A randomised controlled trial to assess the efficacy of Silirum vaccine in control of paratuberculosis in young farmed deer was carried out in 2008–2009 in six New Zealand herds with a history of clinical disease. Vaccination with Silirum was carried out in four-month-old deer, and vaccinates (n=1671) and controls (n=1664) were weighed at vaccination and at 8 and 12 months old, when faecal samples were collected from 125 vaccinates and 123 controls on five farms. Deer were slaughtered between 11 and 20 months of age, and the incidence of gross visceral lymph node (VLN) pathology typical of paratuberculosis in deer, that is, enlarged and/or granulomatous VLN, was recorded. Clinical disease was confirmed in 18 controls and seven vaccinates, representing a vaccine efficacy estimate of 60 per cent (95% CI 3 per cent to 83 per cent, P=0.04). Forty-seven percent (95% CI 38 per cent to 56 per cent) of faecal samples from vaccinates and 55 per cent (95% CI 46 per cent to 64 per cent) from controls were Mycobacterium avium subspecies paratuberculosis positive (P=0.5). Average daily liveweight gain did not differ between the cohorts. At slaughter, 1.4 per cent of vaccinates and 4.5 per cent of controls had VLN pathology, RR=0.32 (95% CI 0.19 to 0.54, P<0.001). These data indicate that vaccination with Silirum may be useful as an aid to control losses associated with clinical paratuberculosis in young deer.
Statistics from Altmetric.com
Paratuberculosis or Johne's disease (JD), a granulomatous enteropathy caused by Mycobacterium avium subspecies paratuberculosis (MAP), occurs globally in a range of species. Clinical disease was first recognised in farmed deer in New Zealand in 1979 (Gumbrell 1987) and since then, MAP infection has been confirmed in an increasing number of deer herds (de Lisle and others 2003). Young red deer have a higher annual clinical disease incidence than adults, with a median of 1.2 per cent in deer up to 12 months of age (weaner), 2 per cent in 12–24-month-old, and 0.93 per cent and 1.3 per cent in adult hinds and stags, respectively, on farms with confirmed MAP infection (Glossop and others 2008). The same study reported disease incidence up to 21.5 per cent in weaner deer, thus on some farms, JD can result in substantial economic losses (Bell 2006).
Vaccination as a control measure for JD has been used worldwide since 1926, primarily in young livestock since effectiveness of vaccination appears best when used before or soon after infection with MAP (Emery and Whittington 2004). When vaccine was used in cattle in the UK (Stuart 1965), Hungary (Kormendy 1994) and Spain (Garcia-Pariente and others 2005), a reduction in clinical disease incidence resulted, while Wilesmith (1982) reported that clinical JD had been eliminated in some UK cattle herds that used vaccine long-term. A recent field trial of Gudair, a whole-cell killed vaccine, in Australia (Reddacliff and others 2006) demonstrated a 90 per cent reduction in mortalities due to JD in vaccinated sheep.
Experimental studies with oil-adjuvant vaccines in deer have shown that some reduce the severity of subclinical JD (Mackintosh and others 2008). An experimental challenge trial using Silirum, an oil-adjuvant vaccine licensed for use in cattle, found significantly less gross pathology and a non-significant reduction in clinical disease (1/40 vaccinates vs 4/40 controls) in vaccinated deer (Mackintosh and Thompson 2007). Those observations justified investigation of vaccination on commercial farms, involving larger numbers of deer under natural challenge conditions.
This paper describes a randomised controlled trial of Silirum vaccine on commercially farmed young deer under field management conditions in a naturally infected population. The primary objective was to estimate vaccine efficacy (VE) against clinical JD. The effect of vaccination on average daily liveweight gain (ADG), faecal shedding of MAP and pathology at slaughter was also assessed.
