Cohorte 1, mais était numériquement plus élevé pour les porcs témoins comparativement aux porcs vaccinés dans la Cohorte 2. La vaccination contre SI, dans un troupeau au prise avec une épidémie de PCVAD, avait une valeur questionnable.
(Traduit par Docteur Serge Messier)
Swine influenza virus (SIV) is a prevalent respiratory pathogen of swine with worldwide distribution. In addition to its significance as an important pathogen of swine, it is of concern for its potential human health risk (1). Infection with SIV in naïve swine herds usually manifests as an outbreak of febrile respiratory disease with high morbidity and low mortality. Loeffen et al (2) attributed 7 of 16 investigated outbreaks of respiratory disease in finisher pigs in The Netherlands solely to SIV infection. Circulation of H1N1 and H3N2 SIV among pigs is documented in herds experiencing atypical, mild, or no clinical signs (2–4). Additionally, SIV infection is a component of the porcine respiratory disease complex (PRDC) (5). Since the late 1990’s, both antigenic drift and shift have been recorded in SIVs circulating in the US swine population, causing concern from a production perspective. Swine influenza vaccines appear effective in preventing clinical disease under experimental conditions (6,7) but their efficacy has been questioned in field reports (8). In addition, vaccine efficacy under conditions of natural exposure to SIV has been rarely evaluated.
The objective of the study was to evaluate the direct effect of a commercial inactivated bivalent (H1N1/H3N2) influenza vaccine on growth of finisher pigs. To accomplish this we used a randomized, blinded, field trial in a commercial multi-site swine production system in Ontario, in which the producer reported persistent coughing in successive finisher groups. Further objectives were to evaluate serological response, mortality, and development of clinical signs suggestive of SI over time, with the expectation of comparing these parameters between treatment groups to better understand within-herd SI dynamics. However, during the trial, the study herd had an outbreak of porcine circovirus type-2 associated disease (PCVAD), characterized by weight loss, increase in mortality, and respiratory distress of growing pigs.
The farm selected for this study was a multi-site swine production system consisting of 1 nursery barn and 3 finisher barns, each located at a different site. The nursery barn was divided into 2 major sections. The section used for the first stage nursery consisted of 5 rooms, each divided into 16 pens, housing approximately 35 pigs per pen. The section for the second stage nursery consisted of 4 rooms, each divided into 8 pens with slatted floors, housing approximately 70 pigs per pen. Nursery rooms were ventilated with a negative pressure system. The finisher barn used for this study was located approximately 2 km from the nursery barn, and consisted of 4 equally sized rooms, each with a capacity of 550 pigs. Doors between rooms were kept closed. Each room contained 5 pens on each side of a central alley. Rooms were individually ventilated with a combination of negative pressure (winter ventilation) and an electronically monitored natural ventilation system with curtains (summer ventilation). All pens in the nursery and finisher barns were washed with hot water and disinfected between groups. The nursery barn and the finishing barn selected for this study were intended to be managed as all-in/all-out (AIAO) by room facilities. Once weekly, the nursery barn received 500 to 600 pigs weaned at 17 to 19 days of age from 3 different supplier farms. Weaned pigs stayed in the first-phase nursery for 4 wk, and were then sorted by size and moved to the second-phase nursery for an additional 3 wk. The second-phase pen housing the smallest pigs was also used to house pigs that were treated during this phase. Once per week, pigs at the end of the nursery phase were shipped by truck to a finisher barn and redistributed into finisher pens by size. The finisher barn used in this study was populated with pigs that had come from 4 consecutive weekly cohorts of pigs. Once populated, pigs from the nursery barn would be shipped to other finisher sites. Frequently, 1 or 2 rooms in the finisher barn were filled with light-weight pigs from previous cohorts waiting to be shipped to market, and pigs entering the finisher barn from the nursery were moved through these rooms. Each finisher room contained 1 pen that housed light-weight pigs and pigs that required treatment, extra care, or both (room-hospital pen). An additional pen in the storage area of the barn was used for pigs that required extra heat, care, and feed (barn-hospital pen).
