This thesis investigates the supply chain of Vancomycin and Piperacillin/Tazobactam in order to understand why the two antibiotics have been exposed to back orders during recent years in Sweden. The health economical consequences due to these back orders of the two antibiotics was also examined. The used methods were literature search and elementary calculation methods. The supply chains for the two antibiotics consists of multiple manufacturing actors, both primary and secondary. The manufacturing actors are mostly located in low and middle income countries, which increases risks for the supply chains. The Swedish market is unattractive due to its small size and ineffective purchasing system, which also increases risks of shortages. The unattractive market is a probable cause of the lower amount of market authorisation holders which sell the antibiotics in Sweden. Furthermore, a financial model was created to assess the health economic impacts of shortages. The costs were calculated as the sum of the additional labor required to deal with shortages along with the costs of the alternative medicines. It was estimated that a shortage of Vancomycin can cost up to SEK 1 600 000 in fixed costs followed by up to SEK 202 997 per additional day of shortage and that a shortage of Pipetazo can cost up to SEK 1 600 000 in fixed costs followed by up to SEK 923 650 per day. There are also other negative aspects of these consequences, such as worsening of patient health and contributions to increased AMR.

No / Piperacillin, a potent ß-lactam antibiotic, is effective in a large variety of Gram+ and Gram¿ bacterial infections but its administration is limited to the parenteral route as it is not absorbed when given orally. In an attempt to overcome this problem, we have synthesized a novel series of charged liposaccharide complexes of piperacillin comprising a sugar moiety derived from d-glucose conjugated to a lipoamino acid residue with varying side-chain length (cationic entity) and the piperacillin anion. A complete multiple reaction monitoring LC¿MS/MS method was developed to detect and characterize the synthesized complexes. The same method was then successfully applied to assess the in vitro apparent permeability values of the charged liposaccharide complexes in Caco-2 monolayers. / BBSRC

Introduktion Antibiotikaresistens är ett globalt växande problem. Till gruppen β-laktamantibiotika hör piperacillin-tazobaktam och cefotaxim som båda verkar genom att försvaga cellväggen med kovalenta bindningar till peptidoglykanlagret som lyserar cellen. E. coli och K. pneumoniae tillhör gruppen Enterobacteriaceae, som är en del av den humana tarmfloran och ofta förekommande vid urinvägsinfektion och sepsis. Utvidgat Spektrum β-Laktamas (ESBL) är ett enzym som finns hos Enterobacteriaceae och som hydrolyserar β-laktamantibiotika. Matrix assisted laser desorption ionization time of flight mass spectrometry (MALDI-TOF MS) är en kvalitativ analysteknik för detektion av kemiska föreningar i avseende på massa och laddning. Kännedom om antibiotikametaboliters molekylvikt vid hydrolys möjliggör detektion. Liquid Chromatography Tandem Mass Spectrometry (LC-MS/MS) är en högsensitiv kvantifieringsmetod som separerar molekyler i avseende på polaritet för vidare detektion i avseende på massa och laddning. Syfte Syftet med denna studie var att vidareutveckla en snabb och effektiv metod för att påvisa nedbrytning av piperacillin-tazobaktam och cefotaxim i blodplasma med LC-MS/MS. Material och Metod Tiofaldigt sjunkande koncentrationer av piperacillin-tazobaktam från 2000 till 2 µg/ml, och cefotaxim med koncentrationerna 500 till 0,5 µg/ml analyserades med MALDI-TOF MS, dels intakt men även med bakterierna E. coli och K. pneumoniae med uttryck av olika resistensmekanismer. Vid optimerade koncentrationer spikades plasmaprover med nedbrutet antibiotika som sedan kvantifierades med LC-MS/MS. Resultat Lägsta detektionsgräns med MALDI-TOF MS för intakt och hydrolyserat piperacillin-tazobaktam var 20/2,5 µg/ml. För cefotaxim var lägsta gränsen 5 µg/ml. Med kliniskt relevanta blodkoncentrationer gick hydrolys inte att detektera för. Med tre bakteriekolonier/50 µl kunde dock hydrolys detekteras och kvantifieras med LC-MS/MS. Slutsats Detektion av β-laktamantibiotika är möjligt med både MALDI-TOF MS och LC-MS/MS. För att påvisa hydrolys krävdes större mängder