Materials and methods
The study was conducted between March 2008 and August 2009 as a randomised controlled trial in commercial finishing deer herds with a history of clinical JD in young stock. Six study farms were selected on the basis of a history of clinical JD in young (<12-month-old) deer of at least 5 per cent in the previous two years, and with no recent changes in control interventions or source of stock, that is, they were anticipated to continue to have similar levels of disease. Additional inclusion criteria were previous confirmation of JD by culture, and/or a history of typical JD pathology at postmortem examination, based on the incidence of gross visceral lymph node (VLN) pathology consistent with JD in deer, that is, enlargement (≥55 mm circumference) and/or granulomatous lesions, in commercially slaughtered deer as recorded in a national deer JD database (Lynch 2007).
Regulatory compliance required that all study animals had to be slaughtered, at around 12–15 months of age, to eliminate the potential for confounding future testing for tuberculosis (TB). Candidate farms were identified via previous participation in a nationwide case-control study (Glossop and others 2007) or were nominated by veterinary practitioners. Any herds with an annual TB testing interval, considered at risk of bovine TB, were excluded, due to concerns about cross-reaction with TB diagnostic tests. All farms selected were in the South Island, located in the Canterbury, Southland and Otago regions and represented a reasonable geographic distribution of South Island deer finishing farms.
Vaccinated and control deer were grazed in the same management groups to avoid potentially confounding management and environmental effects, and ensuring each cohort experienced similar levels of exposure to MAP during the trial. The deer were all venison production stock comprising red deer (Cervus elaphus), and wapiti (C.e.canadensis)×red deer crossbreds, and were slaughtered, according to normal management practice, around 11-20 months of age. Farmers and their veterinary practitioners, who performed diagnostic investigations, were blinded to the vaccination status of the animals.
The vaccine (Silirum, Batch No. 06/001, Pfizer Animal Health) is a whole-cell inactivated vaccine of MAP strain 316F in an oil adjuvant. Vaccination was completed between March 15, 2008 and April 2, 2008, when the animals were around four months old, by the lead author and assistants. Whole mobs were yarded, and deer were handled in holding pens in small groups of 6–10. Half of each small group was selected for vaccination by systematic random sampling of every other animal, starting from the right or left side of the pen alternately for each group. Those selected for vaccination were injected subcutaneously in the right upper neck with 0.5 ml of Silirum, using a 1 ml syringe and 18-G×1/4 inch needle. The dose rate was determined from an earlier study assessing immunological responses at dose rates of 0.5 ml, 1 ml and 2 ml (Goodwin-Ray and others 2008). The control animals received no intervention.
The primary outcome measure was clinical disease incidence. The secondary outcomes were daily liveweight gain, faecal MAP shedding and gross VLN pathology. Data was also collected on the occurrence of injection site lesions resulting from vaccination and their effect on carcase quality.
The definition of a suspected clinical case was ‘loss of body condition and weight relative to herd-mates in animals that are otherwise eating well; weakness; poor coat appearance; and/or diarrhoea; unresponsiveness to treatment.’ Farmers were asked to contact their veterinarian when they identified suspected cases of JD in trial animals. A clinical examination and on-farm euthanasia and necropsy were then carried out if considered appropriate by the attending veterinarian. A standardised report on pathological findings was completed and samples were taken from grossly affected lymph nodes when present, or a pool of anterior, mid and posterior jejunal lymph node segments and ileocaecal lymph node, for MAP culture. Suspect clinical cases were confirmed only if typical gross carcase pathology, particularly of the gastro-intestinal tract, was identified at necropsy and MAP was isolated from tissue samples.
At vaccination in March 2008, the ear tag number, vaccination status and gender of each animal were manually recorded and each deer was weighed. On the four farms where electronic tags had been applied, individual weights were automatically recorded to electronic file while the remainder were manually recorded. Data were recorded to a relational database (Microsoft Access), with double entry of manually recorded data to ensure accuracy of transcription.
All deer were individually weighed in July/August and again in November, when faecal samples were collected from the five herds retaining most of their animals at that time. Twenty-five deer from each cohort were selected by systematic randomisation and approximately 10 g of faeces was collected per rectum using a fresh glove for each animal. Samples were immediately chilled and were despatched in an insulated container to the bacteriology laboratory by courier the same day.