Pelleted feed was fed in the first-phase nursery and mash feed in the second phase. Nursery feed was medicated with penicillin (trade name not recorded) and 60 mg/kg of salinomycin (Posistac 6% Premix; Phibro Animal Health, Regina, Saskatchewan). However, in nursery-week 3 for Cohort 1, a new protocol was introduced so that feed contained 275 mg/kg of chlortetracycline hydrochloride (Chlor 100 Granular Premix; Bio Agri Mix LP, Mitchell, Ontario) and 31 mg/kg of tiamulin (Denagard Medicated Premix; Novartis Animal Health Canada, Mississauga, Ontario). Finisher pigs were fed liquid feed, pulse-medicated for 14 d at a time with chlortetracycline hydrochloride (Chlor 100; 550 mg/kg feed) as required in response to respiratory problems. Weaned pigs were vaccinated with modified-live PRRS vaccine (Ingelvac PRRS ATP; Boehringer Ingelheim Vetmedica, St. Joseph, Missouri, USA) at the time of entry into the nursery barn.
Finisher pigs in this herd experienced outbreaks of respiratory disease during several months prior to this study; the disease started at approximately 110 days of age (4 wk after entering the finisher barns). Clinical signs included a persistent barking cough and lack of response to treatment. Diagnosis of SI was confirmed by isolation of influenza A virus using Madin-Darby canine kidney cells, typed as H1N1, and the demonstration of almost 100% seropositivity to H1N1 SIV by enzyme-linked immunosorbent assay (ELISA) in finisher pigs prior to the study. At the same time, seropositivity to porcine reproductive and respiratory syndrome virus (PRRSV) in finisher pigs was also 100%, with sample-to-positive (SP) ratios ranging from 0.5 to 4 when tested by HerdChek PRRS ELISA, (IDEXX Laboratories, Westbrook, Maine, USA).
The SI status of the first-phase nursery pigs before study initiation was dependent on the source. While first-phase nursery pigs from the first source barn were 80% positive, the pigs coming from the second and the third source barn were 25% and 0% positive, respectively.
Only one finisher barn was selected for this study in an attempt to minimize the potential effect of site-specific variables on the investigated associations and because of concerns regarding biosecurity. The study site was selected because it started receiving nursery pigs at the time the study was initiated. This field trial was designed to determine if there was a difference of at least 3.5 kg in the market live body weight between the influenza-vaccinated (vaccinated) and non-vaccinated (control) animals with 95% confidence and 80% power, and under the assumption that the standard deviation (s) in live body weight for this farm was assumed to be 12 kg and the mean final market body weight was 115 kg. The initial estimate of 185 animals per intervention group was increased to 224 animals per group to accommodate for possible losses; to equalize the number of pigs per pen due to stratification; and, to adjust for clustering of pigs within a finisher pen. For logistical reasons, and to minimize the potential effect of herd immunity, the required number of pigs was divided into 2 cohorts. Cohort 1 entered the trial January 20, 2004, and Cohort 2, 4 wk later. Each cohort included 112 vaccinated and 112 control pigs (Table I).
A sampling frame was constructed, whereby pigs were randomized into intervention groups, and a subset from each intervention group was selected to be blood sampled. All randomization procedures were performed in MINITAB 14 (Minitab, State College, Pennsylvania, USA). All pigs housed in each of the 7 pens of the second-stage nursery that met the inclusion criteria were marked with sequential numbers to facilitate random selection (target population). Pigs that had been in the hospital pen, pigs with hernias or that appeared sick on the day of selection, and non-castrated male pigs were excluded from the study. Thirty-two pigs per pen for a total of 224 pigs per cohort (study population) were randomly selected from the target population using numbers from the sampling frame and the numbers recorded on pigs. These pigs were randomly allocated to the intervention group (vaccinated and control). In addition, 5 pigs per pen per intervention group were randomly selected for the sequential blood sampling for a total of 35 pigs per intervention group per cohort. This number was designed to detect the seroconversion of at least 10% in the control group with 95% confidence (assuming a perfect test and accounting for possible losses). Hence, pigs included for both weight and serology evaluation were selected by a stratified random sampling based on pen strata. Animals were tagged with 4-colored, uniquely numbered ear tags whereby the colors represented the vaccination and blood sampling status of the pig.
