A randomized, masked, 2-arm parallel trial was conducted to assess the efficacy of a Moraxella bovis (M. bovis) autogenous vaccine to prevent naturally occurring infectious bovine keratoconjunctivis (IBK) in beef calves.
A randomized, masked, 2-arm parallel trial was conducted to assess the efficacy of a Moraxella bovis (M. bovis) autogenous vaccine to prevent naturally occurring infectious bovine keratoconjunctivis (IBK) in beef calves.
The null hypothesis was that treatment group was not associated with either risk of IBK or last observed weight.
The trial was conducted between May and November 2009 and 2010 on a university-owned farm in Iowa. The vaccine contained 2 randomly selected M. bovis from IBK cases that occurred at the farm in 2008. Calves born between January and May 2009 and 2010 without visible corneal lesions were randomly allocated to receive vaccine (n = 191) or placebo (n = 178).
Two SC doses were administered 21–28 days apart. Allocation to treatment was concealed using bottles marked A or B. Staff observing the animals for IBK could not determine the treatment grouping. The herd met the “at-risk” criteria (ie, >15% IBK in unvaccinated calves and M. bovis detection in IBK cases). Analysis was “per-protocol”.
The risk of IBK was 58/185 (31%) in vaccinated calves and 66/173 (38%) in unvaccinated calves (adjusted risk ratio = 0.78; 95% CI, 0.49–1.24). Average weight before sale did not differ between the vaccinated calves (196.6 kg, SD ± 39.9) and unvaccinated calves (198.1 kg, SD ± 42.7) (P value = .19). No adverse effects were noted.
Combination of the study results with previous studies suggests that autogenous M. bovis vaccines often are ineffective in controlling naturally occurring IBK.
Infectious bovine keratoconjunctivitis (IBK) is a recurring disease problem for cow-calf producers. Calves with IBK exhibit a wide range of clinical signs that include lacrimation, photophobia, corneal edema, ocular pain, and corneal ulceration. IBK lesions have been associated with a 15–30 lb decrease in weaning weight compared to calves with no recognized IBK lesions.[1-3] Although most animals recover from infection without serious sequelae, corneal scarring and vision loss have been associated with acute infection.
Moraxella bovis (M. bovis) is considered the primary causal organism associated with IBK and all commercially available vaccines contain only M. bovis antigens, but the effectiveness of M. bovis vaccines as an IBK management tool is questionable. Two suggested reasons for failure of M. bovis vaccines to prevent IBK include the failure of commercial vaccines to contain the farm-specific M. bovis isolates or the presence of another causal organisms.
Autogenous vaccines can legally be used in the United States under the 1995 Virus-Serum-Toxin Act. Moraxella bovis autogenous vaccines are produced by a variety of companies and have been used widely by veterinarians and cattle producers. Four other published trials have explicitly evaluated an autogenous M. bovis vaccine for the control of IBK.[5-8] These studies failed to report true randomization to treatment cohorts or blinding of outcome assessment and all reported that vaccination was not associated with a statistically significant decrease in cumulative incidence of IBK. Despite these publications, autogenous M. bovis vaccines are routinely recommended and used.
The primary objective of this study was to conduct an independent evaluation of an autogenous M. bovis vaccine as a management tool to control IBK in a randomized blinded trial. The primary hypothesis tested was that the use of a M. bovis autogenous vaccine is not associated with IBK occurrence in beef calves. The secondary hypothesis tested was an autogenous vaccine is not associated with weight gain in beef calves. A secondary objective was to report the results from the previous studies and the current studies in a meta-analysis to provide a combined evidence summary regarding the effectiveness of autogenous IBK vaccines in beef calves.
The trial was approved by the ISU Institutional Animal Care and Use Committee (Log # 4-08-6550-B).
