Search

1. General introduction: In the agricultural sector, antimicrobial drugs are widely used as curative or metaphylactic treatment in food producing animals. One side effect of these usages concerns the development of resistance among the commensal bacteria of animals, mainly within guts, and their potential spread between animals. Then, the resistant bacteria could spread in all the food chain, from the animal towards the food and finally towards the consumers intestines. The risk could be worrying when looking at the plasmid-mediated resistance, like Extended-spectrum β-lactamases (ESBL) for beta-lactamin/cephalosporin or mcr for colistin, due to the possibility that a resistant bacteria originating from animals could transmit its plasmid to human commensal bacteria or pathogen bacteria that are present within guts (Madec, et al., 2017; Nordmann, Poirel, 2016). The link between antimicrobial drug (AMD) use and the development of antimicrobial resistance (AMR) in bacterial strains has been mainly studied in vitro experiments and sometimes modelled thanks to pharmacokinetics-pharmacodynamics (PKPD) approaches. Many within‐host models of antibiotic resistance are based on this pharmacokinetics/pharmacodynamics (PK/PD) approach as opposite to the between-host models, which are mainly epidemiological models. Within‐host models are useful for understanding bacteria dynamics and response to an AMD treatment, within a single individual (Tetteh, et al., 2020). PKPD models are a good basis to explore mechanistically the relationship between drug concentrations and the development/selection of resistant bacteria and may be extrapolated to in vivo situations (Nielsen, Friberg, 2013). However, there are only few PKPD models that describe the impact of AMD within the gut for several reasons: (i) these mechanisms occur in a complex environment (intestinal microbiota) with various interactions and external factors (ii) we face a lack of observed data to inform the PKPD model (need of longitudinal data); (iii) this could be time and computer-intensive. Hence, the goal of this work was to refine and develop a “within-host PKPD model” (hereafter simplified as PKPD model) to assess and predict the impact of an AMD treatment (as an input of the model) on the emergence/selection of resistant bacteria within guts and excretion towards faeces (output of the model) for pig, at the individual level and population level (taking into account the inter-individual variability).

International audience

International audience

Picosecond optoacoustics is a technique that can probe properties of matter on a submicronscale. The aim of this work is to synthesize shear waves at GHz frequencies in a thin isotropiclayer in order to analyse its shear properties. Transverse waves are synthesized using diffraction.In the first part we develop theoretical tools in order to analyse the spatial repartition of theacoustic field generated by the laser-matter interaction. Directivity patterns predict the particulardirections for which amplitude of the shear waves is higher. Thus, we synthesized shearwaves in these particular directions using a dedicated post processing method. To investigateshear properties of a film lying on the sample, we study the reection of shear waves betweenthe transducer and the lying film. Starting with either simulated or experimental recordedwaveforms, we demonstrate the method in the specific case of the titanium/glycerol interface., L'optoacoustique picoseconde est une technique qui permet de sonder la matière à l'échelle submicronique. L'objectif de ce travail est la synthèse d'ondes de cisaillement de fréquences GHz dans des couches minces isotropes pour l'analyse des propriétés transverses de la matière. C'est grâce à la diffraction du champ acoustique à l'intérieur du transducteur que ces ondes de cisaillement sont synthétisées. Nous exposons dans une première partie le développement des outils théoriques nécessaires à l'analyse de la répartition spatiale du champ acoustique généré lors de l'interaction laser-matière. Les diagrammes de directivité ainsi établis permettent de prévoir les directions de rayonnement privilégiées. Par une méthode de post traitement adaptée,nous synthétisons ensuite des ondes transverses dans ces directions particulières. Lorsque le transducteur est chargé par un milieu dont on cherche à connaître les propriétés transverses, les ondes de cisaillement synthétisées sont réfléchies à l'interface entre le transducteur et ce milieu. L'analyse du coefficient de réflexion permet alors de déterminer les propriétés mécaniques du milieu à sonder. Nous illustrons cette méthode numériquement et expérimentalement dans le cas d'une interface titane/glycérol.