Pigs in the control group, pigs not selected for the study, and pigs excluded from the study were vaccinated with Mycoplasma hyopneumoniae bacterin (RespiSure; Pfizer Animal Health, Exton, Pennsylvania, USA) according to the manufacturer’s instructions. Pigs in the vaccinated group were vaccinated according to the manufacturer’s instructions with FluSure/RespiSure RTU (Pfizer Animal Health), which is a combination of Mycoplasma hyopneumoniae bacterin and SI. The SI component is a bivalent (H1N1/H3N2) inactivated vaccine with an oil-in-water adjuvant containing antigens from strains claimed to match the ones currently circulating in North American swine. Booster doses were administered in the control and vaccinated groups 3 wk later, just before the pigs were moved to the finisher barn.
Only 2 observers were aware of the experimental designation of the pigs, and neither of them participated in the initial marking or weighing of pigs. Observers who weighed pigs and farm personnel were blinded throughout the study. Baseline measurements for each pig included body weight, gender, and for the pigs in the sequential blood sampling group, SIV serological status. This protocol was approved by the Animal Care Committee of the University of Guelph and was in accordance with guidelines for the care and use of experimental animals (9,10).
Each pig (n = 2 × 224) was weighed 4 times: baseline (6.5 wk of age) and 44, 88, and 108 d later, corresponding to approximately 0, 6, 12, and 15 study wk, respectively. Pigs were weighed on an electronic scale (Pelouze Model 4010; Pelstar LLC, Bridgeview, Illinois, USA) for the baseline weight and on a spring scale for all other weights. For the spring scale, frequent weighing of a constant weight was performed throughout the duration of the weighing, and observers recording weight were blinded with respect to vaccination status. A standardized approach was used to read the scale. Sixteen pigs in Cohort 1 (7.9%) and 22 pigs in Cohort 2 (10.9%) were shipped to market 3 to 9 d before the final weighing. These 38 pigs were weighed just before shipping, and they accounted for 9.2% and 6.7% of vaccinated and control pigs, respectively, in Cohort 1; and 11.1% and 10.7% of vaccinated and control pigs, respectively, in Cohort 2. Some of these pigs were faster growing pigs shipped earlier to avoid penalties; the other pigs in this group could be considered as a random sample since they were sold to a niche market.
Blood was collected from the orbital venous sinus after weight was recorded from pigs selected for blood sampling (70 per cohort). Sera were submitted to the Animal Health Laboratory (AHL; University of Guelph, Guelph, Ontario) for testing with a commercial SIV H1N1 ELISA assay (IDEXX Laboratories). A sample-to-positive ratio of ≥ 0.4 was considered positive. This assay was used because only H1N1 was detected by virus isolation prior to the study.
Herd managers in the nursery and finisher barns were provided with a log book to record information about mortality, medication, and moving of pigs into room-hospital pens. For any treated animal, the date of treatment, reason for treatment, dosage of drug used, and identification of the treated animal were recorded. Pigs that were moved into the barn-hospital pen were not weighed nor blood sampled, but treatments were recorded in the log book.
Investigators visited the farm each week to observe clinical signs, to score coughing, and to collect dead pigs if they were in suitable condition for postmortem examination. Mortalities were submitted to AHL for necropsy. Two observers recorded the number of pigs per pen and counted the number of coughs per pen per 5-min period after pigs were encouraged to move. One episode of coughing was counted as a single cough. Pigs showing clinical signs of a barking or paroxysmal cough were marked, and nasal swabs were taken. If no barking or paroxysmal coughing was observed, nasal swabs were collected from pigs that showed any kind of coughing or other respiratory distress, regardless of whether or not they were part of the trial. Swabs were transported in virus transport media (VTM; AHL, University of Guelph) at 4°C. At least 3 swabs per week were collected until SIV was detected. Swabs were frozen at −80°C and submitted the following week to AHL for SIV isolation in Madin-Darby canine kidney cells and embryonated eggs. Isolated SIVs were typed by reverse-transcriptase polymerase chain reaction.
