A Systematic Review of Campylobacter jejuni Vaccine Candidates for Chickens

Campylobacter jejuni infection linked to the consumption of contaminated poultry products is one of the leading causes of human enteric illness worldwide. Vaccination of chickens is one of the potential strategies that could be used to control C. jejuni colonization. To date, various C. jejuni vaccines using potential antigens have been evaluated, but a challenge in identifying the most effective formulation is the wide variability in vaccine efficacies reported. A systematic review was undertaken to compare C. jejuni vaccine studies. Based upon specific selection criteria eligible papers were identified and included in the analysis. Vaccine efficacy reported from different C. jejuni antigens, vaccine types, and vaccination regimens reported in these papers were reviewed. Our analysis shows that total outer membrane proteins and cysteine ABC transporter substrate-binding protein were among the most efficacious vaccine antigen candidates reported. This review also highlights the importance of the need for increased consistency in the way C. jejuni vaccine studies in poultry are designed and reported in order to be able to undertake a robust comparison of C. jejuni vaccine candidates.


Introduction
Campylobacter jejuni is considered an important zoonotic pathogen causing enteric illness in humans globally [1][2][3]. Outbreaks are commonly linked to the consumption of contaminated poultry products [4][5][6]. Poultry is considered a reservoir host of C. jejuni because this pathogen commensally colonizes the intestines where it can be present in large bacterial loads [7]. Based on quantitative risk assessment and regression models in previous studies, a low C. jejuni prevalence (a percentage/proportion of colonized chickens in a flock) between chicken flocks or a 1 to 2 log10 reduction of C. jejuni loads in broiler intestines could lead to a decrease in public health risk [8][9][10]. Thus, both the reduction in C. jejuni concentration and prevention of campylobacter colonization of chickens on farms are the most effective approaches to reduce the risk of campylobacter contamination of chicken meat [9]. To date, researchers have endeavored to develop and evaluate several interventions in primary broiler production including biosecurity monitoring [11], use of feed additives [12][13][14], drinking water sanitation [15], use of bacteriophage [16], probiotics [17,18], and bacteriocins [19]. Although some of these interventions have led to significant reductions in C. jejuni loads in the intestines of chickens, none of them have eliminated or prevented C. jejuni colonization of poultry.

Selection Criteria
Eligibility of studies for inclusion in this review was determined using a two-step process (primary and secondary inclusion/exclusion criteria) ( Table 2). Table 2. Inclusion and exclusion criteria in this study.

Process Inclusion Criteria Exclusion Criteria
Vaccine trials using layers were excluded in this review due to reported differences in C. jejuni colonization and immune responses between layer and broiler chicken breeds [32,33] and an expectation of reducing the public health risk [9,34]. If more than one sample type (i.e., ceca and cloaca) were evaluated in a single study, the result for the cecal sample was selected for the purposes of this study [30]. In applying this, we have not differentiated trials using conventional (bacterial culture) and/or molecular techniques (i.e., qPCR) to determine the colonization status of chickens at the end of vaccination studies [35,36]. Previous studies reported that both bacterial culture and qPCR methods had a high correlation (>99%) for enumerating C. jejuni in intestinal samples [22] and no significant difference in the detection of Campylobacter in chicken faecal samples [37]. Studies where the required details for the vaccine efficacy data were not provided in a usable format, the author (P.P.) contacted the study corresponding author of the original articles requesting the missing data via email two times. None of the corresponding authors responded to these requests.

Defining Vaccine Efficacy for Selecting Eligible Studies for This Review
The effectiveness of various C. jejuni controlling interventions at broiler farms is commonly evaluated using reduction levels in the prevalence of colonized chickens in a flock and/or reductions of C. jejuni loads in the broiler intestine at slaughterhouse [9]. Previous studies, EFSA [9] and Rosenquist et al. [34] have reported that a decrease in the prevalence of C. jejuni between and within broiler flocks could reduce bacterial loads in carcasses at slaughter and consequently reduce the incidence of human campylobacteriosis. Moreover, Nauta et al. [10] reported that a 1-2 log reduction in C. jejuni loads in gut contents had an impact on the human health risk of campylobacteriosis with a relative risk reduction by at least 44% based on regression and risk assessment models. Therefore, articles reporting vaccine efficacy based upon prevalence or proportion of "diseased" (i.e., colonized) chickens in a flock or group [38] and/or the reduction levels of C. jejuni colonization in vaccinated and unvaccinated chicken groups after C. jejuni challenge, were included in this review.