The culture procedure applied a decontamination step using cetylpyridinium chloride and BACTEC 12B liquid culture medium containing egg yolk and mycobactin (de Lisle and others 2003). Growth indices (GI) were recorded as a semi-quantitative measure of the number of viable organisms in the sample (Reddacliff and others 2003). Sample GIs were measured weekly, and time to first detection of MAP, at a cumulative growth index of 15, was recorded for MAP-positive vials.
In November, the sites of vaccination of trial deer on three farms were palpated and callipers were used to measure any reactions identified.
The animals were sent for slaughter at a Deer Slaughter Premises (DSP) once they reached optimal slaughter weight, at the discretion of the farmer. Data on gross VLN pathology at postmortem examination were accessed from the national deer JD surveillance database, which records the incidence of enlarged (≥55 mm) or granulomatous VLN in deer slaughtered in New Zealand, as observed by meat inspectors.
All procedures on live animals were approved by the Massey University Animal Ethics Committee.
The null hypothesis for the primary outcome was that the cumulative incidence of clinical JD during the trial was the same in the vaccinated and control groups. The unit of analysis was the individual deer. Sample size calculations were based on a confidence level of 95 per cent, and a power of 80 per cent to detect a 50 per cent difference in disease incidence between the cohorts, postulating an incidence of 5 per cent in the controls and 2.5 per cent in the vaccinates. To examine the effect of vaccination on the shedding of MAP in faeces, sample size was estimated based on an anticipated prevalence of 50 per cent of culture-positive samples in controls, and 25 per cent in vaccinates. A design effect of two was included, to take account of an expected moderate intra-class correlation due to grouping of animals within the study herds. The required sample size to estimate efficacy against clinical disease was 3600 deer, while faecal shedding analysis required 232 deer. Liveweight and necropsy data were recorded for all trial deer.
Analyses were conducted using Stata V.10 (StataCorp, College Station, Texas, USA). Crude associations between vaccination and dichotomous outcomes (confirmed clinical cases, faecal MAP shedding and lymph node pathology) were examined for significance with the χ2 or Fisher's exact test on 2×2 contingency tables. Continuous outcomes (mean liveweight gain and time to positivity of faecal culture) were tested for significance of association with vaccination in univariate analysis using the Student's t test.
Multivariable analysis was carried out using general linear models to examine and control for the effects of covariates (gender, herd, breed and weight at vaccination) on the association between vaccination and each outcome. Binary outcomes were analysed with logistic regression, and estimates of relative risk were then derived using modified Poisson regression with robust error variance (Zou 2004). ADG was calculated using the difference between individual liveweights measured in March, July and November, divided by the number of intervening days using the actual weighing dates for each herd. Analysis of ADG used generalised estimating equations to adjust for repeated measures on the same subject. The potential effect of clustering of the observations within herds on the variance of the estimates was assessed by calculation of the intra-class correlation coefficient (ICC).
The VE estimate was derived from the preventable or attributable fraction comparing the risk of clinical disease (R) in vaccinated compared with unvaccinated individuals, assuming equal exposure to the infectious agent in each cohort (Halloran and others 1999). The denominators used were the number of individual animals in each cohort that were enrolled in the trial.
A total of 3335 deer were enrolled in March 2008, with 1671 randomised to the vaccinated and 1664 to the control cohort groups comprising 1206 hinds and 2127 stags; the gender of two animals was not recorded. Liveweights in March 2008 ranged from 25 to 76 kg (mean 53.7, sd 8.0) in vaccinates and 22.4–77.5 kg (mean 53.6, sd 8.5) in controls. The majority of the trial deer (n=2107) were red/wapiti crossbred, while the remainder (n=1228) were red deer. Management groups (mobs) on each farm consisted of a single breed type. No adverse events were reported following vaccination.
On-farm mortalities, including clinical suspects, numbered 88, of which 37 were vaccinates and 51 were controls. The incidence of confirmed clinical cases on each farm is presented in Table 1. All trial farms experienced a lower incidence of clinical disease in the 2008/2009 season than in previous seasons, evident in the trial animals and other similar mobs on the farms. Twenty-two controls and nine vaccinates were investigated as suspect cases of JD. Eighteen controls and seven vaccinates were confirmed by necropsy and culture. No culture results were available for two vaccinated suspect cases due to severe autolysis of the carcase. Two suspect cases from the control group showed no evidence of pathology and were negative at culture; one suspect control had typical pathology but MAP was not isolated, while another suspect control was found to have a chronic leg injury and no pathological evidence of JD, despite being MAP culture positive. Five of the confirmed clinical cases were hinds, 20 were stags, while 13 were red deer and 12 were wapiti hybrids. No further analysis was possible on the breed variable, as it could not be separated from herd effects.
