Dynamics of Susceptibility and Transmission of Campylobacter jejuni in Chickens

Lead Research Organisation: University of Cambridge
Department Name: Veterinary Medicine

Abstract

Campylobacter jejuni is the leading cause of bacterial food poisoning in the UK, the largest source of which is believed to be contaminated poultry products. For this reason, the primary focus of regulatory authorities is directed towards measures that will limit the numbers of campylobacter-positive birds in commercial flocks. Studies that observe the burden of Campylobacter within different flocks provide important insights into the dynamics of transmission at the flock level. In particular it is almost impossible to distinguish whether flocks have lower levels of Campylobacter due to the individuals being more difficult to colonise or due to the rate of transmission between individuals being lower. However, from observational data alone it is difficult to distinguish the mechanisms responsible for any observed differences. Controlled animal challenge experiments can resolve these questions systematically. Traditionally this would require separate controlled experiments to be carried out to assess transmission and susceptibility independently. Recent statistical models - based on mathematical models of transmission and developed by the CIDC - offer a new approach whereby susceptibility and transmission can be simultaneously estimated from a single experiment. In this project we will develop this theory further and promote this approach that should lead to reductions in the number of experimental animals required to be killed to achieve a given scientific objective. Although this research is being developed specifically with regards to the colonisation of chickens with Campylobacter, we anticipate that this theory will be more generally applicable and has the potential to reduce the number of animals used in the investigation of a broad range of host-pathogen systems. Mathematical models are also useful for helping us to understand and utilise surveillance data more effectively and inform our understanding of the biological mechanisms in action with colonised flocks. A unique data set generated by our collaborators at the University of Oxford suggests that birds may naturally begin to clear Campylobacter when they have grown beyond the average slaughter-age of 30-40 days in Broiler production. We set out to use mathematical models to assess this hypothesis and determine the simplest possible set of mechanisms that can explain this fascinating new data set. This data set also provides a snapshot of the diversity of different strains of Camyplobacter and how it changes over time. Understanding the within- and between-host competition of different strains of bacteria will allow the development of new rational interventions to reduce colonisation. We will use this unique resource to begin to develop a framework by which the competition dynamics of different strains can be described, and the design of future studies improved. High-throughput sequencing is rapidly becoming fast and cheap enough that we will be able to study how the population of different strains evolves over time under natural transmission conditions. We propose to begin the development of theory that will facilitate the collection and analysis of such data to improve our understanding of the basic colonisation biology of this important zoonotic pathogen.

Technical Summary

Modelling of the transmission dynamics of C. jejuni in chickens has been limited by the lack of suitable data. Likewise, the role that transmission dynamics play in the empirical assessment of differences in production methods or application of preventative treatments has been neglected. An important example of this is the routine use of groups of co-housed birds in dose-response studies for C. jejuni in chickens. We have developed a framework that combines classical dose-response models with simple transmission models that has the potential to estimate both susceptibility and subsequent rates of transmission of an infectious agent simultaneously within a single experimental design. Building on our previous work we propose to derive optimal experimental designs to achieve this objective using likelihood-based optimal design theory. We will use stochastic individual-based transmission models to produce simulated datasets to assess these designs in the face of incomplete observational data - e.g. due to imperfect sensitivity and specificity of cloacal swabs. We also propose to apply mathematical models to understand the dynamics of colonisation at the flock level. A unique data set quantifying the prevalence and strain diversity of C. jejuni within a single free-range flock suggests that clearance may occur within individual hosts over a long-enough time-scale of observation. Supporting this hypothesis requires a systematic exploration of the competing mechanisms consistent with these observed epidemic dynamics. We will develop a hierarchy of nested, stochastic transmission models and perform model selection using information theoretic methods to select the most parsimonious model consistent with the data. In addition we propose to use this data set to inform the development of next-generation strain competition models for C. jejuni, adapting stochastic meta-population models originally developed for macro-parasite transmission through the faeco-oral route.

Planned Impact

This proposal applies mathematical modelling to the host-colonisation dynamics of an important zoonotic pathogen, Campylobacter, in a natural host population - the Chicken. Through modelling we seek to improve our understanding of this host-commensal relationship and provide novel experimental designs to reduce the number of animals used in research. The BBSRC will be a beneficiary, as this proposal addresses current BBSRC priorities in both research (animal health, which encompasses zoonotic pathogens such as Campylobacter that are commensal with their farmed animal host) and policy (replacement, refinement and reduction (3Rs) in research). The modelling approaches developed in this proposal aim to provide experimental designs to quantify the impact of different production models (e.g. stocking density, free-range housing) on susceptibility and transmission of Campylobacter. Understanding the mechanisms underlying the differences observed between these different production models is essential if new rational interventions are to be developed. The production of such quantitative data will be of interest to the farming industry, the FSA and Defra in terms of forming policy and the adoption of new farming practices. Ultimately we would hope that successful promotion of production models that reduce the prevalence of Campylobacter at slaughter would benefit the general public, through a reduced exposure to contaminated poultry products. Recognition as researcher co-investigators will provide important career development opportunities for Dr Colles and Dr Conlan, allowing them to build on their previous research successes in order to develop their own research agendas through ongoing multi-disciplinary collaboration.
 
Description The transmission of Campylobacter in commercial poultry flocks has been almost exclusively modelled as a direct transmission process between susceptible and infectious birds. However, indirect transmission through birds encountering contaminated faeces in the environment is also known to be an important component of bird-to-bird transmission. We developed models to systematically compare the relative importance of direct and indirect transmission in an attempt to understand the so-called lag-phase of colonisation where commercial flocks are rarely found to be Campylobacter positive during the first 2-weeks of production. In contrast to what is currently believed, we found that the lag phase can potentially be explained without any changes in susceptibility of individual birds provided indirect environmental transmission, rather than direct transmission, is the dominant transmission route.
Exploitation Route Outputs are being prepared for publication in the peer-reviewed literature. Data and source code for models developed will be shared on publication to allow future research to build on these outputs.
Sectors Agriculture

Food and Drink

 
Description Outputs are still being prepared for publication, so it is to soon for impact from findings to be realised.