Data were entered into a database (Access; Microsoft Corporation, Redmond, Washington, USA) and imported into SAS version 9.1 (SAS Institute, Cary, North Carolina, USA) for further manipulation and analysis. The effect of influenza vaccine on average daily gain (ADG) was analyzed using a linear mixed effect model (Proc Mixed; 11), with randomly varying intercepts and slopes among individual pigs nested within the finisher pens. The model assumptions for the “jth” pig nested within the “ith” finisher pen, at the “kth” measurement occasion are listed in Equation 1. The outcome consisted of 3 sequential weight measurements in grams, and the effect of covariates of interest on weight gain over time was evaluated using their interaction with the time variable measured as days since the initial weight measurement (11). Hence, the coefficients containing the interaction with time was interpreted as average daily gain in grams.
Linear mixed effect model used to assess the pig’s weight over time where:
yijk = the weight of the jth pig in the ith pen at the kth measurement.
x1ij = time-invariant covariates (vaccine status, gender, cohort, weight at day 0, interactions)
x2ijk = time-variant covariates (time in days since vaccination)
beta = coefficients associated with fixed effects (vaccine, gender, cohort, initial weight, interaction terms)
b = random effects (intercept and random coefficient of days since the first vaccination)
i = pen id
j = pig id
k = repeated measurement (days: 44, 88, 108)
Outcome is measured as weight in grams. Slope of interaction terms between the time variables measured in days and other fixed effects is interpreted as average daily gain in grams.
The final fixed effect model was fitted using the maximum likelihood method (ML). Contrasts were performed to test for equality of slopes (ADG). Outliers and influential observations were assessed for fixed and random effects at the pig and pen level (12). After identifying observations with undue influence, models were reanalyzed without them.
Changes in the SP ratio over time were modeled using a linear mixed effect model, with unstructured covariance to account for repeated measurements within a pig (SP ratio model). Least square means of SP ratios were evaluated at each time point that sera were taken. Logistic regression, with pen as a random effect using penalized quasilikelihood (Glimmix macro) was used to assess the mean probability of seroconversion to SIV between weeks 0 and 6, and between weeks 6 and 12, for each cohort (Empty model). The intra-cluster correlation coefficient was calculated using a latent class model approach (13) and interpreted as the proportion of variation in seroconversion that resided at the pen level. Seroconversion was defined as a change from negative status (SP < 0.4) to positive status (SP ≥ 0.4).
Effect of vaccine and cohort on mortality was evaluated using a Cox’s proportional hazard model. In addition, effect of cohort on mortality in the target population was evaluated by logistic regression in a model that included all pigs (Table I).
The incidence rate of coughing in the finisher barn was evaluated in a Poisson regression, using penalized quasilikelihood approach (Glimmix macro), and with the random intercept and time at the pen level.
Observers noted a paroxysmal, barking cough in pigs even when they were resting. The mean incidence rate of coughing in finisher pens tended to be higher [incidence rate ratio, 1.73; 95% confidence interval (CI), 1.0–3.0] in Cohort 1 than in Cohort 2 throughout the study (Table II). In addition, the incidence rate of coughing increased consistently by 21% (95% CI, 7.5%–36%) on a weekly basis. However, there was significant random variation both in the initial rate of coughing and in its development over time (Table II, Figure 1). The pattern of coughing in 2 pens of Cohort 1 differed both from the pattern in most of the other pens in either cohort, and from the pattern in the average pen in each cohort predicted by the model (Figure 1). The pigs in one of these pens were 1 wk younger than the others in that room. Nasal swabs collected in week 9 from 2 coughing pigs in this pen were positive for type A SIV by virus isolation, typed as H1N1.