Colistin is an old antibiotic, used in both human and veterinary medicine. However, the antibiotic arsenal being more and more limited, colistin appears as one of the last ramparts in the fight against multidrug-resistant bacteria in humans. In order to preserve the efficacy of colistin, two issues were addressed in this thesis:(i) the risks of selecting colistin resistance related to the discovery in 2015 of a gene carried by a plasmid (mcr-1). Thus, the impact of the use of colistin by the oral route in pig production was evaluated in vivo and an absence of selection was observed in our experimental conditions. Similarly, the (minority) use of colistin in human medicine as a prophylaxis for selective digestive decontamination (SDD) was studied in rats harboring a human intestinal microbiota. Preliminary results also show no selection effect.(ii) the development of a physiology-based pharmacokinetic model (PBPK) in pigs for the systemic use of colistin and its prodrug, colistin methanesulfonate (CMS). This model allowed to explore the tissue distribution of CMS and colistin, especially in the kidneys where toxicity is most frequent. As an application of this model, the estimation of waiting times before slaughter when using CMS in pigs was carried out. Finally, the ability of PBPK models to perform intra and interspecies extrapolations was used to adapt this model in adults and children, in order to be able to predict plasma concentrations of colistin during treatment., La colistine est un vieil antibiotique, utilisé à la fois en médecine humaine et vétérinaire. Cependant, l’arsenal antibiotique étant de plus en plus limité, la colistine apparait comme un des derniers remparts dans la lutte contre les bactéries multi-résistantes chez l’Homme. Afin de préserver l’efficacité de la colistine, deux problématiques ont été abordées dans cette thèse :(i) les risques de sélection de résistance à la colistine en lien avec la découverte fin 2015 d’un gène porté par un plasmide (mcr-1). Ainsi, l’impact de l’usage de colistine par voie orale en production porcine a été évalué in vivo et une absence de sélection a été observée dans nos conditions expérimentales. De façon similaire, l’usage (minoritaire) de colistine en médecine humaine comme prophylaxie de décontamination digestive sélective (SDD) a été étudié chez des rats hébergeant un microbiote intestinal humain. Les résultats préliminaires ne montrent pas non plus d’effet de sélection.(ii) le développement d’un modèle pharmacocinétique basé sur la physiologie (PBPK) chez le porc pour l’usage par voie systémique de la colistine et de sa prodrogue, le colistine méthanesulfonate (CMS). Ce modèle a permis d’explorer la distribution tissulaire du CMS et de la colistine, notamment au niveau rénal où la toxicité est la plus fréquente. Comme application de ce modèle, l’estimation des temps d’attente avant abattage lors d’usage de CMS chez le porc a été effectué. Enfin, la capacité des modèles PBPK à réaliser des extrapolations intra et inter-espèces a été utilisé pour adapter ce modèle chez l’adulte et l’enfant, afin de pouvoir prédire les concentrations plasmatiques de colistine lors d’un traitement.

The distribution of metronidazole in the central nervous system has only been described based on cerebrospinal fluid data. However, extracellular fluid (ECF) concentrations may better predict its antimicrobial effect and/or side effects. We sought to explore by microdialysis brain ECF metronidazole distribution in patients with acute brain injury. Four brain-injured patients monitored by cerebral microdialysis received 500 mg of metronidazole over 0.5 h every 8 h. Brain dialysates and blood samples were collected at steady state over 8 h. Probe recoveries were evaluated by in vivo retrodialysis in each patient for metronidazole. Metronidazole and OH-metronidazole were assayed by high-pressure liquid chromatography, and a noncompartmental pharmacokinetic analysis was performed. Probe recovery was equal to 78.8% ± 1.3% for metronidazole in patients. Unbound brain metronidazole concentration-time curves were delayed compared to unbound plasma concentration-time curves but with a mean metronidazole unbound brain/plasma AUC 0–τ ratio equal to 102% ± 19% (ranging from 87 to 124%). The unbound plasma concentration-time profiles for OH-metronidazole were flat, with mean average steady-state concentrations equal to 4.0 ± 0.7 μg ml −1 . This microdialysis study describes the steady-state brain distribution of metronidazole in patients and confirms its extensive distribution.