Calves from an Iowa State University (ISU) owned and managed research herd were enrolled in the study. The herd was conveniently identified by one of the authors (AOC). The herd had a long history of yearly IBK outbreaks, defined as more than 15% of calves affected with IBK in a single season. The ISU owned and managed herd has approximately 300 Angus cow-calf pairs and is located in Lucas County, Iowa. Calves born from January 1, 2009 to May 30, 2009 and 2010, with no visible eye lesions, were eligible for enrollment.
The autogenous vaccine was created from 2 strains of M. bovis isolated at Iowa State University Veterinary Diagnostic Laboratory (VDL). In 2008, eye swabs were obtained from all pinkeye cases in the herd and cultured by the ISU VDL. From the isolates identified, 6 randomly chosen strains were submitted to a commercial laboratory that markets autogenous vaccines to confirm these species as M. bovis isolates, and then 2 of the 6 isolates were randomly selected (RAND function in Excel) to be included in the vaccine.1 One thousand doses of vaccine were purchased at a single time, and used in 2009 and 2010 (product no. 12886, serial no. 009-48214, expiration date 11-28-2010). The vaccine was stored continuously in a 4°C refrigerator in the author's (AOC) laboratory.
At enrollment, each calf's weight was recorded and an eye swab collected using a sterile cotton swab rolled over the dorsal and ventral surfaces of the eye and under the third eyelid.
At enrollment, calves were randomly allocated to 1 of 2 treatment cohorts: Treatment 1: Two doses of 2 mL autogenous M. bovis vaccine SC, 21–28 days apart; and, Treatment 2: Two doses of 2 mL adjuvant (base product without antigen) SC, 21–28 days apart.
The follow-up period for recording IBK incidence started after the 2nd injection. The study continued until the animals were last handled at the farm. After this handling, calves were either sold or moved to another management group if retained as replacement animals. The weight collected at the last time handled is referred to as the last observed weight. As part of routine management, the calves also received 2 doses of clostridial vaccine SC,2 a topical anthelmintic3 and vaccines to prevent bovine respiratory diseases,4,5,6 during the study period.
The primary objective and hypothesis tested the association between autogenous vaccine and IBK occurrence. Determination of the primary outcome of interest, IBK occurrence, was determined by the farm staff who routinely observed calves, while on pasture, as part of routine farm management during the summers. The farm staff was provided with pictures of IBK lesions, a narrative description of the lesion, and the score that should be assigned. Each time an animal was observed, the corneal lesions were rated on a 5-point scale: normal eye = 0, lesions involving less than one-third of the cornea = 1, lesions involving between one-third and two-thirds of the cornea = 2, lesions involving more than two-thirds of the cornea = 3, lesions involving the entire cornea or ruptured eyes = 4. All animals with lesions were treated with florfenicol at the manufacturer's recommended dose. The farm staff had worked on IBK projects in the previous 6 years and was familiar with the clinical signs of IBK including corneal lesions, blepharospasm, and lacrimation. Not only was there additional routine monitoring for lesions on pasture, but the research project staff also conducted whole herd inspections, when the calves were handled for other management practices performed during the study period (Table 2). To evaluate the secondary hypothesis that vaccination with an autogenous vaccine is not associated with weight gain, the secondary outcome was last weight recorded on the individual calf (Table 2).
To confirm the presence of M. bovis in IBK cases during the study periods, each year cases were tested for M. bovis. Each summer when cattle were handled for typical management practices, swabs were collected from both eyes of all calves enrolled in the study. The swabs were transported to the laboratory for identification of M. bovis, M. ovis, and M. bovoculi. In 2009, samples were cultured on 5% sheep blood agar within 24 hours after being collected on the farm. Plates were checked each day from day 1 to day 3 and gray-white, friable, circular, and β hemolytic colonies typical of Moraxella spp. were selected, passaged, and stored at –80°C for later processing. In 2010, an in-house multiplex PCR was used to detect M. bovis, M. ovis, and M. bovoculi.7
The expected sample size to test the primary hypothesis was based on the previously documented 30% cumulative incidence of IBK at the ISU farm. Vaccine efficacy was estimated at 0.5, based on results reported previously of 3 randomized blinded trials (ie, cumulative incidence of disease was 15% in the vaccinated calves and 30% in the unvaccinated calves). The resulting sample size of 120 per treatment cohort allowed for a type 1 error rate of 0.05 and power of 0.8. Adjustments to the proposed sample size were made for a possible 5% loss of calves between enrollment and study conclusion attributable to non-study associated factors and therefore we aimed to enroll 126 animals per treatment group per vaccine.