Data Extraction
All research articles identified from the three databases were entered in Microsoft Excel datasheets and duplicate studies were removed by one author (P.P.). One author (P.P.) initially inspected the titles and abstracts from the individual articles to select articles for inclusion in the review. If those titles and abstracts fitted the selection criteria, the full text of each potential article was further examined for the final determinations of eligible studies. At this stage, the full text was reviewed to classify the eligible studies and trials based on the vaccine-controlled efficacy trials described and to extract the relevant information. Subsequently, another independent reviewer (T.V.) validated the data and results. For any disagreements, all conflicts were resolved by consensus, and a third author (T.J.M. or R.A.H.) was asked to confirm whether the articles should be included or excluded. The final lists of the eligible article were imported to the EndnoteX9 program for storage, and consolidation (P.P.).
The extracted information of the individual eligible studies included article identification (authors and publication year), the title of each article, study type, information of animal models (poultry species), vaccine types, vaccine regimen (dosages of vaccine and adjuvants, ages of chickens, antigen candidates, frequency of vaccination, and bacterial challenge strains), samples collected, isolation test and outcome measurements of vaccine efficacy between the vaccinated and unvaccinated groups at the end of study. The concentration of C. jejuni loads in cecal contents and/or reduction levels of C. jejuni colonization reported in text and tabulation and/or estimated from figures provided in the original papers were included in this review. For multiple trials reported in each paper, each trial was considered as a separate trial unless the trials using the same vaccine protocols and evaluation methods. If only one control group was used to compare with more than one type of vaccine in the same experiment, this control group was used for each comparison. If two control groups were used in the DNA vaccine study, one control group with the parent plasmids (no insertion of an antigen of interest) was used as the control group [39]. The extracted information was summarized in Microsoft Excel datasheets.

Data Analysis
The extracted data were analysed with the aim of conducting a systematic review and/or meta-analysis. Based on the data extracted using a definition of prevalence detected in vaccinated broilers, six eligible articles (18 trials) reporting different C. jejuni antigens, vaccine types, and vaccine protocols were identified. Consequently, it was possible to undertake a systematic review, but the data were insufficient to conduct a meta-analysis. The outcomes of individual vaccine trials of the eligible studies were extracted, analysed, and reported as a percentage (proportion) of colonized broilers and relative risk (RR) with 95% confidence interval (CI). Trials with RR < 1.00 were further analysed with respect to vaccine efficacy as it indicates that the exposed (vaccinated) group could reduce a ratio of the risk or possibility of disease (C. jejuni) occurrence, compared to the unexposed (non-vaccinated) group [40]. While trials with RR is ≥ 1.00 were reported as having no effect in this review. The efficacy of vaccine was calculated as (1 − RR) and reported as a percentage [39,41,42]. The R software program (Version 1.3.1056, the R Foundation, Vienna, Austria) was used for calculating these estimates [43].
Trials demonstrating a log10 reduction of C. jejuni loads between vaccinated and non-vaccinated broilers and reporting this as a geometric or arithmetic mean or median of log10 (CFU/gram) loads of each treatment group were included. Extracted data from 62 trials reported in 16 papers were included for this review.

Search Results
A total of 1556 articles were retrieved from the three electronic databases (PubMed Central, Scopus, and Elsevier ScienceDirect). Of these, 1488 articles (95.6%) were assessed using the text of the title and abstract after the removal of duplicates. The selection process used in the current study is shown in Figure 1.

