Les aminosides sont majoritairement utilisés pour le traitement d’infections causées par des bactéries Gram-négatif. En raison de leur index thérapeutique étroit, les aminosides doivent être administrés avec des doses adéquates afin d’optimiser la guérison clinique tout en minimisant les risques de toxicité. De plus, le suivi thérapeutique posologique est d’autant plus important pour les populations spéciales. En effet, ces dernières, telles que les patients aux soins intensifs, peuvent présenter des conditions physiopathologiques pouvant influencer la pharmacocinétique des aminosides. Ce projet, séparé en trois volets, a permis la description des habitudes de posologies et de suivi thérapeutique des aminosides auprès des patients aux soins intensifs du Québec à l’aide d’un questionnaire. De plus, ce projet inclut également une revue des modèles pharmacocinétiques par approche populationnelle (PopPK) des aminosides pour des patients aux soins intensifs. Finalement, ce projet consiste en l’évaluation de la performance prédictive des modèles de gentamicine avec une base de données-patients provenant de deux établissements de santé du Québec. Le volet 1, sous forme d’un questionnaire, a obtenu un taux de réponse de 64.7%, représentant 42% des lits aux soins intensifs de la province. Les régimes posologiques administrés de façon uniquotidienne, sont plus utilisés que l’administration multiquotidienne avec des doses allant de 5 à 7 mg/kg pour la gentamicine et la tobramycine. L’amikacine est très peu utilisé dans les établissements du Québec. Les cibles thérapeutiques respectaient généralement les cibles recommandées dans la littérature. Le volet 2 a permis la description de six, onze et cinq modèles PopPK d’amikacine, de gentamicine et de tobramycine respectivement. Les modèles à deux compartiments décriraient mieux la pharmacocinétique de l’amikacine et de la tobramcyine, tandis que les modèles à un compartiment décriraient mieux la pharmacocinétique de la gentamicine. Les covariables les plus souvent considérées comme significatives étaient la clairance rénale et le poids corporel. Dans le volet 3, malgré qu’une performance prédictive adéquate a été déterminée auprès des 4 modèles évaluées avec la base de données-patients du Québec, de la variabilité demeure présente concernant la prédiction des concentrations et l’application de ces modèles dans un contexte doit ainsi se faire avec prudence. À partir du meilleur modèle, des régimes posologiques a priori ont pu être simulés. / Aminoglycosides are mostly used for treatment of severe Gram-negative infections. Due to their narrow therapeutic index, aminoglycosides should be administered following adequate dosing regimens in order to optimize clinical efficacy while minimizing the risks of toxicity. Moreover, therapeutic drug monitoring is even more important for frail populations such as the critically ill patients. In fact, the latter often present pathophysiological changes that may influence aminoglycosides’ pharmacokinetics. This project was divided in three parts. Firstly, a survey was developed to describe the usual dosing and monitoring practices of aminoglycosides in critically ill patients in the province of Quebec. This project also includes a literature review of aminoglycosides population pharmacokinetic (PopPK) models in critically ill patients. Finally, this project also consists of evaluating the predictive performance of gentamicin PopPK models with a validation dataset composed of patients from two Quebec institutions. The survey had a response rate of 64.7%, therefore representing 42% of all intensive care unit beds in the province. Once-daily-dose regimens are more used than multiple-daily-dose regimens. Most common gentamicin and tobramycin administered dose regimens ranged from 5 to 7 mg/kg. Amikacin is rarely used in Quebec’s institutions. Therapeutic targets were generally in-line with findings from the literature. The literature review described six, eleven and five amikacin, gentamicin and tobramycin PopPK models, respectively. Amikacin and tobramycin pharmacokinetics were mostly described by bi-compartment models whereas gentamicin pharmacokinetics were mostly described by single-compartment model. Most common covariates used were renal clearance and bodyweight. In the third part of this project, although an adequate predictive performance was determined in all four evaluated models, variability in the predicted concentrations by the model still remains. Therefore, usage of these models in clinical settings should be done cautiously. Based on the best performing model, a priori dosing regimens were simulated.