The unit of treatment allocation and randomization was the individual calf. Before the ISU farm visits, containers holding the autogenous vaccine and placebo were relabeled injection A or B by staff who would not enroll animals at the farm. A chute processing order sheet was created, and a corresponding random allocation number between 0 and 1 was generated by an investigator not involved with enrollment.7 Random numbers between 0 and < 0.5 received injection A. Random numbers ≥ 0.5 received injection B. Students, including the authors (SB and SG), allocated the animals to treatment cohorts. Calves were restrained and both eyes examined for corneal lesions, and only calves with normal eyes were allocated to a treatment group. Calf identification numbers, weight, and sex were recorded. Revaccination of calves was performed 21–28 days after initial vaccination.
Initial data analysis was conducted using the coding as injection A and B by one of the investigators (AOC). The data were uncoded by the authors (AOC and SG) after analysis of the primary outcome. Injection A was the active vaccine and injection B was the placebo.
The primary outcome variable was IBK occurrence coded for the analysis as a binary variable. Calves with both eyes recorded as normal throughout the study period were considered IBK-negative and assigned the value 0. Calves diagnosed with at least 1 eye with a score ≥ 1 were considered IBK-positive and assigned the value 1. The outcome did not differentiate between calves with unilateral or bilateral lesions or severity of lesion. The secondary outcome variable was last observed weight, a continuous variable not transformed for analysis. For both the primary and secondary outcome, only animals that completed the study per protocol with complete data for treatment cohort, weight at enrollment, management group, sex, and last observed weight were included in the analyses.
Cumulative incidence of IBK was determined for vaccinated and un-vaccinated calves. Mean ± standard deviations (SD) were calculated for last observed weights for vaccinated and un-vaccinated calves. The distribution of the baseline characteristics, management group, sex, and enrollment weight, was calculated as either frequency or mean ± SD.
Because corneal lesions occur for a variety of reasons in beef cattle, and not all are of infectious origin, we defined at-risk herds as those with cumulative incidence of IBK ≥15% and M. bovis recovered from IBK cases. Fifteen percent was chosen by one of the authors (AOC) as indicative of an infectious process. A lower prevalence may be associated with a noninfectious process such as trauma, and therefore the population may not be at risk for IBK associated with M. bovis and evaluation of the vaccine would be inappropriate.
Descriptive analyses were calculated for each year the study was conducted and combined. To evaluate the null hypothesis for the primary outcome, IBK cumulative incidence over both years, univariable and multivariable logistic models were used. The explanatory variable of interest was treatment cohort. We modeled the risk of IBK occurrence (IBK = 1), and used un-vaccinated calves as the referent. We used a nonlinear mixed model to estimate the risk of IBK (β) for each year and vaccine group. The management group was included as a random effect. The standard errors of all estimates were calculated using the delta method. We then calculated the log of risk of IBK using formula (1):
We then calculated the log of the relative risk, using the log of the risk of IBK in vaccine group as the numerator and the log of the risk of IBK in unvaccinated group as the denominator. We then back transformed the log of the relative risk to calculate the point estimate and 95% confidence intervals for the preventive fraction in the vaccinated animals using formula (2):
The 95% confidence for the preventive fraction in the vaccinated animals (exposed) was back transformed from the 95% confidence limits of relative risk.
To evaluate the null hypothesis for the secondary outcome, the last observed weight, univariable and multivariable generalized linear models were used. The outcome variable was last observed weight and the explanatory variable of interest was treatment cohort. Unvaccinated calves were the referent; therefore if the autogenous vaccine was effective, the β from the model would be positive (ie, increased at the last time a weight was recorded associated with vaccination). For the adjusted analysis, the only covariate included in the model was year as a fixed effect. Model fit was assessed by evaluating the distribution of residuals of last weaning weight for a normal distribution and by looking for a characteristic plot of the observed versus predicted last weaning weight.