bakterier än förväntat med LC-MS/MS. / Introduction Antibiotic resistance is a global growing problem. Piperacillin-tazobactam and cefotaxime are parts of the group β-lactam antibiotics. The common feature is to inhibit the cell wall synthesis by covalent bindings to the peptidoglycan layer and thereby causing lysis of the bacterial cell. E. coli and K. pneumoniae are members of the Enterobacteriaceae which is a part of the human normal flora but also are commonly associated with urinary tract infections which sometimes develops into to sepsis. Extended Spectrum β-Lactamases (ESBLs) are enzymes with hydrolytic abilities acting on β-lactam antibiotics, expressed by Enterobacteriaceae. The qualitative, Matrix assisted laser desorption ionization time of flight mass spectrometry (MALDI-TOF MS) can be used to detect chemical compounds in the ratio of mass to charge in accordance to their molecular weight. Liquid Chromatography Tandem Mass Spectrometry (LC-MS/MS) is a highly sensitive two-step method of quantification which first separate molecules by their polarity attraction force and then by the ratio of mass to charge. Aim The aim of this study was to develop a fast and efficient method to determine degradation of piperacillin-tazobactam and cefotaxime in blood plasma by LC-MS/MS. Method Tenfold dilution of piperacillin-tazobactam in concentrations of 2000 to 2 µg/ml, and cefotaxime in concentrations of 500 to 0,5 µg/ml where analyzed by MALDI-TOF MS, intact and also with the bacteria E. coli and K. pneumoniae with different expression of antibiotic resistance. Optimized concentrations where fixed in blood plasma and then quantified by LC-MS/MS. Result The detection limit by using MALDI TOF MS of hydrolyzed as well as non-hydrolyzed piperacillin-tazobactam was 20/2,5 µg/ml. The detection limit in cefotaxime was 5 µg/ml. Hydrolysis could not be detected in clinically fixed blood concentrations. Detection and quantification of hydrolysis by LC-MS/MS was possible in a concentration of three bacteria colonies/50 µl. Conclusion It is possible to detect hydrolysis in both MALDI TOF MS and LC-MS/MS. A larger amount of bacteria than expected was needed to demonstrate hydrolysis In LC-MS/MS.

In critically ill patients, the pathophysiology of sepsis can affect the interactions between the antibiotic, the bacteria and the patient, leading to potential therapeutic failure and the development of antibiotic resistance. It is well acknowledged that research that optimises antibiotic exposure will assist improvement of outcomes in this patient group. Although beta-lactam antibiotics, such as piperacillin and meropenem, are commonly selected for empiric therapy of sepsis, dosing is unlikely to be optimal. In patients without renal dysfunction, data suggests that disease-induced alterations to pharmacokinetic parameters result in low trough concentrations for significant periods within a dosing interval. Administration of these time-dependent antibiotics by continuous infusion has been suggested to improve the pharmacokinetic-pharmacodynamic profile. Knowledge of concentrations in the extracellular fluid of human tissue, which is the target site of most pathogens, is particularly instructive. Extracellular fluid concentrations can be determined using techniques such as microdialysis. Therefore, the principal aims of this Thesis were to determine the plasma and subcutaneous tissue pharmacokinetics of piperacillin and meropenem administered by bolus dosing and continuous infusion in critically ill patients with sepsis; and to use Monte Carlo simulations to compare the ability of different dosing strategies to achieve pharmacodynamic endpoints. This Thesis also sought to compare the clinical outcomes of bolus dosing and continuous infusion of a beta-lactam antibiotic, ceftriaxone, in a prospective randomised controlled trial and to perform a meta-analysis on clinical outcomes from other similar published studies. Finally, this Thesis aimed to systematically review the published literature to determine any correlation between antibiotic dosing and the development of antibiotic resistance. The results of the pharmacokinetic studies, using piperacillin and meropenem, indicate that beta-lactam distribution into subcutaneous tissue, in critically ill patients with sepsis, is less than that observed in previous studies in healthy volunteers yet superior