Nonlinear Mixed effect models are often used to describe population pharmacokinetics (PK) and Pharmacodynamics (PD) and play an important part of drug development both from regulatory and industry point of view. However, they can be time consuming and computationally expensive to develop. This thesis is a part of a larger collaboration between Uppsala University and two pharmaceutical companies, with the aim to develop a suite of software that can automate the model building process with more efficiency. One aspect that is important during the model building process is to detect how much the population parameter estimates are influenced by particular individuals. The results of this might lead to reconsideration of the model structure, as well as exclusion of these individuals from the dataset. The current tools available to detect this use case deletion diagnostics (CDD) to run the model multiple times with each subject removed from the dataset to examine whether the population estimates alter when that individual is removed. Another important aspect is whether an individual is an outlier from the population parameter predictions, which is obtained from simulating the model and evaluating the residuals (simeval). Both of these tools are computationally expensive and can take a lot of time, in particular CDD. Therefore, we developed a tool using machine learning (ML) algorithms that can predict these individuals based on other criteria, which will decrease the runtime in an automated model building procedure, whilst maintaining the robustness of the current methods described above.  To create a training database for the ML models, predictors were extracted from 27 previously published models and the CDD and simeval diagnostic tools were run on these models to obtain that true values we want the ML model to predict. The database was then used to train two artificial neural networks (ANN) which is an efficient and powerful method in ML. To enable ‘on-the-fly’ predictions, the developed ANN models were deployed using tflite into pharmpy. The resulted ANNs were able to predict outlying individuals with 79% sensitivity, 83% precision, and 99.1% specificity. While the influential individuals ANN was able to predict with 58% sensitivity, 63% precision and 99.6% specificity. Both ANNs offered a rapid assessment of influential individuals and outlying individuals and were able to make predictions in a matter of sub-seconds compared to hours using traditional methods.

Tuberculosis is one the leading causes of death globally and was before the COVID-19 pandemic the leading cause of death from a single infectious agent. Developing active tuberculosis is life threatening and therefore is the rise of drug-resistant tuberculosis alarming as this risk causing current treatments to become ineffective. Linezolid is a promising drug for treatment of drug-resistant pulmonary tuberculosis, but the effect of linezolid treatment for pulmonary tuberculosis subjects is still not understood well enough and the World Health Organization has requested this knowledge gap to be filled. In this project we support the closing of this knowledge gap by describing the pharmacokinetics of linezolid for treatment of pulmonary tuberculosis using data collected from a phase two clinical trial in a South African population. This was done by creating a pop-PK model and resulted in the PK of linezolid in pulmonary tuberculosis patients from South Africa was best described using a one-compartment model, with first-order absorption process preceded by a series of transit compartments and saturable elimination. However, the diagnostics of the model still show that there are room for improvements and future work is necessary to further optimize the model.