An ancillary analysis conducted was an evaluation of the association between weaning weight and IBK incidence. This analysis was planned before the study; however, determining weight loss associated with IBK occurrence was not a primary objective of the study. A multivariable generalized linear model was used. The outcome variable was last observed weight and the explanatory variable of interest was IBK. Covariates included in the model were weight at enrollment, management group, sex, and treatment cohort. Calves without IBK were used as the referent. Therefore, if IBK negatively affected weight gain the β from the model would be negative (ie, decreased weight gain associated with IBK incidence). Model fit was assessed by assessing the distribution of residuals of last weaning weight for a normal distribution and by looking for a characteristic plot of the observed versus predicted last observed weight. Descriptive and regression analyses were conducted using SAS (Version 9.2).
For the discussion, we included a forest plot of the current and previous studies on autogenous IBK vaccines. Pubmed and CAB abstracts were searched using the string “autogenous Moraxella bovis” and data were extracted from publications that reported using an autogenous M. bovis vaccine and naturally occurring IBK. Data from each study arm were extracted. A forest plot was created using data from the current trial and data from previously published trials of autogenous M. bovis vaccines. In the forest plot, each year of the current trial was treated as an independent study. When studies had more than 2 treatments, the size of the control group was split between the 2 active arms. We added 2 times the incremental increase of 0.5 to the number of events in the vaccine and control groups for the calculation of the relative risk for studies with a zero cell as used in RevMan 5, the Cochrane Collaboration's program for preparing and maintaining Cochrane reviews. A summary fixed effect measure was not calculated because the purpose was to provide a graphic description of all available studies, and a summary effect may not be appropriate because the vaccine doses administered and route differed.
The flow of study units is reported in Table 1. Observations were available on 358 animals. Because of low pregnancy rates at the ISU herd in 2009, we did not achieve the desired sample size per group, enrolling 93 and 88 animals per treatment group instead of the preferred 126, and therefore animals were enrolled in a second year. Because of management practices it was not possible to partially enroll the herd, so all animals were enrolled, which explains why the sample size is larger than required. The study started in May 2009 and was completed in November 2010.
|Available for enrollment||228||239|
|Not eligible at 1st vaccination||19||14|
|Not eligible at 2nd vaccination||28||37|
|Allocation to group||93||88||98||90|
|Lost to follow-up (death)||0||0||4||5|
|Included in analysis||91||88||94||85|
Recruitment of animals was coordinated with the farm managers and depended on parturition dates and availability of staff. Dates for allocation of treatment and observations are reported in Table 2. Cows and calves were managed in 2 groups: Group 1 which contained cows with their 2nd, 3rd or 4th calf and Group 2 which contained the cows with their 1st calf and cows with their 5th or more than the 5th calf. These groups were used for handling animals and for planned management practices.
|Group 1||Group 2||Group 1||Group 2|
|1st vaccination||June 23||June 30||June 22||June 29|
|2nd vaccination||July 14||July 21||July 13||July 20|
|1st observation||July 28||September 15||September 8*||August 12|
|2nd observation||August 20||October 7||September 29||August 30|
|Last observation||September 9||October 27||October 20||September 22|
Baseline characteristics of the treatment cohort in each year and combined results are reported in Table 3. Animals with abnormal eyes at the 1st and 2nd vaccination were not eligible for allocation to treatment. Over the 2 years of the study, animals allocated to treatment at 1st vaccination, 65 were excluded from the analysis because they developed corneal lesions between the 1st and 2nd vaccination or received the wrong treatment at the 2nd vaccination. Nine animals receiving the correct protocol were lost to follow-up because of death. Two animals were allocated to treatment and completed the study, but did not have a valid weight during the last time of observation; these animals were excluded from analysis. Data from 195 animals that received the autogenous vaccine and 173 that received the placebo were used in the final analyses.