to studies in patients with septic shock. This supports the notion that the peripheral concentration of drugs may be inversely related to the level of sickness severity. Administration by continuous infusion was found to maintain statistically significantly higher trough beta-lactam concentrations in both plasma and subcutaneous tissue. Further analysis of the plasma data using population pharmacokinetic modeling and Monte Carlo simulations described significant pharmacodynamic advantages for administering meropenem or piperacillin by continuous infusion to organisms with high minimum inhibitory concentrations. Given the documented pharmacodynamic advantages for administering beta-lactams by continuous infusion, a prospective randomized controlled clinical trial was conducted using the beta-lactam antibiotic ceftriaxone. In 57 critically ill patients, we found equivalence between continuous infusion and bolus dosing in the intention-to-treat analysis. However, our a priori analysis criteria, requiring patients receive at least 4-days antibiotic treatment, found significant clinical and bacteriological advantages for administration by continuous infusion. To further investigate any clinical differences between bolus dosing and continuous infusion of beta-lactam antibiotics, we performed a meta-analysis of all published studies. Our analysis of the 13 published prospective randomized controlled trials (846 hospitalised patients) showed equivalence of continuous infusion and bolus dosing. Possible confounders observed within, and between the studies, make interpretation of these results challenging. However, two large retrospective cohorts not included in the meta-analysis, found definitive clinical and bacteriological advantages suggesting further research may be appropriate. The possible relationship between antibiotic dosing, or exposures, on the development of resistance was investigated using a structured review of the published literature. Our analysis of relevant papers found a wealth of data describing increasing levels of resistance with sub-optimal antibiotic dosing, particularly for fluoroquinolone antibiotics, but also for other classes including beta-lactams. These results demonstrate the importance of optimizing antibiotic dosing to decrease the development of antibiotic susceptibility from sub-optimal dosing, particularly for critically ill patients who are likely to have low drug concentrations. The results of this Thesis, suggest that a large, prospective, multi-centre randomised controlled trial in critically ill patients with sepsis, is required to definitively determine the clinical utility of administration of beta-lactam antibiotics by continuous infusion.

Objetivos: Avaliar a adequabilidade do modelo farmacocinético-farmacodinâmico (PK-PD) (NOLTING et al., 1996b) para modelar o efeito bactericida da piperacilina (PIP) em ratos Wistar infectados experimentalmente com Escherichia coli ATCC 25922. Metodologia: Experimentos de Farmacocinética: Determinou-se as concentrações plasmáticas totais e livres teciduais de PIP, através de microdiálise (MD), após administração de 240 mg/kg i.v. bolus a ratos Wistar granulocitopênicos (ciclofosfamida) infectados no músculo esquelético (105 UFC/mL) com E. coli. As amostras de plasma e de MD foram analisadas por CLAE. As sondas de MD foram calibradas por retrodiálise. Experimentos de Farmacodinâmica: Os animais imunodeprimidos e infectados foram tratados com PIP nas doses de 120 ou 240 mg/kg, em intervalos de 4/4, 6/6 e 8/8 horas por 24 h. Em tempos pré-determinados, os animais foram sacrificados (n = 3/tempo), o músculo infectado foi retirado, homogeneizado e o número de UFC/mL foi determinado em placas de ágar-sangue, após diluições sucessivas. Um grupo não tratado foi utilizado como controle. Modelagem PK-PD: A partir dos dados farmacocinéticos e farmacodinâmicos obtidos, avaliou-se efeito de morte bacteriana em função do tempo com o auxílio do programa de regressão não-linear SCIENTIST® v.2.0. Resultados e Discussão: Os parâmetros farmacocinéticos após a administração de PIP (240 mg/kg) foram t½ de 40 ± 8 min; CL de 0,46 ± 0,021 (L/h/kg) e um Vdss de 0,30 ± 0,06 L/kg. O perfil de PIP livre tecidual foi previsto a partir dos parâmetros plasmáticos utilizando ajuste simultâneo dos dados de plasma e tecido e um fator de proporcionalidade de 0,342 ± 0,101. Os