Other pathogens identified during the course of the trial included porcine parvovirus (PPV; virus isolation in cell culture) from pigs exhibiting severe dyspnea; PRRSV (RT-PCR); and porcine circovirus type-2 (PCV-2; immunohistochemistry) in lungs, spleen, liver, and kidney of vaccinated and control pigs. Salmonella enterica, Actinobacillus suis, Streptococcus suis, Haemophillus parasuis, and Pasteurella multocida were cultured in a subset of pigs submitted for complete necropsy. In pigs submitted for gross examination, the 4 most common tentative diagnoses were pneumonia, serositis, wasting, and lymphadenopathy. Room-hospital pens were filled with pigs with severe (expiratory) dyspnea and tachypnea, which was especially noticeable after nursery pigs were moved to the finisher barn. Enlarged inguinal lymph nodes were observed in almost all pigs examined in the room-hospital pen. Severe wasting was observed in pigs with and without respiratory signs. Clinical signs and lesions discovered on gross and histological examination and immunohistochemistry were consistent with PCVAD (14). The start of the PCVAD outbreak in the study finisher barn apparently coincided with the start of the trial, although clinical signs consistent with the systemic PCVAD were observed in nursery pigs prior to randomization.
Descriptive statistical results are presented in Table I. Coefficients of variation (CV) of initial body weight at room level were 16.2 in Cohort 1 and 13.6 in Cohort 2. At the pen level, body weight CV ranged from 12 to 15.3 in Cohort 1 and from 9.3 and 13.1 in Cohort 2.
Rate of growth of vaccinated and control pigs differed between Cohorts 1 and 2, as suggested by the significant three-way interaction between vaccine, cohort, and time variable (day). Evaluation of residuals and influential statistics at the pig- and pen-levels revealed that pigs coming from the room-hospital pens had an undue impact on the model estimates. The final model, therefore, was fitted without animals that originally occupied room-hospital pens at Week 6 of the trial (7 pigs), but it still indicated a different vaccine effect in the 2 cohorts (Table III; P = 0.02). Least squares differences in mean average daily gain of vaccinated and control pigs are presented in Table III. Briefly, the model suggested that vaccinated pigs in Cohort 1 grew 44.0 g/day faster than control pigs (P = 0.03). In contrast, Cohort 2 control pigs grew at 24.6 g/day faster rate than the vaccinated pigs, but this difference was not statistically significant (P = 0.24).
The final model for the development of a serological response contained the linear and quadratic effect of time represented by week (P < 0.01), interaction of vaccination with the linear and quadratic effect of week (P < 0.01), and three-way interaction of cohort, vaccination, and linear effect of week (P = 0.01). The expected SP ratios in different groups predicted by this model are shown in Figure 2. The least squares means of SP ratios of vaccinated pigs did not differ (P > 0.5) at weeks 6, 12, and 15 of the trial, although at week 0, vaccinated pigs in Cohort 1 tended to have higher SP ratios (P = 0.06) than vaccinated pigs in Cohort 2. However, the least squares means of SP ratios of control pigs tended to be higher in Cohort 1 than in Cohort 2 at Week 6 (P = 0.09), and were higher at weeks 12 (P = 0.03) and 15 (P = 0.02).
The proportions of vaccinated and control pigs that were positive for SI at each measurement are presented in Table I. For each cohort, the mean probability of seroconverting to SI between weeks 0 and 6 and between weeks 6 and 12, and the approximate proportion of variation at the finisher-pen level, are listed in Table IV. The highest mean probability of seroconversion occurred in Cohort 1 in the period between weeks 6 and 12.
In the study population, the hazard of dying did not differ between vaccinated and control pigs (P > 0.05). In the overall cohorts, the total mortality was higher in Cohort 1 than in Cohort 2 (OR = 1.81; 95% CI, 1.12–2.98). Development of clinical signs indicative of PCVAD over time was reported elsewhere (15).
The major clinical problem reported by the producer that motivated this study was persistent coughing in successive finisher groups; however, during the study period, the most important disease concern was an outbreak of systemic PCVAD. In addition, overall disease dynamics in the 2 cohorts were different, as indicated by a higher mortality risk in Cohort 1 pigs.
Under experimental conditions, the efficacy of influenza vaccine is usually evaluated by measurements related to the expression of clinical signs and to viral replication (6,7). Both types of measurements provide information about varying degrees of vaccine protection against clinical disease and SIV infection (6,7). Under field conditions, however, these measurements are impractical. Hence, we decided to evaluate vaccine efficacy by measuring growth, which is, from the producers’ perspective, the most practical parameter.