The transmission of antimicrobial resistance (AMR) between animals, their environment, food and humans is a complex issue. Previous pharmacokinetic-pharmacodynamic (PKPD) models indicate that extended-spectrum β-lactamase (ESBL) resistant bacterial populations may be self-sustaining through horizontal and vertical gene transfer, even in the absence of antimicrobial pressure. However, models focusing purely on the biochemical aspects fail to incorporate the complicated host population dynamics which occur within a farm environment. Models of disease transmission within commercial farm environments can provide further insight to the on-farm transmission dynamics of AMR between animals and their environment, as well as predict the effect of various on-farm interventions. Here, we present a risk assessment which predicts the likelihood that slaughter-aged pigs would carry resistant bacteria after a single introduction of ESBL E. coli on commercial pig farms. We incorporate outputs from a PKPD model which explores the complex host/gastrointestinal bacteria interplay after antimicrobial treatment; with an on-farm model of bacterial transmission. The risk assessment is designed to be adaptable for the simultaneous transmission of multiple bacteria and resistant strains. We predicted that after introduction onto a pig farm, ESBL E. coli bacteria are likely to persist on the farm for more than a year, leading to a high batch prevalence (39.4% slaughter pigs, 5th and 95th percentiles: 0.0-57.5) and high faecal shedding. A comparison of different farm management types suggested that all-in-all-out housing was a protective measure for both prevalence in slaughter-aged pigs and faecal shedding rates. We applied two main interventions at the farm level, an enhanced cleaning and disinfectant (C&D) protocol and isolation of pigs in sick pens for the duration of their antibiotic treatment. Both interventions were able to reduce the number of pigs shedding more than 2 log10 ESBL E. coli from 18.7% (5th and 95th percentiles: 5.9-30.4) in the baseline scenario, to 7.2% (5th and 95th percentiles: 0.0-21.5) when an enhanced C&D protocol was applied, 0.1% (5th and 95th percentiles: 0.0-0.3) when sick pens were used and 0.1% (5th and 95th percentiles: 0.0-0.3) when a combination of enhanced C&D plus sick pens was used. Both scenarios also reduced the prevalence in batches of pigs going to slaughter. This effect was largest when sick pens were used, where 75% of batches had 0% positive pigs. The results suggest that a single introductory event is sufficient to cause a substantial risk of carriage in slaughter-aged pigs. Further quantitative microbial risk assessments (QMRA) are needed to consider the onwards risk posed to later parts of the food chain.

The transmission of antimicrobial resistance (AMR) between animals, their environment, food and humans is a complex issue. Previous pharmacokinetic-pharmacodynamic (PKPD) models indicate that extended-spectrum β-lactamase (ESBL) resistant bacterial populations may be self-sustaining through horizontal and vertical gene transfer, even in the absence of antimicrobial pressure. However, models focusing purely on the biochemical aspects fail to incorporate the complicated host population dynamics which occur within a farm environment. Models of disease transmission within commercial farm environments can provide further insight to the on-farm transmission dynamics of AMR between animals and their environment, as well as predict the effect of various on-farm interventions. Here, we present a risk assessment which predicts the likelihood that slaughter-aged pigs would carry resistant bacteria after a single introduction of ESBL E. coli on commercial pig farms. We incorporate outputs from a PKPD model which explores the complex host/gastrointestinal bacteria interplay after antimicrobial treatment; with an on-farm model of bacterial transmission. The risk assessment is designed to be adaptable for the simultaneous transmission of multiple bacteria and resistant strains. We predicted that after introduction onto a pig farm, ESBL E. coli bacteria are likely to persist on the farm for more than a year, leading to a high batch prevalence (39.4% slaughter pigs, 5th and 95th percentiles: 0.0-57.5) and high faecal shedding. A comparison of different farm management types suggested that all-in-all-out housing was a protective measure for both prevalence in slaughter-aged pigs and faecal shedding rates. We applied two main interventions at the farm level, an enhanced cleaning and disinfectant (C&D) protocol and isolation of pigs in sick pens for the duration of their antibiotic treatment. Both interventions were able to reduce the number of pigs shedding more than 2 log10 ESBL E. coli from 18.7% (5th and 95th percentiles: 5.9-30.4) in the baseline scenario, to 7.2% (5th and 95th percentiles: 0.0-21.5) when an enhanced C&D protocol was applied, 0.1% (5th and 95th percentiles: 0.0-0.3) when sick pens were used and 0.1% (5th and 95th percentiles: 0.0-0.3) when a combination of enhanced C&D plus sick pens was used. Both scenarios also reduced the prevalence in batches of pigs going to slaughter. This effect was largest when sick pens were used, where 75% of batches had 0% positive pigs. The results suggest that a single introductory event is sufficient to cause a substantial risk of carriage in slaughter-aged pigs. Further quantitative microbial risk assessments (QMRA) are needed to consider the onwards risk posed to later parts of the food chain.