Contexte: L’insuffisance cardiaque (IC) est un syndrome clinique complexe regroupant un large spectre de mécanismes pathologiques qui peuvent altérer le fonctionnement de multiples organes, affectant ainsi la pharmacocinétique (PK) des médicaments. La modélisation pharmacocinétique de population (Pop-PK) consiste à appliquer des modèles non linéaires à effets mixtes dans le but de décrire l’exposition au traitement et quantifier la variabilité au niveau des paramètres PK. Objectif: Ce travail vise à évaluer par approche populationnelle la PK du candesartan en IC et à déterminer les covariables décrivant d’une façon statistiquement et cliniquement significative la variabilité au niveau de la clairance. Méthodes: Les données d’une étude pharmacogénomique ouverte, multicentrique et prospective ont été récupérées pour amorcer notre analyse. Le processus de modélisation et les simulations nécessaires sont réalisés à l’aide du logiciel NONMEM (Nonlinear Mixed Effects Modeling). Les covariables préliminaires ont été sélectionnées par des tests statistiques tels que la régression linéaire et l’ANOVA. Enfin, l’élaboration du modèle final est effectuée en utilisant le processus de sélection séquentielle « forward/backward ». Résultats: Un total de 281 patients caucasiens ont été inclus pour développer le modèle Pop-PK. Les données du candesartan ont été caractérisées par un modèle à un compartiment avec absorption de premier ordre et temps de latence. Le poids, l'âge, la fraction N-terminale du pro-peptide natriurétique de type b (NT_proBNP), le débit de filtration glomérulaire (DFG), le diabète, l'utilisation du furosémide et le sexe étaient les covariables sélectionnées préliminairement pour la clairance apparente (CL/F). Le modèle final développé pour la clairance apparente est représenté par l'équation suivante : CL/F (L/h) = 8.63*(Poids/82.45)0.963 * (DFG/74)0.56 * (0.682) Diabète * EXP0.138 Les simulations ont révélé qu'une diminution importante de la clairance orale (diminution de plus que 25 %) est obtenue en combinant les facteurs significatifs retenus dans le modèle final (patients ayant un faible poids corporel avec une insuffisance rénale modérée à sévère et patients diabétiques avec une insuffisance rénale faible à modérée). Nous avons constaté que les patients ayant ces combinaisons dans notre base de données présentaient des concentrations comparables à celles des autres patients malgré qu’ils aient toléré de plus faibles doses pendant la titration. Conclusion: La modélisation PK de population a servi comme une approche efficace pour caractériser la PK du candesartan en IC et pour identifier une sous-population à risque d’une exposition élevée. Le poids, le DFG et le diabète sont des prédicteurs indépendants de la clairance du candesartan en IC. Considérant ces facteurs, une approche plus individualisée de l'administration du candesartan est nécessaire chez les patients atteints d’IC. / Context: Heart failure (HF) is a clinical condition that causes pathological changes all over the body affecting hence the pharmacokinetic of drugs. Population pharmacokinetic modeling (Pop-PK) consists in applying non-linear mixed-effects models to characterize treatment exposure and quantify PK parameters variability. Objective: The aim of this study was to investigate the pharmacokinetic (PK) of candesartan in HF patients while examining statistically and clinically significant covariates on estimated clearance using population pharmacokinetics (Pop-PK) modeling approach. Methods: Data from a prospective, multicenter, open label, pharmacogenomic study were available for this analysis. Modeling and simulations were conducted using Nonlinear Mixed-Effect Modeling software NONMEM. Preliminary selection of covariates was accomplished with statistical tests (linear regression and ANOVA). Final model development was performed using forward/backward selection approach on the preliminarily selected covariates. Results: A total of 281 Caucasian patients were included to develop the Pop-PK model. Candesartan data were characterized by a 1 compartment model with first order absorption and lag time. Weight, age, N-terminal pro b-type natriuretic peptide (NT_proBNP), estimated glomerular filtration rate (eGFR), diabetes, use of furosemide and sex were the preliminarily selected covariates for apparent clearance (CL/F). The final model developed for apparent clearance is represented by the following equation: CL/F (L/h) = 8.63*(Weight/82.45)0.963 * (eGFR/74)0.56 * (0.682) Diabetes * EXP0.138 Simulations revealed that an important decrease in oral clearance (decrease of more than 25%) is obtained with the combination of the significant factors retained in the final model (patients having low weight with moderately to severely impaired renal function and diabetic with mildly to moderately impaired renal function). Patients having these combinations in our database were found to achieve comparable concentrations to the rest of patients despite tolerating only lower doses. Conclusion: Population pharmacokinetic modeling provided an effective approach to characterize the PK of candesartan in HF and to identify a subpopulation at potential risk of high exposure. Weight, eGFR and diabetes are independent predictors of candesartan clearance in patients with HF. Considering these factors, a more individualized approach of candesartan dosing is needed in HF patients.

No description available.

Pharmacometrics is widely used in drug development. Models are developed to describe pharmacological measurements with data gathered from a clinical trial. The information can then be applied to, for instance, safely establish dose-response relationships of a substance. Glycated hemoglobin (HbA1c) is a common biomarker used by models within antihyperglycemic drug development, as it reflects the average plasma glucose level over the previous 8-12 weeks. There are five different nonlinear mixed-effects models that describes HbA1c-formation. They use different biomarkers such as mean plasma glucose (MPG), fasting plasma glucose (FPG), fasting plasma insulin (FPI) or a combination of those. The aim of this study was to compare their performances on a population and an individual level using model averaging (MA) and to explore if reduced trial durations and different treatment could affect the outcome. Multiple weighting methods were applied to the MA workflow, such as the Akaike information criterion (AIC), cross-validation (CV) and a bootstrap model averaging method. Results show that in general, models that use MPG to describe HbA1c-formation on a population level could potentially outperform models using other biomarkers, however, models have shown similar performance on individual level. Further studies on the relationship between biomarkers and model performances must be conducted, since it could potentially lay the ground for better individual HbA1c-predictions. It can then be applied in antihyperglycemic drug development and to possibly reduce sample sizes in a clinical trial. With this project, we have illustrated how to perform MA on the aforementioned models, using different biomarkers as well as the difference between model weights on a population and individual level.