Efficacy of vaccination in 2 cohorts was different, as indicated by significant interaction between vaccine, cohort, and time in a model for ADG. In Cohort 1 vaccinated pigs had higher ADG compared with control pigs, whereas in Cohort 2 vaccinated pigs had numerically lower ADG than control pigs. The following factors may have contributed to the observed measures of vaccination efficacy in 2 cohorts: 1) spread of influenza virus in the study populations and consequent level of exposure, 2) initial level of maternally derived antibodies against influenza virus, 3) similarity between the vaccine and the field strain, 4) potential detrimental effect of vaccination, and 5) effect of other co-infections on vaccine efficacy.
The spread of influenza virus was assessed using serological response and further supported by evaluating counts of coughs over time. The SP ratios of vaccinated pigs developed similarly in both cohorts. This was expected, as vaccination commonly induces high HI titers which are protective (6,7,16). However, the SP ratios of the control pigs developed differently in the 2 cohorts, suggesting that Cohort 1 had a greater exposure to SIV. Our estimates of serocon-version risks for both six-week periods in Cohort 2 (0 to 6 wk and 6 to 12 wk) and for the initial six-week-period in Cohort 1 were in agreement with the results of the study by Loeffen et al (17), who estimated an incidence risk of 16% to 17% for a four-week period in nursery pigs. Seroconversion risk in the second six-week period of Cohort 1 was higher, suggesting more rapid transmission. In this cohort, a tendency for higher ADG and weight of vaccinated pigs began at week 12, coinciding with this period of greater SIV challenge.
Results of both the SP-ratio and empty models suggested slower SIV transmission than had been previously recorded in the herd. Such atypical SIV transmission has been previously reported (3), characterized by either partial seroconversion or seroconversion without clinical signs. During the first 6 wk of the study period, more variation in the probability of seroconversion at the pen-level was observed than during the second 6 wk in both cohorts. This might be explained by clustering of SIV transmission by pen in the early phase of the epidemic, followed by more uniform transmission among pens in the later phase. Different microclimates associated with weather conditions and air-exchange might have existed in this barn during the study, with a subsequent impact on SIV transmission. Slow SIV transmission might also be associated with a larger number of resistant individuals in the population (herd immunity). We do not believe, however, that this is likely to have played a role at the room level in this study, as < 25% of the population was vaccinated. It is, however, intriguing to consider whether herd immunity might have played a role at the pen level. Although we initially blocked by pen, the population was redistributed into new pens once they entered the finisher barn, which might have resulted in an unequal ratio of vaccinated to non-vaccinated individuals by pen.
In addition to serological responses, development of coughing over time was used to assess respiratory disease dynamics in the barn. The clinical expression of coughing in this herd, at least in the second part of the finisher period, suggested involvement of SIV, and SIV was isolated from pigs that coughed. Because of presumed low specificity of cough for influenza classification we consider development of cough only as a supporting evidence for influenza spread in a barn. Fixed parameters in a model for coughing suggested a higher incidence in Cohort 1 pigs, and an increase in incidence with time. In addition, both covariance parameters and pen-specific estimates of intercept and slope suggested considerable variation in the initial pen incidence of coughing and its development over time. This variation might have been influenced by 2 pens in Cohort 1 that had a high initial incidence of coughing that decreased with time, a pattern opposite to that in all other pens. Swine influenza virus was first isolated from 2 coughing pigs in one of these 2 pens, suggesting that it was the index pen for SIV infection. Although the cough model suggested an increase in the mean incidence of coughing over time, no pigs seroconverted between weeks 12 and 15. If the increase in coughing indeed indicated an increasing rate of SIV infection, this would probably not be evident on ELISA SP ratios because of the lag time between infection and seroconversion (18).
Atypical transmission of SIV may also be associated with maternally derived antibody (MDA) in pigs. An absence of typical clinical signs of SI, prolonged shedding time, and a tendency for a slower growth rate after exposure to SIV were identified in pigs with MDA, in a study by Loeffen et al (19). In addition, H1N1 vaccine efficacy was reduced after pigs were vaccinated in the presence of MDA and challenged with a heterologous H1N1 strain (20). Hence, the impact of MDA on clinical signs, transmission pattern, and on reduced vaccine efficacy in our study cannot be completely disregarded, but was not likely, since a maximum of 6% of pigs in a group were seropositive at the beginning of the study, and the mean SP ratio in each intervention group was below the cut-off recommended for vaccination (21). Moreover, the higher proportion of SIV-positive pigs was detected in Cohort 1, in which vaccine efficacy was higher than in Cohort 2.