parâmetros do modelo PK-PD obtidos foram: EC50 de 1,31 ± 0,27 μg/mL e kmax 1,39 ± 0,20 h-1. Os valores dos parâmetros da modelagem PK-PD obtidos in vivo diferiram dos descritos na literatura para o mesmo antibiótico e bactéria quando simulados in vitro. Conclusões: O modelo Emax-modificado descreveu os perfis de crescimento e morte bacteriana em função do tempo obtidos nas diferentes posologias testadas sendo adequado para modelagem PK-PD da piperacilina nas condições experimentais investigadas. / Purpose: The objective of this study was to model the killing effect of a β-lactam antibiotic, piperacillin (PIP), in neutropenic and E. coli ATCC 25922 infected rats after different dosing regimens using a modified Emax PK-PD model. Methodology: Pharmacokinetic studies: Total plasma and free tissue concentrations of PIP, determined by microdialysis, were investigated after i.v. bolus of 240 mg/kg of the drug to immunecompromised (cyclophosphamide) and E. coli infected (107 CFU) Wistar rats. Microdialysis probes recoveries were determined by retrodialysis. Plasma and tissue samples were analyzed by HPLC. Pharmacodynamic studies: The infected rats were treated with iv bolus PIP 120 mg/kg or 240 mg/kg q8h, q6h, q4h. Three animals were sacrificed at predetermined times up to 24 hours. The infected muscle was removed, homogenized and the number of CFU/mL was determined by plate counting after 24 hours of incubation at 37ºC. A control group without treatment was used. PK-PD modeling: PIP killing effect as a function of time was fitted using the Emax-modified model with the aid of a non-linear regression computer program SCIENTIST® v.2.0. Results and Discussion: The pharmacokinetic parameters determined for PIP 240 mg/kg iv bolus were: t½ of 40 ± 8 min; CL of 0.46 ± 0.021 (L/h/kg) and Vdss of 0.30 ± 0.06 L/kg. Piperacillin free tissue levels were predicted using plasma data ina a simultaneous fitting with a proportionality factor of 0.342 ± 0.101. The parameters derived from PK-PD modeling were: bacterial killing rate (kmax) of 1.39 ± 0.20 h-1 concentration to produce 50% of de maximum effect (EC50) of 1.31 ± 0.27 μg/mL. The PK-PD parameters determined in vivo were different from those reported for the same bacteria and drug in vitro. Conclusions: The Emax model adequately described PIP antibacterial effect in animals for the different dosing regimens investigated.

Objetivos: Avaliar a adequabilidade do modelo farmacocinético-farmacodinâmico (PK-PD) (NOLTING et al., 1996b) para modelar o efeito bactericida da piperacilina (PIP) em ratos Wistar infectados experimentalmente com Escherichia coli ATCC 25922. Metodologia: Experimentos de Farmacocinética: Determinou-se as concentrações plasmáticas totais e livres teciduais de PIP, através de microdiálise (MD), após administração de 240 mg/kg i.v. bolus a ratos Wistar granulocitopênicos (ciclofosfamida) infectados no músculo esquelético (105 UFC/mL) com E. coli. As amostras de plasma e de MD foram analisadas por CLAE. As sondas de MD foram calibradas por retrodiálise. Experimentos de Farmacodinâmica: Os animais imunodeprimidos e infectados foram tratados com PIP nas doses de 120 ou 240 mg/kg, em intervalos de 4/4, 6/6 e 8/8 horas por 24 h. Em tempos pré-determinados, os animais foram sacrificados (n = 3/tempo), o músculo infectado foi retirado, homogeneizado e o número de UFC/mL foi determinado em placas de ágar-sangue, após diluições sucessivas. Um grupo não tratado foi utilizado como controle. Modelagem PK-PD: A partir dos dados farmacocinéticos e farmacodinâmicos obtidos, avaliou-se efeito de morte bacteriana em função do tempo com o auxílio do programa de regressão não-linear SCIENTIST® v.2.0. Resultados e Discussão: Os parâmetros farmacocinéticos após a administração de PIP (240 mg/kg) foram t½ de 40 ± 8 min; CL de 0,46 ± 0,021 (L/h/kg) e um Vdss de 0,30 ± 0,06 L/kg. O perfil de PIP livre tecidual foi previsto a partir dos parâmetros plasmáticos utilizando ajuste simultâneo dos dados de plasma e tecido e um fator de proporcionalidade de 0,342 ± 0,101. Os parâmetros do modelo PK-PD obtidos foram: EC50 de 1,31 ± 0,27 μg/mL e kmax 1,39 ± 0,20 h-1. Os valores dos parâmetros da modelagem PK-PD obtidos in vivo diferiram dos descritos na literatura para o mesmo antibiótico e bactéria quando simulados in vitro. Conclusões: O modelo Emax-modificado descreveu os perfis de crescimento e morte