The crude risk difference in clinical disease was 0.007 (95% CI 0.0008 to 0.012) with a risk ratio of 0.39 (95% CI 0.16 to 0.92, P=0.03). The risk ratio estimate and statistical significance of the effect of vaccination was altered only slightly by multivariable analysis, with the adjusted measure of VE estimated at 60 per cent (95% CI 3 per cent to 83 per cent, P=0.04). The ICC for these data was 0.004, so no further adjustment for clustering was made.
While univariable analysis of the risk ratio of confirmed clinical disease in stags (0.9 per cent) vs hinds (0.4 per cent) was not significant (RR=2.3, 95% CI 0.85 to 6.0, P=0.09), after adjusting for the effects of covariates, the RR for stags was 3.3 (95% CI 1.2 to 9.2, P=0.02).
Liveweight and daily liveweight gain
Liveweights measured at vaccination and mid-trial in July/August differed by less than 0.1 kg between cohorts across all farms (P=0.4). ADG was 119 g/day for vaccinates and 118 g/day for controls over the entire measurement period of March to November. No statistically significant differences in ADG were observed within or across farms (P=0.6) (Fig 1).
Multivariable analysis found no significant association between vaccination and ADG, either over the whole measurement period or during the specific time periods of March to July or July to November.
The proportion of faecal samples MAP positive for each farm is presented in Fig 2. MAP was isolated from 127 of 248 faecal samples (51 per cent) comprising 47 per cent (95% CI 38 per cent to 56 per cent) of vaccinated and 55 per cent (95% CI 46 per cent to 64 per cent) of control animals. The prevalence on individual farms ranged from 20 per cent to 80 per cent. The adjusted effect of vaccination on faecal MAP positivity across all herds was not statistically significant (RR 0.9, 95% CI 0.7 to 1.1, P=0.5).
Sixty-one per cent (94/155) of stags sampled were culture positive compared with 34 per cent (33/92) of hinds (RR 1.4, 95% CI 1.02 to 1.9, P=0.04).
The mean time to detection of MAP in culture was five weeks for each cohort (P=0.4).
Injection site lesions
Examination of vaccination sites of 486 deer in November revealed 181 (38 per cent) with palpable subcutaneous lesions. Reaction size ranged from 5 to 39 mm (mean 14.4, sd 5.93). Most were firm circumscribed reactions, although two discharging sinuses were found.
At slaughter, meat inspection staff reported that although vaccination site reactions were evident, these were generally circumscribed and were removed with the hide or were easily trimmed. Recording was not consistent between plants, so the data is not presented. No difficulties were experienced with access to export markets, nor was there any apparent loss of carcase value of vaccinated deer due to injection site lesions. The New Zealand Food Safety Authority (NZFSA) currently requires additional inspection of deer vaccinated against JD: ‘Palpation and deep incision of the muscles lateral and parallel to the ligamentum nuchae at or about the likely site of injection. Lengthen the incisions when suspicious lesions have migrated along the lymphatics of fascial planes’. No tracking lesions were identified in vaccinated animals as a result of these additional procedures. A granulomatous lesion within the prescapular lymph node of one vaccinated animal resulted in detention of the carcase in accordance with routine TB suspect procedures. Histological examination was unable to rule out TB, and the lesion was cultured for Mycobacterium bovis with negative results, allowing release of the carcase.