Eligibility
The remaining 39 articles fulfilled the initial selection criteria for further assessing the full text for eligibility and were published within the search period. From the 37 articles, 186 trials were identified, and these trials involved the evaluation of C. jejuni vaccines conducted in chickens (layers and broilers). Of the 186 trials, 66 trials conducted in layer chickens were excluded. Thus, 120 trials conducted in broilers were included for further review using two different focuses of vaccine efficacy (the prevalence of colonized broilers and significant log10 reduction levels) reported in text/tabulations of the original papers.
Based on the full-text evaluation using the investigation on the C. jejuni loads in cecal contents, 58 of 120 vaccine trials in broilers were excluded as they were a seeder colonization challenge model (n = 22), immunogenicity studies (n = 18), vaccine efficacy evaluated from ileum and cloaca (n = 15), trials using co-administration of vaccine and probiotics (n = 2), vaccine efficacy reported as prevalence (n = 1). Consequently, 62 vaccine trials fulfilled the selection criteria of this review. A summary of the details of these studies is shown in Table 3.
Based on the full-text evaluation using the prevalence of colonized chickens, 75 of 120 trials (85.0%) were excluded. The majority of the excluded trials were vaccine trials using a seeder colonization challenge model (n = 22), followed by data of number of individual colonized broiler not reported or unable to estimate from figures (n = 18), immune responses reported (n = 18), vaccine efficacy evaluated from ileum and cloaca (n = 15; 4 articles) [22,28,44,45], and trials using co-administration of vaccine and probiotics (n = 2; 1 article) [46]. The remaining 45 trials (10.1%) from 13 articles fulfilled the selection criteria and were included in the systematic review.

Vaccine Types
Overall, eight vaccine types (bacterial vector-based, subunit, DNA, a combination of vaccine, killed-whole cells, cell lysate, crude cell lysate, and conjugated vaccines) were identified in this review using the two definitions of vaccine efficacy. Based on the 62 trials with the C. jejuni loads in ceca of the vaccinated and non-vaccinated broilers investigated, the bacterial vector-based vaccines were the most frequently used in 19 trials from seven papers. Of these, Salmonella Typhimurium (ST) and Salmonella Enteritis (SE) vectors were used in 16 trials, followed by Lactobacillus lactis (L. lactis) (n = 2) and Escherichia coli (E. coli) (n = 1). Subunit vaccine (n = 12) and DNA (prime) with subunit (boost) vaccine (n = 12) were identified in six and two papers. Crude lysate vaccine was used in eight trials from one paper. DNA, Whole-cell lysate, and conjugated vaccines were used in nine trials from four papers. The remaining vaccine type used in two trials from one paper was a killed-whole cell vaccine.
Based on the 45 trials with the prevalence of colonized broilers reported, subunit vaccine was the most frequently used in 13 trials from seven papers, followed by a combination of DNA (prime) and subunit (boost) vaccines from 12 trials of two papers. The crude cell lysate vaccine was found in eight trials of one paper [47]. The bacterial vector-based vaccines used in six trials from three papers were E. coli wzy::kan strain [46], ST χ9088 strain [56], and Avirulent ST χ3987 strain [26]. While the use of whole-cell lysate (n = 3) and DNA (n = 3) vaccines were found in three papers.