bacteriana em função do tempo obtidos nas diferentes posologias testadas sendo adequado para modelagem PK-PD da piperacilina nas condições experimentais investigadas. / Purpose: The objective of this study was to model the killing effect of a β-lactam antibiotic, piperacillin (PIP), in neutropenic and E. coli ATCC 25922 infected rats after different dosing regimens using a modified Emax PK-PD model. Methodology: Pharmacokinetic studies: Total plasma and free tissue concentrations of PIP, determined by microdialysis, were investigated after i.v. bolus of 240 mg/kg of the drug to immunecompromised (cyclophosphamide) and E. coli infected (107 CFU) Wistar rats. Microdialysis probes recoveries were determined by retrodialysis. Plasma and tissue samples were analyzed by HPLC. Pharmacodynamic studies: The infected rats were treated with iv bolus PIP 120 mg/kg or 240 mg/kg q8h, q6h, q4h. Three animals were sacrificed at predetermined times up to 24 hours. The infected muscle was removed, homogenized and the number of CFU/mL was determined by plate counting after 24 hours of incubation at 37ºC. A control group without treatment was used. PK-PD modeling: PIP killing effect as a function of time was fitted using the Emax-modified model with the aid of a non-linear regression computer program SCIENTIST® v.2.0. Results and Discussion: The pharmacokinetic parameters determined for PIP 240 mg/kg iv bolus were: t½ of 40 ± 8 min; CL of 0.46 ± 0.021 (L/h/kg) and Vdss of 0.30 ± 0.06 L/kg. Piperacillin free tissue levels were predicted using plasma data ina a simultaneous fitting with a proportionality factor of 0.342 ± 0.101. The parameters derived from PK-PD modeling were: bacterial killing rate (kmax) of 1.39 ± 0.20 h-1 concentration to produce 50% of de maximum effect (EC50) of 1.31 ± 0.27 μg/mL. The PK-PD parameters determined in vivo were different from those reported for the same bacteria and drug in vitro. Conclusions: The Emax model adequately described PIP antibacterial effect in animals for the different dosing regimens investigated.

Objetivos: Avaliar a adequabilidade do modelo farmacocinético-farmacodinâmico (PK-PD) (NOLTING et al., 1996b) para modelar o efeito bactericida da piperacilina (PIP) em ratos Wistar infectados experimentalmente com Escherichia coli ATCC 25922. Metodologia: Experimentos de Farmacocinética: Determinou-se as concentrações plasmáticas totais e livres teciduais de PIP, através de microdiálise (MD), após administração de 240 mg/kg i.v. bolus a ratos Wistar granulocitopênicos (ciclofosfamida) infectados no músculo esquelético (105 UFC/mL) com E. coli. As amostras de plasma e de MD foram analisadas por CLAE. As sondas de MD foram calibradas por retrodiálise. Experimentos de Farmacodinâmica: Os animais imunodeprimidos e infectados foram tratados com PIP nas doses de 120 ou 240 mg/kg, em intervalos de 4/4, 6/6 e 8/8 horas por 24 h. Em tempos pré-determinados, os animais foram sacrificados (n = 3/tempo), o músculo infectado foi retirado, homogeneizado e o número de UFC/mL foi determinado em placas de ágar-sangue, após diluições sucessivas. Um grupo não tratado foi utilizado como controle. Modelagem PK-PD: A partir dos dados farmacocinéticos e farmacodinâmicos obtidos, avaliou-se efeito de morte bacteriana em função do tempo com o auxílio do programa de regressão não-linear SCIENTIST® v.2.0. Resultados e Discussão: Os parâmetros farmacocinéticos após a administração de PIP (240 mg/kg) foram t½ de 40 ± 8 min; CL de 0,46 ± 0,021 (L/h/kg) e um Vdss de 0,30 ± 0,06 L/kg. O perfil de PIP livre tecidual foi previsto a partir dos parâmetros plasmáticos utilizando ajuste simultâneo dos dados de plasma e tecido e um fator de proporcionalidade de 0,342 ± 0,101. Os parâmetros do modelo PK-PD obtidos foram: EC50 de 1,31 ± 0,27 μg/mL e kmax 1,39 ± 0,20 h-1. Os valores dos parâmetros da modelagem PK-PD obtidos in vivo diferiram dos descritos na literatura para o mesmo antibiótico e bactéria quando simulados in vitro. Conclusões: O modelo Emax-modificado descreveu os perfis de crescimento e morte bacteriana em função do tempo obtidos nas diferentes posologias testadas sendo adequado para modelagem PK-PD da piperacilina nas condições experimentais investigadas. / Purpose: The objective of this study was to model the killing effect of a β-lactam antibiotic, piperacillin (PIP), in neutropenic and E. coli ATCC 25922 infected rats