Pathology at slaughter
Data on gross VLN pathology for 2516 of the trial deer are presented in Fig 3. There were missing data resulting from a failure to adequately record individual animal identification in one DSP due to problems with barcode scanning equipment. The missing data were evenly distributed between vaccinates (n=367) and controls (n=364). The overall cumulative incidence of pathology was 2.9 per cent (74/2516) comprising 18/1267 vaccinates (1.4 per cent) and 56/1249 controls (4.5 per cent), (P<0.0001). Granulomatous lymph node lesions were recorded in two vaccinates and eight controls (P=0.02). Fifty stags (3.1 per cent) and 24 hinds (2.6 per cent) (P=0.5) were recorded with VLN pathology. The adjusted risk ratio for gross lymph node pathology in vaccinates compared with controls was 0.32 (95% CI 0.19 to 0.54, P=0.001). The ICC for these data was 0.004, so no further adjustment for clustering was applied.
This field trial found that vaccination with Silirum significantly reduced the incidence of clinical JD and gross VLN pathology in young deer. There were no significant differences between vaccinates and controls in ADG, or in prevalence or degree of faecal MAP shedding.
The low incidence of clinical disease recorded during the study (1.1 per cent in controls, compared with approximately 5 per cent reported previously on the trial farms) limited the precision of the estimate of VE, although the power of the study was still sufficient to demonstrate statistical significance. Incidence rates are known to vary between years, but another possible explanation is that disease incidence was influenced by the herd immunity effect. Direct protective effects of vaccination occur at the individual animal level and indirect effects at the herd level. Herd immunity thus refers to the protection of non-vaccinates due to the presence of immune individuals and the resultant reduction in sources of infection (Fine 1993). At the design stage of this study, we decided to maintain the vaccinated and control deer mixed together in the same management mobs. This was to ensure that all trial animals had the same environmental, management and infection exposures. The effect of herd immunity in this design may bias the estimate of VE towards the null, if vaccination reduces transmission of infection and controls have reduced exposure as a result. The estimate is similarly biased towards the null if vaccinates are heavily exposed to the agent by the presence of infectious controls, thus overcoming their immunity (Dohoo and others 2003). However, in deer, transmission of MAP may be via the intra-uterine route (van Kooten and others 2006, Thompson and others 2007) and young deer on infected farms may also be exposed to MAP via infected colostrum or milk (Thompson and others 2007) or from the environment in the first few months of life. The finishing deer were brought on to the individual trial farms at around four months of age. They were likely to have already been exposed to infection on their farms of origin before the trial started, thus limiting the potential influence of herd immunity.
The presence of a herd immunity effect is not supported by results of individual faecal sampling, since vaccination had no significant effect on the proportion of deer shedding MAP in faeces or on the semi-quantitative measurement of MAP. This demonstrates the likelihood of continuing exposure of all trial deer to the organism. Furthermore, other (non-trial) mobs of deer on the farms experienced similarly low levels of clinical JD during the trial period. All the deer herd managers considered that the mild winter and good spring pasture growth in 2008 contributed to lower than previous disease rates. The combined evidence thus suggests that herd immunity was not a significant factor in this trial.
Failure to find an effect of vaccination on faecal MAP shedding prevalence is consistent with other studies of JD vaccination in deer (Mackintosh and others 2008) and in dairy cattle (Kalis and others 2001). By contrast, Reddacliff and others (2006) reported a 90 per cent reduction in faecal shedding in vaccinated sheep. The power of the study presented here was designed to detect a 25 per cent difference in prevalence of faecal shedding between the cohorts. A larger sample size may indeed have found statistical significance, but with 47 per cent of vaccinates shedding MAP in faeces, it would not have altered the biological relevance in terms of continuing transmission of the agent.
The isolation of MAP from the faeces of so many deer on farms with little clinical disease was notable. On Farm 3, for example, MAP was isolated from 80 per cent of all faecal samples, yet the incidence of clinical disease in trial deer was 0.4 per cent, and this was the best-performing herd in terms of mean November liveweight. Farms with the lowest disease incidence had the lowest prevalence of faecal MAP positivity, although on farms with no reported clinical disease, the proportion of culture positive control animals was still up to 30 per cent. These proportions of potentially infectious animals raise questions about the impact of infection per se, and the specific factors that influence progression from infection to clinical disease in deer.