Vaccine Antigens and Vaccine Regimens
The C. jejuni antigens evaluated in the vaccine efficacy studies included in this review are summarized in Table 4.
A total of 23 C. jejuni antigens used as single and/or multiple antigens in vaccine trials were identified using both terms of vaccine efficacy in this review. Based on the 62 trials with evaluations of C. jejuni loads reported, variations of antigens used in the trials, vaccine regimens, and the broiler age at the end of study (ranging from 25 to 128 days) were identified (Table 3). Of these, total outer membrane proteins (OMP) used in the crude lysate vaccine were the most frequent antigen evaluated after homologous challenge in 8 trials from one paper. This antigen was used either encapsulated with biodegradable and biocompatible poly (lactide-co-glycolide) nanoparticles (OMP-NP) or non-encapsulated via oral or subcutaneous vaccinations with a booster. Following this, cysteine ABC transporter substrate-binding protein (CjaA) used in five trials from three papers were evaluated in the oral vaccination with booster(s) using three different bacterial vectored vaccines: ST (∆aroA) mutant-1 (STM-1) (n = 3), L. lactis NZ3900 strain (n = 2) and, avirulent ST χ398 strain (n = 1). Of these five trials, the only vaccine trial using the avirulent ST χ398 strain expressing CjaA was challenged with heterologous C. jejuni strains. Flagellin A protein used in 4 trials (two papers) were evaluated three different vaccine types (DNA, subunit, and DNA (prime)/subunit (boost) vaccines) and routes of administrations (intramuscularly or subcutaneously with booster). Hypothetical protein YP99817.1 protein used in four trials was evaluated using three different vaccine types booster as well, but only one vaccinated route (intramuscularly with/without booster) was used and these trials were from only one paper. Three antigens (i.e., Campylobacter adhesion protein to fibronectin (CadF), whole-cell lysate, and capsular polysaccharide (CPS)) were used and evaluated upon homologous challenge in nine trials from four different papers. CadF identified in two papers was evaluated in subunit (intramuscularly with booster) and STM-1 vectored vaccines (orally with booster) after homologous challenge. Whole lysate and CPS were used in six trials from two papers. The whole-cell lysate was orally administrated with/without E-CpG, whereas CPS was conjugated with diphtheria toxoid of Corynebacterium diphtheriae vaccine (CPSconj) and mixed with 10 µg of CpG ODN 2007 for subcutaneous vaccination with a booster. The remaining trials using other antigens were less frequent studies (less than 2 trials) ( Table 3). Based on the prevalence of colonized broilers reported, several antigens were also identified in the 45 eligible trials conducted in various ages of broiler at the end of study ranging between 25 and 44 days old ( Table 5). The antigens used in the 45 trials were a subset of the antigens used, based on the investigation of C. jejuni loads criteria except for an extra trial which was a subunit vaccine using recombinant flagellin A protein (FlaA) fused to heat-labile enterotoxin (LT-B) of E.coli (FlaA-LT-B) mixed with sodium carbonate, delivered orally with a booster (designated as Trial no. 63) in this review (Table 5). Of the 45 trials, the OMP (with/without NP) used in the crude lysate vaccine were the most common antigens used in eight trials, following this, FlaA (n = 5, three papers), hypothetical protein YP_999817.1 (n =4, one paper), whole-cell lysate (n = 3, one paper), and CjaA (n = 3, two papers). Seven antigens were used in 14 trials from five papers using different vaccine formulations and regimens (Table 5). These seven antigens were DNA binding protein for biofilm formation (Dps), flagellin, hemolysin co-regulated protein (Hcp), flagellin protein family (FlgL), Campylobacter hemolysin activation/secretion protein (YP_001000437.1), C. jejuni surface-exposed lipoprotein A (JlpA), and hypothetical protein YP99838.1. While the remaining eight trials were utilized eight different antigens (six individual and two multiple antigens).