Similarity between a vaccine strain and a field strain could also impact vaccine efficacy. Vaccine was matched at the subtype level as field isolate was classified as H1N1. However, previous research indicates that vaccine efficacy could be reduced after challenge with a virus with the identical HA, but low antigenic cross-reactivity to the vaccine strain. For example, an inactivated vaccine based on a H1N1 strain was not clinically efficacious after challenge with a heterologus H1N2 strain (22); and 3 different commercial vaccines containing cluster 1 H3N2 strains reduced clinical signs but not SIV shedding after challenge with a cluster 3 H3N2 strain (23). Assessment of genetic or antigenic similarity between vaccine and field strains in this study was not done, precluding us from further investigating this question. Given the epidemiological situation at the time study was done, involvement of other undetected subtypes was possible, but likely minimal because triple-reassortant H3N2 strain, that could contribute to further reassortments, emerged in Ontario approximately 1 y (in 2005) after this study ended (24,25), and pandemic H1N1 emerged in human population in 2009.
Detrimental effect of vaccination in the absence of sufficient challenge cannot be disregarded. This could occur through at least 2 different mechanisms, apart from unmatched vaccine and field strain. First, in an effort to match vaccine response to expected challenge the booster vaccination was scheduled before transport to the finisher barn. Such practice alone might have had detrimental effect on weight gain of animals. This stresses the importance of careful vaccination protocol design at times where pigs are exposed to other stressful events. Second, in other studies immune stimulation induced by vaccines, adjuvants, timing of vaccination, or age of animals at vaccination were suggested as factors that may contribute to expression of PCVAD (26–29), although this has not been consistently demonstrated under experimental conditions (30,31). In this study, there was no detectable difference in mortality or probability of treatment over time [data not shown; (15)] between groups vaccinated with different vaccines. This question could not be fully investigated due to the nature of experimental groups.
Of other infections that could influence vaccine efficacy, PRRS was likely the most important. In a recent study, Kitikoon et al (8) suggested that SI vaccination protected almost completely against SIV challenge in the absence of PRRSV infection, but only partially when pigs were infected with PRRSV, even in pigs with high SIV HI titers. The study herd was PRRSV-positive and nursery pigs were vaccinated with attenuated live PRRSV vaccine in the nursery. Thus, infection with PRRSV might have also lowered vaccine efficacy. Additionally, a recent study showed that PCV-2 infection resulted in higher severity of macroscopic lung lesions in pigs that were vaccinated with modified live PRRSv vaccine and subsequently challenged with PRRSv (32). Strict adherence to inclusion criteria, blocking, and randomization likely helped to equalize these factors between the treatment groups, but could not eliminate overall detrimental effect of these factors on the performance of entire population as well as on vaccine efficacy.
In conclusion, the results of this study suggest that vaccinating for SIV is of questionable value in a herd experiencing relatively low exposure to H1N1 influenza and a PCVAD outbreak. Influenza vaccination protocols should be considered carefully, not only to target the decline in maternal immunity, but also to avoid coinciding with other stressful events or concurrent disease outbreaks. In this study, vaccination induced a high serological response. Transmission of SIV was relatively slow and clinical signs were subtle and easy to miss, in agreement with observations of other authors (2,4). It is possible that this was the typical spread of H1N1 SIV in Ontario pig herds at the time study was done, in contrast to the outbreak form with dramatic coughing and inapettence.
The authors thank the cooperating swine producer and staff in the nursery and the finisher barn for participation in this study. Cooperation of Mitchell Veterinary Services and herd veterinarians is greatly appreciated. We are also thankful to Pfizer Animal Health for funding and other support for this study, and The University of Guelph — Ontario Ministry of Agriculture, Food, and Rural Affairs animal research program for funding.