Search

International audience; Purpose: Cocktail approach using a combination of probes to phenotype several cytochromes P450 or transporters is of high interest in anticipating drug-drug interactions and personalized medicine. Its clinical use remains however limited by the intensive sampling scheme required to obtain phenotyping indexes (PI) which consists in calculating the area under the concentration-time curves. We proposed to use maximum a posteriori Bayesian estimation (MAPBE) that incorporates available information from the whole population to derive PI from a few individual observations. The performance of a limited sampling strategy (LSS) based on MAPBE was evaluated for a five-probe cocktail. Methods: The studied cocktail included midazolam, tolbutamide, caffeine, dextromethorphan, omeprazole and their relevant metabolites. Prior information for MAPBE was obtained by nonlinear mixed-effect modelling of data from a pilot study. Sampling times were chosen based on optimal design theory using the Bayesian Fisher information matrix. Through a simulation study, we investigated the predicted PI in terms of bias and imprecision for varying number and timing of samples. Results: Some three-point Bayesian designs gave mean prediction errors in [-5 %, 5 %], root mean square errors below 30 % for all probes, except dextromethorphan whose model should be consolidated further with additional data. This approach gave overall less outlier predicted values than single-point metrics and was more flexible to the timing of the latest sampling. Conclusions: MAPBE is accurate for predicting simultaneously several PI while being flexible in terms of integrating clinical constraints. Therefore, LSS based on MAPBE could help reduce the time of presence in hospital for individuals to be phenotyped.

International audience; Untargeted metabolomics of human plasma with mass spectrometry is of particular interest in medical research to explore pathophysiology, find disease biomarkers or for the understanding of the response to pharmacotherapy. Since analytical performances may be impacted by the laboratory environment and the acquisition method settings, the objectives of this study were to assess the role of interfering compounds and to propose an acquisition method to maximize the metabolome coverage for human plasma metabolomic analysis. Human plasma samples were processed with liquid/liquid extraction then analysed with HILIC-high resolution mass spectrometry. A method with a single m/z range was compared to four methods with different split acquisition ranges and four sets of ionization source parameters were compared. The data were analysed with the R software and on the Worklow4Metabolomics online platform. The major interfering compounds were identified in blank samples where they accounted for up to 86% of the signal intensity. Splitting the acquisition range into 3 m/z ranges improved the number of detected features, the number of features with proposed annotation in the Human Metabolome Database, as well as signal intensity throughout the whole m/z range. The method performing best was the one using three m/z ranges of approximatively the same extent. Ionization source parameters also strongly affected the number of detected features. Splitting the acquisition range into 3 m/z ranges with optimized ionization source parameters allows a comprehensive analysis of the human plasma metabolome with perspectives for applications to pathophysiological studies.

International audience; Models for the polymerization process involved in prion self-replication are well-established and studied [H. Engler, J. Pruss, and G.F. Webb, Analysis of a model for the dynamics of prions II, J. Math. Anal. Appl. 324 (2006), pp. 98-117; M.L. Greer, L. Pujo-Menjouet, and G.F. Webb, A mathematical analysis of the dynamics of prion proliferation, J. Theoret. Biol. 242 (2006), pp. 598-606; J. Pruss, L. Pujo-Menjouet, G.F. Webb, and R. Zacher, Analysis of a model for the dynamics of prions, Discrete Cont. Dyn. Sys. Ser. B 6(1) (2006), pp. 215-225] in the case where the dynamics coefficients do not depend on the size of polymers. However, several experimental studies indicate that the structure and size of the prion aggregates are determinant for their pathological effect. This motivated the analysis in Calvez et al. [Size distribution dependence of prion aggregates infectivity, Math Biosci. 217 (2009), pp. 88-99] where the authors take into account size-dependent replicative properties of prion aggregates. We first improve a result concerning the dynamics of prion aggregates when a pathological state exists (high production of the normal protein). Then we study the strain phenomena and more specifically we wonder what specific replicative properties are determinant in strain propagation. We propose to interpret it also as a dynamical property of size repartitions.

Regenerative medicine and cell therapy provide great hopes for the use of adult and stem cells. The latter are far less present in tissue than the former and must be expanded using cell culture. Stem cells culture requires the conservation of their proliferation and self-renewal capabilities. Still, the complex interaction between cell populations, for example in primary cell cultures, are not well-known and may account for part of the variability of such cultures. In order to represent and understand the evolution of cultured stem cells, we present here a mathematical model of cell proliferation and differentiation. Based on the formalism of cellular automata, this model simulates the evolution of several cell classes (which may represent either different levels of differentiation or different cell types) in an environment modeling the growth medium. We model the cell cycle as on the one hand a quiescence phase during which a cell rests, and on the other hand a division phase during which the cell starts the division process. In order to represent cell-cell interaction, the transition probability between those phases depends on the local composition of the growth medium depending itself on neighboring cells. An interaction between cellular populations is represented by a quantitative parameter which has a direct impact on cellular proliferation. Differentiation results in a change of the cell class and depends on the biological model studied : it may result from an asymmetric division or be a consequence of the local composition of the growth medium. This mathematical model aims at a better understanding of the interactions between cell populations in a culture. By defining constraints on the potential or the type of the cells at the end of a culture, it will then be possible to find optimal experimental conditions for cell production.