Levels of C. jejuni Loads (log10 CFU/g) in Cecal Contents as Vaccine Efficacy
The four different outcomes of vaccine efficacy reported from the 62 trials were identified in this review: no reduction (n = 9), log10 reductions (n = 15), non-significant log10 reductions (n = 20), and significant log10 reductions (n = 18). The four different reporting outcomes of log10 CFU/gram were also identified in this review (i.e., genome equivalents per reaction per gram, median log10 reductions, geometric log10 reduction, and arithmetic mean log10 reduction).
The 18 trials reported significant log10 reductions ranging between 0.5 log10 and 6.0 log10 of C. jejuni cecal loads upon homologous/heterologous challenge. Of these, an avirulent ST χ3987 strain vectored vaccine expressing CjaA (10 8 CFU) administrated orally and a booster was the most significant levels of reductions (~6.0 log10) after heterologous challenge reported [26]. Following this, the crude cell lysate vaccine contained 125 µg of OMP or OMP-NP subcutaneously with booster (two trials) provided approximately 5.7 log10 reductions after homologous challenges, compared to the broiler vaccinated with NP alone estimated from the figure provided in the original paper [47]. Eight trials reported significant reduction levels of C. jejuni loads varied between 2.0 log10 and 4.27 (mean or median) log10 CFU/g after homologous/heterologous challenge. Of the eight trials, four trials were a combination of DNA (prime) with subunit (boost) vaccines using four antigens: Campylobacter hemolysin activation/secretion protein, FlaA, fibronectin-like protein A (FlpA), flagellin protein family (FlgL), hypothetical protein YP99838.1, or hypothetical protein YP99817.1, and a CadF-FlaA-FlpA) were from one paper [50]. Another four trials (from three papers). Another four trials were cell lysate vaccines using 4.3 µg of C. jejuni cell lysates with/without 5 µg of E-CpG (orally), E. coli wzy::kan strain vectored vaccine expressing C. jejuni protein glycosylation (N-glycan) (orally with a booster, Salmonella Typhimurium strain χ9088 vectored vaccine expressing DNA binding protein for biofilm formation (Dps) orally with a booster. The significant log10 reductions reported in seven trials were less than 1.0 log10 CFU/g. Moreover, 15 trials from two papers reported levels of log10 CFU/g reductions without significant or non-significant reported. Of these, five trials with subunit vaccines contained recombinant CadF, FlaA, FlpA, a component of multidrug efflux pump (CmeC), a fusion protein of CadF-FlaA,-FlpA emulsified with MONTANIDE™ ISA70 VG intramuscularly with booster provide various reduction levels (between 1.37 and 3.16 median log10 reductions) and the original paper reported the subunit with CmeC did not prevent C. jeuni colonization after homologous challenge due to a wide range of C. jejuni loads in the individual vaccinated broilers [25]. Ten trials (from one paper) used STM-1 vectored vaccine expressing various antigens from the inserted plasmids or ST chromosome orally with booster reported the reduction levels between 0.4 log10 CFU/g and 2.2 log10 CFU/g after homologous challenge, estimated from the figures provided in the original paper [24].                 Twenty trials from seven papers reported non-significant (geometric, arithmetic, or median) log10 reduction (CFU/g or GenEq/g) upon homologous/heterologous challenge. The levels of log10 reduction from 15 of 20 trials were reported between 0.1 and 1.6 (mean or median) log10 reduction after homologous challenge. These trials were the crude lysate vaccines with OMP, OMP-NP orally with booster [47], DNA vaccine with purified FlaA cloned into pcDNA3 plasmids mixed with adjuvant subcutaneously or intramuscularly with booster [51], formalin-killed whole-cell vaccine mixed with oil adjuvants [52], subunit vaccine with 40 µg of recombinant NHC flagellin mixed adjuvant in ovo [53], and SE vectored vaccine expressing Omp18 protein (Cj0013), peptidoglycan associated lipoprotein of Salmonella (PAL of Salmonella), and high mobility group box 1 protein (HMGB1) orally [58]. Five trial used a combination of DNA (prime) and subunit (boost) vaccine with four individual antigens and combinations of these antigens, delivered intramuscularly with booster upon heterologous challenge reported reduction levels ranging between 0.12 and 2.2 log10 CFU/g (using qPCR or bacterial culture methods) and between 1.06 and 1.92 log10 reductions in genome equivalents per gram (qPCR) [50].

Prevalence of Colonized Broilers in Vaccine Efficacy
To further evaluate vaccine performance in this review, the trial outputs from eligible studies were used to estimate vaccine efficacy using relative risks to enable comparisons to be made between studies. A wide range of vaccine efficacies in vaccinated broilers was identified, ranging from no effect of C. jejuni colonization to 100.0% prevention (Table 5). Of 45 eligible vaccine trials, three trials reported C. jejuni detection was unculturable (below detection limit) in all vaccinated broilers after homologous challenge (RR < 0.11) and vaccine efficacy was approximately 90%.The three trials were 125 µg of crude cell lysate vaccine with total OMP (subcutaneously with booster), 125 µg of crude cell lysate vaccine with total OMP encapsulated with lactide-co-glycolide nanoparticles (subcutaneously with booster), and 10 8 cells of an avirulent Salmonella enterica χ3987 strain vectored vaccine expressing CjaA (orally with booster). Following these, one trial using a subunit vaccine with recombinant FlaA-LT-B mixed with sodium carbonate reported a significant reduction of the number of colonized vaccinated broilers after heterologous challenge with the prevalence of 27.6% [78] but the efficacy was 44% (Table 5).
In contrast, 28 trials failed to prevent C. jejuni colonization as all vaccinated broilers were positive of C. jejuni with a relative risk of ≥1.00 (Table 5). Based on a comparison of log10 CFU/g reductions, 10 trials report significant log10 reductions between 0.5 and 4.2 log10 CFU/g and 10 trials were non-significant log10 reductions ranging between 0.1 and 2.2 log10. While six and two trials were non-reduction and not reported, respectively (Table 5).