after different dosing regimens using a modified Emax PK-PD model. Methodology: Pharmacokinetic studies: Total plasma and free tissue concentrations of PIP, determined by microdialysis, were investigated after i.v. bolus of 240 mg/kg of the drug to immunecompromised (cyclophosphamide) and E. coli infected (107 CFU) Wistar rats. Microdialysis probes recoveries were determined by retrodialysis. Plasma and tissue samples were analyzed by HPLC. Pharmacodynamic studies: The infected rats were treated with iv bolus PIP 120 mg/kg or 240 mg/kg q8h, q6h, q4h. Three animals were sacrificed at predetermined times up to 24 hours. The infected muscle was removed, homogenized and the number of CFU/mL was determined by plate counting after 24 hours of incubation at 37ºC. A control group without treatment was used. PK-PD modeling: PIP killing effect as a function of time was fitted using the Emax-modified model with the aid of a non-linear regression computer program SCIENTIST® v.2.0. Results and Discussion: The pharmacokinetic parameters determined for PIP 240 mg/kg iv bolus were: t½ of 40 ± 8 min; CL of 0.46 ± 0.021 (L/h/kg) and Vdss of 0.30 ± 0.06 L/kg. Piperacillin free tissue levels were predicted using plasma data ina a simultaneous fitting with a proportionality factor of 0.342 ± 0.101. The parameters derived from PK-PD modeling were: bacterial killing rate (kmax) of 1.39 ± 0.20 h-1 concentration to produce 50% of de maximum effect (EC50) of 1.31 ± 0.27 μg/mL. The PK-PD parameters determined in vivo were different from those reported for the same bacteria and drug in vitro. Conclusions: The Emax model adequately described PIP antibacterial effect in animals for the different dosing regimens investigated.

Le sepsis sévère est l’une des causes de mortalité les plus fréquentes à travers le monde. L’étiologie la plus fréquente étant des infections causées par des bactéries, le traitement repose sur l’administration rapide d’un traitement antibiotique adapté. Toutefois, la diminution de la sensibilité des bactéries observée au cours des dernières années nous pousse à repenser notre utilisation des antibiotiques. Parmi les options envisageables, on retrouve l’utilisation de nouveaux antibiotiques et l’optimisation des posologies d’antibiotiques couramment utilisés. Dans les deux cas, la modélisation pharmacocinétique est un outil indispensable pour caractériser la pharmacocinétique des agents antimicrobiens et ainsi guider les posologies. Les études pharmacocinétiques comportent toutefois de nombreux défis en pédiatrie. Afin de les contourner, nous avons utilisé des méthodes à risques minimaux pour étudier deux molécules chez les enfants : le linézolide, un nouvel antibiotique de la classe des oxazolidinones, qui a été administré chez des nouveau-nés prématurés, et la pipéracilline-tazobactam, une bêta-lactamine fréquemment utilisée en pédiatrie, qui a été administrée en utilisant une nouvelle posologie sous forme d’infusions prolongées. Premièrement, nous avons effectué une étude pharmacocinétique rétrospective du linézolide aux soins intensifs néonataux du CHU Sainte-Justine. Le linézolide est un antibiotique qui peut être utilisé pour traiter les infections causées par le staphylocoque à coagulase négative chez les nouveau-nés prématurés. Il s’agit d’une pratique relativement nouvelle et un programme de surveillance des concentrations plasmatiques avait été instauré il y a quelques années pour encadrer l’utilisation du linézolide dans cette population. Nous avons utilisé les données de ce programme et construit un modèle pharmacocinétique de population en utilisant des méthodes de modélisation non-linéaire à effets mixtes. Nous avons ainsi pu démontrer que les posologies utilisées chez les 26 nouveau-nés inclus dans notre étude atteignaient la cible préalablement déterminée (aire sous la courbe/concentration minimale inhibitrice [ASC/CMI 0-24] > 80), et qu’elles étaient donc probablement efficaces. De plus, nous avons observé que le linézolide semblait sécuritaire dans cette population. Nous nous sommes ensuite intéressés aux infusions prolongées de pipéracilline-tazobactam en pédiatrie. Déjà bien décrite dans la population adulte, l’utilisation d’infusions prolongées permet d’optimiser l’efficacité des bêta-lactamines puisque cette dernière dépend du temps où les concentrations plasmatiques sont supérieures à la concentration