|Vaccine (n = 91)||Placebo (n = 88)||Vaccine (n = 94)||Placebo (n = 85)|
|Enrollment weighta||108 ± 22||107 ± 21||231 ± 41||227 ± 45|
|Sex||Heifer||50 (45.0%)||34 (38.6%)||40 (47.1%)||45 (52.9%)|
|Bull||41 (55.0%)||54 (61.4%)||54 (57.4%)||40 (42.6%)|
Table 4 provides a summary of the primary and secondary objectives for the treatment cohorts. In both years, the farm met the criteria for an at-risk herd (ie, >15% cumulative incidence of IBK in unvaccinated calves and M. bovis in IBK cases).
|Year||Vaccine||Placebo||Relative Risk (95% CI)||Preventive Fraction in Exposed (95% CI)|
|2009||23 of 91 (25.3%)||31 of 88 (35.2%)||0.67 (0.33–1.37)a||0.32 (−0.37–0.66)a|
|2010||35 of 94 (37.2%)||35 of 85 (41.2%)||0.88 (0.48–1.63)a||0.12 (−0.63–0.51)a|
|2009–2010||58 of 185 (31.3%)||66 of 173 (38.1%)||0.78 (0.49–1.24)b||0.22 (−0.25–0.51)b|
The relative risk for occurrence of IBK occurrence over the 2 years based on the multivariable analysis was 0.78 with 95% CI limits from 0.49 to 1.24. The preventive fraction in the vaccinated animals was 22% (95% CI limits: 21–55%). These intervals clearly include the null values (1 for the relative risk and 0 for the preventive fraction) and provide no evidence to reject the null hypothesis that the cumulative incidence of IBK was different in the vaccinated and unvaccinated animals. Results for each year were similar and are provided in Table 4.
There was no significant difference for the secondary outcome, last observed weight, in the unadjusted or adjusted analysis (P = .19) (Table 5). The fit of the adjusted model was reasonable (ie, plots identified abnormal patterns).
|2009||219 ± 39 (n = 91)||223 ± 38 (n = 88)||NA|
|2010||175 ± 28 (n = 94)||173 ± 32 (n = 85)||NA|
|2009 + 2010||197 ± 40 (n = 185)||198 ± 43 (n = 173)||.19a|
The average unadjusted last observed weight of the calves diagnosed with IBK was 215 kg (SD = ± 47 kg) (Table 6). Average last observed weight of the unaffected calves was 224 kg (SD = ± 34 kg). After adjustment for covariates, this difference was statistically and biologically significant (β = −8.2 kg; 95% CI, 4.04–12.52). The fit of the model was reasonable (ie, plots identified no abnormal patterns).
|Year||IBK||No IBK||β (95% CI) (kg)||Wald P-value|
|2009||214 ± 47 (n = 54)||223 ± 34 (n = 125)||−6.4 (−25.4, −2.9)||.0134|
|2010||171 ± 33 (n = 70)||176 ± 27(n = 109)||−5.8(−28.2, 2.5)||.1008|
|Combined||190 ± 45 (n = 124)||201 ± 39 (n = 234)||−10.4(−35.5, −10.4)||.0003|
The PubMed search conducted on November 24, 2010 retrieved 5 publications, of which 4 were relevant. The CAB abstracts search conducted on November 24, 2010 retrieved 8 publications of which 1 was a relevant but did not include a control group and was therefore excluded, 3 were not relevant, and 4 were relevant duplicates with the PubMed search. None of the trials included reports of blinding of the outcome assessment, 1 study described systematic allocation to treatment group, and the other studies did not describe how animals were allocated to treatment group. The forest plot, with 95% CI, is provided in Figure 1.