Passive shedding, in which ingested bacteria are shed in faeces without infecting the host, has been reported in sheep, goats, cattle and deer (Whittington and Sergeant 2001). How frequently passive shedding occurs and how important it is in disease transmission have not been established in deer. However, it is possible that passive shedding may have occurred in a proportion of trial deer, biasing the measure of effect of vaccination to the null (Copeland and others 1977), as both vaccinates and control deer were similarly exposed to MAP in the environment.
The quantitative measure for MAP concentration in faeces applied in this study was crude, with analysis limited to time to first evidence of MAP in culture media. Further work to derive a more precise quantification measure is required before robust conclusions can be made.
Gender was a risk factor in this study, with stags having over three times the risk of clinical disease and a higher risk (RR 1.4) of shedding MAP in faeces. The explanation for this finding may lie in the effects of environmental and nutritional stress impacting more on faster-growing animals.
No effect of vaccination on ADG was shown in this trial, consistent with the findings of a previous experimental challenge study of Silirum vaccination in New Zealand deer (Mackintosh and Thompson 2007). A sheep vaccine trial (Reddacliff and others 2006) recorded significantly higher liveweights (0.73 kg, P<0.05) in young control sheep than vaccinates only at 12 months postvaccination. The study herds were selected to represent herds with a high incidence of JD in young deer, and results may thus not be generalisable to all infected herds. However, failure to find an effect in this population suggests that a production effect is even less likely to be observed in herds with lower infection prevalence.
The data on gross VLN pathology at meat inspection should be interpreted with some caution. The sensitivity of meat inspection to detect enlarged and/or granulomatous VLN is estimated at 13 per cent (Hunnam and others 2013b), while the predictive value of enlarged VLNs as an indicator of MAP infection is estimated at 95 per cent (Hunnam and others 2013a). However, the identification and recording of findings is known to be variable between individuals and DSPs. Any misclassification is likely to be non-differential between the cohorts, thus biasing estimates of effect towards the null. The NZFSA requirement to carry out additional inspection on vaccinated animals meant that individual DSPs needed advance notification if vaccinated animals were to be presented for slaughter. In most DSPs, though, it was logistically simpler to treat the whole consignment as vaccinated rather than separate the cohorts in the lairage, and meat inspectors were thus effectively blinded to the individual carcase vaccination status. However, one DSP limited the number of vaccinates that could be presented each day and inspected them separately. Although the status of the carcases would then have been known, differential bias in reporting was considered unlikely but, if present, was more likely to bias the estimate of effect to the null since carcases of vaccinates were subjected to a higher degree of inspection. The finding of an effect of vaccination on the incidence of gross VLN node pathology is thus considered robust. It is also important in terms of efficiency of carcase processing. At present, the similarity of granulomatous lesions associated with MAP infection to those caused by M bovis in deer causes difficulty for the meat industry (Campbell 1995), as carcases in which these lesions have been identified require detention pending laboratory investigation to exclude M bovis as a diagnosis. There are logistical implications and increased costs for the DSP and the authorities in sample transport and laboratory procedures, and detained carcases may miss the opportunity of sale to export markets. A reduction in levels of such pathology, therefore, benefits the producer and the processor.
The deer in this trial were vaccinated around weaning, around four months of age, as this was the first opportunity to do so on most of the trial farms. It is possible that vaccination may be more effective if administered earlier in life, but on many breeding farms deer calves are not handled until this stage. Further research to assess the effect of the vaccine when used at a younger age may be useful to inform those deer managers who are able or willing to implement vaccination earlier.
In conclusion, Silirum vaccine was effective in reducing the incidence of clinical JD and gross VLN pathology in young deer vaccinated at approximately four months of age, suggesting that Silirum may be useful as an aid to control losses associated with clinical JD in young deer.
We would like to thank the large number of students who helped with the trial field work, as well as staff at AgResearch, Wallaceville and the Disease Research Laboratory, Otago University, and veterinary practitioners for diagnostic support. Special thanks go to the farmers for contributing their animals, time and enthusiasm to the study.
Funding The study was funded by Pfizer Animal Health Limited (now Zoetis NZ Ltd) and supported by Massey University and International Doctoral Research Scholarships.
Provenance: Not commissioned; externally peer reviewed
If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.