Discussion
The development of efficacious C. jejuni vaccines for poultry is potentially an effective intervention strategy to reduce the risk of campylobacter infections in humans. In this review, our goal was to evaluate the results of published C. jejuni vaccine studies with the view to identifying the most efficacious antigens for further development. The effective outcomes of controlling C. jejuni at farms are commonly evaluated using the reduction of prevalence (proportion) of colonized broilers or the reduction of C. jejuni loads in the intestine. When undertaking this review, it became apparent that the variability of how C. jejuni vaccine studies have been reported prevent direct comparisons of vaccine efficacy from being made. Most studies report the outcome of vaccination as either antigen-specific immune responses and/or reductions of the C. jejuni loads in the intestines [22,25,45,[47][48][49][50][51]54,55,[79][80][81][82]. While many studies report significant reductions in C. jejuni loads, the actual reductions are highly variable. Consequently, it is difficult to estimate the potential impacts of these studies on the risk of C. jejuni transmission to humans.
Based on this review, 62 trials from 16 papers fulfilled the selection criteria and were included using C. jejuni loads in vaccinated broilers [24][25][26][46][47][48][49][50][51][52][53][54][55][56][57][58]. The variations of C. jejuni log10 reduction within these studies were estimated from different C. jejuni loads (log10) in ceca between vaccinated and non-vaccinated chickens [49,50,80,82]. High variations of significant log10 reductions of C. jejuni loads in the intestines of vaccinated broilers were reported between 0.5 and 6.0 log10 reductions among the studies using different variables for statistical comparisons (i.e., geometric mean, arithmetic mean, or median) [25,26,49,50,[54][55][56]. Highly variable data of C. jejuni loads in the individual vaccinated broilers were reported in some trials of the original papers was identified in this review [25,50]. In some cases, where levels of log10 reductions were identified, the outcomes of vaccine efficacy were reported as non-significant reduction or no decrease in C. jejuni colonization [25,50]. While other studies reported levels of CFU/g reduction of C. jejuni colonization between~0.5 and~1.9 were significant [49,54,55], other studies reported similar reductions (<1.9 log10 CFU/g) as non-significant [25,50,58]. These suggest that the statistical power of some studies was insufficient to discriminate between treatment groups where the log10 reductions of colonization were modest. Nauta et al. [10] estimated that a one or two log10 reduction of Campylobacter loads in cecal contents of broilers at slaughterhouses could potentially reduce the risk of transmission to humans by at least 44%. Therefore, more studies are needed to define the vaccine trial parameters required to enable the robust measurement of log10 reductions and how these reductions impact on the risk of human transmission. Defining these parameters is important as assessing the efficacy of C. jejuni vaccines as it is likely to remain reliant on challenge studies. Several studies have reported poor correlations between immune responses and reductions in the C. jejuni loads in the intestines of chickens in vaccination/challenge studies [22,25,45,47,50,51,54,79,81,82].
As a result of these factors, the quantitative risk assessment model reported by Rosenquist et al. [34] was adopted for this review. The model predicts that a 30-fold reduction in the broiler flock prevalence of C. jejuni would result in a 2-log10 reduction of carcass contamination. The outcome of reducing carcass contamination by this amount could result in a 30-fold decrease in the incidence of human campylobacteriosis. Similarly, EFSA [9] using a model for C. jejuni prevalence targets to analyse the quantitative microbiological risk assessment estimated that setting targets of 25% and 5% between broiler flock prevalence would reduce to 50% and 90% of the public health risk, respectively. Thus, these models enable the critical evaluation of published vaccine efficacy studies in the context of public health outcomes. Consequently, in this review, we used the proportionate number (prevalence) of C. jejuni positive/negative broiler chickens between vaccinated and unvaccinated after challenge to evaluate the included studies as another definition of vaccine efficacy. The prevalence of colonized broilers was taken from the text/tabulations reported and/or estimated from figures provided in the original papers. Consequently, a total of 45 trials from 13 papers fulfilled the inclusion criteria [25,26,[46][47][48]50,51,[53][54][55][56][57]78]. This highlights the need for future studies to consider the models of Rosenquist et al. [34] and EFSA [9] to determine the impact of reducing C. jejuni loads in ceca of chickens on the risk of carcass contamination. When considering the vaccine efficacy based upon