The primary hypothesis tested was that vaccination with a M. bovis autogenous vaccine was not associated with the cumulative incidence of IBK in beef calves. We found no evidence to reject the null hypothesis. In both years, the farm met the definition of an at-risk herd and the difference in the cumulative incidence of IBK in vaccinated and unvaccinated calves was not statistically significant (0.78; 95% CI, 0.49–1.24). Our secondary hypothesis stated that vaccination with an M. bovis autogenous vaccine was not associated with weight at last observation in beef calves. Again, there was no evidence to reject the null hypothesis. Weight at last observation of the treatments cohorts did not differ meaningfully. Ancillary analysis of data did support previous reports that IBK infection can be associated with significantly decreased calf weight at weaning (Table 6).[1-3]
There are several possibilities why vaccination was not associated with the cumulative incidence of IBK or last observed weight. First, M. bovis, although frequently recovered from IBK lesions, may be incidental and not causative. This seems unlikely. There are many criteria for establishing causation, and one of the most credible is ability to reproduce the lesion with deliberate exposure and numerous studies have been able to do so. A second possible explanation for failure to observe a treatment effect may be failure of antigen presentation. Failure of the vaccine to present “correct” M. bovis antigen may render the vaccine ineffective or induce a nonprotective immune response. The vaccine was a killed product delivered by SC injection and likely elicited a systemic humoral (IgG) response to either the proteins on the bacterial membrane or internally produced proteins.[2, 12] The response to the vaccine therefore may not have been protective. There is no consensus about the immunoglobins and antigens needed to induce a protective immune response.[11, 13, 14] The results of the meta-analysis suggest that, despite good evidence that M. bovis is a cause of IBK, M. bovis autogenous vaccines consistently fail to offer protection. Given the positive publication bias in peer-reviewed literature, it is surprising that none of the published studies identified by the search reported that M. bovis autogenous vaccines were effective, and provides even stronger evidence that this type of vaccine is not effective for prevention of naturally occurring IBK.
Finally, it is also possible that this study design introduced a bias because of herd immunity. The design employed a vaccine strategy that no one recommends (ie, to vaccinate only a subset of the at-risk population). The assumption implicit with this study design is that protection against IBK is purely a function of the individual immune response. This ignores the potential role of herd immunity in preventing pathogen transmission. If a producer vaccinated only half of the cattle against IBK, we might not be surprised that disease occurred, even in vaccinated cattle. Therefore, among the reasons that this vaccine may have failed to protect cattle is failure to protect a sufficient proportion of the population. One finding that perhaps decreases the likelihood of this explanation is that herd immunity is associated with protection of all animals in the herd; however, in this study a high incidence of disease was observed, indicating that even if protection occurred it was not very effective.
Major sources of internal bias in the trial are unlikely to have affected the study results. Study design features such as randomization and masking were included to decrease the potential impact of confounding, misinformation, and selection bias on the outcome. Randomized and masked trials should provide the gold standard for assessing an intervention. Furthermore, loss to follow-up and lost records were minimal in both treatment cohorts (6/185 in the vaccinated cohort and 5/173 in unvaccinated cohort) and the reasons for loss to follow-up were unrelated to treatment cohort.
It would be overstating the importance of this single study to suggest that autogenous M. bovis vaccines are universally not effective, but in this study population a vaccine using M. bovis isolates from active IBK cases was not an effective management tool for controlling, preventing, or decreasing the incidence of IBK. The forest plot of previous studies conducted to answer the same question suggests that others have similar observations.
The authors thank Dr B. Fergen for help with the analysis. The authors thank Dennis Maxwell, Kevin Maher, and staff of Iowa State University's McNay Research and Demonstration Farm for their cooperation during the study. Assistance from J.R. Tait, Yimin Liu, and Lucas Funk with sample collection is gratefully acknowledged.
Financial support: Partial funding for this project was provided by Iowa Livestock Health Advisory Council.
Newport Laboratories, Worthington, MN
Ultrabac7/Somubac, Pfizer, New York, NY
Bovishield Gold 5, Pfizer
TSV 2 Naselgen, Pfizer
One Shot, Pfizer
Excel; Microsoft, Redmond, WA