Sustainable control of antimicrobial resistance (AMR) requires a One Health approach, because humans and animals can exchange bacteria and genetic mobile elements encoding for AMR. Responsible use of antimicrobials in animals is essential, but AMR genes may also spread in the absence of selection pressure. Thus, AMR surveillance has been implemented to evaluate AMR trends in animal populations. Current surveillance programs are, however, not suitable for early detection of newly emerging AMR. Early detection is important in case of resistance to antimicrobials of critical importance to humans. When only a few farms are positive at the time of first detection, measures like quarantine or culling could be implemented to minimize exposure of humans. In case many farms are affected at first detection, implementation of such measures is not feasible. Methicilin-resistant Staphylococcus aureus (MRSA), Vancomycin-resistant enterococci (VRE) and Extended Spectrum Beta-Lactamase Enterobacteriaceae (ESBL) are all examples of AMR’s with a large impact on human health that emerged in animal populations in the Netherlands and many other countries and that were already widespread at the time of detection.
The goal of BEWARE is to develop a blueprint for detecting emerging AMR in livestock when the number of affected farms is still low. This early warning blueprint is unique, because it takes account of the most likely routes of introduction, the transmission of AMR between animals and between farms and the diagnostic quality of the assay used. Although the blueprint will be applicable to all AMR’s and animal sectors, BEWARE will focus on carbapenemase producing Enterobacteriaceae (CPE) in veal calves, pigs and broilers. These are the most relevant livestock sectors regarding AMR and CPE emerging in animals is of critical importance in human health care. Specific goals are: 1) identify AMR introduction routes, their association with human behaviour, and rank them according to their importance; 2) calculate transmission parameters for AMR-carrying bacteria within and between animal populations; 3) develop a sensitive and specific assay for early detection of CP-genes; and 4) generate a dynamic mathematical model to integrate the results to predict the numbers of affected farms at first detection as function of the selected sampling strategy.
Developing such an early warning blueprint requires an interdisciplinary project subdivided in 4 work packages (WP).
In WP1 AMR introduction routes will be identified, including their association with human behaviour, and combined into a risk model of AMR introduction into the livestock populations that allows to rank the sites of possible introduction according to their relevance. The model will include both national and international routes of introduction and information will be collected from literature, available databases and by expert elicitation.
Transmission of AMR carrying bacteria within and between populations determines the speed of AMR emergence. In WP2 parameters for AMR transmission in veal calf, broiler and pig farms will be estimated from available data sets, assuming transmission of emerging CPE carrying bacteria is like transmission of ESBL carrying Enterobacteriaceae. We will test this assumption in vivo in broilers. The network of animal movements between farms will be created from data available to the research consortium.
In WP3 an assay for sensitive and specific metagenomics detection of CPE will be developed. Following enrichment, DNA will be isolated and used for RT-PCR and for targeted metagenomics. Furthermore, we will perform a second enrichment step using the ResCap platform and samples will be sequenced using Illumina and Nanopore sequencing. Finally, test quality parameters will be estimated.
In WP4 an early detection surveillance framework will be developed using a dynamic mathematical model. The framework will consist of three parts that integrate the results of WP 1-3: introduction of AMR genes/plasmids in the livestock population, transmission between animals and between farms, and detection. Using the fully parameterized framework, the sensitivity and specificity of early warning is evaluated. By varying the sampling strategy (i.e. frequency and sample size) the effects on the performance of the early detection system are studied. The blueprint can be easily adapted to other types of emerging AMR, provided their transmission and test quality parameters are available. The flexibility of the model regarding selected sampling strategy will enable policy makers to develop an AMR early warning program according their needs.