minimale inhibitrice (ƒt > CMI). Comme aucune posologie n’était établie en pédiatrie, nous avons d’abord effectué une étude de simulation où nous avons déterminé les posologies dites « optimales » en utilisant les paramètres pharmacocinétiques décrits en pédiatrie. Nous avons par la suite effectué une étude pharmacocinétique prospective où les posologies préalablement établies ont été administrées à 89 enfants de deux mois à six ans, duquel 79 ont eu des prélèvements sanguins pour déterminer les concentrations plasmatiques. Deux modèles pharmacocinétiques de population distincts (pipéracilline et tazobactam) ont été développés en utilisant la modélisation non-linéaire à effets mixtes. Des simulations ont par la suite été effectuées en utilisant le modèle final de la pipéracilline pour déterminer les posologies optimales selon l’âge. Pour des bactéries avec une CMI à 16 mg/L, nous avons observé que des infusions prolongées étaient nécessaires pour atteindre notre cible préalablement déterminée (ƒt > CMI > 50%) chez les enfants de six mois à six ans (130 mg/kg/dose toutes les 8 heures administré sur 4 heures), alors que des durées d’infusion standard de trente minutes étaient suffisantes chez les nourrissons de deux à six mois (75 mg/kg/dose toutes les 4 heures administré sur 30 minutes). Notre étude supporte également la faisabilité et l’innocuité des infusions prolongées en pédiatrie. / Severe sepsis remains one of the most important causes of pediatric mortality around the world. Bacterial infections represent the most common cause, and effective treatment depends on the prompt administration of antibiotics. However, we observe a concerning decrease in susceptibility to antibiotics over the last decades, prompting us to reevaluate our antibiotics use. New antibiotics or novel ways of administering currently available antibiotics more efficiently are the two main alternatives when facing increased antibiotic resistance. In both cases, pharmacokinetic (PK) modeling represents an invaluable tool to guide dosing. However, PK studies in children are challenging. We used minimal risk methods to study two different antibiotics in children: Linezolid, a new oxazolidinone antibiotic that was administered to premature neonates, and piperacillin-tazobactam, a frequently used beta-lactam that we administered in a novel way using extended infusions. First, we conducted a single-center retrospective PK study of linezolid in premature neonates in the neonatal intensive care unit of the CHU Sainte-Justine. We built a population PK model using nonlinear mixed-effects modeling with plasmatic concentrations collected for therapeutic drug monitoring per standard of care. We were able to demonstrate that the dosing regimens used in the 26 neonates included in our study reached our established target (area under the curve over the minimal inhibitory concentration [AUC/MIC 0-24] > 80), and, therefore, were deemed efficient. Moreover, we collected adverse events and found that linezolid administration appeared safe in this population. We then focused on piperacillin-tazobactam extended infusions in children. Beta-lactams efficacy depends on the fraction of time that concentrations are above the MIC (ƒt > MIC). Extended infusions are a simple way to achieve higher ƒt > MIC and are well studied in adults. Based on published piperacillin-tazobactam PK parameters in children, we first conducted a simulation PK study to establish optimal extended infusions dosing in children. We then conducted a single-center prospective PK study where the established dosing regimens were administered to 89 children from two months to six years old. Of those, 79 children contributed plasma PK samples. Two PK models (piperacillin and tazobactam) were developed using nonlinear mixed-effects modeling. Simulations were conducted using our final piperacillin model, allowing us to determine optimal dosing regimens according to age. For bacteria with MICs up to 16 mg/L, extended infusions (130 mg/kg/dose every 8 hours infused over 4 hours) were needed in children six months to six years old to reach our established target (ƒt > MIC > 50%), whereas standard 30 minutes infusion (75 mg/kg/dose every 4 hours infused over 30 minutes) were adequate in infants two months to six months old. Our study also supported the feasibility and safety of extended infusions in young children.