prevalence, OMP, OMP-NP, and CjaA antigens from three different vaccine trials (crude cell lysate and avirulent ST χ3987 strain vectored vaccines) were demonstrated to clear C. jejuni colonization in the vaccinated broilers with RR < 0.11 and vaccine efficacy greater than 90%, compared with the control groups. These outcomes were comparable to significant levels between 5.7 and 6.0 log10 reductions reported. Following this, a subunit vaccine with 1 mg of recombinant FlaA-LT-B mixed with sodium carbonate reported significant reductions in the number of colonized broiler with prevalence of 27.59% [78], but the RR was 0.56 with the vaccine efficacy of 44%. Thus, based on the data reviewed using both definitions of vaccine efficacy, significant reduction levels more than 5.7 log10 reductions could provide the vaccine efficacy more than 90%.
One of the potential challenges for using vaccination to control C. jejuni colonization is the lifespan of commercial broilers. The current review identified that the many eligible vaccine efficacy studies used broilers with a wide age range, ranging from 24 to 46 days by the end of the study (Tables 3 and 5). Commercial broiler chickens are commonly slaughtered between 35 and 86 days of age, depending on the target market weight and the type of farming system [83,84]. It has been reported that chicken B cell populations do not fully mature until 42 days of age, which may also affect vaccine efficacy [85]. Chicken age is of further importance to vaccine efficacy with respect to timing of C. jejuni colonization. Recent studies have reported that commercial broilers were colonized by C. jejuni and/or C. coli by 10 days of age [86,87], suggesting that vaccination of chicks would be of benefit to the poultry industry. However, maternal antibodies can interfere with vaccine efficacy when live vectored vaccines are applied in young chicks [88]. To overcome this issue, a subunit vaccine or a vectored vaccine with various routes of immunization (i.e., intranasal or in ovo) that are not neutralized by maternal antibodies would be worthwhile exploring [89][90][91]. Thus, the ideal C. jejuni vaccine will need to confer rapid immune responses to antigens associated with preventing colonization and provide protection to chickens from early in the production cycle through to slaughter.
Based upon the inclusion/exclusion criteria of this review using the prevalence of colonized broilers, a meta-analysis could not be performed due to highly variable data. Thus, it is recommended that future studies reporting C. jejuni efficacy studies are supported by datasets that include, the numbers of colonized/non-colonized broiler chickens in treatment groups. Where the outcomes of trials are reported as a degree of colonization (e.g., CFU/g of fecal matter) individual chicken data should be reported to enable future meta-analyses of vaccine studies.
The compiled dataset of published C. jejuni poultry vaccine studies reviewed here has highlighted the highly variable nature of how these prototype vaccines have been evaluated and reported. However, it is clear from the results of these vaccine studies, some of these could potentially lead to a commercial vaccine in the future. Thus, it is recommended that a standardized evaluation model and reporting system be developed for C. jejuni vaccination studies. The standardized evaluation model would need to include, bird type (e.g., broiler and layer), age of bird, type of vaccine, antigen (source and dose), type of adjuvant where applicable, route of vaccination, method of challenge, time to challenge, and challenge dose(s) being the minimal reporting requirements. In terms of evaluating efficacy, while various outcomes would be acceptable, such as protected/not protected or reductions in colonization loads, based on bacterial culture and/or molecular (i.e., quantitative PCR or mass spectrometry) detection, it is crucial that individual bird data should be made readily available. Standardization, particularly of efficacy trial outcome reporting, would enable a more robust evaluation of putative antigens and their formulations between studies.

Conclusions
Of the C. jejuni antigens evaluated in this study, it was concluded that the OMP (125 µg) formulated with and without PLGA-NP delivered subcutaneously and the oral vaccination with subunit vaccine with recombinant FlaA-LT-B mixed with sodium carbonate were the most efficacious candidate vaccines to reduce C. jejuni colonization of broilers identified to date. Further evaluation of this "antigen complex" is clearly warranted, perhaps using OMP preparations from gene deletion mutants to identify which components are contributing to the protection, using the proposed evaluation model described above. Overall, the data assessed in this review supports the conclusion that the development of a C. jejuni vaccine to prevent the colonization of poultry is feasible. Such a vaccine would be crucial in helping the global poultry industry minimize risks to the consumers of their products.