La pipéracilline-tazobactam est une combinaison d’un antibiotique bêta-lactamine et d’un inhibiteur des bêta-lactamases fréquemment prescrite aux soins intensifs. Les patients admis aux soins intensifs présentent souvent une réponse très variable au traitement en raison des multiples changements pathophysiologiques présents dans cette population qui modifient le profil pharmacocinétique du médicament. La modélisation pharmacocinétique de population est une approche qui permet d’expliquer une partie de cette variabilité au moyen d’équations mathématiques. À l’aide d’un modèle pharmacocinétique, il est possible de décrire le devenir systémique du médicament à l’aide de paramètres clés comme la clairance et le volume de distribution. Également, ce type de modèle offre la possibilité d’effectuer des simulations de régimes posologiques pour faciliter l’atteinte des cibles thérapeutiques, dans une optique d’individualisation de la pharmacothérapie. Ce projet de maîtrise avait trois objectifs. Le premier était de documenter la variabilité associée à la pharmacocinétique de la pipéracilline-tazobactam aux soins intensifs en réalisant une revue de la littérature. Cette revue a pu relever certaines covariables significatives dans les modèles qui expliquaient une partie de la variabilité observée, comme la clairance de la créatinine et le poids. Le second avait pour but d’évaluer la performance prédictive des modèles pharmacocinétiques déjà disponibles dans la littérature pour la pipéracilline-tazobactam à l’aide d’une base de données indépendante. Parmi les modèles évalués, le meilleur a été retenu afin d’optimiser les régimes posologiques aux soins intensifs. Ainsi, grâce à ce modèle, un nomogramme prenant en considération la fonction rénale du patient a été développé pour faciliter l’atteinte des cibles thérapeutiques lors de l’administration de la pipéracilline-tazobactam. Finalement, le troisième objectif était d’évaluer l’impact d’une variation de la fraction libre de la pipéracilline-tazobactam sur la performance prédictive du modèle ainsi que son impact sur la pharmacocinétique du médicament. Cette évaluation a fait ressortir l’importance d’utiliser la concentration libre plutôt que d’utiliser la concentration totale de la pipéracilline-tazobactam pour le suivi thérapeutique, car l’assomption d’une fraction libre théorique unique pour tous les patients peut nuire à la prédiction adéquate des concentrations par un modèle pharmacocinétique en milieu clinique. / Piperacillin-tazobactam is a beta-lactam/beta-lactamase inhibitor antibiotic combination frequently prescribed in intensive care units. Admitted patients often show a large variability in treatment response due to multiple pathophysiological changes that alter the pharmacokinetic profile of the drug. Population pharmacokinetic modeling is an approach that can explain some of this variability using mathematical equations. Using a pharmacokinetic model, key parameters such as clearance and volume of distribution can be retrieved to describe the systemic exposure of the drug. Also, this type of model offers the possibility to perform simulations to find optimized dosing regimens that may facilitate the achievement of target concentrations to individualize drug therapy. This master's project had three objectives. The first was to document the variability in the pharmacokinetics of piperacillin-tazobactam in the intensive care unit by conducting a literature review. This review was able to highlight key covariates, such as creatinine clearance and body weight, that could explain the variability observed in this population. The second was to evaluate the predictive performance of pharmacokinetic models available in the literature for piperacillin-tazobactam using an independent database. Among the models evaluated, the best one was selected to offer optimized dosing regimens for critically ill patients. Thus, with this model, a nomogram that accounts for the patient's renal function was developed to facilitate the achievement of therapeutic targets of piperacillin-tazobactam. Finally, the third objective was to evaluate the impact of fluctuations in the unbound fraction of piperacillin-tazobactam on the predictive performance of the model as well as its impact on the pharmacokinetics of the drug. This evaluation highlighted the importance of using unbound piperacillin concentrations for drug monitoring over total concentrations, as applying a theoretical unbound fraction to every individual may hinder the predictive performance of pharmacokinetic model if it is used in a clinical setting.